Datasets:

griffin

/

ChemSum

like 13

Clustering of catalytic nanocompartments for enhancing an extracellular non-native cascade reaction

Compartmentalization is fundamental in nature, where the spatial segregation of biochemical reactions within and between cells ensures optimal conditions for the regulation of cascade reactions. While the distance between compartments or their interaction are essential parameters supporting the efficiency of bio-reactions, so far they have not been exploited to regulate cascade reactions between bioinspired catalytic nanocompartments. Here, we generate individual catalytic nanocompartments (CNCs) by encapsulating within polymersomes or attaching to their surface enzymes involved in a cascade reaction and then, tether the polymersomes together into clusters. By conjugating complementary DNA strands to the polymersomes' surface, DNA hybridization drove the clusterization process of enzyme-loaded polymersomes and controlled the distance between the respective catalytic nanocompartments. Owing to the close proximity of CNCs within clusters and the overall stability of the cluster architecture, the cascade reaction between spatially segregated enzymes was significantly more efficient than when the catalytic nanocompartments were not linked together by DNA duplexes. Additionally, residual DNA single strands that were not engaged in clustering, allowed for an interaction of the clusters with the cell surface as evidenced by A549 cells, where clusters decorating the surface endowed the cells with a nonnative enzymatic cascade. The self-organization into clusters of catalytic nanocompartments confining different enzymes of a cascade reaction allows for a distance control of the reaction spaces which opens new avenues for highly efficient applications in domains such as catalysis or nanomedicine.

clustering_of_catalytic_nanocompartments_for_enhancing_an_extracellular_non-native_cascade_reaction

Introduction<!>Generation and characterization of catalytic nanocompartments<!>Single CNC activity<!>CNC clustering<!>Cascade within clusters<!>Cluster localization and functionality on cells<!>Conclusions<!>Materials<!>Synthesis of diblock copolymers<!>Preparation of GOX-and LPO-CNCs<!>AMG(GOX)-CNCs<!>Catalytic nanocompartment characterizationtransmission electron microscopy (TEM)<!>Nanoparticle tracking analysis (NTA)<!>Edge Article Chemical Science<!>Enzyme quantication<!>DNA functionalization and clustering<!>Fluorescence (cross) correlation spectroscopy<!>Enzymatic assays<!>Cell viability (proliferation) assay<!>Live cell imaging of A549 cells<!>Colocalization analysis<!>Extracellular cluster functionality

<p>In nature, compartmentalization is a prerequisite for the spatiotemporal control of signalling pathways and for intra-and intercellular communication. The distance between compartments is critical for intercompartmental interactions, 1,2 with distances ranging between 20-50 nm in intracellular and synaptic communication 3,4 to above 250 mm in paracrine signalling. 5 In addition, there are various reactions or conditions that require bio-compartments such as organelles or cells to directly interact in order to communicate or transfer molecules. Signicant efforts have been made to exploit nature's designs and develop compartments in which specic enzymatic reactions take place, 6,7 or that can support cascade reactions. 8,9 Of particular interest are nanometric polymer vesicles (polymersomes), with their hollow spherical architecture, as they allow insertion of hydrophilic molecules inside their lumen (e.g. enzymes) and/or hydrophobic molecules such as membrane proteins in their membrane. 10,11 Polymersomes show improved stability compared to lipid-based vesicles, and allow tuning properties such as membrane thickness, polarity or toxicity based on the chemical versatility of polymers. [12][13][14] Moreover, it is possible to modify their surface with biological molecules that mediate cellular targeting or surface attachment or, more recently, the self-organization of polymersomes into clusters. 10,[15][16][17][18][19][20][21] When loaded with enzymes and made permeable to substrates and products, polymersomes serve as effective catalytic nanocompartments (CNCs) with a broad range of applications, such as organelle models, 12,22,23 biosensors, 10 detoxifying agents, 24 and production or release/activation of prodrugs. 10,25 More complex setups rely on enzymatic cascades, where the product of one enzyme becomes the substrate of another, similar to many biochemical pathways in cells. 26 As in communicating cellular compartments, the distance between the enzymes has been demonstrated to be a fundamental parameter in cascade reaction efficiency. 26,27 However, coencapsulation of enzymes in the same polymersome was shown to have some limitations, in particular a lack of modularity, 28 issues that are avoided in CNCs harbouring only one kind of enzyme at a time. As polymersomes are colloidal systems, the main way to modulate the inter-vesicle distance is to vary their concentration, which affects the cargo concentration as well. Another strategy to keep the distance between polymersomes small and constant is to link them together, for example by DNA, 20 resulting in polymersome clusters which, unlike their liposome counterparts, [29][30][31][32] do not precipitate. Polymersome clusters have been applied to the co-delivery of enzymes and dyes for theranostic applications. 15,33 However, their unique potential to increase the efficiency of cascade reactions by mimicking the conditions in which natural organelles or cells are in close contact has not yet been explored.</p><p>Here we present how different CNCs can be tethered together at a controlled distance and carry out a cascade reaction involving segregated compartments (Fig. 1). We advance from compartments acting in tandem without xed distance 26, 34 to a new construct where the distance between compartments is pre-determined. This architecture facilitates the diffusion of molecules between compartments: the products of the rst reaction (located in one type of CNC) will reach the second type of compartment where they will become substrates for the in situ reaction, thus supporting the overall cascade reaction. Catalytic nanocompartments are tethered together by DNA hybridization of complementary single-stranded DNA (ssDNA) exposed on the surface of different polymersomes, to promote clusterization. 20 As a model cascade, we used glucose oxidase (GOX) and lactoperoxidase (LPO): GOX oxidizes glucose into gluconic acid and H 2 O 2 which is used by LPO to oxidize a variety of substrates. More generally, the oxidase-peroxidase system is an antibacterial cascade found in many animal secretions. Based on the hydrogen peroxide coming from organic substrates, peroxidases can produce bacteriostatic compounds such as hypothiocyanites from thiocyanates, and thus have been suggested for biomedical applications such as oral plaque treatment or to counter to opportunistic infections developed in cystic brosis. [35][36][37] Moreover, indications of the GOX-LPO cascade in anticancer and antiviral activities 38,39 make it an interesting enzyme pair for possible applications.</p><p>To provide one step further in the cascade complexity, we conjugated amyloglucosidase (AMG), an exoenzyme that hydrolyses amylose into small glucose units, 48 to the surface of GOX-nanocompartments. Due to the external location of AMG, thousands of g mol À1 large amylose could be hydrolyzed to glucose that could enter the compartment and serve as substrate for the GOX encapsulated.</p><p>The co-localization of enzymes via DNA had been used for several enzymes in bulk, [40][41][42] but never for catalytic compartments. Our approach has the advantage of allowing the enzymes to move freely inside the polymersomes where they are protected from harmful environmental conditions, while inspired by organelles and cells, the compartments are kept at a close but constant distance. 26 In addition, the distance between the CNCs within a cluster can be easily controlled by modifying the length of DNA strands which affects the overall performance of the cascade reaction.</p><p>More importantly, CNC clusters were not limited to colloidal suspensions: DNA single strands that are not engaged in linking CNCs act as ligands for cell surface receptors and thereby attach the clusters to the surface where they are too big for uptake. 15,33 Such decoration endows the cell with a novel compartment, a satellite organelle bound to their plasma membrane with which cells can perform cascade reactions at the surface. With our clusters, cells could break down an otherwise inadequate substrate (amylose) to feed into a totally bio-orthogonal cascade reaction. Both features of our clusters, distance control between consecutive steps of cascade reactions and cell association are essential for developing applications in catalysis or medicine.</p><!><p>As building block for the CNCs, we synthesized an amphiphilic diblock copolymer, poly(2-methyl-2-oxazoline)-block-poly(dimethylsiloxane) PMOXA 10 -PDMS 29 according to an established procedure (Fig. S1 †). 20 Copolymers with PDMS as hydrophobic block and PMOXA as hydrophilic block have been used previously for producing catalytic compartments because generally the membrane of polymersomes is impermeable to small molecules but sufficiently exible to enable the insertion of membrane proteins. 10,11 The same polymer was also functionalized with poly(ethylene glycol)-N 3 (PEG 4 -N 3 ) (Fig. S2 †) on its PMOXA block to both provide a clickable moiety on the surface of the self-assembled structures and to ensure maximum miscibility. PEG 4 -N 3 allowed the conjugation of dibenzocyclooctyne (DBCO)-DNA strands, via strain-promoted alkyne-azide cycloaddition (SPAAC). 20 Subsequently, 50% (mol%) PMOXA 10 -PDMS 29 were mixed with 50% N 3 -PEG 4 -PMOXA 10 -PDMS 29 to have available a high density of accessible azides upon the self-assembly of polymersomes. 15,20 Fig. 1 Concepts of a GOX-LPO cascade between two clustered CNCs, tethered via complementary ssDNA, in order to facilitate the diffusion of H 2 O 2 and thus improve the overall reaction efficiency. Similarly, an AMG-GOX-LPO cascade achieves an improved diffusion of the glucose derived from amylose, and the enzyme on the surface allows the access to bulky substrates that would otherwise be out of reach for encapsulated enzymes.</p><p>Self-assembly of polymersomes encapsulating either GOX or LPO was achieved via lm rehydration and resulting nanoassemblies were extruded to decrease the size polydispersity. Melittin, a pore-forming peptide derived from bee venom, was added to the polymersomes in order to render membranes permeable to molecular ow to and from the conned enzymes. Melittin has been shown to induce pore formation in membranes formed by triblock PMOXA-PDMS-PMOXA and diblock PMOXA-PDMS without destabilizing the vesicles. 21,43 In addition, melittininduced pores are less prone to hindering to the diffusion of substrates compared to outer membrane porin (OmpF) pores, 26,34,45 making it well-suited for the permeabilization of polymersomes. The characterization of the assemblies by static and dynamic light scattering (SLS/DLS) showed that GOX-CNCs had a hydrodynamic radius (R h ) of 119 AE 8 nm and a radius of gyration (R g ) of 110 AE 2 nm, whereas LPO-CNCs were slightly bigger at 170 AE 24 nm and 150 AE 11 nm, respectively. The ratio between R h and R g , 0.9 in both cases, conrmed the production of vesicular assemblies. The difference in size between CNCs could be explained by the fact that LPO interacts with the PMOXA block and thus preferentially localizes to the membrane surface. Similarly, such interaction was seen with the related horseradish peroxidase, 45 and recently reported for LPO in giant unilamellar vesicles as well. 44 The vesicular architecture of LPO-and GOX CNCs was conrmed by TEM micrographs revealing the deformed spherical morphology typical of this kind of polymersome (Fig. 2A, B and S1 †).</p><p>The polymersome concentration, determined by nanoparticle tracking analysis (NTA), showed a 30% higher concentration for GOX-CNCs than for the smaller LPO-CNCs. At approximately 200 mg mL À1 , the concentration of encapsulated enzyme was determined to be similar for both types of CNCs (Table S1 †). To further characterize CNCs, we encapsulated uorescently labelled GOX (with Atto-488) or LPO (with DyLight 633), and analyzed the diffusion times of CNCs compared to free labelled enzyme and free dye by uorescence correlation spectroscopy (FCS). 19 The corresponding shi of the FCS autocorrelation curves to higher diffusion times indicated that both enzymes were labelled (reduced diffusion time of labelled enzyme compared to free dye) and that polymersomes were associated with labelled enzyme (signicant reduction of the diffusion time of CNCs compared to free enzyme) (Fig. 2C, D and Table S2 †). Moreover, the brightness intensity of the single species (free enzyme, CNC) allowed us to quantify the average number of dye molecules per enzyme and the enzyme molecules per polymersome. 19 The resulting 11 AE 4 GOX/ polymersome and 52 AE 32 LPO/polymersome indicated that there was no overcrowding within the CNCs (Table S2 †). Calculations of the average space in the cavity of polymersomes occupied by the enzymes revealed the total volume of GOX molecules to be around 1900 times smaller than that of the compartment, and the volume of LPO 330 times smaller. 46,47 Furthermore, only 2% of the total enzyme in the samples remained free aer purication (Table S2</p><p>Having characterized the lumen of the CNCs, we moved on to their surface (Fig. S3 and Table S3 †). By conjugating a small uorescent dye (Atto-488 DBCO) via SPAAC to azide groups of vesicles and measuring the uorescence intensity per polymersome via FCS, we estimated the number of easily accessible azides on the CNC surface to be 104 AE 24 which is in line with published data. 15 For linking together the CNCs, we designed two DBCOcoupled, 22 nucleotides long, complementary ssDNA, each comprising a short, 5 0 non-complementary thymidine sequence as spacer to improve the DNA hybridization (Table S4 †). 20 DBCO-ssDNA was then conjugated to the azide-functionalized polymersome membrane via SPAAC. We quantied the number of ssDNA conjugated to each type of CNC by the means of hybridizing complementary ssDNA labelled with a uorescent dye (Atto-488 for GOX-and Cy5 for LPO-CNCs) and FCS (Fig. S3 †). While the surface density of the ssDNA on the polymersomes was quite disperse (ranging from few to above 100 ssDNA per polymersome), the numbers were in the range shown to promote the clustering of CNCs (Table S3 †). 15 Finally, amyloglucosidase (AMG) was conjugated via an NHS-PEG 3 -DBCO linker to free azide moieties on preformed, ssDNA-GOX-CNCs. The resulting AMG(GOX)-CNCs hold one type of enzyme (GOX) in their cavity and expose another (AMG) on the outer surface where it is available to substrates too big to enter CNCs through melittin pores. By further clicking Atto-488-DBCO to the remaining azides not occupied by either ssDNA or AMG, we estimated that about 35 AE 4 AMG were present on each GOX-CNC (Table S3 †).</p><p>The molecular weight 48 and dimensions of AMG from Aspergillus niger (75 Å Â 45 Å Â 40 Å, PDB code 6YQ7, manually measured using Pymol) indicate that the size of AMG is orders of magnitude smaller than that of polymersomes. Accordingly, FCS measurements did not reveal any signicant shi in the diffusion times if AMG was conjugated to GOX-CNCs (6000 ms for AMG-conjugated CNCs versus 5486 AE 2510 ms for CNCs without AMG).</p><!><p>We tested the enzyme functionality of the single CNCs (without DNA conjugates) based on the oxidative conversion of Amplex Red (AR) to resorun. LPO-CNCs were directly incubated with AR and H 2 O 2 (Fig. 3A), whereas GOX-CNCs were incubated with glucose, AR and free LPO, ensuring that the H 2 O 2 produced by glucose oxidation can directly diffuse to the peroxidase catalyzing the conversion of AR to resorun (Fig. 3B). Both types of CNC showed activity only when rendered permeable by melittin (blue curve), conrming that substrates need to have access to the conned enzymes. The small amount of residual nonencapsulated LPO detected in FCS measurements was probably inactivated when testing enzyme activity of LPO-CNCs. The absence of LPO activity outside of LPO-CNCs corroborates the protection of enzymes inside polymersomes, which is in line with previously reported CNCs. 26,34,49 AMG(GOX)-CNCs behaved similarly: when supplied externally with amylose, the glucose released by AMG-catalyzed hydrolysis could enter the CNC only if the membrane was permeabilized by melittin (Fig. 3C). We could detect only minimal non-specic oxidation of AR by free LPO (Fig. 3B and C). Notably, the inuence of permeabilization via melittin, reected by the difference between blue (permeabilized) and red (non-permeabilized) curves, was less evident as the system gained complexity: from a large difference when only LPO was involved, to a relatively small difference with a 3-enzyme system.</p><!><p>Having conrmed the enzymatic activity of individual CNCs, we next aimed to establish clusters of CNCs by taking advantage of DNA hybridization. For this purpose, complementary ssDNAs were conjugated to the azide groups on the membranes of the respective polymersomes. Aer mixing the functionalized CNCs at an equal ratio, we followed the clustering process over time via dynamic light scattering (DLS) (Fig. 4A). The number of ssDNA per polymersome was sufficient to induce clustering. Aer 14 h, a plateau was reached where LPO-GOX clusters had an average diameter, D H of about 700 nm. AMG(GOX)-LPO-CNCs formed smaller clusters of around 500 nm in diameter, possibly due to AMG-associated repulsion. We then used uorescence cross-correlation spectroscopy (FCCS) to characterise GOX-LPO cluster formation (Fig. 4B), as both conned enzymes could be uorescently labelled with different uorophores.</p><p>FCCS can be used to detect the association of uorescent species in Brownian motion when their separate signals are correlated (higher G(s)). 50 We observed an increase in the crosscorrelation between the uorescently-labelled CNCs upon clustering (Fig. 4B, blue curve). The signal was absent if uorescent CNCs without ssDNA were mixed (Fig. 4B, black curve), conrming that DNA hybridization is key to cluster formation. In addition, the association of uorescently labelled CNCs into small clusters (3-4 vesicles) was visualized by TEM (Fig. 4C).</p><p>Once the clusters were formed, we determined the mean distance between GOX-and LPO-polymersomes, both when clustered and un-clustered, as the distance between surfaceconjugated AMG and encapsulated GOX is given. For clustered polymersomes, the distance for productive molecule transfer (i.e. H 2 O 2 from GOX to LPO) depends on the length of ssDNA. Having both paired and unpaired bases, the average DNA length, L in nm, (corresponding to the vesicle-to-vesicle direct surface distance) is based on eqn (1): 51</p><p>where n BP and n S are the number of paired and unpaired bases, respectively. L was estimated to be 14.9 nm, which is in the range of the width of a synaptic cle 4 and of some interorganelle distances found in cells. 27,52 For non-clustered vesicles, we adapted a previously developed equation for lattices of heterogeneous, rigid particles (eqn (2)):</p><p>where the mean inter-vesicle distance hDi depends on the mean size of polymersomes d, the spatial distribution parameter x (a measure of the degree of dispersion of the system, xed to 1.1 (ref.</p><p>53)), the total volume fraction occupied by polymersomes f and the geometric standard deviation s of the polymersome's sizes.</p><p>However, in our system, not all polymersomes were equal, as the "bridging molecule", H 2 O 2 , could only productively go from a GOX-to an LPO-CNC, but we had to consider both their relative concentrations and sizes with the weights W 4 and W d to sum the two contributions of GOX-and LPO-CNC, obtaining eqn (3). A more in-depth explanation of the geometric meaning of the equations and the rationale behind the further derivation of eqn (3) are presented in Fig. S4 and the related section in ESI. † Applying the enzyme concentrations used to test the untethered CNC cascade (see Materials and methods), we calculated hDi ¼ 2.3 AE 0.2 mm, in agreement with the distance calculated for other CNCs working in similar conditions. 26 This means that the same type of polymersome could carry the same concentration of enzymes, but then communicate at distances ranging from 10 (when untethered) to less than 0.1 times (when tethered) their average diameter. Interestingly, the longest distance is in the order of magnitude of autocrine signalling, the shortest in that of synaptic signalling. 4,5 Thus, these CNCs could be an interesting way to mimic distance effects in cellular communication.</p><!><p>Since polymersome suspensions can be easily nebulized and inhaled, 54 airways and lungs could be possible targets for in situ prodrug metabolism or anti-microbial protection. To mimic the conditions of the lungs, we monitored the oxidation of AR catalyzed by clustered and unclustered LPO at 37 C using glucose concentrations close to those found in the airway surface liquid. 55 As the concentration of enzymes was constant, both with and without clusters, only the distance between CNCs of a different kind varied.</p><p>The inter-cluster distance was calculated with eqn (2), without weights, as all clusters had both types of CNC and thus, no effect would come from the different relative populations of GOX and LPO-CNC. The inter-cluster distance was calculated 2.3 AE 0.1 mm at the same enzyme concentrations used for untethered CNCs. Therefore, CNCs clusters were effectively acting like a "2-enzyme system", as inter-cluster distances were comparable to un-clustered inter-polymersome distances. The comparison between CNCs inside clusters and when freely moving un-clustered, clearly shows an increase in enzymatic activity. On the contrary, un-clustered CNCs were threefold slower than CNC clusters (Fig. 5A), an effect of controlled cocompartmentalization already observed in giant</p><p>polymersomes and other cell-mimics, but never in colloidal vesicle networks. 44,56 In a previous study we had showed that in the case of un-clustered CNCs, inter-polymersome distances above 1.5 mm signicantly hinder the cascade reaction due to the slow diffusion of the "bridging" molecule (the product of the rst reaction that is released from the rst type of CNC and serves as substrate for the second reaction inside the second type of CNC). 26 On the contrary, CNC clustering averted this limitation, resulting in a considerably increased reaction rate at a constant amount of encapsulated enzymes. 19 We observed the same effect with AMG(GOX)-LPO clusters (Fig. 5B). The effect was less pronounced due to the inherently low catalytic efficiency of AMG which therefore presents an additional bottleneck in the cascade. Interestingly, we found that GOX-LPO clusters in the presence of non-conjugated AMG showed a lower activity than in clusters with conjugated AMG (magenta curve): evidently, they appeared to accelerate the kinetics of the rst, slowest reaction. Despite AMG being ratelimiting, cascade efficiency depended on this step and could not be bypassed by additional glucose, as shown by clusters supplied with glucose and amylose at the same time (Fig. S5 †).</p><!><p>Envisioning possible biological applications, we rst tested the effect of unclustered, empty polymersomes up to a concentration of 1 mg mL À1 (Fig. S6A †), unclustered LPO-and GOX-CNCs at 0.2 mg mL À1 (Fig. S6B †), and clustered (AMG)GOX-LPO and GOX-LPO at 0.5 mg mL À1 (Fig. S6C †) on the viability of A549 cells, a lung carcinoma cell line used as a model for lung protection from bacterial infections and ROS therapy. 57 Notably, GOX-LPO slightly decreased cell viability compared to (AMG) GOX-LPO which had no effect (Fig. S6C †). This difference may be attributed to the slower H 2 O 2 production associated with AMG (see Fig. 5). Polymersome clusters were previously shown to accumulate at the surface of epithelial cells by the interaction of DNA strands that remain unpaired on the polymersomes aer clustering. 15,33 We probed the interactions of GOX-LPO clusters where GOX-and LPO-CNCs were labelled with different uorophores, with A549 cells. Atto488-GOX/DyLight633-LPOclusters were incubated at 0.2 mg mL À1 with A549 cells for 24 hours and examined by confocal laser scanning microscopy (CLSM) (Fig. 6A-D, S7 and S8 †). Confocal Z-stacks revealed a colocalization of uorescence signals at the surface of the cells. The colocalization analysis of the dyes (Fig. 6C, S7, S8 and Table S5 †) yielded a Pearson's coefficient of 0.27 (P ¼ 1) for clusters, and À0.73 for non-clustered vesicles. These results show that clustered CNCs are present on the cell surface where they appear to remain linked, 15,33 whereas unlinked CNCs, known to be taken up by cells, 19,34 are independently distributed throughout the confocal planes. The relatively low Pearson's and Manders' coefficients (Table S5 †) -the closer to 1, the higher the colocalization-are not surprising: if the clusters are wellspaced and big enough, the resolution might be sufficient to resolve the separate polymersomes, at least partially (Fig. 6), as observed in previous studies. 15,20,33 The association of clusters with the cell surface allowed us to test whether the cascade reaction could take place ectopically, in the immediate vicinity of the cell surface. Our group has previously described the use of clustered CNCs to localize conned enzyme activity to the cell surface. 15,33 However, in these studies, clusters comprised only one type of CNC, i.e. the enzymatic activity was limited and the full potential of clustering was not exploited. In contrast, by positioning clustered GOX-LPO CNCs and AMG(GOX)-LPO at the cell surface, the cells were endowed with an extracellular, two-and three-step cascade reaction. The latter in particular, provided A549 cells with the ability to metabolize amylose (Fig. 7). 58 Our ndings support the potential of clusters as co-delivery systems to epithelial cells, where they modulate the microenvironment at the cell surface, or if internalized, function as articial organelles. Moreover, this very cascade might be suitable for future on-site bio-catalysis based on starch, as peroxidase bio-transformations have been developed for environmental and industrial applications. 59,60</p><!><p>The distance between enzymes is an important factor in the optimization of natural and man-made cascade reactions. Adequate spacing aids in offsetting limiting factors, such as diffusion across membranes, and increases the overall efficiency of the system. Inspired by organelles and intra-and intercellular interactions, we developed clusters of catalytic nanocompartments that sustain efficient cascade reactions. Separate nanocompartments were rst independently loaded with enzymes that are able to act in tandem, and then tethered together by complementary DNA strands such as to selforganise into clusters. In addition, a three-step cascade reaction has been successfully established by coupling an upstream enzyme to the surface of GOX-CNCs. Linking compartments by DNA hybridization lends itself to precise tuning of the intercompartment distance, ranging from distances typical of paracrine signalling for non-adjacent compartments, to those found in some inter-organelle interactions or synaptic signalling for neighbouring compartments. The ability to control the distance between the CNCs supports their use as non-living models for bio-communication. In addition, clusters can modulate the extracellular microenvironment by harnessing cells and improving the efficiency of cascade reactions by segregating reaction steps in physically dened spaces. Besides, such compartment clusters are generated in a modular fashion, such that they can readily be expanded by changing the encapsulated enzymes, by combining more than two types of compartments or by modifying the DNA links. Owing to the combinatorial und functional diversity, clusters of catalytic nanocompartments open new avenues in various domains including bio-catalysis, therapeutics and other biomedical applications.</p><!><p>DyLight 633 NHS ester and Atto 488-NHS ester were purchased from ThermoFisher Scientic (USA). All other reported compounds were purchased from Sigma-Aldrich (USA) unless stated otherwise.</p><!><p>The OH-terminated diblock PMOXA 10 -b-PDMS 29 was synthesized according to the already reported procedure. 20 Briey, OHterminated PMOXA 10 -b-PDMS 29 was dissolved into 5 mL anhydrous chloroform, then succinic anhydride (6.5 mg, 0.066 mmol), 4-dimethylaminopyridine (1.32 mg, 0.011 mmol) and TEA (8.7 mg, 0.088 mmol) were added. Aer deoxygenating by three vacuum-argon cycles, the mixture was stirred for another 72 h at RT under Ar atmosphere. Finally, 180 mg of a colorless solid product was obtained aer the ultraltration, yield 90%. 1H NMR (500 MHz, CDCl3) d 3.65-3.23 (m, 39H), 2.28-2.00 (m, 28H), 1.30 (tt, J ¼ 7.7, 4.5 Hz, 5H), 0.87 (t, J ¼ 6.8 Hz, 4H), 0.56-0.44 (m, 4H), 0.06 (s, 171H) (Fig. S1 †). To produce PMOXA 10 -b-PDMS 29 -PEG 4 -N 3 , the polymer (100 mg) was then rst dissolved into anhydrous chloroform, then 11-azido-3,6,9-trioxaundecan-1-amine (11.80 mg, 0.055 mmol), N,N 0 -dicyclohexylcarbodiimide (15.6 mg, 0.078 mmol) and 4-dimethylaminopyridine (1.2 mg, 0.01 mmol) were added into the above solution. Aer deoxygenating three times, the mixture was further stirred for another 48 h, at RT. Finally, a colorless solid product was obtained aer ultraltration. The polymer was characterized again via NMR and GPC. 1H NMR (500 MHz, CDCl3) d 3.65-3.28 (m, 43H), 2.23-2.03 (m, 29H), 1.59-1.52 (m, 2H), 1.30 (tt, J ¼ 7.5, 3.8 Hz, 5H), 0.87 (t, J ¼ 6.8 Hz, 3H), 0.50 (ddd, J ¼ 15.9, 8.9, 4.5 Hz, 4H), 0.06 (s, 172H) (Fig. S2 †).</p><!><p>CNCs were prepared at RT, with 50% (molar ratio) of the azidefunctionalized polymer. Films were rehydrated to a nal polymer concentration of 10 mg mL À1 with 1 mg of GOX or LPO in PBS (pH 7) and 25 ml of melittin 1 mM (from bee venom). Samples were extruded through an Avanti mini-extruder (Avanti Polar Lipids, USA) with a 200 nm pore diameter polycarbonate membrane for GOX; LPO-CNCs were rst extruded through 400 nm and then 200 nm, 11 times each. Non-encapsulated enzyme was removed through size exclusion chromatography (SEC) (Sepharose 4B column; 30 cm length) and recovered for quantication.</p><!><p>1 mL of amyloglucosidase from Aspergillus niger (AMG) $260 U mL À1 aqueous solution, was mixed with 200 mL of dibenzocyclooctyne-PEG 4 -N-hydroxysuccinimidyl ester (DBCO-PEG 4 -NHS ester) in DMF and 800 mL of 0.2 M sodium bicarbonate at pH 8-9 was added. The solution was let it stir at room temperature for 16 h. The functionalized protein was puried with Amicon Ultra-0.5 mL-30 kDa cutoff by washing 3 times in PBS for 5 min at 10 000 rpm. 200 mM stock solutions of the DBCO-modied DNA strands (Microsynth, Switzerland) were prepared in nuclease-free water. 50 mL of each solution was added to 150 mL of corresponding CNCs, and made to react at 37 C overnight. The vesicles were thus puried with a 10 cm-Sepharose 2B column, mixed 1 : 1 (volumetrically) and let to rest at 37 or 4 C overnight to allow clustering for further experiments and then used with no further purication.</p><p>Catalytic nanocompartment characterizationstatic and dynamic light scattering SLS and DLS experiments were performed on a setup from LS instruments (Switzerland), equipped with a He-Ne 21 mW laser (l ¼ 632.8 nm) at scattering angles from 30 to 55 at 25 C. The radius of gyration (R g ) was obtained from the SLS data with a Guinier plot. The intensity versus angle curve of a diluted sample (to suppress multiple scattering) was t with a linear regression and the slope of the curve m was used to calculate R g according to the equation</p><p>The error was calculated on the standard error of the slope.</p><p>In the case of SLS, second order cumulant analysis of the data between 30 and 155 was performed to obtain the R h .</p><p>Clustering was followed on a Zetasizer Nano ZSP (Malvern Instruments, UK) at 20 C, where 50 mL of each DNAfunctionalized CNC were added to 200 mL of PBS, measuring the R h for 14 hours.</p><!><p>CNC suspensions in PBS at 0.25 mg mL À1 were deposited on glow-discharged carbon grids (Quantifoil, Germany) stained with 1.5% uranyl acetate solution and deposited on carboncoated copper grids. A transmission electron microscope (Philips Morgagni 268D) at 293 K was used.</p><!><p>NTA was used as further analysis of particle size and concentration, on a NanoSight NS300 (Malvern Panalytical Ltd., UK), using a ow cell (100 mL min À1 ), 1 : 1000 concentration in freshly ltered PBS, yielding particle R h and concentration (particle per mL).</p><!><p>Open Access Article. Published on 31 August 2021. Downloaded on 7/6/2022 2:26:52 AM. This article is licensed under a Creative Commons Attribution-NonCommercial 3.0 Unported Licence.</p><!><p>Unencapsulated enzyme was recovered from melittin-less samples, and quantied at 280 nm, using a Nanodrop 2000 UV-vis spectrophotometer (ThermoFisher, USA).</p><!><p>200 mM stock solutions of the DBCO-modied DNA strands (Microsynth, Switzerland) were prepared in nuclease-free water. 50 mL of each solution was added to 150 mL of corresponding CNCs, and made to react at 37 C overnight. The vesicles were thus puried with a 10 cm-Sepharose 2B column, mixed 1 : 1 (volumetrically) and let to rest at 37 or 4 C overnight to allow clustering for further experiments and then used with no further purication.</p><!><p>Stock solutions of GOX (2 mg mL À1 ) and LPO (2 mg mL À1 ) were prepared in 0.1 M Na 2 CO 3 buffer. 5 mL of a 1.5 mM Atto-488 NHS ester in DMSO solution was added to 1 mL of the GOX stock solution and 5 mL of 1.5 mM DyLight 633-NHS ester in DMSO was added to 1 mL of LPO solution. Both labelling reactions were mixed overnight at 4 C. Free dye was removed by spin ltration with Amicon Spin Filters 30 MWCO (Merck, Germany). Upon purication, labelled enzymes were used directly and polymersomes were formed as previously described, with no melittin added. A 488 nm argon laser was used to excite ATTO 488 and a 633 nm HeNe laser was used for DyLight633. The two lasers were passed through MBS488 and MBS488/561/633 lters and the signals were detected in the range of 500-532 nm and 657-690 nm, respectively. The pinholes were adjusted to maximize the count rate using the respective free dye in PBS and the sample volumes were 15 mL. Fluorescent uctuations over time were recorded for 20 Â 5 s. The raw data was processed and analyzed using ZEN soware. Autocorrelation curves were tted by a two-component model, except for dye-only samples.</p><p>where f 1 and f 2 are, respectively, the fraction of the particles of the corresponding component 1 (dye) or 2 (vesicles), s D1 represents the diffusion time of the dye and s D2 the diffusion time of the vesicles, T the fraction of uorophores in triplet state with triplet time s trip , N is the number of particles and R the structural parameter, xed at 5, according to the manufacturer's guidelines. The s trip and s D of free dye were determined independently, and subsequently xed in the tting procedure for dye-stained vesicles. The degree of labelling (DOL) was obtained from the ratio of the counts per molecule (CPM)</p><p>Similarly, the enzymes per vesicle were obtained by</p><p>To quantify DNA, 11T-less strands (thus, 22a and 22b) were used, labeled with Cy5 and Atto-488, respectively. An excess amount (10 mL of a 200 mM stock) was added to vesicles with complementary strands, clustered and then puried via SEC.</p><p>Similar to FCS, dual-colour uorescence cross-correlation spectroscopy (FCCS) was performed, with the same system, on CNCs containing the labelled enzymes and either nonfunctionalized, or conjugated to their respective DNA strands, then mixed, incubated at 37 and measured with both lasers simultaneously, in FCCS mode.</p><!><p>GOX-CNC. 20 mL of GOX-CNCs (with or without melittin), glucose (nal concentration 60 mM), free LPO (nal concentration 2 mg mL À1 ) and 2 mL of Amplex Red (AR, 100 mM) were added to each well of a 96-well plate. The enzymatic kinetic reaction was determined by monitoring the formation of resorun at 560 nm for 20 minutes using a Spectramax id3 plate reader.</p><p>AMG(GOX)-CNC. 20 mL of GOX-CNCs (with or without melittin), amylose from potato (5 mg mL À1 in DMSO stock; nal concentration 50 mg mL À1 ), free LPO (nal concentration 2 mg mL À1 ) and 2 mL of AR (100 mM) were added to each well. The enzymatic kinetic reaction was determined by monitoring the formation of resorun for 20 minutes.</p><p>LPO-CNC. For LPO-CNCs, 20 mL of LPO-CNCs (with or without melittin), H 2 O 2 (nal concentration 10 mM) and 2 mL of AR (100 mM) were added to each well. The enzymatic kinetic reaction was determined by monitoring the formation of resorun for 20 minutes.</p><p>GOX-LPO cascade. Knowing the sample concentration aer workup, GOX-CNC (free or clustered) were added at a nal concentration of 8 mg mL À1 and LPO-CNC (likewise) to 7 mg mL À1 , with 60 mM glucose and 2 mL of AR 100 mM. To mimic</p><p>a biological setting, the reaction was followed at 37 C for 4 hours. AMG-GOX-LPO cascade. Knowing the sample concentration aer workup, (AMG)GOX-CNC (free or clustered) were added at a nal concentration of 8 mg mL À1 and LPO-CNC (likewise) to 7 mg mL À1 , with 50 mg mL À1 amylose and 2 mL of AR 100 mM. To mimic a biological setting, the reaction was followed at 37 C for 4 hours.</p><!><p>For cell viability assessment, a CellTiter 96 Aqueous One Solution Cell Proliferation Assay (MTS, Promega) was used according to manufacturer instructions. A549 cells were seeded (5000 cells per well in 100 mL of cell culture medium) in a 96-well plate and incubated for 24 h (n ¼ 4 per sample). Aer 24 h, empty polymersomes, CNCs, CNCs with ssDNA, and CNC clusters were added to the cells at the concentrations indicated to reach a nal volume of 200 mL per well. Aer 24 h incubation, 20 mL of MTS reagent was added to each well. Aer 3 h, absorbance was measured at 490 nm using a Spectramax id3 plate reader. Background absorbance from control wells containing all assay components without cells was subtracted from each well, and data were normalized to control cells containing all components and PBS instead of CNCs. One-way ANOVA followed by Tukey's post hoc test was performed to determine whether cell viability was signicantly affected by the treatment.</p><!><p>Freshly trypsinized A549 human carcinoma cells were seeded at a density of 6 Â 10 4 cells per well in an 8-well glass bottom ibidi plate. Aer 24 h, the cell supernatant was removed and replaced with uorescently-labelled CNCs, CNC clusters at a nal concentration of 0.2 mg mL À1 , or PBS as control.</p><p>Aer carefully removing unbound CNCs/clusters, cells were imaged by confocal laser scanning microscopy (CLSM) on a LSM 880 confocal laser microscope with a 40Â, 1.2 water immersion C-Apochromat objective lens, using Atto-488 laser and DyLight-633 light path parameters.</p><!><p>CLSM micrographs were analysed with ImageJ Coloc2 plugin, using both Pearson's (PCC) and Manders' threshold coefficient for the red to the green channel (tM1).</p><p>For PCC,</p><p>where Ch1 and Ch2 are the ratios between colocalized pixels and total pixels for their respective channels. Similarly, for tM1,</p><p>PCC quanties the degree to which two channels follow a simple linear relationship of intensity. Values can range from À1 (an inverse or "anti-colocalization" relationship), to 0 (a random cloud of no relationship), or +1 (a perfect linear slope); tM1 is similar to PCC but ranges from 0 to +1. It does not incorporate a relationship to mean intensity (as with Pearson's), so it is mostly sensitive to only the overlap alone above the threshold. 61 Costes' p-value was calculated by the plugin to determine whether the result was statistically signicant.</p><!><p>A549 cells were seeded in a 96-well plate at a density of 5000 cells per well in 100 mL cell culture medium, and incubated for 24 h at 37 C. Then, cells were dosed for 24 h with 50 mL of GOX-LPO and AMG(GOX)-LPO clusters. AR (1 mM, nal concentration in 200 mL nal volume) and AR/amylose (1 mM/50 mg mL À1 ) was added to each well containing GOX-LPO and AMG(GOX)-LPO clusters, respectively. The enzyme kinetics of clusters on the cell surface was determined by monitoring the formation of resorun at 37 C for 4 hours.</p>

Royal Society of Chemistry (RSC)

Periodic DFT study of acidic trace atmospheric gas molecule adsorption on Ca and Fe doped MgO (001) surface basic sites

The electronic properties of undoped and Ca or Fe doped MgO (001) surfaces, as well as their propensity towards atmospheric acidic gas (CO2, SO2 and NO2) uptake was investigated with an emphasis on gas adsorption on the basic MgO oxygen surface sites, Osurf, using periodic Density Functional Theory (DFT) calculations. Adsorption energy calculations show that MgO doping will provide stronger interactions of the adsorbate with the Osurf sites than the undoped MgO for a given adsorbate molecule. Charge transfer from the iron atom in Fe doped MgO (001) to NO2 was shown to increase the binding interaction between adsorbate by an order of magnitude, when compared to that of undoped and Ca doped MgO (001) surfaces. Secondary binding interactions of adsorbate oxygen atoms were observed with surface magnesium sites at distances close to those of the Mg-O bond within the crystal. These interactions may serve as a preliminary step for adsorption and facilitate further adsorbate transformations into other binding configurations. Impacts on global atmospheric chemistry are discussed as these adsorption phenomena can affect atmospheric gas budgets via altered partitioning and retention on mineral aerosol surfaces.

periodic_dft_study_of_acidic_trace_atmospheric_gas_molecule_adsorption_on_ca_and_fe_doped_mgo_(001)_

Introduction<!>Theoretical methods<!>Structural and electronic properties of undoped and Ca or Fe doped MgO (001) surfaces<!>Adsorption of CO2, SO2 and NO2 on undoped and Ca and Fe doped MgO (001) surfaces<!>Conclusions and Atmospheric Implications

<p>MgO is an important compound in the environment and is found in the form of mineral periclase with measured concentrations in the continental crust of ~3.7%.1 MgO is also a typical atmospheric mineral dust aerosol component.2 Mineral aerosol is produced over deserts by natural processes3 and, once airborne, can be transported over long distances.4 In addition, mineral dust can originate from anthropogenic sources with small ambient particulate matter levels higher in large city environments.5 As natural minerals are rarely pure in the environment, naturally-occuring MgO can have varying concentrations of other elements substituted into its crystalline lattice with Ca and Fe reported as main dopants with concentrations of Fe as high as 5–10 wt%.6,7 Currently, the reactivity of doped mineral dust and their components, such as Ca and Fe doped MgO, towards atmospheric gas surface uptake is not well understood.</p><p>It is well known that mineral aerosol particles will act as reactive sinks for trace atmospheric gases via heterogeneous uptake. Thus, surface reactions on mineral dust particles can affect atmospheric chemistry and gas phase budgets via reactions on aerosol particles.8 Additionally, alterations to mineral dust surface composition upon gas/surface uptake can change the reactive nature and the radiative impacts of the aerosol particles on climate.9 Furthermore, uptake of trace atmospheric acidic gases, such as CO2, SO2 and NO2, on basic mineral surface sites can reduce the acidity of rain as well as direct climate effects due to removal of important greenhouse gases from the atmosphere, of which CO2 is the most significant greenhouse gas10 and is considered in the current study.</p><p>Adsorption of CO2, SO2 and NO2 on mineral dust components has been thoroughly investigated using laboratory11,12 and theoretical methods13,14 or combination thereof. Additionally, adsorption properties and speciation of CO2,15–20 SO215,18,19,21–23 and NO219,24,25 on undoped MgO has been investigated extensively both experimentally and theoretically, providing a foundation for further investigations on more complex and environmentally relevant, doped MgO systems. Fundamentally, MgO surfaces possess both acidic and basic surface sites. Acidic gas adsorption on MgO, however, will predominantly be determined by the interaction of the acidic adsorbate atoms with basic MgO (001) surface sites.26 Due to the purely ionic Mg-O bond character, electron rich surface oxygen sites, Osurf, will act as strong bases by donating electrons to the adsorbing molecule, whereas electron deficient magnesium sites, Mgsurf, will act as weak electron acceptors.27 Hence, to a first approximation, Osurf can be considered as the most important reactive site of the MgO surface responsible for the binding of acidic atmospheric gases via electron pair donation. Other factors that can affect MgO surface acidity/basicity are adsorbed hydroxyl groups resulting from water dissociation at the surface.28,29 These surface hydroxyls are amphoteric in nature. For example, surface hydroxyls will act as electron donors via nucleophilic attack on an acidic CO2 molecule. Furthermore, surface hydroxyl sites, depending on their location on steps, corners and kinks, will have different coordination numbers, thermal stabilities and, consequentially, reactivities.30 Collectively, acidic gas adsorption on natural MgO containing surfaces will be complex due to the variety of reactive surface sites available on the MgO surface, the degree and nature of hydroxylation, the presence of natural dopants and the coordination number of surface sites. Thus, systematic computational studies are necessary for elucidating the various mechanisms of surface adsorption.</p><p>In this work we focus on the effects of doping the MgO surface using common naturally-occurring dopants, such as Ca and Fe, on the propensity of the MgO surface to reactive uptake of three of the most common acidic gases found in the atmosphere, including CO2, SO2 and NO2. It has already been shown that dopants change MgO surface properties.31 A typical example was reported by Orlando et al.32 where Li doping of MgO resulted in the formation of a very reactive O− species in the surface layer that exhibited a relatively small energy barrier (18 kcal/mol) for methane deprotonation. This type of enhanced reactivity due to doping within the mineral surface layer is expected to occur in natural environments where MgO is present as an atmospheric aerosol component. The resulting doped MgO surfaces can affect adsorption properties of acidic gases, such as CO2, SO2 and NO2, thus altering their removal rate from the atmosphere. With this in mind, we performed periodic density functional theory (DFT) calculations to elucidate initial adsorption mechanisms of acidic gases on Ca and Fe doped MgO surfaces.</p><!><p>The periodic ab initio solid state program suite CRYSTAL'09 was used in all calculations.33,34 This program uses functions localized at atoms as the basis for expansion of the crystalline orbitals via linear combination of atomic orbitals (LCAO). Basis sets used for Mg and O are based on previous work35 with an additional d-exponent of 0.5 added to O, whereas those for Ca, Fe, C, S and N were obtained from the University of Torino CRYSTAL basis set library36 (accessed Fall 2011). Namely, Ca (86-511d21G),37 Fe (86-411d41G),38 C (6-31d1G),39 S (86-311d1G),40 and N (6-31d1G)39 all electron basis sets were used.</p><p>Spin unrestricted Hartree-Fock formalism and hybrid UB3LYP41,42 Hamiltonian was used in all calculations. B3LYP functional has been shown to reproduce well MgO properties, including bandgap and vibrational frequencies.43 The DFT exchange–correlation contribution is evaluated by numerical integration over the unit cell volume. Radial and angular points of the grid were generated through Gauss–Legendre radial quadrature and Lebedev two-dimensional angular point distributions with a pruned grid of 75 radial and 974 angular points. The level of accuracy in evaluating the Coulomb and Hartree–Fock exchange series was controlled by five parameters,33 and values of 7, 7, 7, 7, 14 were used. The reciprocal space integration was performed by sampling the Brillouin zone with the 6×6×1 Pack-Monkhorst net.44 The Fermi energy (EF) was defined as the top of the valence band, while the bandgap was defined as the difference between the bottom of the conduction band and the top of the valence band. In all figures, EF was aligned with zero x-axis value.</p><p>By far the most stable MgO surface plane is (001) with 1.16 J/m2 and the closest higher index plane higher in surface energy by 0.15 J/m2.45 As such, the MgO (001) surface was modeled as an infinite slab with periodicity in two dimensions as a reactive acidic gas adsorption surface. The initial structure was built using experimental crystallographic MgO (periclase) data with a 4.21 Å cell parameter and a 2.11 Å Mg-O bond length.46 A supercell was constructed with dimensions of 8.93×8.93 Å having 9 Mg-O units in every layer and was propagated parallel to the (001) plane. Dopant position was selected in the second, subsurface layer, similar to Li doped MgO model used earlier.32 More importantly, doped MgO modeled here and present as atmospheric aerosol component originates from the Earth's crust under high temperature-low oxygen pressure conditions. Under these conditions, intrinsic (self-diffusion)47 or extrinsic (counterdiffusion)48 diffusion mechanisms, facilitated by the structural oxygen defects, proceed with the resulting dopant penetration depth in the order of microns47. Any long range interactions between the dopant atom and the adsorbing molecule will vanish quickly with dopant localization depth while dopant atom in the surface layer will have almost the same properties towards adsorbing molecule as its pure oxide. Thus of interest in this work was dopant-surface oxygen-adsorbate interaction which would possess properties different from that of bulk MgO or dopant oxide. Structure optimizations were performed using analytical energy gradients with respect to atomic coordinates within a quasi-Newton scheme combined with the Broyden–Fletcher–Goldfarb–Shanno scheme for Hessian updating.49–52 Convergence was checked on both gradient components and nuclear displacements and was signaled when RMS gradient was 0.0003 Hartree/Bohr and RMS displacement was 0.0012 Bohr. Symmetry, where available, was fully exploited in all calculations.</p><p>The adsorption energies of the adsorbed molecules on undoped and Ca or Fe doped MgO (001) surfaces were calculated as the energy difference between the total energies of the MgO (001) surface with molecule adsorbed and those of the molecule in gas geometry and unreconstructed MgO (001) surface themselves, such that, Eadsorption = Etotal − Esurface − Emolecule. Adsorption energies were basis set superposition error (BSSE)53 corrected by removing nuclear charge and the shell electron charges of the selected atoms but leaving the basis set at the atomic position.</p><!><p>Atomic position optimization was performed using a 5 layer MgO supercell propagated parallel to (001) plane. Figure 1 shows the representative computed structure with the view oriented perpendicular to the surface. A box outlines the region of interest where Mg substitution with Ca and Fe and gas adsorption was modeled. As a result of the optimization, Osurf atoms underwent almost negligible outward relaxation of 0.01 Å, whereas Mgsurf atoms relaxed inwards by 0.05 Å with an Mg-O bond length of 2.06 Å (Figure 1b). The Mg-O bond length between the 2nd and 3rd layers remained 2.11 Å, whereas the O-Mg bond length was slightly shorter at 2.10 Å. The total surface relaxation energy during the atomic position optimization per 90 atom supercell was −18.01 kcal/mol. Lateral Mg-O bond lengths, e.g. those parallel to (001) plane, remained essentially unchanged, close to 2.11 Å. From this perspective, bulk-to-surface relaxation did not perturb the symmetry of the MgO610− coordination unit with the Mg atom situated in the 2nd layer maintaining an octahedral environment. The calculated total Mulliken charges were 9.89, 9.92, 10.10 and 10.09 e− for Osurf, Obulk, Mgsurf and Mgbulk, respectively, which can be used for comparison with gas adsorbed structures. The total Mg-O bond population overlap between the atoms in the 1st layer, as well as those in the 1st and 2nd layers was 0.01 e− indicating ionic bonding.</p><p>Substitution of the 2nd layer Mg atom with Ca (Figure 1c) resulted in the elongation of all metal-oxygen bonds in the vicinity of the Ca atom. Equatorial Ca-O bonds increased from 2.11 Å in MgO to 2.21 Å in the Ca doped MgO. Even further distortion was observed in the axial directions with the Ca-O bond length between 1st and 2nd layer of 2.28 Å and that between 2nd and 3rd layers of 2.24 Å. Although these Ca-O bond lengths are much shorter than those in CaO (lime) at 2.41 Å,54 they introduce partial stress into the doped MgO structure. This can be attributed to the different ionic radii of Mg2+ and Ca2+ atoms (86 and 114 pm, respectively).55 The Ca atom, while doped in MgO, can be considered to have a distorted octahedral coordination with axial Ca-O at equivalent bond lengths. More importantly, upon doping the MgO surface with Ca, the Osurf site coordinated to the Ca atom (O5 in Figure 1c), changes its electronic properties to accommodate the longer Ca-O bond and relaxes outwards. The calculated total Mulliken charge was 9.82 e−, lower than that of other oxygen atoms in the 1st layer (9.89 e−) and the undoped MgO. Finally, the Osurf atom (O5 in Figure 1) coordinated to Ca has larger displacement, ~0.16 Å higher in the z-axis, above the other (001) surface oxygen atoms in the MgO (001) surface relative to the undoped MgO. These results collectively indicate that the reduced electron charge and the "enhanced" sterics will make the Osurf atom react differently with respect to gaseous uptake relative to the Osurf atom in the undoped MgO surface.</p><p>Upon substitution of an Mg for Fe in the 2nd layer and optimization of atomic positions, the octahedral coordination around the dopant atom was altered. The axial Fe-O bond lengths changed very little from 2.11 Å in the undoped MgO to 2.11 and 2.12 Å with respect to Osurf and Obulk in the Fe doped MgO. However, equatorial bonds elongated to 2.15 Å. The total Mulliken charge on Fe1 atom was 24.22 e− effectively showing a Fe2+ electron configuration. The O5 atom was situated slightly higher than the rest of Osurf atoms by 0.04 Å relative to the undoped MgO surface. Due to the effective Fe2+ configuration, the Fe1 atom had 3.74 unpaired electrons with the partial electron spin donated to the surrounding oxygen atoms of ~0.04 e−.</p><p>Total, difference with respect to the superposition of ionic densities, and spin electron charge density maps were plotted (Figure 2) to better understand the observed unpaired electron localization in the Fe doped MgO. Unpaired electrons in Fe doped MgO were mostly localized on Fe1 atoms with some density donated to the octahedrally coordinated oxygen atoms. Importantly, the Osurf atoms in the Fe doped MgO surface possess some spin density, thus potentially making it available to an adsorbing molecule. Difference maps with respect to the superposition of ionic densities (Figure 2, right) show the basicity of the Osurf atoms with an increased electron density extending into the surface due to the absence of counterbalancing attraction by the Mg atom, absent above Osurf. Additional electron polarization between Ca, Fe and equatorial oxygen atoms can be observed with a very distinct change in the region between Osurf and Fe atoms. The latter shows that substitutional doping in the 2nd layer can profoundly affect electronic properties of the Osurf.</p><p>Further electronic structure information can be obtained from the density of state (DOS) plots shown in Figure 3. The UB3LYP calculated bandgap for MgO in this work was 10.3 V, larger than the experimentally-determine value of 7.81.56 This overestimation is due to the incomplete Gaussian basis set used in this work to avoid linear dependencies. Nevertheless, the hybrid functional describes bandgap related properties much better than Hartree-Fock or pure functionals alone and qualitative description is provided here.57 The MgO valence band is primarily comprised from O2p and the conduction band is due to the mixing of empty Mg and O states. With Ca atom doping, a very small change is introduced into the valence band composition and the conduction band is primarily the combination of Ca4s and Ca3d states. It is worth noting that the bottom of the conduction band is now a sharp band from the Ca atom. Thus, the nature of charge acceptance from the adsorbing gas molecule would change.</p><p>A more complicated electronic structure emerges from Fe doping of the MgO (001) surface as inferred from the atomic DOS plot shown in Figure 3c. The nature of the valence band is altered due to the defect states in the bandgap localized on Fe and neighboring O atoms. Band energy at the Fermi level for Fe doped MgO is due to the mixing of O and Fe states with several empty bands situated 4–6 eV above it. Contributions to Fe DOS from 3d-orbitals are shown in Figure 3d with Mulliken AO populations tabulated in Table 1. It can be seen that the state at the Fermi level is due to the beta spin x2-y2 Fe 3d-orbitals. These orbitals point towards the nearest oxygen neighbors thus maximizing electron-nuclei attraction. Thus the Fermi level is due to the electrons in the axial Fe-O bonds. From Table 1 it can also be seen that the x2-y2 subshell is doubly populated. While Mg atom is octahedrally coordinated in bulk undoped MgO, presence of the surface causes symmetry to be decrease and thus the t2g-eg degeneracy is broken. It can be seen from earlier discussion and Figure 1d, the Fe coordination symmetry is now reduced to C4v, in agreement with equal 2.15 Å equatorial bond lengths. In that case, Fe 3d xy, yz and xz (t2g) orbitals split in energy with xz and yz still remaining degenerate as can be seen in Figure 3d.</p><!><p>CO2 adsorbed on undoped and Ca or Fe doped MgO (001) surfaces are shown in Figure 4 and the corresponding bond lengths, total Mulliken population, bond overlap population and the charge and spin density parameters are tabulated and reported in Table 2. Only a representative section of the slab pertaining to the adsorbed molecule is shown for clarity. From Figure 4, we find that CO2 adsorbs via electron donation from the surface basic site, Osurf, to form an adsorbed monodentate-like carbonate structure. This adsorption mechanism involves electron density donation to the CO2 2πu orbital with its components 6a1 molecular orbital (out of plane bending) becoming lower and energy.58 The geometry of the adsorbed monodentate-like is similar in all three cases with a C1-O5 bond length of 1.44, 1.42 and 1.43 Å for undoped, Ca and Fe doped MgO, respectively. The electron density donation from the O5 atom to C1 in Figure 4 can be also seen in Table 2, where the total Mulliken population on the O5 atom is ~9.34 e−. This value is much lower relative to the case with no adsorbed CO2 (~9.86 e−). Hence, ~0.5 e− is being donated to the adsorbing CO2 molecule to form a bond. This electron donation has a profound effect on the local coordination of the O5 atom. In nearly all three adsorbed CO2 cases, the Mg-O5 bond lengths increase from 0.10 to 0.15 Å due to the decrease in charge difference between the atoms involved in the Mg-O5 bond. Population overlap of Mg-O5 is zero still showing purely ionic bonding character. In general, C1s total Mulliken population remains almost the same (~5.20 e−) in CO2 adsorbed on undoped and Ca or Fe doped MgO (001) surfaces. For the Fe doped MgO, the total Mulliken population on Fe (24.21 e−) and the number of unpaired electrons (3.73 e−) remain approximately the same relative to the surface without adsorbed CO2. The elongation of Fe1-O5 bond from 2.11 to 2.28 Å upon CO2 adsorption, as well as change of two equatorial bond lengths from 2.15 to 2.08 Å is an evidence of Jahn-Teller distortion and the profound effect of the adsorbing molecule on the local structure, as well as the electronic properties of the Fe doped MgO. Here x2-y2 Fe 3d-orbital becomes only half-occupied with the other electron now shared between xz and yz.</p><p>Adsorption energies of adsorbed CO2 were calculated to evaluate doping effects on the adsorption strength between the MgO surface slab and the CO2 molecule. Results are reported for CO2, SO2 and NO2 adsorption in Table 3, including results for individual total energies of the surface models with adsorbates. It can be seen that the BSSE corrected adsorption energy, Eadsorption, for CO2 decreases by 5.81 and 3.25 kcal/mol from undoped to Ca and Fe doped MgO with −8.77, −14.58 and −12.02 kcal/mol, respectively, hence showing a lower calculated adsorption energy (stronger adsorption) of CO2 on Ca and Fe doped MgO than in undoped case. As an internal test, Eadsorption was also calculated using larger basis sets on all atoms, including two additional d-exponents on Mg, and the calculated Eadsorption was found to increase in magnitude by ~4 kcal/mol (weaker interaction) but the difference of ~5 kcal/mol between CO2 adsorbed on MgO and that of Ca doped MgO was still maintained. Recent reported calculated adsorption energies of CO2 on MgO (001) terrace ranged from 3.1 kcal/mol using cluster type17 to −3 kcal/mol using GGA plane wave with PW91 functional,59 showing a weak attractive or repulsive coordination due to the lower basicity of highly coordinated terrace site60 and in agreement with values reported in early work.18 Thus, the CO2 binding interaction with the basic Osurf sites in Ca and Fe doped MgO is enhanced relative to the undoped MgO surface. However, the total energy change in the CO2 configuration upon adsorption is small (~0.6 kcal) in all three cases, as well as the difference in C1-O5 bond length (~1.43 Å) between the three models. This result rules out the possibility that the stronger interaction energy between adsorbed CO2 and the undoped, Ca and Fe doped MgO (001) surface is due to adsorbate rearrangement itself. Accordingly, the next probable largest contributor to the adsorption energy could be the rearrangement of the surface slab, particularly in the proximity of the O5 atom. However, this is not the case since the total energy difference between the optimized MgO (001) supercells (shown in Figure 1) and those optimized with CO2 adsorbed (but adsorbate removed in single point calculations) was found to be very close at −25.08, −23.29 and −25.82 kcal/mol for undoped, Ca and Fe doped MgO (001) supercells, respectively, e.g. the difference was much smaller than that between the calculated adsorption energies.</p><p>The last probable contributor to the increase in interaction energy could be due to polarization effects upon doping. To assess this effect, we plotted the total electron density maps and those computed as a difference between the MgO (001) surface with CO2 adsorbed and their respective components, adsorbed CO2 and MgO (001) in the same atomic coordinates. These maps are shown in Figure 5, left and right, respectively. The total energy maps show adsorbed CO2 forming a carbonate ion, CO32−, on the surface. Additionally, a small electron density overlap between CO2 oxygen atoms and Mgsurf can be seen which could further stabilize the adsorbed molecule. Indeed, O-Mgsurf bond lengths of 2.16 to 2.20 Å can be observed as a secondary interaction between CO2 and the MgO (001) surface in Figure 4. This effectively shows polydentate bridging binding configuration, similar to previously reported of CO2 in various organometallic compounds.61 Most interestingly, the electron density difference maps (Figure 5, right) show clear polarization of the electrons towards the adsorbing CO2 molecule, with negative density (less electrons) situated below the O5 atom in the undoped MgO (001) surface. In Ca and Fe doped electron density difference maps, there is additional polarization between the dopant atom and O5, also including axial oxygen atoms surrounding the dopant atom. We believe that this additional electron polarization is responsible for the ~ 5 kcal/mol increase in the CO2 interaction energy with the Ca and Fe doped MgO (001) relative to the undoped MgO case.</p><p>Figure 6 shows the optimized structures of adsorbed SO2 on undoped and Ca or Fe doped MgO (001) surfaces. The corresponding bond lengths and the charge and spin density parameters are reported in Table 2. SO2 adsorption on MgO basic sites proceeds via Osurf (O5) electron density donation to the S1 atom, thus forming an adsorbed sulfite species. This is in agreement with previous experimental data62 and confirmed by the total Mulliken population analysis reported in Table 2 where it can be seen that O5 possesses ~9.30 e− in all three cases. This result is very similar to that of CO2 adsorption. Bond lengths in the adsorbed SO2 molecule are 1.56 Å while those of the O5-S1 bond are ~1.84 Å. Thus, in contrast to CO2 adsorption, adsorbed SO2 does not form a planar complex perpendicular to the MgO surface making the molecular geometry of the adsorbed SO2 trigonal pyramidal. Calculated adsorption energy for SO2 on Osurf atoms was −17.06, −24.60 and −19.91 kcal for undoped, Ca and Fe doped Mg (001) (Table 3, Eadsorption column). In addition, a secondary interaction between Mg3-O6 and Mg4-O7 could be inferred from the short ~2.09 Å bond length between these atoms. This bond length is within the order of Mg-O bond in the crystal (2.11 Å), thus drawing some electron density towards O6 and O7 atoms and weakening O5-Mg3 and O5-Mg4 bonds, increasing their lengths to 2.38 Å. As shown in Table 3, the calculated adsorption energy was larger for Ca and Fe doped MgO relative to undoped MgO, similar in the energy ordering to the case of CO2 adsorption. These values are consistent with the literature reported of −16.8 kcal/mol19 indicating SO2 adsorption stronger than that of CO2 primarily due to the longer S-O bonds in SO2 and resulting secondary acid-base interaction with surface Mg atoms.59</p><p>NO2 coordinated to undoped, Ca and Fe doped MgO (001) surfaces was modeled and the resulting structures are shown in Figure 7. The corresponding bond lengths and charge and spin density parameters are reported in Table 2. NO2 adsorption on MgO (001) Osurf sites should yield surface nitrate species. This interaction was reported to have a low adsorption energy of −10.2 kcal/mol and a large surface-adsorbate separation of 2.62 Å using PW91 functional.24 Our calculated value of −0.19 kcal/mol reported in Table 3 shows no interaction between the NO2 molecule and the surface, consistent with the rich NO2 coordination chemistry reported earlier.59 The magnitude of this interaction can be described as physisorption, well in agreement with experimental observations where adsorbed NO2 was only transformed into nitrate at elevated temperatures.63 This is also substantiated by the total Mulliken population analysis reported in Table 2 where NO2 adsorbed on the undoped and Ca doped MgO surfaces result in a smaller electron density transferred from O5 to the NO2 molecule relative to the case in which CO2 and SO2 are the adsorbates. Secondary interactions of O6 and O7 atoms with the surface Mg3 and Mg4 were observed with the bond length of 2.38 Å in all cases and adsorption energy of −2.67 kcal/mol in Ca doped MgO. The adsorption energy determined for NO2 on the Ca doped MgO surface is higher by ~2.5 kcal/mol than for NO2 adsorbed on the undoped MgO surface.</p><p>NO2 adsorption on the Fe doped MgO (001) surface was found to be dramatically different from adsorption on the undoped and Ca doped MgO (001) surfaces. The adsorption interaction for NO2 on the Fe doped MgO surface was also very different from interactions theoretically observed for CO2 and SO2 on the same surface. As shown in Figure 7, upon NO2 adsorption on Fe doped MgO, a strong inward relaxation of the O5 atom was observed with the Fe1-O5 bond length at 1.89 Å. This bond length is much smaller than in the case without the NO2 molecule interacting with the surface (2.18 Å). All equatorial Fe-O bonds also became shorter and very similar in length (2.07 to 2.10 Å). The total Mulliken charge of the O5 atom became 9.70 e− upon accounting for NO2/surface interactions. This value is larger than that for CO2 and SO2 adsorption on the same surface and much closer to that of the Fe doped MgO (001) surface (9.86 e−) with no molecules adsorbed. This result possibly indicates that electron density is not transferred from the O5 to N1 to facilitate a bonding interaction or that the O5 acted as an electron density transfer atom from Fe1 to N1. The most prominent change happens to the Fe1 atom and manifests itself as the increase in total spin from 3.72 e− to 4.29 e− and the total Mulliken population decreasing from 24.21 e− to 23.77 e− (Table 2). This, concurrently with the increase in total Mulliken charge on the N1 atom from 6.54 e− to 6.85 e−, shows charge redistribution (transfer) from the Fe1 atom located in the 2nd layer of the Fe doped MgO (001) surface to the NO2 molecule. There is no spin density left on the N1 atom with the total Mulliken population (6.85 e−) closer to that of a neutral nitrogen atom (7 e−). Furthermore, the total number of electrons on the NO2 molecule is 23.99 e− whereas for NO2 adsorbed on the undoped MgO (001) surface there is a total of 23.35 e− localized on NO2. This effectively shows that NO2 has obtained some electron density from Fe1, thus effectively becoming NO2−, as confirmed by no unpaired electrons on N1. Most importantly, charge transfer results in an increased adsorption energy of −37.64 kcal/mol in Fe doped MgO (001) when compared to −2.67 kcal/mol in Ca doped MgO (001). It cannot be excluded that, in the NO2 adsorption case (especially Fe doped MgO (001)) due to the inherently weak adsorbate-surface interactions, there are other adsorbate configurations with respect to the O5 atom that might be located during optimization, depending on the initial geometry guess. One other NO2 optimization on Fe doped MgO (001) resulted in an Mg-O coordinated structure that was 11.58 kcal/mol lower in energy (more stable) than the bonded N1-O5. This result is not surprising as Mg-O bonds are stronger than those of N-O. All reported structures, however, resulted in the absence of unpaired electron density on N1 and ~4.26 e− localized on Fe1, thus showing electron transfer from the NO2 molecule to the surface regardless of the adsorbate configuration. Collectively, our results show a profound effect on both electronic and structural properties of MgO (001) surface during NO2 adsorption on Osurf sites.</p><!><p>The structural and electronic properties of undoped, Ca and Fe doped MgO (001) surfaces, and the differences in their propensity towards the acidic atmospheric gas (CO2, SO2 and NO2) adsorption were investigated with emphasis on gas adsorption on basic oxygen surface sites. Monodentate-like coordination was observed for all adsorbates with SO2 and NO2 exhibiting secondary interactions with surface Mg atoms. Thus, it can be inferred that trace atmospheric gas adsorption onto Osurf basic sites is a preliminary step for further adsorbate transformation, such as more stable Mg-O bond formation. Importantly, adsorption energy calculations showed that MgO doping will provide stronger interactions between the adsorbate and the surface than in the undoped MgO case for a given adsorbate molecule. Since mineral dust in the environment is rarely pure and will have structural defects and impurities, this can affect trace atmospheric gas partitioning rates and stability and implies potentially enhanced uptake relative to previous laboratory studies on pure minerals as model mineral aerosol components. Additionally, our results show that charge transfer from the iron atom in the Fe doped MgO (001) surface to NO2 results in the formation of adsorbed NO2− and an increase the binding attraction between adsorbate and the surface by an order of magnitude relative to the undoped and Ca doped MgO (001) surface. This result suggests that the nature of NO2 adsorption on MgO surfaces can be significantly altered depending on the identity of the dopant atom. In this case, adsorption changes from physisorption to chemisorption via charge transfer, contrary to the currently accepted weak NO2 adsorption on dust surfaces. This is in agreement with previous XAFS data where Fe mixing into MgO was found to alter the nature of CCl4, atmospheric pollutant molecule, adsorption.64 Thus, our results indicate that trace metal doping in atmospheric mineral aerosol as it occurs naturally in the environment needs to be accounted for in laboratory and theoretical studies as well as global atmospheric chemistry models in order to better understand the heterogeneous chemistry of naturally-occurring mineral dust aerosol.</p>

PubMed Author Manuscript

N-Benzoimidazole/Oxadiazole Hybrid Universal Electron Acceptors for Highly Efficient Exciplex-Type Thermally Activated Delayed Fluorescence OLEDs

Recently, donor/acceptor type exciplex have attracted considerable interests due to the low driving voltages and small singlet-triplet bandgaps for efficient reverse intersystem crossing to achieve 100% excitons for high efficiency thermally activated delayed fluorescence (TADF) OLEDs. Herein, two N-linked benzoimidazole/oxadiazole hybrid electron acceptors were designed and synthesized through simple catalyst-free C-N coupling reaction. 24iPBIOXD and iTPBIOXD exhibited deep-blue emission with peak at 421 and 459 nm in solution, 397 and 419 nm at film state, respectively. The HOMO/LUMO energy levels were −6.14/−2.80 for 24iPBIOXD and −6.17/−2.95 eV for iTPBIOXD. Both compounds could form exciplex with conventional electron donors such as TAPC, TCTA, and mCP. It is found that the electroluminescent performance for exciplex-type OLEDs as well as the delayed lifetime was dependent with the driving force of both HOMO and LUMO energy offsets on exciplex formation. The delayed lifetime from 579 to 2,045 ns was achieved at driving forces close to or larger than 1 eV. Two TAPC based devices possessing large HOMO/LUMO offsets of 1.09–1.34 eV exhibited the best EL performance, with maximum external quantum efficiency (EQE) of 9.3% for 24iPBIOXD and 7.0% for iTPBIOXD acceptor. The TCTA containing exciplex demonstrated moderate energy offsets (0.88–1.03 eV) and EL efficiency (~4%), while mCP systems showed the poorest EL performance (EQE <1%) and shortest delayed lifetime of <100 ns due to inadequate driving force of 0.47–0.75 eV for efficient exciplex formation.

n-benzoimidazole/oxadiazole_hybrid_universal_electron_acceptors_for_highly_efficient_exciplex-type_t

Introduction<!>Synthesis and Characterization<!><!>Photophysical Properties<!><!>Theoretical Calculations and Electrochemical Properties<!><!>Theoretical Calculations and Electrochemical Properties<!><!>Electroluminescence Properties<!><!>Electroluminescence Properties<!><!>Conclusion<!>Author Contributions<!>Conflict of Interest Statement

<p>Organic light-emitting diodes (OLEDs) have been developed rapidly in recent years since the pioneer work on low-voltage fluorescence electroluminescence by Tang in 1987 (Tang and VanSlyke, 1987; Ma et al., 1998; Gong et al., 2010; Park et al., 2013; Zhang et al., 2016). According to spin statistics, the ratio for singlet and triplet excitons recombined from electrogenerated holes and electrons is 1:3 (Baldo et al., 1999; Segal et al., 2003). Thus, the first generation of traditional fluorescent OLEDs which solely harvest singlet excitons only shows 25% of maximum internal quantum efficiency (IQE) (Wen et al., 2005). On the other hand, the second generation of phosphorescent OLEDs (PHOLEDs) based on heavy metal complexes and third generation of thermally activated delayed fluorescence (TADF) OLEDs could both reach 100% IQE in theory by utilizing all singlet and triplet excitons through intersystem crossing (ISC) and reverse inter-system crossing (RISC), respectively (Baldo et al., 1998; Adachi et al., 2001; Su et al., 2008; Lo et al., 2009; Goushi et al., 2012; Uoyama et al., 2012; Zhang and Forrest, 2012; Li et al., 2016; Cao et al., 2017; Huang et al., 2018; Wu Q. et al., 2018). However, to avoid consuming noble metals and achieving reliable true-blue light, TADF OLEDs based on low-cost pure organic emitters have attracted increasing interests as an alternative mechanism to PHOLEDs. TADF emission is realized by an up-conversion process from lower energy triplet states to slightly higher energy singlet states by endothermic reverse inter-system crossing process (Li et al., 2016; Cao et al., 2017; Huang et al., 2018; Wu Q. et al., 2018). Therefore, a small singlet-triplet energy bandgap (ΔEST) is required for TADF emitters.</p><p>It is reported that the small ΔEST could be attained in (i) intramolecular charge transfer featured single molecule with twisted donor-acceptor structured for effective spatial isolation between the highest occupied molecular orbital (HOMO) and lowest unoccupied molecular orbital (LUMO) on the relevant hole and electron transporting moieties, and (ii) bimolecular-exciplex which contains an electron-donor material mixed with an electron-acceptor material through intermolecular charge transfer characteristics (Cai and Su, 2018; Liu et al., 2018; Sarma and Wong, 2018). High external quantum efficiency (EQE) of 20% for red, 29% for orange, 38% for green, and 37% for light-blue TADF OLEDs have been achieved in single-molecule TADF emitters (Lin et al., 2016; Chen et al., 2018; Wu T.-L. et al., 2018; Zeng et al., 2018). However, the development of bimolecular TADF lags far behind. Most exciplex-type TADF OLEDs showed maximum EQE close to 10% (Jankus et al., 2014; Liu et al., 2015a,b; Oh et al., 2015; Zhang L. et al., 2015; Hung et al., 2016, 2017; Jeon et al., 2016), with only one example approaching to 18% (Liu et al., 2016).</p><p>In electron donor/acceptor formed exciplex systems, compared with commercially available various electron-donor materials, such as 4,4′,4″-tris[3-methylphenyl(phenyl)amino]-triphenylamine (m-MTDATA), N,N′-bis(1-naphthyl)-N,N′-diphenyl-[1,1′-biphenyl]-4,4′-diamine (NPB), 4,4′-(cyclohexane-1,1-diyl)bis(N-phenyl-N-p-tolylaniline) (TAPC), 4,4′,4″-tris(N-carbazolyl) triphenylamine (TCTA), 4,4′-bis(N-carbazolyl)-1,1′-biphenyl (CBP), and N,N′-dicarbazolyl-3,5-benzene (mCP) etc., the types of efficient and low-cost electron-acceptor materials are scarce (Goushi and Adachi, 2012; Goushi et al., 2012; Sun et al., 2014; Lee et al., 2015; Liu et al., 2015b). Thus, the exploration of electron accepting materials is essential for constructing exciplex systems with outstanding optoelectronic performance. Therefore, in this work, we designed and synthesized two new electron-acceptors of 2-(2,4-bis(2-phenyl-1H-benzo[d]imidazol-1-yl)phenyl)-5-phenyl-1,3,4-oxadiazole (24iPBIOXD), and 2-phenyl-5-(2,4,6-tris(2-phenyl-1H-benzo[d]imidazol-1-yl)phenyl)-1,3,4-oxadiazole (iTPBIOXD) through a simple one-step catalyst-free aromatic nucleophilic substitution reaction. The electron-withdrawing oxadiazole (OXD) unit has been extensively applied in donor-acceptor type bipolar transport host materials, single molecule intramolecular charge transfer type TADF emitters as well as electron transport materials (Tao et al., 2011; Mondal et al., 2013; Olivier et al., 2017; Cooper et al., 2018; Yao et al., 2018; Zhang et al., 2018). By combining OXD building block with our previously reported isomeric N-linkaged benzoimidazole (Hu et al., 2017), both 24iPBIOXD, and iTPBIOXD exhibited deep HOMO level of ~-6.15 eV, facilitating the exciplex formation with general electron donor materials of TAPC, TCTA, and mCP due to the compatible HOMO and LUMO energy levels between donor and acceptor materials. The gradient energy offsets ranging from 0.47 to 1.34 eV correlated well with the delayed lifetime and EL efficiencies in exciplex type TADF OLEDs. The TAPC:24iPBIOXD exciplex with the largest HOMO/LUMO offsets exhibited the best EL performance, with maximum EQE of 9.3% for green TADF OLEDs.</p><!><p>Scheme 1 shows the synthetic routes and molecular structures of 24iPBIOXD and iTPBIOXD. The two compounds could be facilely synthesized by a simple one-step catalyst free C-N coupling reaction. This nucleophilic substitution reaction was carried out in DMSO solvent with K2CO3 base at high yields over 80% by using di/tri-fluorine substituted oxadiazole derivatives as electrophiles and 2-phenyl-1H-benzo[d]imidazole as nucleophiles. The considerably high yields and environmentally eco-friendly conditions demonstrated the superiority than common metal-catalyzed Ullman reactions (Son et al., 2008; Liu et al., 2011; Volz et al., 2013). In addition, the directly connection of the isomeric N-linked benzoimidazole to the central phenyl ring avoided the complicated multistep ring-closing synthetic process for the normal C-linked benzoimidazole in traditional electron transport material of 2,2,2-(1,3,5-phenylene)-tris(1-phenyl-1H-benzoimidazole) (TPBI) or its derivatives. The chemical structures of the new compounds were fully characterized by 1H NMR, 13C NMR, mass spectrometry (MALDI-TOF) and element analysis (Figure S1). The good thermal stability of the two compounds was confirmed by thermogravimetric analysis (TGA) and differential scanning calorimetry (DSC) (Figure 1). The decomposition temperatures (Td, corresponding to a 5% weight loss) from TGA curves for 24iPBIOXD and iTPBIOXD were determined at 443 and 461°C, respectively. Additionally, the melting point (Tm) of iTPBIOXD was observed at 327°C, which was much higher than 286°C of 24iPBIOXD. The glass transition temperature (Tg) of both materials can be detected from the second heating cycles from DSC, with values of 126°C for 24iPBIOXD and 165°C for iTPBIOXD, indicating their reasonable thermal stability.</p><!><p>Synthesis of compounds 24iTPBIOXD and iTPBIOXD.</p><p>(A) Thermogravimetric analysis (TGA) and (B) differential scanning calorimetry (DSC) (solid symbols represent for first heating scan and open symbols for second heating scan) curves for 24iPBIOXD and iTPBIOXD.</p><!><p>The room temperature UV-Vis absorption and photoluminescence (PL) spectra of 24iPBIOXD and iTPBIOXD in CH2Cl2 solution are shown in Figure 2A. Both compounds exhibited an intense absorption with peaks at 289 and 283 nm in solution, 297 and 288 nm in film, respectively, which can be ascribed to the π-π* transition of molecules. The optical bandgap (Eg) was calculated to be 3.34 eV for 24iPBIOXD and 3.22 eV for iTPBIOXD, according to the film-state absorption edge. On the other hand, 24iPBIOXD and iTPBIOXD showed unimodal photoluminescence peaking at 421 and 459 nm in solution, whereas significantly blue-shift to 397 and 419 nm in film state (Table 1). By analyzing the highest-energy vibronic sub-band of low-temperature fluorescence and phosphorescence spectrum (Figure 2B), the singlet (ES) and triplet (ET) energy levels could be determined to be 3.31/2.55 and 3.18/2.53 eV for 24iPBIOXD and iTPBIOXD, respectively. In addition, the ES/ET energy levels of three hole-transport electron donor materials were also calculated to be 3.54/2.95 eV for mCP, 3.79/2.82 eV for TAPC, and 3.66/2.84 eV for TCTA (Figure 2C). The PL spectra for the neat film of electron donors such as mCP, TAPC, and TCTA, the two new electron-acceptors of 24iPBIOXD and iTPBIOXD as well as their corresponding mixtures in a 1:1 weight ratio were investigated. As shown in Figure 3 and Figure S2, all blended films showed bathochromic shifted PL spectra compared with the emission of neat 24iPBIOXD/iTPBIOXD and the corresponding donor-material, indicating the successful formation of exciplex (Zhang T. et al., 2015). In addition, it is found that exciplex based on 24iPBIOXD acceptors all exhibited about 20–30 nm blue-shifted emission than iTPBIOXD based exciplex systems. The exciplex emission color could be tuned from deep-blue of mCP:24iPBIOXD with peak at 419 nm to light-blue of TCTA:24iPBIOXD (501 nm) and further to green of TAPC:24iPBIOXD (518 nm). Besides, transient photoluminescence (PL) measurements were carried out for all six exciplexes (Figure 4). The exciplexes comprising TAPC or TCTA donor all possessed significantly longer delayed decay lifetime, with values of 579 ns for TCTA:24iPBIOXD, 1,907 ns for TCTA:iTPBIOXD, 1,520 and 2,045 ns for TAPC:24iPBIOXD, TAPC:iTPBIOXD exciplex, respectively. However, the mCP:24iPBIOXD and mCP:iTPBIOXD exciplex systems displayed greatly shorter delayed decay lifetime of only 42 and 72 ns (Table 1). Besides, the temperature dependent PL transients for the representative TAPC:24iPBIOXD and TCTA:iTPBIOXD exciplexes (Figure S3) both demonstrated a more significant decay from 100 to 300 K at the longer lifetime range, suggesting the potential existence of endothermic reverse inter-system crossing. It is expected the obvious variations on delayed decay time for different exciplexes may demonstrate some relationships with the device efficiency in exciplex-TADF OLEDs.</p><!><p>(A) Normalized UV-Vis absorption and PL spectra for 24iTPBIOXD and iTPBIOXD in CH2Cl2 solution, (B) 77 K fluorescence (Fl) and phosphorescence (Ph) spectra for 24iTPBIOXD and iTPBIOXD in neat film, (C) 77 K fluorescence (Fl) and phosphorescence (Ph) spectra for three donor materials in neat film.</p><p>Physical properties of compounds 24iPBIOXD and iTPBIOXD.</p><p>Measured in CH2Cl2 solution at room temperature.</p><p>Measured in film.</p><p>Singlet energy and triplet energy was calculated from low temperature (77 K) fluorescence spectra and phosphorescence spectrum.</p><p>Optical bandgap (Eg) calculated from the absorption edge of film state UV-Vis spectra.</p><p>LUMO measured from the onset of reduction curves from CV and HOMO calculated from the difference between LUMO and Eg, values in parentheses from DFT calculations.</p><p>Glass transition temperature/melting point/decomposition temperature.</p><p>Emission maxima, charge transfer state energy, delayed decay lifetime and HOMO/LUMO energy offsets for various exciplexes.</p><p>PL spectra of (A) mCP, (B) TAPC, (C) TCTA with different electron acceptors and the relevant exciplexes in film.</p><p>Transient decay curves of (A) 24iPBIOXD and (B) iTPBIOXD-based exciplexes at film state.</p><!><p>In order to gain insights into the frontier molecular orbital and excited states level distribution of 24iPBIOXD and iTPBIOXD, density functional theory (DFT) calculation was conducted at the B3LYP level (Francl et al., 1982; Becke, 1988; Lee et al., 1988). From the optimized geometry shown in Figure 5, the dihedral angles between the central phenyl and oxadiazole ring were 22.0 and 50.3° for 24iPBIOXD and iTPBIOXD, respectively, the values between the benzoimidazoles and the central phenyl rings ranged from 50.4 to 77.4°, indicating a twisted structure for both compounds. Furthermore, in the ground state, the highest occupied molecular orbital (HOMO) were almost completely located on one of the ortho-positioned phenylbenzoimidazole units, indicating the electron-donating characteristics of N-linked phenylbenzoimidazole, which was quite different from the C-isomerized phenylbenzoimidazole containing TPBI (Hu et al., 2017). And the lowest unoccupied molecular orbital (LUMO) were mainly localized on 2,5-diphenyl-1,3,4-oxadiazole, along with mildly distribution over the penta-heterocyclic imidazoles, suggesting the weak electron-withdrawing property to gently participate electron-transport for the imidazoles. Similar distribution can be observed in the highest occupied natural transition orbital (HONTO) and the lowest unoccupied natural transition orbital (LUNTO) at singlet excited state. It should be noted that the HONTO distribution at triplet excited state was completely different from S0 and S1 for both compounds, which was mainly delocalized through the 2,5-diphenyl-1,3,4-oxadiazole skeleton, similar with the LUNTO distribution.</p><!><p>Optimized 3D structures, spatial distributions of ground state HOMO and LUMO, singlet (S1), and triplet excited state (T1) HONTO and LUNTO for compounds 24iPBIOXD and iTPBIOXD.</p><!><p>The electrochemical features were measured by cyclic voltammetry (CV) (Figure 6). Both compounds exhibited reversible reduction whereas undetectable oxidation behavior. The LUMO energy levels calculated from the onset of reduction curves for 24iPBIOXD and iTPBIOXD were measured to be −2.80 and −2.95 eV, while the HOMO energy levels calculated from the different between the LUMO and optical bandgaps (Eg) were evaluated to be −6.14 and −6.17 eV, respectively. The values were in good agreement with the theoretical calculation. Besides, the energy levels for electron donor materials were also measured, with HOMO estimated from the onset of electro-oxidation curves and LUMO calculated from HOMO and optical bandgaps. The HOMO/LUMO energy level values for mCP, TAPC, and TCTA were −5.67/−2.20, −5.05/−1.61, and −5.19/−1.92 eV, respectively. The deep HOMO and LUMO for the two new electron acceptors of 24iPBIOXD and iTPBIOXD, provided sufficient driving forces on HOMO/LUMO energy offsets for the exciplex formation (Figure 6A). As shown in Figure 7A, the HOMO energy level offsets between the electron donor of TAPC, TCTA, or mCP and the electron acceptors of 24iPBIOXD/iTPBIOXD were calculated to be 1.09/1.12, 0.95/0.98, or 0.47/0.5 eV, and the corresponding LUMO offsets were 1.19/1.34, 0.88/1.03, or 0.6/0.75 eV, respectively. It is noted in both acceptor systems, the TAPC donor based exciplex presented the highest driving force, followed by TCTA, while the mCP donor demonstrated the lowest HOMO/LUMO offsets.</p><!><p>Cyclic voltammograms of (A) 24iPBIOXD and iTPBIOXD in THF solution for reduction scan; (B) conventional electron donors (mCP, TAPC, and TCTA) in CH2Cl2 solution for oxidation scan.</p><p>(A) HOMO, LUMO energy level diagrams and (B) singlet and triplet excited-state energies for 24iPBIOXD, iTPBIOXD, and various donor materials, as well as the charge transfer state energy (ECT) for the corresponding exciplex systems, red for 24iPBIOXD, and blue for iTPBIOXD.</p><!><p>To investigate the charge transport properties of the two new N-linked isomeric benzoimidazole containing electron acceptors, single carrier electron-only device was prepared to find out the electron inject and transport properties of 24iPBIOXD and iTPBIOXD. The device structure was ITO/24iPBIOXD, iTPBIOXD, or TPBI (50 nm)/LiF (1 nm)/Al (150 nm), where the commercial electron transport material of 2,2,2-(1,3,5-phenylene)-tris(1-phenyl-1H-benzoimidazole) (TPBI) with C-linkage in benzoimidazole was selected for comparison. As shown in Figure 8, at the same operating voltage, TPBI based device exhibited the highest current density among all the three devices. Since the LUMO energy of TPBI (2.7–2.9 eV) (Bian et al., 2018; Jou et al., 2018) was almost the same as 24iPBIOXD and iTPBIOXD, which manifested their similar injection barrier for efficient electron injection. Therefore, the significantly higher current for TPBI indicated better electron transporting property than 24iPBIOXD and iTPBIOXD. On the other hand, the current density in iTPBIOXD device was slightly higher than 24iPBIOXD, as depicted in Figure 8, the LUMO level of iTPBIOXD was 0.15 eV lower than 24iPBIOXD, therefore a mildly efficient electron-injection could be attained in iTPBIOXD device due to its lower injection barriers. Thus, the electron-transport performance for both electron acceptors may be comparable.</p><!><p>J-V characteristic of nominal single-electron-only devices based on compounds 24iPBIOXD, iTPBIOXD, and TPBI [device structures: ITO/EML (50 nm)/LiF (1 nm)/Al (150 nm)].</p><!><p>To conduct a comprehensive comparison on the EL performance for the exciplex-TADF OLEDs among diverse electron-donor and acceptor systems, a series of vacuum deposited devices A-F were fabricated. Due to the highest HOMO level of TAPC for efficient hole-injection, the device configuration for TAPC-based OLEDs was ITO/MoO3 (1 nm)/TAPC:24iPBIOXD or iTPBIOXD (1:1, 70 nm)/LiF (1 nm)/Al (100 nm). To reduce the hole-injection barrier, TCTA-based device was constructed by ITO/MoO3 (1 nm)/TAPC (40 nm)/TCTA:24iPBIOXD or iTPBIOXD (1:1, 30 nm)/TmPyPB (40 nm)/LiF (1 nm)/Al (100 nm), while a further 10 nm TCTA thin film was inserted between the TAPC layer and emissive layer (EML) in mCP-based devices. Among them, MoO3 and LiF were used as hole- and electron-injection materials, respectively; TAPC and 1,3,5-tri[(3-pyridyl)-phen-3-yl] benzene (TmPyPB) were functionalized as hole- and electron-transport materials, respectively, an extra TCTA layer was aimed to promote the hole-injection and block the electrons.</p><p>The current density-voltage-luminance (J-V-L), electroluminescence (EL) spectra, together with the current and power efficiency, external quantum efficiency vs. luminance curves are shown in Figure 9. The device fabrication details are stated in Supporting Information. According to the key EL data listed in Table 2, the turn-on voltage for TAPC, TCTA, and mCP containing devices was gradiently increased from 2.8, 3.0 to 3.8 eV. The as high as 0.48 eV hole-injection barrier between TCTA and mCP lead to the highest operating voltage in devices E and F. The EL performance trend was in consistent with the values of HOMO/LUMO energy offsets, and TAPC-analog bearing the highest driving forces for convenient exciplex formation demonstrated the best highest EL efficiency. The best performance was attained from device A with TAPC:24iPBIOXD exciplex, corresponding to a maximum current efficiency (CE), power efficiency (PE), and external quantum efficiency (EQE) of 28.8 cd/A, 32.3 lm/W, and 9.3%. And the TAPC:iTPBIOXD based device B demonstrated slightly poorer EL performance with maximum current efficiency, power efficiency and external quantum efficiency of 22.1 cd/A, 23.4 l m/W, and 7.0%, respectively. Device C and D based on TCTA electron donor showed comparable EL efficiency, with maximum EQE of 4.0 and 3.9% for 24iPBIOXD and iTPBIOXD electron acceptors, respectively. The EL performance for mCP device was rather poor, with maximum EQE of <1% in both device E and F. As depicted in Figure 9B, devices A and B with TAPC donor depicted smooth exciplex-TADF emission, with EL peak at 519 and 556 nm, respectively, which is in agreement with the relevant PL spectra. Commission Internationale de L'Eclairage (CIE) values for device A and B was measured at (0.31, 0.55) and (0.43, 0.54), corresponding to green and yellow emission, respectively. The TCTA based device C and D both exhibited blueish-green emission, with a gentle shoulder peak at around 400 nm for 24iPBIOXD. The two mCP-based devices displayed blue emission with CIE x, y each at ~0.20. However, the EL spectra of device E and F revealed bimodal emission showing comparable intensity for the two peaks. It is hypothesized that the inadequate HOMO and LUMO energy offsets (< ~1 eV) for TCTA:24iPBIOXD, mCP:24iPBIOXD, and mCP:iTPBIOXD, resulted in the unexpected shorter wavelength EL emission peak, which was ascribed to pure mCP emission (Chiu and Lee, 2012; Shahalizad et al., 2017). In addition, since the ECT of TAPC and TCTA based exciplexes were lower than the triplet energy of both donor and acceptor materials, which was beneficial to restrict triplet excitons in the exciplex states for the efficient RISC. However, ECT of 2.8–2.96 eV (Figure 7B) for mCP based exciplex was significantly higher than the triplet energy (~2.55 eV) of the two electron acceptors, which provided a potential way for energy leakage from exciplex states to the T1 excited state of 24iPBIOXD and iTPBIOXD. Thus, devices based on mCP donors demonstrated the lowest EL efficiency and inadequate TADF emission.</p><!><p>(A) L-V-J characteristics; (B) normalized electroluminescent (EL) spectra; (C) external quantum efficiency (EQE) vs. luminance curves; (D) current efficiency and power efficiency vs. luminance curves of device A–F.</p><p>Electroluminescence characteristics for the devices.</p><p>Turn-on voltage at 1 cd/m2.</p><p>Maximum current efficiency.</p><p>Maximum power efficiency.</p><p>Maximum external quantum efficiency.</p><!><p>In summary, we have designed and synthesized two universal N-linked benzoimidazole/oxadiazole hybrid electron acceptors through a simple nucleophilic substitution reaction. Diverse deep-blue to yellow emissive exciplex could be formed between various conventional donor materials and the two acceptors due to their deep HOMO levels of ~6.15 eV. The HOMO and LUMO energy level offsets which were also named as the driving forces for exciplex formation were gradiently increased from 0.47 to 1.12 and 0.6 to 1.34 eV in mCP, TCTA, to TAPC based exciplexes. We have found that both HOMO and LUMO offsets ≥1 eV was required to form efficient and stable intermolecular charge transfer exciplex. When the driving forces were as low as 0.47–0.75 eV, which is far <1 eV, the two mCP based exciplex demonstrated considerably short delayed component lifetime, with values of only 42 ad 72 ns for 24iPBIOXD and iTPBIOXD acceptors, respectively. Additionally, the exciplex-type device EQE was lower than 1%. When the driving forces were slightly lower or approaching 1 eV, the two TCTA exciplexes displayed moderate EL efficiency of about 4%. And the best EL performance was achieved in TAPC containing exciplex-type TADF OLEDs, with relatively low turn-on voltage of 2.8 V, maximum efficiency of 28.8 cd/A CE, 32.3 lm/W PE, and 9.3% for 24iPBIOXD acceptor and 22.1 cd/A CE, 23.4 lm/W and 7.0% for iTPBIOXD acceptor. Our results provide guidance on the exploration of efficient exciplex type TADF OLEDs.</p><!><p>WY, MZ, and DH designed and synthesized the materials. WY and MZ did most of the experimental work and data analyses. OLED device fabrication and electroluminescent performance studies were carried out by HY and NS. YT had the idea, led the project. WY and YT prepared the manuscript. All authors contributed to the manuscript preparation.</p><!><p>The authors declare that the research was conducted in the absence of any commercial or financial relationships that could be construed as a potential conflict of interest.</p>

PubMed Open Access

Modular Continuous Flow Synthesis of Orthogonally Protected 6-Deoxy Glucose Glycals

An efficient, modular continuous flow process towards accessing two orthogonally protected glycals is described with the development of reaction conditions for several common protecting group additions in flow, including the addition of benzyl, naphthylmethyl and tert-butyldimethylsilyl ethers. The process affords the desired target compounds in 57\xe2\x80\x9374% overall yield in just 21\xe2\x80\x9337 minutes of flow time. Furthermore, unlike batch conditions, the flow processes avoided the need for active cooling to prevent unwanted exotherms and required shorter reaction times.

modular_continuous_flow_synthesis_of_orthogonally_protected_6-deoxy_glucose_glycals

<!>Conclusions

<p>Recently, continuous flow processes have been adapted by many groups in the synthesis of different natural products1 and active pharmaceutical ingredients (APIs).2,3 Flow chemistry can provide many benefits,4–6 including the ability to run continuous processes enabling large scale production7 of material and increased reaction efficiencies8 as compared to batch processes. Reactions can be more efficiently heated and cooled and highly reactive intermediates can be made transiently to avoid the safety hazards of larger scale production.</p><p>The design of efficient flow reactions requires consideration of several parameters. Although microfluidic systems require careful consideration of parameters to get reproducible temperature transfers and mixing efficiencies,9 larger tubing-based systems flow systems can be easier to implement. The benefits of microfluidic and larger flow systems have seen applications to chemical glycosylation,10–15 but almost none of the necessary glycosyl donors and acceptors needed for glycosylation reactions have been produced with the development of continuous flow processes.16</p><p>Glycals—1,2-unsaturated monosaccharide derivatives —have a wide variety of uses in organic synthesis. From natural product synthesis to the generation of novel structural motifs, glycals are versatile building blocks.7 In carbohydrate chemistry, glycals can be directly used as donors in chemical glycosylation18 by a variety of different activation conditions.19–26 Glycals also serve as important intermediates in the construction of glycosyl donors, as they can readily be made into hemiacetals27,28 or thiogylcosides.29–33</p><p>A major hurdle in carbohydrate chemistry remains the ability to quickly and efficiently access multiple donor and acceptor building blocks.34 Our solution to this problem is to take advantage of the faster timescales of continuous flow processes. In continued studies on deoxy-sugar oligosaccharide synthesis, we had need for large quantities of glycal precursors. In an effort to circumvent issues associated with batch synthesis (time consuming reaction sequences, limited scale, potential exotherms, etc.), we chose to examine if these substrates could be produced in a continuous flow system. As an initial foray into this chemistry, we chose the synthesis of orthogonally protected L-rhamnals 1 and 2, which serve as precursors for many of the deoxy-sugars found in natural products, such as those in the anthracycline35 and angucycline36–38 families of antibiotics (Figure 1).</p><p>The synthesis of glycals 1 and 2 commenced with translating a Zemplén deacetylation that we had previously run in batch reaction conditions28 into a flow process. As is the case with converting manual to automated batch process,29 the conversion of batch to flow processes is not trivial. Ideally, solvents are found in which all reagents and reactants are soluble and no products or byproducts precipitate during the course of the reaction. We started with 0.4 equivalents (0.24 M in MeOH) of sodium methoxide (Table 1, entry 1). While the desired diol 4 was formed in 76% yield, the reaction did not go to completion. By an increase in the amount of sodium methoxide (NaOMe) to 0.8 equivalents (0.48 M in MeOH), compound 3 was completely consumed and the formation of the desired product 4 was affected in 98% yield (Table 1, entry 2). Importantly, we were able to run the deacetylation on a five gram scale of 3, at a rate that could produce 35.8 g/h of diol 4.</p><p> </p><p>With 4 in hand, we next sought to convert the traditional Corey silylation conditions39 to a flow process by regioselectively protecting the allylic alcohol as a tert-butyldimethylsilyl (TBS) ether. Following optimization, we found that using 1.1 equivalents of TBSCl (0.66 M in DMF) and 1.5 equivalents of imidazole (0.89 M in DMF) was sufficient to selectively protect the C3 alcohol over the C4 position in 40 minutes of retention time to afford 5 in 91% yield (Table 2, entry 2). Efforts to reduce the retention time by increasing flow rate resulted in formation of product in 49% yield with recovery of unreacted starting material (Table 2, entry 3). In order to increase the rate of the reaction, we looked into using a different auxiliary base such 4-dimethylaminepyridine (DMAP) which is known to have improved rates of silylation relative to imidazole.40 Using 1.5 equivalents (0.89 M in DMF) of DMAP, we were able to get complete conversion of diol 4 to corresponding silyl ether 5 in 88% yield in 10 minutes of total retention time (Table 2, entry 5). After increasing the scale of the reaction to 2.97 g of diol 4, we were able to achieve regioselective protection of the allylic alcohol at a rate of 29.4 g/h. These conditions are an improvement to traditional TBS silylation conditions that typically use a large excess of base to increase the rate of reaction. Interestingly, we also found that the reaction could be run at ambient temperature and it was not necessary to cool the reaction down to 0 °C in order to selectively protect the C3 position.</p><p> </p><p>With the allylic alcohol successfully protected, we next chose to protect the C4 alcohol with an orthogonal protecting group such as an alkyl ether. Recently, the Pohl lab has demonstrated continuous flow benzylation, acetylation, and thioglycoside formation in the synthesis of protected levoglucosan and glucose.16 While acetylation and thioglycoside formation proceeded smoothly, the use of BaO packed bed reactors had issues with pressure buildup and possibly product absorption to the solid phase that warranted a reenvisioning of this reaction.41 In our case, only a single hydroxyl group was free for protection, which expanded the choice of solvents and conditions that could be considered for the benzylation reaction.</p><p>To address this, we chose to examine the use of alternative bases in the reaction. Typical Williamson ether conditions include sodium hydride, which is incompatible with flow-based setups due to it being a heterogeneous solution. However, benzylation has successfully been demonstrated under a variety of homogeneous conditions, such as LiHMDS/BnBr/TBAI42 or benzyl trichloroacetimidate/TfOH.43 To this end, we pictured glycal 5 first going through a deprotonation sequence with KHMDS, followed by treatment with BnBr and TBAI to achieve benzylation (Table 3, entry 1). Through optimizations, we found that TBAI was unnecessary in the reaction, perhaps due to the high reaction concentration leading to fast rates of reaction (Table 3, entry 2). Further optimization of the reaction led to finding that 5.0 equivalents of BnBr (1.8 M in DMF) was optimal for benzylation, affording 1 in 86% yield without any observed migration of the silyl ether (Table 3, entry 5). With a total retention time of only 6 minutes, this is a large improvement to traditional batch Williamson ether conditions that we have observed in our previous work where benzylations can take between 3–16 hours.44–46 Increasing the scale of the reaction to 2.7 grams of 5 enabled production of target glycal 1 at a rate of 31.6 g/h. Importantly, to prevent clogging due to the formation of insoluble salts that was observed through the course of the reaction, we chose to exclude a static mixer at the T-junction between the deprotonated glycal and solution of benzyl bromide. Furthermore, the reaction proceeded smoothly at ambient temperature, thus avoiding a cooling bath for another reaction that is typically run at 0 °C.</p><p>We chose to next investigate replacing the silyl ether with a naphthylmethyl (Nap) ether47 to further illustrate the general applicability of this method. Following similar reaction conditions as Corey,39 TBS removal proceeded smoothly with 2.0 equivalents of tetra-butylammonium fluoride (TBAF, 1.0 M in THF) to afford 7 in 93% yield in 10 minutes of total retention time (Figure 2). Upon scale up to 1.7 g of 1, we were able to produce 6 at a rate of 6.4 g/h. Importantly, while typical silyl ether removals are run in THF, we found that it was possible to run this reaction in DMF, which was used for the previous two reactions in this sequence. This use of the same solvent opens up the future possibility of telescoping these reactions into a single step if only a particular product is needed.</p><p>To finish the synthesis of L-olivose glycal 2, we protected the C3 hydroxyl with naphthylmethyl ether following similar conditions as we did with the benzyl ether alkylation. Gratifyingly, naphthylmethyl ether synthesis proceeded smoothly to afford 2 in 82% yield in 6 minutes of total retention time (Figure 3). Scaling up to 1.1 g of 6 proceeded without incident, allowing for production of 2 at a rate of 14.4 g/h. Similar to the benzylation, we found that this reaction could be run at ambient temperatures and did not require cooling to 0 °C.</p><!><p>In summary, we have demonstrated the first modular syntheses of two orthogonally protected glycals of the 2,6-dideoxy glucose L-olivose using only continuous flow processes for each reaction step. Glycal 1 was synthesized in 74% overall yield over three steps with a total retention time of 21 minutes. Similarly, glycal 2 was synthesized in 57% overall yield in 5 steps in 37 minutes of total retention time. These processes are in stark contrast to batch approaches to these molecules, which in our experience can take over 1 week.44–46 All of the reactions in question are run at ambient temperature and most are in the same solvent (DMF), again in contrast to batch approaches to these molecules. Through increased reaction concentration, all reactions were able to be run in less than 10 minutes of total retention time. Furthermore, due to the efficient heat transfer enabled by the flow processes, both TBS silyl protection and alkyl ether installations could be run at ambient temperatures rather than 0 °C, which is typically required for these transformations in batch. We anticipate that this flow-based approach will accelerate glycal synthesis and also provide an invaluable tool to organic chemists who are looking to adopt these steps in their own synthetic pathways to increase rates of substrate production. Efforts to automate this chemistry and telescope it into a single sequence for the continuous production of protected glycans from commercial materials are currently underway.</p>

PubMed Author Manuscript

A Rh(III)-Catalyzed Formal [4+1] Approach to Pyrrolidines from Unactivated Terminal Alkenes and Nitrene Sources

We have developed a formal [4+1] approach to pyrrolidines from readily available unactivated terminal alkenes as 4-carbon partners. The reaction provides a rapid construction of various pyrrolidine containing structures, especially for the diastereoselective synthesis of spiro-pyrrolidines. Mechanistic investigation suggests a Rh(III)-catalyzed intermolecular aziridination of the alkene and subsequent acid-promoted ring expansion for the pyrrolidine formation.

a_rh(iii)-catalyzed_formal_[4+1]_approach_to_pyrrolidines_from_unactivated_terminal_alkenes_and_nitr

<p>Pyrrolidine is a prevalent structural unit in pharmaceuticals, biologically active natural products and materials.1 Consequently, significant efforts have been devoted to the development of efficient routes to pyrrolidines. Early examples can be traced back to the classical Hofmann-Löffler-Freytag reaction in the 19th century.2 Given the ubiquity of the heterocycle, myriad cyclization methods have been developed for its assembly involving nearly every reaction class.3</p><p>Convergent synthetic strategies involving two or three unique components has also received great attention. Among potential reaction partners, olefins are perhaps the most abundant, general and desirable. The [3+2] cycloaddition between azomethine ylides and alkenes represents a well-studied method among them (Figure 1).4 Other recent examples also include the reaction of β-amino aldehydes, protected cinnamyl amines and aziridines with various types of alkenes as two-carbon sources.5 Strategies utilizing alkenes as three-carbon coupling partners are relatively rare and often require prefunctionalization. For example, 2-((trimethylsilyl)methyl) allyl acetate has been reported as a three-carbon source for the formal [3+2] cycloaddition with imines (Figure 1).6 Most recently, a visible light-mediated photoredox catalyzed synthesis of spiropyrrolidines was reported using homoallylic amines as 3-carbon/1-nitrogen sources with aliphatic ketones as 1–carbon partners.7 The use of alkenes as 4-carbon partners is unknown. Herein, we report a Rh(III)-catalyzed formal [4+1] approach to pyrrolidines from α-olefins as 4-carbon sources and hydroxylamine derivatives as nitrogen sources.</p><p>Recently, we reported a branch-selective allylic amination of simple a-olefins using Ir(III) catalysis.8 During the course of this work, we noted that heptamethylindenyl (Ind*) Rh(III) catalyst9 delivers the pyrrolidine side product (3aa) and linear allylic amination product (4) in low yield from 1-hexene (1a) and N-pivalolyloxy tosylamide (2a) (Table 1, entry 1). Inspired by this preliminary result and the rapid access to pyrrolidines, we sought to optimize the transformation. Although desired product could be formed by Rh(III) alone, we found that the yield of pyrrolidine (3aa) is greatly improved by pre-stirring the reaction mixture until full consumption of 1-hexene (1a) and then subjecting the reaction mixture with AgOTf at 80 °C (entry 2).</p><p>Systematic examination of acid additives revealed that Sc(OTf)3 is more effective than AgOTf, promoting the cyclization even at room temperature (entry 3, for detailed information see SI). Eventually, TfOH was chosen as optimal, delivering 66% yield of desired 2-ethyl-1-tosylpyrrolidine (3aa) (entry 4). Electron-deficient heptamethylindenyl (Ind*) ligand on Rh(III) catalyst is crucial with the more common [Cp*RhCl2]2 providing product in much lower yield (entry 5).10 Moreover, other nitrene precursors11 such as dioxazolone or tosyl azide (entry 6, 7) lead to no pyrrolidine product formation.</p><p>We next sought to examine the scope of this method with various terminal alkenes (Table 2). The reaction proceeds smoothly with alkenes containing substituents at C4 and C5 positions, giving desired products (3aa-3da) in good yield. Interestingly, in case of 4,4-dimethyl-1-pentene (1e) where C4 position is blocked, one of the methyl groups migrates to C3 position, forming 2,2,3-trimethyl-1-tosylpyrrolidine (3ea) selectively. Additionally, a variety of functional groups such as -Br, -OAc, -OTs, -NPhth, -CONR2 and –CO2R are tolerated when located far from the reaction center,12 providing corresponding products (3ea-3ja) in moderate to good yield.</p><p>In consideration of the importance of spirocyclic pyrrolidines among pharmaceuticals, we tested synthetic utility of this method for the synthesis of spiropyrrolidines with a variety of allyl cyclohexane substrates (Table 3). Allyl cyclohexane (1l) was successfully converted to the 1-tosyl-1-azaspiro[4.5]decane (3la) product in excellent yield. A conformational bias in the cyclohexane ring (4-tert-butyl, 4- phenyl, 3-methyl and 3-trifluoromethyl) produced corresponding spiro-pyrrolidine products with excellent diastereoselectivity (3na, 3oa, 3pa, and 3qa). It is also worth noting that >20:1 diastereoselectivities are achieved regardless of trans/cis ratio of starting cyclohexane. The relative stereochemistry of 3na and 3pa was unambiguously determined by x-ray crystallography. Interestingly, the reaction with 2-allyl adamantane (1r) causes a 1,2-hydride shift before cyclization, leading to the formation of an unprecedented structure 3ra. A similar hydride shift event was also observed with 1-allyl-2-methylcyclohexane (1s), giving rise to 6,6-fused bicyclic, decahydroquinoline (3sa). Moreover, naphthyl sulfonyl group was introduced as an alternative tosyl protecting group due to its milder deprotection conditions (3nb).13 Last, regiodivergent synthesis of pyrrolidine-based spirocycle (3la) and fused bicyclic compound (5la) was achieved from allyl cyclohexane (1l) with exquisite selectivity by simply changing reaction temperature.</p><p>To obtain mechanistic insights, we first conducted the reaction with 1-hexene (1a) using the standard reaction conditions without the sequential treatment of TfOH. N-tosyl aziridine 6aa was isolated in quantitative yield. As far as we know, the aziridination of unactivated alkenes catalyzed by Ind*Rh(III) is unprecedented, although it is a well-explored research area.14 Further subjecting the resulting aziridine to TfOH leads to pyrrolidine product in 75% yield (3aa) (Scheme 4a).5c,15 These results suggest that the Rh(III)-catalyzed intermolecular aziridination of the alkene and subsequent acid-promoted ring expansion is likely responsible for the reaction mechanism.</p><p>In order to investigate the reaction mechanism, allylic and homoallylic tosyl amides (4 & 7) were subjected to TfOH (Scheme 4b). Allylic amine (4) does not lead to pyrrolidine product, while a quantitative yield was achieved from homoallylic amine (7).16 Furthermore, we continued to test the mechanism with C1 to C4 deuterated alkenes (Scheme 4c). In the case of C3 deuterated aziridine (6ca-C3-d2), extensive deuterium incorporation is observed at C2 position. In addition, significant erosion of deuterium is observed in the case of 6ca-C4-d2.</p><p>Taken together, these experiments suggest a plausible reaction mechanism (Scheme 5). N-pivalolyloxy tosylamide (2a) first coordinates to Rh(III) catalyst and undergoes Rh-nitrene formation (II).17 Alkene aziridination proceeds subsequently to form N-tosyl aziridine (6aa). It is also possible that the aziridine (6aa) is formed through the alkene migratory insertion of intermediate I, followed by a concerted C-N bond formation and N-O bond cleavage. Upon treatment with triflic acid, aziridine ring expansion is initiated by 1,2-hydride shift, followed by a combination of 1,2-hydride shift and elimination/protonation pathways, and eventually quenches the carbocation at C4 position to form the pyrrolidine product (3aa). The possible intervention of homoallylic amide VI explains the modest loss of deuteration in the deuterium labelling experiments noted above.</p><p>In summary, we have developed a Rh(III)-catalyzed formal [4+1] synthesis of pyrrolidines from readily available unactivated alkenes. Mechanistic studies show that the reaction proceeds through a Rh(III)-catalyzed aziridination of the alkene and subsequent ring expansion from aziridine to pyrrolidine promoted by the acid. This method offers a new strategy for pyrrolidine synthesis by employing a simple alkene as a four-carbon source. With this method, various types of pyrrolidines, especially spiro-pyrrolidines were rapidly constructed. Further efforts at elucidating the mechanism and expanding this chemistry are currently underway.</p>

PubMed Author Manuscript

Engineering tyrosine electron transfer pathways decreases oxidative toxicity in hemoglobin: implications for blood substitute design

Hemoglobin (Hb)-based oxygen carriers (HBOC) have been engineered to replace or augment the oxygen-carrying capacity of erythrocytes. However, clinical results have generally been disappointing due to adverse side effects linked to intrinsic heme-mediated oxidative toxicity and nitric oxide (NO) scavenging. Redox-active tyrosine residues can facilitate electron transfer between endogenous antioxidants and oxidative ferryl heme species. A suitable residue is present in the α-subunit (Y42) of Hb, but absent from the homologous position in the β-subunit (F41). We therefore replaced this residue with a tyrosine (βF41Y, Hb Mequon). The βF41Y mutation had no effect on the intrinsic rate of lipid peroxidation as measured by conjugated diene and singlet oxygen formation following the addition of ferric(met) Hb to liposomes. However, βF41Y significantly decreased these rates in the presence of physiological levels of ascorbate. Additionally, heme damage in the β-subunit following the addition of the lipid peroxide hydroperoxyoctadecadieoic acid was five-fold slower in βF41Y. NO bioavailability was enhanced in βF41Y by a combination of a 20% decrease in NO dioxygenase activity and a doubling of the rate of nitrite reductase activity. The intrinsic rate of heme loss from methemoglobin was doubled in the β-subunit, but unchanged in the α-subunit. We conclude that the addition of a redox-active tyrosine mutation in Hb able to transfer electrons from plasma antioxidants decreases heme-mediated oxidative reactivity and enhances NO bioavailability. This class of mutations has the potential to decrease adverse side effects as one component of a HBOC product.

engineering_tyrosine_electron_transfer_pathways_decreases_oxidative_toxicity_in_hemoglobin:_implicat

Introduction<!>Tyrosine electron transfer pathways in globins.<!>Protein preparation<!>HPLC analysis<!>Oxygen affinity measurements<!>Liposome preparation<!>Conjugated diene and singlet oxygen formation in liposomes<!>Autoxidation measurements<!>Reactions with the lipid hydroperoxide HPODE<!>NO dioxygenation rate<!>Heme release from Hb<!>Kinetic reactions of deoxy Hb with nitrite<!>Statistical analysis<!>HPLC comparison of recombinant and native Hb.<!>Functional properties of βF41Y Hb<!><!>Functional properties of the βF41Y mutant<!>Heme loss from recombinant Hb.<!>Hb nitrite reductase activity.<!>Effect of tyrosine mutations on Hb-catalyzed lipid peroxidation.<!>Singlet oxygen production catalyzed by recombinant Hb.<!>Reaction of the lipid peroxide HPODE with recombinant Hb.<!>Tyrosine mutations and lipid oxidation<!>βF41Y as a component of a HBOC<!>Conclusion<!>Abbreviations<!>Author Contribution<!>Funding<!>Competing Interests

<p>Hemoglobin (Hb)-based oxygen carriers (HBOC, colloquially termed 'blood substitutes') have the potential to be transfused in place of packed red blood cells to restore impaired oxygen transport [1]. HBOC offer the potential advantages of universal compatibility, enhanced shelf life, no risk of disease transmission, enhanced and controllable oxygen delivery, improved rheological properties, and more reliable availability. They are also available to individuals who cannot receive conventional blood transfusions for clinical or religious reasons. HBOC have also been used in addition to standard therapy in attempts to augment oxygen delivery in the microvasculature.</p><p>There have been a variety of different approaches to optimize recombinant Hb as a suitable starting material for a HBOC product [2]. Historically, Olson's group has been instrumental in introducing a variety of globin mutants to enhance stability, increase affinity, and decrease nitric oxide (NO) dioxygenase activity [2,3]. Nagai and co-workers [4] engineered genetically linked Hb dimers to increase vascular retention time and also created new interactions with plasma effectors, such as bicarbonate, to enhance oxygen offloading in tissue [5]. Many groups used novel cysteine residues to create sulfhydryl bridges and stabilize higher order protein structures [6,7].</p><p>In a recent review bemoaning the lack of viable approved products, blame was largely placed on side effects due to the redox reactivity of Hb [1]. However, to date, specific issues relating to the oxidative stress induced by recombinant Hb have focused almost entirely on two different areas. First, avoiding significant increases in the autoxidation rate — the conversion of oxyhemoglobin (oxyHb) to methemoglobin (metHb) and superoxide radical. Secondly, avoiding oxidation during the manufacturing/protein production process; for example, Levine et al. [8] showed that modifying the His-2 residue could decrease oxidative modification in the β-subunit and suggested that recombinant proteins that maintained their N-terminal methionine could have additional antioxidative properties [9].</p><p>There is a general acceptance [10] that cell-free Hb induces oxidative stress in vivo by reacting with peroxides (H2O2 or lipid-derived). In the ferrous state (oxy or deoxy), this creates ferryl iron; in the ferric (met) state, both ferryl iron and a globin radical are produced. Note, given that ferryl Hb is autoreduced back to ferric, the eventual production of globin radicals is inevitable even if the starting material contains negligible metHb. We have shown that the plasma antioxidants, ascorbate and urate, act synergistically to maintain extracellular Hb in a functional state [11].</p><p>Following peroxide addition to ferric globins, ascorbate can quantitatively capture both the ferryl iron and the free radical species [12]. However, kinetic limitations, or a decrease in antioxidant levels, may result in partial reduction and a consequent increase in tissue oxidative damage. Tyrosine residues are able to act as redox mediators by one-electron cycling between oxidized (radical) and reduced forms. Indeed, many enzymes, such as catalases and peroxidases, make use of tyrosine redox centers [13,14]. We have shown that myoglobin [15] and the Hb α-subunit [16] have electron transfer pathways that are able to enhance the rate of ferryl reduction by plasma antioxidants. These proteins have a high-affinity saturable pathway where the initial electron acceptor is a tyrosine residue. The Hb β-subunit lacks such a pathway [16]. Introducing such a pathway into the β-subunit in the homologous site where one is present in the α-subunit (βF41Y) resulted in enhanced ferryl reduction in tetrameric Hb [16].</p><p>Myoglobin from the sea hare (Aplysia fasciata) lacks any tyrosine residues. Therefore, like the β-subunit of human Hb, it lacks a high-affinity ferryl reduction pathway. In this model system, we were able to create a variety of phenylalanine to tyrosine mutations with enhanced electron transfer pathways for ascorbate [17]. Furthermore, in the presence of ascorbate, but not in its absence, these mutants had significantly decreased lipid peroxidation activity compared with the wild-type (wt) protein. The purpose of the present study was therefore to determine if introducing a new tyrosine-based electron transfer pathway in human Hb resulted in similar antioxidant properties.</p><!><p>Active site structure of Hb α-subunit (A), Hb β-subunit (B), and myoglobin (C); sites of redox-active tyrosine residues are indicated. Note that in the Hb β-subunit, the tyrosine is substituted with a phenylalanine. Mutating this to a tyrosine (βF41Y) creates a new ascorbate reducible site. This diagram was adapted from a figure originally published in The Journal of Biological Chemistry. Reeder, B. J. et al. Tyrosine residues as redox cofactors in human hemoglobin: Implications for engineering non toxic blood substitutes. J. Biol. Chem. 2008; 283:30780–30787 © the American Society for Biochemistry and Molecular Biology.</p><!><p>The cloning, expression, and purification of the recombinant Hb proteins were carried out as previously described [16], and the proteins were stored in the stable ferrous-CO adduct forms. Native Hb was purified from adult volunteers [11]. To make ferric Hb, the CO forms were incubated with potassium ferricyanide (10 mM) at 4°C, and the CO was removed by constant strong light. The excess ferri/ferrocyanide was removed from the ferric proteins by filtration through a size-exclusion Sephadex G-25 column (∼5 × 0.5 cm). The concentration of Hb was determined by reduction in an aliquot of the ferric protein to the deoxy form using sodium dithionite and using the extinction coefficient of 133 mM−1 cm−1 at 430 nm [19]. The oxy form of the Hb proteins was made by the addition of a slight excess of sodium dithionite to the ferric proteins, followed by passage down a Sephadex G-25 column.</p><!><p>Hb samples were analyzed by reverse-phase HPLC using an Agilent 1290 HPLC fitted with a diode array spectrophotometer. The column used was a Zorbax StableBond 300C3 250 mm × 4.6 mm fitted with a 12 mm × 4.6 mm guard column. Solvents were (A) 0.1% TFA (trifluoroacetic acid) and (B) acetonitrile-containing 0.1% TFA. The gradient was initially 35% solvent B, stable for 10 min, increasing to 37% solvent B over 5 min. This was increased to 40% solvent B over 1 min and then to 43% solvent B over 10 min. The flow rate was 1 ml/min, and the temperature was 25°C.</p><!><p>OxyHb solutions were left overnight under helium flow and in the presence of the Hayashi reducing system before titrations [20]. The oxygen p50 and Hill coefficients were obtained by measuring optical spectra after equilibration at varying pO2 gas mixtures as previously described [21]. For in vitro conditions, p50 values were measured in 100 mM HEPES, 1 mM EDTA, pH 7.0, 15°C; conditions more akin to that in the vasculature (pseudo-physiological conditions) were 100 mM HEPES, 1 mM EDTA, 100 mM sodium chloride and 1.2 mM sodium phosphate, pH 7.0, 25°C. The fractional saturation was calculated by fitting the data to a linear combination of oxy, deoxy, and met Hb reference spectra.</p><!><p>α-Phosphatidylcholine from soybean (Type II-S, from Sigma) was added to the reaction buffer (sodium phosphate, 20 mM, pH 7.4) at a concentration of 5 mg/ml and then sonicated in a water bath for 15 min, until no particulates could be seen. The final liposome solution was maintained a pearlescent appearance. This suspension was then forced through a Northen Lipids extruder containing a membrane of size cutoff 0.1 μm using 25 mm Whatman filtration membranes [22]. A minimum of 10 extrusions per sample preparation was used to produce unilamellar liposomes of ∼100 nm.</p><!><p>The concentration of ferric proteins used in all liposome oxidation experiments was 2 µM heme. The amount of extruded liposomes used in all lipid oxidation assays was ∼30 µl/ml. Lipid oxidation was measured in two ways using UV and fluorescence spectroscopies. Conjugated diene formation was measured by following the increase in absorbance at 234 nm using a Varian Cary 5E spectrophotometer. Data were collected every 7–10 min for up to 12 h. Sample volumes were 120–150 µl. Studies were performed in the presence and absence of physiological levels of ascorbate (30 µM). Dunnett's many-to-one comparison test was used to determine if a mutant was different from control. In addition, due to the variability in the lipid oxidation properties between liposome preparations, in separate experiments, each protein was also compared directly with each other in the same batch of liposomes (n = 3) using unpaired t-tests. Singlet oxygen (1O2) production was measured as fluorescence intensity of the fluorophore Singlet Oxygen Sensor Green (SOSG, Molecular Probes) using a FLUOstar OPTIMA fluorescence plate reader from BMG LABTECH. The stock solution of SOSG (5 mM in methanol) was diluted into the reaction mixture to a final concentration of 0.25 µM. The reactions containing SOSG were excited at 480 nm and emission was followed at 520 nm. Data were collected every ∼3 min for up to 12 h. Sample volumes were 200 µl.</p><!><p>The rate of conversion of ferrous oxy Hb proteins to ferric Hb was monitored by UV–visible spectroscopy in 20 mM sodium phosphate (pH 7.4); for autoxidation measurements at 25°C, spectra (375–700 nm) were collected for up to 48 h. For autoxidation measurements at 37°C, whole spectra were collected for up to 3 h. Sample volumes were 1 ml. Protein concentrations were 10 µM heme. The kinetic traces were analyzed by fitting to single exponential fits.</p><!><p>The met forms of the Hb proteins were reacted with the 13-S hydroperoxy 9-cis, 11-trans octadecadienoic acid (HPODE) in a stopped-flow diode array spectrophotometer (Applied Photophysics, model SX-20). On rapid mixing of the proteins with HPODE, spectral changes in the Soret and visible regions were monitored. Heme destruction was monitored by the bleaching of the Soret peak at 405 nm. Time courses were analyzed by fitting to a double exponential curve, although in the presence of the β41 mutation the two rates were indistinguishable.</p><!><p>An NO solution was prepared by dissolving the NO donor Proli NONOate (Alexis Biochemicals) in a lightly buffered degassed solution at alkaline pH. The concentration of Proli NONOate was determined using an extinction coefficient of 8500 M−1 cm−1 at 250 nm [23]. Proli NONOate was then added to a strong buffer at neutral pH (100 mM sodium phosphate, pH 7.4) to release free NO gas; under these conditions, 1.8 molecules of NO are released per Proli NONOate molecule. The oxy form of the relevant Hb solution (in 100 mM sodium phosphate, pH 7.4) was placed in one syringe and the NO released from the Proli NONOate in the other. Both the solutions were then mixed rapidly in equal volumes in the stopped-flow spectrophotometer. After mixing, the Hb concentration was 4 μM and the NO concentration was 30 μM. The absorbance changes were monitored at single wavelengths of 405 and 430 nm over a 0.1 s time range. Owing to the speed of the reaction, the temperature was set to 15°C.</p><!><p>The vector pEMBL19-containing sperm whale myoglobin (SW Mb) H64Y/V67F was a kind gift from John Olsen (Rice University) [24] and was then His-tagged to simplify purification. The gene for SW Mb (H64Y/V67F) was amplified by overhanging PCR with the primers 5′-CCACATATGGTTCTGTCTGAAGGTGAATGGCAGCTG and 5′-CATGGATCCTCATTAACCCTGGTAACCCAGTTC (Eurofins) to introduce novel restriction sites for NdeI and BamHI (shown in bold). The PCR fragment was restricted with NdeI and BamHI then ligated into pET28a vector cut with the same restriction endonucleases to generate pET28a-SWMb (H64Y/V67F). To prepare His-tagged SW Mb (H64Y/V67F), BL21 (DE3) cells were first freshly transformed with the expression vector pET28a-SWMb (H64Y/V67F) and subsequently grown in Luria–Bertani media at 37°C to an optical density of ∼0.8 at 600 nm. Protein expression was then induced with 0.5 mM isopropyl β-d-1-thiogalactopyranoside, 0.25 mM aminolevulinic acid, and 0.1 mM ferric citrate. Cultures were then bubbled with pure CO gas, flasks sealed thoroughly with rubber bungs and grown for a further 18 h at 27°C and 90 rpm. Cells were then harvested by centrifugation at 4000 rpm for 20 min at 4°C. The cell paste was resuspended in 40 ml of buffer A [50 mM NaPi (pH 7.2) and 100 mM NaCl] and cells were lysed by using an Avestin C3 Emulsiflex homogenizer. The cell lysate was cleared by centrifugation at 18 000 rpm for 30 min at 4°C and the supernatant was filtered using 0.45 μm filters before being loaded onto a 5 ml HisTrap-HP column (GE Healthcare) pre-equilibrated with buffer A. The protein was eluted with a 50 ml linear gradient of buffer B [50 mM NaPi (pH 7.2), 100 mM NaCl, and 1 M imidazole] and fractions containing the myoglobin were pooled. The heme was then extracted to create apomyoglobin as described previously [25]. Heme release from the met forms of wt and βF41Y was then measured by incubating the proteins with an excess of the heme-binding apomyoglobin mutant H64Y/V67F and monitoring absorbance changes at 37°C in 100 mM NaPi pH 7.2, containing 0.15 M sucrose [24]. The fast phase of heme loss (β-subunit) was measured using single exponential fits to the first 60 s of data in a stopped-flow spectrophotometer looking at the appearance of the 600 nm peak. The slow phase (α-subunit) was also monitored at 600 nm using a Cary 5000 spectrophotometer, fitting to a double exponential and reporting the slower rate.</p><!><p>Formation of the nitrosyl complexes from the deoxy forms of the recombinant Hb proteins wt and βF41Y was carried out in excess sodium dithionite [26]. Spectral changes were monitored in the Soret and visible regions, and time courses were fitted to 432–413 nm as single exponential fits (min−1) over a 20 min time range.</p><!><p>The program KaleidaGraph 4 (Synergy Software) was used for analysis. In experiments involving two conditions, unpaired t-tests were used to determine significance. Where multiple tests or treatments were involved, ANOVA followed by Dunnett's many-to-one comparison test was used to determine if any condition was different from the control.</p><!><p>Reversed-phase HPLC separation of heme cofactor and protein subunits from native Hb, wt recombinant Hb and the βF41Y mutant.</p><!><p>Mean ± SD (n = 3).</p><!><p>*P < 0.05 compared with wt. For detailed conditions, see the Experimental section.</p><!><p>Ferrous oxyHb is susceptible to autoxidation, producing the nonfunctional ferric (metHb) form and the superoxide radical. Enhanced autoxidation rates and accompanying heme loss from the ferric protein have the potential to prove problematic for HBOC applications [27]. The introduction of the βF41Y had no deleterious effect on autoxidation rate (Table 1), either at ambient temperature or at 37°C.</p><!><p>Heme loss from metHb and binding to the apo H64Y/V68F mutant. The spectral changes show the reaction following mixing metHb with the apo H64Y/V68F mutant for wt (bottom) and βF41Y (top). The main figure shows the spectral changes every 13 min from t = 0 to t = 4 h. An expanded region (×7) of the visible spectra in the 600 nm region is also shown. The inset shows the time course for the reaction in the visible region used for further kinetic analysis. Conditions: 100 mM NaPi, pH 7.2, containing 0.15 M sucrose; T = 37°C; [Hb] = 2.5 μM; [apo mbH64Y/V68F] = 30 μM.</p><!><p>Formation of nitrosyl Hb from deoxy Hb, wt, and βF41Y recombinant forms, in the presence of 0.2 mM nitrite. Conditions: sodium phosphate buffer (20 mM, pH 7.4, 25°C); [Hb] = 5 μM. Inset: the rate (mean ± SD, n = 6) of conversion from deoxy to nitrosyl. Asterisk indicates significant difference from wt (P < 0.0001).</p><!><p>Liposome oxidation monitored by conjugate diene formation following the addition of recombinant metHb in the presence or absence of ascorbate. Assay conditions: sodium phosphate (20 mM, pH 7.4, 30°C); [heme] = 2 µM; data presented as mean ± SD, n = 9. Asterisks indicate significant difference from wt (P < 0.05).</p><!><p>Changes in singlet oxygen (fluorescence) and conjugated diene (absorbance) formation following the addition of metHb to liposomes (A). Comparison of wt and βF41Y metHb addition in the absence (B) and presence (C) of 50 μM ascorbate. Addition of wt and βF41Y oxyHb to liposomes (D). Assay conditions: sodium phosphate (20 mM, pH 7.4, 30°C); [heme] = 2 µM; [SOSG] = 0.25 µM. Data presented as mean ± SD, n = 8.</p><!><p>Addition of 32 μM HPODE to metHb (wt and βF41Y). Time courses of heme damage (inset) were fit to double exponentials and plotted against [HPODE]. Conditions: sodium phosphate (20 mM, pH 7.4); [heme] = 2 μM after mixing; T = 25°C.</p><!><p>The present study demonstrates the potential advantages of engineering tyrosine-based electron transfer pathways as a component of a HBOC, especially if combined with antioxidants. The exemplar protein studied was the βF41Y mutation, creating a novel protein with similar electron transfer sites facilitating ferryl reduction in both the α- and β-subunits. It is not a priori true that introducing a new pathway for electrons to enter and leave ferryl Hb will have new antioxidative properties; it could be argued that a new pathway could facilitate oxidative damage induced by Hb, making it easier for nearby lipids and proteins to be modified by the ferryl species. However, at least in the case of lipids, that did not seem to be the case. The onset of liposome oxidation by metHb was unchanged whether the electron transfer pathway was removed in the α-subunit or added to the β-subunit. However, in the presence of plasma antioxidants, such as ascorbate, the βF41Y mutation protected the liposomes from damage. This is consistent with our studies in Aplysia myoglobin, where we introduced two novel, fast, tyrosine-based electron transfer pathways [17]. Both of these protected against liposome oxidation, but again only in the presence of ascorbate.</p><p>Rather than being catalyzed by Hb ferric/ferryl redox cycling [10], liposome oxidation could be initiated directly by the ferric ion alone in metHb or indirectly via heme iron releasing from the protein and entering the lipid bilayer. The results in the present paper argue against this. Ascorbate is protective against lipid peroxidation in the presence of our mutants. Yet, ascorbate reduction in metHb [11] is slow compared with ascorbate reduction in ferryl Hb, and βF41Y does not enhance this already very slow rate (results not shown). In the case of heme insertion into the liposomes, the βF41Y shows slightly enhanced heme release (Table 1) and so, if anything, would be expected to be more pro-oxidative in this system.</p><p>There are two alternative mechanisms by which tyrosine mutations could increase the lag phase of lipid oxidation in the presence of ascorbate. The first is by direct reduction in the ferryl species that react with lipid peroxides. However, it is also theoretically possible that the tyrosine residue could provide a pathway whereby the unpaired electron on a lipid peroxide radical is transferred to ascorbate via Hb. In the latter case, electron transfer to/from the heme iron would not be directly involved. These mechanisms are not mutually incompatible and could operate synergistically. This could explain why in Aplysia Mb introducing new tyrosine residues enhances the reduction in ferryl Mb and decreases lipid oxidation, but the two activities do not closely correlate, that is, the mutant that reduces ferryl fastest is not the one that has the greatest inhibition of lipid oxidation. It could also explain why moving the tyrosine residue from α42 to β41 is partially protective in the presence of ascorbate (the present paper), despite only limited effects on the ferryl reduction rate [17].</p><p>Liposome oxidation catalyzed by oxyHb appears to be slower in βF41Y even in the absence of ascorbate. The reason for this could also be linked to the idea that βF41Y is in a particularly apposite environment to donate electrons. The reaction of oxyHb with liposomes is more complex than that of metHb, as not only does the heme react with lipid peroxides, but there is also a contribution from the autoxidation rate (which itself forms superoxide radicals). The autoxidation rate is slightly slower in βF41Y (Table 1) and this could at least partially explain the observed protection in the oxyHb mutant. However, there also seems to be a direct effect of lower reactivity with lipids. This is shown by the slower reaction of βF41Y metHb with the lipid peroxide HPODE. Globin reactions with HPODE are mechanistically complex [29]. Therefore, in the present paper, we limited ourselves to looking at the simpler endpoint of heme destruction. In wt Hb, this is biphasic. However, in the mutant, the faster phase is lost. The simplest explanation for this is that the presence of a redox-active tyrosine close to the heme enables the safe removal of an oxidizing species. Although we have previously suggested that the final stable protein radical following peroxide interaction with Hb is at βTyr145 [30], it is likely that the tyrosine 42 in the α-Hb subunit (or its homologous location in other globins) is the initial destination for the radical once it leaves the porphyrin ring [14,31,32]. Lacking such a tyrosine, the β-subunit may be more susceptible to heme destruction by HPODE. In this simple model, the biphasic reactivity seen in the present paper would be a simple consequence of differing reactivity at the α- and β-subunits, with the βF41Y mutation equalizing the relative rates.</p><p>Taken together, these results strongly suggest that the addition of a redox-active heme-accessible tyrosine to the β-subunit of human Hb would be a useful component of a future recombinant HBOC designed to decrease ferryl heme reactivity, especially when combined with a suitable reductant.</p><!><p>The βF41Y mutation is functional in vivo as it has been described in a human polymorphism (Hb Mequon). The clinical subject was not anemic, and the blood oxygen affinity was normal and sensitive to the effector BPG [18]. The Hb Mequon patient was suffering from severe hemolysis following a viral illness treated with acetaminophen. However, this was an isolated incident and none of the family relatives carried any hematological disorder. Although it is possible that there is some latent instability in Hb Mequon, it remains unclear whether there is any causal link between βF41Y and the hemolytic event. In any event, the existence of such a stable polymorphism in an otherwise healthy 34-year-old patient suggests that βF41Y is a relatively benign polymorphism. Although there may be subtle differences between the recombinant and native version of this protein, the data in the present paper support the idea that the βF41Y mutation is relatively benign; perhaps, the tendency for faster heme loss is ameliorated by enhanced antioxidative properties? In any event, there appears no reason a priori to dismiss βF41Y as a component of an acutely administered HBOC.</p><p>The oxygen affinity and autoxidation rate of βF41Y have been studied extensively under a range of conditions by the groups of Baudin-Creuza and Marden [33,34]. They showed a decrease in oxygen affinity compared with Hb A (the native protein purified from blood) and attributed this to stabilization of the T-state conformation. Our study is consistent with theirs in that a decrease in oxygen affinity was observed in the presence of 100 mM chloride. However, a slight increase in affinity in the absence of chloride was seen. Part of the difference is possibly because we are comparing our protein to the recombinant wt rather than Hb A; indeed, most of the difference seen in the absence of chloride is due to differences in our wt compared with their Hb A, rather than differences in the mutant itself. In general, when analyzing mutant Hbs, we caution against comparing solely against native Hb A, as recombinant proteins can be modified during growth or incompletely processed (for example, due to incomplete cleavage of the N-terminal methionine). In this regard, we found some reduction in co-operativity in βF41Y compared with previous publications [34]. It is possible that this was due to protein heterogeneity as in our hand values of the Hill coefficient in different preparations are more variable than those for the oxygen affinity; measurements on wt Hb from five different preparations showed robust p50 values under our 'pseudo-physiological' conditions (5.66 ± 0.42 mm Hg), but rather variable numbers for the Hill coefficient (2.23 ± 0.42).</p><p>In contrast with previous results, which showed a small increase in the autoxidation rate when compared with Hb A [35], we found that when compared with wt, βF41Y had a slightly decreased rate of autoxidation. Measurement of this rate is highly dependent on protein concentration and the nature of the analysis [35], and we would not put too much weight on these small differences. Of more concern for an HBOC is the rate of heme loss, which has not previously been measured in βF41Y. Heme loss from extracellular Hb is a concern in HBOC, particularly as free heme has recently been shown to act as a damage-associated pattern molecule, activating the immune response system via the TLR-4 receptor [36]. A small increase in this rate in the β-subunit was observed, which is of some concern. It is likely that this increased heme loss will need to be addressed in any final HBOC product, either by the additional mutants designed to 'waterproof' the heme pocket [37] or by subsequent chemical modifications. Alternatively, F41Y is not the only place, where a tyrosine can be introduced into the β-chain to enhance ferryl reduction. We have shown the same effect with another mutation introduced on the surface of Hb, βK66Y [38]; although this mutant is not oxidatively stable, it has decreased heme loss compared with wt. Therefore, the ability to enhance ferryl reduction does not automatically correlate with enhanced heme loss, suggesting that a different tyrosine mutation to βF41Y could be an even better starting material for a new HBOC.</p><p>It appears unlikely that surface-accessible tyrosine, even those close to the heme like βF41Y, would significantly affect the NO dioxygenation activity as this generally requires modification of the heme pocket [28]. This is indeed what was found; F41Y only showed a small decrease in NO scavenging. However, it did significantly enhance the nitrite reductase rate. Although this is known to be under allosteric control, and thus a function of the oxygen p50 [39,40], the βF41Y mutation affected the intrinsic nitrite reductase rate of T-state Hb independent of allostery. The three-fold increase in this rate in βF41Y is similar to the increase seen in fetal compared with adult Hb [41,42]; it is not clear if these differences have a kinetic or thermodynamic explanation.</p><p>The co-administration of nitrite has been suggested as being beneficial for maintaining NO levels following HBOC addition. However, enhanced metHb formation makes this problematic and in vivo results have not been promising [43]. The results from βF41Y suggest that — like NO dioxygenation — the nitrite reductase activity of Hb may be amenable to manipulation, raising the possibility of maintaining NO levels without enhancing metHb formation.</p><p>The idea of using βF41Y as part of a HBOC has been suggested previously [34]. However, the design rationale [35] was the engineering of a lower intrinsic oxygen affinity (by stabilizing the T-state), rather than enhancing the interaction with external reductants. The work here suggests that F41Y possesses an additional key property; in the presence of suitable reductants, βF41Y has superior antioxidative properties compared with wt Hb. This property is likely to be shared by other tyrosine mutations [17].</p><!><p>In summary, the addition of a new surface-accessible tyrosine residue in the β-subunits renders Hb less oxidatively damaging without compromising its ability to act as an extracellular oxygen carrier. This enhanced reactivity with plasma antioxidants is not just a function of βF41Y in human Hb. We have shown even larger effects after introducing tyrosine residues to a variety of novel sites on the surface of Aplysia myoglobin [17]. Historically, the lack of a clinically viable HBOC product has been seen as due to adverse side effects rather than a lack of functionality in oxygen transport [44]. This report demonstrates that the introduction of new tyrosine electron transfer pathways in human Hb renders it less pro-oxidant. Alternative attempts to render Hb less oxidatively damaging have included cross-linking to antioxidant enzymes [45] or cellular antioxidants [46]. Protein engineering has the potential to deliver similar benefits in a more defined and homogenous product. We suggest that appropriate redox-active tyrosine mutations could form the basis for a new generation of HBOC that retain the functionality of previous products, but limit the adverse side effects.</p><!><p>Hb, hemoglobin; HBOC, hemoglobin-based oxygen carrier; HPODE, 13-S hydroperoxy 9-cis, 11-trans octadecadienoic acid; metHb, met(ferric) hemoglobin; NO, nitric oxide; oxyHb, oxygenated hemoglobin; SOSG, Singlet Oxygen Sensor Green; TFA, trifluoroacetic acid; TLR-4, Toll-like receptor 4; wt, wild-type recombinant.</p><!><p>C.E.C., M.T.W and B.J.R conceived and initiated the project. G.G.A.S performed the majority of experiments with help from R.S.S., M.S., K.K., K.R., L.R. and B.J.R. G.G.A.S. analysed the data with additional contributions from C.E.C., R.S.S., L.R., L.B., A.M. and B.J.R. C.E.C. and B.J.R. designed the experimental strategy. C.E.C. wrote the paper which was critically reviewed by G.G.A.S, B.J.R., A.M., L.R., L.B. and M.T.W.</p><!><p>We thank the UK Biotechnology and Biological Sciences Research Council for financial support [BB/L004232/1].</p><!><p>C.E.C., B.J.R., and M.T.W. have a patent relating to modification of hemoglobin amino acids designed to render a blood substitute less toxic. C.E.C., B.J.R., M.T.W., and G.G.A.S. are shareholders in a related company (CymBlood).</p>

PubMed Open Access

Improved Templated Fluorogenic Probes Enhance the Analysis of Closely Related Pathogenic Bacteria by Microscopy and Flow Cytometry

Templated fluorescence activation has recently emerged as a promising molecular approach to detect and differentiate nucleic acid sequences in vitro and in cells. Here we describe the application of a reductive quencher release strategy to the taxonomic analysis of gram-negative bacteria by targeting a single nucleotide difference in their 16S rRNA in a two-color assay. For this purpose, it was necessary to develop a release linker containing a quencher suitable for red and near-infrared fluorophores, and to improve methods for the delivery of probes into cells. A cyanine-dye labeled oligonucleotide probe containing the new quencher-release linker showed unprecedentedly low background signal and high fluorescence turn-on ratios. The combination of a fluorescein-containing and a near-IR emitting probe discriminated E. coli from S. enterica despite nearly identical ribosomal target sequences. Two-color analysis by microscopy and the first successful discrimination by two-color flow cytometry are described.

improved_templated_fluorogenic_probes_enhance_the_analysis_of_closely_related_pathogenic_bacteria_by

Introduction<!>Preparation of Reactive Oligonucleotide Probes<!>Kinetic Measurements<!>Fluorescence Microscopy<!>Flow Cytometry Analysis<!>Bactericidal Activity of Surfactants<!>Surfactant-Mediated Effects on Probe Signals<!>Design, Preparation and Evaluation of Quenched Near-IR Probes<!>Fluorescence Activation and Sequence-Specificity<!>Effect of Probe Delivery Protocols on Bacteria Growth<!>Effect of Delivery Agents on Probe Background<!>Effect of Detergents and Polymers on Probe Uptake<!>Two-Color Discrimination of Bacteria by Fluorescence Microscopy<!>Two-Color Analysis of Bacteria Cultures by Flow Cytometry<!>Discussion

<p>The development of methods for rapid detection and identification of human pathogenic bacteria is widely regarded as important both in medical diagnostics and in monitoring food safety. The standard analytical approach for most bacteria involves growing cultures followed by microbiological assessment, which is time consuming, labor intensive and requires expert personnel. Consequently, there is significant interest in developing methods to identify microbes directly from clinical samples with a straightforward readout. Of particular appeal with this respect are fluorescence in situ hybridization (FISH) methods,1,2 which target specific RNA or DNA sequences with labeled hybridization probes followed by microscopic analysis or flow cytometry.3 Bacterial FISH assays frequently target ribosomal RNAs (rRNAs) because these markers are highly expressed and the target accessibility is well established.4,5 However, it is challenging to distinguish closely related pathogens with standard-length hybridization probes, because rRNAs are evolutionarily conserved and a difference of only one or two nucleotides in a target sequence has an insignificant influence on the binding of a standard-length probe of 25 nucleotides or more.6 Furthermore, standard FISH assays involve cell fixation and washing steps, extending the time and care required for analysis.</p><p>One possible approach to avoid these limitations of standard FISH probes involves the use of quenched probes that report the presence of a target with a fluorescence turn-on signal.7 Prominent examples of fluorescence turn-on probes include molecular beacon (MB) probes,8 forced intercalation (FIT) probes,9 and templated reactive probes.10 Because such probes are non-fluorescent before binding their target, they can (at least in principle) be used without washing steps, thus simplifying the genetic analysis considerably. Indeed, MB probes have been employed in the analysis of bacterial RNAs.11,12 In a different molecular strategy, templated reactive probes allow the identification and analysis of bacteria in a single step requiring as little as 45 min incubation time.13 In this approach, two modified oligonucleotide probes bind at adjacent sites on a nucleic acid target sequence, bringing two chemically reactive groups together to generate a fluorescence signal.10 The possibility to use short probes provides this strategy with excellent mismatch selectivity on highly expressed, conserved RNA sequences such as rRNAs.13–17 This high degree of sequence specificity is a major advantage over standard FISH probes for which single base specificity is possible18 but is often challenging. Two-color sets of templated quenched autoligation (QUAL) probes enabled discrimination of bacteria by a simple fluorescence color-call.19 In preliminary experiments, this technique has been applied to distinguish Escherichia coli from Salmonella enterica13–17 and Shigella19 pathogens by fluorescence microscopy and single-color flow cytometry.15 Several laboratories have recently reported advances in templated reactive oligonucleotide designs;20–25 examples included probes that are based on templated Staudinger reactions.26–30 Among these, we have described quenched Staudinger reaction triggered α-azidoether release (Q-STAR) probes, which release a quencher after an azide reduction reaction on a linker carrying it.31–33</p><p>Despite the promise of early experiments, challenges remain for the general application of templated probes in bacteria, including spectral limitations and problems with probe delivery into cells. Previous two-color reactive probe systems for the detection of bacterial rRNAs relied on a FRET-reporter strategy, in which one probe contained fluorescein and the second one fluorescein combined with a TAMRA FRET-acceptor.17,19,31 This strategy offers direct visual readout under a fluorescence microscope with single-wavelength excitation. However, low emission quantum yield and poor spectral resolution render such FRET-based probes unsuitable for quantitative analytical methods such as flow cytometry, which may benefit from probes with fluorophores that are excited at separate wavelengths. Multi-color QUAL probes have been described and applied to imaging of bacteria,34 however the dabsyl group intrinsic to the QUAL design quenches red and near-IR emitting dyes inefficiently. Other examples of two-color reactive probes based on two distinct profluorophores have been described but their potential for cellular RNA probing remains untested.35,36 Delivery of oligonucleotide probes into bacteria is another challenge for the direct detection of genetic markers in microbes. Cell fixation is one possible solution; however, this adds extra time and effort to the experimental protocol, renders bacteria less amenable to separation by flow cytometry, and excludes the possibility of isolating live cells for further analysis. One report described the use of a detergent to aid in delivery of FISH probes into bacteria, offering a simple alternative to fixation.15 However, it is unknown to what extent various detergents might aid probe uptake into bacteria and how such reagents affect bacterial viability.</p><p>Here we describe the preparation and evaluation of quencher release probes suitable for red and near-IR emitting fluorophores and their potential for pathogen analysis in a two-color assay with fluorescein-labeled probes (Figure 1).31 We found that these probes enable the discrimination of E. coli and S. enterica with improved signal contrast by fluorescence microscopy and by flow cytometry. Moreover, we investigated the use of various detergents and delivery agents on probe uptake, probe background signals and bacterial viability. Furthermore, we outline and discuss the remaining challenges for development of practical whole-cell microbial detection assays based on templated reactive probes and light-up probes in general.</p><!><p>Probes were prepared by solid-phase conjugation of the quencher release linker (synthesis described in Supporting Information) to the protected, fluorophore labeled (Quasar 670-dT, Biosearch Technologies) NIR STAR probes and (Fluorescein-dT, Glen Research)) green STAR probes oligonucleotides containing a 5′-terminal amino modifier (5′-amino-modifier 5, Glen Research). NIR STAR probes were synthesized with UltraMILD deprotection phosphoramidites (Glen Research); labeling of these oligonucleotides with 3 provided the NIR STAR probes in lower yield than green STAR probes. The major side product had a mass corresponding to the amine-modified DNA plus a Pac group, possibly resulting from a transfer of a Pac protecting group from nucleobases to the terminal amine under the quencher conjugation conditions. For the introduction of the 2,6-diaminopurine (DAP) DNA monomer, we used 2-Amino-dA-CE phosphoramidite (Glen Research). The trityl protecting group of the terminal amine was removed by repetitive cycles of 2 % trichloroacetic acid in DCM on the DNA synthesizer. Controlled pore glass (CPG) containing the oligonucleotides was incubated with a solution of the quencher release linker (25 mM), PyBOP (25 mM) and DIEA (50 mM) in DMF (250 μL) and gently shaken at room temperature for 5 h, protected from light. The supernatant was removed and the beads washed with DMF (2×) and MeCN (3×) until the solution was colorless. For cleavage from the solid support and deprotection of green STAR probes, the CPG were incubated for 90 min with concentrated aqueous ammonia/methylamine at 55 °C; for NIR STAR probes, beads were incubated in 0.5 M potassium carbonate solution in methanol for 4 h at room temperature. The purity of the probes was assessed by analytical HPLC chromatography (the probes were found to be >95% pure) and the molecular structure was verified by MALDI-TOF mass spectrometry analysis. The triphenylphosphine (TPP)-DNA probe was prepared as previously described.31 Unmodified template and helper DNAs were obtained from the Stanford Protein and Nucleic Acid Facility.</p><!><p>Quenched probes (200 nM) and the corresponding template (200 nM) were incubated at 37 °C in Tris-borate buffer (70 mM, pH 7.55) containing MgCl2 (10 mM). TPP-DNA (600 nM) was added and the fluorescence emission (λex = 494 nM; λem = 521 nm for green STAR; λex = 644 nm and λem = 670 nm for NIR STAR) was measured as a function of time on a Fluorolog 3 Jobin Yvon spectrometer equipped with an external temperature controller.</p><!><p>E. coli K12 and S. enterica were grown to mid-log phase (OD600 = 0.4) in Luria-Bertani (LB) media with rapid shaking at 37 °C. Aliquots of the media (100 – 200 μL) were carefully centrifuged and the supernatant decanted. Pellets were washed with 1× PBS buffer (pH 7.4) and resuspended in 6× saline sodium citrate (SSC) buffer (pH 7.4) containing 0.05% SDS. Green STAR (200 nM), NIR STAR (200 nM), TPP-DNA (2 μM), and helper DNAs (3 μM) were added and the samples incubated at 37 °C for 3 h without shaking and protected from light. Aliquots of incubated bacteria suspensions were mixed with 2% agarose solution and spotted on cover slides without washing or fixation. Imaging was performed on a Nikon Eclipse E800 epifluorescence microscope equipped with a Nikon Plan AP 100×/1.40 oil immersion objective and a SPOT RT digital camera; filter sets were B-2A (excitation: 450–490 nm, dichroic mirror: 500 nm, emission: >500 nm) for green STAR and Cy5 HYQ (excitation: 590–650 nm, dichroic mirror: 660 nm, emission: 663–738 nm) for NIR STAR probes.</p><!><p>Bacteria were grown and treated with reactive probes as described for fluorescence microscopy experiments. Bacteria were diluted 10-fold in 6× SSC and the cell suspension immediately analyzed using an LSR I flow cytometer (BD Biosciences). Side angle light scattering was used as the triggering event. Forward angle light scatter, side angle light scattering were recorded as well as fluorescein (λex = 488 nm laser; λem = 530 – 560 nm) and Q670 (λex = 640 nm laser; λem = 666 – 690 nm) fluorescence emission. Flow cytometry data was analyzed with FlowJo software (Tree Star).</p><!><p>Bacteria cells were grown to mid-log phase (OD600 = 0.4), concentrated 10-fold by mild centrifugation and dispersed in SSC buffer (pH 7.0). Bacteria were incubated with different additives for 4 h at 37 °C without shaking. Aliquots of the bacteria samples were spread over solid LB-agar medium in Petri dishes and the dishes incubated at 37 °C overnight. The degree of bacteria growth was assessed visually.</p><!><p>A mixture of Q-STAR (green STAR SE, 200 nM) and TPP-DNA (2.0 μM) probes was incubated in SSC buffer (1×, pH 7.0) containing different additives (1.0% w/v) at 37 °C for 4 h. The fluorescence intensity was measured with a Flexstation II microplate reader (Molecular Devices) in a 96-well quartz microplate (λex = 494 nm, λem = 525 nm). The fluorescence intensity was normalized by dividing by the fluorescence intensity of a sample containing the probes in SSC buffer only.</p><!><p>We aimed to develop a quenched probe with a red-light emitting fluorophore that could be used in combination with fluorescein-labeled green STAR probes31 for two-color instrumental analyses. The choice of a new long-wavelength dye required pairing it with a compatible quencher, since the original dabsyl quencher (see 1 in Chart 1) is limited to shorter-wavelength fluorophores (λem < 550 nm). Thus we designed the new quencher-linker 2 containing a Black Hole Quencher™ 1 (BHQ-1) and the same α-azidoether linker structure as the preceding dabsyl-probe 1. The synthesis of 2 was identical to that of 1 except for the quencher attachment step, which involved HBTU-mediated amide bond formation (Supporting Information). To test the cleavage kinetics, the linker 2 was incorporated at the 5′-terminus of a labeled Q-STAR probe specific to a target sequence of 16S rRNA of E. coli (green STAR EC BHQ-1, Table 1).</p><p>Evaluation of the quencher-linker 2 showed that the DNA-templated reaction was rapid but that background quencher release is elevated compared with other probes. The presence of the EC DNA template induced a rapid reaction between green STAR EC BHQ-1 and TPP-DNA, eliciting a strong fluorescence turn-on signal (Figure 2a). The reaction kinetics were similar to those observed previously for probes containing linker 1.31 Control experiments (mismatch template SE DNA, no template, no TPP-DNA) confirmed the target-specificity of the probes (Figure 2a). Interestingly, the fluorescence-activation was similar for all control experiments, indicating significant phosphine-independent breakdown of probes containing the BHQ-1 release linker 2. Mass spectrometry analysis revealed cleavage of the α-azidoether group upon incubation of probe alone at room temperature in buffer.</p><p>We hypothesized that the carboxamide moiety of linker 2 is responsible for the significant spontaneous decomposition of these probes, possibly by stabilizing the transition state of α-azidoether hydrolysis37 via an unidentified cyclic intermediate. Based on this assumption, we synthesized linker 3 (Chart 1) with an additional methylene group to disfavor a possible intramolecular interaction. Concomitantly, we changed the quencher to Black Hole Quencher™ 2 (BHQ-2) because this dye quenches red and near IR fluorophores more efficiently than BHQ-1.38 The synthesis of BHQ-2 release linker 3 is similar to those of 1 and 2, affording the product in 6 steps (Supporting Information).</p><p>Using the new quencher-linker 3, we prepared NIR STAR probes containing a Cy5-derivative (Quasar™ 670, Q670) as the fluorophore (Table 1). Probes were complementary to a target site on 16S rRNA containing a single site difference between E. coli (EC) and S. enterica (SE).15</p><!><p>Incubation of NIR STAR EC with TPP-DNA and the matched template EC DNA rapidly elicited a strong increase of fluorescence (λex = 644 nm, λem = 670 nm) with a kinetic trace similar to that of green STAR EC BHQ-1 (Figure 2b). The measured fluorescence turn-on ratio (fluorescence intensity after quencher release, divided by fluorescence intensity before addition of TPP-DNA) exceeded 200-fold, establishing that the fluorophore is quenched with >99.5% efficiency. This quenching effect compares favorably with probes containing a dabsyl/fluorescein pair (67-fold fluorescence activation)31 and is presumably caused by the high quenching efficiency of BHQ-2.38 Importantly, the generation of background fluorescence for NIR STAR EC was distinctly slower in the absence of TPP-DNA than for green STAR EC BHQ-1 and was below the rate of the reaction in the absence of the template or in the presence of the singly mismatched template SE DNA (Figure 2b). A second NIR STAR probe complementary to the sequence corresponding to S. enterica (NIR STAR SE, Table 1) also exhibited sequence-selective fluorescence activation of similar rate and magnitude (Figure S2, Supporting Information) although with significantly lower single-mismatch specificity because of the small energetic difference between the A–T match and the A–G mismatch, one of the most challenging mismatches to discriminate.39</p><p>To enhance sequence discrimination of SE DNA at this target site, we investigated the use of the 2,6-diaminopurine (DAP) base surrogate, which reportedly enhances single nucleotide discrimination over canonical adenine.40 We prepared a DAP-containing green STAR probe1 (green STAR SE DAP, Table 1) and compared its sequence-specificity to a probe with an adenine at this position (green STAR SE, Table 1). The SE DNA-mediated reaction was faster with green STAR SE DAP than with green STAR SE and the reaction mediated by the mismatch-containing EC DNA target was reduced for green STAR SE DAP relative to green STAR EC (Figure 2c). These two favorable effects synergistically yielded a 2-fold enhancement of mismatch selectivity for the DAP probe; after 9.5 min incubation, the sequence-specificity reached a maximum of 19.2 for green STAR SE DAP and only 8.8 for green STAR SE (Figure 2c inset).</p><!><p>Protocols that enable cellular probe delivery without complex and time-consuming fixation steps would be advantageous to harness the simplicity of fluorogenic probes in bacteria. Previous studies described the introduction of oligonucleotide probes into gram-negative bacteria by incubating the prokaryotes in hyperosomotic buffer (6× saline-sodium citrate buffer, SSC) containing the detergent SDS (sodium dodecyl sulfate).15,34 Although early studies suggested that bacteria remain viable under these conditions,15,41 this conclusion remains uncertain.42 The survival of the bacteria is unnecessary for many diagnostic applications; however, it would be useful for example for culturing colonies from genetically sorted bacteria.</p><p>We tested the toxicity of SDS on the bacteria and compared it with the bactericidal effect of a small array of selected surfactants also to be tested for probe delivery. We examined cationic (HDTA), anionic (SDS, STDC), neutral (Triton X-100; Tween 20) and zwitterionic (CHAPS; DPAPS) detergents and selected polymers previously used for cell delivery purposes (polyethylene imine, PEI;43 Pluronic F-6844) (for full names and molecular structures see Table S1, Supporting Information). As anticipated, different additives exhibited widely varying toxicities (Figure 3 and Figure S1, Supporting Information). HDTA was highly toxic at all tested concentrations, in agreement with the bactericidal activity of cationic lipids.45 The neutral surfactants exhibited low toxicity in the investigated concentration range. Zwitterionic and anionic detergents, including SDS, exhibited a dose-dependent toxicity. SDS significantly inhibited colony formation in 6× SSC buffer; colonies formed only for the sampled with 0.01% ot lrdd SDS but were absent in samples for 1.0% and 0.1% SDS (Figure 3). The data reveal that the previously employed delivery protocol (0.05% SDS in 6× SSC) is toxic to E. coli, leaving only a few colonies. The hyperosmotic medium partially accounts for this bactericidal effect; bacteria treated with 1× SSC solutions formed significantly more colonies than those treated with 6× SSC.</p><!><p>For assessment of probe delivery agents, it was important to evaluate whether the reagents engender background signals, for example by sequence-independent aggregation of probes with polymers or lipids. To test this possibility, we incubated Q-STAR (green STAR SE) and TPP-DNA probes for 4 h in the presence of the additives (1% w/v) without template DNA and compared the fluorescence level to a control sample of probes without TPP-DNA. All additives with the exception of Pluronic F-68 generated higher fluorescence levels than the control (Figure 4). For SDS and several other detergents (Triton X-100, Tween 20, CHAPS, DPAPS) the level was approximately 2-fold higher than the control; this level of background fluorescence is acceptable for RNA detection in bacteria. In contrast, background fluorescence for PEI (15.5-fold), HDTA (6.8-fold) and STDC (5.8-fold) substantially exceeded the level of the control sample, ruling out these reagents for use with templated probes. The template-independent background fluorescence may arise from condensation of the polyanionic DNA reactive probes with cationic or zwitterionic surfactants.</p><!><p>We next tested whether milder SDS conditions (1× SSC buffer, low SDS concentrations) or alternative delivery agents promote the uptake of oligonucleotide probes into gram negative bacteria (Figure S3, Supporting Information). For this purpose, we incubated E. coli with the reactive probes (green STAR EC and TPP-DNA*) and unlabeled helper oligonucleotides (Helper 1 and 2, Table 1; helper oligonucleotides enhance target accessibility by hybridizing adjacent to the probe binding sites and unwinding the local secondary structure5,15) in the presence of varying concentrations of the additives (for structures see Table S1, Supporting Information). From the tested surfactants, only SDS provided detectable probe uptake as judged by fluorescence signal associated with cells (HTDC and PEI were not tested because of high toxicity and non-specific fluorescence background, respectively). We then evaluated the effect of SDS and SSC buffer concentration on the bacterial staining (Figure S3, Supporting Information). Samples with 6× SSC buffer showed the characteristic green fluorescence staining of E. coli at 0.5% and 0.05% SDS and to a lesser degree at 0.005% SDS. Fluorescence was also observable for the 1× SSC sample with 0.5% SDS but the signal faded considerably with decreasing SDS concentration. Therefore, the nontoxic conditions deliver oligonucleotide probes with only low efficiency into E. coli. For further experiments with Q-STAR probes in bacteria, we used the previously described protocol (6× SSC with 0.05% SDS) with the knowledge that the viability is significantly affected for the majority of the cells.</p><!><p>We tested species-specific Q670-containing NIR STAR probes in combination with fluorescein-containing green STAR probes for two-color discrimination of E. coli and S. enterica (Figure 5). Bacteria were incubated with the Q-STAR and TPP-DNA probes and the helper oligonucleotide probes in hybridization buffer (6× SSC, 0.05% SDS) for 3 h and analyzed by fluorescence microscopy with two filter sets (see Experimental Section). A mixture of NIR STAR SE and green STAR EC probes resulted in a strong and exclusive Q670-fluorescence staining of S. enterica, while E. coli exhibited strong fluorescein signal without detectable Q670-emission (Figure 5a). The inverse staining pattern was present when we reversed the dye labeling of the probe sequences (green STAR SE DAP and NIR STAR EC). Co-incubation of mixed microbes with NIR STAR and green STAR probes provided images of bacteria stained either red or green, enabling the discrimination of single bacterial cells by a fluorescence color call.</p><p>Interestingly, while fluorescein staining for the combined bacteria was selective for the assumed E. coli in the mixture, all bacteria exhibited some degree of Q670-emission (Figure 5a, middle row). Possibly under the given conditions, NIR STAR probes after being activated in a sequence specific manner (after release of the BHQ-2 quencher) can exit the host bacteria and reenter another bacterium.</p><!><p>Next we evaluated the optimized two-color probes for use in flow cytometry. Previously we described the application of a single monocolor probe using simple fluorescence intensity as the flow cytometry readout.15 However, a variety of experimental parameters influence the fluorescence intensity besides the presence and the sequence of the target rRNA, for example efficiency of probe uptake or bacteria aggregation. Thus a ratiometric two-color analysis is desirable and is expected to provide more reliable results.</p><p>To this end, we employed fluorescein- and Q670-labeled probes (green STAR SE DAP and NIR STAR EC) as described for the microscopy experiments. The bacteria were incubated with the reactive probes and helper oligonucleotides in the delivery buffer (6× SSC, 0.05% SDS) and analyzed by two-color flow cytometry. The signal measured in the fluorescein channel was distinctively higher for S. enterica than for E. coli (3.7 ± 0.5-fold) for green STAR SE DAP (Figures 6 and 7). Inversely, the NIR STAR EC probe produced a stronger Q670-fluorescence signal in E. coli relative to S. enterica (6.7 ± 1.6-fold difference). Neither of the probes generated a significant fluorescence signal in the nonspecific detection channel, as expected because of the high spectral resolution of the chosen fluorophore labels. When incubated with a mixture of green STAR SE DAP and NIR STAR EC, bacteria colonies developed distinct fluorescence patterns. E. coli exhibited strong Q670-fluorescence, whereas for S. enterica fluorescein emission was predominant (Figures 6 and 7). The combination of the probes also provided a better discrimination signal than single probe measurements; taking the ratio of Q670- and fluorescein emission provided a 15.7 ± 3.4-fold difference between the two bacterial strains.</p><p>E. coli treated with NIR STAR EC separated into two major subpopulations in flow cytometric analysis, one having approximately 10-fold higher Q670-emission than the other (distinguishable as a shoulder on the histogram in Figure 6, right panel). The fraction of the low-emission population increased with incubation time and such an effect was absent for green STAR SE DAP. This result agrees with the observation of a small population of doubly stained bacteria in mixtures of E. coli and S. enterica and points towards incomplete retention of the NIR STAR probes after quencher release.</p><!><p>The present data demonstrate improved methods for identification of two bacterial pathogens with a two-color pair of templated reductive quencher release probes targeting the 16S rRNA. The results show that bacteria can be successfully discriminated with fluorescence microscopy by a simple fluorescence color-call and, for the first time, by two-color flow cytometry. The performed experiments constitute a significant advance over previous assays targeting rRNAs. Standard FISH probes are simple label-containing oligonucleotide hybridization probes, requiring cell fixation and extensive washing steps. However, the necessity for cell fixation makes the protocol more cumbersome, can produce artifacts, and is incompatible with the long-term goal of culturing bacteria obtained from fluorescence-activated cell sorting. Light-up probes provide a powerful tool to overcome these limitations; for example, MB probes have been applied to the analysis of bacteria RNA with FISH and flow cytometry.11,12 However, those reports used fixed cells and their value for the analysis of intact cells remains to be evaluated. In addition, background fluorescence from nonspecific interactions of MB probes with cellular macromolecules or degradation of the probes complicates analysis.46 Templated reactive probes in general and the described two-color set of probes in particular offer several possible advantages in bacterial RNA targeting. First, the dual probe design minimizes nonspecific activation of reactive probes, as demonstrated here by the minimal fluorescence background observed in microscopy and flow cytometry experiments. Second, the short length of the probes enables excellent sequence specificity, demonstrated by the ability to discriminate two bacteria by a single nucleotide difference. Finally, templated reactive probes can yield isothermally amplified signals, while MB and FIT are stoichiometric at best.</p><p>The current probe set with two distinct fluorophores also offers several advantages over previous templated probe designs that employed either a single color or a FRET pair. Two-color approaches allow for ratiometric analysis of the readout, which in this study provides a significantly better discrimination than a single probe. The two-color ratio further reduces the uncertainty deriving from experimental parameters, such as variations in bacteria size, copy number of rRNAs, and probe delivery efficiency, obviating the necessity of an external standard. During this study, we also noted that under the given experimental conditions, the bacteria tend to aggregate and the E. coli more strongly so than S. enterica, but again the use of a two-color ratio alleviates problems associated with this effect.</p><p>The use of two spectrally separated dyes can be advantageous relative to FRET based probes when using sophisticated equipment such as modern flow cytometers. The current fluorescein and Q670 dyes are bright fluorophores, while FRET-based probes can suffer from low emission quantum yields47 and suboptimal spectral resolution. Previous attempts to analyze bacteria by flow cytometry were hampered by low fluorescence intensity. Nevertheless, the earlier FRET probes may remain advantageous for certain applications with simpler equipment, such as in cases where there is only a single excitation wavelength and a single filter set available.</p><p>A major challenge for in situ hybridization-based experiments with live bacteria remains the difficulty of delivering oligonucleotides into cells. The current protocols enable bacterial analysis by microscopy and flow cytometry without fixation; however, our results show that cells are rendered largely nonviable by the studied delivery agents. While many diagnostic applications require no live bacteria, it may be useful for applications such as sorting and selecting bacteria genetically. Based on the present experiments, it seems possible that conditions for probe delivery and cell viability are orthogonal, likely because the pores induced in the cell wall to allow for probe delivery are also lethal to the cells. In this light, it has been reported that the conjugation of cationic peptides to some oligonucleotide analogues (such as PNAs48 or morpholino DNAs)49 enhances bacterial uptake. Whether such an approach is effective with light-up probes in general or with template reactive probes in particular remains to be determined.</p><p>Finally, the development of the new quencher-linker 3 expands the utility of templated quencher release (Q-STAR) probes. In combination with the previously established dabsyl release linker, the newly developed BHQ-2 (NIR STAR) probes are highly versatile allowing straightforward design of quenched probes with fluorophores emitting in colors ranging from blue to near-infrared. Furthermore, because of the efficiency of the new quencher, NIR STAR probes reach turn-on values exceeding 200-fold, which is significantly better than previous quencher release probes and rivals the highest turn-on ratios reported to date for any template reactive probes.21,30 Such multicolor probes may find applications in multiplex detection of nucleic acids in complex samples.</p>

PubMed Author Manuscript

Novel Extraction Method for Combined Lipid and Metal Speciation From Caenorhabditis elegans With Focus on Iron Redox Status and Lipid Profiling

Parkinson´s disease progression is linked to iron redox status homeostasis via reactive oxygen species (ROS) formation, and lipids are the primary targets of ROS. The determination of iron redox status in vivo is challenging and requires specific extraction methods, which are so far tedious and very time-consuming. We demonstrated a novel, faster, and less laborious extraction method using the chelator ethylene glycol l-bis(β-aminoethyl ether)-N,N,N′,N′-tetra acetic acid (EGTA) as a stabilizing agent and synthetic quartz beads for homogenization under an argon atmosphere. Additionally, we combined the metal extraction with a well-established lipid extraction protocol using methyl-tert-butyl ether (MTBE) to avoid the problems of lipid precipitation in frozen samples and to determine lipid profiles and metal species from the same batch. The nonextractable matrix, such as the debris, is removed by centrifugation and digested to determine the total metal content of the sample as well. Lipid profiling using RP-LC–MS demonstrated high accordance of the modified extraction method to the reference method, and the organic solvent does not affect the iron redox status equilibrium. Furthermore, rigorous testing demonstrated the stability of the iron redox status equilibrium during the extraction process, secured by complexation, inert atmosphere, fast preparation, and immediately deep frozen extracts.

novel_extraction_method_for_combined_lipid_and_metal_speciation_from_caenorhabditis_elegans_with_foc

Introduction<!>Materials<!>C. elegans Culture<!>Extraction According to Matyash et Al<!>Modified Extraction for Metal Species and Lipids<!>Metal Quantification in Aqueous and Organic Worm Extracts and the Pellet<!><!>Lipid Profiling by RP-LC–MS/MS<!>Iron Redox State Speciation<!><!>Iron Redox State Speciation<!>Data Processing and Lipid Annotation<!>Statistical Analysis<!>Lipid Profiling<!><!>Lipid Profiling<!><!>Lipid Profiling<!><!>Lipid Profiling<!><!>Lipid Profiling<!>Metal Extraction<!><!>Iron Redox State Speciation<!><!>Iron Redox State Speciation<!><!>Iron Redox State Speciation<!>Conclusion<!>Data Availability Statement<!>Author Contributions<!>Conflict of Interest<!>Publisher’s Note

<p>Parkinson´s disease is the second most common neurodegenerative disorder, which affects about 1 percent of the population above 65 years. Clinical symptoms occur after about 60 percent of dopaminergic neurons in the substantia nigra are lost (Adams Jr, Chang et al., 2001). Risk factors are commonly connected to genetic background, pesticides and heavy-metal exposure, drug intake, viral infection, or physical damage. Mostly, a combination of different risk factors leads finally to the onset of disease (Adams Jr, Chang et al., 2001; Klein and Westenberger 2012). The underlying mechanisms are not fully understood, but an imbalance in forming and scavenging of reactive oxygen species (ROS) is accepted to play a crucial role in Parkinson´s disease progression (Kolodkin, Sharma et al., 2020). ROS are formed enzymatically by the electron transport chain reaction in the inner mitochondrial membrane or via cytochrome P450 by one-electron reduction of O2 (Quinlan, Perevoshchikova et al., 2013). The formed superoxide anion (O2 - ) is dismutated spontaneously or catalyzed by superoxide dismutase (SOD) to H2O2 (Wang, Branicky et al., 2018). Ferrous iron [Fe (II)] has a crucial role in the formation of the hydroxyl anion (OH−), and the hydroxyl radical (OH ∙ ) forms H2O2 via the Fenton reaction. The OH ∙ is highly reactive and mainly responsible for oxidative damage in lipid membranes (Graf, Mahoney et al., 1984; Gaschler and Stockwell 2017). Disruption in the iron redox state results in formation of Fe (II) in elevated concentration, thus accelerating the formation of the hydroxyl radical. Therefore, the iron redox state and the possibly caused lipid peroxidation in biological systems are of high interest for enlightening cellular mechanisms in Parkinson´s disease progression. The nematode Caenorhabditis elegans (C. elegans) has become a widely used model organism to probe for underlying mechanisms in neurodegenerative disorder progression, since basic cellular mechanisms are well-conserved in the nematode with 60–80% of homologous genes (Kaletta and Hengartner 2006; Leung, Williams et al., 2008). The nervous system of the nematode is relatively simple with around 302 neurons including dopamine (DA), acetylcholine (ACh), gamma-aminobutyric acid (GABA), serotonin (5-HT), glutamate, and others (White, Southgate et al., 1986). Short reproduction cycles of only 3 days allow to generate sufficient biomass for several treatment experiments in a short period. Hence, C. elegans is the perfect organism to establish new analytical methods and enable the enlightenment of basic mechanisms in neurodegenerative disorders (Chakraborty, Bornhorst et al., 2013). However, biological variability regarding iron redox and lipid metabolisms lead to relatively high batch to batch variances, which worsens the significance of the results. A usual but tedious solution to overcome this problem is performing many replicates, sample pooling, and extended comparison with control groups. So far, for each analytical method, separate sample preparation protocols are required increasing the number of samples as well as the amount of time. The extraction and speciation of the iron redox state from biological systems, such as C. elegans, is challenging because of low iron concentration and the sensitivity of Fe (II) to environmental oxygen during extraction. Nevertheless, the extraction method developed by J. Diederich and B. Michalke (Diederich and Michalke 2011) achieves the extraction of native iron and manganese species from the brain and liver tissue by working under the argon atmosphere using a Potter-Elvehjem tissue homogenizer. While the method secures the stability of iron species during extraction, the preparation is time-consuming and the throughput is low, since only one sample can be prepared at the same time. The speciation is carried out by capillary electrophoresis–induced coupled mass spectrometry (CE–ICP-MS) using 20 mM HCl as an electrolyte to prevent iron precipitation or adsorption on the capillary wall (Michalke, Willkommen et al., 2019) as well as by high-pressure liquid chromatography–ICP-MS (HPLC–ICP-MS) (Solovyev, Vinceti et al., 2017). A widely used and accepted extraction method was developed by Matyash et al. (Matyash, Liebisch et al., 2008), which used methyl-tert-butyl ether (MTBE) and methanol as extraction media instead of carcinogenic halogenated solvents. Yields and recoveries were similar to the lipid extraction protocols by Bligh and Dyer (Bligh and Dyer 1959) and Folch (Folch, Lees et al., 1957). Consequently, use of this extraction protocol over the last decade has increased.</p><p>Lipid profiling is carried out with liquid chromatography–MS (LC–MS) with the normal or reversed-phase. The identification of lipids is usually achieved by MS/MS spectra and matching with databases (Hermansson, Uphoff et al., 2005; Sommer, Herscovitz et al., 2006; Yetukuri, Katajamaa et al., 2007). Until now, no combined extraction protocol for ion redox speciation and lipid analysis has been published. However, there is an eminent need since both are strongly interlinked. Here, we report a combined extraction protocol, which allows simultaneous extraction of metal species. We particularly focused on iron redox state and lipid extraction in increased throughput. This extraction protocol is currently the missing tool for monitoring the lipidome of the same nematode population, specifically the oxidized lipids and iron redox state homeostasis. Previously mentioned limitations such as batch to batch variations and the amount of time spent on extraction are highly reduced. Additionally, the problematic of precipitating lipids and proteins in the metal containing aqueous phase after freezing and thawing, by separating lipids and metal species forehand will be avoided.</p><!><p>Ethylene glycol l-bis(β-aminoethyl ether)-N,N,N′,N′-tetra acetic acid (EGTA) tetra sodium salt was purchased from Santa Cruz Biotechnology, Inc. (Dallas, United States). 3-morpholinopropane-1-sulfonic acid (MOPS) from MP Biomedicals (Illkirch, France) and sodium chloride (NaCl) of analytical grade from Fluka® Analytical (Munich, Germany) were purchased. Agar granulate was ordered from BD Diagnostics (Franklin Lakes, United States). MTBE, acetonitrile (ACN), formic acid (Fa), and methanol (MeOH) of LC–MS grade were purchased from Sigma-Aldrich (Darmstadt, Germany). Butylhydroxytoluol (BHT) and cetyltrimethylammonium bromide (CTAB) were ordered from Th. Geyer GmbH and Co. KG (Renningen, Germany). Synthetic quartz beads (0.315–0.5 mm) were purchased from Gaßner Glastechnik GmbH (Oberhaching, Germany). HNO3 was purchased from Carl Roth GmbH und Co. KG (Karlsruhe, Germany) and purified by sub-boiling distillation. The metal extraction buffer (MEB) was prepared by dissolving exactly weighted respective amounts of MOPS, NaCl, and EGTA in MiliQ® water to achieve the final concentration of 5, 6, and 1 mM, respectively. Fe (II) and Fe (III) standard solutions were prepared by dissolving 1 g/L FeCl2 in 100 mM HCl and FeCl3 in 1 mM HCl, respectively. Stock solution was aliquoted, kept under argon at −20°C, and were thawed and diluted shortly prior to usage. The dilution was carried out in the MEB.</p><!><p>The wild-type N2 strain of C. elegans was maintained and handled at 20°C according to previously published protocol (Brenner 1974). Escherichia coli OP50 (E. coli), cultivated in the Luria–Bertani medium (LB-medium), was used as a food source. Worms were bleached for age synchronization with basic hypochlorite solution as described previously by WB (WB 1988). Hatched L1 was fed with E. coli for 2 hours, washed two times with the 10 ml M9 buffer (22 mM KH2PO4, 22 mM Na2HPO4, 85 mM NaCl, and 1 mM MgSO4), incubated for 20 min in M9, and washed three times with 10 ml Milli-Q® water. An amount of 100 k L1 was aliquoted, the liquid was aspirated, and the samples were frozen in liquid nitrogen and stored at -80°C until extraction.</p><!><p>For the lipid extraction, 50 µl MeOH was added to a pellet corresponding to 100 k L1 larvae and mixed. The suspension was transferred to 2-ml beading tubes filled with 200 mg quartz beads. This step was repeated once to transfer the pellet quantitatively. Then, 300 µl MTBE was added, the lids were closed, and the samples were incubated at 25°C for 1 h. By adding 100 µl water and incubating for 10 min at room temperature (RT), the phase separation was induced. The upper organic phase was collected after centrifugation at 10.000 g for 15 min, and the lower phase was reextracted with 200 µl MTBE/MeOH/H2O (10:3:2.5, v/v/v). Additionally, 50 µl BHT solution (58.5 mM in MeOH) was added to the combined organic phases. The combined organic phase was dried in a vacuum centrifuge for 30 min. The pellet was dissolved in isopropanol/acetonitrile/water (6:3.5:0.5, v/v/v) for LC–MS measurement.</p><!><p>All buffers and solvents were prepared freshly and stored in the dark at 4°C until use. Before extraction, all buffers and solvents were degassed, and additionally, all buffers, solutions, and vials were flushed with argon for 1 h. Extraction was carried out by adding 100 µl MEB/Methanol (2:1, v/v) and 50 µl BHT solution (58.5 mM in MeOH) to a pellet corresponding to 100 k L1 larvae and mixed. The suspension was transferred to 2-ml beading tubes filled with 200 mg quartz beads. This step was repeated once to transfer the pellet quantitatively. Then, 300 µl MTBE was added; the samples were overlaid with Ar and stored on ice, lids were closed, and the sample was homogenized at −10°C at 6,000 rpm with 3 × 20 and 30 s pause, using a Precellys bead tube homogenizer, equipped with a cryo unit. Phase separation was induced by adding 100 µl MEB and incubated for 5 min at RT and centrifuged at 10.000 g for 10 min. The upper organic phase was collected, and the lower phase was reextracted with 200 µL MTBE/MeOH/MEB (10:3:2.5, v/v/v). The aqueous phases were carefully aspirated and immediately frozen in liquid nitrogen, whereas 300 µl MEB was added to the pellet. After mixing and centrifugation at 10.000 g for 5 min, the aqueous phase was aspirated, combined with the previous one, and frozen in liquid nitrogen. The frozen samples were overlaid with Ar and stored in liquid nitrogen. The combined organic phase was dried in a vacuum centrifuge for 30 min. The pellet was dissolved in isopropanol/acetonitrile/water (6:3.5:0.5, v/v/v) for LC–MS measurement.</p><!><p>400 µl aqueous and organic phases were diluted with 2% HNO3 up to 4 ml. The worm pellet and the quartz beads were transferred to 10-ml microwave digestion tubes using 1 ml mixture of subboiled HNO3, 30% H2O2, and water (1:1:2, v/v/v) and digested in total for 20 min (5 min heat up, 10 min at 200°C and 27 bar, and 5 min cooling down). The digested samples were diluted up to 10 ml with 2% subboiled HNO3, mixed with three pipette strokes, and 5 ml was aspirated for measurement, after quartz beads have settled down. The quantification was carried out with inductively coupled plasma atomic emissions spectroscopy (Optima 7300 DV, PerkinElmer, Toronto, Canada) for the elements Fe, Al, Cu, Mn, and Zn (Table 1). Control (QC) and blank measurements were performed periodically after 10 measurements as well as before and after measuring the samples with the high organic content.</p><!><p>Lines used for ICP-AES determination of selected elements.</p><!><p>Lipid profiling was performed on a Sciex ExionLC AD system coupled to a Sciex X500R QToF-MS (MDS Sciex, Concord, Canada) equipped with an electron spray ionization source (ESI) operated in the negative and positive mode. Lipid separation was achieved on a Waters CORTECS UPLC C18 column (150 mm × 2.1 mm ID, 1.6 µm particle size) using a linear gradient eluant A (40% H2O, 60% ACN, 0.1% NH4FA) to eluant B (90% isopropanol, 10% ACN, 0.1%, NH4Fa) with a flow rate of 0.25 ml/min and column temperature of 40°C (Witting, Maier et al., 2014).</p><p>The MS was calibrated daily and every six samples to maintain high mass accuracy and operated in information dependent acquisition (IDA) using an ionization voltage of -4.5 and 3.8 kV. The m/z range was set from 50 to 1,500 Da. MS/MS information was acquired with IDA selecting 10 precursors and a collision energy of −35 and 35 eV with a collision energy spread of 15 eV.</p><!><p>The aqueous phase was thawed, mixed, transferred to 300-µl sample vials, and capped. The experimental setup, including the interface between capillary electrophoresis (CE-700, prince technologies, Emmen, Netherlands) and ICP-MS (NexION 300 D, PerkinElmer, Toronto, Canada), was arranged as previously described (Michalke, Willkommen et al., 2019). An uncoated fused silica capillary with the dimensions of 100 cm × 50 µm inner diameter (ID) was used. The electrolyte was 50 mM MOPS (pH 6.2) with 0.5 mM CTAB. The capillary voltage was set to -15 kV, with an external pressure of 700 mbar to induce an additional flow. Before each run, the capillary was flushed at 4 bar with 10% HCl for 1 min, 0.5 M NaOH for 1 min, MiliQ water for 2 min, and the background electrolyte (BGE) for 5 min. Ammonium acetate in the concentration of 20 mM was used as a sheath liquid (Table 2).</p><!><p>Conditions for capillary electrophoresis and Fe (II)/Fe (III) Speciation.</p><!><p>The ICP-MS was used in the dynamic reaction cell (DRC) mode with ammonia as a reaction gas. The isotopes 56Fe and 57Fe were measured. The flow parameters of lenses, collision cell, carrier gas, plasma gas, and auxiliary gas were tuned for maximum sensitivity, low oxide ratio of <1.0% (140CE16O/140CE+), and double-charged ratio <1.5% (140Ce++/140Ce+) with background counts of <0.1 cps. The RF power was set to 1250 W, the plasma gas was set to 16 L Ar/min, the nebulizer gas was set to 0.94 L Ar/min, the dwell time was set to 50 ms, and the ammonia gas flow was set to 0.58 ml NH3/min. Operating software for MS was Syngistix from PerkinElmer (Toronto, Canada) and for CE was DAx-3D from prince (Emmen, Netherlands).</p><!><p>RP-LC–MS data were reprocessed using Genedata Expressionist for MS 15.5 (Genedata AG, Basel, Switzerland), which included chemical noise subtraction, peak detection, isotope and adduct clustering, and MS/MS peak detection. MS1 data were exported to .xlsx for further processing. MS2 data were exported as .mgf files and further processed in R using an in-house annotation workflow (https://github.com/michaelwitting/MetaboAnnotationGenedata). Lipids were putatively annotated by matching m/z values against an in-house C. elegans database and LipidMaps CompDB (Fahy, Subramaniam et al., 2009) as well as MS2 spectra data against LipidBlast (Kind, Liu et al., 2013) and an in-house C. elegans lipid database. Annotations were manually verified, and only verified annotations were reported. Any further data processing and generating graphs were carried out with the open source tool RStudio (Version February 1, 1335, RStudio, Inc.); and packages: ggplot2, ggpubr, openxlsx, reshape2, gridExtra, scales, grid, plyr, dplyr and tidyverse, mcr, and outliers.</p><!><p>For comparing two methods, we used the statistical test according to Pearson with p values below 0.05 considered to be statistically significant (Benesty, Chen et al., 2009). The Passing and Bablok regression was performed to compare the intensities of features for both methods with a confidence interval (CI) of 95% for the intercept and the slope (Passing and Bablok 1983). The error bars in bar plots present the standard error of the mean (SEM) obtained from three measurements. The confidence area was calculated according to the following equation (Eq. 1) with the number of samples (n), the student factor for n variables and confidence probability of 0.95 (t), and the standard deviation (s): μu/d= x¯±t sn (1)</p><p>For the lipid profiling, only annotations identified in all triplicate and with a coefficient of variation (CV) <30% were accepted as valid features and were used for further statistics. The intensities of features in one lipid class were tested for normal distribution using the Shapiro–Wilk test. In case of normal distribution, the features of both extraction methods were compared using the independent two-tailed Student´s t-test (Student 1908).</p><!><p>The lipidome of C. elegans includes a wide range of different lipid classes and sub classes such as fatty acids and amides, glycerophospholipids, sphingolipids, glycerolipids, steroids and related substances, glycolipids, prenol lipids, and additionally small classes (Witting and Schmitt-Kopplin 2016). The glycerophospholipids are the main components of biological membranes and therefore of high interest for determining potential cell damages caused by reactive oxygen species (ROS), whose alteration is typical for Parkinson's disease progression (Kolodkin, Sharma et al., 2020). For the determination of changes in the lipid profile of the C. elegans, it is mandatory that the combined extraction method extracts the main lipid classes as good as the reference method (Matyash, Liebisch et al., 2008). Therefore, we compared the total number of annotated lipids and their intensities of selected lipid classes, with predominant focus on the glycerophospholipids: phosphatidylcholine (PC), lysophosphatidylcholine (LPC), phosphatidylethanolamine (PE), ether-linked PCs (PC-O) and PEs (PE-O), lysophosphatidylethanolamine (LPE), phosphatidylinositol (PI), phosphatidylglycerol (PG), phosphatidylserine (PS), diglyceride (DG), triacylglyceride (TAG), and other lipid classes.</p><p>In total, 849 lipid features were detected. A total of 822 in Matyash and 845 in the combined extraction protocol were detected. From these, 818 were found in both. The number of annotations based on MS2 data in the positive mode was 525 for the modified version and 505 for the reference method. The predominant species are the TAGs with 243 and 231 annotations, followed by the PCs with 125 and 123 annotations, and 63 and 47 annotations for the PEs. PC-Os, PEOs, LPEs, and other annotations are <16 for each of the classes (Table 3). The differences in the number of annotated lipids are considered as marginal, and the modified protocol has more annotations in total as well as more or the same annotations in the sub groups.</p><!><p>Number of annotations in the positive mode for lysophosphatidylcholine (LPC), phosphatidylcholine (PC), ether-linked PCs (PC-O), phosphatidylethanolamine (PE), ether-linked PEs (PE-O), phosphatidylinositol (PI), diglyceride (DG), triacylglyceride (TAG), and other lipid classes.</p><!><p>In the negative mode, the total annotations for the combined extraction method were 320 and for the reference method were 317. The predominant species are PEs with 106 and PCs with 94 annotations. Three PSs, which were annotated for the combined extraction method, could not be annotated for the reference extraction because of intensities lower than the threshold of 200 (Table 4). Conclusively, the comparison of the number of annotations, in the positive and negative ionization mode, shows no significance difference between the reference and the modified protocol. The marginal changes in the positive mode in the number of annotations might be explained by differences in the intensity, which lead to undergoing the threshold and conclusively no annotation. Therefore, we compared the absolute intensity for the annotated lipids for both extraction methods.</p><!><p>Number of annotations in the negative mode for lysophosphatidylcholine (LPC), phosphatidylcholine (PC), ether-linked PCs (PC-O), phosphatidylethanolamine (PE), ether-linked PEs (PE-O), lysophosphatidylethanolamine (LPE), phosphatidylinositol (PI), phosphatidylserine (PS), phosphatidylglycerol (PG) and other lipid classes.</p><!><p>Nonvolatile or less volatile compounds, such as salts, ion-pairing agents, endogenous compounds, drugs, and metabolites, change the electron spray droplet properties, which results in ion suppression (Thomas 2003). Therefore, we studied the effect of the MEB and the shortened incubation time on the intensity of the features. The range of intensities was from 200 to 1 × 107 counts, which are multiple orders of magnitude. Consequently, we applied the logarithmic intensities against the lipid class (Figure 1). Interestingly, the average intensities for all lipid classes were higher in the modified extraction method for the positive mode and without noticeable differences in the negative mode. We further applied Student´s t-test for the intensities and the coefficient of variation within one lipid class. The differences between the intensities within the lipid classes were not significant except for the PCs in the positive mode (p < 0.05). Differences in the coefficient of variation were significant for DGs, PCs, PEs, PE-Os, TGs, and others in the positive mode and PE-Os, PGs, PCs, and PEs in negative mode. The coefficients of variation for all of those were lower for the combined method than those of the reference. Basically, this indicates a higher specificity for the combined extraction method but not necessarily a higher efficiency. But, it was presumed that the homogenization with glass beads for the modified protocol would affect the extraction efficiency, by supporting the disruption of the epidermis of C. elegans. This effect seems to be more significant than the longer incubation time of 5 min than 1 h in the reference method. This is also the main difference that affects the CV positively. In addition, our results indicate that the MEB is not affecting the intensity of annotated features negatively.</p><!><p>Intensities of chosen lipid classes generated in positive (A) and negative (B) modes: diglyceride (DG), lysophosphatidylethanolamine (LPE), phosphatidylcholine (PC), phosphatidylethanolamine (PE), phosphatidylinositol (PI), phosphatidylglycerol (PG), phosphatidylserine (PS), sphingomyelin (SM), triacylglyceride (TAG), and other lipid classes. The intensities of identified lipids are represented as averages from triplicate determination.</p><!><p>The comparison with the reference method for the number of annotations shows high agreement, as discussed before. Additionally, the intensities of all annotated lipids show a high level of positive agreement with p values <0.001 and Pearson correlation coefficients being 0.94 in the negative mode and 0.98 in the positive mode. The Passing and Bablok regression shows no significant difference between both methods (CI: 95%) regarding the slope and intercept in the negative mode, and no significant difference for the slope in the positive mode. In the positive mode, we obtained a significant shift to higher intensities for the combined method (Figure 2A,B). Nevertheless, the modifications made to the extraction method of Matyash et al. showed no negative effect on the lipid extraction efficiency and no effect on the lipid variety for the selected lipid classes. Therefore, a negative effect of the MEB on the extraction efficiency and variety can be excluded. Additionally, the modified method reduces the time spent on the extraction protocol and opens the possibility of simultaneous extraction of multiple samples. Our results clearly demonstrate that the incubation of 1 h in MeOH/MTBE, which is described in Matyash et al., is not necessary if a homogenization technique such as the cryo bead milling is used. The argon atmosphere in combination with BHT acts as a scavenger for oxygen radicals, and the low temperature prevents lipid oxidation during homogenization, which is necessary for the study of oxidized lipid species in vivo (Fujisawa, Kadoma et al., 2004). Furthermore, quartz beads are applied for the use in lipid extraction from C. elegans, which in this context has not been described before. Consequently, the modified extraction protocol is valid for the use of lipid extraction.</p><!><p>Passing–Bablok regression fits for the logarithmic intensities of identified lipids in negative (A) and in positive (B) modes. The dashed red line represents the half line with a slope of one and in blue is the the confidence band of the Passing and Bablok regression with an α of 0.05.</p><!><p>Even though, the modifications made to the extraction protocol including the use of the cryo unit and the additional steps to extract the aqueous phase are increasing the costs of the method and the amount of laboratory time are almost the same compared to Matyash et al. and 48 times less for the metal extraction. The modifications made are needed to keep the redox iron status stable and are shown to have no significant effect on the extraction efficiency compared to the reference method. Conclusively, the additional effort is necessary for the combined extraction of lipids and the metal species of interest.</p><!><p>Based on the polar character of metal ions and EGTA complexes, the ions will sustain in the aqueous phase. Complexation with EGTA increases this effect and stabilizes the redox species Fe (II), which is susceptible to oxidation (Wilson and Carbonaro 2011). In addition, adsorption of Fe (III) on the surface of the quartz beads is minimized through complexing with EGTA by electrostatic repulsion. The determination of iron concentration in the organic phase with ICP-OES results in concentration level below the limit of quantification (LOQ) of 1 μg/L, whereas the aqueous extracts contain 18 ± 1.7 μg/L, which confirms that the aqueous phase is the metal enriched phase. The recovery of the extraction method was calculated by dividing the iron content of the extract by the total iron content, which is the sum of the extracts and digested pellet content, and multiplication with 100%. Therefore, we determined the metal concentration in the organic phase, the aqueous phase, and in the pellet–quartz beads mix after microwave digestion for 10 replicates. We achieved an iron recovery of 62 ± 8.2%, which is limited by remaining liquid in contact with the quartz beads and inclusions of metal in the remaining debris and precipitating proteins. If we take into account that we washed the pellet only one time after extraction to minimize the time at room temperature in the oxygen atmosphere, in respect of the susceptibility of Fe (II) to oxidation, this recovery is very acceptable. In addition, we determined the metal content in the aqueous phase and pellet for Al, Zn, Mn, and Cu in respect of their important function or toxicity in neurological disease progression (Figure 3) (Kozlowski, Luczkowski et al., 2012; Page, White et al., 2012; Guilarte 2013; Szewczyk 2013). Mn shows the highest recovery with 96% followed by Zn (80%), Fe (62%), Cu (47%), and Al (25%). The dependence on the recovery of the metal might be explained by the diversity of metal species. This extraction method is optimized for free ions, low molecular weight, and polar metal species. The membrane proteins such as cytochrome p450 oxidase (Šrejber, Navrátilová et al., 2018) will be partly removed by centrifugation, if the membrane disruption is incomplete. In addition, included or adsorbed metals in the debris and precipitating proteins due to the MeOH and MTBE in the extraction media will decrease the total recovery. However, the recovery of iron is acceptable, since we are interested in the free metal and low molecular metal species. For Mn, Zn, Al, and Cu, the recovery is very promising, and even though we optimized the protocol only for selected Fe and Mn species, these findings open new perspectives for future applications of this extraction protocol regarding speciation of metals other than Fe.</p><!><p>Obtained recoveries for extracted metals: Fe, Mn, Al, Cu, and Zn.</p><!><p>The sensitivity of Fe (II) to surrounding or solved oxygen is the major problem in iron redox state speciation. Under inappropriate storage conditions such as room temperature (RT), neutral pH, and air contact, Fe (II) oxidizes within minutes (Michalke, Willkommen et al., 2019). Therefore, it is required to exclude oxygen during sample preparation by the argon or nitrogen atmosphere and keep the pH low. By flushing all solvents with Ar and overlaying the caps with Ar or N2, the excluding of oxygen during extraction was achieved. In order to extract the sample close to the physiological pH (intracellular: 7–7.4) to keep metal species close to the native environment, EGTA and BHT were added, which have a stabilizing effect on the Fe (II) species as demonstrated previously (Wilson and Carbonaro 2011). The complexation of Fe (II) and Fe (III) with EGTA has two effects. First, it inhibits the electron transfer by sterical stabilization, and second it prevents Fe (III) to form hydroxyl complexes, which accelerate the oxidation of Fe (II) (Morgan and Lahav 2007). In contrast to ethylenediaminetetraacetic acid (EDTA), which acts like a catalyst for the Haber–Weiss reaction, EGTA showed no catalytic activity (Sutton 1985). Additionally, EGTA prevents the iron species from adsorbing at the surface of the quartz beads and capillary wall by electrostatical repulsion of the negatively charged complex Fe-EGTA-/2-. The adsorption to the capillary wall occurs through the high specific surface area of about 8 m2 l−1 for a capillary of 100 cm length and 50 µm ID, which we identified as a major problem in iron redox state speciation. Additionally, the dynamic coating with CTAB and reversed current prevent adsorption of Fe (II) and Fe (III) to free silanol groups on the capillary surface. The BHT scavenges free radicals, such as the hydroxyl radical, which would alter the iron redox state equilibrium by Haber–Weiss reaction as mentioned before (Black 2002). The extraction media components such as EGTA, NaCl, and MOPS as well as the snap-freezing do not alter the iron redox status as demonstrated before (Diederich and Michalke 2011; Wilson and Carbonaro 2011). However, organic solvents during the extraction and remaining organic solvents in the aqueous phase might cause an alteration of the iron redox status during extraction or have an effect on the iron determination with CE–ICP-MS. Therefore, we studied the effect of MeOH and MTBE remaining in the aqueous phase on the Fe (II)/Fe (III) ratio compared with an extract containing only the MEB. Accordingly, the extraction was performed five times using a mixed culture of C. elegans as a model organism, following our modified extraction without adding organic solvents and straight according to the protocol. We determined the area under the curve (AUC) for Fe (II) and Fe (III), which we identified by standard addition. The observed average ratios were 0.27 ± 0.06 for the MTBE free extraction and 0.25 ± 0.11 for the extract with MTBE (Figure 4A).</p><!><p>Fe (II)/Fe (III) ratios for the modified extraction protocol with and without traces of methyl-tert-butyl ether (MTBE) to study the effect of MTBE/MeOH on the iron redox state equilibrium (A) and Fe (II)/Fe (III) ratios for a 5-mg/L Fe (II) and Fe (III) standard solution in the MEB after following the extraction protocol compared to the standard solution (B). The determination was carried out by CE–ICP-MS.</p><!><p>Lipid precipitation was observed for the extraction without organic solvents, which might lead to accelerated capillary alteration and therefore should be avoided (Dawod, Arvin et al., 2017). The modified homogenization technique using the glass beads and high frequency of shaking might alter the iron redox state. Therefore, we proved the stability of iron redox state during extraction by determining the iron redox state equilibrium of five 5 mg/L standard solutions Fe (II)/Fe (III) in the MEB after following the extraction protocol and comparing it to five 5 mg/L standard solutions before extraction. The samples and control were stored under the same condition and time under argon, frozen in liquid nitrogen for transport, and were measured directly after the extraction, to prevent time-related oxidation of Fe (II). The Fe (II)/Fe (III) ratio for the standard solutions after extraction was 0.49 ± 0.02 and for the control 0.52 ± 0.04 (Figure 4B). Consequently, we proved the stability of the iron species during the extraction and showed that MTBE and MeOH do not affect the extraction of the iron species or interfere the determination with CE–ICP-MS. Additionally, a slight change in the peak shape was noticed, which was caused by the remaining MTBE and MeOH in the aqueous phase toward more narrow peaks (Figure 5). This positive effect improved the resolution of performed measurements. However, a decrease in the AUC was noticed, which is explained by the narrowing peak width, while the peak height was constant. The lower integrated signal might explain the higher variation of the Fe (II)/Fe (III) ratio, since this is proportional to the amount of data points generated for each peak. The limit of generated data points was set by the dwell time of 50 ms and the number of measured isotopes (Fe-56 and Fe-57). A reduction in the dwell time might result in less variation, but it will also decrease the intensity and LODs. For our purpose, we already optimized this parameter regarding variation and intensity, but this will differ depending on the metal concentration in the sample.</p><!><p>Electropherogram of the extracts obtained according to the modified protocol with (blue) and without (red) organic solvent.The electropherogram was obtained after smoothing the data by applying the Savitzky–Golay filter. The determined isotope was Fe-56 in the DRC-ammonia mode.</p><!><p>Traces of remaining oxidative species, such as H2O2, formed by UV-C radiation (Piskarev 2018), might alter the iron species, and we noticed problems to keep the iron species stable, especially in the standard solutions. This, we solved by using low temperature, freshly prepared buffer storage in the dark, BHT as an antioxidant, EGTA as a complexing agent, Ar atmosphere, and short extraction times. The final composition of 6 mM NaCl, 5 mM MOPS, and 1 mM EGTA in the MEB was based on the previous work (Wilson and Carbonaro 2011) and slightly modified to match the ionic strength of the electrolyte used in the CE determination. In summary, the extraction media are stabilizing the labile iron redox status and optimizing the performance in the following determination process. Interestingly, the biological matrix seemed to stabilize the iron species. The reason for that observation is likely the cellular ROS defense mechanisms, such as superoxide dismutase in combination with catalase and glutathione peroxidase and glutathion (GSH) as an electron donor (Hanukoglu 2006). Additionally, GSH forms Fe (II)-GSH complexes at a physiological pH, which are discussed to be responsible for the stabilization of free iron (II) in the labile iron pool (Hider and Kong 2011). This gives some evidences why the iron redox state in biological samples is more stable than the iron standard in the MEB. In summary, the main differences to the extraction protocol of Matyash et al. (Matyash, Liebisch et al., 2008) are the extraction solvents, the liquid nitrogen freezing, the use of a cryo unit, and glass bead homogenizer as well as Ar-flushed solvents. These modifications are essential for the stability of the redox iron status during the extraction and the performance of the following determination. Additionally, the extraction time is reduced by a factor of 48 compared to the existing protocol (Diederich and Michalke 2011), since up to 24 samples can be homogenized simultaneously in 5 min compared to only one sample using a glass Dounce homogenizer under the Ar atmosphere in 10 min.</p><!><p>We have successfully established a modification of the lipid extraction protocol previously described from Matyash et al. (Matyash, Liebisch et al., 2008) for the new application in iron redox state speciation for the model organism C. elegans. The results showed similar extraction efficiencies compared to the reference method for the important lipid classes. We reduced the laboratory time spent on the extraction and opened up the opportunity to extract multiple samples simultaneously for the extraction of the labile iron redox equilibrium. Using this combined protocol, we enabled the extraction of lipid species as well as free iron and other metals simultaneously. Additionally, lipid and protein precipitation in the aqueous phase as well as associated alteration of the capillary, caused by protein adsorption on the silica surface, is prevented. We reduced the alteration of the iron species during extraction, with the combined effect of the MEB buffer, argon atmosphere, BHT, and liquid nitrogen cooling. Consequently, the determination of free metal species and the ratio between iron (II) and iron (III) together with lipid species from the same sample was successfully established.</p><!><p>The raw data supporting the conclusions of this article will be made available by the authors, without undue reservation.</p><!><p>BB designed and performed the experiments including sample preparation and measurements as well as the data processing, creation of graphics, and writing of the manuscript. MW performed the lipid annotation with RStudio and largely contributed to the introduction and the lipid profiling parts of this manuscript. BM and PS-K provided the laboratories and used systems as well as advising and directing during the project development. All authors contributed to the manuscript by proofreading.</p><!><p>The authors declare that the research was conducted in the absence of any commercial or financial relationships that could be construed as a potential conflict of interest.</p><!><p>All claims expressed in this article are solely those of the authors and do not necessarily represent those of their affiliated organizations, or those of the publisher, the editors, and the reviewers. Any product that may be evaluated in this article, or claim that may be made by its manufacturer, is not guaranteed or endorsed by the publisher.</p>

PubMed Open Access

Pathogenic Troponin T Mutants with Opposing Effects on Myofilament Ca2+ Sensitivity Attenuate Cardiomyopathy Phenotypes in Mice

Mutations in cardiac troponin T (TnT) associated with hypertrophic cardiomyopathy generally lead to an increase in the Ca2+ sensitivity of contraction and susceptibility to arrhythmias. In contrast, TnT mutations linked to dilated cardiomyopathy decrease the Ca2+ sensitivity of contraction. Here we tested the hypothesis that two TnT disease mutations with opposite effects on myofilament Ca2+ sensitivity can attenuate each other\xe2\x80\x99s phenotype. We crossed transgenic mice expressing the HCM TnT-I79N mutation (I79N) with a DCM knock-in mouse model carrying the heterozygous TnT-R141W mutation (HET). The results of the Ca2+ sensitivity in skinned cardiac muscle preparations ranked from highest to lowest were as follow: I79N > I79N/HET > NTg > HET. Echocardiographic measurements revealed an improvement in hemodynamic parameters in I79N/HET compared to I79N and normalization of left ventricular dimensions and volumes compared to both I79N and HET. Ex vivo testing showed that the I79N/HET mouse hearts had reduced arrhythmia susceptibility compared to I79N mice. These results suggest that two disease mutations in TnT that have opposite effects on the myofilament Ca2+ sensitivity can paradoxically ameliorate each other\xe2\x80\x99s disease phenotype. Normalizing myofilament Ca2+ sensitivity may be a promising new treatment approach for a variety of diseases.

pathogenic_troponin_t_mutants_with_opposing_effects_on_myofilament_ca2+_sensitivity_attenuate_cardio

INTRODUCTION<!>Mice \xe2\x80\x93<!>Cardiac Skinned Preparations \xe2\x80\x93<!>Echocardiography (ECHO) Measurements \xe2\x80\x93<!>Isolated heart perfusion and arrhythmia measurements \xe2\x80\x93<!>Histopathology \xe2\x80\x93<!>Gross morphology \xe2\x80\x93<!>Statistical Analysis \xe2\x80\x93<!>Myofilament Ca2+ sensitivity \xe2\x80\x93<!>Heart Size and Histopathology \xe2\x80\x93<!>Echocardiography \xe2\x80\x93<!>Ventricular Arrhythmias \xe2\x80\x93<!>QRS Complex \xe2\x80\x93<!>Discussion

<p>Hypertrophic cardiomyopathy (HCM) is the leading cause of sudden cardiac death in young adults. HCM affects approximately 1 in 500 individuals who exhibit a wide range of symptoms including diastolic dysfunction, fibrotic tissue formation and impaired contractile properties [1, 2]. Heart failure and arrhythmias are clinical staples of this disease [3], with the pathology of HCM including left ventricular (LV) hypertrophy as a tell-tale sign [2, 4]. In contrast, primary dilated cardiomyopathy (DCM) exhibits systolic dysfunction and LV chamber dilation. Interstitial fibrosis and abnormal contractile properties are also present [5]. Currently, there are no effective pharmacological treatments for either disease.</p><p>The contractile properties of the heart are primarily dictated by the myofilament proteins in conjunction with the sarcoplasmic reticulum [6]. The turnover and recruitment of actomyosin cross-bridges as well as Ca2+ binding to, and dissociation from the thin filament can determine the kinetics and force of contraction [6]. The troponin complex, a crucial Ca2+ regulatory nexus for cardiac muscle contraction, is thought to play a role in both diseases. Located on the actin filament, the troponin complex consists of the subunits Troponin T (TnT), Troponin I (TnI), and Troponin C (TnC). Together with tropomyosin (Tm), the complex acts as a "molecular switch" to expose myosin-to-actin interaction sites upon Ca2+ binding to TnC [7]. This allows thick and thin filaments to slide past each other causing the sarcomere to shorten [6]. Mutations in the Tn subunits can cause either primary HCM or primary DCM. Evidence for the deleterious effect of Tn mutations comes from in-vitro and animal models that show abnormalities in Ca2+ sensitivity of contraction, maximal force development, regulation of actomyosin ATPase activity, Ca2+ handling proteins, contractile transients, among others [8].</p><p>Among the Tn subunits, human disease mutations occur most commonly in TnT [8]. Interestingly, mutations in TnT can result in either HCM or DCM, depending on whether the mutation increases (HCM) or decreases (DCM) myofilament Ca2+ sensitivity. Prior studies have looked specifically at the TnT I79N mutant, previously established as a human cardiac mutation found in HCM patients. Tested in mouse, the mutation leads to an increase in myofilament Ca2+ sensitivity without causing hypertrophy [9]. Our group has suggested that the Ca2+ sensitized myofilament is the primary reason for the high risk of arrhythmia caused by the TnT-I79N mutation [10]. Ventricular arrhythmia and sudden death are also a primary disease manifestation in HCM patients with TnT mutations [11]. Conversely, the missense TnT mutation R141W is associated with DCM in humans and mice [12–14]. This DCM associated mutation has been shown to induce Ca2+ desensitization of the myofilament [15–18].</p><p>To test the hypothesis that abnormal myofilament Ca2+ sensitivity is a primary cause for the pathogenicity of TnT mutations, here we investigated whether two pathogenic TnT mutations that cause opposite effects on myofilament Ca2+ sensitivity can rescue each other's cardiomyopathy phenotype. Transgenic mice expressing the HCM TnT-I79N mutation (I79N) were crossed with a knock-in (KI) mouse expressing the DCM TnT-R141W mutation (HET). The new double-transgenic mice (I79N/HET) were subsequently studied for cardiac myofilament function as well as heart morphology and physiology. The I79N/HET mice exhibited an overall normalization of myofilament Ca2+ sensitivity, left ventricular dimensions and reduction in susceptibility to arrhythmias. These results suggest an independent role of myofilament Ca2+ sensitivity in the development of morphological and functional changes within the diseased heart.</p><!><p>All protocols and experimental procedures followed NIH guidelines and were approved by both Vanderbilt University and Florida State University Animal Care and Use Committee (ACUC). Transgenic mice expressing the HCM TnT-I79N mutation (replacing ~ 50% of the endogenous cTnT-WT, line 8) were crossed with a DCM knock-in TnT-R141W (~ 60% of total is the mutant mouse cardiac TnT transcripts). The founders were heterozygous for the TnT- R141W mutation and overexpressed the TnT-I79N mutation in the heart. Although we have not measured relative TnT expression levels in the double crossed mice (I79N/HET), functional findings indicate that both mutant TnT proteins should be present in the heart in addition to one allele that expresses the mouse cardiac TnT-WT. Non-transgenic (NTg), TnT-I79N transgenic (I79N) and Knock-in TnT-R141W heterozygote mice (HET) were used as controls. The I79N mice were maintained in B6/SJLF1 background strain, while the HET mice were originally generated in 129/SvEv background strain. The I79N/HET is therefore a hybrid background strain. The generation and characterization of the I79N and HET mouse models have been previously published [9, 12].</p><!><p>Skinned cardiac preparations were prepared according to established protocols [10, 19]. Papillary muscle preparations were isolated from the left ventricles of 15–36 week old male and female mice. The membranes were skinned using 1% Triton X-100, rinsed and then stored in a relaxing solution (pCa 8.0 = 10−8 M free Ca2+, 150 mM ionic strength, 2.5 mM MgATP2-) containing 50% glycerol (v/v) at −20°C. Skinned cardiac preparations were mounted onto a force transducer and immersed in pCa 8.0 solution. Muscle preparations were stretched at pCa 8.0 by 20% over slack length. The Ca2+ dependence of tension was measured in a series of solutions with increasing Ca2+ concentrations at room temperature (~20°C). Methods for solving the free and bound metal ion equilibria in our pCa solutions were provided by the computer program, pCa Calculator [20].</p><!><p>ECHO was performed using a Vevo 2100 high-resolution in vivo imaging system. Male and female mice (14–61 weeks old) were put under light anesthesia with isoflurane (~2%) during the procedure. M-mode imaging of the parasternal short axis view allowed evaluation of left ventricular systolic and diastolic function and morphology. The mitral valve flow parameters were acquired through a four-chamber view using the pulsed wave spectral Doppler to be used as an index of diastolic function [21, 22].</p><!><p>Hearts were removed from 15–61 week old male and female mice after anesthesia (3% Isoflurane with 100% O2). Then the aorta was cannulated for retrograde perfusion as previously described [23]. The tyrode solution used for perfusion contained in mmol/l: NaCl 139, KCl 4, NaHCO3 14, NaH2P04 1.2, MgCl2 1, CaCl2 1.5, Glucose 10, S-propranolol 0.0002. Tyrode solution was filtered before use and oxygenated with carbon dioxide (95% O2/5% CO2) to achieve pH 7.4. Hearts were lowered into a warm bath and volume conducted ECGs were recorded using two platinum wire electrodes. Two-prong platinum pacing wires contacted the right ventricles for recording. ECGs were recorded using a custom built pre-amplifier and amplifiers connected to a PowerLab (ADI instruments). Hearts were stabilized for 15 min with sinus rhythm, then paced 12 Hz, 15 Hz, and 18 Hz every 5 min. Arrhythmias (premature ventricular contraction and ventricular tachycardia) observed during 18 Hz pacing and thereafter as previously reported [23].</p><!><p>Mouse hearts were excised and perfused with a 10% buffered formaldehyde solution. The hearts were sliced, stained with Masson's trichrome (MTX) or periodic acid-Schiff (PAS) and examined using a Nikon AZ 100M. Analysis was blinded and performed as previously described using NIH ImageJ program [24].</p><!><p>Body weight was recorded at time of sacrifice and mouse hearts excised and weighed.</p><!><p>All values are presented as mean ± S.E. Error bars represent S.E. Statistical significance was determined by one-way ANOVA using Tukey or Student-Newman-Keuls post- hoc analysis for the myofilament Ca2+ sensitivity, maximal steady-state force developed and echocardiography data, respectively. Histopathology was tested by independent T-test. The statistical analysis of the prior experiments were completed using the program Sigma Plot 12. The incidence of arrhythmias was determined by non-parametric Mann-Whitney test. p < 0.05 was considered to be significant. The arrhythmia data statistical analysis was performed using Origin Lab Pro 2018.</p><!><p>We first tested the Ca2+ sensitivity of contraction in NTg, I79N, HET and I79N/HET mice skinned cardiac muscle preparations. The I79N cardiac muscle preparations showed increased Ca2+ sensitivity of contraction compared to NTg (Figure 1A), while the HET cardiac preparations displayed decreased Ca2+ sensitivity of contraction (Figure 1B). The I79N/HET cardiac preparations showed a promising Ca2+ sensitivity closer to NTg than the I79N mutation alone (Figure 1C). Furthermore, the HET cardiac skinned preparations displayed increased cooperativity of thin filament activation (nHill) compared to NTg (Table 1). Interestingly, the I79N/HET cardiac skinned preparations did not show a significant change in nHill compared to NTg, implying that the I79N mutation rescued the increased nHill imposed by the R141W (HET) mutation (Table 1). The maximal steady-state forces developed among the four groups were not statistically significant. The pCa50, nHill, and Fmax values for all groups are reported in Table 1.</p><!><p>There was no significant difference in heart weight (HW) between NTg vs HET (Figure 2A), however mice expressing the TnT I79N mutation (I79N and I79N/HET) had smaller hearts compared to NTg and HET group mice. The HW normalized to body weight (BW) ratio followed the same trend as observed in HW alone (Figure 2B). Thus, the I79N/HET mice show a heart and body weight phenotype similar to I79N. MTX histopathological staining of the I79N/HET mice hearts showed a trend in the reduction of fibrosis when compared to the I79N mice depicted in Figure 2C and corresponding Figure 2D.</p><!><p>Echocardiography of each mouse group showed that the ejection fraction was not significantly different in any of the groups (I79N, HET and I79N/HET) compared to NTg mice (Figure 3A). The I79N mice had a statistically significant lower stroke volume compared to NTg, which was normalized in the I79N/HET mice (Figure 3B). I79N and HET mice displayed statistically significant lower and higher volumes, respectively, compared to NTg mice (Figure 3C and3D). Interestingly, the I79N/HET mice showed a normalization of both systolic and diastolic volumes (Figure 3C and3D). The diameter of the left ventricle (LV) during diastole and systole followed the same trend as observed for the volumes, e.g., while I79N and HET mice have small and large hearts, respectively, the I79N/HET mice showed similar dimensions to NTg (Figure 3E and3F). These data are further exemplified with the LV mass shown in Table 2. While HET mice display increased LV mass, the I79N/HET showed further normalization. Additional echocardiography parameters are presented in Table 2.</p><!><p>I79N mice displayed an increased susceptibility to ventricular tachycardia compared to NTg mice (Figure 4A and4B) during rapid pacing protocol, consistent with previous findings reported by our group [23]. Ventricular arrhythmias occurred in 5 out of 6 (83% incidences) I79N mouse hearts, but only in 1 out of 4 (25% incidences) I79N/HET mouse hearts (Figure 4A and4B). Therefore, the presence of the Ca2+ desensitizing R141W mutation in I79N hearts attenuated the incidence of ventricular tachycardia during rapid pacing protocol.</p><!><p>In I79N mice, the arrhythmia susceptibility induced by rapid heart rates is in part caused by regional conduction slowing, which manifests itself as QRS widening on the ECG [23]. Hence, we next analyzed the QRS duration during the rapid pacing protocol that elicited the ventricular arrhythmia in I79N mice. Importantly, only I79N mice exhibited QRS widening (30 % compared to sinus rhythms) after fast pacing (Figure 5). QRS widening was blunted in I79N/HET mice (Figure 5), suggesting that expression of the R141W mutation prevents the intraventricular conduction delay caused by the I79N mutation.</p><!><p>Our study shows that two pathogenic TnT mutations — which by themselves cause cardiomyopathy in mice but manifest as opposite effects on myofilament Ca2+ sensitivity — can ameliorate each other's "disease" phenotypes. When transgenic I79N mice that exhibit increased myofilament Ca2+ sensitivity are crossed with knock-in R141W mice that exhibit a decrease Ca2+ myofilament Ca sensitivity and increased nHill, the resulting double-transgenic I79N/HET mice show an overall decrease in the severity of the cardiomyopathy phenotype caused by either mutation alone: Co-expression of the pathogenic R141W mutation prevented to arrhythmia phenotype caused by the TnT-I79N mutation (Figure. 4). Conversely, the I79N mutation prevented the chamber dilation and reduced contractile function caused by the R141W mutation (Figure. 3). Since the two pathogenic mutation have opposite effects on myofilament Ca2+ sensitivity, our results suggest that regardless of the mutation, there exists an independent role of myofilament Ca2+ sensitivity in the development of morphological and functional changes within the diseased heart.</p><p>Over 1000 mutations that cause cardiomyopathies in patients have been found constituting the cardiac sarcomere [25, 26]. Mutations that can affect normal myofilament mechanism of activation and relaxation include multiple TnT domains. The TnT I79N transgenic mouse mutation used in this study is a mutation found in HCM patients. Although the I79N related patients and murine model do not exhibit left ventricle hypertrophy, it does show an increased Ca2+ sensitivity of contraction and an increase in the arrhythmic profile—the latter being a hallmark of HCM. The reduction of cardiac output without increased LV thickness suggest that this mutation may also be implicated in the development of a phenotype that resembles restrictive cardiomyopathy (RCM). This mixed HCM and RCM phenotype has been acknowledged in the contemporary classification of inherited cardiomyopathy and highlights the clinical overlap of these two diseases [27]. The KI TnT R141W mouse mutation, used as a "counter effecting" sarcomeric mutation in this study, is also found in humans, and manifests in decreased myofilament Ca2+ sensitivity.</p><p>Both I79N and R141W mutations are located in TnT1, the tail domain of TnT, known to interact with Tm. Tropomyosin is known to control cooperativity of thin filament activation. Previous studies report on the TnT1 domain's crucial flexibility and its role in regulating Tm positioning for contraction [28–30]. Further evidence for the importance of this domain in our study of TnT1 mutations lies in their ability to alter the affinity of troponin complex for Tm [31]. Therefore, we suggest that the most targeted approach to "rescuing" either HCM or DCM phenotype would be to have a mutation within that same domain to counteract manifestation of the disease, probably through TnT1's flexibility and eventually altered contraction. The increased binding affinity between R141W troponin and Tm [31] possibly underlies the increased cooperativity of thin filament activation displayed by the HET mice within this study (Table 1). Interestingly, I79N/HET mice displayed a normalization of cooperativity of thin filament activation (Table 1). These data support a role for R141W (HET)'s rescuing potential towards the HCM phenotype displayed in the I79N'mice.</p><p>Ultrasound cardiac imaging allowed for several cardiac parameters to be measured and compared between the groups. There was an increase in overall hemodynamic parameters for I79N/HET compared to I79N mice, i.e., cardiac output and stroke volume (Figure 3B and Table 2), suggesting a betterment in heart function. The I79N/HET also showed a normalization of left ventricular volume and dimension during both diastole and systole compared to I79N and HET. Furthermore, I79N/HET mice exhibited reduced susceptibility to arrhythmia during ex-vivo testing, thus lending evidence to the attenuation of ventricular tachycardia in the double mutant. Taken together, these measurements of improved cardiac blood flow, morphology and arrhythmia incidences in I79N/HET demonstrate positive support for the balance between the two studied TnT mutations with opposing effects on the myofilament Ca2+ sensitivity of contraction.</p><p>Myofilament Ca2+ sensitivity has been previously shown to be essential in the manifestation of DCM pathophysiology [32–35]. To note, the TnT R141W mutation showed significantly Ca2+ desensitization and also displayed an increase in Ca2+ transient amplitudes as a consequence [12]. Our data are consistent with prior studies that corrected myofilament Ca2+ sensitivity in mice rescued the cardiomyopathy phenotype. Li et al have shown that a TnI RCM mutation crossed with an N-terminal truncated TnI reversed Ca2+ hypersensitivity and improved cardiac function [36]. In addition, an HCM thin filament mutation crossed with a pseudophosphorylated TnI displayed myofilament Ca2+ desensitization and "associated correction" of systolic dysfunction [37]. Our understanding that manifestations, specifically within the thin filament that affect Ca2+ sensitivity, could have the potential to be corrected mechanistically was supported. This fueled the idea of the potential to "reverse" or at least dampen the effect of opposing thin filament mutations to allow for an improvement in disease presentation. Although studies on the breeding of mice with opposing myofilament Ca2+ sensitivity responses have been previously completed, this study would be the first to test the idea of breeding HCM and DCM mouse models for normalization of cardiac pathophysiology. This study is also novel in its achievement of a genetically engineered mouse model that mitigates the arrhythmias displayed in an HCM model through the desensitization of the myofilament. This had been previously shown with the I79N mice, however this was attained using the unspecific drug Blebbistatin to accomplish that myofilament desensitization ex vivo [10].</p><p>The experimental design used here is based on the hypothesis that thin filament changes will impact the disease phenotype directly. However, there is also the hypothesis that thick filament changes affecting myofilament Ca2+ sensitivity can alter the manifestation of thin filament associated cardiomyopathies as well. For example, permeabilized muscle preparations isolated from a DCM Tm mouse model treated with a small molecule cardiac myosin activator showed Ca2+ an increased myofilament Ca2+ sensitivity that was restored to physiological levels compared to control [38]. This small molecule cardiac myosin activator has also been shown to increase Ca2+ sensitivity of force production in healthy permeabilized rat cardiomyocytes [39]. More recently, human induced pluripotent stem cells-derived cardiomyocytes containing a mutant TnT, associated with DCM, showed improved sarcomere function when treated with the cardiac myosin activator [40]. This modulation of myofilament Ca2+ sensitivity through the thick filament may also be accomplished by enhancing myosin regulatory light chain phosphorylation. Interestingly, a DCM Tm mouse model treated with angiotensin II type I receptor biased ligand showed an improved cardiac contractility that was correlated with an increased ventricular myosin light chain-2 phosphorylation [41, 42]. It is also possible that the mutual rescue observed in I79N/HET mice can involve the remodeling of Z-disk-related proteins. A recent study reported that the alterations in desmin and α-actinin which are pivotal sensor molecules of the sarcomere are directly related to cardiac dysfunction [43].</p><p>Our study revealed that pathogenic TnT mutations (I79N and R141W) that by themselves cause cardiomyopathy but have opposite effect on myofilament Ca2+ responsiveness can cancel out each other's effects during development of diseases. The experiments provide evidence in support of the double transgenic mouse as a mutually "rescued" cardiac phenotype. Our results further emphasizes the function of TnT having an imperative role in the manifestation of the disease. Finally, our study points to the importance of TnT as a cardiomyopathy susceptibility protein and its role in regulating heart function, specially arrhythmias—a major hallmark of HCM disease.</p>

PubMed Author Manuscript

Regulation of a long noncoding RNA MALAT1 by Aryl Hydrocarbon Receptor in pancreatic cancer cells and tissues.

Environmental toxicants such as dioxins and polycyclic aromatic carbons are risk factors for pancreatitis and pancreatic cancer. These toxicants activate aryl hydrocarbon receptor (AHR), a ligand-activated transcription factor, of which activation regulates many downstream biological events, including xenobiotic metabolism, inflammation, and cancer cell growth and transformation. Here, we identified that environmental toxicant-activated AHR increased expression of metastasis associated lung adenocarcinoma transcript 1 (MALAT1) in pancreatic cancer cells and pancreatic tissues. The MALAT1 is a long noncoding (lnc) RNA which interacts with Enhancer of Zeste 2 (EZH2), a histone methyltransferase with epigenetic silencer activity, and the MALAT1-EZH2 interaction increased its epigenetic silencing activity. In contrast, AHR antagonist, CH223191 or resveratrol, counteracted the AHR-mediated MALAT1 induction and MALAT1-enahnced EZH2 activity. Collectively, these results revealed a novel pathway of how environmental exposure leads to epigenetic alteration via activation of AHR-MALAT1-EZH2 signaling axis under pancreatic tissue- and cancer cell-context.

regulation_of_a_long_noncoding_rna_malat1_by_aryl_hydrocarbon_receptor_in_pancreatic_cancer_cells_an

Introduction<!>Cell culture and Chemicals.<!>Animal experiment.<!>Chromatin immunoprecipitation assay.<!>RNA immunoprecipitation (RIP) assay.<!>Real-time quantitative PCR (RT-qPCR).<!>Histological analysis.<!>Histone extraction.<!>Nuclear extraction and H3K27me3 enzymatic assay.<!>Immunoblotting.<!>Immunohistochemical staining.<!>Statistical analysis.<!>Environmental toxicant induced MALAT1, a lncRNA that interacted with EZH2.<!>Role of AHR in the MALAT1 induction<!>Role of AHR-MALAT1 signaling in the regulation of EZH2 activity.<!>Effects of TCDD on Malat1-EZH2 signaling axis in vivo<!>Discussion

<p>Pancreatitis is a prerequisite for development of pancreatic ductal adenocarcinoma (PDA), which is one of the deadliest diseases with no effective prevention strategy currently available [1, 2]. Accumulated epidemiological studies showed a strong connection between cigarette smoking and PDA. Cigarette smoking is considered as a major risk factor for pancreatitis and PDA, which contains various environmental toxicants, including dioxins and benzo(a)pyrene (BaP), agonists of AHR [3-6].</p><p>The AHR is a ligand-activated transcription factor essential for mediating xenobiotic metabolism, immune responses, inflammation, differentiation as well as cancer cell growth and malignancy [7, 8]. It is activated by various environmental toxicants such as 2,3,7,8-tetrachlorodibenzo-p-dioxin (TCDD), BaP, and phytochemicals as well as tryptophan metabolites. Interestingly, depending on type of ligand, AHR can be either activated or inhibited and this selective feature makes AHR an important pharmacological target [9, 10].</p><p>The MALAT1 is a lncRNA which acts molecular scaffolds for various riboprotein complexes and functions as a transcriptional and epigenetic regulator [11]. MALAT1 interacted with EZH2, a histone methyltransferase that mediates gene silencing via tri-methylation of histone 3 lysing 27 (H3K27me3) [12, 13]. Indeed, EZH2 can interact with many lncRNAs other than MALAT1, including HOTAIR [14, 15], H19 [16], MEG3 [17] and XIST [18]. These interactions likely regulate the distribution, level, and activity of EZH2. Furthermore, it was reported that MALAT1 is highly expressed and associated with poor prognosis in pancreatic cancer [19, 20]. However, the underlying mechanisms for MALAT1 regulation by environmental toxicant exposure and MALAT1-directed regulation of EZH2 function in pancreatic cancer cell or tissue remain unclear.</p><p>In the present study, we showed that environmental toxicant-activated AHR induced MALAT1 and that MALAT1 interacted with EZH2 and increased its activity in pancreatic cancer cells and tissues. In contrast, treatment of AHR antagonist, CH223191[21] or resveratrol [22], inhibited the increase of MALAT1, EZH2 activity, and H3K27me3 levels. Taken together, our findings revealed a novel pathway that links environmental exposure to epigenetic regulation via activation of AHR-MALAT1-EZH2 signaling axis in pancreatic cancer cells and tissues.</p><!><p>Panc-1 and AsPC-1 cells were purchased from the American Type Culture Collection (ATCC, Manassas, VA) and were maintained at 37°C in the presence of 5% CO2 in DMEM medium supplemented with 5% fetal bovine serum (FBS) and 1% penicillin/streptomycin solution (Sigma Aldrich, St Louis). TCDD was purchased from AccuStandard, Inc. (New Heaven, CT); BaP was purchased from Sigma Aldrich (St. Louis, MO); CH223191 and Resveratrol were purchased from Cayman Chemical (Ann Arbor, MI). Small interfering RNAs (siRNAs) targeting MALAT1 (ID #: n511399 and n511398) and negative control siRNAs were purchased from Thermo Fisher Scientific. The siRNAs targeting AHR (siRNA ID: SASI_Hs02_00332181 and SASI_Hs01_00140198) were purchased from Sigma Aldrich.</p><!><p>C57BL/6J wild type mice (9-12 weeks old, 25-30 g body weight) were purchased from Jackson Laboratory (Bar Harbor, ME). Mice were allocated for treatment with corn oil as control, TCDD dissolved in corn oil, or Resveratrol in corn oil. Each group contains 4-5 mice. Before any treatment, mice were fasted up to 12 hr. TCDD (30 μg/kg) intraperitoneally (i.p.) was injected into mice which was then anesthetized after 24 hr. In TCDD plus resveratrol treatment group, mice were pre-treated with resveratrol (20 mg/kg, i.p.) and after 90 min, mice were treated with TCDD (30 μg/kg, i.p.). After 24 hours, pancreatic tissue samples were collected for further analyses. The animal experiments were performed in compliance with the guidelines established by the Animal Care Committee of University of Cincinnati. The animals were acclimated to temperature- and humidity-controlled rooms with a 12-h light/dark cycle for 1 week prior to use.</p><!><p>A chromatin immunoprecipitation (ChIP) assay was carried out with AHR (Enzo Life Sciences, Inc., Cat#: ALX-804-423-R100) and RNA polymerase II (Active Motif, Carlsbad, CA, Cat# 39097) antibodies using Panc-1 cell ChIP lysate treated with TCDD for 2 hour with the ChIP-IT Express Chromatin Immunoprecipitation Kit (Active Motif), according to the manufacturer's protocol. The ChIP primer set that covers MALAT1 gene promoter proximal region (−442~−201): (forward): 5'-AGGAGAGAGGTGGGAAAGGAAG-3' and (reverse) 5'-TGGTTCTAACCGGCTCTAGC-3'. All the ChIP-PCR reactions were carried out using a 7300HT Real-Time PCR system or a QuantStudio 3 Real Time PCR system with a 96-well block module (Applied Biosystems). The cycling conditions were 56°C for 30 min and 95°C for 10 min, followed by 48 cycles of 95°C for 25 s and 60°C for 60 s.</p><!><p>To detect the level of interaction between lncRNAs and EZH2, we performed the RIP assay as previously established [23]. Briefly, cells were fixed with formaldehyde and sheared by sonication. After DNase treatment, the fixed chromatins were immunoprecipitated with IgG or EZH2 antibody (Active Motif, Cat # 39933) with RNAse inhibitor. The purified lncRNAs and input RNA lysate were analyzed by qRT-PCR with the individual lncRNA primer set (Supplementary Table 1).</p><!><p>The cDNAs were synthesized with the obtained RNA lysate from cells or tissues using a high-capacity cDNA Reverse Transcription Kit (Applied Biosystems) and RT-qPCR was performed. The obtained results were normalized to GAPDH or β-actin control. All the primer set sequences for detecting specific mouse or human genes by qRT-PCR are listed (Supplementary Table 1).</p><!><p>Paraffin-embedded liver sections were sectioned, deparaffinized in xylene, and rehydrated through a series of graded ethanol solutions. The morphological changes in pancreatic tissue sections stained with hematoxylin and eosin (H&E) were examined under a light microscope for histological analysis.</p><!><p>Histones were extracted from cells or tissues using EpiQuik total histone extraction kit according to the manufacturer's protocol (EpiGentek, Farmingdale, NY).</p><!><p>Nuclear extract from cells or pancreatic tissues were collected using EpiQuik Nuclear Extraction kit (Cat # OP-0002-1, EpiGentek, Inc., Farmingdale, NY). By using obtained nuclear extracts, we performed H3K27me3 enzymatic assay with EpiQuik Histone Methyltransferase Activity/Inhibition Assay Kit (EpiGentek, Inc., Cat # P3005). All the procedures were performed according to the manufacturer's protocol.</p><!><p>Total proteins extracted from either cells or minced pancreatic tissues (30-100 μg per lane) were resolved by sodium dodecyl sulfate-polyacrylamide gel electrophoresis (SDS-PAGE) and transferred onto polyvinylidene difluoride (PVDF) membranes. After being blocked with TBST or PBST buffer containing 5% non-fat milk, the membranes were incubated with proper primary antibody. Proteins of interest were detected with either anti-rabbit or anti-mouse horseradish peroxidase (HRP)-conjugated secondary antibodies (Cell Signaling Technology) and visualized with an enhanced chemiluminescence (ECL) detection kit using a C-DiGit Blot Scanner from LI-COR (Lincoln, NE).</p><!><p>After pancreatic tissue sections were deparaffinized, the sections were immersed in 0.01 mM sodium citrate (pH 6.0) and heated in a microwave oven (98°C) for antigenic retrieval. The deparaffinized sections were incubated with peroxidase-blocking reagent (Biogenex, CA, USA) to block endogenous peroxidase activity and then incubated with nonspecific staining blocking reagent (Vector Laboratories, Burlingame, CA, USA). The sections were incubated with primary antibodies at 4°C overnight. Anti-EZH2 antibody (Cell Signaling Technology, Danvers, MA, Cat # 5246) and anti-H3K27me3 antibody (Active motif, Cat #: 39155) were employed. The sections were subsequently incubated with peroxidase-conjugated secondary antibodies (Vector Laboratories) and 3, 3-diaminobenzine-tetrachloride (DAB; Vector Laboratories), according to the manufacturer's instructions. The sections were counterstained with hematoxylin and observed under a microscope.</p><!><p>Statistical significance was determined by the unpaired Student's t-test with a two tailed distribution or one-way analysis of variance (ANOVA) with a post-hoc Tukey's test. Data are presented as the mean ± SD. Statistical analysis was performed with Graph Pad Prism 6.0 software. Statistical significance was set at a P-value of <0.05.</p><!><p>Many previous reports showed that various lncRNAs formed complexes with EZH2 and regulated activity or distribution of EZH2, an epigenetic writer and silencer with histone methyltransferase function [24, 25]. By employing RIP assay, we determined the types and levels of lncRNA-EZH2 interactions under pancreatic cancer cell context. After Panc-1 human pancreatic cancer cells were fixed with formaldehyde, EZH2-associated lncRNAs were pull down by EZH2 antibody (Fig 1A). The qRT-PCR was performed using the obtained RNA lysate with appropriate primer sets for determining the levels of binding interactions between EZH2 and these lncRNAs, including MALAT1, HOTAIR, MEG3, H19, and XIST, as previously reported as EZH2 interacting lncRNAs. Among lncRNAs, we identified that MALAT1 most strongly interacted with EZH2. MEG3, H19, and HOTAIR also interacted with EZH2 with different levels however no interaction was detected between XIST and EZH2 (Fig. 1B), indicating existence of various lncRNA-EZH2 complexes. The induction levels of CYP1A1, a representative AHR downstream target gene, by TCDD or BaP treatment in Panc-1 and Aspc-1 cells were presented (Supplementary Fig. 1). The primer sequences for all the lncRNAs were listed (Supplementary Table 1).</p><p>In parallel, we determined which lncRNA can be induced by environmental exposure. TCDD, a potent AHR agonist, was used as a model compound. We found out that only MALAT1 was significantly induced by TCDD treatment, not others (Fig. 1C, left). MALAT1 was also induced in AsPC-1 human pancreatic cancer cells (Fig. 1C. right). Furthermore, treatment of BaP, another AHR agonist and ubiquitous environmental toxicant, induced MALAT1 in both Panc-1 and AsPC-1 cells (Fig. 1D). These results together indicated that MALAT1 is an EZH2-interacting lncRNA which is significantly induced by environmental toxicant exposure in pancreatic cancer cells.</p><!><p>Since TCDD or BaP are AHR agonists, we examined a role of AHR in the MALAT1 induction. Cells were treated with TCDD alone or together with AHR antagonist either CH223191 or resveratrol. The induction of MALAT1 was markedly inhibited by AHR antagonist (Fig. 2A). To further determine whether AHR proteins is required for the MALAT1 induction, we depleted AHR proteins using small interfering RNA (siRNA) targeting AHR. Cells were transfected with two different siRNAs targeting different regions of AHR (siAHR I and II) or scrambled control siRNA (siCT). The knockdown of AHR was confirmed by immunoblotting and RT-qPCR. The reduced AHR mRNA and protein levels were observed (Fig. 2B). Furthermore, the AHR depletion using this interfering siRNA approach significantly decreased MALAT1 induction, confirming that AHR was essential for the MALAT1 induction (Fig. 2C). Next, we performed gene promoter analysis using PROMO program [26] whether there is any potential AHR binding sequence located in the MALAT1 gene promoter region. We identified a potential AHR binding site, namely Dioxin Response Element (DRE), in the proximal region of MALAT1 gene promoter (5'-GCGTGCGCAGTCACGC-3') (Fig. 2D, top). To examine whether TCDD-activated AHR is actually recruited to the DRE, we performed chromatin immunoprecipitation (ChIP) assay. The recruitment of AHR to the DRE site was detected and in parallel, RNA polymerase II, a positive control for transcriptional activation, also interacted with the DRE, indicating that AHR plays a role in transcriptional activation of MALAT1 via the DRE site (Fig. 2D, bottom). These results demonstrated that AHR is essential for the TCDD-induced MALT1 induction.</p><!><p>Previous report showed that MALAT1 interacted with EZH2 and regulated its function in pancreatic cancer cells [27, 28]. However, overarching effects of MALAT1 knockdown on EZH2 level and activity remain elusive. Therefore, we transfected cells with siRNAs targeting MALAT1 (siMalat1 I and II) or control siRNA (siCT) in Panc-1 cells and determined effects of MALAT1 depletion on EZH2 and H3K27me3 levels. We observed decreased levels of EZH2 and H3K27me3 (Fig. 3A, top). The MALAT1 depletion efficiency was determined by RT-qPCR (Fig 3A, bottom). Next, using nuclear extracts from the control and MALAT1-depleted cells, we performed the enzymatic assay for measuring H3K27 trimethylation activity since H3K27me3 is a substrate of EZH2 (EpiGenteck, Inc.). Consistently, there was greatly reduced H3K27me3 enzymatic activity observed with the MALAT1-depleted nuclear extracts (Fig. 3B).</p><p>Next, we examined effects of TCDD with or without AHR antagonist treatment on MALAT1 and EZH2 levels, and H3K27me3 enzymatic activities by employing the same approaches above. TCDD increased MALAT1, EZH2, and H3K27me3 levels whereas co-treatment of TCDD with AHR antagonist either CH223191 or resveratrol inhibited the increases (Fig. 3C, top and bottom). Similarly, the TCDD-enhanced H3K27me3 enzymatic activities and its inhibition by CH223191 or resveratrol were observed (Fig. 3D). Collectively, these results indicated that TCDD increased MALAT1, EZH2, and H3K27me3 levels as well as the enzymatic activity of EZH2.</p><!><p>In order to determine effects of TCDD on the Malat1-EZH2 signaling activation in vivo, TCDD was administered to mice with or without pre-treatment of resveratrol. After 24 hours, pancreatic tissues were collected and then immunostaining with anti-EZH2 or anti-H3K27me3 antibody was performed. Significantly increased EZH2 and H3K27me3 staining were observed in the TCDD-treated pancreatic tissues while pre-treatment of resveratrol prevented them (Fig. 4A). The result from RT-qPCR analysis consistently showed that TCDD treatment markedly induced MALAT1 expression however pre-treatment of resveratrol inhibited it (Fig. 4B). Representative images of haemotoxylin and eosin (H&E) staining were shown (Supplementary Fig. 2A). The induction of CYP1A1 by TCDD and its inhibition by resveratrol in the pancreatic tissues were confirmed by RT-qPCR (Supplementary Fig. 2B). Similarly, results from immunoblotting showed that TCDD treatment increased EZH2 and H3K27me3 levels while resveratrol counteracted them (Fig. 4C). The H3K27me3 enzymatic assay was performed with tissue nuclear extracts. TCDD treatment enhanced H3K27me3 enzymatic activities whereas the pre-treatment of resveratrol prevented the enhancement (Fig. 4D). Lastly, to determine effects of the enhanced EZH2 activity on downstream target gene expression, we examined expression levels of miR-200b, a known EZH2 target gene [29, 30]. We found out that miR-200b was significantly downregulated by TCDD but restored its expression level by resveratrol, validating effects of TCDD-mediated MALAT1-EZH2 signaling activation on downstream target gene expression. Taken together, results clearly demonstrated that AHR-activated MALAT1-EZH2 signaling pathway in vivo.</p><!><p>Accumulating epidemiological evidences have indicated that environmental exposure is a main risk factor in pancreatic inflammation and cancer. Our results in this study demonstrated that environmental exposure-activated AHR induced a lncRNA MALAT1 and this induction subsequently increased epigenetic silencing function of EZH2. Overarchingly, these results showed a novel molecular mechanism of how environmental exposure leads to epigenetic alteration in pancreatic cancer cell and tissue context, which, in part, occurs via AHR-MALAT1-EZH2 signaling activation.</p><p>Moreover, given that higher expression or dysregulated activities of EZH2 or MALAT1 played an important roles in cancer stem cell property and malignancy [31, 32], these results provide the possibility for a pro-oncogenic role of AHR in the epigenetic dysregulation. AHR can function as a linchpin molecule that connects environmental exposure to the EZH2-mediated epigenetic dysregulation via MALAT1 induction in pancreatitis and pancreatic cancer. In addition, as shown in Fig. 4D, low levels of miR-200b, one of downstream target gene of MALAT1-EZH2 signaling pathways, is highly correlated with epithelial-mesenchymal transition, cancer stem cell features, and drug resistance in pancreatic cancer [33, 34].</p><p>It is also of interest that AHR can selectively be activated or inhibited, depending on ligand-type. Our results demonstrated that either AHR antagonist or depletion inhibited the MALAT1 induction and prevented the increased EZH2 enzymatic activity in both in vitro and in vivo (Fig. 3 and 4), suggesting that AHR antagonism has a potential to be a novel prevention strategy. However, it is highly conceivable that depending on environmental toxicant- or ligand-type, AHR induced other lncRNAs that differentially regulates EZH2 function. Therefore, environmental toxicant- or ligand-dependent regulation of lncRNAs via AHR need to be further systematically investigated. Lastly, in line with our results as shown in this study, it is highly interesting to investigate a role of AHR-MALAT1-EZH2 signaling under experimental settings of pancreatitis and pancreatic cancer in the future.</p>

PubMed Author Manuscript

Rational designing of glyco-nanovehicles to target cellular heterogeneity

The aberrant expression of endocytic epidermal growth factor receptors (EGFRs) in cancer cells has emerged as a key target for therapeutic intervention. Here, we describe for the first time a state-of-the-art design for a heparan sulfate (HS) oligosaccharide-based nanovehicle to target EGFR-overexpressed cancer cells in cellular heterogeneity. An ELISA plate IC 50 inhibition assay and surface plasma resonance (SPR) binding assay of structurally well-defined HS oligosaccharides showed that 6-O-sulfation (6-O-S) and 6-Ophosphorylation (6-O-P) of HS tetrasaccharides significantly enhanced EGFR cognate growth factor binding.The conjugation of these HS ligands to multivalent fluorescent gold nanoparticles (AuNPs) enabled the specific and efficient targeting of EGFR-overexpressed cancer cells. In addition, this heparinoid-nanovehicle exhibited selective homing to NPs in cancer cells in three-dimensional (3D) coculture spheroids, thus providing a novel target for cancer therapy and diagnostics in the tumor microenvironment (TME).

rational_designing_of_glyco-nanovehicles_to_target_cellular_heterogeneity

Introduction<!>Results and discussions<!>Conclusions<!>Conflicts of interest

<p>Cancer is a devastating and multifactorial disease. Recent studies have conrmed that over-expression of certain cell-surface receptors, and growth factors, and suppression of tumor-specic genes are primary causes of cancer phenotype development. 1 Therefore, developing suitable markers for these functional changes may provide new diagnostic tools and lead to new delivery systems. Over the past two decades, epidermal growth factor receptors (EGFRs) have emerged as potential oncogenes that are commonly found in various cancer types. 2 Thus, EGFR targeted small molecules and EGFR-neutralizing monoclonal antibodies have impressive clinical signicance. 3 EGFRs bind to and are activated by their autocrine growth factors, such as EGF and heparin-binding EGF-like growth factors (HB-EGFs) that are oen governed by heparan sulfate (HS), which is ubiquitous on both cell surfaces and the extracellular matrix. 4 Hence, deciphering the structure-function relationship between HB-EGF or EGF and HS could be a new tool for selectively targeting cancer cells in tumor microenvironments.</p><p>Structurally, HS is composed of a(1-4)-linked disaccharide repeating units of D-glucosamine and hexuronic acid, which could be either D-glucuronic acid (GlcA) or L-iduronic acid (IdoA).</p><p>The structural diversity of HS comes from the degree of Osulfation and N-sulfation/acetylation on the glucosamine and hexuronic acid ligands. 5 Several research groups have synthesized broad well-dened HS oligosaccharide libraries to elucidate the active ligand for growth factors, chemokines and biologically active molecules. 6 The majority of these HS libraries are able to determine the sulfation code, uronic acid composition and oligosaccharide lengths of the HS during protein recognition. However, to the best of our knowledge, it is still unclear as to what is the HS chemically dened epitope(s) of EGFR specic growth factors.</p><p>Herein, we report a new set of HS-tetrasaccharides with well-dened sulfation codes to determine the structure-functional relationship between EGFR binding growth factors. An ELISA assay with HB-EGFs and EGFs rationalized the molecular code of HS for EGFR activation. SPR binding conrms the strength of HS binding affinity. We have also synthesized the phosphate analog of super-active HS-tetrasaccharides in order to determine why nature prefers the sulfate pattern of HS during molecular recognition. Active ligands were functionalized on uorescent gold nanoparticles to highlight the endocytosis process in different cancer cell lines with variable EGFR expressions. Finally, we developed a three-dimensional coculture spheroid model to demonstrate selective targeting of cancer cells in the presence of stromal cells and the extracellular matrix (Fig. 1).</p><p>Although HS/heparin polymer-based nanoparticles targeting cancer cells have been reported in the literature, 7 however, structure heterogeneity of the HS polymer rendered native ligands less specic to cancer cells. Here, we report the rst example of well-dened HS oligosaccharide-based nanovehicle construction to target specic cancer cells, followed by a 3D-coculture assay to establish selective targeting of cancer cells in the tumor microenvironment.</p><!><p>With the aim of identifying HS oligosaccharide ligands for EGFR specic growth factors, we synthesized N-unsubstituted and N-acetate derivatives of glucosamine with sulfation substitution at 6-OH, 3-OH and 2-OH of iduronic acid residue respectively. The rationale behind this sulfation patterns is that they exhibit a common binding pocket for various growth factors and chemokines. We adopted the [2 + 2] glycosylation strategy comprising glycosyl donor 15 and acceptor 16 to synthesize tetrasaccharide precursors 22 and 23 (Scheme 1). The disaccharide assemblies 13 and 14 (ref. 8) were obtained from orthogonally protected D-glucosamine donors 11 and 12 and acceptor L-idopyranosyl 9 (ref. 9) under standard glycosylation conditions. The thiodonor 11 carries C-4 chloroacetyl and C-6 silyl as a facile protecting group which can be removed easily at the later stage for further modication and chain elongation. The 4-O-chloroacetate of 13 was selectively deprotected in the presence of thiourea to afford acceptor 16 in quantitative yield. The disaccharide donor 15 was obtained by acetolysis of 14 using acetic anhydride and copper(II) tri-uoromethanesulfonate as a catalyst followed by phenyl trimethylsilyl sulphide and ZnI 2 treatment to generate corresponding thioglycoside in excellent yield. 10 Next, we conrmed the HB-EGF binding affinity of 1-7 HStetrasaccharides. To this end, native heparin was immobilized on an ELISA plate and standard competition assay with heparinoids (1-7) was performed using HB-EGF, and EGF proteins. ELISA analysis revealed that the only 6-O-S HS tetrasaccharide (7) showed strong inhibition with HB-EGF binding to native HS (IC 50 ¼ 126.6 mM) (Fig. S2 †) compared to 2-O-S, 3-O-S and nonsulfated HS-tetrasaccharide analogs (1-6) at a concentration between 0 and 1 mg l À1 . In contrast, EGF protein showed no binding with synthetic HS-tetrasaccharides. An additional surface plasmon resonance (SPR) binding experiment of 7 with HB-EGF revealed a K D of 11.12 mM (Fig. 2a and Table S1 †), clearly illustrating that the 6-O-S HS tetrasaccharide functioned as an active ligand for the HB-EGF. To determine whether comp. 7 and HB-EGF interactions differentiate between charge species geometrically close to the sulfate group, 11 we synthesized 6-Ophosphated HS tetrasaccharide by phosphorylating 31 using diphenyl phosphoryl chloride to yield 60% of 34 (Scheme 3). Finally, LiOH mediated lactone ring opening and hydrogenolysis yielded 45% of 8 (Scheme 3). The ELISA assay and SPR binding showed strong binding with the HB-EGF (IC 50 ¼ 97.13 mM) (Fig. S2 †) and a K D of 11.06 mM (Fig. 2b and Table S1 †). These results demonstrated that the sulfated and phosphated HS tetrasaccharides functioned as active ligands of the HB-EGF. It would be interesting to determine whether both ligands target cancer cells by activating the EGFR through autocrine HB-EGF signaling.</p><p>To assess the efficacy of the HS-tetrasaccharide ligands in activating EGFRs, we functionalized them using commercial Nhydroxysuccinimide-active uorescent AuNPs (AF 555 Au), which served as optical and non-toxic probes. 12 The nanoparticle functionalization was performed by mixing ligands 7 and 8 at RT in 0.01 M PBS buffer with a pH of 7.5 for 12 h (Scheme 4). The remaining NHS groups were then neutralized with ethanolamine to afford heparinoid-capped uorescent AuNPs (AF 555 Au@1 and AF 555 Au@2), collectively represented as heparinoid-AuNPs. The physical properties of AF 555 Au@1 and AF 555 Au@2 were conrmed by transmission electron microscopy (TEM), UV-visible and uorescence spectroscopy and zeta potential measurements (Table S2, Fig. S3 and S4 †). As a control, native heparan sulfate was conjugated with Texas Red (T-HP) and characterized (Fig. S1 †).</p><p>Next, a cellular uptake assay was performed using the standard protocol described in the ESI. † Breast cancer cell lines were selected based on the EGFR expression level (MDA-MB-468 high degree; MDA-MB-231, T-47D and MCF-7: moderate to low degree; and SK-BR-3 least EGFR expression 13 ). The cancer cells and NIH-3T3 (as normal cells) were seeded on eight-well glass chamber slides and allowed to grow until they reached 70-80% conuency at 37 C in a 5% CO 2 incubator. Heparinoid-AuNPs (15 mg ml À1 ) and T-HP (10 mg ml À1 ) were added to the wells, and live images were recorded at two different time intervals (4 h and 24 h) (Fig. 3(ii) and S5 †). To demonstrate the HB-EGFmediated uptake, the uptake mechanism was tested in the presence of 0.1 ng ml À1 of the HB-EGF proteins (Fig. S6 †). Hierarchical clustering (HCA) was developed based on the uorescence intensity of the heparinoid-AuNPs inside the cells (Fig. 3(i)). To ensure the consistency of the ndings, all tests were performed in triplicate. The HCA of the heparinoid-AuNP cellular internalization assay indicated the presence of a distinct disparity in uptake rates. As expected, the MDA-MB-468 cell line showed stronger cellular internalization responses than the other breast cancer cell lines and normal cells. Among the heparinoids, the uptake rate of AF 555 Au@1 was approximately 70-80% stronger aer 4 h (Fig. 3(ii)), and the uptake rate signicantly increased in the presence of the HB-EGF protein (Fig. S6 †). This trend continued aer 24 h. MDA-MB-468 showed a preferential uptake of the sulfatedheparinoid over the phosphate analog, conrming the critical role of 6-O-sulfation of glucosamine in HB-EGF/EGFR signaling.</p><p>The uptake rate in the other breast cancer cell lines was weak as compared to that of MDA-MB-468. Moreover, T-HP exhibited a weak cellular uptake rate when compared to the heparinoid-AuNPs, indicating that synthetic ligand 7 might be a better functional ligand in terms of targeting the EGFR than the native HS sequence. We also performed FACS assays with MDA-MB-468 and NIH-3T3 to quantify the percentage cell uptake of the heparinoid-AuNPs. FACS analysis clearly revealed the potential uptake of AF 555 Au@1 in MDA-MB-468 and no uptake in the normal NIH-3T3 cells (Fig. 3(iv)). On the basis of these results, we hypothesized that AF 555 Au@1 could serve as a potential nanovehicle for targeting breast cancer cells.</p><p>To elucidate the mechanism underlying endocytosis, we performed live confocal imaging studies in the presence of endocytic pathway inhibitors. First, we evaluated the energydependent pathway. To this end, we incubated MDA-MB-468 with sodium azide so as to deplete the ATP and then administered the AF 555 Au@1 treatment for 4 h. We observed a substantial decrease in cellular internalization of the nanoparticles, indicating the presence of receptor-mediated endocytic pathways (Fig. 3(iii-b)). To analyze EGFR-mediated endocytosis, getinib (an EGFR inhibitor) (30 mM) was added. The blockage of the receptor resulted in a strong decrease in the cellular uptake of AF 555 Au@1 (Fig. 3(iii-c)). These ndings suggested that AF 555 Au@1 undergoes receptor-mediated endocytosis.</p><p>Recent research established that a two-dimensional (2D) monolayer cell assay did not replicate the in vivo tumor model to evaluate the efficacy of nanovehicles in cancer therapy. 14 Alternatively, three-dimensional (3D) spheroids can provide an attractive in vitro model that accurately mimics the tumor microenvironment (TME) for the purposes of drug discovery and tumor targeting. 15 The TME comprises tumor cells encapsulated by a dense extracellular matrix (ECM) as well as heterogeneous cell types such as stromal cells, immune cells, and endothelial cells. These heterogeneous cellular environments, together with the composition of the ECM, support the enhancement of cancer cell motility, activate different signaling pathways, and reduce the targeting efficacy of nanovehicle delivery to the cancer cells. 16 Thus, it is important to examine the activity of heparinoid-AuNPs in a 3D-spheroid model to demonstrate the efficacy. We rst constructed MDA-MB-468 and SK-BR-3 cell microspheroids using ECM Matrigel. Aer eight days of culture, we observed the formation of spheroids that were uniform in size (50-52 mm) and contained approximately 10-15 cells per spheroid. We added AF 555 Au@1 at an optimum concentration of 50 mg ml À1 . Confocal imaging revealed a strong uptake of AF 555 Au@1 by the MDA-MB-468 cells (Fig. 4). However, no uptake of AF 555 Au@1 was observed in the SK-BR-3 cells. Well-organized spheroids composed of hoechst stained nuclei (Fig. 4a) and the red uorescent AF 555 Au@1 (Fig. 4b) were co-localized as shown by a 3D reconstruction of the confocal Zstack images (Fig. 4c and d). These results conrmed that AF 555 Au@1 diffused into the ECM and targeted the MDA-MB-468 cells.</p><p>Next, we investigated whether the cancer cells were selectively targeted in the presence of stromal cells such as broblast cells. Fibroblast cells are one of the prominent cell types in the TME, and they play a pivotal role in ECM remodeling and inducing resistance to the uptake mechanism of nanovehicles by cancer cells. 17 To shed light on selective targeting of cancer cells in the presence of stromal cells, we constructed 3D coculture models by mixing an optimum 2 : 1 concentration of MDA-MB-468 and GFP stable NIH-3T3, which also replicate the TME. 18 Aer ve days, we observed the presence of multicellular spheroids. These spheroids were much larger in size then the MDA-MB-468 alone, and they were composed of 15-20 cells per spheroid with a maximum size of 100-110 mm. In addition, as shown in the Z-stack images (Fig. 5b-d) NIH-3T3 encapsulated the cancer cell spheroids thereby, creating a broblast layer outside the tumor cells. To these co-culture models, we added AF 555 Au@1 (50 mg ml À1 ). Aer 4 h, as shown in merged Z-stack images and 3D reconstruction (Fig. 5d and f, respectively), AF 555 Au@1 successfully crossed the broblast layer and targeted only the cancer cells. These results suggest that AF 555 Au@1 targeted the cancer cells in the presence of broblast cells, thus demonstrating the efficacy of the nanovehicle in targeting cancer cells in the TME.</p><!><p>In conclusion, we synthesized a structurally well-dened HS library using a divergent strategy. The 6-O-S HS tetrasaccharide was the most active ligand among the series when the binding affinity with EGFR cognate growth factors was tested. This study also conrmed that 6-O-phosphate is a potential ligand of HB-EGFs. Using these two ligands, we have constructed HS-based uorescent-nanovehicles that are intended to target breast cancer cells in tumor microenvironments. Confocal imaging studies conrmed the enhanced uptake of 6-O-sulfated HStetrasaccharide by EGFR-overexpressed cancer cells, whereas their phosphate derivatives showed weak EGFR-mediated uptake rates. These results conrmed the critical role of HS sulfation groups in EGFR activation. The HS-nanoparticles targeted the breast cancer cells via HB-EGF/EGFR-mediated interactions in both the 2D-monolayer and in the 3D-complex coculture tumor model. Overall, these results represent a major step forward toward designing a HS-based nanovehicle for targeted cancer therapies.</p><!><p>There are no conicts of interest to declare.</p>

Royal Society of Chemistry (RSC)

Longitudinal investigation of neuroinflammation and metabolite profiles in the APP swe×PS1Δe9 transgenic mouse model of Alzheimer's disease

AbstractThere is increasing evidence linking neuroinflammation to many neurological disorders including Alzheimer's disease (AD); however, its exact contribution to disease manifestation and/or progression is poorly understood. Therefore, there is a need to investigate neuroinflammation in both health and disease. Here, we investigate cognitive decline, neuroinflammatory and other pathophysiological changes in the APP swe×PS1Δe9 transgenic mouse model of AD. Transgenic (TG) mice were compared to C57BL/6 wild type (WT) mice at 6, 12 and 18 months of age. Neuroinflammation was investigated by [18F]DPA‐714 positron emission tomography and myo‐inositol levels using 1H magnetic resonance spectroscopy (MRS) in vivo. Neuronal and cellular dysfunction was investigated by looking at N‐acetylaspartate (NAA), choline‐containing compounds, taurine and glutamate also using MRS. Cognitive decline was first observed at 12 m of age in the TG mice as assessed by working memory tests . A significant increase in [18F]DPA‐714 uptake was seen in the hippocampus and cortex of 18 m‐old TG mice when compared to age‐matched WT mice and 6 m‐old TG mice. No overall effect of gene was seen on metabolite levels; however, a significant reduction in NAA was observed in 18 m‐old TG mice when compared to WT. In addition, age resulted in a decrease in glutamate and an increase in choline levels. Therefore, we can conclude that increased neuroinflammation and cognitive decline are observed in TG animals, whereas NAA alterations occurring with age are exacerbated in the TG mice. These results support the role of neuroinflammation and metabolite alteration in AD and in ageing.

longitudinal_investigation_of_neuroinflammation_and_metabolite_profiles_in_the_app_swe×ps1δe9_transg

<!>Animals<!>Study design<!><!>Novel object and smell recognition tests<!>Y‐maze spontaneous alteration task<!>Behaviour to predator and non‐predator urine<!>[18f]dpa‐714 pet<!>Magnetic resonance spectroscopy acquisition and analysis<!>Tissue collection and immunohistochemistry<!>Statistical analysis<!>TG mice display increased cognitive decline<!><!>TG mice display increased cognitive decline<!>[18F]DPA‐714 binding significantly increases in vivo as a result of AD‐like pathology and age<!><!>Metabolite profiles of APPswe×PS1Δe9 and WTs are affected by age<!><!>Immunohistochemistry<!><!>In vivo studies<!>Ex vivo immunohistochemistry<!>Conclusion<!>

<p>Alzheimer's disease</p><p>β‐amyloid plaques</p><p>integrin αM chain of the Macrophage‐1 antigen</p><p>cerebrospinal fluid</p><p>creatine & phosphocreatine</p><p>discrimination index</p><p>time spent exploring the familiar object/scent</p><p>glial fibrillary acidic protein</p><p>glutamate</p><p>magnetic resonance spectroscopy</p><p>microtubule‐associated protein‐2</p><p>mild cognitive impairment</p><p>magnetic resonance</p><p>myo‐inositol</p><p>N‐acetylaspartate</p><p>neurofibrillary tangles</p><p>normalized (to the cerebellum) uptake value</p><p>novel object recognition</p><p>novel smell recognition</p><p>positron emission tomography</p><p>standard uptake value</p><p>synaptic vesicle glycoprotein 2A</p><p>taurine</p><p>time spent exploring the novel object/scent</p><p>total choline</p><p>total time exploring objects/scents</p><p>transgenic</p><p>translocator receptor 18 kDa</p><p>wild type</p><p>Alzheimer's disease (AD) is the most common form of dementia which has become a huge socio‐economic burden because of an increased ageing population. Despite considerable research, the mechanisms leading to its manifestation remain unclear. AD is defined pathologically by amyloid plaques (Aβ), neurofibrillary tangles and neuronal loss (Braak and Braak 1991); however, the aetiology remains elusive. Hence, a need exists to find biomarkers that can help understand disease progression and mechanisms, aid diagnosis and track the efficacy of therapeutic interventions.</p><p>AD is a multifactorial disease associated with complex neuronal and neuroinflammatory alterations (Halliday et al. 2000; Rubio‐Perez and Morillas‐Ruiz 2012; Perry and Holmes 2014; Heneka et al. 2015; Mhatre et al. 2015). Activated microglia have been shown to surround amyloid plaques (Haga et al. 1989) ex vivo and increased inflammatory cytokine expression is observed post‐mortem in brain and in vivo in the serum and CSF of AD and patients with mild cognitive impairment (MCI) (Forlenza et al. 2009; Swardfager et al. 2010; Rubio‐Perez and Morillas‐Ruiz 2012; Varnum and Ikezu 2012; Westin et al. 2012). In addition, treatments which experimentally reduce pro‐inflammatory cytokines have been shown to improve memory performance (Lee et al. 2013; Song et al. 2013), suggesting increased inflammation could contribute to memory impairments seen in AD.</p><p>The ability to accurately assess neuroinflammation in vivo is important to our understanding of the impact it has in AD. The discovery that translocator receptor 18 kDa (TSPO) expression is up‐regulated on activated microglial cells (Papadopoulos et al. 2006) has enabled the generation of TSPO radio‐ligands to image neuroinflammation in vivo using positron emission tomography (PET). TSPO expression in the brain is mainly restricted to endothelial cells and activated microglia, and is low in healthy brain tissue where inflammation is absent (Papadopoulos et al. 2006; Scarf and Kassiou 2011); therefore TSPO PET permits the identification and tracking of region‐specific neuroinflammation throughout disease progression. It is important to note that TSPO is also expressed in endothelial cells and hence in blood vessels (Turkheimer et al. 2007).</p><p>PK11195, a TSPO ligand, has been successfully labelled with 11C and has shown good correlation to microglial activation ex vivo (Venneti et al. 2009); however, contradictory reports exist regarding the detection of neuroinflammation in AD using [11C]PK11195 imaging. Some studies have demonstrated increased [11C]PK11195 binding in AD and MCI patients (Edison et al. 2008; Yokokura et al. 2011), whereas others have reported no differences in any regions between patients and age‐matched controls (Wiley et al. 2009; Schuitemaker et al. 2013). Short half‐life, low signal to noise ratio, high non‐specific binding and modelling issues of [11C]PK11195 may have contributed to these contradictory results (Boutin and Pinborg 2015; Varrone and Lammertsma 2015); hence, new generation tracers with improved affinities and kinetics such as [18F]DPA‐714 (James et al. 2008; Chauveau et al. 2009; Doorduin et al. 2009; Boutin et al. 2013) have been developed to assess. Recently, using [18F]DPA‐714, Hamelin et al. (2016) have demonstrated that AD patients with slower cognitive decline had higher level of neuroinflammation, hypothesizing that neuroinflammation may also exert neuroprotective functions early in the disease. However, this is the first in vivo study showing a positive correlation between neuroinflammation and cognitive decline, suggesting a potential beneficial role for neuroinflammation in AD. However, this needs to be further investigated as TSPO expression is only a surrogate marker of microglial activation in a broad sense and does not provide information on the functional phenotype of activated microglia. Similarly, TSPO expression has been reported to be increased with ageing (Gulyas et al. 2011; Kumar et al. 2012) but the role and meaning of this potential increase with age is yet to be fully understood.</p><p>Similarly, magnetic resonance spectroscopy (MRS), a method that allows the detection of metabolite profiles in vivo, can also be used to investigate neuroinflammation through measurement of myo‐inositol (mI) levels. mI has been suggested to be a glial specific marker (Brand et al. 1993) indicative of increased gliosis. Although increased levels of mI have been repeatedly reported in AD and MCI subjects in regions associated with degeneration in AD (Parnetti et al. 1997; Rose et al. 1999; Kantarci 2007; Shinno et al. 2007; Foy et al. 2011; Shiino et al. 2012; Murray et al. 2014), evidence suggests that mI does not necessarily correlate with increased microglial activation ex vivo (Murray et al. 2014; Pardon et al. 2016) and may be more directly associated with plaque load. MRS can also measure other metabolites of interest related to neuronal function including N‐acetylaspartate (NAA), glutamate (Glu), taurine (Taur), total choline (tCho) and creatine + phosphocreatine (Cr). NAA is a neuronal marker (Birken and Oldendorf 1989; Urenjak et al. 1993) and decreases in its levels are consistently reported in AD patients (Parnetti et al. 1997; Rose et al. 1999; Kantarci 2007; Shinno et al. 2007; Watanabe et al. 2010; Foy et al. 2011; Shiino et al. 2012; Murray et al. 2014), suggesting neuronal dysfunction or death. Moreover, mI and NAA levels have been shown to be correlated with Aβ burden (Klunk et al. 1992; Murray et al. 2014) and performance in some cognitive tasks (Rose et al. 1999; Foy et al. 2011), implicating both in AD pathology and disease severity.</p><p>Although, the majority of AD cases are sporadic, so far most of the animal models use transgenes present in the inheritable familial form. Nevertheless, these models give us insight into the pathological mechanisms underlying AD. Hence, here we used the human double mutation APPswe×PS1Δe9 mouse model to investigate inflammatory and neuronal integrity biomarkers as indicators of disease progression using MRS, PET, immunohistochemistry and cognitive assessment.</p><p>A longitudinal study was conducted in male TG APPswe×PS1Δe9 and WT C57BL/6 mice to investigate neuroinflammation using PET, neurochemical profile using MRS and cognitive function using several behavioural assessments. This was done with the aim of characterizing inflammatory, metabolite and cognitive differences between TG and control animals with age. We hypothesized that there would be increased neuroinflammation (demonstrated by increased DPA‐714 uptake and mI levels), neuronal dysfunction (decreased NAA levels), neurotransmitter disturbances (altered Glu) and cognitive decline with age in the TG compared to WT mice.</p><!><p>Fourteen TG male APPswe×PS1Δe9 mice (RRID: MGI:5701399) with a C57BL/6 (RRID: IMSR_JAX:000664) background and 17 WT C57BL/6 mice were acquired from the Jackson laboratory (Bar Harbor, ME, USA) at 8 weeks of age and allocated to the study. Additionally, 5 WT and 7 TG mice were bred in‐house and complemented the initial groups (see details below). As the mice are per se WT or TG, the WT and TG cannot be randomized. All mice were kept in the Biological Sciences Facility in the University of Manchester. Animals were housed in individually ventilated cages in groups of 2–5 in a 12 h:12 h light and dark cycle with environmental enrichment and 24 h access to food and water. In total, 22 male WT and 21 TG male mice were used in this study. Eight WT and 10 TG were measured repeatedly for MRS (two WT and five TG of these were bred in‐house); a total of 20 WT and 16 TG (five WT and seven TG of these were bred in‐house) mice were scanned non‐repeatedly with PET. This was as a result of seven TGs and five WTs dying of unknown causes or during anaesthesia, and one WT was excluded after abnormal brain morphology was noticed during MR imaging. In addition, three animals in each group were used for analysis ex vivo at both 6 and 12 months. At 18 months, all mice were killed and the brains removed for analysis ex vivo. All experiments were carried out in accordance with the Animal Scientific Procedures act 1986 and approved by the University of Manchester Local Ethical Review Committee.</p><!><p>The study was not pre‐registered. Mice underwent longitudinal behavioural testing and imaging (Fig. 1) at 6, 12 and 18 months of age (body weights (in grams, mean ± SD): 32.9 ± 1.71, 37.9 ± 3.45, 38.9 ± 3.69 in WT and 34.5 ± 3.19, 40.2 ± 4.28, 41.1 ± 6.17 in TG at 6 m, 12 m, 18 m of age respectively). Mice were also tested at 3 (11 WT and 15 TG) and 9 months (9 WT and 12 TG) of age to check for potential changes in behaviour and potentially guide changes in the choice of imaging time‐points; however, as there was no significant differences between 3, 6 and 9 months' time‐points, data from the 3 and 9 months' time‐points are not shown here and only the time‐points matching the imaging time‐points are shown. Imaging experiments began at 6 months of age because of the development of detectable amyloid pathology from this age in this model (Jankowsky et al. 2004). Immunohistochemistry was carried out on separate animals at 6 and 12 months (see 'animals' above). Animals followed longitudinally were culled and processed for ex vivo analysis at 18 months of age. A week gap was given between imaging and behavioural experiments to allow animals to recover from the potential stress of imaging/anaesthesia and limit interference between experiments. All experiments were carried out between 9 am and 5 pm, with all behaviour carried out between 9 am and 2 pm for each time‐point. For all behavioural tests animals were placed in the testing room 30 min prior to experimentation to allow for habituation to the environment.</p><!><p>Gantt chart of the study. Behaviour tests and positron emission tomography (PET) and magnetic resonance spectroscopy (MRS) scans were performed at 6, 12 and 18 months of age. Behaviour tests were also performed at two additional time‐points to evaluate potential cognitive changes and had cognitive deficit been detected at 9 months, the imaging time‐points would have been brought forward.</p><!><p>A plastic circular arena (30 cm diameter, 21 cm height) was used for behavioural testing. Animals were subjected to 2 days of habituation and 1 day of novel object recognition (NOR) testing. Habituation involved single mice exploring an empty arena for 5 min and was carried out in a random order within the same time‐frame each day. NOR utilizes the natural behaviour of rodents to explore novelty to test non‐associative working memory (Ennaceur and Delacour 1988). This was assessed by investigating the ability to discriminate between novel and familiar objects and was carried out in both TG and WT mice at 6, 12 and 18 months of age. Small plastic objects of varying shapes and colours (e.g. Lego blocks) were used as NOR objects. Test day consisted of two phases. In phase 1, a mouse was placed in the arena with two identical unfamiliar objects for 10 min. After this time, the mouse was returned to its home cage and a 1 h delay was implemented. In phase 2, one familiar object was replaced by a novel object and the mouse was given 4 min to investigate the objects.</p><p>Novel smell recognition (NSR) utilizes the natural exploration of novelty to test working memory in rodents although through olfaction. The experimental design was the same as the NOR; however, a 3‐min delay was implemented and identical plastic scent balls with sniffing holes were stuffed with cotton wool and filled with 0.5 mL of a certain scent (e.g. vanilla) were used.</p><p>For both the NOR and NSR, behaviour was recorded and analysed retrospectively. In addition, in between trials, all objects and areas were cleaned with 70% ethanol to remove any scent of the previous mouse which may alter results.</p><p>Time spent investigating left and right identical objects/smells in phase 1 were analysed to assess side bias. Mice were to be excluded if significant side bias (> 60% of time spent investigating one of two identical objects) was observed in phase 1. Time spent exploring the novel and familiar objects in phase 2 was used to generate a discrimination index (DI), defined as the difference between time spent exploring the novel (Tn) and the familiar (Tf) object/scent divided by the total time (T) (DI = (Tn‐Tf)/T). This resulted in values ranging from −1 to +1. A negative value indicated more time spent with the familiar, a positive value indicated more time investigating the novel and a zero value indicated no preference.</p><p>The re‐use of novel objects for each mouse was kept to a minimum to prevent potential memorization of objects/scents between time‐points and ensure the test was unique (i.e. a truly novel object or smell they were never exposed to before). The only exception to this was a small cohort of mice that underwent behavioural testing at 3 months of age to investigate whether early time‐points were needed for baseline. No significant differences were observed at this age in this small group and therefore only 6, 12 and 18 months were chosen going forward with behavioural testing.</p><!><p>The Y‐maze spontaneous alteration task utilizes the natural exploratory behaviour of rodents to assess spatial learning and short‐term memory. The Y‐maze test was carried out as described previously shown by Martins et al. (2017). A black Perspex maze with three arms (15 cm length × 10 cm width × 10 cm depth per arm) labelled with A‐C and with different internal visual cues was used. The mice were placed inside the maze and arm entries were recorded manually over an 8‐min time period, with entries only valid if the whole body of the mouse entered the arm. Successful spontaneous alternation was defined by consecutive entry into all three arms in any order. Analysis was carried out by calculating overlapping triplet sets relative to successful arm entries as previously described (Hiramatsu et al. 1997; Knight et al. 2014). Percentage of successful alternation was compared between WT and TG mice at all ages.</p><!><p>A Y‐maze with either predator or non‐predator urine at the end of each arm was used to assess the general olfactory ability of 12‐month‐old WT and TG mice. The maze had no visual cues. Vented containers were injected with 1 mL of bobcat, fox or rabbit urine and were placed at the end of each arm. Mice were allowed to explore the maze for 8 min and time (in seconds) spent in each arm was quantified. Time spent in the middle of or rearing on the side of the maze was not counted. Both bobcats and foxes are predators for mice, whereas rabbits are not, hence an increased amount of time in the arm containing rabbit urine would be expected in mice with intact olfaction.</p><!><p>Neuroinflammation was investigated using the TSPO tracer [18F]DPA‐714. [18F]DPA‐714 was produced as previously described (James et al. 2008). Animals were anesthetised, cannulated (via tail vein) and injected with [18F]DPA‐714 (12.3 ± 1.9 MBq). Respiratory rate and temperature were monitored throughout the experiment and body temperature was maintained at 37 ± 0.5°C (BioVet® system m2 m Imaging Corp., Cleveland, OH, USA). Images were acquired on a Siemens Inveon® PET‐CT scanner using a 60‐min dynamic acquisition. CT scans were performed prior to PET acquisition to obtain the attenuation correction map. The time coincidence window was set to 3.432 ns and levels of energy discrimination to 350 keV and 650 keV. List mode data from emission scans were histogrammed into 16 dynamic frames (5 × 1 min; 5 × 2 min; 3 × 5 min and 3 × 10 min) and emission sinograms were normalized, corrected for attenuation, scattering and radioactivity decay and reconstructed using an OSEM3D protocol (16 subsets and 4 iterations) into images of dimensions 128 (transaxially) ×159 (axially) with 0.776 × 0.776 × 0.796 mm voxels. The PET images segmented using the local means analysis method and the organ mean time activity curves were corrected for partial volume effect as previously described (Maroy et al. 2008, 2010; Boutin et al. 2013). The correction method combined the geometric transfer matrix method and the regions of interest (ROI)‐opt method. Dynamic PET images were analysed using Brainvisa and Anatomist software (http://brainvisa.info/web/index.html) and quantified using the magnetic resonance imaging (MRI) template (Waxholm space) created by Johnson et al. (2010). This MRI mouse brain template was used to create three brain ROIs (all cortical areas and whole hippocampus, subcortical regions and cerebellum) large enough to be accurately quantified based on the spatial resolution of the PET scanner (Figure S1a–c). Data are expressed as uptake values normalized to the cerebellum (NUVcb) as previously used in the same mouse model (Serriere et al. 2015).</p><!><p>Animals were anesthetised using isoflurane (3% induction and 1–2% maintenance) and medical oxygen at a rate of 2 L/min. Respiratory rate and temperature were monitored throughout the experiments which were conducted using a 7 Tesla magnet connected to a Bruker Avance III console (Bruker Biospin Ltd, UK). An anatomical multi‐slice FLASH MRI sequence was used to enable positioning of the hippocampal volume for MRS. Spectra were acquired using a water‐suppressed PRESS sequence (Bottomley 1987) (TR 2500 ms, TE 20 ms, 512 averages) from a 2.5 × 4.5 x 3 mm3 voxel that covered the hippocampus and the most dorsal part of the thalamus (Figure S1d). Prior to acquiring the spectrum the localized voxel was shimmed using 'FASTMAP' (Gruetter 1993) and water suppression was optimized using VAPOR (Griffey and Flamig 1990). A non‐water‐suppressed reference PRESS spectrum was also acquired (1 average).</p><p>A metabolite basis set was simulated using NMRScope with the same spectroscopic parameters used for the PRESS acquisition (jMRUI version 5) (Stefan et al. 2009). Metabolites included in the basis‐set were: NAA, Glu, mL, Cre, GABA, scyllo‐inositol (Scy‐I), Gln, Tau and tCho. Peaks at 0.9 and 1.3 ppm were included to model lipid/macromolecules. Another additional peak was added at 3.76 ppm to cover the detection of α‐protons of amino acids not otherwise included in the basis‐set. These additional peaks were added to help with spectral fitting as significant residual signal had previously been found at these resonances (Forster et al. 2013). An example of an in vivo spectrum with labelled metabolites is shown in Fig. 1e. Spectra were pre‐processed by applying a HLSVD (Hankel Lanczos Singular Values Decomposition) filter to suppress the residual water signal (van den Boogaart et al. 1994). Metabolite concentration in vivo was measured using the jMRUI version 5 algorithm QUEST (Ratiney et al. 2005). QUEST compiles the metabolite theoretical signals into a basis‐set, and then fits the signals to the spectra in vivo, allowing detection and measurement. QUEST was run without background handling. Results were referenced to Cr. Referenced data was used to compare metabolite levels in WT compared to TG mice.</p><!><p>Immunohistochemistry was carried out on TG and WT animals (n = 3–5) to visualize integrin αM chain of the Macrophage‐1 antigen (CD11b) (microglial marker), glial fibrillary acidic protein (GFAP) (astrocytic marker), TSPO, 6E10 (Aβ marker), SVA2 (synaptic vesicle marker), MAP2 (microtubule marker) and NeuN (neuronal marker). Animals were culled by isoflurane overdose confirmed by cervical dislocation. The brains were collected, snap frozen using isopentane on dry ice and stored at −80°C. Coronal brain sections (20 μm thick), from the rostral to caudal part of the brain, were taken using a cryostat (Leica CM3050s, Leica Biosystems Nussloch GmbH, Germany) and stored at −80°C. Sections were allowed to defrost and dry at ∼20°C for 20 min and then fixed with 4% paraformaldehyde for 10 min before being washed (6 × 5 min) in phosphate‐buffered saline (PBS) and incubated for 30 min in 2% normal donkey serum and 0.1% Triton X‐100 in PBS to permeabilize and block non‐specific binding. TSPO, 6E10, SV2A and neurogranin immunohistochemistry required an extra step of antigen retrieval done by incubating the slides in 10‐mM citrate buffer at 90°C for 20 min and then washed 2 × 3 min in PBS. Primary antibody incubation was carried out overnight at 4°C with one of the following primary antibodies in 2% normal donkey serum and 0.1% Triton X‐100 in PBS: rat anti‐mouse CD11b (AbD Serotec (MCA711), RRID: AB_321292, 1 : 1000); rabbit anti‐mouse TSPO (Abcam, Cambridge, UK (EPR5384), RRID: AB_10862345, 1 : 250); rabbit anti‐mouse GFAP (DAKO (Z0334), RRID: AB_10013382, 1 : 1000); mouse anti‐human 6E10 amyloid (BioLegend, London, UK (803001), RRID: AB_2564653, 1 : 1000); rabbit anti‐mouse SV2A (Abcam (ab32942), RRID: AB_778192, 1 : 500); chicken anti‐mouse MAP2 (Abcam (ab5392), RRID: AB_2138153, 1 : 1000); rabbit anti‐mouse NeuN (Abcam (ab177487), RRID: AB_2532109, 1 : 500); rabbit anti‐mouse Neurogranin (Abcam (ab23570), RRID: AB_447526, 1 : 500). Following incubation in primary antibody, PBS washes were repeated (3 × 10 min) and incubated with one of the following secondary antibodies was carried out: Alexa Fluor 594 nm Donkey anti‐rat IgG 1 : 500 (for CD11b); Alexa Fluor 488 nm Donkey anti‐rabbit IgG 1 : 500 (for TSPO, GFAP, SV2A); Alexa Fluor 594 nm Donkey anti‐mouse IgG 1 : 500 (for 6E10); Alexa Fluor 488 nm Goat anti‐chicken Double staining was carried out for CD11b+TSPO, CD11b+GFAP and 6E10 + TSPO and to allow the visualization of microglia and astrocytes with TSPO expression and Aβ burden. Double staining was also carried out for MAP2 + NeuN to look at neuronal density.</p><p>Images of the hippocampus and cortex were collected between bregma −2.06 mm and −2.30 mm on an Olympus BX51 upright microscope using a 10 × /0.30 or 20 × /0.50 UPlanFLN objective and captured using a Retiga 6000 Color camera through QCapture Pro 7 Software (QImaging Inc., Surrey, Canada). Specific band pass filter sets were used to prevent bleed through from one channel to the next.</p><!><p>No blinding stricto sensu was performed; however, animals were identified and recorded only using a unique code number during behavioural tests and image analysis and only identified as TG or WT post‐analysis so that the observers could not know whether the animal being analysed was a WT or a TG. Sample size were calculated to n = 9–10 per group using anterior data obtained in our laboratory using the following online tool: https://www.stat.ubc.ca/~rollin/stats/ssize/n2.html (with α = 0.05, β = 0.2, with anticipated mean difference of 8% and SD ~5–8%).</p><p>The data were statistically analysed using GraphPad Prism version 5.04. Behavioural data are expressed as mean ± SEM and imaging data are expressed as mean ± SD. Paired t‐tests were used to determine differences in exploration in phase 1 of both the NOR and NSR tests. Two‐way analysis of variance (anova) were carried out to test the effect of strain and age on DI, percentage of successful alternation in the Y‐maze, metabolite levels and [18F]DPA‐714 uptake in WT and TG mice. Post hoc analysis was carried out using Dunnet's and Sidak's tests. Main effects were considered significant if p ≤ 0.05. Interactions were deemed significant if p ≤ 0.1. Significance was not adjusted for comparisons of multiple region of interest as those were not compared between them, but if a Bonferroni correction had been applied to the DPA‐714 to account for the number of regions analysed, the appropriate adjusted p value would be p < 0.017.</p><!><p>To test whether short‐term working and recognition memory were affected at different ages, Y‐maze and novel recognition tests were carried out (Fig. 2). No significant differences in DI were seen between WT and TG mice in the NOR at 6 months of age (Fig. 2a). Both groups displayed positive DI results suggesting a preference for the novel and indicating good short‐term memory at this age. However, analysis revealed a significant interaction between gene and age (Fig. 2a, p = 0.0225). This effect was because of a significant decrease in cognitive performance in the TG mice compared to the WT mice at 12 months of age as assessed by DI scores (p ≤ 0.01). This effect was not replicated at 18 months of age, whereby both groups displayed low DI scores suggesting poor cognitive performance in both WT and TG by this age. No significant differences were seen in the exploration of identical objects in phase 1 of the NOR (Fig. 2b) indicating that there was no side bias in the test.</p><!><p>Discrimination index (DI) of exploration of the novel and the familiar object (a) and smell (c) in phase 1 and 2 (WT n = 11, TG=15 at 6 months, WT n = 6, TG=9 at 12 months, WT n = 6, TG = 6 at 18 months). Exploration times of right and left objects in phase 1 of the novel object recognition (NOR) (b) and novel smell recognition (NSR) (d) (t‐tests per genotype at each age). Time spent in the presence of predator and non‐predator urine (n = 4) (e). Alternation in the Y‐maze (WT n = 16, TG = 18 at 6 months, WT = 12, TG = 11 at 12 months, WT = 9, TG = 10 at 18 months) (f). Results are shown as mean ± SEM. Statistical analysis was performed using two‐way anovas and Sidak's multiple comparisons post hoc tests (***p ≤ 0.001).</p><!><p>No significant differences in performance were seen between groups in the NSR (Fig. 2c). At 6 months of age, both WT and TG were able to discriminate between novel and familiar smell, demonstrating good working memory and olfaction. Although no side bias was observed in phase 1 (Fig. 2d), high variation was seen in this test from 12 months of age onwards, therefore the olfactory ability of these mice was tested at this age. In the olfaction test, the behaviour to predator and non‐predator urine was tested. Neither WT nor TG spent significantly more time in the arms containing rabbit urine when compared to fox or bobcat urine (Fig. 2e) indicating olfactory dysfunction from 12 months of age.</p><p>TG mice also displayed increased cognitive decline in the Y‐maze test (Fig. 2f). Overall decreased percentages of successful alternation were seen in the TG mice as an effect of gene (p = 0.0001). In addition, a significant interaction was observed between gene and age (p = 0.0233). This effect did not reveal any significant differences in Y‐maze performance at 6 or 12 months of age. In contrast, at 18 months, a significant decrease in the percentage of alternation was observed in TG mice compared to age‐matched WT mice (p ≤ 0.001) indicating accelerated decline in short‐term working memory in TGs at this age.</p><!><p>To assess neuroinflammation differences between APPswe×PS1Δe9 and WT mice, [18F]DPA‐714 NUVcb was compared at 6, 12 and 18 months in the hippocampal+cortical and subcortical ROIs (Fig. 3). Statistical analysis of the cerebellum standard uptake value revealed a significant effect of age only, which the Sidak's post hoc test revealed to be between 6 and 18 months old WT only. No other difference, particularly between WT and TG, could be found in the cerebellum standard uptake value justifying the use of the cerebellum to normalize the uptake values as previously done in this type of study (Serriere et al. 2015; Takkinen et al. 2016). A two‐way anova revealed a significant effect of gene (p = 0.02) and age (p = 0.03) on [18F]DPA‐714 uptake in the hippocampus and cortex, which resulted in a modest but significant increase in [18F]DPA‐714 NUVcb uptake in TG mice at 18 months of age (0.930 ± 0.059) when compared to both age‐matched WT (Fig. 3b, 0.870 ± 0.044; +7%, p = 0.04) and 6‐month‐old TG mice (0.866 ± 0.051; +7%, p = 0.03). These results suggest that both age and disease increase the neuroinflammatory status of TG mice. An increasing effect was seen with age on [18F]DPA‐714 NUVcb values in the other subcortical region (p = 0.0008) and a significant increase in [18F]DPA‐714 NUVcb uptake was seen in other subcortical regions of 18 months old TG mice (0.638 ± 0.034) when compared to 6‐month‐old TG mice (Fig. 3b, 0.726 ± 0.067; +14%, p = 0.002). The same trend was observed in the subcortical regions of WTs but was not significant (Fig. 3c, p = 0.058).</p><!><p>Positron emission tomography (PET) images showing [18F]DPA‐714 uptake in WT and APP swe×PS1Δe9 mice at 6 (WT n = 10, TG=7), 12 (WT n = 8, TG=9) and 18 months (WT n = 10, TG=9) of age (a). NUV cb values in the hippocampus and cortex (b) and other subcortical regions (c). Results are expressed as mean±SD. Statistical analysis was performed using two‐way anova followed by Sidak's and Dunnet's post hoc analysis (*p ≤ 0.05, **p ≤ 0.01).</p><!><p>Single voxel 1H MRS was repeated in the same cohort of TG and WT mice at 6, 12 and 18 months of age to investigate changes in metabolite profile. A 3 × 3×3 mm voxel was placed to encompass the hippocampus and the most dorsal part of the thalamus. Example spectra from this region can be seen in Fig. 4a. No significant differences were seen in Cr concentration referenced to water at any age group for either WT or TG mice (Fig. 4d), allowing metabolite data to be expressed as a ratio to Cr. This has been previously reported in this model (Jansen et al. 2013); however, it was important to assess this as elevated Cr levels, compared to WT mice, have previously been reported in the double‐mutant APPswe × PS1.M146V (TASTPM) mouse model of AD in vivo at an early age (Forster et al. 2013) and at a later age analysing brain extracts in vitro (Forster et al. 2012). No overall significant effect of gene alone was seen on any metabolite in our study. However, a significant effect of age (p = 0.0006) and a significant interaction between gene and age (p = 0.0866) were observed for NAA. This effect resulted in significantly lower NAA in 18‐month‐old TG mice compared to 6‐month‐old TG mice using multi‐comparisons analysis (−58%, Fig. 4c, p ≤ 0.0001). This ageing effect was seen in the WT mice (−20%) but was not statistically significant. Significant ageing effects were also observed on Glu (p = 0.0003) and tCho (p = 0.0016) levels, resulting in reduced Glu (−53% average across groups from 6 to 18 months) and increased tCho (+71% average across groups from 6 to 18 months) levels with age. No gene effect or gene×genotype interaction was associated with these changes suggesting that these alterations are an effect of normal ageing.</p><!><p>Example spectra from WT and TG mice at 6, 12 and 18 months (a). Metabolites expressed as ratios to creatine in WT (n = 6) (b) and TG (n = 8) (c) mice at 6, 12 and 18 months of age. No differences were observed in the Cr concentration (mean ± SD) referenced to water (d). Reduced Glu/Cr and increased tCho levels were observed with age in both WT and TG (independently of genotype). Results are shown as mean±SD. Statistical analysis was performed using two‐way anova followed by Dunnet's post hoc analysis. (****p ≤ 0.0001).</p><!><p>To verify the in vivo imaging results, ex vivo immunohistochemistry was performed to assess the presence of neuroinflammation, Aβ burden and neuronal integrity. GFAP‐ and CD11b‐positive staining were seen in the hippocampus (Fig. 5a) and cortex (Fig. 5b) of TG mice but not WT mice. Low levels of immunostaining were evident at 6 months of age and increased with age in both regions in TG mice. Similarly, double staining for TSPO and CD11b revealed an increase in both proteins with age in the hippocampus (Fig. 6a) and cortex (Fig. 6b) of the TG mice. In the TG mice, there was regional co‐expression of TSPO and CD11b in the hippocampus and cortex, with a modest expression at 6 months of age increasing at 12 months and further at 18 months. In WT animals, TSPO staining was only seen in the vessels and not the parenchyma in both regions. No CD11b staining was evident in the WTs at any age. Immunostaining for TSPO and Aβ pathology (6E10) revealed similar results. An age‐dependent increase was seen in Aβ burden in both the hippocampus (Fig. 7a) and cortex (Fig. 7b) of TG mice only. Aβ staining was sparse at 6 months but increased with age revealing a heavy burden by 18 months. CD11b, TSPO and 6E10 demonstrated good regional co‐expression from 6 to 18 months in the cortex of TG mice, indicating increased microglial activation around Aβ plaques. No Aβ staining was evident in the WT mice at any age. No differences were seen in staining for MAP2, NeuN (Fig. 8a‐b) and SV2A and neurogranin (Figure 9a and b) between TG and WT mice or with age in the hippocampus or cortex indicating that neuronal death could not be detected in this model with this method.</p><!><p>Immunoreactivity of glial fibrillary acidic protein (GFAP) (green) and CD11b (red). Representative images of double staining in the hippocampus (a) and cortex (b) of WT and APP swe×PS1Δe9 mice at 6, 12 and 18 months of age. Pictures were taken at 10 × magnification between bregma −2.06 mm and −2.30 mm. Scale bar represents 200 μm.</p><p>Immunoreactivity of translocator receptor 18 kDa (TSPO) (green) and CD11b (red). Representative images of double staining in the hippocampus (a) and cortex (b) of WT and APP swe×PS1Δe9 mice at 6, 12 and 18 months of age. Pictures were taken at 10 × magnification between bregma −2.06 mm and −2.30 mm. Scale bar represents 200 μm.</p><p>Immunoreactivity of translocator receptor 18 kDa (TSPO) (green) and Aβ (red). Representative images of double staining in the hippocampus (a) and cortex (b) of WT and APP swe×PS1Δe9 mice at 6, 12 and 18 months of age. Pictures were taken at 10 ×  magnification between bregma −2.06 mm and −2.30 mm. Scale bar represents 200 μm.</p><p>Immunoreactivity of MAP2 (green) and NeuN (red). Representative images of double staining in the hippocampus (a) and cortex (b) of WT and APP swe×PS1Δe9 mice at 6, 12 and 18 months of age. Pictures were taken at 10 × magnification between bregma −2.06 mm and −2.30 mm. Scale bar represents 200 μm.</p><p>Immunoreactivity of neurogranin (red) and SV2A (green). Representative images of single SVA2 staining in the hippocampus (a) and cortex (b) and neurogranin staining in the hippocampus (c) and cortex (d) of WT and APP swe×PS1Δe9 mice at 6, 12 and 18 months of age. Pictures were taken at 10 × magnification between bregma −2.06 mm and −2.30 mm. Scale bar represents 200 μm.</p><!><p>Cognitive performance in the TG group at 6 months of age was comparable to WT mice and has been previously reported in a variety of memory‐based tests in this (Chen et al. 2012; Jansen et al. 2013) and other AD models (Nagakura et al. 2013; Webster et al. 2013). This supports the low levels of Aβ load observed at this age in this and other studies (Jankowsky et al. 2004; Garcia‐Alloza et al. 2006). TG mice displayed deficits in non‐associative recognition memory and working memory from 12 months, with reduced cognitive performance in the NOR test and Y‐vmaze by 12 and 18 months respectively. Cognitive deficits in these tests have been previously reported in both this (Petrov et al. 2015) and other amyloid‐based models of AD (Forster et al. 2013; Daniels et al. 2016; Martins et al. 2017), further supporting the link between pathology burden and cognitive decline. WT mice retained memory later in both tests, displaying increased working memory in the Y‐maze test compared to TG at 18 months. However, no differences in cognition were identified in the NSR test. Both WT and TG were able to discriminate between novel and familiar scents at 6 months of age, demonstrating good working memory and olfaction. However, high variation in both groups was observed in this test from 12 months onwards, therefore we hypothesized that these animals may have altered olfaction by this age, which was proven by the lack of preference of exposure to non‐predator (rabbit) urine over predator (bobcat and fox) urine. An increased amount of time in the arm with non‐predator urine would be expected in mice with intact olfaction; hence the changes observed our NSR study are most likely down to an alteration in olfaction rather than cognitive deficits. These results are, in part, in line with previous publications showing various alterations of olfaction in Tg2576 (Wesson et al. 2011) and in APPswe×PS1Δe9 mice (Yao et al. 2016) using different tests and with clinical reports showing that loss of olfaction is common in neurodegenerative diseases and is an early sign of AD (Mesholam et al. 1998). We, however, also observed a loss of olfaction in our WT mice. Similarly, some previous studies reported early cognitive dysfunction prior to pathology development using different behavioural assessments such as the Morris water maze (Zhang et al. 2012). Here, we report decline in memory with age in APPswe×PS1Δe9 mice, which is in support of our working hypothesis that cognitive performance decreases with age at a faster rate in the TG mice compared to the WT.</p><p>It has been hypothesized that neuroinflammation precedes clinical manifestation of cognitive dysfunction in AD. In this study, we observed the presence of neuroinflammation in TG mice via immunohistochemistry (CD11b and GFAP) from 6 months which increased with age. However, increased neuroinflammation and altered metabolite profile, assessed via in vivo PET and MRS, respectively, occurred after the emergence of cognitive decline in this animal model.</p><p>Increased [18F]DPA‐714 NUVcb was found in the hippocampal and cortical region of old TG mice when compared to both age‐matched WT mice and young TG mice, suggesting that increased neuroinflammation in hippocampal and cortical region is driven by the presence of AD pathology in TG mice. This increase was only statistically significant at 18 months of age despite increased Aβ, GFAP and CD11b immunostaining evident from 12 months of age. Although a significant increase is also seen with age in [18F]DPA‐714 uptake in the other subcortical regions, no significant differences were found between genotypes. A trend towards significance was also observed in the WTs in the subcortical ROI. As these regions have little to no amyloid burden, these data suggest that neuroinflammation in these regions is increased mostly as a result of normal ageing and independently of amyloid pathology. It is, however, notable that Yokokura et al. (2017) recently reported increased uptake in the thalamus of AD and elderly patients using [11C]DPA‐713. These results suggest that neuroinflammation is present prior to cognitive dysfunction; however, current in vivo methods are not sensitive enough (because of a combination of tracer sensitivity, resolution capabilities of small animal PET and non‐specific relevance of the mI signal) to detect the more subtle inflammatory changes in the earlier stages of disease that are achievable via ex vivo methods.</p><p>Our data are in agreement with previous reports in mouse models using [18F]DPA‐714. Serriere et al. (2015) investigated [18F]DPA‐714 uptake in the same mouse model at 6, 9, 12, 15 and 19 months of age. NUVCb were increased in the cortex of TG mice at 12 and 19 months (but not 15 months of age) and in the hippocampus at 19 months of age. Increased cortical [18F]DPA‐714 NUVcb has also been reported in the APP/PS1‐21 mouse from 6 months of age, with hippocampal increases emerging from 12 months of age (Takkinen et al. 2016). In our analysis methods, we pooled cortical and hippocampal areas as one region because of AD pathology present in both regions and because the size of the pooled ROI is more realistically compatible with the resolution of PET imaging in preclinical scanners (~1–1.6 mm). Therefore, by pooling the hippocampus and cortex into a single ROI, we may have masked the possible differential time‐effect between each region, and hidden the earlier increase in neuroinflammation in the cortex. However, such difference would be small as our immunohistochemistry results show only a slightly stronger signal in CD11b and TSPO in the cortex than in the hippocampus in this model at 6 and 12 months of age. On the other hand, it can also be argued that thin ROIs including only the cortical hippocampal areas (~1.5 mm thick) are far more subject to partial volume effects, which could consequently have biased the quantification in previous studies. A significant effect of both age and gene on TSPO PET has been also recently reported in the same model using [18F]GE‐180 (Liu et al. 2015). Increased [18F]GE‐180 uptake was seen in the hippocampus of old TGs compared to age‐matched WTs and young TG and WT mice. In addition, increased [18F]GE‐180 uptake was also seen in the whole brain of old TGs compared to young TGs. This age effect was replicated in WT mice with significant increases in uptake from young to old mice in both hippocampal and whole brain, implicating both normal ageing and AD pathology on neuroinflammatory status. Increased neuroinflammation has also been observed as an effect of ageing in both humans (Gulyas et al. 2011; Kumar et al. 2012; Yokokura et al. 2017) and in the WTs of other AD models (Brendel et al. 2016). Altogether, our results further support previous findings that age can significantly alter microglial responses, which are modified further in the presence of neurodegenerative diseases such as AD. Increases in TSPO expression have been found in other models of AD (Ji et al. 2008; James et al. 2015; Brendel et al. 2016; Mirzaei et al. 2016) and in human AD (Forlenza et al. 2009; Swardfager et al. 2010; Rubio‐Perez and Morillas‐Ruiz 2012; Varnum and Ikezu 2012), indicating that elevated neuroinflammation is a consistent characteristic of this disease. The modest increases and the relative overlap between WT and TG mice demonstrated by the present data and others' are also in agreement with the PET quantification in AD patient versus healthy controls which generally reports non‐significant (Varrone et al. 2013) or small to more substantial increases (+10−35%) (Edison et al. 2008; Okello et al. 2009; Schuitemaker et al. 2013; Varrone et al. 2015; Hamelin et al. 2016) in neuroinflammation. This demonstrates that (i) measuring neuroinflammation in vivo in AD is challenging because the amplitudes of changes are overall modest when compared with the changes induced by ageing only and (ii) that, at least from this point of view, animal models are actually reflecting the clinical situation quite well. Finally, a recent study by Owen et al. (2017) suggests that, at least in human, TSPO expression may reflect microglia density rather than microglia phenotype. Although obtained from a purely in vitro setting in which cells may behave differently than in their in vivo environment, these observations taken together with the known TSPO expression by endothelial cells may explain the inconsistencies between imaging studies and the difficulties encountered over the years to truly assess microglial activation in AD patients using TSPO PET.</p><p>In contrast to increased neuroinflammation assessed by PET and immunohistochemistry, no alterations in mI levels were observed between TG and WT mice in this study. This lack of agreement between neuroinflammatory status and mI expression questions the role of mI as a putative marker for gliosis with the specific biological significance of mI needing further investigation to be truly elucidated. In support of this, reported mI level alterations are inconsistent in this model, with both increased and stable levels reported (Dedeoglu et al. 2004; von Kienlin et al. 2005; Jansen et al. 2013). Moreover, in a clinical study, mI was found to be associated with amyloid pathology and not neuroinflammation (Murray et al. 2014), therefore it is possible that mI levels represent more closely amyloid load than neuroinflammation. On the other hand, we have extensive amyloid burden in this model without seeing mI alterations. These results are in contrast to many other reports demonstrating increased mI levels in clinical AD (Kantarci et al. 2007; Shinno et al. 2007; Foy et al. 2011; Shiino et al. 2012; Murray et al. 2014) and AD models (Marjanska et al. 2005, 2014; Jack et al. 2007; Oberg et al. 2008; Chen et al. 2009, 2012; Choi et al. 2010, 2014; Yang et al. 2011; Forster et al. 2013). Chen et al. (2009) reported significantly higher mI/Cr levels in the APPswe×PS1Δe9 compared to WT mice as early as 3 months of age. However, the voxel used in this study included cortical as well as hippocampal tissue. Here, we report the appearance of inflammation and plaques in cortical regions prior to and more frequently than in hippocampal regions, so the discrepancies between studies might be because of differences in size and location of the voxel used as well as the age studied.</p><p>Although no qualitative changes were observable in neuronal markers (MAP2 and NeuN) using immunohistochemistry, a significant effect of ageing and a significant genotype×age interaction was observed on NAA levels, resulting in a ‐58% decrease in NAA/Cr in 18‐month‐old TG mice when compared to 6‐month‐old TG. This effect was not mirrored in the WT mice and suggests that decreased NAA levels, although influenced by both age and gene, are slightly more pronounced in TG mice. This underlines the importance of understanding the pattern of normal ageing within brain metabolites or other potential biomarkers and the possible confounding effects of age.</p><p>Chen et al. (2009) reported a small but significant decrease (−11% change, 1.16 ± 0.07 in WT vs. 1.03 ± 0.06 in TG) in NAA/Cr from 5 months of age. In contrast, we report NAA alterations post‐cognitive decline and is in line with the results from Xu et al. (2010) in the same animal model. Xu et al. (2010) found a significant decrease in hippocampal NAA/Cr in 16‐month‐APPswe×PS1Δe9 TG mice compared to younger TG mice, which was not seen in WT mice. Similar decreases in NAA/Cr associated with age and AD pathology have been reported in other models of AD, at varying ages (Marjanska et al. 2005; Jack et al. 2007; Oberg et al. 2008; Choi et al. 2010, 2014; Xu et al. 2010; Forster et al. 2013; Jansen et al. 2013). We also report age to have a significant effect on Glu and tCho levels without an effect of genotype effect or an interaction genotype × age, in which Glu levels were decreased and tCho levels increased in both WT and TG mice. Similarly, specific reduced Glu and increased tCho levels have been reported in mouse models of AD (Dedeoglu et al. 2004; von Kienlin et al. 2005; Marjanska et al. 2005; Jack et al. 2007; Oberg et al. 2008; Choi et al. 2010; Chen et al. 2012; Esteras et al. 2012) and in the clinical situation (Hattori et al. 2002; Kantarci 2007; Griffith et al. 2008; Foy et al. 2011; Shiino et al. 2012) versus WT or healthy controls respectively. This result suggests that these effects are a result of normal ageing in this strain.</p><!><p>To verify the in vivo imaging results, ex vivo immunohistochemistry was performed to assess the presence of microglia, astrocytes, TSPO expression and Aβ pathology. Aβ pathology was sparse in young TG mice with increasing burden evident by 12 months and abundant plaques by 18 months. This level of Aβ burden is in line with many previous reports in this model (Jankowsky et al. 2004; Garcia‐Alloza et al. 2006; van Groen et al. 2006). Similarly, low levels of CD11b, GFAP and TSPO staining were evident in the hippocampus and cortex of TG mice at 6 months, which increased with age. Co‐localization of CD11b and TSPO was observed in the cortex and the hippocampus, confirming that in this model, TSPO is mostly expressed by microglia. Similarly, TSPO and Aβ co‐localization was observed in the cortex of TG mice, indicating increased glial activation around Aβ plaques. Similarly, strong astrogliosis was detected around microglial cells which were found only around Aβ plaques supporting the presence of both astrocytes and microglia around Aβ plaques. Microglial cells are known to surround plaques in human AD and have been shown to co‐localize with pathology in AD models further supporting the role of neuroinflammation in AD. In contrast to our results, Ji et al. (2008), did not find co‐expression of CD11b and TSPO in the APP23 mouse model, but found co‐localization of GFAP and TSPO which were in close proximity to amyloid plaque staining. However, the opposite was identified in a tau model, with CD11b but not GFAP‐positive cells expressing TSPO. These results suggest that different forms of tau and amyloid pathology may alter TSPO expression differently in vivo (Jansen et al. 2013). This may be because of potential differences in type of amyloid, aggregation properties, expression levels and age of pathology emergence between different models of AD, which may differently affect the glial response (Stalder et al. 1999; Xiong et al. 2011; Huang et al. 2016). In contrast, TSPO expression was only visible in the vessels in WT animals which are in agreement with previous report demonstrating the expression of TSPO in vessels (Turkheimer et al. 2007).</p><p>We did not identify any striking reductions in neuronal, synaptic vesicle or microtubule markers in this study. This is in line with previous reports (Jansen et al. 2013) suggesting that APP mutations are not sufficient to cause neuronal loss that is observed in human AD and models with tau abnormalities but in contrast with Huang et al. (2016) who reported a decrease (~20%) in NeuN staining and hippocampal atrophy (~20–30%). On the other hand, amyloid pathology induced cognitive decline and resulted in accelerated NAA loss with age in TG compared to WT mice. This lack of congruency could be explained by many factors such as assessing the wrong neuronal/synaptic markers, changes being too small or regionally specific or as a result of changes in certain neuronal populations (e.g. cholinergic) to be detectable using the methods and markers used here.</p><!><p>It has become evident that Alzheimer's disease is a complex multifactorial disease in which neuroinflammation plays a pivotal role. In support of this, we report increased neuroinflammation in the form of increased [18F]DPA‐714 uptake, confirmed through the use of immunohistochemistry, and reduced neuronal function in the form of accelerated NAA reductions in the APPswe×PS1Δe9 similar to those measured in AD patients. We here show that TSPO PET can detect changes in neuroinflammation in this mouse model, however, not as early as detected using ex vivo techniques. This suggests that, although TSPO PET is a viable imaging technique to study AD in animal models, the current restrictions because of resolution and brain size in mice hamper earlier detection. Moving to larger species such as rats may address this. Overall, these results support the role of neuroinflammation in the pathogenesis of AD and the potential use of metabolite alteration to monitor disease progression or response to treatments.</p><!><p>Figure S1. Whole brain was segmented into the hippocampus and cortex (a), other subcortical (b) and cerebellum (c) regions of interest for PET quantification using a modified version of the Waxholm space template (Johnson et al. 2010). The MRS voxel was centred at bregma ‐2.30 mm according to the Paxinos mouse brain atlas and encompassed hippocampal and thalamic regions (d). Example of an MRS spectrum in vivo with the main metabolites highlighted (e). AAP‐amino acid proton, mI‐myo‐Inositol. Tau‐taurine, Cho‐choline containing compounds, Cre‐ creatine+phosphocreatine, Glu‐ glutamate, GABA‐ γ‐aminobutyric acid, NAA‐ N‐acetylaspartate, Lipid MMS‐ lipid and macromolecules.</p><p>Figure S2. [18F]DPA‐714 standard uptake value in the cerebellum of WT and APPswe×PS1Δe9 mice at 6 (WT n = 10, TG=7), 12 (WT n = 8, TG=9) and 18 months (WT n = 10, TG=9) of age. Results are expressed as mean±SD. Statistical analysis was performed using two‐way ANOVA followed by Sidak's post hoc analysis (*p ≤ 0.05).</p><p>Figure S3. Immunoreactivity of GFAP (green) and CD11b (red). Representative images of double staining in the hippocampus (a) and cortex (b) of WT and APPswe×PS1Δe9 mice at 6, 12 and 18 months of age. Pictures were taken at 20 ×  magnification between bregma −2.06 mm and −2.30 mm. Scale bar represents 200 μm.</p><p>Figure S4. Immunoreactivity of TSPO (green) and CD11b (red). Representative images of double staining in the hippocampus (a) and cortex (b) of WT and APPswe×PS1Δe9 mice at 6, 12 and 18 months of age. Pictures were taken at 20 ×  magnification between bregma −2.06 mm and −2.30 mm. Scale bar represents 200 μm.</p><p>Figure S5. Immunoreactivity of MAP2 (green) and NeuN (red). Representative images of double staining in the hippocampus (a) and cortex (b) of WT and APPswe×PS1Δe9 mice at 6, 12 and 18 months of age. Pictures were taken at 20 ×  magnification between bregma −2.06 mm and −2.30 mm. Scale bar represents 200 μm.</p><p>Click here for additional data file.</p><p>Data S1. Supplementary materials.</p><p>Click here for additional data file.</p>

PubMed Open Access

Reactivity of Nucleic Acid Radicals

Nucleic acid oxidation plays a vital role in the etiology and treatment of diseases, as well as aging. Reagents that oxidize nucleic acids are also useful probes of the biopolymers\xe2\x80\x99 structure and folding. Radiation scientists have contributed greatly to our understanding of nucleic acid oxidation using a variety of techniques. During the past two decades organic chemists have applied the tools of synthetic and mechanistic chemistry to independently generate and study the reactive intermediates produced by ionizing radiation and other nucleic acid damaging agents. This approach has facilitated resolving mechanistic controversies and lead to the discovery of new reactive processes.

reactivity_of_nucleic_acid_radicals

1. INTRODUCTION<!>2. RADICAL FORMATION IN NUCLEIC ACIDS<!>3. THE NORRISH TYPE I PHOTOREACTION<!>4.1 C1\xe2\x80\xb2-Radical Formation<!>4.2 C1\xe2\x80\xb2-Radical Reactivity<!>4.3 Utility of C1\xe2\x80\xb2-Radical Generation as a Source of 2-Deoxyribonolactone in Mechanistic Studies<!>4.4 Probing DNA Repair Enzyme Activity Using Independently Generated 2\xe2\x80\xb2-Deoxyuridin-1\xe2\x80\xb2-yl Radical (3)<!>5.1 C2\xe2\x80\xb2-Radical Formation Following Irradiation of 5-Halopyrimidine Nucleotides in DNA<!>5.2 Generation and Reactivity of the 2\xe2\x80\xb2-Radical in RNA<!>5.3 C2\xe2\x80\xb2-Radical Formation and Reactivity Following Irradiation of 5-Bromouridine in RNA<!>6. C3\xe2\x80\xb2-RADICAL GENERATION AND REACTIVITY IN DNA<!>6.1 C3\xe2\x80\xb2-Radical Formation Following Irradiation of Transition Metal Coordination Complexes<!>6.2 Independent Generation and Reactivity of Thymidin-3\xe2\x80\xb2-yl Radical (35)<!>7.1 C4\xe2\x80\xb2-Radical Formation<!>7.2 C4\xe2\x80\xb2-Radical Reactivity in DNA<!>7.3 Independent Generation and Reactivity of C4\xe2\x80\xb2-Radicals in DNA<!>7.4 Double-Strand Cleavage via a Single C4\xe2\x80\xb2-Radical<!>7.5 The Role of Independent C4\xe2\x80\xb2-Radical Generation in Understanding Electron Transfer in DNA<!>7.6 C4\xe2\x80\xb2-Radical Reactivity in RNA<!>8.1 C5\xe2\x80\xb2-Radical Formation<!>8.2 C5\xe2\x80\xb2-Radical Reactivity in DNA<!>8.3 Independent Generation and Reactivity of C5\xe2\x80\xb2-Radicals<!>9.1 Nucleobase Radical Formation<!>9.2 Nucleobase Radical Reactivity<!>9.3 Independent Generation and Reactivity of DNA Nucleobase Radical Adducts<!>9.4 Independent Generation and Reactivity of RNA Nucleobase Radical Adducts<!>9.5 Independent Generation and Reactivity of DNA 5-(2\xe2\x80\xb2-Deoxyuridinyl)methyl and 5-(2\xe2\x80\xb2-Deoxycytidinyl) methyl Radicals<!>9.6 Independent Generation and Reactivity of Neutral Purine Radicals<!>10. SUMMARY AND FUTURE CONSIDERATIONS

<p>Within physical organic chemistry, independent generation of reactive intermediates is a powerful method for proving their intermediacy in chemical processes and unambiguous characterization of their reactivity.1 Photochemistry is often the method of choice and under appropriate conditions the use of lasers and spectroscopic methods (eg ultraviolet (UV)–visible absorption, infrared) together enables their direct observation and kinetic characterization.2,3 In the absence of laser flash photolysis, product analysis, sometimes in conjunction with isotopic labelling, and competitive kinetics experiments of reactive intermediates generated under steady-state conditions have shed valuable light on their reactivity. For instance, investigations of independently generated carbon-centred reactive species, including radicals, radical ions and carbenes, have enhanced our understanding of the effects of substituents on reactivity, the effects of structure on ground state spin states and the effects of the latter on reactivity.4–6 A greater understanding of the connection between reactive intermediate structure and reactivity facilitates their use in organic synthesis and novel materials.7–11 During the past two decades, our understanding of oxidative damage of nucleic acids (DNA, RNA) has been greatly improved by independently generating reactive intermediates. Experiments carried out on nucleosides and oligonucleotides have resolved mechanistic controversies and uncovered novel chemical pathways.12–16</p><p>The importance of nucleic acid damage in aging, disease development, as well as the treatment of cancer, provides a part of the motivation for such investigations.17–21 However, nucleic acid oxidation is also useful for determining RNA structure and its folding dynamics, and may have applications in material science.22–26 Reactive oxygen species, especially hydroxyl radical (HO•), play a role in the therapeutic aspects of nucleic acid damage and disease etiology.17–19,27 Hydroxyl radical is also a powerful probe for determining biomacromolecular interactions and RNA folding dynamics.28–31 Much has been learned about the reactivity of HO• with DNA and RNA using various forms of ionizing radiation in conjunction with product analysis and various spectroscopic methods, including electron paramagnetic resonance (EPR) to detect radical intermediates.17,32 These investigations are limited by the lack of control over HO• reactivity, resulting in heterogeneous mixtures of reactive intermediates. Consequently, the formation of putative reactive intermediates produced by HO• from synthetic precursors both simplifies elucidating the chemistry of this important reactive oxygen species, and facilitates revealing complexities hidden by the formation of multiple reactive species within the biopolymers.</p><!><p>Radicals are directly produced in nucleic acids predominantly via hydrogen atom abstraction from the carbohydrate moiety or radical addition to the nucleobase ρ-bonds.17,32 Hydroxyl radical is produced by metal complexes, most notably Fe•EDTA, and is a major source of DNA damage by γ-radiolysis.32,33 Hydrogen atom abstraction by diffusible species such as HO• is believed to be governed by solvent accessibility and not bond dissociation energies due to the radical's high reactivity. Solvent accessibility favors reaction at the C4′- and C5′-positions (Fig. 1A).34 Computational experiments also favor C5′-hydrogen atom abstraction in DNA, followed by reaction at C4′ and C1′ (Table 1).35 However, the C1′-hydrogen atom is buried in the minor groove and inaccessible to diffusible species such as HO•. Small molecules that bind in the minor groove of DNA, many of which have antitumor activity can access the C1′-position, as well as the hydrogen atoms bonded to the C4′- and C5′-carbons.36 Hydrogen atoms bonded to the C2′-carbon in DNA (Fig. 1) have considerably stronger bond dissociation energies (Table 1), and HO• is the rare exception of an oxidant that reacts at this position. Another example is the σ-radical derived from 5-halo-2′-deoxyuridines.37–43 The C3′-hydrogen atom (Fig. 1B) is abstracted less frequently due to the smaller number of oxidants that bind in the major groove of DNA, and possibly a surprisingly high calculated bond dissociation energy (BDE) (Table 1). The weaker bond strengths of the C2′- and C3′-carbon–hydrogen bonds in RNA (Table 1) compared to DNA are the greatest differences in potential hydrogen atom abstraction sites between the two biopolymers.</p><p>Direct hydrogen atom abstraction occurs less frequently from the nucleobases, despite the expected modest carbon–hydrogen bond dissociation energy of the carbon–hydrogen bonds in the methyl groups of thymidine and 5-methyl-2′-deoxycytidine due to resonance stabilization of the incipient radicals. The respective radicals are also formed by deprotonation of the nucleobase radical cations, intermediates involved in electron transfer that are produced via one-electron oxidation. Amine radicals are also postulated as intermediates produced from the spontaneous decomposition of chloramines that arise from reactions of nucleosides with hypochlorous acid.44 However, the majority of nucleobase radical intermediates arise from the addition to nucleobase ρ-bonds. In fact, this is the kinetically preferred pathway for HO•. Although estimates vary, nucleobase addition may account for more than 90% of the reactions between nucleic acids and HO•.17 Significantly more data are available concerning the reactivity of pyrimidine nucleobases than purines. In fact, as discussed later, many questions regarding the reactivity of purine nucleobases remain.</p><!><p>A considerable number of examples described below in which nucleic acid radicals are independently generated take advantage of the α-photo-cleavage (Norrish type I) of ketones (Scheme 1 ).45 Most of the examples that will be cited involve photolysis of t-butyl or benzyl ketones (R′). This is consistent with the general quantum yield efficiency and rate constant for α-cleavage of the triplet excited state ketone. Consideration of the rate constant for decarbonylation of the ground state acyl radical is also relevant, with the efficiency of decarbonylation correlating with radical stability.46</p><!><p>Abstraction of the C1′-hydrogen atom in duplex DNA by diffusible species (eg HO•) is limited by its position in the minor groove (Fig. 1A), despite the relatively modest C1′-H BDE (Table 1). The solvent inaccessibility of the C1′-hydrogen is overcome by DNA oxidizing agents that bind in the minor groove. For instance, the antitumor agent neocarzinostatin (NCS), which forms a biradical upon reductive activation in the minor groove abstracts hydrogen atoms from the C1′-position.47 Activated forms of some coordination compounds, such as manganese porphyrins (MnPy) and copper bis-phenanthroline (Cu•OP2) also abstract the C1′-hydrogen atom.48–50</p><p> </p><p>The C1′-radical is believed to form indirectly via reactions of initially formed intermediates. For instance, photolysis of menadione (2-methyl-1,4-napthoquinone, MD) in the presence of 2′-deoxycytidine (dC) produces the pyrimidine radical cation (1, Scheme 2 ).51,52 The radical cation of dC is proposed to yield the C1′-radical (2) upon deprotonation, which ultimately leads to 2-deoxyribonolactone (L), presumably via a mechanism discussed in more detail below. This process has not been detected in DNA, possibly because 1 is too short lived for deprotonation to compete with hole migration (electron transfer). Other pathways that do not involve radical cations produce 2-deoxyribonolactone via C1′-hydrogen atom abstraction by nucleobase radical adducts and are discussed in more detail below. Irradiation of DNA containing 5-bromo-(BrdU) or 5-iodo-2′-deoxyuridine (IdU) yields the highly reactive σ-radical (uracil-5-yl radical), which abstracts the C1′-hydrogen (and C2′-hydrogen) atom from the 5′-adjacent nucleotide.37,38,53,54 (This topic is discussed in more detail below in the section concerning C2′-radical reactivity.)</p><p> </p><!><p>The mechanism for transformation of the C1′-radical to 2-deoxyribonolactone (L) under aerobic conditions was examined using a photochemical precursor to generate 3 directly (Scheme 3 ).55–58 2′-Deoxyuridin-1′-yl radical (3) was generated via Norrish type I photocleavage of 4. Steady-state and laser flash photolysis experiments supported transformation of the C1′-radical into 2-deoxyribonolactone (L) under aerobic conditions via the carbocation (6). Superoxide formation was detected spectrophotometrically during steady-state generation of 3 from 4. However, the use of competitive kinetics using thiol and isotopic labelling (H218O) in conjunction with one another under steady-state photolysis conditions resulted in a gross underestimation of the rate constant for superoxide elimination from 5.57 Using laser flash photolysis, Newcomb detected the release of superoxide via its reduction of tetranitromethane, which is observed directly (350 nm).58 Deconvolution of these data yielded a rate constant for superoxide elimination from 5 of ~1 × 104 s−1, which is comparable to the rate constants reported for similarly substituted peroxyl radicals.59 Although comparable experiments were not reported in DNA, thiol trapping of 5 would not be expected to compete with superoxide elimination, suggesting that release of the reactive oxygen species will accompany 2-deoxyribonolactone (L) formation from 2′-deoxyuridin-1′-yl radical (3) under aerobic conditions.</p><p>Laser flash photolysis of 4 under anaerobic conditions provided rate constants for thiol trapping of 2′-deoxyuridin-1′-yl radical (3, Scheme 4 ) by following the decay of the 3 at 320 nm β-Mercaptoethanol (BME), cysteine and glutathione (GSH) reacted with 3 between 2 and 4 × 106 M−1s−1.58 The rate constants for thiol trapping of 3 are slightly lower than those typically reported for reactions with other alkyl radicals, and may be a consequence of the stabilization of 2′-deoxyuridin-1′-yl radical (3) by two α-heteroatoms.60 These absolute rate constants were corroborated in single- and double-stranded DNA by competitive kinetic studies under aerobic conditions in which BME (and separately dithiothreitol) concentration was varied and the ratio of α,β-2′-deoxyuridine versus 2-deoxyribono-lactone (L) used to estimate the rate constant thiol trapping of 3.61 (Please note that reactive intermediates are referred to by the same descriptor whether they are monomeric or within biopolymers throughout this chapter.) The stereoselectivity of thiol reduction of 3 was also determined in single- and double-stranded DNA by generating the radical under anaerobic conditions. Product ratios were determined by HPLC analysis of nucleosides released upon enzymatic digestion of the DNA. β-2′-Deoxyuridine was favoured over the anomer by BME and dithiothreitol in single- and double-stranded DNA. The ratio of β-2′-deoxyuridine (β-dU) to α-2′-deoxyuridine (α-dU) varied between 4.1 and 4.5 in single-stranded DNA, but increased to between 6.2 and 8.3 in double-stranded substrates. Preferential formation of β-dU from 3 is relevant to the role of thiols as radioprotecting agents.62,63 Thiols not only need to compete with O2 for radicals but they need to restore the nucleic acids to guard against formation of potential pro-mutagenic nucleotides, such as α-dU.</p><!><p>Independent generation of 2′-deoxyuridin-1′-yl radical (3) from 4 and its transformation into 2-deoxyribonolactone (L) under aerobic conditions provided a valuable tool for mechanistic studies on nucleic acid damage, although more efficient methods for generating L were subsequently developed.64–67 As noted above, a variety of DNA damaging agents abstract the C1′-hydrogen atom, ultimately producing L. The lactone is an example of an alkali-labile lesion, indicating that it yields strand breaks upon treatment with mild base. In fact, L is so labile that it yields strand breaks upon incubation in aqueous buffer, albeit with a half-life on the order of days.68,69 The copper bis-phenanthroline complex (Cu•OP2) was a notable exception. Although C1′-oxidation was the predominant pathway proposed by its pioneering discoverer, David Sigman, the copper complex yielded direct strand breaks (Scheme 5 ).70 Despite the absence of L in DNA damaged by Cu•OP2, the formation of 5-methylene-2-furanone (8), which could be construed to arise from 2-deoxyribonolactone, as well as a labile intermediate detected by gel electrophoresis that could be butenolide 7 were consistent with C1′-oxidation.50,71</p><p>The absence of L and formation of direct strand breaks when DNA is treated with Cu•OP2 was investigated by several laboratories. Several reports by Sigman implied that 2-deoxyribonolactone (L) was an intermediate en route to strand scission.49,50 However, subsequent mechanistic studies using C1′-deuterated DNA and 18O-labelling led to a mechanism that avoids formation of L (Scheme 6 ).72,73 Although this mechanism still began with C1′-hydrogen atom abstraction, the initially formed radical was oxidized to the carbocation, which then yielded a ketene acetal (9) that undergoes hydrolysis of the allylic 3′-phosphate to yield a strand break. Formation of 9 to explain 18O incorporation from H218O was unnecessary because of the reactivity of 3 described above (Scheme 3 ).</p><p>Experimental doubt was cast on the kinetic viability of 9 using a model system in which 11 was independently synthesized by oxidizing 10 (Scheme 7 ).74 The nucleoside model (11) was stable for days under physiologically relevant pH and temperature. This model study provided an alternative explanation for the formation of direct strand breaks by Cu•OP2 and the absence of L. β-Elimination from 12 was first-order in Cu•OP2 and the rate constant was such that an effective molarity of 10 M in an intramolecular reaction would yield a half-life of <1 min for L in DNA.74 Hence, the model study suggested that while a variety of oxidants damage DNA by abstracting the C1′-hydrogen, the differing products (direct strand breaks versus the alkali-labile lesion 2-deoxyribonolactone, L) were a consequence of the instability of L in the presence of Cu•OP2.</p><p>Sugiyama put forth a completely different mechanism to explain the lack of 2-deoxyribonolactone formation from Cu•OP2.75 Using hexanucleotide duplexes and electron spray ionization-mass spectrometry (ESI-MS) to analyze products, Sugiyama confirmed that Cu•OP2 forms L in relatively high yield but the lactone does not yield strand breaks via β-elimination in these short DNA substrates. Furthermore, no evidence for a 1′,2′-dehydronucleotide (eg 9) was observed. The authors detected strand scission products consistent with C4′- and C5′-hydrogen atom abstraction, which led them to conclude that these were the positions that Cu•OP2 reacted with to yield direct strand breaks. The significance of the stability of L in these substrates was questioned because duplexes containing the lactone lesion melted below the temperature at which the aforementioned experiments were carried out.12 The Cu•OP2 would not be expected to bind to the single-stranded oligonucleotides and catalyze β-elimination.76</p><p> </p><p>The latest mechanistic proposal was derived from experiments that combine the ability to independently synthesize duplex DNA containing 2-deoxyribonolactone (L) at a specific position from 2′-deoxyuridin-1′-yl (3) with sequence-selective delivery of Cu•OP2.12,77 The metal complex was delivered to the position where L was produced by tethering the metal complex to a minor groove DNA-binding molecule (13).78,79 The half-life for L in a duplex (14a) bound by minor groove-binding distamycin tethered to Cu•OP2 was 20.6 min (kElim = 5.6 ± 0.7 × 10−4 s−1), and was ~100 times shorter than when the metal complex was not delivered to the lesion. Cu•OP2-induced strand scission at 2-deoxyribonolactone was supported by experiments containing dA at the position where L was independently generated in an otherwise identical duplex. Oxidation at this position (A13 in 14b) by the distamycin-Cu•OP2 conjugate (13) was recorded by measuring the combined yields of alkali-labile lesions (presumably mostly L) and direct strand breaks as a function of time. The overall rate constant for oxidation (kOx = 1.9 ± 0.6 × 10−5 s−1) was slower than the rate constant measured for cleavage at L incorporated at the comparable position. In addition, fitting the growth and decay of alkali-labile lesions as a function of time to a sequential mechanism yielded a rate constant for decay of these lesions (kDecay = 4.7 ± 0.9 × 10−4 s−1, Eq. [1]) that is within experimental error of the independently measured rate constant for cleavage at L (kElim). Unless, more than one alkali-labile lesion reacts with similar rate constants, the similarity in kDecay and kElim indicates that 2-deoxyribonolactone (L) is the pre-dominant alkali-labile lesion that is cleaved in the presence of Cu•OP2. Similarly, the rate constant for the formation of the alkali-labile lesions (kGrow = 1.8 ± 0.4 × 10−5 s−1, Eq. [1]) was within error of the overall rate constant for oxidation (kOx). This latter point supports the suggestion that alkali-labile lesion formation accounts for the majority of Cu•OP2-mediated DNA oxidation. In summary, the ability to independently generate 2′-deoxyuridin-1′-yl radical (3) at a defined position within DNA enabled carrying out kinetic experiments on 2-deoxyribonolactone (L) reactivity that support the original mechanism put forth by Sigman explaining why Cu•OP2 produces direct strand breaks following C1′-hydrogen atom abstraction, whereas other damaging agents yield L as a final product following the same initial oxidation event.</p><!><p>Independent formation of 2-deoxyribonolactone (L) in DNA using ketone 4 was also useful for examining the interactions of the DNA lesion with repair enzymes. Base excision repair (BER) is a general pathway used by cells to replace damaged nucleotides with the correct native one (Scheme 8 ).80–82 (Tomas Lindahl received a share of the 2015 Nobel Chemistry Prize for his work on BER.) BER typically proceeds through abasic site (AP) intermediates that are produced by glycosylases, which recognize damaged nucleobases (NDam). Some glycosylases are bifunctional and are equipped with the ability (lyase activity) to induce β-elimination, or even β,δ-elimination of the AP sites. The primary BER pathway in mammalian cells involves a similar lyase reaction by DNA polymerase β (Pol β) following hydrolysis of the 5′-phosphate of the AP site by apurinic endonuclease 1 (Ape1) to produce dRP. Pol β then translocates the DNA to its polymerase active site and fills in the now missing nucleotide before passing the substrate off to DNA ligase, which stitches the DNA together.</p><p>Iminium ions (Schiff bases) are common intermediates in the lyase reactions carried out by various repair enzymes (Scheme 9 ). Consideration of elementary organic chemistry raised the question concerning how BER enzymes that possess lyase activity would cope with 2-deoxyribonolactone (L). Biochemically, this is potentially important because failure to correctly repair damaged DNA can lead to mutagenesis and/or cell death. Indeed, incubation of duplex DNA containing 2-deoxyribonolactone (L) produced from photolysis of 4 revealed that none of a series of eight BER glycosylases incise the lesion.83,84 In addition, the presence of L when part of a tandem lesion 15 (defined as two or more contiguously damaged nucleotides) inhibited repair of the accompanying damaged nucleotide as well.85 Moreover, 2-deoxyribonolactone (L) irreversibly inhibits one of the enzymes, endonuclease III (Nth) by forming cross-links with the nucleophilic lysine residue that is responsible for Schiff base formation.83,84 This was the first demonstration that a DNA lesion inactivates a repair enzyme, a process that has since been characterized for a small number of other damaged nucleotides and enzymes.86–90</p><p>2-Deoxyribonolactone (L) inactivation of Nth may be of minor importance because excision of an incised AP site by Pol β is the primary pathway for removing this lesion after incision by the phosphodiesterase, apurinic endonuclease I (Ape 1). Similar to the interaction with Nth, incised 2-deoxyribonolactone (16) cross-links to Pol β in a manner that is dependent upon the presence of the lysine residue believed to be responsible for Schiff base formation.91 Recently, evidence was provided for cross-linking between 16 and Pol β in mammalian cells.92 This latest report illustrates the possibility that the BER enzyme inactivation by DNA lesions provides a chemical basis for the cytotoxic effects of the therapeutic agents and other modalities that produce them.93</p><p> </p><!><p>As mentioned above, C2′-hydrogen atom abstraction in DNA is a rare occurrence due to the strong carbon–hydrogen bond (Table 1) and relative inaccessibility of the hydrogen atoms to diffusible species, such as HO•.34 C2′-Radical formation is indirectly detected when DNA containing 5-bromo-(BrdU) or 5-iodo-2′-deoxyuridine (IdU) is irradiated. UV irradiation generates the uracil-5-yl radical (17) via photoinduced electron transfer from a 2′-deoxyguanosine within the duplex (Scheme 10 ).53 The efficiency of σ-radical (17) generation is highly sequence dependent due to competition between halide ion loss and back electron transfer.38,54 The uracil-5-yl radical (17) abstracts hydrogen atoms from the C1′- and C2′-positions of the 5′-adjacent nucleotides (Scheme 10 ). Hydrogen atom abstraction from these positions by 17 was determined using deuterium isotope effects, tritium transfer and product studies.40–43,53 2-Deoxyribonolactone (L) formation is indicative of C1′-oxidation. Whether the initially formed C1′-radical is oxidized via the distal oxidized purine or by O2 as described above (Section 4.2) is uncertain, as H2O is the ultimate source of the carbonyl oxygen in either scenario.38,57,58 Formation of the C2′-oxidized abasic site (18) is attributed to O2 trapping of the radical produced upon C2′-hydrogen atom abstraction.41 The details for obtaining 18 from the peroxyl radical are uncertain, including the identity of the reducing agent. A Criegée rearrangement is drawn here but the original report suggested a different mechanism.41 Independent generation of the C2′-radical (20) from 2′-iodo-2′-deoxyuridine (19, Scheme 11 ) in DNA or a nucleoside provided 18 but 2-deoxyribonolactone (L) was also observed. Photolysis of C1′-deuterated 19 indicated that the lactone was formed via oxidation of the C2′-radical (21) and subsequent 1,2-hydride rearrangement (Scheme 11 ).94 Although not proposed at the time, one could invoke a photoinduced single electron transfer mechanism for the generation of 20 analogous to that substantiated for BrdU. In that situation the oxidized purine nucleotide formed upon irradiation could carry out the oxidation of the C2′-radical in DNA. Moreover, this chemistry suggests that C2′-hydrogen atom abstraction by a uracil-5-yl radical (17) could yield the formal C1′-oxidation product (L) in DNA.</p><p>The reactivity of the uracil-5-yl radical (17) is strongly dependent upon the DNA structure and is also affected by whether the complementary strand is RNA.95,96 The reactivity of the C2′-radical is also affected by the nucleic acid structure and the distance between it and the electron-deficient nucleobase.97,98 The C2′-oxidized abasic site (18) is observed in B-DNA, whereas the respective ribonucleotides are produced in Z-DNA. The product distribution obtained upon irradiating BrdU and/or IdU in DNA has been used by Sugiyama to probe nucleic acid structure.99 As discussed below, the chemistry is different still in RNA100</p><!><p>The significant radical stabilizing energy of a hydroxyl group weakens the C2′-carbon–hydrogen bond in RNA relative to that in DNA and may be enhanced due to hydrogen bonding to the 3′-phosphate (Table 1).35,101 Stabilized radicals containing β-leaving groups undergo heterolytic fragmentation to produce radical cations, a process that is facilitated in polar solvents such as water.102–106</p><p>Norrish type I photocleavage has been used to generate the adenosin-2′-yl radical (22) and uridin-2′-yl radical (23) from 24 and 25, respectively (Scheme 12 ).107–109 The reactivity of each radical was examined with thiol. Trapping of 22 in aqueous buffer by GSH yielded adenine and a 3.5:1 ratio of arabinoadenosine and adenosine, although the yield was not reported.107 A competition study in which the ratio of reduction products (arabinoadenosine + adenosine) versus adenine were measured as a function of [GSH] suggested that the rate constant for loss of the nucleobase was ~2 × 105 M−1 s−1. This ratio of deglycosylation versus thiol trapping products is significantly different than what was observed from 23. When 25 was photolyzed in phosphate buffered (pH 7.2) saline (100 mM), uracil (56–59% based upon unrecovered 25) was the only product detected in the presence of BME (0.25 M).108 The inability of thiol to trap 23 in aqueous buffer suggests that loss of uracil is significantly faster than that of adenine. Thiol trapping (BME, 0.1 M) of 23 was observed in H2O and greater yields of arabinouridine were obtained as increasing amounts of acetonitrile were added. Another difference between 23 and 22 was that arabinouridine was the only reduction product formed, which is typical of reductions of nucleoside radicals.110–112</p><p>Uridin-2′-yl radical (23) was also independently generated from 25 in single- and double-stranded RNA.109 The radical rapidly yields direct strand breaks in an O2-independent manner (Scheme 13 ). Thiol (BME, 1 M) also has no effect on strand scission. Assuming rate constants for the reactions of O2 (kO2 = 2 × 109 M−1s−1) or BME (kBME = 1–10 × 106 M−1s−1)113 with 23 and the respective concentrations of the traps ([O2] = 0.2 mM, [BME] = 1 M), the rate constant for strand scission was estimated to be >106 M−1s−1. Based upon precedents in DNA (more below when the chemistry of C4′-hydrogen atom abstraction is discussed), and the general reactivity of alkyl radicals containing β-leaving groups noted above, strand scission was postulated to occur via heterolytic cleavage of the C3′-carbon–oxygen bond in single- and double-stranded substrate to form an olefin cation radical (26, Scheme 13 ).106 Product analyses using gel electrophoresis in conjunction with enzymatic and chemical reactivity are consistent with this. The 3′-fragment is composed solely of a 5′-phosphate terminus (27). The major products detected at the 3′-terminus of the 5′-fragment are the phosphate (28) and 2′-keto-3′-deoxyuridine (29). The ratios of these products depend upon O2 and thiol concentration. The ketone product (29) dominates (>3:1) under anaerobic conditions, even at low thiol concentration (5 mM). It is not known whether the initially formed radical cation (26) undergoes deprotonation (30), followed by hydrogen atom transfer to the α-keto radical or is reduced directly to the enol (31), which tautomerizes to 29 (Scheme 14 ). Radical cation 26 may also be reduced by guanosine within the RNA. This should be thermodynamically favourable. However, no evidence has been presented in support of this. 3′-Phosphate (28) formation is favoured over 29 under aerobic conditions but its mechanism of formation is less obvious. One speculative mechanism that requires further investigation involves O2 trapping of the α-keto radical (30, Scheme 14 ). Other questions, including uracil loss in the oligonucleotides, which was not reported on also need to be addressed.</p><!><p>Recently, the photochemistry of 5-bromouridine (BrU) was examined in a series of sequences in which BrU was flanked on its 5′-side by either adenosine or guanosine (Scheme 15 ).100 In contrast to studies involving BrdU, no evidence for C1′-hydrogen atom abstraction by the uracil-5-yl radical (32) was obtained. Conformational differences between duplex RNA and DNA could contribute to this difference, but certainly the more favorable C2′-carbon–hydrogen BDE in RNA than DNA (Table 1) plays a role. The final product resulting from C2′-oxidation (characterized by MS) is also different in RNA than DNA. The C2′-oxidized abasic site (18, Scheme 11 ) is not observed, nor do the authors report any strand scission resulting from heterolytic cleavage of the 3′-phosphate (Schemes 13 and 14). The sole C2′-oxidation product observed in RNA when the adjacent nucleotide is adenosine or guanosine is the 2′-keto purine (33). The authors attribute 33 to oxidation of the C2′-radical by the one-electron oxidized guanosine that provided the initial electron used to reduce BrU. Based upon the reactivity of uridin-2′-yl radical (23) described above, electron transfer to the oxidized purine must occur faster than 106 s−1 in order to prevent strand scission.</p><!><p> </p><!><p>The C3′-hydrogen exposure to diffusible species is considerably less than hydrogen atoms at the C4′- or C5′-positions (Fig. 1B) and the calculated carbon–hydrogen BDE is surprisingly high given its substitution by a phosphate group (Table 1). In addition, the C3′-hydrogen lies in the major groove of DNA, while most small molecules bind in the minor groove. Consequently, molecules that abstract the C3′-hydrogen atom from DNA are largely confined to coordination compounds. The Barton group has described a number of Rh complexes (eg Rh(phi)2(bpy)3+, 34) that bind in the major groove and oxidatively damage DNA upon photoexcitation.114,115 C3′-Hydrogen atom abstraction (35) is supported by the binding preference and product analysis. Strand scission was observed under aerobic and anaerobic conditions (Scheme 16 ). 3′-Fragments containing 5′-phosphate termini were formed exclusively regardless of whether O2 was present or not, as were free nucleobases. The 5′-fragments were composed of a mixture of 3′-phosphate and 3′-phosphoglycoaldehyde (37) termini. The remaining three carbons of the 2′-deoxyribose ring were released in the form of base propenoic acids (38) and were only detected under anaerobic conditions. Peroxyl radical formation (36) from the C3′-radical and subsequent Criegée rearrangement of hydroperoxide reduction product are consistent with the O2 dependence of 37 and base propenoic acid (38) generation. Under anaerobic conditions, the C3′-radical (35) ultimately must be oxidized and trapped by H2O (39). However, as is often the case with studies in nucleic acids, it is unclear what oxidizes the C3′-radical. Similarly, it is unknown what reduces the peroxyl radical to the hydroperoxide (36).</p><!><p>The observations and mechanistic proposals put forth by Barton have largely been corroborated by experiments in which thymdin-3′-yl (35) was independently generated via Norrish type I photocleavage from 40a,b.116–119 A preliminary report established that 35 produced in single-stranded oligonucleotides was efficiently trapped by GSH under anaerobic conditions. The reduction products were accompanied by small amounts of strand scission but the mechanistic source of the fragmentation was not ascertained.116 Strand scission resulting in the formation of fragments containing phosphate groups at their termini (Scheme 17 ) was the major pathway when 35 was generated under anaerobic conditions in the absence of thiol. This was consistent with the chemistry discovered by Barton (Scheme 16 ), and the metastable ketone (39) was proposed as an intermediate, although the authors note that the source of the oxidant of 35 is unknown. In addition, an unspecified yield of cleavage product resulting from oxidation of the 3′-adjacent nucleotide was observed and it was suggested that radical transfer occurs via hydrogen atom abstraction by 35.</p><p>Greater detail was provided using mass spectrometry to characterize products when 35 was generated in single-stranded oligonucleotides via Norrish type I photocleavage under aerobic conditions and dilute GSH (6 mM).117 The major products were consistent with those detected by Barton. For instance, the majority of 3′-fragments contained 5′-phosphate termini (82%) and the majority of the 5′-fragments (78%) were attributable to the 3′-keto thymidine (35, Scheme 16 ). Ketone 39 and elimination product 41 were observed in a combined 21% yield. The 3′-phosphate product, which also presumably arises from 35 was the major product (57%). In contrast to Barton's experiments (Scheme 16 ), the 3′-phosphoglycoaldehyde (37) was only formed in 9% yield under aerobic conditions, and the authors did not report the base propenoic acid (38) that would be expected to accompany formation of this product. This difference may indicate that the coordination complexes influence the Criegée rearrangement, possibly by acting as an acid catalyst. Alternatively, the GSH present in the experiments where 35 is independently generated from 40 may reduce the hydroperoxide derived from 36 (Scheme 16 ) before it can rearranges, which would result in greater amounts of 39.</p><p>The C3′-peroxyl radical (36) derived from 35 was also suggested to abstract hydrogen atoms in an intranucleotidyl and internucleotidyl manner (Scheme 18 ). (The detailed reactivity of the subsequently formed radicals is discussed in Sections 7.2–7.5 and 8.2 and 8.3.) Intranucleotidyl abstraction of the C4′-hydrogen atom is proposed to explain the formation of small amounts (4%) of phosphoglycolate (42), leaving <10% of the 5′-fragment unaccounted for. A greater yield of 3′-fragments (18%) is attributed to 5′-hydrogen atom abstraction from the 3′-adjacent nucleotide. The diagnostic C5′-aldehyde (6%, 43) and 5′-phosphorylated 3′-fragment lacking the 3′-adjacent nucleotide (12%) are observed by high pressure liquid chromatography (HPLC) and matrix assisted laser desorption ionization-fime of flight (MALDI-TOF) MS. The latter is presumably formed via elimination from 43.120 The proposed sites of hydrogen atom abstraction are consistent with O2 trapping from the α-face of 35. These product observations were corroborated in a later study in which the GSH concentration was varied from 0 to 30 mM where there was a complex product dependence on thiol.118 More limited studies have been carried out in which 35 was generated in double-stranded substrates, and it is unknown how this more conformationally restricted environment affects the reactivity of the C3′-radical (35) and subsequently formed intermediates.119</p><!><p>Due to its accessibility on the outer edge of the minor groove (Fig. 1A) and its relatively favourable C–H bond dissociation energy (Table 1), the C4′-hydrogen is abstracted by a number of DNA damaging agents, including HO• and the antitumor agent bleomycin.34,121 Several members of the enediyne family of antitumor agents also form the C4′-radical.122–124</p><p> </p><!><p>There are three general reaction pathways associated with the C4′-radical in DNA (44, Scheme 19 ). Much of the mechanistic details for two of these pathways were worked out using bleomycin as a means for generating the radical.125 Bleomycin is a peptide that coordinates iron. The resulting complex carries out redox chemistry in the presence of O2, and selectively abstracts the C4′-hydrogen atom from nucleic acids. The coordination complex, and in particular the metal is believed to be involved in subsequent steps.126 Inferential support for the latter is based upon metal-independent methods for generating the C4′-radical, such as γ-radiolysis.103 However, C4′-oxidation in DNA by other reagents has not been examined as thoroughly as that by bleomycin. Studies carried out using bleomycin made extensive use of isotopic labelling. The C4′-radical has also been independently generated in DNA and RNA from stable photochemical precursors. It has been produced most frequently from a t-butyl ketone (45), although a phenyl selenide (46) was also employed in some investigations.</p><p>C4′-hydrogen atom abstraction yields the alkaline-labile, oxidized abasic site (C4-AP). Isotopic labelling ruled out bleomycin formation of C4-AP via an oxygen rebound-type mechanism, and instead oxidation of 44 and trapping of the carbocation (47) by H2O was invoked (Scheme 20 ).127,128 However, the species responsible for oxidizing 44 is uncertain. Other damaging agents, such as the enediynes produce C4-AP as well via hydrogen atom abstraction. However, in these instances the mechanism proceeding from 44 to C4-AP is also unknown. (As discussed in Section 7.3, it is unlikely that superoxide elimination from 48 yields 47.) Strand breaks are produced at C4-AP upon mild treatment with hydrazine or 0.1 M NaOH at 37 C.129,130 This molecule has rich chemistry in its own right. It is an irreversible inhibitor of the vital DNA repair enzyme DNA polymerase β.90 In addition, strand scission at C4-AP is significantly accelerated in nucleosomes.131,132 The histone proteins, which are modified during the process, catalyze strand scission.</p><p>The formation of direct strand breaks containing 3′-phosphoglycolate termini (42) in the 5′-fragment is a signature product indicative of C4-oxidation (Scheme 21 ). (Please note that as described in Section 6.2 C4′-oxidation can be a secondary process when the C3′-hydrogen atom is abstracted.) The 3′-DNA fragment contains a phosphate group at its 5′-terminus and the remaining three carbons of the 2′-deoxyribose that underwent oxidation are accounted for by the formation of base propenals. Kinetic isotope effects of C4′-isotopically labelled substrates (3H, 2H) definitively documented rate limiting hydrogen atom abstraction from this position by bleomycin.121,133,134 A key step in the overall transformation is a putative Criegée rearrangement of an intermediate hydroperoxide, which is believed to be catalyzed by the metallopeptide. As mentioned previously, this is another instance in which the reducing agent responsible for hydroperoxide formation from the peroxyl radical is unidentified. One possibility is that a Fe+3 species could be the reductant, which would reactivate bleomycin. Bleomycin reactivation while bound to DNA has been proposed as a means for producing some double-strand breaks.135 Nonetheless, the steps leading up to and including the formation of 48 are widely accepted.</p><p>A greater number of proposals were proffered explaining formation of the final observed products following the Criegée rearrangement. Extensive experimentation that included determination of the sequence in which strand scission, 3′-phosphoglycolate (42), 5′-phosphate and base propenal (50) release occur gave rise to the currently accepted mechanism (Scheme 21 ).126 This mechanism is consistent with incorporation of a single 18O atom from 18O2 into the glycolate. Importantly, the 18O is not from the activated bleomycin but comes from 18O2 trapping of the C4′-radical.136 In contrast, the 18O incorporated into the base propenal carbonyl is derived from H218O.126 In addition, elegant experiments in which the C2′-position was stereoselectively labelled with 3H revealed that the 2′-pro-R proton is removed en route to base propenal and that deprotonation occurs prior to DNA strand scission.137,138</p><p>Direct strand breaks attributable to the C4′-radical are produced that yield 3′- and 5′-phosphate termini instead of 3′-phosphoglycolate termini and base propenal (Scheme 22 ). This phenomenon was originally observed in DNA subjected to γ-radiolysis under deoxygenated conditions.103 Interestingly, the mechanism was originally proposed by von Sonntag based upon the products derived from the 2′-deoxyribose component of the DNA. The phosphate termini, while undoubtedly produced by one or more of the myriad pathways populated upon γ-radiolysis of DNA were proposed to rationalize the formation of the sugar degradation products. As seen below, this pathway proved prescient and the intermediate formation of olefin radical cations (eg 51) was immensely useful for studying electron transfer in DNA (Section 7.5).</p><!><p>Strong evidence for the von Sonntag mechanism was obtained by independently generating C4′-radicals (Scheme 19 ) via Norrish type I photocleavage of 45 and to a lesser extent phenyl selenide 46.139,140 Initial studies under anaerobic conditions on monomeric models and single-stranded oligonucleotides containing 45 or 46 supported initial 3′-phosphate cleavage, followed by regioselective addition of H2O to regenerate a C4′-radical that then eliminated the 5′-phosphate (Scheme 22 ). Subsequent studies confirmed preferential H2O addition to the C3′-position of 51 to yield a diastereomeric mixture of 52.141 The products were characterized by MALDI-TOF MS, which was useful for subsequent 18O studies under aerobic conditions.</p><p>Under aerobic conditions, the authors observed an ion whose m/z corresponded to the hydroperoxide (eg 49) derived from O2 of the initially formed C4′-radical (44). 18O2 labelling, which showed that two 18O atoms were incorporated, corroborated this assignment. Interestingly, the respective hydroperoxide was detected from the oligonucleotide containing 45 but not 46. The latter precursor yielded a strand scission product that based upon MS analysis in conjunction with 18O2 and H218O labelling was proposed to be the hydroperoxide containing a 3′-hydroxyl group (53, Scheme 23 ). The difference in products from the two precursors was suggested to be due to formation of a radical pair from 46, which prevented O2 trapping of the initially formed C4′-radical.140 Both hydroperoxides produce the 3′-phosphoglycolate (42) signature product of C4′-oxidation (Schemes 21 and 23). The cleaved oligonucleotide containing a 3′-hydroxyl group was proposed to undergo a Grob-type fragmentation (Scheme 23 ), whereas the intact DNA containing hydroperoxide was believed to undergo a Criegée rearrangement as discussed above for bleomycin-induced strand scission (Scheme 21 ). Unfortunately, kinetic studies on 3′-phosphoglycolate formation were not reported, as this would have been useful for evaluating the role of the metallated bleomycin in the respective rearrangement.</p><p>C4′-Radical generation from 45 and 46 under anaerobic conditions in the presence of GSH yielded the expected diastereomeric mixture of reduction products consisting of the native nucleotide and the C4′-epimer (54, Scheme 24 ).106 Hydrogen atom delivery from the α-face to restore the naturally occurring stereochemistry of the DNA was slightly favoured in single-stranded substrates (≤2:1), but much more so (<8:1) when the C4′-radical was generated in duplex DNA. Competition studies between strand scission and GSH (kGSH) trapping provided estimates of the rate constant for strand scission from DNA containing C4′-radical via phosphate elimination and concomitant radical cation (51, Scheme 24 ) formation. The ratio of hydrogen atom trapping products (native nucleoside + 54) relative to strand scission was measured as a function of GSH concentration. In single-stranded DNA the rate constant (kCleave) was estimated to be from ~0.8 to 1.9 × 103 s−1. Moreover, the C4′-radical was observed to yield strand breaks ~10-fold more slowly in double-stranded DNA (kCleave ~0.2–2.1 × 103 s−1). Although fundamentally the same type of process described above for strand scission from the C2′-radical in RNA, strand scission from the C4′-radical in DNA is at least 1000-times slower.109</p><p>The rate constants for strand scission via phosphate elimination from the C4′-radical in DNA are also considerably slower than the expected rate of radical trapping by O2 (kO2 ~2 × 109 M−1s−1, [O2] ~0.2 mM). However, strand scission products ascribable to phosphate elimination from the C4′-radical were reported. This apparent discrepancy was resolved by competitive kinetic studies with GSH under aerobic conditions from which O2 trapping was determined to be reversible (Scheme 25 ).142 In contrast to the discussion above regarding elimination of superoxide from a C1′-peroxyl radical (See Section 4.2), the C4′-peroxyl radical fragments much more slowly (kO2 ~ 1 s−1) and does so homolytically. However, the estimated rate constant is consistent with that for related radicals containing a single oxygen substituent.59</p><!><p>Molecules that produce double-strand breaks are rare and highly valued due to the biological significance of this form of DNA damage.135,143–147 Hence, the possibility that a single radical can result in a double-strand break via interstrand hydrogen atom transfer by one or more intermediates is very interesting. Radiation scientists debated this possibility in the 1990s due the observation that double-strand break yield varied linearly with dose (which HO• yield is proportional to) at low doses of radiation.148,149 Ultimately, the formation of locally high HO• concentrations ('spurs'), which led to clusters of DNA lesions and in turn double-strand breaks, became widely accepted.150–152 Recently, Taverna Porro discovered a chemical pathway by which a single C4′-radical yields double-strand breaks in an O2-dependent manner.13,153</p><p>Reversible peroxyl radical formation from the C4′-radical (Scheme 25 ) was crucial for providing an explanation for the O2-dependent production of double-strand breaks from this species.13,153 Generation of 44 from 45 in duplex DNA yielded double-strand breaks under aerobic but not anaerobic conditions. ESI-MS analysis and deuterium kinetic isotope effects indicated that complementary strand scission resulted from C4′-hydrogen atom abstraction at nucleotides opposite the three most proximal 5′-adjacent nucleotides (Scheme 26 ). This produces double-strand breaks biased in the 3′-direction, which is consistent with reaction in the minor groove of right-handed helical DNA.154 Double-strand scission is made possible via cleavage of the initial strand containing a C4′-radical via the von Sonntag mechanism. As described by Giese, the radical cation (51) yields a second peroxyl radical via sequential addition of H2O and O2.140,141 Four different peroxyl radicals can be produced from 51 with those resulting from H2O addition at C3′-favoured.141 Mechanistic studies using oligonucleotide substrates that contain the C4′-radical precursor (45, Scheme 19 ) at their 3′-termini provided an independent method for producing the C3′-hydration product of 55.153 This radical did not yield opposite strand cleavage, suggesting that the (minor) regioisomeric C4′-hydration product that yields the C3′-peroxyl radical(s) (56) is responsible for interstrand hydrogen atom abstraction.</p><!><p>Although this subject has been reviewed, one would be remiss not to briefly mention how independent C4′-radical generation in DNA contributed to this important problem.155,156 The efficiency, distance and sequence dependence of DNA hole migration was vigorously debated in the 1990s.20,157–160 Elucidation of the rules that govern this process are useful to this day as DNA electron transfer is a continued source of chemical innovation.26,161–165 Briefly, Barton reported the rapid, long-range (40 Å) transfer of electrons between two metal complexes bound to DNA.166–168 In addition, initial reports indicated that in contrast to electron transfer in proteins there was very small dependence of hole migration on distance. These observations raised the possibility that a DNA duplex could behave like a molecular wire in its ability to conduct electrons. Although one can transfer electrons over large distances through DNA, it is now accepted that the efficiency is strongly dependent on sequence.169</p><p>Giese made significant contributions to this contested subject by taking advantage of 44 generation from 45 (Scheme 19 ). Irreversible formation of olefin cation radical 51 from 44 (Scheme 22 ) was critical for the success of this approach. The reduction potential of the radical cation (51) is sufficient to oxidize 2′-deoxyguanosine. Hence, C4′-radical generation provides a site-specific means for irreversibly introducing a hole in DNA. Using chemically synthesized oligonucleotides containing 45, Giese was able to demonstrate the importance of sequence on hole transfer efficiency. The interested reader is referred to the references noted above to learn more about this topic.</p><!><p>Bleomycin also cleaves RNA, and it has been postulated that this contributes to the biological activity of the antitumour agent.170–173 However, the details of the chemistry, including C4′-hydrogen atom abstraction are not as well understood as in DNA. Giese independently generated a C4′-radical in single-stranded RNA (57, Scheme 27 ) under aerobic and anaerobic conditions via Norrish type I photocleavage of 58.174 The rate constant for strand scission (kCleave, Scheme 27 ) from 58 was estimated by measuring the ratio of thiol trapping products (59, 60) versus phosphate cleavage product as a function of GSH concentration. By assuming that the thiol traps 57 with a bimolecular rate constant kTrap = 1 × 107 M−1s−1, kCleave was estimated to be 5 × 102 s−1. This is slightly more than threefold slower than the comparable cleavage step in DNA, assuming that the rate constants for GSH trapping of the DNA and RNA C4′-radicals are equal.106 The decrease in reactivity of 57 compared to the comparable DNA radical is modest compared to a related model study containing phosphate triesters (instead of phosphate diesters that are present in DNA and RNA) that was carried out in aqueous alcohol solvent (as opposed to H2O).175 The modest difference in the rate constants for phosphate elimination from 57 compared to its DNA counterpart (44, Scheme 24 ) was attributed to solvation of the radical cation by the aqueous solvent, such that the proximal 2′-hydroxyl group in RNA had a small effect on the intermediate's stability.174</p><p>Aerobic photolyses yielded the expected fragmentation products following peroxyl (62) radical formation, a 5′-fragment containing a 3′-phosphoglycolate (42) and a 3′-fragment containing a 5′-phosphate terminus (Scheme 28 ).174 The ratio of these products was ~3:1 in favor of phosphoglycolate 42 over phosphate. This is distinct from what happens in DNA where 42 forms concomitantly with base propenal (50) and phosphate product (Scheme 21 ). Although the authors present a mechanism rationalizing the formation of these products, evidence for the formation of the crucial product (61) is not presented. Presumably, elimination from the Criegée product (63) is slower than in DNA due to the presence of the hydroxyl group. This enables H2O to trap the carbocation and altering the final decomposition pathway to phosphoglycolate (42).</p><!><p>The C5′-hydrogen atoms are even more accessible to diffusible species than the C4′-position (Fig. 1A) and this site has been suggested to be the preferred position for hydrogen atom abstraction by HO•.34 A variety of minor groove-binding DNA oxidizing agents abstract the C5′-hydrogen atom, including manganese porphyrins and the enediyne antitumor antibiotics, resulting in strand scission.120,176–178 C5′-Deuterium-labelled DNA substrates were commonly used for detecting hydrogen atom abstraction from this position by the antitumour antibiotics.123 Frequently, the deuterium was used to reduce strand scission and/or shift reaction to another nucleotide position (eg C4′) resulting in different products, both of which were detected by gel electrophoresis.177–179 However, in some instances the deuterium was even traced to its incorporation into the final product obtained from the antitumor agent.180</p><!><p>The major product formed following C5′-oxidation is the C5′-aldehyde (43), which releases furfural from DNA upon heating (Scheme 29 ). Under most experimental conditions 43 is believed to arise via the peroxyl radical (65) (see below). 18O-Labelling was employed to obtain additional information on the mechanism for formation of 43 resulting from manganese porphyrin oxidation. However, the results were inconclusive and it was proposed that this was due to rapid exchange of the aldehyde in H2O.181 In contrast, 18O-labelling experiments indicated that O2 was the source of oxygen in 43 when neocarzinostatin reacts with DNA.182 These experiments also ruled out oxidation of the initially formed C5′-radical (64) to a carbocation. The aldehyde was reduced in the neocarzinostatin experiments to prevent solvent exchange of the carbonyl oxygen. However, reduction was unsuccessful in preventing equilibration in experiments using the manganese porphyrin.</p><p>Various proposals have been put forth to explain how peroxyl radical 65 is transformed into the aldehyde (43, Scheme 29 ), including the dimerization of two peroxyl radicals (not shown).183 The tetroxide pathway is improbable in DNA because of the unlikeliness that two peroxyl radicals will collide. In addition, the 18O-labelling experiments carried out using neocarzinostatin as a source of 65 suggest that superoxide elimination is also unlikely.182 Ferric ion was reported to rapidly oxidize (k ~4 × 109 M−1s−1) the nucleoside 2′-deoxyadenosin-5′-yl radical (64).184 Although this could be a viable pathway under anoxic conditions, 18O-labelling indicates that this process also does not compete with formation of 65 without adding exogenous metal ion. Hence, the most general pathway to 43 from 64 under aerobic conditions would appear to require reduction of the peroxyl radical (65), most likely by thiol.</p><p>The putative hydroperoxide (66) from 65 generated by reaction with neocarzinostatin has also been proposed to undergo a Criegée rearrangement to yield a strand break concomitantly with free base release and a 5′-fragment containing a labile formate bonded to the 3′-terminal phosphate (67, Scheme 30 ).185 The 3′-fragment contains chemically unstable 5′-(2-phosphoryl-1,4-dioxobutane) (DOB) at its 5′-terminus.186,187 Like C4-AP (Section 7.2), DOB irreversibly inhibits DNA polymerase β and yields histone proteins containing modified lysines in nucleosomes.87–89,188 The yield of free base obtained from neocarzinostatin reactions varies from 16% to 20% depending upon the thiol that activates the antibiotic.186 The DOB yield determined by measuring the stabilized reduction product (68) is about half this and may be due to reaction of the lesion with nucleophilic thiol.187,189</p><p>Under anaerobic conditions, the distinctive 5′,8-cyclonucleos(t)ide DNA lesions (69, 70) are produced from 64 of purine nucleos(t)ides(Scheme 31 ).107,184,190–192 The 5′,8-cyclonucleotides have significant effects on replication and repair.193–196 They are strong replication blocks and are repaired via the nucleotide excision repair pathway instead of BER. The repair pathway is unusual for modified nucleotides formed via oxidative stress. The mechanism and kinetics of their formation are described below (Scheme 32 ).</p><!><p>The C5′-radical has not yet been independently generated in oligonucleotides but the respective 2′-deoxy- and ribonucleoside radicals have been produced via Norrish type I photocleavage and a less conventional manner using pulse radiolysis.107,184,190,192,197 Pulse radiolysis of various 8-bromopurine nucleosides provided access to the C5′-radicals via initial reaction with the solvated electron, which yields the C8-σ-radicals on the sub-microsecond timescale (Scheme 32 ). The C8-purine σ-radicals are believed to rapidly abstract the C5′-hydrogen atom providing 64. The investigators typically monitor the growth of the subsequent C8-cycloadduct (71) on the microsecond timescale by absorption spectroscopy. C5′-radical cyclization rate constants range from ~1 × 104 s−1 to ~1 × 106 s−1.107,190,192 However, it should be noted that the upper range of these rate constants was estimated using competitive kinetics by generating a protected form of 2′-deoxyguanosin-5′-yl radical via Norrish type I generation in tetrahydrofuran (THF) solution.190 These rate constants suggest that cyclization should be competitive with O2 trapping of the C5′-radicals, and this was demonstrated for 2′-deoxyadenosin-5′-yl radical (64) that was generated via pulse radiolysis of the corresponding 8-bromopurine nucleoside (Scheme 32 ).184 5′-Aldehyde yields correlated with increased O2 concentration at the expense of cyclonucleoside product formation (Scheme 32 ). C5′-Radical cyclization of pyrimidines was not detected. In addition, no mention of free base release or other evidence for DOB formation was reported. There is still quite a bit to learn about the reactivity of the C5′-nucleos(t)ide radicals, particularly within the biopolymer.</p><!><p>Nucleobase radicals are largely the domain of HO• and to a lesser extent hydrogen atoms, that are produced by ionizing radiation.17,32 Hydroxyl radical is also produced by metal complexes such as Fe•EDTA.33 Unlike the various molecules mentioned above that bind in the minor groove and abstract hydrogen atom(s) from the 2′-deoxyribose backbone, HO• addition to the nucleobases is the major pathway for this species. Estimates for the partitioning of HO• reactivity with nucleic acids range from ~80% to more than 90% addition to nucleobase π-bonds. The electrophilic HO• preferentially adds to the C5-position of pyrimidines, with the ratio of the resulting C6- and C5-radicals ranging from 2:1 to 4–5: 1.17,198–200 Hydrogen atom abstraction from the methyl group of thymidine and 5′-methyl-2′-deoxycytidine is also a minor pathway for HO• and other diffusible species.201,202 Direct hydrogen atom abstraction form nitrogen–hydrogen bonds in purines have also recently been proposed but this has been questioned.203–205 In general, the reaction of HO• with purines is less well understood than with pyrimidines.</p><p>Hydroxyl radical formation via ionizing radiation is typically referred to as the indirect effect because HO• is the product of initial H2O ionization. The energy released upon γ-radiolysis can also interact directly with nucleic acids, and the corresponding direct effect of ionizing radiation accounts for approximately one-half of the induced damage. The direct effect of ionizing radiation initially produces radical cations, which as briefly mentioned when describing the reactivity of C4′-radicals, are responsible for electron transfer ('hole migration') in DNA (Section 7.5). The radical cations also give rise to other forms of DNA damage. The purines are more readily oxidized than the pyrimidines, and dG has a more favorable redox potential than does dA.206 Hydration of the nucleobase radical cations (eg 72, Scheme 33 ), followed by deprotonation, yields the same radical species resulting from HO• addition, which yield products such as thymidine glycol (Tg).207 In contrast, deprotonation produces the formal hydrogen atom abstraction products (eg 73), which yield O2 trapping products (eg 74, 75).208 This pathway has been observed in thymidine (Scheme 33 ) and 5-methyl-2′-deoxycytidine using photosensitization and illustrates that the direct and indirect (HO•) effects of ionizing radiation produce many of the same products.209–211 The purine radical cations tautomerize212 in competition with deprotonation to yield the neutral radicals whose direct formation, as mentioned above, is a topic of current interest. The neutral purine radicals, formal products of hydrogen atom abstraction, are believed to be the thermodynamic sinks during electron transfer and are unreactive with O2.213 Neutral nucleobase radicals are also believed to result from decomposition of the respective chloramines, which are generated during oxidative stress.44</p><!><p>A great deal of research has been carried out on the formation of various modified nucleotides following direct ionization or reaction with HO• under aerobic and anaerobic conditions.214,215 Pyrimidine nucleosides yield the hydrates (78, 79) and glycols (Tg) via regioisomeric HO• adducts (76, 77) under anaerobic and aerobic conditions, respectively (Scheme 34 ). The cytidine molecules are unstable upon saturation of the π-bond and hydrolyze to the corresponding modified uridines.216 The reaction of dG is especially rich because the initial two-electron oxidized product, 8-oxodGuo is more reactive than the native nucleoside and produces a variety of DNA lesions.217,218 Many of the DNA lesions produced by oxidation of native nucleotides are mutagenic.219–221</p><p> </p><p>To produce a strand break, a nucleobase radical or respective peroxyl radical must abstract a hydrogen atom from the carbohydrate backbone. Approximately 40% of the reactions between HO• and RNA result in strand scission.222 Since a minimum of 80% of HO• reactions occur with the nucleobases, at least 20% of the nucleobase radicals must ultimately transfer spin to the ribose ring.223 In contrast, strand scission in DNA following HO• to a nucleobase is significantly less efficient (≤5%) and nucleobase (peroxyl) radicals do not have to react with a 2′-deoxyribose ring to account for direct strand scission.224 A variety of mechanisms have been proposed for the requisite spin transfer from nucleobases to the (2′-deoxy)ribose that is required for strand scission, some of which involve initial one-electron oxidation of the nucleobase followed by hydration to form the formal HO• adduct.199,200,225–229</p><p>5-(2′-Deoxyuridinyl)methyl radical (74) and the respective radical from 5-methyl-2′-deoxycytidine have not been proposed to yield strand breaks. However, 74 and other nucleobase radicals have been implicated in reactions with adjacent nucleotides, resulting in two contiguously damaged nucleotides (eg 15). Such lesions are typically referred to as tandem lesions.230–235 Tandem lesions are a subset of clustered lesions, which are defined as two or more damaged nucleotides within one to two helical turns. Clustered lesions are of particular interest to radiation scientists because of their formation via HO• spurs and the difficulties that they present for DNA repair.85,236–239 In addition, 74 has been invoked as an intermediate in DNA electron transfer where it arises from deprotonation of the radical cation (Scheme 33 ).240,241 Stable products resulting from O2 trapping of 74 (76, 77) are observed under these conditions.</p><!><p>Several DNA and RNA pyrimidine nucleobase radicals have been generated from ketone precursors by the Norrish type I cleavage, aryl sulfides and phenyl selenides (Scheme 35 ). Formal hydrogen atom addition adducts of thymidine (80) and 2′-deoxyuridine (83) have been produced using these types of precursors, as has the C5-thymidine HO• adduct (76, Scheme 35 ).46,233,242–251 In each instance, the chemical integrity of the photochemical precursor with respect to generating the respective radical is characterized at the nucleoside level using standard product analysis conditions.</p><p>Direct strand breaks or alkali-labile lesions are not observed from 76, 80 or 83 under anaerobic conditions. (Reminder: Radicals are referred to using the same descriptor whether they are nucleosides or in biopolymers.) Small amounts of direct strand breaks are detected at the 5′-adjacent nucleotide when the formal C6-hydrogen atom adduct of thymidine (80) is produced under aerobic conditions.243 Strand scission is reduced approximately fourfold when the 5′-adjacent nucleotide is deuterated at the C1′-position.242 Deuteration of the C2′-, or C4′-positions has no effect suggesting that peroxyl radical 86 selectively abstracted the C1′-hydrogen atom from the 5′-adjacent nucleotide (Scheme 36 ). Additional support for internucleotidyl C1′-hydrogen atom abstraction was obtained when the tandem lesion containing 2-deoxyribonolactone (15) was observed in low yield when 80 was generated from 82 in a dinucleotide under aerobic conditions (Scheme 36 ).233 It is possible that the observed direct strand scission was an artifact of sample handling, as C1′-oxidation does not typically result in this product. Direct strand scission could also have resulted from hydrogen atom abstraction from another position that was not detected using deuterium labelling.</p><p>Independent generation of C6-radicals 76 and 83 (Scheme 35 ) failed to yield any direct strand scission under aerobic conditions.244,247,249 These radicals do produce a variety of alkali-labile tandem lesions that were detected by denaturing polyacrylamide gel electrophoresis and mass spectrometry. The peroxyl radical of the formal C5-hydrogen atom adduct of 2′-deoxyuridine (87) produces tandem lesions by reacting with the 5′- and 3′-adjacent nucleobases, and abstracting hydrogen atom(s) from the 5′-adjacent nucleotide but not the 3′-adjacent nucleotide (Scheme 37 ). Selective C1′-hydrogen atom abstraction from the 5′-adjacent nucleotide was based upon several observations, including a significant observed 2H kinetic isotope effect (KIE) (4.4 ± 0.1). Additional information was gleaned from chemical fingerprinting using a series of reactions that are diagnostic for 2-deoxyribonolactone (L), reaction with a biotinylated sensor that selectively tags L, as well as observation of 2-deoxyribonolactone upon MS analysis of photolyzed single-stranded dodecamer and trinucleotides containing 84.244,247,252,253 MS experiments also indicated that L is formed at the original site of the radical, presumably via intranucleotidyl C1′-hydrogen atom abstraction.244 There is no evidence for hydrogen atom abstraction from the 2′-deoxyribose ring of the 3′-adjacent nucleotide. The absence of reactivity at this carbohydrate component is consistent with the greater distance from the peroxyl radical center (87).247 Reaction at the adjacent nucleotides is also dependent upon the peroxyl radical configuration at C6 and the conformation about the glycosidic bond (Scheme 37 ). Kinetic evidence suggested that the syn- and anti-conformations of 6R-87 and 6S-87 are involved in tandem lesion formation (Scheme 37 ), and that proximity dictated by the right handed α-helical duplex structure strongly influences reactivity (Fig. 2).244,245,247 Molecular modelling reveals that the C1′-hydrogen atom of the 5′-adjacent nucleotide is accessible to anti-5R-87 (Fig. 2A). In contrast to 5R-87, the C1′-hydrogen atom of the 3′-adjacent nucleotide is not readily observable when viewed from the major groove when anti-5S-87 is present (Fig. 2B).</p><p>Incorporation of 5,6-dihydrothymidine (dHT) at the 5′-adjacent nucleotide to where 87 is generated significantly reduced the contribution of piperidine-labile lesions at that position, which are indicative of nucleobase lesions, indicating that the majority of 5′-tandem lesions involved addition to the π-bond.247 Some of the tandem lesions detected by MS are attributable to peroxyl radical addition to adjacent nucleotides.244,247 Competitive kinetics using GSH as a competitor provided an estimate of the rate constants for reaction of 87 with the 5′-adjacent nucleotide. The rate constant for reaction with the adjacent nucleotide was almost 30 times greater for a 5′-dG (1.2 ± 0.2 × 10−1 s−1) than with a 5′-dT (4.4 ± 0.6 × 10−3 s−1).244,249 This is consistent with the more electron-rich nature of dG and its relatively favorable oxidation potential compared to thymidine.206</p><p>Although tandem lesions were detected from the peroxyl radical (88, Scheme 38 ) of the C5-hydroxyl radical adduct of thymidine (76) and some were characterized by MS (eg 89, 90, Scheme 38 ), the reactivity of this peroxyl radical was considerably different than that of 87.249 For instance, there was no evidence for intramolecular hydrogen atom abstraction or from the 5′-adjacent nucleotide. In addition, competitive kinetics using GSH indicated that 88 reacted with a 5′-adjacent dG approximately one-half as fast (7.3 ± 0.9 × 10−2 s−1) as did 87. The slower reactivity of 88 than 87 is consistent with the rate constants for the respective nucleobase radicals (Scheme 35 ). The thymidine C5-hydroxyl radical adduct (76, Scheme 35 ) reacts with β-mercaptoethanol (BME) unusually slowly (k = 4.7–5.7 ± 0.1 × 105 M−1s−1) and considerably more slowly than does 83 (k = 8.8 ± 0.5 × 106 M−1s−1).246,248 Although it is not known why 88 reacts more slowly and selectively within DNA than does 87, the possibility that disubstitution at C5 of the pyrimidine increases steric hindrance and/or disrupts base stacking making it more difficult for the former to access the adjacent nucleotide was considered.249 The disruption of base stacking by 5,6-dihydrothymidines that are disubstituted at C5 is an outcome of computational studies.254,255</p><p>Independent generation of 88 was also used to test a proposal that pyrimidine peroxyl radicals contribute to DNA electron transfer by oxidizing purine nucleotides (Fig. 3).256 Independent generation of 88 at a defined site of a carefully chosen duplex sequence provides a facile means for detecting electron transfer by monitoring DNA damage at distal sites. The 5′-dGGG sequence is often used as a tool for monitoring DNA oxidation (Fig. 3). Outer sphere oxidation of dG by a DNA peroxyl radical was expected to be uphill by ~0.23 eV.206,257 Although a variety of oxidation products were detected, independent generation in a variety of duplex sequences that are frequently used to probe for electron transfer provided no evidence for this pathway.249,258–260</p><p>Overall, independent generation of HO• (and formal H•) addition products indicate that these radicals and their respective peroxyl radicals do not yield direct strand breaks. However, the respective peroxyl radicals produce a variety of tandem lesions. Although it is not possible to generate the analogous 2′-deoxycytidine radicals due to anticipated hydrolysis of the 5,6-dihydropyrimidine precursors, consideration of sterics suggests that their reactivity will be similar to the 2′-deoxyuridine adducts described above, which are more reactive than that of thymidine.</p><!><p>The formal C5- (91) and C6-hydrogen atom adducts (92) of uridine have been generated from the respective t-butyl ketones (93, 94, Scheme 39 ).261–264 Direct strand breaks were observed at the 5′-adjacent ribonucleotides when 91 or 92 were generated (Scheme 40 ).261–263 The C6-radical (91) also yielded intranucleotidyl cleavage, whereas a low level of cleavage at the nucleotide where 92 was generated was attributed to an artifact. Strand scission was as much as four-fold more efficient in duplex RNA than in single-stranded substrates. In addition, direct strand scission was at least seven-fold more efficient under anaerobic conditions than when O2 was present.</p><p>Strand scission products were characterized by gel electrophoresis with the aid of chemical and enzymatic reactions, and MALDI-TOF MS. Radicals 91 and 92 gave rise to common products (Scheme 40 ). In both instances a significant deuterium isotope effect was observed when the C2′-position of the 5′-adjacent nucleotide was labelled. Radical 92 yielded a KIE = 3.6 ± 0.7 but no effect when the C3′-position was deuterated.261 5′-Internucleotide strand scission via 91 was even more strongly affected by C2′-deuteration.263 The mechanism for strand scission was proposed to involve C2′-hydrogen atom abstraction (23), followed by heterolytic cleavage of the 3′-phosphate (Scheme 13 ). As discussed above (See: Generation and reactivity of the 2′-radical in RNA, Section 5.2), how the radical cation is transformed into the respective ketone (29, Scheme 14 ) and 3′-phosphate termini products is uncertain. The possibility that 26 generated from 91 is reduced by a guanosine within the RNA duplex was examined using a variety of sequences but no evidence for this process was obtained.262</p><p>Independent generation of 91 and 92 was carried out prior to reports on C2′-radical generation.109 At that time it was uncertain whether more efficient strand scission under anaerobic conditions was due to O2 trapping of 23 in competition with strand scission or less efficient hydrogen atom abstraction by the corresponding nucleobase peroxyl radicals (95, 96). It was also uncertain whether C2′-hydrogen atom abstraction or fragmentation from 23 was the rate determining step in strand scission. The recent independent generation of 23 from 24 (Scheme 12 ) revealed that O2 trapping does not compete with strand scission from this reactive intermediate and that hydrogen atom abstraction by the nucleobase radicals (91, 92) must be the rate determining step.109 Competitive kinetics revealed that 92 abstracts the C2′-hydrogen atom from the 5′-adjacent nucleotide almost 25-times faster than does 91.261,262 Computational studies corroborated these findings.261 The C5-carbon–hydrogen bond in N1-methyl-5,6-dihydrouracil (97) was determined to be at least 2.8 kcal/mol stronger (95.3 kcal/mol) than the C6-carbon–hydrogen bond.</p><p> </p><p>The results are consistent with γ-radiolysis experiments that showed that RNA is significantly more susceptible to strand scission than is DNA.223,224 Furthermore, the mechanism of strand scission is in line with expectations set by computational studies that indicate that the C2′-hydrogen atom, whose respective carbon–hydrogen bond is on average 4.8 kcal/mol weaker than any other in a ribonucleotide, should be the most readily abstracted.35 However, mechanistic questions remain. How the 3′-phosphate product is formed is still unclear. In addition, the transformation of 26 into the 5′-fragment containing a 3′-terminal 3′-deoxy-2′-ketouridine (29) is also unknown.</p><!><p>5-(2′-Deoxycytidinyl)methyl radical (98) was produced from the respective phenyl sulfide (99) via 254 nm photolysis.265,266 Initial studies in dinucleotides revealed that 98 adds to the C8-position of dG to form intrastrand cross-linked products (Scheme 41 ). These products were also observed when 98 was generated from 99 in chemically synthesized oligonucleotides, and when DNA was subjected to γ-radiolysis.266 In DNA, 100 is favored over 101 and is even detected, albeit in almost 20-fold lower yield when 98 is generated under aerobic conditions.</p><p>5-(2′-Deoxyuridinyl)methyl radical (102) was independently generated from a methoxy substituted arylsulfides (103a–c) and phenyl selenide (104) via 350 nm irradiation (Scheme 42 ).267–269 The Norrish type I photo-cleavage of 105 also produced 102.270 Thiol and O2 radical trapping products were observed under the appropriate anaerobic and aerobic conditions when monomeric 102 was generated.267 Although intrastrand cross-links analogous to those from 98 (Scheme 40 ) were not detected, interstrand cross-links with the opposing dA (Scheme 44 ) were formed upon photolysis of duplex DNA containing 103c or 104 (Scheme 41 ).268,271,272 The formation of interstrand cross-links from 102 was surprising given the absence of a similar product from 98, as well as in DNA electron transfer experiments where products attributable to 5-(2′-deoxyuridinyl)methyl radical (102) were observed.241,273,274</p><p>Another surprise was that interstrand cross-links were also formed independently of O2. However, this was initially rationalized by demonstrating that O2 reacted reversibly with 102 (Scheme 43 ).268 Nonlinear regression analysis of the ratio of thymidine to oxygenated products (eg 107) as a function of GSH concentration provided an estimated rate constant for GSH trapping of 102 (kGSH = 6.9 × 106 M−1s−1) consistent with expectations for reaction of an alkyl radical with the thiol.60 The accuracy of the thiol trapping rate constant validated the extracted rate constant for O2 loss from 106 to reform 102 (kO2 = 3.4 s−1), which is consistent with rate constants reported for O2 loss from other peroxyl radicals.59,142</p><p>The isolated cross-link resulted from formal addition of the radical to the N6-amine of dA, an improbable process. It was proposed that 102 adds to the N1-position, and following a second one-electron oxidation, that the N1-alkylated product (108) rearranges to the final product (109, Scheme 44 ).275,276 Rearrangement of the initially formed N1-alkylation product to 109 was supported by nuclear magnetic resonance (NMR) experiments using momomeric substrates, as well as doubly-13C and 15N labelled DNA substrate.272 The same product was formed from 104 under mild oxidative conditions (eg NaIO4, H2O2, 1O2) and the mechanism involving oxidation to 110 and its rapid rearrangement to 111 (Scheme 44 ) was supported by NMR experiments using 15N and 13C labelled DNA substrates.272,277,278 Oxidatively induced cross-linking from 104 has proved useful in a variety of applications, including single nucleotide polymorphism detection, triple helical (triplex) DNA detection and as a tool for examining DNA–protein interactions.279–281</p><p>Despite the interesting and useful chemistry emanating from 104 (and 103c), the attribution of DNA interstrand cross-links from 5-(2′-deoxyuridinyl)methyl radical (102) was inconsistent with DNA electron transfer experiments in that formation of the same radical intermediate failed to yield cross-links.21,240,273,274 This discrepancy was resolved by Weng who determined that in addition to generating 102 upon photolysis, 104 and 103c also produce the carbocation (112, Scheme 45 ).282 Carbocation 112 may arise via heterolytic cleavage of the excited state precursor or electron transfer with the initially formed radical pair.283 The carbocation is trapped by nucleophiles, including methoxyamine (113) and t-butyl thiol. The latter is important because it helps explain why initial studies supported interstrand cross-link formation from 5-(2′-deoxyuridinyl)methyl radical (102).268 Importantly, while methoxyamine had no effect on the yield of radical trapping products (eg 107, Scheme 43 ), it quenched DNA interstrand cross-link formation upon irradiation of DNA containing 103c in a concentration-dependent manner. In contrast, Tempo derivative 115 had no effect on cross-linking but did trap 102 (114) when it was produced in oligonucleotides. These experiments clearly indicated that, although 104 and 103c produce 5-(2′-deoxyuridinyl)methyl radical (102), the precursors also produce the carbocation (112), and it is this latter species that is responsible for DNA interstrand cross-link formation. Computational experiments on free nucleobases corroborated these findings.282 Addition of (5-uracil)methyl radical to adenine was determined to face a 50 kJ mol−1 barrier, whereas reaction of the corresponding carbocation analogous to 112 was barrierless.</p><!><p>Limited data are available on the species that result from formal hydrogen atom abstraction from either dA or dG. As mentioned above, these radicals play a pivotal role in DNA electron transfer and may also be important species that are produced directly from the respective chloramines via oxidative stress (See Sections 2 and 7.5). Photolyses of 117 and 118 were proposed to produce the formal N1-hydrogen atom abstraction product from dG (116, Scheme 46 ).284,285 Data from 118 were limited to quantification of dG and the proposed generation of 117 is complicated by a required rearrangement to 119 and subsequent photolysis.284 Independent synthesis of 118 was reported to produce 116 upon Pyrex filtered photolysis, and is supported by transient spectroscopy.285 Studies on the formation of 116 from 118 in oligonucleotides have not yet been reported.</p><p>The N6-aminyl nucleoside radical of dA (120) has also been produced from a photochemical precursor (121) but not within DNA (Scheme 47 ).286 Photolysis (315 ± 2 nm) of phenylhydrazone 121 in the presence of GSH (50 mM) provides a 73% yield of dA. In the absence of excess thiol, the yield of dA is reduced (32%) as expected and is accompanied by the radical recombination product resulting from the dimerization of 120.</p><!><p>Although the ability to independently generate reactive intermediates has increased our understanding of nucleic acid damage processes, the brief description of neutral purine radical generation (Section 9) illustrates that unanswered questions remain regarding the roles of these reactive intermediates in DNA. Nucleobase radical cations, the family of species produced via the direct effect of ionizing radiation have also never been independently generated in nucleic acids. Furthermore, little is known about the effects of protein binding on the reactivity of nucleic acid reactive intermediates.287,288 Other gaps in our knowledge were pointed out throughout the above review. Clearly, there is still much to be done.</p>

PubMed Author Manuscript

4S-Hydroxylation of insulin at ProB28 accelerates hexamer dissociation and delays fibrillation

Daily injections of insulin provide lifesaving benefits to millions of diabetics. But currently available prandial insulins are suboptimal: The onset of action is delayed by slow dissociation of the insulin hexamer in the subcutaneous space, and insulin forms amyloid fibrils upon storage in solution. Here we show, through the use of non-canonical amino acid mutagenesis, that replacement of the proline residue at position 28 of the insulin B-chain (ProB28) by (4S)-hydroxyproline (Hzp) yields an active form of insulin that dissociates more rapidly, and fibrillates more slowly, than the wild-type protein. Crystal structures of dimeric and hexameric insulin preparations suggest that a hydrogen bond between the hydroxyl group of Hzp and a backbone amide carbonyl positioned across the dimer interface may be responsible for the altered behavior. The effects of hydroxylation are stereospecific; replacement of ProB28 by (4R)-hydroxyproline (Hyp) causes little change in the rates of fibrillation and hexamer disassociation. These results demonstrate a new approach that fuses the concepts of medicinal chemistry and protein design, and paves the way to further engineering of insulin and other therapeutic proteins.

4s-hydroxylation_of_insulin_at_prob28_accelerates_hexamer_dissociation_and_delays_fibrillation

<p>Blood glucose levels are tightly controlled in mammals through a sensitive regulatory system mediated by insulin, a 51-amino acid endocrine hormone composed of two disulfide-linked polypeptide chains (designated A and B). Upon binding to its receptor, insulin initiates a signaling cascade that accelerates glucose uptake and glycogen production. In diabetic patients, this system malfunctions, and glucose levels must be controlled through subcutaneous injections of insulin1. The C-terminus of the B-chain is important in mediating dimerization of the hormone2–3, and the flexibility of the B-chain C-terminus is believed to contribute to aggregation through formation of amyloid fibrils4–6. Pharmaceutical formulations of insulin are stabilized with respect to fibrillation by addition of zinc and phenolic preservatives, which drive assembly of the R6 hexamer (Fig. 1a)7–9. The R6 form of insulin is inactive and dissociates slowly to its active monomeric form after subcutaneous injection; the lag time for dissociation delays the onset of action10. Mutation of ProB28 yields rapid-acting insulins (RAIs) by disrupting contacts that are critical for dimer formation8, but replacement of Pro through conventional mutagenesis also increases the flexibility and perturbs the trajectory of the protein backbone (Fig. 1B). We sought a means to disrupt the dimer interface without releasing the conformational constraints characteristic of proline by using non-canonical amino acid (ncAA) mutagenesis11–13. Specifically, we introduced hydroxyl groups at the 4-position of ProB28 (Fig. 1B, C) by replacing Pro with Hzp or Hyp. In addition to introducing a polar functional group and the capacity for hydrogen-bonding (including transannular hydrogen bonding), hydroxylation at the 4-position is known to alter the endo/exo preference of the pyrrolidine ring and the cis/trans equilibrium of the backbone amide bond14–16.</p><p>We expressed modified proinsulins (PIs) in the proline-auxotrophic E. coli strain CAG18515 in M9 minimal media supplemented with Hyp or Hzp. The extent of replacement of Pro by either Hyp or Hzp was approximately 90%17–18 as determined by matrix-assisted laser desorption/ionization mass spectrometry (MALDI-MS; fig. S1). Denatured PIs were purified by Ni-NTA affinity chromatography in yields of 32 mg/L for Hzp-PI and 29 mg/L for Hyp-PI (table S1) from the inclusion body fraction. The PIs were refolded and cleaved with trypsin and carboxypeptidase B19. The resulting mature insulins were purified by reversed phase HPLC18–19, and proper proteolytic processing of each variant was verified by MALDI-MS (table S1). Wild-type insulin (ProI) and RAI Aspart (AspI, in which ProB28 is replaced by aspartic acid) were produced similarly. All of the variants caused similar reductions in blood glucose upon subcutaneous injection into diabetic mice (Fig. 1D)2, 20–22. RAIs cannot be distinguished from ProI in rodent models23.</p><p>In the absence of Zn2+ and phenolic preservatives, insulins dimerize with a dissociation constant (KD) of approximately 10 µM. In contrast, KD for RAIs is typically >500 µM, and it is believed that destabilization of the dimer interface causes the accelerated onset of insulin action after subcutaneous injection20, 24–25. Monomeric forms of insulin give rise to characteristic circular dichroism (CD) spectra with distinct minima at 208 and 222 nm26–27 (e.g., AspI; Fig. 2A). Dimerization causes a loss of negative ellipticity at 208 nm27 (e.g., ProI; Fig. 2A). At a concentration of 60 µM, Hypl appears to be monomeric (with a CD spectrum nearly identical to that of AspI; Fig. 2A) while the spectrum of HzpI suggests a dimeric insulin (Fig. 2A). Sedimentation velocity (SV) and sedimentation equilibrium (SE) experiments were consistent with the results of the CD analysis (fig. S3). SE data were fitted to a model of monomer-dimer-hexamer self-association (SEDPHAT) 28–29, and yielded monomer-dimer dissociation constants (KD) of >200 µM and 25 µM for HypI and HzpI, respectively.</p><p>Previous studies of RAIs have shown that destabilization of the dimer interface correlates with accelerated dissociation of the hexamer and rapid onset of insulin action8, 13. Triggered dissociation of Zn2+-hexamers by addition of terpyridine30 revealed nearly identical rates of dissociation for HypI and ProI, (τ1/2 = 87.0 ± 10 s and 90.4 ± 4.2 s, respectively; Fig. 2B and fig. S4) while HzpI exhibited kinetics similar to those of AspI (τ1/2 = 53.6 ± 3.7 s and 42.7 ± 4.3 s, respectively; Fig. 2B and fig. S4). We found these results surprising – replacement of Pro by Hyp destabilizes the dimer but has essentially no effect on hexamer dissociation, while introduction of Hzp causes little change in dimer stability but a substantial increase in the rate of hexamer disassembly.</p><p>Each of the insulin variants was subjected to fibrillation lag time analysis (Fig. 2C)31. We found similar times to onset of fibrillation for HypI, ProI and AspI; in contrast, HzpI is markedly more resistant to aggregation, with a mean time to onset more than three-fold longer than that observed for ProI. The behavior of HzpI is especially striking, in that it combines fast hexamer dissociation with enhanced stability toward fibrillation.</p><p>Each subunit in the insulin hexamer adopts one of two conformational states (T or R), depending on the concentration of phenolic ligand (Fig. 1A)13. Pharmaceutical formulations are prepared in the more stable R6 form, whereas the T-state is observed in the absence of phenolic ligands, most commonly in the form of T2-dimers32. To elucidate the molecular origins of the dissociation and fibrillation behavior of HypI and HzpI, we examined crystal structures of both states.</p><p>Hydroxylation at ProB28 does not cause substantial perturbation of the overall insulin structure (Fig. 3, fig. S5). In comparison with ProI, the backbone RMSD values of HypI and HzpI are 0.31 Å (T2- HypI), 0.44 Å (T2- HypI), 0.38 Å (R6- HypI) and 0.69 Å (R6- HypI)33. The most notable feature of the HzpI structures is the proximity of the hydroxyl group of Hzp to the backbone carbonyl oxygen atom of GluB21′, which lies across the dimer interface (denoted by prime; Fig. 3B, E). The inter-oxygen distances (2.8 Å in the T2 structure, 2.7 Å in R6), are consistent with the formation of strong hydrogen bonds between the hydroxyl group of HzpB28 and the backbone carbonyl of GluB21′ in both structures. An analogous hydrogen bond has been observed in a structure (PDB ID: 1ZEH) of R6-AspI co-crystallized with m-cresol34; here the phenolic ligand serves as the hydrogen-bond donor (fig. S6). Although the significance of this hydrogen bond has not been discussed in the literature, we suggest that it may play an important role in determining the relative stabilities of the insulin species involved in dissociation and fibrillation. In contrast to the (4S)-hydroxyl group of Hzp, the (4R)-hydroxyl of HypB28 does not contact any crystallographically resolved hydrogen bond acceptor in the T2-structure (Fig. 3C), and appears to bond to an ordered water molecule in the R6-hexamer (Fig. 3F). The absence of new hydrogen-bonding interactions is consistent with the unaltered dissociation and fibrillation kinetics of HypI.</p><p>Taken together, our results show that replacement of Pro by Hzp at position 28 of the insulin B-chain introduces a new hydrogen bond across the inter-subunit interface, accelerates hexamer dissociation and delays the onset of fibrillation (Table 1). We suggest that the hydrogen bond between Hzp and Glu21′ may stabilize the dimer relative to the hexamer, or perhaps reduce the energy of the transition state for the conformational change from the R-state to T-state, and thereby speed disassembly. The delayed onset of fibrillation may reflect changes in the structure and dynamics of the HzpI monomer or in the kinetics of fibril nucleation. Subtle conformational effects caused by 4S-substitution on the pyrrolidine ring may also contribute to the observed behavior.35 Whether or not these hypotheses are correct, the results described here demonstrate the power of ncAA mutagenesis to control functionally relevant biophysical properties of therapeutic proteins. We anticipate that this approach will find increasing application in the design of antibody-drug conjugates, bispecific antibodies, and other novel protein therapeutics.</p>

PubMed Author Manuscript

hIAPP forms toxic oligomers in plasma

In diabetes, hyperamylinemia contributes to cardiac dysfunction. The interplay between hIAPP, blood glucose and other plasma components is, however, not understood. We show that glucose and LDL interact with hIAPP, resulting in \xce\xb2-sheet rich oligomers with increased \xce\xb2-cell toxicity and hemolytic activity, providing mechanistic insights for a direct link between diabetes and cardiovascular diseases.

hiapp_forms_toxic_oligomers_in_plasma

<p>Diabetes is a widespread disease affecting more than 415 million people and claiming more than 5 million lives worldwide only in 2015.1 Type 2 diabetes (T2D) is the most common form of the disease and accounts for approximately 90% of all cases.2 While the exact cellular pathomechanism of T2D remains elusive, at least six principal factors have been identify to contribute to T2D, which include insulin resistance, lipotoxicity, endoplasmic reticulum oxidative stress, tissue inflammation response, amyloid deposition and β-cell failure.3, 4 Pancreatic amyloid deposits are composed of the 37 residue peptide hormone human Islet Amyloid Polypeptide (hIAPP).5, 6 hIAPP is thought to be involved in the slowdown of post-meal increase of plasma glucose concentration.7 The mechanism of its cytotoxicity in T2D is, however, only partially understood.8 Recent studies have shown that soluble oligomers of hIAPP are responsible for cell toxicity and cell death.9–11 There is only little knowledge about the factors that lead to hIAPP aggregation in vivo. hIAPP is stored in a functional and soluble form in insulin granules in β-cells at concentrations up to 1–4 mM.12 The cellular environment has thus a strong impact on the rate of hIAPP aggregation. Factors affecting aggregation include pH,13 divalent cations such as Zn2+,14 insulin,15, 16 and the redox environment.17, 18 It is known that the serum of diabetic patients contains low concentrations of high-density lipoproteins (HDL) and high concentrations of very low-density lipoproteins (VLDL), low-density lipoproteins (LDL) and plasma triglycerides (TG).19 In addition, the concentration of sugars and insulin is increased. Whereas the blood glucose concentration in healthy individuals is on the order of 6 mM, it ranges between 10 mM and >14 mM in people affected by T2D.20 T2D is linked with other diseases such as Alzheimer's disease,21, 22 cardiac dysfunction in obesity,23 kidney disease and other renal failures.24 The aim of this study is to investigate the impact of major plasma components known to be disturbed in T2D on the kinetic and structural aspects of hIAPP oligomerization and fiber formation. We find that LDL stabilizes high molecular weight hIAPP species and effects hIAPP aggregation. We observe further that sugars induce liquid-liquid phase separation and yield increased cellular toxicity. We propose that these oligomers may be the link between T2D and associated complications such as cardiac dysfunction in obesity.</p><p>In the past, we have studied the aggregation behaviour of hIAPP in vitro.18 We now want to understand how hIAPP behaves under physiological conditions. To address this question, we employ blood plasma of diabetic transgenic mice. The mouse models harbor either one (+/TG mice) or two copies (TG/TG mice) of human IAPP.25 As a control, we used plasma of +/+ wild type mice. Although pancreatic amyloid fibrils are present in both the +/TG and TG/TG mice, only TG/TG mice are affected by diabetes.26 It is known that the plasma of diabetic patients becomes lactescent due to an increase of dispersed lipids27 (increasing lactescence +/+ < +/TG < TG/TG) (Fig. 1A). In addition, previous results showed decreased HDL-cholesterol and insulin levels and high levels of plasma glucose, urea and LDL-cholesterol in hIAPP transgenic mice.26</p><p>To understand the impact of plasma composition on the conformation and behavior of hIAPP, we dissolved isotopically labeled hIAPP in plasma. As a control, the peptide was measured in a standard buffer at pH 7.5. To estimate the stability of hIAPP in solution, we followed the kinetics by monitoring the solution-state NMR signal intensities over time. We find that hIAPP is most stable in plasma of TG/TG animals (Fig. 1B). In general, the NMR signal intensities decrease more slowly if the peptide is dissolved in plasma than in reference buffer, in agreement with previous studies which showed that crowding agents can induce a retardation of aggregation.28 Diabetic (TG/TG) and non-diabetic (+/+, +/TG) plasma differs mostly in the concentration of LDL and sugars.29 We therefore analyzed the effect of these two components on the aggregation kinetics, as well as on the conformation of hIAPP. Aggregation of hIAPP was monitored by NMR. As illustrated in Fig. 1C, LDL and sugar solutions (fructose or glucose) had a similar stabilizing effect as TG/TG plasma. To obtain a better understanding of the aggregation kinetics, we performed Thioflavin T (ThT) assays (Fig. 1D,E). In presence of LDL, we observe an increase of the lag time. The maximum fluorescence intensity is, however, unchanged. By contrast for fructose, the maximum fluorescence is significantly reduced. The results suggest that LDL, glucose and fructose have a direct impact on the aggregation state of hIAPP. This is in agreement with results from Kedia et al. who showed that sugars favor Aβ42 oligomer formation.30 To characterize the oligomerization state in more detail, we performed Dynamic Light Scattering (DLS) experiments. DLS reveals an increase of the hydrodynamic size of the hIAPP assemblies formed in the presence of LDL, glucose and fructose (Fig. 1G–I). In addition, we performed Western blot experiments using the hIAPP specific antibody A133 (Fig. 1F).31 As shown previously,18,32 hIAPP yields bands in the molecular weight range from 15 kDa to approximately 100 kDa. We find that +/+, +/TG, and TG/TG plasma induced oligomers have a similar molecular weight as the oligomers that are formed in the presence of 2 μM LDL. By contrast, glucose or fructose induced oligomers yield a band corresponding to a molecular weight on the order of 15 kDa, suggesting that hIAPP interacts differently with sugars and LDL.</p><p>To test the hypothesis whether LDL, glucose and fructose are able to induce hIAPP oligomeric structures, we carried out fluorescence and Differential Interference Contrast (DIC) microscopy experiments. We find that sugar and LDL induced hIAPP aggregates yield an intrinsic fluorescence (Fig. 2), which has been observed previously for amyloid fibrils (Fig. S3).33,34 hIAPP alone yields only few colloidal structures. In the presence of sugars, however, larger and more aggregate structures are observed. In DIC experiments, LDL particles appear spherical and non-fluorescent. In the presence of hIAPP, however, LDL particles show an intrinsic fluorescence indicating that LDL interacts with hIAPP. To find out if these soluble oligomers are toxic in vivo, we carried out cellular toxicity assays using the pancreatic β-cell line b-tc3. B-tc3 cells were incubated with hIAPP in combination with glucose, LDL or fructose. Increasing concentrations of hIAPP yield a decrease of cellular viability in all cases (Fig. 2I), suggesting that LDL and glucose/fructose increase cellular toxicity. In addition, we performed a hemolytic assay to investigate whether hIAPP is able to lyse red blood cells (Fig. 2J). We find that the presence of LDL or sugars such as glucose or fructose increases the hemolytic activity strongly.</p><p>To obtain deeper insight into the interaction mechanism, we carried out solution-state NMR experiments. First, we dissolved isotopically labeled hIAPP35 in plasma (+/+, +/TG and TG/TG) and recorded HMQC experiments (Fig. 3A). As a control, spectra of hIAPP dissolved into phosphate buffer (pH 7.5) are recorded. Fig. 3C represents the chemical shift changes observed for the three different plasma samples with respect to the control sample. We find that chemical shifts of residues located in the N-terminal region of hIAPP are perturbed. To identify the molecules that are responsible for these shift changes, we recorded experiments for hIAPP dissolved in a 2 μM LDL solution (Fig. 3B). Many of the chemical shift changes in the N-terminal region of hIAPP observed in plasma are also seen in the presence of LDL, suggesting that lipid containing particles are responsible for these spectral changes. This observation is consistent with a previous study where it was shown that the N-terminal part of hIAPP interacts with membrane nanodiscs.36 To characterize the structural changes in hIAPP which are induced by TG/TG plasma, we performed 3D HNCA-experiments. The analysis of the hIAPP Δδ(13Cα) chemical shift differences (Fig. 3D) suggests that the peptide is converted from a random coil structure with an α-helical propensity in phosphate buffer18 to a conformer which is rich in β-sheet structure. The Random Coil Index (RCI) supports this finding (Fig. 3E). Based on the Δδ( 13Cα) data and the RCI, we propose a secondary structure model for hIAPP in buffer and TG/TG plasma (Fig. 3F,G).</p><p>After intake of food, the blood glucose level rises, stimulating the pancreatic secretion of insulin and hIAPP. In diabetes, these two peptide hormones become overexpressed, which can drive aggregation and in turn cellular toxicity.10 So far, aggregation of hIAPP has been observed to occur only in pancreatic beta-cells, without affecting significantly other organs. We find that key molecules in plasma which are known to be elevated in T2D, such as LDL and glucose induce a stabilization of a non-aggregation prone state of the peptide. hIAPP solubilized in plasma and LDL yields high molecular weight oligomer complexes with an apparent molecular weight in the range of ~15 kDa to ≥ 100 kDa. LDL and sugar induced hIAPP oligomers yield an increased hemolytic activity as well as cellular toxicity. Interestingly, the LDL/sugar induced hIAPP oligomeric assemblies display an intrinsic fluorescence. This might indicate that β-sheets are already preformed in these oligomeric structures, as otherwise no intrinsic fluorescence would be observed. Intrinsic fluorescence of oligomeric Aβ assemblies has been observed previously by super-resolution microscopy.37 By NMR, we have observed that the peaks originating from residues located in the N-terminal region of the peptide are changing in chemical shift in the presence of diabetic plasma and LDL, suggesting that this part of hIAPP is involved in interactions with the respective plasma components. At the same time, the C-terminal part of the peptide adopts a certain propensity for β-sheet structure. For the hIAPP - glucose/fructose mixtures, no chemical shift changes are observed after mixing. We find, however, that aggregation of hIAPP is significantly retarded in the presence of sugars. Under these conditions, hIAPP populates oligomers that yield a reduced ThT fluorescence signal, but a very pronounced intrinsic fluorescence. In EM, hIAPP adopts very thin and fragile protofibrillar structures in presence of fructose and glucose (Fig. S2). We speculate that sugars might induce formation of hIAPP-rich colloidal structures. In fact, glucose is a cosmotropic osmolyte,38,39 coordinates water and gets excluded from interactions with the hIAPP backbone, thereby inducing structure. At the same time, hIAPP might exchange between a high molecular weight and a monomeric state which would allow to explain the relatively high intensities observed in the NMR experiments. In case of LDL, EM shows that hIAPP aggregates are lined up along a series of LDL particles (Fig. S2). The presence of a lipid particle might catalyze the transition between a monomeric and a high molecular weight state of hIAPP, which might facilitate the detecton of a hIAPP monomeric state. The plasma concentration of hIAPP has been reported to be in the pM range.40, 41 Nevertheless, antibodies which specifically recognize oligomeric hIAPP assemblies were identified in the serum of diabetic patients,10 suggesting that hIAPP oligomers can assemble even at these very low concentrations.</p><p>In conclusion, we have shown that hIAPP, LDL, and sugars mutually interact, suggesting a direct link between TD2 and cardiovascular diseases.</p>

PubMed Author Manuscript

Long-range electrostatic corrections in multipolar/polarizable QM/MM simulations

Taking long-range electrostatic effects into account in classical and hybrid quantum mechanics\xe2\x80\x93molecular mechanics (QM/MM) simulations is necessary for an accurate description of the system under study. We have recently developed a method, termed long-range electrostatic corrections (LREC), for monopolar QM/MM calculations. Here, we present an extension of LREC for multipolar/polarizable QM/MM simulations within the LICHEM software package. Reaction barriers and QM\xe2\x80\x93MM interaction energies converge with a LREC cutoff between 20 and 25 \xc3\x85, in agreement with our previous results. Additionally, the LREC approach for the QM\xe2\x80\x93MM interactions can be smoothly combined with standard shifting or Ewald summation methods in the MM calculations. We recommend the use of QM(LREC)/MM(PME), where the QM region is treated with LREC and the MM region is treated with particle mesh Ewald (PME) summation. This combination is an excellent compromise between simplicity, speed, and accuracy for large QM/MM simulations.

long-range_electrostatic_corrections_in_multipolar/polarizable_qm/mm_simulations

1 Introduction<!>2.1 Multipole moments<!>2.2 QM/MM Hamiltonian<!>2.3 Polarization<!>3 Long-range electrostatics<!>3.1 Shifting<!>3.2 Ewald and PME<!>3.3 LREC<!>4.1 Effective multipole moments<!>4.2 Mixed LRE methods<!><!>4.2 Mixed LRE methods<!>5.1 Computational details<!>5.2 QM\xe2\x80\x93MM interactions<!>5.3 Reaction barriers<!>6 Conclusion

<p>Long-range electrostatic (LRE) interactions are important for the accurate determination of enzyme structures and reaction paths [7, 8, 38, 51]. Unfortunately, directly incorporating LRE interactions in quantum mechanical–molecular mechanical (QM/MM) simulations can be challenging due to the necessity of modifying the Fock matrix elements.</p><p>The most straightforward approach for the treatment of electrostatic interactions in QM/MM simulations is to simulate only a small portion of the solvent around a protein or enzyme active site [5] (i.e., embedding a protein in a water droplet). The droplet's liquid-like environment can be maintained by either freezing the positions of the molecules in the outer edge or applying a stochastic boundary potential [5]. Droplet simulations are efficient due to the small number of atoms, although, the LRE interactions are neglected. However, some of the missing LRE effects can be implicitly included by embedding the droplet in a classical continuum [2–4, 11, 23, 41].</p><p>An alternative approach involves using Mulliken charges to approximate the QM charge distribution in periodic images, so that the QM/MM calculation can be carried out using the Ewald summation or smooth particle mesh Ewald (PME) methods [18, 27, 34, 47]. QM/MM calculations with Ewald/PME methods are effective and can readily be improved by replacing Mulliken charges with charges fit from the electrostatic potential. Similar approaches have been developed to extend the Ewald, Wolf, and isotropic periodic sum methods to calculations with semi-empirical, density functional tight binding, Hartree–Fock, and density functional theory [22, 35, 36, 40, 50]. Very recently, Giese and York have reported the ambient-potential composite Ewald method, where no approximations are employed to represent the QM charge density in reciprocal space [20]. Most of these QM/MM LRE methods require modifications to the self-consistent-field (SCF) matrix elements or post-SCF corrections. While these approaches can be quite accurate, modifying SCF routines can limit which QM packages can be used for the QM/MM calculations.</p><p>Recently, Fang et al. [14] introduced a cubic smoothing function to include long-range electrostatic corrections (LREC) in QM/MM simulations. The LREC approach uses a combination of a smoothing function and the minimum image convention to scale the electrostatic interactions such that the potential and forces smoothly decrease to zero at a finite cutoff radius. While energy and force shifting approaches also smoothly truncate the potential at a finite radius [16, 28, 29, 35, 36, 44, 49], LREC has an exceptionally simple implementation in the QM/MM Hamiltonian. In the LREC approach, the external MM monopoles are scaled based on their distance from the QM region, which does not require modifications to the matrix elements or post-SCF corrections. The LREC method has been shown to calculate energies and forces that are within 0.2 % of the PME results when using cutoffs of 20–25 Å [14].</p><p>The complications of treating LRE interactions of monopoles also affect potentials that include multipole moments and explicit polarization. Truncation, shifting, and smoothing approaches are appealing for multipolar/polarizable QM/MM simulations due to the relative simplicity of introducing these methods into the Hamiltonian [14]. However, each type of shifted multipole–multipole interaction requires a different correction term. As will be shown below, the LREC approach produces a single scale factor for each MM atom, and hence, does not require modifications to the underlying QM software.</p><p>In this paper, we extend the LREC approach to atomic multipole moments within the LICHEM software package and implement QM(LREC)/MM(LRE) calculations, where the QM and MM regions are treated with a different LRE approaches. In Sect. 2, we review multipolar/polarizable QM/MM simulations, followed by a discussion of LRE methods in Sect. 3. Our extension of the LREC method to multipolar/polarizable QM/MM simulations is presented in Sect. 4, and we conclude our discussion by examining the performance of multipolar LREC.</p><!><p>The electron density and nuclei in a molecule create a continuous, and often, anisotropic electrostatic potential. In classical simulations, interactions due to the diffuse electron density are generally approximated by a set of pair-wise electrostatic potentials. One of the simplest methods for modeling the molecular electrostatic field is to place point-charges (monopoles) on the atomic centers. Atomic monopoles are often inadequate [8, 25], and some models augment the field by adding massless charged dummy atoms to the molecule [24, 33]. However, the dummy atoms can complicate the potential/dynamics during the simulations.</p><p>An alternative approach is to add higher-order moments (multipoles) of the electrostatic potential to the model [8, 30, 43, 45]. A multipole expansion is a Taylor series representation of the continuous electrostatic potential (Velst) [45], (1)Velst(r)≈∑n∞dn!drnVelst(r)rn, where r is the distance from the point of interest (e.g., nucleus, center of mass, etc) and n is the order of the moment. Typically, MM force fields truncate the Taylor series at the second order [8, 30, 43], and the moments can be determined from an analysis of the electron density or electrostatic potential.</p><p>The zeroth moment of the electrostatic potential is the monopole, which is equivalent to a point-charge. The first and second moments are referred to as the dipole and quadrupole, respectively. Multipole moments can produce an anisotropic electrostatic potential around individual atoms, as opposed to the spherical potential wells produced by monopoles. Thus, a set of atomic multipole moments, through the quadrupole moment, can reproduce the molecular electrostatic potential with a higher degree of accuracy than monopole moments alone [8, 45]. Atom centered multipolar models provide a reasonable compromise between computational cost and accuracy in the reproduction of the electrostatic potential at medium and long-range.</p><!><p>The polarizable QM/MM Hamiltonian may be expressed as (2)H=Hqm+Vmm+Vqmmm+Vpol,where Hqm is the unperturbed Hamiltonian for the QM subsystem, Vmm is the potential for the MM–MM interactions, Vqmmm is the potential for the QM–MM interactions, and Vpol is the potential due to the polarization of the QM/MM system.</p><p>In practice [26], Eq. 2 can be expressed as (3)H=Hqm′+Vmm′+Vmm,pol, where (4)Hqm′=Hqm+Vqmmm,mtp, and (5)Vmm′=Vmm+Vqmmm,bnd+Vqmmm,vdw.Here Hqm′ adds the MM multipole moments (Vqmmm,mtp) to the QM Hamiltonian, Vmm′ adds the QM–MM bonded (Vqmmm,bnd) and van der Waals (Vqmmm,vdw) interactions to the MM potential, and Vmm,pol is the many-body MM polarization energy. Thus, Eq. 3 divides the QM–MM interactions and polarization between the QM and MM calculations.</p><p>The QM/MM Hamiltonian becomes more complicated when the system is periodic. Since the Coulomb interactions have a relatively long range, particles 20 Å or more from the QM region can have non-negligible contributions to the energy and forces [6, 14, 44, 48]. One would expect that monopole–monopole interactions in a neutral system would begin to cancel with interactions of the opposite sign. In fact, Wolf demonstrated [48] that the r−1 dependence of isolated charged particles behaves as a r−5 potential in a homogeneous neutral environment. However, since the pair-wise terms in the electrostatic potential often have relatively large magnitudes beyond 10 Å, the potential cannot simply be truncated [1, 6, 8, 15, 16, 37, 38, 42, 44, 48, 51]. As will be discussed below, many LRE methods can be integrated into the periodic QM/MM Hamiltonian.</p><!><p>Polarizable QM/MM simulations combine quantum polarization (due to many-body electrostatic, exchange, and dispersion interactions) with classical polarization (induced dipoles, Drude oscillators, etc). Within LICHEM, the electron density is polarized by both the QM region and the static MM multipole moments [26]. A separate MM calculation is then performed to determine the response of the MM polarizable sites to the QM and MM electrostatic fields. In this manner, only one QM and two MM calculations ( Vmm′ and Vmm,pol) are required to calculate the relaxed QM/MM total energy and the underlaying QM and MM packages do not need to be modified. Additionally, this approach can easily be implemented for a variety of classical polarizable models.</p><!><p>In this section, some common approaches to calculate long-range electrostatic interactions will be reviewed. While the higher-order moments are neglected in most of this discussion, these long-range methods can readily be extended to include the full multipole–multipole and polarizable interactions [28, 29].</p><!><p>The Coulomb potential produces relatively long-range interactions due to the r−1 distance dependence. The potential is given by (6)Vij,cl(rij)=qiqjrij, where Vij,cl is Coulomb's law in atomic units, qi is the monopole on atom i, and rij is the distance between atoms i and j. For periodic systems, the Coulomb potential is often only calculated up to a distance of 9–20 Å to reduce the computational cost [8, 31]. However, due to the length scale of the potential, the residual energy is often non-negligible at the cutoff radius (Rc) [6, 42, 44]. To mitigate some of the truncation artifacts, the potential can be shifted by a constant; such that it is equal to zero at the cutoff [44]. The energy shifted potential, Vij,es, is given by (7)Vij,es(rij)=qiqjrij−qiqjRc.The force, Fij, due to a general potential, Vij, is given by (8)Fij(rij)=−∇Vij(rij), where ∇ is the gradient operator, and thus, (9)Fij,cl(rij)=qiqjrij2, and (10)Fij,es(rij)=qiqjrij2.Since Fij,cl = Fij,es, the use of energy-shifted potential still results in finite forces at the cutoff radius [44]. Thus, energy-shifted potentials produce artifacts in calculations which require forces, e.g., MD simulations or geometry optimizations.</p><p>The truncated potential can be further improved by shifting the force in a manner analogous to the shifted potential. The shifted force, Fij,fs, is given by [44] (11)Fij,fs(rij)=qiqjrij2−qiqjRc2, and the force-shifted potential can be derived by integrating the shifted force [44], (12)Vij,fs(rij)=−∫Fij,fs(rij)drij+ΔVij,fs, where (13)ΔVij,fs=−2qiqjRc.The final force-shifted potential is given by (14)Vij,fs(rij)=qiqjrij+qiqjrijRc2−2qiqjRc where both the energy and force are now zero at the cutoff.</p><p>Truncation and shifting schemes make the calculation of the Coulomb potential tractable for large systems, but they often produce artifacts due to the artificial nature of the finite cutoff. In principle, the shifting procedure could be continued for higher-order derivatives. However, as will be shown in Sects. 3.3 and 4.1, this approach can drastically alter the electrostatic potential. The energy and force shifting derivations given thus far are by no means the only shifting approaches. Polynomial or exponential damping functions are often applied to the Coulomb potential to shift the energy and accelerate the convergence of the LRE interactions. In general, any shifting approach can be rewritten as (15)Vij,sm=S(rij)Vij,cl,where Vij,sm is the smoothed Coulomb potential and S(rij) is a damping function.</p><!><p>The calculation of the total electrostatic potential for an infinite periodic system involves a conditionally convergent sum [48]. However, this issue can be overcome by separating this sum into two absolutely convergent sums, which is the basis of the Ewald method [10]. Ewald summation methods divide the electrostatic potential into a short-range damped Coulomb potential and a long-range Fourier transformed Coulomb potential. The total Coulomb potential, Vtot,cl, may be calculated by [10] (16)Vtot,cl=Vsr+V∼lr−Vself+εb, where Vsr is the sum of the damped short-range Coulomb potential, V∼lr is the Fourier transformed sum of the long-range Coulomb potential, Vself is the self-interaction potential of the multipoles, and εb is a correction term due to the boundary conditions that arises when the unit cell has a finite charge and/or dipole moment [10, 32].</p><p>The efficiency of the Ewald method can be improved by employing numerical Fourier transforms. One possibility is by using B-spline functions to interpolate the periodic potential onto a mesh of grid points [9]. This approach, termed (smooth) particle mesh Ewald summation, can be used in conjunction with a fast Fourier transform (FFT) algorithm [13]. The FFT can significantly speed up the calculation of the long-range Coulomb potential and reduces the theoretical computational cost from O(N2) to O(NlogN).</p><!><p>Our approach for treating LRE interactions is to scale the electrostatic potential based on the distance between the atoms [14]. This approach smooths the electrostatic potential and forces, which are zero at the cutoff radius. Conceptually, this is equivalent to the procedure used in shifted or Wolf approaches, except that the LREC smoothing function has been designed to produce energies and forces in good agreement with PME.</p><p>The smoothed Coulomb potential used by the LREC method, Vij,lrec, is given by (17)Vij,lrec(rij)=f(rij′,2)qiqjrij, where (18)f(rij′,s)=[1−(2rij′3−3rij′2+1)s], and (19)rij′=(1−rijRc).Here f is the LREC cubic smoothing function and s is an adjustable integer exponent (see Fig. 1). Unlike most LRE approaches where the damped forces are derived analytically from the damped energy, the LREC force calculations simply use a different exponent in the smoothing function, (20)Fij,lrec(rij)=−f(rij′,3)qiqjrij2.The use of different exponents is related to the length scale of the potential and the gradient. Since the forces act over a shorter range than the potential, it is beneficial to move the inflection point of the smoothing function closer to rijRc=1 by increasing the exponent (see Figs. 1, 2). Increasing the exponent reduces the damping at short ranges, while increasing the damping in the regions where the forces are small. In general, the LREC exponent and cutoff can be adjusted to control the convergence, cost, and accuracy of the method, similar to adjusting the short-range cutoff and damping parameter in the Ewald, PME, and Wolf methods. As shown by Fang et al. [14], the LREC approach produces energies (s = 2) and forces (s = 3) in good agreement with PME when the cutoff radius is larger than 20 Å.</p><!><p>To easily implement any shifting or smoothing function in QM/MM simulations, the MM monopoles can be scaled to a value consistent with the smoothed Coulomb potential, (21)qj∗(rj)=S(rj)qj,where qj∗ is the scaled monopole in the MM region, S is a generalized damping function, and rj is the distance between the MM atom and the QM center of mass. Scaling monopoles based on their distance from the QM region is a simple approach which avoids the need to modify the QM integrals or correct the energies. Since the QM external field contains the damped multipole moments, forces on the QM atoms can be calculated using the QM gradients without further corrections. Additionally, there is no need to modify the forces when rj′ is slowly varying, which is the case for iterative QM/MM geometry optimizations [52]. However, one needs to be cautious when performing QM/MM–MD simulations, when there is a small number of QM atoms, or when the cutoff radius is small; since rj′ will no longer be slowly varying. Figure 3 reports scale factors calculated for a selection of smoothing and shifting methods. It is clear that the LREC approach faithfully represents the electrostatic potential over a longer length scale than other shifting approaches. At rj=Rc2, the LREC approach retains 75 % of the original Coulomb energy, while the force shift potential has been reduced to 25 %.</p><p>Higher-order multipole moments can be scaled in the same manner as the monopole moments: (22)μj∗(rj)=Sdpl(rj)μj, and (23)Θj∗(rj)=Sqpl(rj)Θj, where Sdpl and Sqpl are the dipole and quadrupole damping functions, respectively. This approach can easily be implemented to produce smoothed multipole expansions. In the LREC approach, Eqs. 22 and 23 reduce to (24)μj∗(rj)=f(rj′,s)μj, and (25)Θj∗(rj)=f(rj′,s)Θj, where f and r′ have the same form as in Eqs. 18 and 19. The work Fang et al. [14] and our exploratory calculations (see supporting information) have shown that setting s = 2 is sufficient for QM/MM calculations with only monopoles, while setting s = 3 significantly improves LREC performance for multipolar QM/MM calculations. Within LICHEM, the scaled monopole, dipole, and quadrupole moments can be further approximated as a set of 6 point-charges in an octahedral arrangement [12, 19]. As was shown previously [26], the point-charge approximation is a simple and accurate method for including multipole moments in QM software that can only employ point-charges for the external field.</p><!><p>Since QM/MM simulations separate the QM–QM, MM–MM, and QM–MM electrostatic interactions, the long-range electrostatics can be treated with different approximations in each part of Eq. 3. For example, the QM–MM interactions may be treated with the LREC approach while the MM–MM interactions can be calculated with PME. The resulting QM(LREC)/MM(PME) calculations take advantage of the simplicity of the LREC smoothing function in quantum calculations and the speed/accuracy of PME [8, 9, 21] for classical models. It is, perhaps, not intuitively clear why two different LRE approaches can be combined in QM/MM simulations. Separability is key to the validity of smoothly mixing two LRE methods. Two distinct types of separability appear in the QM/MM calculations.</p><!><p>Separation of the optimizations: LICHEM employs the iterative QM/MM optimization algorithm [26, 52] shown in Fig. 4. Since the QM and MM degrees of freedom are optimized independently, only a single LRE method is used to calculate the forces during each optimization procedure. Thus, two different LRE approaches are only used simultaneously while calculating the total energy.</p><p>Separability of the energies: The QM/MM total electrostatic energy, εtot, can be written as (26)εtot=εqm′+εmm′+εmm,pol′, where (27)QM(LREC)→εqm′, and (28)MM(PME)→εmm′,εmm,pol.Here ε represents the electrostatic component of the energies calculated with the QM/MM Hamiltonian (Eq. 3). The QM–MM interaction energies in Eq. 27 are calculated using Gaussian integrals in the QM software, while the electrostatic interactions in Eq. 28 are calculated using the MM force field. Since the QM–MM, MM–MM, and polarization portions of Eq. 26 are already calculated with different algorithms (i.e., Gaussian integrals and force fields), using two sufficiently accurate LRE methods does not significantly affect the total energy. Test calculations with MM(FS)/MM(PME), where the QM region was treated with a force-shifted MM potential, confirmed that there are negligible errors (not reported) compared to full PME calculations.</p><!><p>QM and polarization calculations are inherently many-body and must include enough of the multipole moments for the calculations to converge. On the other hand, the charges on the QM atoms can simply be neglected in the MM–MM calculations. Since the MM potential is pair-wise additive, calculating the electrostatic interactions without the presence of QM charges generally does not affect the MM–MM interactions.</p><!><p>The QM calculations were performed using the 6–31++G(d,p) basis set, and each QM/MM system was solvated using a cubic box of liquid water (32,000 molecules, box length: 98.646 Å). We have tested the LREC method with a variety of cutoff radii up to the maximum cutoff (Rc = 49.323 Å) allowed by the minimum image convention. The QM–MM and MM–MM vdW interactions, on the other hand, were calculated using the default settings of the TINKER software package [39] (polynomial smoothing with Rc = 9 Å). Additionally, we have tested LICHEM's QM(LREC)/MM(PME) implementation using both the non-polarizable TIP3P [24] and the polarizable AMOEBA [43] potentials.</p><!><p>To examine the convergence of multipolar LREC calculations, QM/MM calculations were performed on the box of liquid water using the NWChem [46]–TINKER interface in LICHEM. A collection of 5 water molecules in the center of the box were taken as the QM subsystem, and all remaining water molecules were designated as the MM region. The QM–MM interaction energy, Eint, is given by (29)Eint=Etot−Eqm−Emm,where Etot is the QM/MM total energy, Eqm is the energy of the QM subsystem in the gas phase, and Emm is the MM energy of the system with the QM atoms removed.</p><p>Figure 5 reports the convergence of the QM–MM interaction energy as the electrostatic cutoff is increased. The interaction energies from the long-range corrected QM/MM calculations are in good agreement with the results taken from the AMOEBA force field and, as was shown previously [14], the QM/MM energies converge with a cutoff between 20 and 25 Å. The computational cost of LRE methods is almost entirely determined by the cost of the QM calculations and the LRE cutoff radius. As can be seen in Fig. 6, the computational cost of QM/MM simulations can increase rapidly as the cutoff radius is extended. Since multipolar QM/MM simulations are both more expensive and slower to converge than monopolar QM/MM simulations, it is beneficial to use a higher exponent to accelerate the convergence (see supporting information).</p><p>Additionally, PBE0(LREC)/TIP3P(LREC) calculations, using the in-house modified versions of Gaussian and TINKER, confirmed that the QM/MM total energy was within 0.01 % of the PBE0(LREC)/TIP3P(PME) energy. The excellent agreement between the QM(LREC)/MM(LREC), QM(LREC)/MM(PME), and MM(PME) calculations confirms that the LREC method can be smoothly combined with other LRE methods.</p><!><p>Generally, the determination of reaction barriers is one of the main objectives in QM/MM simulations. To test the QM(LREC)/MM(PME) method on chemical reactions, we have chosen to use the aspartic acid dimer double proton transfer reaction from our previous work [14]. The acid dimer reactant, transition state, and product structures were solvated using the water box from Sect. 5.1 and re-optimized at the ωB97xD/AMOEBA (Rc = 25 Å level of theory.</p><p>The reaction barriers were calculated using LICHEM's Gaussian [17]–TINKER interface for single-point energy calculations with Rc = {2, 5, 10,…, 45, 50} Å. The proton transfer barriers reported in Fig. 7 show large fluctuations with small cutoffs before stabilizing with Rc ≥ 20 Å, around 6.06–6.18 kcal/mol. While the barrier continues to increase beyond Rc = 20 Å, the total change induced by further increasing the cutoff from 20 to 50 Å is ∼0.1 kcal/mol. Furthermore, extrapolating to Rc = ∞ yields a barrier of 6.30 kcal/mol. Thus, the approximate error at Rc = 20 Å is only 0.25 kcal/mol, which is well below the so-called chemical accuracy (1 kcal/mol).</p><!><p>We have demonstrated that the LREC approach can be extended to multipolar/polarizable models in a straightforward manner. The interaction energies and reaction barriers from the LREC approach converge as the cutoff radius is increased and, in agreement with our previous results, Rc = 20–25 Å is sufficient to obtain converged QM/MM energies. Using multipolar LREC for QM/MM simulations is an easy-to-implement approach for calculating long-range electrostatic interactions, which does not require modifications to the underlying QM software. Furthermore, the QM(LREC) calculations can readily be combined with efficient and accurate long-range electrostatics methods, such as PME, for the MM calculations.</p>

PubMed Author Manuscript

Nonlinear photocarrier dynamics and the role of shallow traps in mixed-halide mixed-cation hybrid perovskites †

We examine the role of surface passivation on carrier trapping and nonlinear recombination dynamics in hybrid metal-halide perovskites by means of excitation correlation photoluminescence (ECPL) spectroscopy. We find that carrier trapping occurs on subnanosecond timescales in both control (unpassivated) and passivated samples, which is consistent within a shallowtrap model. However, the impact of passivation has a direct effect on both shallow and deep traps. Our results reveal that the effect of passivation of deep traps is responsible for the increase of the carrier lifetimes, while the passivation of shallow traps reduces the excitation density required for shallow-trap saturation. Our work demonstrates how ECPL provides details about the passivation of shallow traps beyond those available via conventional time-resolved photoluminescence techniques.

nonlinear_photocarrier_dynamics_and_the_role_of_shallow_traps_in_mixed-halide_mixed-cation_hybrid_pe

Introduction<!>Experimental section<!>Time-Resolved Photoluminescence (TRPL).<!>Excitation Correlation Photoluminescence Spectroscopy (ECPL).<!>Linear Spectroscopy<!>Please do not adjust margins<!>Temperature-Dependent ECPL experiments<!>Photophysical model and simulation description<!>Please do not adjust margins<!>Qualitative description of the observed features.<!>Conclusions

<p>Over the last decade, hybrid organic-inorganic perovskites (HOIPs) have emerged as promising low-defect-density semiconductors with useful applications in optoelectronics, such as solar cells [1][2][3][4] , light-emitting diodes 5,6 , and lasers 7,8 . In particular, solar cell devices have reached remarkable power conversion efficiencies (PCEs); many NREL-validated PCEs for single-junction cells are above 20% with the highest PCE over 25%. 9 However, non-radiative charge-carrier recombination prevents HOIPs from approaching their theoretical efficiency. It has been shown that single-crystalline lead-halide materials possess a high tolerance to intrinsic defects 10,11 . The interfaces and grain boundaries of perovskite films have been identified as the source of the dominant trap-assisted nonradiative recombination [12][13][14] . Different post-processing passivation strategies have proven effective in reducing the nonradiative pathways, namely passivation by oxygen exposure 15 , peroxide treatment 16 , and surface modifiers 12,17 . Characterization of the effect of the passivation molecule on the surface is usually limited to time-resolved photoluminescence measurements and PCE comparisons between devices, which affords limited insight into the photophysical nature of carrier trapping processes in HOIPs.</p><p>Several physical models have been proposed previously to describe trap-assisted nonradiative recombination for leadhalide perovskites following the Shockley-Read-Hall (SRH) formalism. 18,19 Stranks et al. describe the interplay between excitons and subgap states in CH3NH3PbI3−xClx; their model proposed the presence of deep traps by means of background doping caused by the photoinduced filling of traps which exhibit nonradiative recombination with very slow dynamics 20 . Dobrolovsky et al. concluded a very different scenario for the same material in which there are distinct types of traps, some of which are "shallow" and emissive 21 . Several authors have noted the importance of considering shallow traps as part of the photophysical description of perovskite nanocrystals, suggesting that trapping and detrapping mechanisms are responsible for the long-lived photoluminescence traces 22,23 . Although, it is not believed that shallow traps are responsible for the long-lived photoluminescent trace in thin films 24 . Recently, shallow traps have been reported to be responsible for degradation of FAPbI 3 25 .</p><p>Because the processing conditions and identities of precursors have been shown to affect the defect physics 21,26 , not only by changing the trap density, but also the intrinsic characteristics of the trap (e.g., trap depth, trapping rate). Here we implement a time-resolved spectroscopic method that characterises carrier trapping dynamics beyond common techniques such as time-This journal is © The Royal Society of Chemistry 20xx</p><p>Please do not adjust margins Please do not adjust margins resolved photoluminescence spectroscopy. Specifically, we employ excitation correlation photoluminescence (ECPL) spectroscopy, which was originally used to describe population mixing dynamics of GaAs p-type semiconductors 27,28 as well as picosecond carrier lifetimes in other inorganic semiconductors [29][30][31][32][33] , and exciton dynamics in carbon nanotubes 34,35 . More recently, it was used to characterise the defect physics in CH3NH3PbBr3 thin films and CsPbBr3 nanocrystals 26 . The technique consists of exciting the sample with two ultrafast pulses modulated at different frequencies Ω 1 , Ω 2 . The measured photoluminescence intensity then includes contributions from recombination between populations generated by the first pulse, the second pulse, and a mixed population term; this mechanism is represented schematically in Fig. 1c. The contribution of the mixed population to the measured photoluminescence intensity is described by:</p><p>By using lock-in heterodyne detection, the signal is demodulated at the sum of the pulse modulation frequencies |Ω 2 + Ω 1 | to isolate this contribution. ECPL is well-suited for measuring sub-nanosecond short-carrier lifetimes without employing any complex electronics for ultrafast detection. Specifically, it is useful in characterizing nonradiative pathways, as was shown by Johnson et al. 29 . Recombination dynamics in which the radiative pathways are predominant results in an ECPL signal of zero. Meanwhile, the signal for ECPL for a trapping-dominated process can be as high as 30% of the total signal 29 .</p><p>In this work, we study the effect of passivation with APTMS, 3-aminopropyl)trimethoxysilane, on the defect scenario of control perovskite thin films with composition FA0.83Cs0.17Pb(I0.85Br0.15)3, employing both ECPL spectroscopy and time-resolved photoluminescence measurements. We describe the defect scenario as a mixture of deep traps and shallow traps. We observe fast ECPL photocarrier dynamics at room temperature for both the passivated and the control sample, suggesting a fast carrier trapping rate resulting from shallow traps. The effect of passivation is observed as a decrease in trap density of both deep traps and shallow traps. This is also supported by temperature-dependent ECPL experiments, which show a shift to a lower excitation threshold in trap saturation. By numerically solving the differential rate equations for both shallow and deep trap cases, we give a qualitative description of the features observed at low temperature, here assigned to Auger recombination. Film preparation and passivation was performed according to the procedure described by Jariwala et al. 36 All precursors were purchased from Sigma, unless stated otherwise. Perovskite precursor solutions (1 M) were formed by adding stoichiometric amounts of precursor salts formamidinium iodide (greatcellsolar), CsI, PbI2 (Sigma), and PbBr2 in DMF:DMSO (1300:640). 60 µL of formamide was added to make a 2 mL 1 M solution (with 3v% formamide). The substrates were plasma cleaned before spin coating and the perovskite solution was filtered before deposition (inside the glovebox). ~25 µL of the filtered perovskite solution was deposited on top of the substrate. The substrates were spun at 4000 rpm for 60 seconds. When ~35 s was remaining, ~50 µL of chlorobenzene (CB) was dropped from above. The films were then annealed on the hotplate at 100 °C for 30 s and at 150 °C for 10 min. Surface passivation with APTMS was done in a vacuum oven (Precision Vacuum Oven Model 19) for 5-10 min at room temperature under vacuum with the gauge pressure reading -30" Hg relative to atmospheric pressure. For further details refer to ref. 36 The films of untreated (herein called Cs17Br15 control, or just control) and APTMS-treated FA0.83Cs0.17Pb(I0.85Br0.15)3 were prepared at the University of Washington and shipped to Georgia Tech for ECPL characterization.</p><!><p>A detailed characterisation of the bandgap, X-ray diffraction, and composition can be found in ref 36 . It is worth mentioning that the characterisation of the surface composition, by XPS and band edge, was cross-validated and showed to be reproducible after the shipping at multiple institutions. 36 UV-Vis Absorption and steady-state Photoluminescence.</p><p>UV-Vis absorption spectra were measured using an Agilent 8453 UV-Vis spectroscopy system, the data corrected for scatter at the longest wavelengths. The PL spectrum was also acquired in the same configuration as the ECPL instrument setup (described Please do not adjust margins Please do not adjust margins below) by collecting the emission with an optical fiber coupled to a spectrometer (USB2000+, Ocean Optics).</p><!><p>TRPL was measured using a Picoharp 300 TCSPC system equipped with a 470 nm pulsed diode laser (PicoQuant PDL-800 LDH-P-C-470B, 300 ps pulse width). The laser was pulsed at repetition rates from 250KHz to 1MHz. The PL emission was filtered using a 580 nm long-pass filter before being directed to the detector.</p><!><p>A schematic representation of the ECPL setup is shown in Fig. S1. In the ECPL instrument setup, 1030 nm, ~220 fs pulses are generated in an ultrafast laser system at a 100 kHz repetition rate (PHAROS Model PH1-20-0200-02-10, Light Conversion). A portion of the laser beam is sent into a commercial optical parametric amplifier (ORPHEUS, Light Conversion), where a 2.64-eV pump excitation pulse is selected. Typical excitation densities used in the experiment range from 3x10 16 to 6x10 18 cm -3 . The excitation intensity is varied by moving a motorized filter wheel (FW212CNEB, Thorlabs). The pulse trains are then split 50/50 by a beam splitter cube, where one of the beams is directed to a motorized linear stage (LTS300, Thorlabs), allowing for control of the delay between the two pulses. Each pulse is modulated with a chopper at the frequencies of 373 and 199 Hz, respectively, and the pulses are then focused onto the perovskite sample with a 150-mm focal length lens. The emitted PL is filtered with a 550-nm long-pass filter to get rid of the pump, then it is focused into a photoreceiver. The photoreceiver is connected to a lock-in amplifier (HF2LI, Zurich Instruments) where the total integrated PL and the nonlinear component ΔPL are obtained simultaneously by demodulating both at the fundamental and the sum frequency. The perovskite films are measured inside a closed-cycle cryostat (Montana Instruments). It is worth mentioning that some degradation was observed at the highest fluences. For all measurements we acquired data from the lowest to highest excitation fluence, all in the same spot.</p><!><p>The PL and absorption spectra for the control and passivated samples are shown in Fig. 2a. From UV-vis absorption spectroscopy, we determine the band gap to be 1.63 eV from Tauc plots. From PL spectra, the control samples exhibit a maximum emission energy peaked at 1.63 eV on average, while the passivated samples exhibit a maximum emission energy 1.60 eV on average. Recently, different lineshapes of PL spectra have been explained in the context of PL reabsorption 37,38 , specifically an apparent redshift caused by the reabsorption and the PL propagation along the film. Under this interpretation the PL in the passivated sample has higher photo recycling, expected from a sample low trap density.</p><p>The APTMS vapour passivation procedure has been demonstrated to reduce non-radiative recombination, increase average PL lifetimes by over an order of magnitude, and increase the external quantum yield by a factor of 60 36 . Due to the effect of APTMS on surface defects, this system is ideal to describe the effect of passivation on the nature of traps and nonradiative recombination dynamics. Typical time-resolved PL measurements for control and passivated samples, collected at a low fluence of < 50 nJ/cm 2 per pulse, are shown in Fig. 2b. The PL average lifetime 〈𝜏〉 increases after APTMS deposition in agreement with previous reports, 36 and is attributed to the surface passivation of traps. Along with an increase in the PL average lifetime, we also observe an increase in the 𝛽 parameter when fitting TRPL to a stretched exponential function, defined by equations 2 and 3. The increase in the 𝛽 parameter suggests a more homogenous distribution of kinetics after APTMS passivation 39 , which we interpret as a distribution of trap environments, consistent with reports by Jariwala et. al 36 . Please do not adjust margins</p><!><p>Room temperature ECPL Fig. 3 shows fluence-dependent ECPL acquired at room temperature for both FA0.83Cs0.17Pb(I0.85Br0.15)3 control and APTMS passivated films. We start by discussing the ECPL dynamics on the control sample Fig. 3a. At the lowest excitation density, the ECPL signal is negative and flat; for the next excitation density it becomes more negative and then monotonically increases with excitation intensity. At a higher excitation density (nD ~ 1x10 18 cm -3 ) the ECPL signal develops dynamics indicative of a picosecond process due to fast charge trapping 31 . At the highest excitation densities, the system reaches a saturation point. This behaviour can be explained by the presence of a high trap density that is only filled at a high excitation density (nD ~ 5x10 18 cm -3 ). For the APTMS passivated sample, the ECPL photocarrier dynamics in Fig. 3c share some similarities with the control sample, but the saturation threshold changes due a lower trap density. The ECPL also starts flat and negative, but the next fluence already has dynamical features and the overall signal saturates at lower fluences. This is expected, given that the passivation procedure would decrease the density of traps causing saturation to occur at lower excitation densities. In this case, we observe a small decrease of the signal at high excitation densities. We attribute this to Auger recombination starting to play a role in the dynamics as we approach the saturation threshold but not enough to observe a significant decrease.</p><p>It is useful to look at the ECPL (t=0) signal, which corresponds to the situation when the two excitation pulses arrive at the same time, to visualise the effect that traps have on the ECPL saturation. Fig. 3b shows ECPL (t=0) signal for the control sample, which increases until it saturates at nD ~ 5x10 18 cm -3 . In Fig. 3d, the ECPL (t=0) signal from the passivated sample initially is negative and then grows until saturation at around nD ~ 1x10 18 cm -3 , and finally decreases due to Auger recombination. These ECPL photocarrier dynamics on control and passivated samples are consistent with a shallow trap model reported before on CH3NH3PbI3 perovskites 26 since they present a picosecond charge trapping process. If the fast trapping were caused by deep defects, then we would expect a sub-nanosecond short recombination lifetime which is inconsistent with the photoluminescence trace observed experimentally, Fig. S11 and S12 show that the population evolution during first nanosecond for deep traps is negligible. Note that this does not imply that deep traps are not present , simply that they are not responsible for the observed positive dynamics. We note that the excitation density values at which the ECPL saturation is observed are slightly higher than previous trap density values reported for similar perovskites 20,39 . However, it is known that the saturation threshold for thermally activated traps is dependent on the trapping/detrapping equilibrium, as will be discussed below.</p><!><p>There are several reports in the literature on the effect of temperature on recombination photophysics in lead-halide perovskites. Stranks et al. 20 reported a change in the trap density with temperature and speculated the origin was due to thermally activated traps for MAPbI 3 films. Kandada et al. 26 also reported thermally activated doping in MAPbBr3. It is also Please do not adjust margins</p><p>Please do not adjust margins worth noting that the detrapping rate is temperature dependent. Therefore, understanding the effect of temperature on ECPL signal would shed light on the physics behind recombination processes. For this reason, we measured ECPL at various temperatures (Fig. 4, S4 and S5).</p><p>We start by discussing the temperature dependence of ECPL photocarrier dynamics for the control sample. At 30 K (Fig. 4a), the ECPL starts negative and flat similar to the 295 K signal. However, dynamics soon develop as excitation density increases, with the ECPL signal approaching zero and appearing to saturate at the highest excitation density. Furthermore, close to time zero with excitation density nD > 1x10 18 cm -3 , ultrafast dynamics appear. In the context of the shallow trap model proposed above, a decrease of the ECPL signal is expected since the trapped electron will lack sufficient thermal energy to hop from the trap state to the conduction band, which will decrease the fluence threshold at which saturation occurs and allow Auger recombination to occur. ECPL photocarrier dynamics at 200 K are complex for the control sample (Fig. 4b) pointing to multiple overlapping regimes. The signal again starts negative and flat at the lowest fluence, then increases with dynamics similar to the 295 K. However, at an excitation density nD ~ 2x10 18 cm -3 , the dynamics turn negative due to Auger recombination and, again, ultrafast dynamics appear. Further increasing the temperature to 240 K (Fig. 4c) leads to ECPL dynamics that are similar to those at 295 K.</p><p>We now discuss the ECPL signal for the passivated sample. At 30 K (Fig. 4d), the ECPL signal is close to zero even at the lowest fluences. As the excitation density increases, the ECPL signal decreases, though to a minuscule extent. In this case, if Auger recombination becomes predominant as is expected for very low temperatures, it is also expected that the ECPL intensity will continue to decrease to negative values as the excitation density continues to increase. It is important to note that the picosecond dynamics appear at a much lower threshold.</p><p>Increasing the temperature to 200 K (Fig. 4e) shows predominantly negative signal at the lowest fluences, with ultrafast dynamics becoming more prominent along with an overall increase in ECPL signal as excitation density increases. Furthermore, an increase to 240 K (Fig. 4c) leads to dynamics like the ones at 295 K for lower excitation density. That is, the signal starts negative and flat at the lowest fluence, then becomes positive until nD ~ 2x10 18 cm -3 , after which the dynamics turn negative and the ultrafast dynamics peak appears. This is somewhat similar to the control at 200 K and further suggests that the appearance of this feature is caused by saturation. The temperature threshold then differs from control to passivated since the density of traps is different. The negative curvature dynamics in the ECPL signal have been reported for hexagonal GaN 32 and was attributed to population saturation.</p><p>A narrow peak appears in most of the ECPL photocarrier dynamics as the excitation density is increased, as we can see from Fig. 4. We believe this is due to an effect termed amplification of spontaneous emission (ASE) 40 . It is optical gain due to light trapped in a waveguide, and air-perovskite-glass interfaces may act as such a waveguide. Interestingly, the presence of this feature is temperature-dependent and at the same time correlates with the very fast dynamics and negative dynamics observed in the ECPL. The correlation also holds when we measure the passivated sample: ASE remains present at higher temperatures (Fig. S2 and S3), and the excitation densities needed to show ASE are much lower. Please do not adjust margins Please do not adjust margins</p><p>When we look at the ECPL (t=0) for the temperature dependence (Fig. S6 and S8), we find that there are two dominant regimes. For temperatures 240 K and above, the ECPL signal increases monotonically with fluence and reaches a saturation point, the saturation point is lower for the passivated sample hinting a lower shallow trap energy, as discussed above. But besides, the saturation threshold seems to shift to lower excitation as the temperature is decreased, this further supports the lower energy of the shallow trap. At T = 200 K and below, there seems to be local maximum in the control sample at lower excitation intensities, this local maximum is closely related to the appearance of the ASE. This effect does not seem to be present for the passivated sample although it may very well be that we are just not covering the region as it will shift to lower excitation densities. It is important to note that these features are present, albeit at a lesser degree when looking at a time away from zero to avoid the ASE peak, (ECPL (t= 250 ps) in Fig. S7 and S9).</p><!><p>The ECPL results obtained are interpreted following the model for shallow traps described by the rate equations shown below: Where 𝛾 𝑡 is the trapping rate, 𝐵 is the bimolecular recombination rate, 𝑁 𝑡 is the density of traps, 𝛾 𝑑 is the detrapping rate, 𝛾 𝑛𝑟 is the nonradiative recombination rate, and 𝛾 𝐴𝑢𝑔𝑒𝑟 is the Auger recombination rate. 𝑃 corresponds to the photocarriers generated by the pulses. The pulses are considered to interact instantaneously both for the interpretation of the results and for the simulation results. The detrapping rate (𝛾 𝑑 ) is of particular interest as it relates to the depth of traps. It is described as 18,26 :</p><p>The temperature dependence study of recombination parameters has been reported for CH3NH3PbI3 41 by using the model 𝑃𝐿 ̇= −𝑘 1 𝑛 − 𝑘 2 𝑛 2 − 𝑘 3 𝑛 3 . This model does not capture the diversity of processes occurring nor their reversibility. The bimolecular recombination rate was shown to change with temperature by two orders of magnitude between 160 K and 300 K. The Auger rate remained constant, with a significant increase at temperatures <150 K related to a change in structure. At temperatures lower than 160 K, a change in the photoluminescence lineshape was observed, indicative of a phase transition from tetragonal to orthorhombic structure. We do not observe such signatures (Fig. S2 and S3) in concordance with phase transition quenching observed for (Cs,MA,FA)Pb(Br,I)3 42 . Due to this observation, we assume that all our measurements were taken for the same phase, therefore we do not consider structural effects as the source of changes in the ECPL experiments. In the temperature-dependent simulation, we assume the various parameters are constant and that the parameters in equation (7) are the only ones affecting the detrapping rate.</p><p>The binding energy for excitons has been estimated for leadhalide perovskites to be 6-60 meV [43][44][45][46][47][48] . Even at room temperature, the role of excitons cannot be overlooked, as they are constantly forming and dissociating 20 . However, we do not include exciton formation in the rate equations to avoid the over-parametrization of the model. This is justified as, in the context of ECPL experiments, the pairing and further recombination of generated photocarriers do not result in a change of the integrated photoluminescence measured by the photodetector, therefore the ECPL response will not be affected by this assumption and recombination from excitons can be safely ignored.</p><p>Our simulation of the ECPL signal solves the initial value problem created by the rate equations following a pulse interaction. This problem is then solved repeatedly over a range of delays and processed to extract simulated ECPL data. As the difference between positive and negative delay times is strictly due to the time-ordering of the pulses, the simulation only simulates positive delay times, and this result is then reflected across the zero-delay line to show negative delay times. This process is repeated for a defined set of parameters based on similar materials. The resulting images are either qualitatively compared to experimental data or studied to determine the qualitative effects of each parameter on ECPL data, depending on the exact simulation run and parameter sets chosen.</p><p>The simulation relies on several assumptions. Firstly, pulses are treated as though they interact instantaneously. This is a valid assumption to make as long as we recognise that the simulation will be inaccurate at timescales on the order of the pulse width.</p><!><p>We also assume total absorption of the pulse to calculate the excitation density. This is inaccurate as we can observe remnants of the excitation pulse reflecting off the sample and transmitting through the sample. However, we are mainly interested in the qualitative behaviour at various excitation densities, so this is sufficient for our purposes.</p><!><p>From the time-resolved photoluminescence data Fig. 2b, the increase in photoluminescence lifetime suggests a decrease in the density of deeper traps between the control and passivated sample. We confirm the presence of shallow traps in both control and passivated films by ECPL at room temperature. The presence of distinct trap scenarios is expected, as thin film microstructure has proven to impact lifetimes of local carriers in the film domains 21,39,49,50 .</p><p>Next, we will justify the assignment of the various features, starting with a fast-trapping approximation. The shift of the saturation threshold with temperature is well-explained in the context of thermal detrapping. After the first pulse excitation, an ultrafast trapping and detrapping process will start. We assume that bimolecular and nonradiative recombination are significantly slower and, for sake of simplicity, a low enough excitation density such that the trapped electrons do not change (𝑁 𝑡 − 𝑛 𝑡 ) < 𝑁 𝑡 , these assumptions simplify the differential equation to an analytically solvable case. Then we can describe the ratio between free electron and trapped electron population evolution as: 𝑛 𝑛 𝑡 = 𝛾 𝑑 + 𝑁 𝑡 𝛾 𝑟 𝑒 −(𝛾 𝑑 +𝑁 𝑡 𝛾 𝑟 )𝑡 (𝛾 𝑟 − 𝛾 𝑟 𝑒 −(𝛾 𝑑 +𝑁 𝑡 𝛾 𝑟 )𝑡 )𝑁 𝑡 .</p><p>(𝟖).</p><p>For the sake of simplicity, we assume 𝑡 → ∞:</p><p>Recall that the temperature dependence is carried entirely by the 𝛾 𝑑 term. It is clear from equation (7) that the value of 𝛾 𝑑 decreases with temperature, resulting in a larger population of traps being filled. Even though the expression was obtained assuming low saturation, the trends discussed do not change.</p><p>The second pulse in the ECPL experiment acts as a probe of the evolution of the carrier populations caused by the first pulse. If the temperature is low, then most of the traps will be filled and the second pulse will result in high electron and hole densities, making Auger recombination and bimolecular recombination the dominant recombination pathways.</p><p>We can visualize the effect of the temperature in the saturation threshold by numerically solving the set of differential equations (2-4) and observing the carrier evolution during the first nanosecond, which is the range we are measuring in the ECPL experiment. This is equivalent to seeing the effect of just one pulse. From Fig S10, it can be observed that for the case of a shallow traps with a depth of 10 meV as we decrease the temperature the trapped electron density increases faster, because of slower detrapping and this causes a decrease in the saturation threshold.</p><p>For a deeper shallow trap, 100 meV, the effect of temperature is almost negligible, Fig S10 . This observation together with the difference of saturation threshold with temperature between control and passivated indicate that, from the initial distribution of shallow traps, the passivation procedure decreased the average trap depth.</p><p>We use the ECPL results obtained from the simulation to further support the assignment of the negative dynamics observed to the Auger recombination rate. We recognize that the ECPL response obtained by the simulation is very high compared to the experiment. We attribute this difference to the inability of a rate equation model to include the statistical description of the defect scenario. Still, we consider the trends regarding fluence and temperature to provide significant information that aids and supports our interpretation of the ECPL response.</p><p>The ultrafast ECPL photocarrier dynamics near zero could also be in part due to hot carrier relaxation. Ultrafast hot carrier cooling has been reported in lead halide perovskites, however it is important to note that relaxation happens in ultrafast time scales less than 1 ps 51 . The ECPL(t=0) then might contain some contribution due to hot carriers, however the ultrafast transient has a FWHM of ~20 ps. For this reason, another ultrafast process, like ASE must be involved. ASE would result in a positive feature in the ECPL response. This can be qualitatively rationalized by analysing the equation (10):</p><p>𝑃𝐿 𝑐𝑟𝑜𝑠𝑠 (𝐼 𝑝𝑢𝑚𝑝 , 𝜏) = 𝑃𝐿(2𝐼 𝑝𝑢𝑚𝑝 , 𝜏) − 2𝑃𝐿(𝐼 𝑝𝑢𝑚𝑝 , 𝜏) (𝟏𝟎).</p><p>In this expression, the cross-correlation photoluminescence intensity is written as the difference between the PL generated by two pulses and twice the PL generated by one pulse. We expect the ASE to be higher for higher fluences therefore it will be more significant in the first term. ASE is a radiative pathway, then it becomes clear that cross-correlation PL is going to have a positive value. This journal is © The Royal Society of Chemistry 20xx</p><p>Please do not adjust margins Please do not adjust margins</p><!><p>Herein, for Cs17FA83Pb(I85Br15) passivated with APTMS, we distinguish between deep trap passivation and shallow trap passivation using excitation correlation photoluminescence spectroscopy. The sub-nanosecond recombination observed is evidence of the presence of shallow traps in both control and passivated samples. We interpret the change in the saturation threshold between the control and the passivated sample as a change of shallow trap density caused by the passivation procedure.</p><p>The correlation between the appearance of ASE in the temperature-dependent photoluminescence and shallow trap saturation suggests shallow traps play an important role by imposing a population inversion threshold. Targeted passivation of the shallow traps in high PLQY perovskite materials might help decrease the fluence threshold for lasing applications. This would be of particular interest in perovskite nanocavities with shallow traps, which present PLQY up to 90% 52 . We also report the observation of Auger recombination dynamics observed when saturation is overcome. We speculate that the direct observation of Auger recombination will help study structure-property relations for lasing applications where Auger recombination dominates and limits the gain.</p><p>With a view to future work, since shallow trap passivation also plays a role in the electron mobility and their transport mechanism has been a recent subject of study 53 , this work also motivates the implementation of excitation correlation spectroscopy with photocurrent detection, which would expand the exploration of the passivation effect of electron/hole transport layers on the sub-nanosecond recombination in the functional devices.</p>

ChemRxiv

Biochemical mechanisms of pathogen restriction by intestinal bacteria

The intestine is a highly complex ecosystem where many bacterial species interact with each other and host cells to influence animal physiology and susceptibility to pathogens. Genomic methods have provided a broad framework for understanding how alterations in microbial communities are associated with host physiology and infection, but the biochemical mechanisms of specific intestinal bacterial species are only emerging. In this review, we focus on recent studies that have characterized the biochemical mechanisms by which intestinal bacteria interact with other bacteria and host pathways to restrict pathogen infection. Understanding the biochemical mechanisms of intestinal microbiota function should provide new opportunities for therapeutic development towards a variety of infectious diseases.

biochemical_mechanisms_of_pathogen_restriction_by_intestinal_bacteria

Emerging mechanisms and applications of microbiota-mediated pathogen resistance<!>Inter-bacterial interactions<!>Metabolic competition<!>Direct bacterial warfare<!>Interference with pathogen virulence<!>Host modulation<!>Activation of epithelial barrier function and other innate immune mechanisms<!>Tuning adaptive immunity<!>Concluding Remarks<!>

<p>The intestinal microbiota (see Glossary) is central to host metabolism and immunity [1, 2]. As a result, the microbiome as a whole broadly impacts host physiology and response to intestinal and systemic diseases. The composition of the intestinal microbiome is dynamic and is influenced by environmental factors including host diet and exposure to drugs, infection, and probiotics, as well as by genetic factors. Advances in our understanding of specific bacterial genes and molecules have revealed a diversity of inter-bacterial interactions and immuno-modulatory roles for intestinal bacteria that influence pathogen fitness and host response to infection [3]. This review will focus on recent studies of specific intestinal bacterial species, their metabolites and potential biochemical mechanisms (see Table 1 for examples presented in this review). Beyond infection, the intestinal microbiome widely influences host physiology, with specific bacterial factors contributing to diseases including obesity, cardiovascular disease, and diverse neurological disorders. These topics have been recently reviewed elsewhere [4, 5] and will not be discussed in this review.</p><p>A molecular understanding of intestinal microbiota-host interactions is of great medical and ecological importance. Insight into microbiota function can help us design targeted therapeutics against a variety of diseases and advance personalized medicine. While the clinical potential of probiotic and other microbiota-based therapies against infectious disease is now being explored, available treatments are still coarse. For example, fecal transplantation has recently emerged as an investigational new drug for recurrent Clostridium difficile infection[6, 7]. Transitioning to more targeted therapeutics, however, requires a better understanding of the specific mechanisms of individual bacterial species. Beyond the clinical potential, characterizing the intestinal microbiome gives us a generalizable framework for evaluating the ecology of other complex host-microbe interactions, which are ubiquitous on earth.</p><!><p>The intestinal microbiome is shaped by metabolic competition and communication. Inter-bacterial interactions within this community can affect the fitness of key species as well as overall community structure. In this section, we will discuss recently described mechanisms by which intestinal bacteria directly restrict enteric pathogens (Figure 1A). Inhibition of pathogen growth can occur through resource competition as well as through direct bactericidal mechanisms. Alternatively, pathogen infection can be attenuated by modulating specific virulence mechanisms.</p><!><p>Although microbial community structure at the taxonomic/species level varies greatly across healthy individuals, functional metabolic capacity at the metagenomic level is largely stable [8]. This highlights how metabolic competition and interdependence builds microbial community structure. Within this framework, host genetics and behavior can alter the intestinal environment and shift microbial composition.</p><p>Niche competition not only defines microbial community structure, but also serves as a barrier for both enteric pathogens and pathobionts. Metabolic competition over resources including diverse carbon sources, trace metals, and vitamins such as B12 [9], shapes the microbiota and likely contributes to colonization resistance. Due to metabolic similarity, competition is often greater between related species than unrelated species. Enterobacteriaceae is a large family of Gram-negative bacteria that includes commensal species as well as several intestinal pathogens including: Salmonella, Shigella, Yersinia, Citrobacter, and Enteropathogenic Escherichia coli. Indeed, metabolic competition between members of the Enterobacteriaceae has been described in vivo. For example, competition over monosaccharide use can restrict colonization by Citrobacter rodentium in wild-type mice [10]. Colonization of mice with commensal E. coli provides greater competition-mediated growth restriction of C. rodentium than colonization with Bacteroides species, which are able to utilize a more diverse profile of mono- and polysaccharides. However, in an experimentally skewed intestinal environment with monosaccharides as the sole carbon source, colonization of mice with Bacteroides thetaiotaomicron can also provide colonization resistance towards C. rodentium.</p><p>Competition over iron is another well-characterized example of inter-species competition in the intestine. Because iron is an essential and limiting nutrient, many intestinal bacteria produce iron-chelating siderophores to increase iron uptake. Host-secreted antimicrobial proteins, such as lipocalin-2, restrict bacterial growth by binding and inactivating diverse bacterial siderophores. Salmonella enterica can evade lipocalin-2-mediated growth inhibition by producing modified siderophores that cannot be bound by lipocalin-2 [11, 12], improving pathogen fitness [13]. In this context, probiotic E. coli Nissle 1917, which expresses four iron uptake systems that are resistant to lipocalin-2, competitively reduces Salmonella colonization and subsequent inflammation [14].</p><p>Many pathogens generate a distinct niche as a strategy to avoid competition with intestinal bacteria and gain access to nutrients. In a simplistic sense, adhesion to the surface of intestinal epithelial cells, or invasion into host cells, could be viewed as a means to evade competing intestinal bacteria. Intestinal pathogens may also have evolved unique nutrient utilization pathways compared to their commensal counterparts. For example, pathogenic strains of E. coli can consume a unique set of intestinal sugars as compared to commensal strains of E. coli [15]. Alternatively, pathogen-induced inflammation can disrupt the microbiome, releasing unique metabolites to give pathogens a growth advantage. For example, Salmonella-induced intestinal inflammation leads to the generation of tetrathionate by host epithelial cells. Tetrathionate respiration by Salmonella confers a growth advantage over other intestinal bacteria that rely on fermentation [16]. In the inflamed intestine, ethanolamine utilization by both Salmonella [17] and pathogenic E. coli [18] can also enhance their colonization [19]. Inflammation-induced dysbiosis can also lead to the outgrowth of specific non-pathogenic species. For example, nitrate released in the inflamed intestine is used in anaerobic respiration by commensal E. coli strains [20]. Whether proliferation of nitrate-utilizing E. coli species contributes to other pathogenesis mechanisms, however, is unclear.</p><p>Antibiotic-induced dysbiosis can also alter nutrient availability to allow the outgrowth of pathogens and pathobionts. For example, antibiotic treatment can alter microbial composition to promote the proliferation of vancomycin-resistant Enterococci [21] as well as C. difficile [22, 23]. In the case of C. difficile infection, antibiotic treatment results in increases in both succinate [23] and sialic acid [22], which are utilized by C. difficile to enhance intestinal colonization.</p><!><p>Beyond metabolic competition, intestinal bacteria can limit the growth of other bacterial species through the production of antibacterial compounds and inhibitory metabolites as well as through contact-dependent killing. Recent studies have highlighted how microbiota-produced antimicrobials can inhibit intestinal colonization by specific enteric pathogens and pathobionts. For example, bacteriocins are diverse secreted antibacterial peptides that target and lyse related bacterial species [24, 25]. In the intestine, colonization by bacteriocin-producing Enterococcus faecalis reduces the numbers of indigenous E. faecalis, as well as infection by vancomycin-resistant E. faecalis [26]. Microcins produced under iron starvation conditions by probiotic E. coli Nissle 1917 reduce intestinal colonization by other commensal E. coli, as well as Salmonella [25]. The large scale bioinformatic and biochemical mining of the microbiome for other antibacterial molecules has highlighted the potential of microbiota-based antimicrobial compounds as therapeutic leads for new antibiotics [27]. Beyond specific antimicrobial compounds, other bacterial metabolites may also directly inhibit pathogen growth. For example, Clostridium scindens inhibits C. difficile growth and pathogenesis in mice [28]. This in vivo protection is correlated with the capacity to generate secondary bile acids, which can inhibit C. difficile growth in vitro.</p><p>In addition to releasing antibacterial compounds into the extracellular space, many Gram-negative bacterial species can directly antagonize neighboring cells by injection of effector proteins using the type-VI secretion system (T6SS). T6SSs are encoded by commensals and pathogens alike, and T6SS-mediated competition is likely prevalent in the intestine. For example, T6SSs are widely encoded amongst the Bacteroidales [29], an order that includes abundant intestinal commensals, and T6SS-encoding Bacteroides can target sensitive Bacteroides species, presumably to limit competition [30–32]. Intestinal pathogens such as S. Typhimurum [33] and Vibrio cholerae also encode T6SSs, which may target intestinal bacteria and enhance pathogen colonization. Ex vivo, V. cholerae uses its T6SS to deliver antimicrobial effectors to intestinal bacteria, such as E. coli and Salmonella [34], and self-immunity to this secretion system is required for robust intestinal colonization, suggesting that V. cholera employs its T6SS in vivo [35].</p><!><p>In addition to restricting pathogen colonization and proliferation, intestinal bacteria can also directly affect pathogen virulence mechanisms. Virulence gene expression involves integrating a diversity of environmental cues, some of which can be modified by intestinal bacteria. For example, in response to a local increase in oxygen close to the intestinal epithelial surface, Shigella flexneri increases the secretion of effector proteins involved in host cell invasion [36]. This raises the possibility that intestinal bacteria adhered to the mucosa could potentially inhibit Shigella invasion by consuming local oxygen.</p><p>Intestinal bacteria may also alter the pool of available host-derived nutrients to affect pathogen virulence. For example, fucosidases expressed by Bacteroides thetaiotaomicron can release fucose from mucins. Fucose is sensed by Enterohemorrhagic E. coli through the FusK/R two-component system, and high fucose leads to a decrease in the expression of virulence genes required for the formation of attaching and effacing lesions [37]. Although B. thetaiotaomicron has not been shown to directly affect EHEC virulence, ΔfusK EHEC exhibit a growth defect in vivo, suggesting that the regulation of virulence through fucose sensing may be relevant to intestinal infection.</p><p>Bacterial fermentation in the intestine results in the production of diverse short-chain fatty acids. Short-chain fatty acids not only modulate host immunity but have also been shown to directly modulate virulence gene expression of various pathogens ex vivo. For example, short-chain fatty acids can differentially regulate the expression of Salmonella virulence genes involved in host cell invasion in a chain-length dependent manner. Specifically, acetate enhances the expression of genes involved in invasion [38], while propionate [39] and butyrate [40] are inhibitory. Other molecules prevalent in the intestinal environment, such as bile acids, have also been shown to modulate virulence gene expression of enteric pathogens in vitro [41–43], but the direct biochemical mechanisms of these metabolites on pathogen virulence is not clear.</p><p>Beyond metabolite regulation of virulence mechanisms, quorum sensing is a widespread form of bacterial "communication" that regulates diverse processes, including pathogen virulence. In quorum sensing, bacteria use secreted small molecules to regulate gene expression in a density-dependent manner. Bacteria can integrate intra-species and inter-species quorum sensing signals to optimize gene expression based on local environment. In the intestine, the extent of quorum sensing-mediated communication is unclear; but intestinal delivery of the inter-species quorum sensing molecule autoinducer-2 (AI-2) alters microbiota composition after antibiotic treatment, suggesting that intestinal bacteria can broadly respond to quorum sensing pathways in vivo [44]. Pathogen sensing of another inter-species quorum sensing molecule class, N-acyl homoserine lactones (AHLs), has been proposed as a possible mechanism by which intestinal bacteria could modulate virulence. Both EHEC and Salmonella encode the transcription factor SdiA, which is required for AHL recognition, but lack the machinery to synthesize AHLs [45, 46]. In the case of EHEC, AHLs repress the expression of virulence genes involved in the formation of AE lesions, but enhance the expression of acid-resistance genes [47]. Consistent with this, SdiA is required for EHEC colonization of the cow rumen, but has no effect on intestinal colonization. Curiously, AHLs are largely absent in the intestine [48, 49], making the relevance of AHL-mediated quorum sensing in the intestine unclear. Host epithelial cells may also engage bacterial quorum sensing pathways to affect intestinal bacteria, as epithelial cells have been shown to produce a partially functional AI-2 mimic in vitro [50].</p><!><p>In addition to inter-bacterial interactions that influence pathogen fitness and overall microbiota composition, intestinal bacteria can also directly affect host susceptibility to infection by regulating epithelial barrier function as well as innate and adaptive immunity (Figure 1B). Modulation of epithelial barrier function can limit pathogen access to relevant host cell types, while activation of other innate and adaptive immune mechanisms can result in direct killing of microbes or pathogen tolerance locally or systemically.</p><p>Recognition of microbial-associated molecular patterns (MAMPs) by both intestinal epithelial cells and lymphoid cells is required for proper immune system development and response to microbial pathogens [51]. MAMPs are conserved microbial factors, such as bacterial cell-surface components, that elicit a host immune response upon recognition by pattern recognition receptors (PRRs). Toll-like receptors, Nod-like receptors, and C-type lectin receptors all respond to MAMPs derived from commensal bacteria and pathogens alike, leading to a variety of downstream innate and adaptive immune responses. The basal immune responses elicited by these receptors in response to intestinal bacteria help establish and maintain immune system homeostasis [52]. Bacterial surface components such as lipopolysaccharide, lipoteichoic acid, peptidoglycan, capsular polysaccharides, flagellin, and other surface proteins can all be recognized by PRRs to modulate intestinal immunity [53].</p><p>In mammals, innate immune recognition of MAMPs is a balance between mechanisms of tolerance versus immune activation. Although the intestine is full of indigenous commensal bacteria, hyper-activation of PRRs may be prevented by controlling the localization [54] and abundance of receptors, as well as by specific negative regulators of downstream pathways. Host immune responses may also be tailored to pathogenic versus commensal bacteria by recognition of species-specific molecules. Downstream responses to MAMP recognition include the modulation of epithelial barrier function, as well as the priming and regulation of other innate and adaptive immune mechanisms. In this section, we highlight recent mechanistic studies that reveal how probiotic or commensal bacteria enhance epithelial barrier function and host immunity to pathogens.</p><!><p>One important downstream response of MAMP recognition is the modulation of epithelial barrier function [55]. Secreted mucins, antimicrobial proteins, and secretory IgA (sIgA), all serve to exclude bacteria from a sterile zone maintained close to the surface of intestinal epithelial cells [56]. Bacterial transgression of the epithelial barrier can lead to pro-inflammatory responses that influence the pathogenesis of diverse chronic diseases including HIV/AIDS [57] and inflammatory bowel disease (IBD) [58]. Not only does the epithelial barrier repel bacteria, but it also reduces permeability to other metabolites that could aggravate systemic extra-intestinal inflammatory responses.</p><p>Recognition of intestinal bacterial MAMPs by both intestinal epithelial and innate immune cell lineages helps to maintain epithelial barrier function. For example, administration of lipopolysaccharide and peptidoglycan can restore the mucous barrier in germ-free mice, and protect against chemically-induced colitis [59]. Intestinal bacteria can enhance barrier function by inducing the production of RegIIIγ, IgA, and other antimicrobial proteins [60–65]. Although several antimicrobial proteins can be produced in an epithelial cell-intrinsic manner, some, such as RegIIIγ, require concomitant innate lymphoid cell-derived IL-22 signaling for full expression [66].</p><p>The bactericidal mechanisms of intestinally-localized antimicrobial proteins are diverse and include both non-enzymatic membrane disruption and enzymatic cell wall hydrolysis[67]. For example, alpha-defensins and cathelicidins are broad-spectrum antimicrobials that form membrane pores to kill bacteria, while RegIII C-type lectins bind peptidoglycan prior to pore formation and therefore preferentially target Gram-positive bacteria [68]. In contrast, lysozyme and secretory phospholipase A2 enzymatically hydrolyze peptidoglycan and phospholipids respectively to disrupt bacterial membranes.</p><p>Microbiota-mediated modulation of key metabolites and MAMPs may contribute to epithelial barrier function and pathogen restriction. Indeed, microbiota-associated metabolites such as taurine, histamine, and spermine have been shown to modulate NLRP6 inflammasome signaling, epithelial IL-18 secretion and the production of antimicrobial peptides. The activation of inflammasome signaling by microbial metabolites is important for preventing gut dybiosis and intestinal inflammation [69]. The intestinal bacteria Enterococcus faecium has been shown to remodel MAMPs to enhance epithelial barrier function and pathogen restriction. E. faecium secretes an NlpC/p60-type peptidoglycan hydrolase, SagA, that is sufficient to protect C. elegans and mice from pathogens such as S. Typhimurium and C. difficile [70, 71]. Based on studies in vitro and in C. elegans, SagA was found to remodel peptidoglycan and generate muramyl-peptide fragments to improve host immunity against pathogens. In mice, SagA enhances host epithelial barrier function through the induction of mucins and antimicrobial peptides, most strikingly RegIIIγ, and protection is dependent on pattern recognition genes MyD88 and Nod2. Interestingly, Lactobacillus rhamnosus, another probiotic bacteria, secretes two peptidoglycan hydrolases, p75 and p40, that enhance epithelial cell proliferation [72–74]. Whether these and other secreted peptidoglycan hydrolases can also induce host responses by remodeling MAMPs has yet to be determined.</p><p>The regulation of tight junctions and epithelial cell turnover by intestinal bacteria and bacterial metabolites also serves to enhance epithelial barrier integrity. For example, indole produced by intestinal bacteria enhances the expression of tight junction proteins, and reduces chemically-induced colitis in mice [75, 76]. Indole may signal via the aryl hydrocarbon receptor (AHR) in intestinal epithelial cells to maintain epithelial barrier integrity[77], similar to dietary indolyl-metabolite-induced AHR signaling in intraepithelial lymphocytes [78] and innate lymphoid cells [79]. Dietary acetate as well as acetate produced by different species of Bifidobacteria has been shown to provide strong protection against EHEC pathogenesis [80]. Acetate is produced as a result of fructose catabolism, and protective Bifidobacteria species encode unique carbohydrate transporters, including a fructose transporter. Heterologous expression of this transporter in non-protective Bifidobacteria species confers a protective phenotype against EHEC that is correlated with an increase in acetate production. In mice and in epithelial cell culture, acetate prevents epithelial cell apoptosis as well as Shiga toxin translocation to inhibit EHEC pathogenesis [81, 82].</p><p>In addition to directly killing extracellular pathogens or limiting their invasion into host cells, intestinal bacteria can also modulate cell-intrinsic microbial clearance mechanisms. Notably autophagy, the intracellular recycling pathway, has emerged as an important part of intestinal barrier function and innate immunity [83]. Epithelial cell-intrinsic autophagy of invasive enteric pathogens, such as Salmonella and E. faecalis, is activated by PRR signaling and serves to limit pathogen spread to distal tissues [84]. The autophagy pathway in Paneth cells also plays a role in intestinal immune homeostasis, through the regulation of inflammation. Hypomorphisms in ATG16L1, a major component of the autophagosome, are associated with Crohn's disease, and decreased expression in mice results in Paneth cell abnormalities including defective granule exocytosis [85] and an increased propensity for inflammation [86]. Whether specific intestinal bacterial species or factors can modulate autophagy to restrict pathogens remains an intriguing possibility.</p><!><p>While the epithelial barrier controls host immunity by regulating pathogen proliferation and/or access to host tissues, the microbiome can also directly regulate adaptive immune responses, which can facilitate antigen-specific clearance of pathogens and/or alter systemic inflammatory responses. In this section, we highlight studies where specific intestinal bacterial species and/or factors directly modulate adaptive immune mechanisms.</p><p>Antibodies provide important antigen-specific protection against potential pathogens. In the gut, the induction of secreted IgA antibodies by intestinal bacteria may facilitate the clearance of pathobionts and enteric pathogens. For example, recent studies using IgA-SEQ have shown that high levels of IgA binding identify pathobionts that can drive intestinal disease [87]. Interestingly, SCFAs produced by the gut microbiota have also been recently suggested to metabolically enhance B cell responses and facilitate homeostatic and pathogen-specific antibody responses [88]. These studies highlight the importance of the intestinal microbiome in stimulating B cell and antibody responses.</p><p>TH17 cells and TReg cells are important T cell subsets that mediate pro-inflammatory and anti-inflammatory responses in the intestine respectively. Specific intestinal bacteria can skew the ratio of these two cell types, which influences immune system maturation and can differentially affect host response to bacterial pathogens, as well as systemic inflammatory diseases. For example, adhesion of segmented filamentous bacteria (SFB) to intestinal epithelial cells induces the differentiation of TH17 cells, which leads to the production of pro-inflammatory cytokines, antimicrobial peptides, and secretion of IgA that cumulatively inhibit C. rodentium infection [89]. SFB appears to induce antigen-specific TH17 cell responses to mediate these effects [90, 91]. Adhesion may also be the molecular cue sensed by host epithelial cells, as other adherent microbes, including EHEC, C. rodentium, and Candida albicans, can induce similar Th17 cell responses [92]. Alternatively, ATP secreted by intestinal bacteria is recognized by purinergic receptors on intestinal epithelial cells and immune cells, which also drives the differentiation of TH17 cells [93]. Intestinal bacteria can also induce the proliferation of anti-inflammatory TReg cells. For example, Bacteroides fragilis capsular polysaccharide A (PSA) stimulation of dendritic cells promotes the development of anti-inflammatory TReg cells, which protect against both chemically- and bacterially-induced colitis [94–98]. Beyond the intestine, SFB and B. fragilis/PSA have opposing effects on inflammation and host outcomes in an experimental autoimmune encephalomyelitis mouse model [99]. Interestingly, although the epithelial barrier restricts most intestinal bacteria to the lumen, certain commensal species naturally reside in innate lymphoid tissues and colonize dendritic cells to elicit anti-inflammatory IL-10 cytokine production [100]. Bacteria-derived short-chain fatty acids have been shown to exert anti-inflammatory effects particularly by promoting the differentiation of TReg cells. G-protein coupled receptors on the surface of epithelial cells and other immune cells can bind to acetate, propionate, and/or butyrate [101] leading to the induction of anti-inflammatory cytokines and TReg cells [102, 103]. Inhibition of histone deacetylases by butyrate results in an increase in histone acetylation, which can also lead to an anti-inflammatory response in TReg cells [104, 105] as well as macrophages [106].</p><p>In addition to modulating B and T cell functions, intestinal bacteria may directly regulate antigen presentation to tune adaptive immune responses. Of course, peptide and protein antigens from intestinal bacteria can be processed and presented by classical MHC class I and II proteins to generate bacteria-specific T cell responses that are important for tolerance or immune activation [107, 108]. However, bacteria-specific metabolites can also directly alter antigen presentation pathways. For example, B. fragilis produces the glycosphingolipid α-galactosylceramide [109], which can be presented by CD1d proteins and inhibit the development of colonic invariant natural killer T (iNKT) cells to reduce intestinal inflammation in mice [110]. Mucosal-associated invariant T (MAIT) cells, another type of innate T cell, are activated by MHC-I related (MR1) presentation of riboflavin (vitamin B2) metabolites [111], which are uniquely bacterial in origin and thus potentially generate an adaptive immune response to enteric pathogens [112]. More recently, the discovery of microbiota-derived lysosomal protease inhibitors (pyrazinones and dihydropyrazinones) [113] suggests that intestinal bacteria can affect antigen presentation by interfering with proteases that degrade proteins into antigenic peptides. These studies highlight how intestinal bacteria may directly affect adaptive immune responses through modulating B cells, T cells, and antigen-presenting cells, which will likely expand to other lymphocytes (i.e. innate lymphoid cells and others).</p><!><p>Recent studies have elucidated diverse biochemical mechanisms employed by the intestinal microbiome to influence host susceptibility to infection. Within the intestine, commensal bacteria can directly antagonize enteric pathogens and pathobionts through metabolic competition, the production of antibacterial compounds, and interference with virulence gene expression. Immune sensing of the microbiome is required for proper immune development, and provides a mechanism for specific intestinal bacteria to shape immune function. MAMPs as well as specific antigens from intestinal bacteria are recognized by intestinal epithelial cells and gut-associated lymphoid tissues to regulate immune system maturation. Regulation of the downstream inflammatory response and epithelial barrier function are two general strategies by which intestinal bacteria can influence host physiology both within and beyond the intestine. With marked progress in the field, questions regarding the challenges, applications, and broader evolutionary and ecological insights of microbiome research have emerged (see Outstanding Questions box).</p><p>Reductionist models have helped winnow the bacterial factors involved in modulating microbiota composition and host physiology that might have been otherwise obscured. Nevertheless, the overall relevance and relative impact of many bacterial factors remains to be determined in the context of more complex natural systems. Integrating metagenomic and experimental findings in a systems biology approach may help clarify the dynamic roles of bacterial species that exhibit diverse functions in different model systems.</p><p>While germ-free and gnotobiotic mouse models continue to be foundational to microbiome research, the utility of non-mammalian model hosts for studying the intestinal microbiome has recently come into focus. Model organisms such as C. elegans, D. melanogaster, and D. rerio offer many experimental advantages and can provide a complementary understanding of the evolutionarily conserved or convergent factors that influence intestinal symbioses.</p><p>From a therapeutic perspective, modulation of the microbiome through clinical interventions such as probiotics, drugs, and diet can have far-reaching effects on host health. Beyond this, the microbiome is a rich reservoir of medically and societally important molecules. Recent genomic and chemical approaches to identify and functionally profile small molecules [114, 115] and regulatory enzymes [116] from the microbiome provide an opportunity to build upon our understanding of the vast molecular interactions occurring in the intestine.</p><!><p>any of the mechanisms (either direct or indirect) by which intestinal bacteria limit the colonization of harmful microorganisms</p><p>colloquially used to refer to any non-harmful symbiotic bacteria, although many species may be beneficial in certain contexts</p><p>any microbial imbalance</p><p>a condition in which all microorganisms are defined.</p><p>all the microorganisms that inhabit the intestine</p><p>the full complement of genetic material, metabolites and other molecules that occupy the intestinal ecosystem</p><p>competition that arises between two species that fill the same ecological role in a given environment</p><p>a potentially pathogenic organism that under normal circumstances exists as a commensal</p><p>a microorganism that causes disease</p><p>the severity with which a pathogen or pathogenic substance causes disease</p><p>Biochemical mechanisms by which intestinal bacterial species and factors affect host susceptibility to enteric pathogens.</p>

PubMed Author Manuscript

An Efficient Synthesis Strategy for Metal-Organic Frameworks: Dry-Gel Synthesis of MOF-74 Framework with High Yield and Improved Performance

Vapor-assisted dry-gel synthesis of the metal-organic framework-74 (MOF-74) structure, specifically Ni-MOF-74 produced from synthetic precursors using an organic-water hybrid solvent system, showed a very high yield (>90% with respect to 2,5-dihydroxyterepthalic acid) and enhanced performance. The Ni-MOF-74 obtained showed improved sorption characteristics towards CO 2 and the refrigerant fluorocarbon dichlorodifluoromethane. Unlike conventional synthesis, which takes 72 hours using the tetrahydrofuran-water system, this kinetic study showed that Ni-MOF-74 forms within 12 hours under dry-gel conditions with similar performance characteristics, and exhibits its best performance characteristics even after 24 hours of heating. In the dry-gel conversion method, the physical separation of the solvent and precursor mixture allows for recycling of the solvent. We demonstrated efficient solvent recycling (up to three times) that resulted in significant cost benefits. The scaled-up manufacturing cost of Ni-MOF-74 synthesized via our dry-gel method is 45% of conventional synthesis cost. Thus, for bulk production of the MOFs, the proposed vapor-assisted, dry-gel method is efficient, simple, and inexpensive when compared to the conventional synthesis method.Metal-organic frameworks (MOFs) have attracted much attention during the last two decades because of their enormous structural and chemical diversity in terms of high surface area, pore volumes, high thermal and chemical stabilities, and variety of pore dimensions/topologies 1-5 . These properties have made them superior to other traditional porous materials, and interest in their use for applications in gas/vapor sorption, molecular separation, and heterogeneous catalysis has increased significantly 1,[6][7][8][9][10][11][12][13] . In spite of their tremendous potential, near-term prospects for commercial applications remain quite limited because of the lack of technologies and processes for synthesizing these materials in quantities required for industrial applications and at a low cost. Facile synthesis of MOFs is very important for lowering the cost and also for achieving fundamental understanding and viable applications. The general synthesis methodology for MOFs is very similar to molecular sieve synthesis, and usually involves hydrothermal or solvo-thermal crystallization of dissolved reactants in suitable solvents using conventional heating methods and autogeneous pressures 14,15 . Recent reports of alternative methods include the use of microwave 16 , sonication 17 , mechano-chemical [18][19][20] , and electrochemical synthesis 21 methods, but all of these methods have issues with scalability, consistency, and cost that prevent their use for practical applications. BackgroundRecently, syntheses of porous materials from dry-gels have attracted considerable attention. The synthesis of porous material using the dry-gel conversion (DGC) method has potential advantages such as minimum waste disposal and reduced reactor size. In the literature, there are many reports about using steam-assisted DGC

an_efficient_synthesis_strategy_for_metal-organic_frameworks:_dry-gel_synthesis_of_mof-74_framework_

<!>Synthesis methodology and experimental details.<!>Results and Discussion<!>Material<!>Solvent recycling.

<p>methods for synthesis of zeolites [22][23][24] , but very few reports about synthesis of MOFs [25][26][27] . Shi et al. synthesized zeolite imidazole framework (ZIF) materials such as ZIF-8 and ZIF-67 by replacing dimethylformamide (DMF) with water as the solvent 28 . Ahmed et al. reported synthesis of iron-based MIL-100(Fe) without adding any hydrofluoric (HF) acid to the reaction mixture 29 . Later, Kim et al. reported the synthesis using water as solvent and HF acid as an additive of the same MIL (Materials Institut Lavoisier) family of MOFs, MIL-101(Cr), which exhibited increased surface area 30 . All of the reported MOFs and ZIFs synthesized thus far using the DGC method involve synthesis chemistry that has a reaction precursor, especially in the case of organic ligand, that is water soluble at reaction temperatures. In the case of Ni-MOF-74, the organic ligand 2,5-dihydroxyterepthalic acid (DHTA) is insoluble in water, which mandates the use of an organic solvent in the solvent mixture. Therefore, the important technical challenge is to demonstrate dry-gel synthesis using hybrid organic-water solvents that have variable boiling points and vapor pressures. To the best of our knowledge, synthesis of the MOF-74 family with the DGC method using hybrid organic-water solvent mixtures has not been reported so far. Here we report the synthesis of Ni-MOF-74 structures in an organic solvent-water mixture using a vapor-assisted DGC synthesis method for the first time. Among all of the MOF materials reported to date, the microporous MOF-74 (CPO-27 or M-DOBDC) structure shows promise for gas sorption applications because of its high density of open metal centers. MOF-74 is typical of MOFs that have a high density of accessible, open metal sites, which have shown remarkable host-guest interactions leading to high storage capacities for CO 2 , CH 4 , H 2 S, xenon, fluorocarbons, etc. [31][32][33] . In this paper, we report on work that focused on synthesizing Ni-MOF-74 using the DGC method. Our results show higher yields, faster kinetics of formation, and improved performance over materials produced using typical batch synthesis methods.</p><!><p>In general, M-MOF-74 (M = Ni, Co) was synthesized mainly under solvo-thermal conditions using two different approaches: 1) a tetrahydrofuran (THF)-water (1:1) mixture at 110 °C for 3 days and 2) a DMF-ethanol-water (1:1:1) mixture at 100 °C for 24-66 hours [34][35][36] . Initially, we attempted to synthesize Ni-MOF-74 using a THF-based procedure whereby the reagents 2,5-dihydroxyterepthalic acid (DHTA) and metal acetate (metal = nickel or cobalt) in a 1:2 molar ratio were ground together and then placed in a pouch made from fluorinated ethylene propylene (FEP) polymer mesh. We chose FEP polymer over the robust polytetrafluoroethylene (PTFE) material because its melt-processability using conventional heating facilitates making the pouches, and it exhibits robust characteristics similar to PTFE. The MOF precursor mixture loaded in the FEP pouch was carefully placed in a Teflon liner containing the solvent mixture (THF-water, 3 mL each), all of which was placed carefully at the bottom of the reactor as shown in Fig. 1.</p><p>The reactor was sealed and allowed to heat at 110 °C for 3 days 35 . After the heating period, the dry solid material was washed with fresh THF solvent to remove any unreacted starting material because an excess of metal salt was used. The product indicated successful formation of the MOF-74 honeycomb structure (now on MOF-74(DGC)) with an improved yield of 90%. Moreover, the liquid mixture at the bottom of the reactor after synthesis was clear and similar to the starting mixture (Electronic Supplementary Information [ESI], Figure S1), unlike the dark brown solvent mixture found after conventional synthesis. Successful formation of the Ni-MOF-74 honeycomb structure via the DGC method was verified by PXRD analysis, which revealed a match for the relative intensity and peak positions of the crystallographic data 37 . For comparison, Ni-MOF-74 also was synthesized in parallel using a conventional solvo-thermal (CS) method in which the precursors were dissolved in a THF/water mixture and heated held at 110 °C for 72 hours. The product of this synthesis (hereafter referred to as Ni-MOF-74(CS) produced an overall yield of ~65%. The PXRD analysis of the MOF-74(DGC) sample revealed an exact match to that of the MOF-74-CS sample of the honeycomb network (Fig. 1c). To check the purity of the reagents and to verify that no mechano-chemical reaction occurred before performing the DGC synthesis, freshly ground reagents (DHTA and the metal salt) also were subjected to PXRD measurements, which showed only starting materials, and no MOF-74 peaks were observed (ESI, Figure S4). Thermogravimetric analysis (TGA) performed on Ni-MOF-74 synthesized by the DGC method showed a weight loss of 18 to 23% as temperature was increased from 25 to 200 °C, which corresponds to the loss of solvent molecules and is comparable to weight loss experienced during the conventional synthesis method (ESI, Figure S6). Brunauer-Emmett-Teller (BET) surface area analysis was performed for both Ni-MOF-74(DGC) and Ni-MOF-74(CS) samples at 77 K using N 2 adsorption; the Ni-MOF-74(DGC) showed a high surface area of ~1350 m 2 /g, while the conventional heating sample showed a surface area of ~1029 m 2 /g, which is in line with the values reported in the literature (ESI, Figures S7-S10) 37 . It is important to note that the MOF-74 synthesized using the DGC method was used "as is" with simple THF washing, while conventional Ni-MOF-74(CS) was tested after multiple solvent activation steps using methanol soaking for 3 days and replacing the methanol every 24 hours.</p><!><p>Adsorption performance. Ni-MOF-74 is known to be a promising candidate for low-pressure CO 2 sorption applications; therefore, we tested the sorption characteristics of both the DGC and CS samples. Both MOF samples were subjected to the same activation procedure before testing their sorption capabilities. Our CO 2 isotherm for the Ni-MOF-74(CS) is very similar to published data within the experimental error. It is interesting to see that the Ni-MOF-74(DGC) showed enhanced CO 2 adsorption performance up to 9% (2.5 wt%) as shown in Fig. 2.</p><p>Though it is a smaller number, the enhanced CO 2 sorption capacity was observed throughout the pressure curve from 100 to 1000 mbar (Fig. 2a). To further elucidate the enhanced sorption capacities of the DGC method, we extended the adsorption towards R12 because Ni-MOF-74 showed extremely high sorption capacities at low pressures (50.8 wt% at 100 mbar) 38 . We attempted the same R12 adsorption studies for both DGC and CS samples at room temperature. Similar to the CO 2 sorption studies, the DGC method showed enhanced sorption capacities, which is close to an ~4.5% increase in R12 adsorption characteristics over Ni-MOF-74(CS) as shown in Fig. 2b. Enhanced sorption capacities are observed throughout the adsorption curve, and for clarity, enhanced sorption can be clearly seen from the enlarged portion of the curve (Fig. 2, inset). Further, to understand the optimized synthesis conditions and to further appreciate the capability of the DGC method over the solvo-thermal synthesis method, we performed a time-dependent kinetics study of Ni-MOF-74 synthesis using the DGC method. In this study, Ni-MOF-74(DGC) synthesis was carried out at variable time durations of 72, 48, 24, and 12 hours under identical thermal and reaction conditions. The PXRD results from all the samples showed the formation of the honeycomb network structure. More surprisingly, the Ni-MOF-74(DGC) synthesized after just 12 hours of heating time also showed successful formation of the honeycomb network (Figure S3, ESI). The BET surface area measurements on these samples reveals that increasing the heating time improves the overall surface area of the material, but not much improvement is observed between the 48-to 72-hour heating time (Table 1). This result was further confirmed by testing the CO 2 sorption performance on the samples where Ni-MOF-74(DGC)-48 h and Ni-MOF-74(DGC)-72h showed similar sorption characteristics (ESI, Figures S11-15). This result implies that the optimal time duration of the DGC synthesis was less than the conventional synthesis method. Similarly, we successfully extended the DGC method to the synthesis of cobalt MOF-74 using the THF-water solvent system. The synthesis and XRD results of Co-MOF-74(DGC) samples were identical to samples produced by solvo-thermal synthesis. The results are shown in Figure S5 (ESI).</p><p>To further demonstrate the vapor-assisted DGC method, we also attempted a DMF-based synthesis procedure developed by the Matzger group for synthesizing the Ni-MOF-74 structure 34 . The MOF precursors (DHTA and nickel nitrate in a 1:3.33 molar ratio) were prepared by grinding them together and then loading the ground mixture in an FEP pouch that was carefully placed in a Teflon liner containing a solvent mixture (DMF-ethanol-water, 2 mL each). Similarly, the reactor was sealed and allowed to heat at 100 °C for 24 hours according to the procedure. The resulting Ni-MOF-74 also had the MOF-74 honeycomb structure (ESI, Figure S2). Interestingly, although the DMF boiling point (~156 °C) was considerably higher than the reaction temperature, ethanol and water vapors can carry DMF to the MOF reagents where it acts as a catalyst for removing the proton from the organic acid for forming the MOF-74 structure 39 .</p><p>MOF-74 structures can be successfully formed in the vapor phase of the DGC method in an organic-water hybrid solvent. The high yield with improved performance might result from a solid-vapor reaction in which no contact exists between the solid reactants and liquid solvent during MOF formation. Thus, unwanted side reactions that generally occur in the liquid phase can be avoided, resulting high purity of MOF and enhanced yield. Regarding the crystallization process in steam-assisted ZIF-8 and also in zeolites, the solvent water heated to 110 °C under autogeneous pressure in the autoclave can generate a pressure close to 1.8 bar 28 . Fig. 3 shows the vapor pressure curves for both water and THF.</p><!><p>We used Antoine's equation to derive the vapor pressure curve.</p><p>In this equation, A, B, C are solvent dependent parameters, and values are obtained from NIST database 40,41 . The synthesis temparature is well above the boiling points of both water and THF. From the vapor pressure curve, the vapor pressures of THF and water, at a synthesis temparature of 110 °C, are 3.49 bar and 1.39 bar, respectively. At this temperature and these pressures, the phase of the water is at the liquid-vapor boundary while the THF is in the vapor phase. Hence, both of components of the THF-water solvent mixture are transported in the vapor phase to the precursor mixture (nickel salt+ DHTA ligand) and penetrate it. Our novel porous FEP polymer pouch bag allows uniform penetration of the solvent vapor mix over the entire salt-ligand mixture compared to the previously reported use of a ceramic cup in which the opening of the cup limits solution transport. The solvent vapor, which is transported to the salt-ligand mixture, condenses and is adsorbed on the precursor particles, creating a "solution-like" phase in which the same chemistry can occur, as in bulk solution, thereby catalyzing the MOF formation reaction 28,42 .</p><!><p>In the dry-gel method the solvent-precursor mixture is physically separated. Moreover, the solvent is transported to precursor mixture in the vapor phase, leaving behind the particulate impurities. Therefore, we thought that the solvent could be recycled without compromising the purity of the synthesized MOF. To verify the solvent reusability, we recycled the same solvent over three synthesis cycles. For cycle 1, we started with a pure solvent mixture (1:1 THF:water ratio, 10 mL) in an autoclave. We loaded the MOF-74 precursor mixture in the FEP pouch and carefully placed it in a Teflon liner containing the solvent mixture (THF-water, 5 mL each); the autoclave was heated at 110 °C for 24 hours. When the reaction was complete, we carefully removed the FEP pouch from the autoclave without disturbing the solvent. For cycle 2, another FEP pouch containing freshly prepared precursor mixture was placed in the autoclave with the recycled solvent. The synthesis process was continued at 110 °C for 24 hours. The same procedure is repeated for third cycle. Figure 4 shows the PXRD results of the Ni-MOF-74(DGC) synthesized in three cycles using recycled solvent where the XRD patterns indicate the formation of MOF-74 honeycomb network. The absence of impurity peaks and x-ray background is indicative of the high purity of the Ni-MOF-74 produced using recycled solvent. Because the solvent is an important cost contributor in bulk MOF synthesis, we performed a cost analysis for scaling up Ni-MOF-74 synthesis using the DGC method with recycled solvent (see ESI Section III). We found the general manufacturing cost of synthesizing Ni-MOF-74 using the conventional method to be $6,523/kg. When synthesis using the new DGC method was analyzed, the manufacturing cost was less than half (i.e., ~$2881/kg) with the solvent recycled for at least three times. The price might decrease even further by increasing the number of cycles (Tables S1 and S2). Thus, the DGC method showed a clear cost advantage over the conventional synthesis method and the method can be easily extended to other methods.</p><p>In conclusion, we successfully synthesized high-performance Ni-MOF-74 using a hybrid organic-water solvent mixture via a DGC method. In comparison to the Ni-MOF-74 produced using the CS method, the dry-gel Ni-MOF-74 showed a higher surface area and improved gas-capture performance for both CO 2 and R12. We also demonstrated that Ni-MOF-74 can be synthesized in 24 hours via the DGC method, and the resulting product exhibits acceptable purity and performance characteristics. Also, physical separation of the solvent mixture from the precursor mixture minimizes unwanted side reactions, thus allowing clean solvent mixture remaining after synthesis to be recycled multiple times. This implies an environmental and cost benefit in MOF synthesis. Overall, we demonstrated a technical advance in MOF synthesis in terms of the synthesis time scale, improved performance and raw material recycling that has important implications for low-cost manufacturing of MOF structures.</p>

Scientific Reports - Nature

On the possibility that bond strain is the mechanism of RING E3 activation in the E2catalyzed ubiquitination reaction

Ubiquitination is a type of post translational modification wherein the small protein ubiquitin (Ub) is covalently bound to a lysine on a target protein. Ubiquitination can signal for several regulatory pathways including protein degradation. Ubiquitination occurs by a series of reactions catalyzed by three types of enzymes: ubiquitin activating enzymes, E1; ubiquitin conjugating enzymes, E2; and ubiquitin ligases, E3. E2 enzymes directly catalyze the transfer of Ub to the target proteinthe RING E3 improves the efficiency. Prior to its transfer, Ub is covalently linked to the E2 via a thioester bond and the Ub~E2 conjugate forms a quaternary complex with the RING E3. It is hypothesized that the RING E3 improves the catalytic efficiency of ubiquitination by placing the E2~Ub conjugate in a "closed" position, which tensions and weakens the thioester bond. We interrogate this hypothesis by analyzing the strain on the thioester during molecular dynamics simulations of both open and closed E2~Ub/E3 complexes. Our data indicate that the thioester is strained when the E2~Ub conjugate is in the closed position. We also show that the amount of strain is consistent with the experimental rate enhancement caused by the RING E3. Finally, our simulations show that the closed configuration increases the populations of key hydrogen bonds in the E2~Ub active site. This is consistent with another hypothesis stating that the RING E3 enhances reaction rates by preorganizing the substrates.

on_the_possibility_that_bond_strain_is_the_mechanism_of_ring_e3_activation_in_the_e2catalyzed_ubiqui

Introduction<!>Structural preparation<!>Molecular dynamics<!>Density Functional Theory<!>DFT calculations show that the strain hypothesis is reasonable<!>MD simulations show that the closed Ubc13~Ub is strained<!>MD simulations show that the closed Ub~E2 has a more rigid binding pocket<!>IV. Conclusion<!>V. Author Contributions<!>VI. Acknowledgements<!>VII. Supporting information<!>VIII. References

<p>Many cellular processes in eukaryotes are initiated by attaching a small protein, ubiquitin (Ub), to a target protein. [1][2][3][4] This process may involve a single ubiquitin or a polyubiquitin chain. Ubiquitin is attached through a series of enzymatic reactions catalyzed by ubiquitin activating (E1), ubiquitin conjugating (E2), and ubiquitin ligase (E3) enzymes. At the E2 step, ubiquitin is covalently bound to the E2 through a thioester linkage. The C-terminal glycine on ubiquitin bonds to a cysteine sidechain on the E2. Next, a lysine sidechain on the target protein attacks the thioester carbonyl carbon to form a zwitterionic, tetrahedral intermediate. Finally, the intermediate collapses and an isopeptide bond forms between the ubiquitin and the target and the E2 is released. This process is shown in Fig. 1, where the target protein is also a Ub. This reaction is assisted by an E3 ligase.</p><p>The E3 serves to recruit the target protein and catalyzes the final ubiquitin transfer. 5 There are several families of E3 enzymes. HECT (Homologous to E6AP C-Terminus) and RBR (RING between RING) E3 ligases transfer the ubiquitin to a target in a two-step process -they first move the ubiquitin from the E2 to a cysteine on the E3, then they move the ubiquitin to its final target. 1,5 RING (Really Interesting New Gene/U-box) E3 ligases, the focus of this paper, catalyze the direct transfer of ubiquitin from the E2 to its target and do not change any chemical steps. 5,6 Figure 1 This is a simplified reaction scheme of an E2-catalyzed formation of a polyubiquitin chain. The substrate ubiquitin is marked Ub*. In the first step, the E2 and Ub are bound via a thioester linkage. The reaction then proceeds through a zwitterionic, tetrahedral intermediate, before it collapses into the products the regenerated E2 enzyme and an isopeptide-linked ubiquitin chain. RING E3s are the largest class of E3 enzymes, 7 and are thought to activate or prime the E2~Ub conjugate for ubiquitin transfer by orienting the substrate lysine in an ideal attack position and by immobilizing the thioester linkage between the E2 and the ubiquitin. 5,8 This preorganization occurs when the RING E3 places the E2~Ub complex in the closed position (see Fig. 2). 6,[9][10][11][12][13][14][15][16][17] It has been shown recently that ubiquitin transfer occurs exclusively when the E2~Ub is in the closed position. 16 Closed conformations have also been observed in RING E3 complexes for other ubiquitin-like proteins like SUMO and Nedd8, 18,19 which hints at a universal mechanism. In addition to preorganization, another intriguing hypothesis for the mechanism of E3 activation is that the closed position strains the thioester bond, "like tensioning a spring," 6 which makes it easier to break.</p><p>In this study, we test both the strain and preorganization hypotheses using molecular dynamics (MD) simulations. Our model E2 is Ubc13, which forms K63-linked polyubiquitin chains, i.e. ubiquitin is also the target protein. 20 K63-linked chains are not directly involved in protein degradation, 21,22 but instead are involved in other processes like inflammation response and DNA repair. 4,[23][24][25] Our simulations also include a ubiquitin E2 variant (UEV), which orients the substrate ubiquitin so that lysine 63 (K63) is positioned for attack. 20 Finally, our model E3 is the RNF4 RING domain. We monitor the thioester strain in the Ubc13~Ub/UEV complex in the open and closed positions. We also observe the effect of the E3 by comparing the thioester strain in the closed complex when the E3 is present to the strain when the E3 is absent. Finally, we measure the how the hydrogen bonding environment in the active site differs between the closed and open positions. In particular, we focus on the behavior of a highly conserved asparagine in the E2 that may stabilize a reaction intermediate, 3,20,[26][27][28][29] or may help preorganize the substrates. [30][31][32]</p><!><p>The initial geometries of the Ubc13~Ub complexes were generated from two crystal structures deposited in the RCSB data bank: pdb codes 2GMI and 5AIT 15,20 . There are two major differences between 2GMI and 5AIT: The Ubc13~Ub complex in 2GMI is in the open position and lacks a RING E3 ligase, whereas the Ubc13~Ub complex in 5AIT is in the closed position and has a RING E3 ligase, specifically the RNF4 RING domain. Figure 2 illustrates the difference between the open and closed conformations. Three initial systems were prepared: 2GMI, 5AIT-noE3, and 5AIT. Only one monomer (chains A-D) was retained in 5AIT, and the E3 (chain A) was also removed in 5AIT-noE3. The non-catalytic, ubiquitin E2 variant (UEV) (MMS2 in 2GMI and Ube2V2 in 5AIT) remained in all three structures. Substituted residue names were changed and their side-chain atoms were deleted. Specifically, in both crystal structures, C87 in Ubc13 was mutated (lysine in 5AIT and serine in 2GMI). Our simulations used the wild type cysteine. The LEaP program 33 was used to generate missing atoms. The final structures are shown in Figs S1-S3 in the SI.</p><p>In each structure, the Ub is covalently linked to Ubc13 via a thioester bond, which needs a parameter set. We developed the parameters using the protocol outlined in Refs 31and 32. Briefly, we generated partial atomic charges by removing the residues that form the thioester linkage between the Ubc13 and Ub from pdb 2GMI (C87 on Ubc13 and G76 on Ub) and capped them with acetyl (ACE) and N-methylamide (NME) groups. The charges were then calculated using the RESP protocol. 34 The cysteine residue within the thioester was labeled as the AMBER residue CYX, the thioester glycine was labeled as CGLY, and the thioester bond between them was described using parameters from the General AMBER Force Field (GAFF) 35 . In an improvement from our previous protocol, we added an improper torsion to the thioester to force planarity. This term was added and fit against the M06-2X/def2TZVP level of theory using gnuplot4.6. 36 The full parameter set used for the thioester bond and a detailed description of the procedure used to generate the new improper torsion can be found in the SI.</p><!><p>All MD was performed using the Amberff12SB force field 37 and the GPU-accelerated PMEMD module in the Amber14MD 33 and Amber20 38 packages. The structures for 2GMI and 5AIT were solvated in a rectilinear box (r= 12.0 Å) of TIP3P water molecules 39,40 and neutralized with K + and Clions (13 K + and 10 Clions for 2GMI and 10 K + , 10 Clfor 5AIT without the E3, and 12 K + and 20 Clfor 5AIT with the E3.). The systems were stabilized by an optimization, heated to 200 K, and subjected to density equilibration (100 ps, with a time step of 0.5 fs).The systems were then simulated at 300 K for 100 ns using a Langevin thermostat with a collision frequency of 2 ps -1 and a time step of 2 fs. The cutoff for non-bonded interactions structure was 8 Å and all covalent bonds to hydrogen atoms were held fixed using the SHAKE algorithm. A full microsecond of simulation data was collected for each of the 2GMI, 5AIT-noE3, and 5AIT-E3 structures. Each microsecond simulation was constructed from ten, independent, 100 ns NVT simulations, which can improve sampling. 41 Snapshots were saved every 2 ps. Trajectory analyses were conducted using CPPTRAJ 42 included in AmberTools14 33 and AmberTools21 38 . All Amber input files and backbone RMSD plots are included in the SI.</p><!><p>Density functional theory (DFT) was used to calculate the stiffness of the thioester when it is bent out-of-plane. For an initial geometry, we randomly selected a snapshot from one of our MD simulations, excised the thioester formed between the C-terminal glycine on Ub and the cysteine sidechain of Ubc13. We capped the dangling bonds using acetamide (ACE) and N-methyl amine (NME) residues (see Fig 3A). Next, we fully optimized the structure. Finally, ran a relaxed scan of the thioester out-of-plane bend defined between the C-S-O-C atoms and kept all other heavy atoms fixed. (Keeping the heavy atoms fixed ensures that the energy changes are solely due to the out-of-plane bending and not bond stretches, for example.) We scanned from -0.815° (fully relaxed and planar) to 34.18° using an increment of 0.5°. These calculations were performed at the M06-2X/def2TZVP level of theory using the Gaussian09 suite of programs 43 . All Gaussian input files are included in the SI.</p><!><p>Kinetics studies show that the presence of a RING E3 enhances the rate of ubiquitination by an E2 by about a factor of 10. 11,30 One hypothesis for the enhancement is that the RING E3 bends the thioester out of plane. 6 Because the carbonyl carbon on the thioester changes from planar to tetrahedral during the first step of the ubiquitination reaction, bending the thioester out of plane will increase the energy of the reactants relative to the transition state, thereby decreasing the ) as a function of out-of-plane bend (C). In (A) and (B), carbon atoms are cyan, nitrogen atoms are blue, oxygen atoms are red, the sulfur atom is yellow, and hydrogen atoms are white. As shown in (B), the improper dihedral is defined so that the last two atoms (O-C) are bonded. The intersection of the blue line and red curve in (C) shows that bending the thioester between 7 and 8 degrees out of plane will increase the rate by a factor of 10. This corresponds to lowering the reaction barrier by 1-1.5 kcal mol -1 (see Eq. 1). A plot of energy vs. out-of-plane is shown in the SI. activation energy. We tested the feasibility of this hypothesis using a combination of transition state theory (TST) and DFT calculations.</p><p>First, we used Eq. 1, which is derived from the TST rate expression, to estimate that at room temperature the reaction barrier should decrease by 1-1.5 kcal mol -1 to increase the rate tenfold.</p><p>Eq. ( 1)</p><p>‡ and is the change in energy barrier; ∆𝐺 2 ‡ is the energy barrier in the presence of the RING E3; ∆𝐺 1 ‡ is the energy barrier in the absence of the RING E3;</p><p>is the ratio of reaction rates for the catalyzed to uncatalyzed reaction; 𝑅 is the universal gas constant; and 𝑇 is temperature.) In this case, the change in the energy barrier does not originate in lowering the transition state energy. Instead, the barrier decreases because the strain on the thioester increases the reactant energy.</p><p>Next, we used DFT (M06-2X/def2TZVP) to estimate how far out of plane the thioester needs to bend to increase its energy by 1-1.5 kcal mol -1 . To simplify the calculation, we used a model thioester and assumed the change in electronic energy would dominate the change in free energy (ΔΔ𝐺 ‡ ), i.e., we ignored vibrational and rotational enthalpy and entropy contributions. We scanned the C-S-O-C out-of-plane bend for the model thioester (see methods) and our calculations show that bending the thioester 7-8° out-of-plane increases the reactant energy by 1-1.5 kcal mol -1 (see Fig. 3C). This is a modest amount of bend, which suggests that the strain hypothesis is reasonable. We analyzed our molecular dynamics simulations of Ubc13~Ub/E3 complexes to determine if the presence of the E3 strained the thioester linkage in a way that is consistent with experimental observations and our DFT calculations. We examined three models, an open structure (2GMI), a closed structure without the E3 (5AIT-noE3) and a closed structure with the E3 (5AIT). We monitored the out-of-plane bend of the carbonyl carbon in the thioester during our simulations. Each structure was simulated with ten independent, 100 ns trajectories. Our results are shown in Figure 4.</p><!><p>On average, there is little difference between 2GMI, 5AIT-noE3, and 5AIT. This result is displayed in Table 1 and in Figure 4A. Although the mean angle is slightly elevated in the closed systems (1.2°/1.3° in 5AIT-noE3/5AIT) versus the open system (0.044° in 2GMI), there is significant overlap in the distributions as is clear from Fig. 4A. Examined from a perspective, Table 1 also shows that the average out-of-plane energies (〈E OPB 〉) for each system are identical. However, since catalysis is a rare event, it is important to examine the extreme ends (the tails) of the distribution. Therefore, we calculated the percent of the trajectory when EOPB was greater than 1.0 and 1.5 kcal mol -1the amount of strain energy required to account for the RING E3-induced rate increase. The results are displayed in Table 1 and Figure 4B.</p><p>The final columns in Table 1 show that there is more out-of-plane bend energy in the closed systems (5AIT-noE3/5AIT) than the open system (2GMI). In fact, EOPB is greater than 1.0 kcal mol -1 for 9.7%/10.4% of the trajectory and greater than 1.5 kcal mol -1 for 4.3%/4.8% of the trajectory in the closed systems. For 2GMI, these numbers are significantly smaller, 7.0% and 2.9%. The box plots in Fig 4B also indicate that there is very little overlap between the open and closed systems and the overall distribution for the closed systems skews much higher than the open system. Interestingly, the presence of the E3 seems to not make much of a difference; there is similar amount of EOPB in 5AIT-noE3 and 5AIT.</p><p>Table 1 Average out-of-plane bend angle (〈Angle〉) and energy (〈E OPB 〉) of 2GMI, 5AIT-noE3, and 5AIT. The last columns show the percentage of the simulations where the strain energy is greater than 1.0 kcal mol -1 and 1.5 kcal mol -1 . All energies are in kcal mol -1 and all angles are in degrees. The error is calculated as the standard error over the entire 1 microsecond data set.</p><p>2</p><p>The error is calculated as the standard error over 10 simulations.</p><!><p>We also examined the hydrogen bonding properties of the binding pocket and the flexibility of the thioester. Specifically, we observed the hydrogen bonding partners of the asparagine (N79 in Ubc13) that is located within the highly conserved HPN motif found in E2 enzymes. 28 N79 has been hypothesized to stabilize the zwitterionic, tetrahedral intermediate of the ubiquitination reaction. 3,20,26 Alternately, this asparagine has been hypothesized to an active site loop and to pre-organize the thioester. [30][31][32] Therefore, we examined how the open/closed Ubc13~Ub state affects the hydrogen bonding in the active site, and we examined if the presence of the E3 had an effect. Our results are shown in Figs. 5 and 6.</p><p>Figure 5 shows the typical hydrogen bonding environment in the active site for both open (A) and closed (B) systems. Both positions show significant hydrogen bonding between the sidechain of N79 and the backbone of N116, and significant hydrogen bonding between the sidechain of H77 and the backbone of N79. We note that this last hydrogen bond contradicts the crystal structure 20 , but confirms previous NMR 28,44 and MD studies 31,32 . However, when Ubc13~Ub is open, there is negligible hydrogen bonding between N79 and the thioester (the carbonyl oxygen on G76). In the closed state, this hydrogen bond is populated. These observations are quantified in Fig 5A . Figure 5 Typical hydrogen bonding environments of 2GMI (A) and 5AIT-noE3/5AIT (B). All residues are from Ubc13, except G76, which is in Ub. In (A) and (B), carbon atoms are cyan, nitrogen atoms are blue, oxygen atoms are red, the sulfur atom is yellow, hydrogen atoms are white, and hydrogen bonds are red, dotted lines. Both Ubc13~Ub configurations show persistent hydrogen bonding between the sidechain of N79 and the backbone of N116, and persistent hydrogen bonding between the sidechain of H77 and the backbone of N79. In the open configuration (2GMI, A), there is no significant hydrogen bonding between N79 and the thioester, C87/G76. The hydrogen bonding increases dramatically when Ub~Ubc13 is in the closed position (5AIT-noE3/5AIT, B) closed.</p><p>The boxplot in Fig. 6A shows that there is consistent hydrogen bonding between N79, N116, and H77, but the N79/G76 hydrogen bond population depends on the state of the system (open vs closed). The average occupancy for the N79/G76 hydrogen bond is 0.14 ±0.07 in 2GMI (open), 0.75 ±0.05 in 5AIT-noE3 (closed), and 0.68 ±0.08 in 5AIT (closed). (The error is the calculated standard error over the ten simulations.) The box plots, which show the variance in more detail, show that although 2GMI clearly has a lower N79/G76 population, the variance in all three systems is high, a further indication that this hydrogen bond is weak 31,32 Finally, we calculated the fluctuations of the thioester in all three systems (see Fig. 6B). First, we aligned the backbone of the entire protein for the trajectory, then measured the RMSF of the thioester using CPPTRAJ. 42 The thioester is clearly more rigid when Ubc13~Ub is in the closed state. In 5AIT-noE3 and 5AIT the RMSFs of the thioester are 1.03 ±0.05 Å and 1.16 ±0.08 Å. In the open, 2GMI simulation, the RMSF is 2.45 ±0.33 Å. This pattern makes sense, since the thioester is weakly held in position by a weak hydrogen bond to N79. Interestingly, there is little difference between 5AIT-noE3 and 5AIT, which further indicates that the E3 is a passive actor in the ubiquitination reaction and activates Ub transfer by promoting the Ubc13~Ub closed state.</p><!><p>It is known that the presences of a RING E3 ligase increases the efficiency of E2-catalyzed ubiquitination. However, the source of this effect is unknown. Our MD simulations support the hypothesis that when the E3 puts the E2~Ub complex into the closed position, it puts tension on the thioester bond, making it easier to break. 6 .</p><p>We examined three model E2~Ub complexes: an open confirmation without the E3 (2GMI), a closed conformation without the E3 (5AIT-noE3), and a closed confirmation with the E3 (5AIT). We monitored the out-of-plane bend of the thioester linking the E2, Ubc13, and Ub and found that this bend is greater in the closed position than it the open. We note that although the average bend is similar across all models, the tail ends of the distribution differ. The closed systems are more likely to experience a high out-of-plane bend than the open system. Interestingly, the presence of the E3 had no effect. This implies that the E3 functions in part by promoting the closed position, but does not play an active catalytic role.</p><p>Next, we measured the flexibility and hydrogen bonding environment of the active site. We saw that the thioester fluctuates less and there is more hydrogen bonding when Ubc13~Ub is closed. When the thioester is held fixed, it can help preorganize the active site, improving catalytic efficiency. Once again, the presence of the E3 had no effect on the hydrogen bond population, which provides more evidence for a passive role for the E3.</p><p>Finally, we note that this study only measures the effect of the RING E3 indirectly. In other words, we did not calculate reaction rates of ubiquitination in the presence and absence of the RING E3. Because we used molecular dynamics, we could not simulate chemical reactions, since MD cannot simulate bonds breaking and forming. We also note that reaction enhancement of the RING E3 (~10x) 11,30 is much smaller compared to the enhancement of the E2 (~10 8 x), 45,46 meaning the RING E3 mechanism is more subtle than the electrostatic stabilization of transition states mechanism typically used by enzymes. 47,48 Placing a bond under slight strain is an example of a subtle enzymatic strategy and may explain why the ubiquitin transfer exclusively from the closed position and why that configuration is promoted by the RING E3 for Ub and other Ubl conjugating enzymes.</p><!><p>IS conceived and supervised the study and designed the experiments. JKJ and IS conducted the experiments. JKJ and IS wrote the manuscript.</p><!><p>This material is based upon work supported by the National Science Foundation under Grant No. CHE-1757874. Computational resources were provided in part by the MERCURY consortium (http://mercuryconsortium.org/) under NSF grants CHE-1229354, CHE-1662030, and CHE-2018427.</p><!><p>The SI contains figures depicting the model systems (2GMI, 5AIT-noE3, and 5AIT); custom force field parameters for the thioester bond, including partial charges and the improper torsion; information about the improper torsion parametrization; backbone RMSDs; AMBER input and parameter files for the MD simulations; and Gaussian09 input files for the thioester improper torsion parameterization. The SI cites references 13,35-37, and 43.</p><!><p>(</p>

ChemRxiv

Photosensitive Ru(II) Complexes as Inhibitors of the Major Human Drug Metabolizing Enzyme CYP3A4

We report the synthesis and photochemical and biological characterization of the first selective and potent metal-based inhibitors of cytochrome P450 3A4 (CYP3A4), the major human drug metabolizing enzyme. Five Ru(II)-based derivatives were prepared from two analogs of the CYP3A4 inhibitor ritonavir, 4 and 6: [Ru(tpy)(L)(6)]Cl2 (tpy = 2,2\xe2\x80\xb2:6\xe2\x80\xb2,2\xe2\x80\xb3-terpyridine) with L = 6,6\xe2\x80\xb2-dimethyl-2,2\xe2\x80\xb2-bipyridine (Me2bpy; 8), dimethylbenzo[i]dipyrido[3,2-a:2\xe2\x80\xb2,3\xe2\x80\xb2-c]phenazine (Me2dppn; 10) and 3,6-dimethyl-10,15-diphenylbenzo[i]dipyrido[3,2-a:2\xe2\x80\xb2,3\xe2\x80\xb2-c]phenazine (Me2Ph2dppn; 11), [Ru(tpy)(Me2bpy)(4)]Cl2 (7) and [Ru(tpy)(Me2dppn)(4)]Cl2 (9). Photochemical release of 4 or 6 from 7\xe2\x80\x9311 was demonstrated, and the spectrophotometric evaluation of 7 showed that it behaves similarly to free 4 (type II heme ligation) after irradiation with visible light but not in the dark. Unexpectedly, the intact Ru(II) complexes 7 and 8 were found to inhibit CYP3A4 potently and specifically through direct binding to the active site without heme ligation. Caged inhibitors 9\xe2\x80\x9311 showed dual action properties by combining photoactivated dissociation of 4 or 6 with efficient 1O2 production. In prostate adenocarcinoma DU-145 cells, compound 9 had the best synergistic effect with vinblastine, the anticancer drug primarily metabolized by CYP3A4 in vivo. Thus, our study establishes a new paradigm in CYP inhibition using metalated complexes and suggests possible utilization of photoactive CYP3A4 inhibitory compounds in clinical applications, such as enhancement of therapeutic efficacy of anticancer drugs.

photosensitive_ru(ii)_complexes_as_inhibitors_of_the_major_human_drug_metabolizing_enzyme_cyp3a4

INTRODUCTION<!>Compound Design and Synthesis.<!>Photochemistry.<!>CYP3A4 Inhibition Studies.<!>Biological Studies.<!>CONCLUSION<!>Materials. General Procedure for Synthesis of Ru(II) Complexes.<!>Synthesis of [Ru(tpy)(Me2bpy)(4)]Cl2 (7).<!>Synthesis of [Ru(tpy)(Me2bpy)(6)]Cl2 (8).<!>Synthesis of [Ru(tpy)(Me2dppn)(4)]Cl2 (9).<!>Synthesis of [Ru(tpy)(Me2dppn)(6)]Cl2 (10).<!>Synthesis of [Ru(tpy)(Ph2Me2dppn)(4)]Cl2 (11).<!>Instrumentation and Methods.<!>Studies on Recombinant CYP3A4.<!>Spectral Binding Titrations.<!>Inhibitory Potency Assays.<!>Crystallization of 7- and 8-Bound CYP3A4.<!>Determination of the X-ray Structures.<!>IC50 Determination Studies.<!>General Viability Assays.<!>EC50 Determination.<!>Chou\xe2\x80\x93Talalay Synergy Determination.

<p>Cytochrome P450s (CYPs) are heme-containing enzymes that play a crucial role in biosynthesis and metabolism. In addition to their activity in the liver, CYPs perform biosynthetic processing and drug oxidation in many other tissues, including the gastrointestinal tract and the brain. Extrahepatic CYP activity reduces local drug bioavailability and fuels resistance and progression of diseases, such as cancer, making CYPs attractive drug targets. Better understanding of the CYP inhibitory mechanism can also help lower the risk of dangerous drug–drug interactions. Genetic diversity of human CYPs leads to pharmacokinetic differences between people of different ethnic backgrounds that make drug responses highly varied. As a result, thorough characterization of small molecule interactions with CYPs is essential; in combination with genetic sequencing, these data will one day lead to better designed and personalized therapies.1</p><p>CYP3A4 is the most abundant liver and intestinal P450 isoform that oxidizes the majority of administered drugs and other xenobiotics relevant to human health.2–9 Fast and overly extensive drug metabolism can reduce treatment efficacy by requiring higher doses to achieve the full therapeutic effect. One way to overcome fast drug metabolism is the inhibition of CYP3A4. Currently, two CYP3A4 inhibitors, ritonavir and cobicistat, are part of multidrug therapies for treating HIV and hepatitis C virus (HCV) infections, whereas ketoconazole is co-prescribed with the quickly metabolized immunosuppressants in organ transplant patients.10–14 Anticancer therapy is another field where targeted CYP3A4 inhibition holds promise. CYP3A4 clears various types of anticancer drugs via both intestinal/hepatic metabolism and enhanced expression/in situ metabolism in solid tumors.15–19 Targeted inhibition of CYP3A4 in tumors has been identified as a potential solution to improve efficacy of chemotherapy by restoring sensitivity of cancer cells.19,20 Since most anticancer drugs have a narrow therapeutic index, potent CYP3A4 inhibition (as part of drug cocktails) has great potential to improve outcomes, lower chemotherapeutic doses, and minimize adverse effects. Importantly, clinicians have already identified an urgent need for localized CYP3A4 inhibition in malignant tissues.21 Localized inhibition was postulated to be more effective than systemic inhibition in colorectal cancer because a widely prescribed class of chemotherapeutics that destabilize micro-tubules are metabolized by CYP3A4 in cancer cells but by other CYPs in the liver.21 Importantly, there are no current methods that achieve tissue-specific blockade of CYP activity. Moreover, unlike the thousands of organic small molecules characterized as CYP inhibitors, inducers, or substrates, only a small handful of metal complexes have been investigated for CYP targeting.22–24</p><p>With the potential benefits in mind, we identified photocaging as a viable strategy to achieve localized CYP inhibition. Photocaging is a powerful method for blocking the action of biologically active molecules and unleashing inhibitory compounds within desired tissues, through which highly controlled and localized CYP inhibition can be achieved.22,23 Toward this goal, Ru(II)-based photocaging can facilitate small molecule release in a noninvasive manner to provide spatial and temporal control over biological activity.25–27 Photocaging has been exploited in basic research and for drug activation during photochemotherapy (PCT),28–30 with recent in vivo validation of Ru(II)-PCT.31 In addition to PCT, Ru(II) complexes show attractive properties for photodynamic therapy (PDT) applications, including high stability and cell permeability,32,33 low inherent toxicity,34–37 and higher light-to-dark ratios for cell death compared to clinically approved PDT compounds.29,38 Due to their rich photochemistry and resistance to photobleaching,39 a common problem with current organic photosensitizers,40 ruthenium complexes are emerging as a promising new class of PDT agents,29,41–43 some of which have advanced to clinical trials.44–47 One recent example is the Ru(II) photosensitizer TLD-1433, which is currently in phase II clinical trials for the treatment of bladder cancer.48–50</p><p>Many small molecules that target CYPs contain N-donor heterocycles that coordinate to the heme iron in the active site (type II ligation) to create strong and stable enzyme–inhibitor complexes.51–53 Ru(II) photocaging is an effective strategy for blocking N-donor heterocycles from binding to their targets, including the hemes found in CYP enzymes, via strong and stable coordination between N-donors, such as imidazoyl and pyridyl groups, and the Ru(II) centers of the photocages.26 Examples include the photochemical release of the CYP17A1 inhibitor abiraterone in PC3 prostate adenocarcinoma cancer cells,23 CYP11B1 inhibitors metyrapone and etomidate caged with the Ru(bpy)2 (bpy = 2,2′-bipyridine) fragment,22 and photocaged analogs of the pan-P450 inhibitor econazole that function as photoactivated cytotoxic and emissive agents in DLD-1 colon adenocarcinoma cancer cells.24</p><p>Herein, we report the design, synthesis, and biochemical characterization of a series of photocaged CYP3A4 inhibitors. Compounds were designed as Ru(II)-caged analogs of the antiretroviral drug ritonavir,54 which is a CYP3A4 inhibitor that binds tightly to the heme iron center via its thiazole ring.51,53 Two types of Ru(II) photocaging groups were employed that show either single action PCT or dual action PCT/PDT behaviors. All compounds were highly stable in solution in the dark but released CYP3A4 inhibitors readily upon irradiation with visible light, enabling type II heme iron ligation. While the main goal of the project was to design and employ light-activated CYP inhibitory molecules, one unexpected and significant finding was that, even without light activation, some Ru(II) compounds could potently inhibit CYP3A4 by binding to the active site without heme ligation. A direct inhibitory action between a large metal complex and a CYP target was verified by X-ray crystallography. Finally, we report that photocaged CYP3A4 inhibitors can function as dual action PDT and PCT agents that can both generate 1O2 and release the inhibitor upon irradiation, respectively. It is shown that these compounds work synergistically with the microtubule-destabilizing drug vinblastine, primarily metabolized by CYP3A4 in vivo. Thus, this work establishes a new paradigm in CYP inhibition and raises the possibility that photoactive CYP3A4 inhibitory compounds can be utilized in clinical applications, such as enhancement of therapeutic efficacy of anticancer drugs.</p><!><p>To begin our studies, we surveyed the literature for known type II inhibitors of CYP3A4. Clinical examples include ketoconazole (1), fluconazole (2), and ritonavir (3) that contain imidazole, triazole, or thiazole N-donors, respectively (Figure 1).51,55,56 Instead, we chose to focus our efforts on CYP3A4 inhibitors containing pyridyl groups that are analogs of ritonavir (4–6).57–59 Pyridine-containing compounds show more favorable properties for Ru(II) photocaging than other heterocyclic compounds, including strong and stable binding to Ru(II) in the dark and facile release when irradiated with low-energy light.23,60–62 Compounds 4–6 inhibit CYP3A4 in the low μM to nM range in in vitro assays with a fluorogenic substrate (vide infra) and, as verified by spectroscopic and X-ray crystallography analyses, inhibit CYP3A4 by ligating directly to the heme iron via the pyridine nitrogen.57–59 Analogs 4 and 6 were chosen over 5, which showed the lowest IC50 value of the series (90 nM) but had the potential to create solubility problems in Ru(II)-caged complexes due to its hydrophobic nature. Compound 4 was obtained using a modified three-step synthetic route that used trityl protection of 3-thiopropanoic acid (Scheme S1).57 Compound 6 was synthesized from S-2-mercapto-3-phenylpropanoic acid63 following a literature protocol.58</p><p>Five Ru(II) complexes containing the caged analogs of CYP3A4 inhibitors 4 and 6 were prepared as shown in Scheme 1. Complexes 7 and 8, coupled to the [Ru(tpy)(Me2bpy)] fragment as the caging group, were designed to demonstrate single action PCT behavior, similar to the pyridine model complex [Ru(tpy)(Me2bpy)(py)](PF6)2,60 as well as caged inhibitors of cysteine proteases64,65 and CYP17A123 reported by us in prior studies. Analogs 9–11, containing the [Ru(tpy)(L)] fragments as the photocaging groups, where L = dimethylbenzo[i]dipyrido[3,2-a:2′,3′-c]phenazine (Me2dppn) and 3,6-dimethyl-10,15-diphenylbenzo[i]dipyrido-[3,2-a:2′,3′-c]phenazine (Me2Ph2dppn), were synthesized to provide dual action PCT/PDT capabilities. The reaction of 4 or 6 with [Ru(tpy)(Me2bpy)(Cl)]Cl66 in a 1:1 mixture of EtOH and H2O at 80 °C gave the photocaged inhibitors 7 and 8 in 75% and 63% yield, respectively, after chromatography over alumina. Complexes 9–11 were obtained by treating 4 or 6 with [Ru(tpy)(Me2dppn)Cl]Cl67 or [Ru(tpy)(Ph2Me2dppn)Cl]Cl68 in a 1:1 mixture of EtOH and H2O at 80 °C in 48–61% yield after chromatography over alumina. The ligands Me2dppn and Me2Ph2dppn found in complexes 9–11 were included to promote ligand dissociation (PCT) from the triplet ligand field (3LF) state(s) and (1O2) generation (PDT) from the dppn-centered 3ππ* excited state(s). Importantly, we were motivated to use these ligands because our prior studies confirmed that dual action PCT/PDT behavior was necessary to achieve efficient death of triple negative breast cancer cells in 3D pathomimetic assays.65</p><p>Complexes 7–11 were characterized by multiple methods, including electronic absorption, 1H NMR, COSY, and IR spectroscopies and electrospray ionization mass spectrometry (ESI-MS). The electronic absorption spectra of 7 and 8 exhibit maxima at 474 nm (ε = 7700 M−1 cm−1) and 470 nm (ε = 9700 M−1 cm−1), respectively, that are in good agreement with the corresponding pyridine model complex [Ru(tpy)(Me2bpy)(py)](PF6)2.60 Likewise, the electronic absorption spectra of 9 (λmax 485 nm, ε = 13 500 M−1 cm−1) and 10 (λmax 480 nm, ε = 12 000 M−1 cm−1) show maxima consistent with [Ru(tpy)(Me2dppn)(py)](PF6)2.69 The electronic absorption spectrum of 11 exhibits a maximum at 491 nm (ε = 13 500 M−1 cm−1) that is slightly red-shifted compared to those of 9 and 10, which agrees well with data for [Ru(tpy)(Ph2Me2dppn)(py)](PF6)2.68 NMR spectra of complexes 7–11 show resonances ranging from 10 to 1 ppm that are consistent with the presence of the Ru(II)-caging groups, as well as peaks that are attributed to inhibitors 4 and 6 present in these structures. In particular, spectra for complexes 7–11 show singlets in the region of 2.5–1.0 ppm that are consistent with the two diasterotopic methyl groups present in the ligands Me2bpy, Me2dppn, and Ph2Me2dppn. Methyl groups on the same face of the Ru(tpy) plane as the monodentate pyridyl ring are shifted upfield by ~0.7 ppm relative to resonances below that plane due to the shielding effect of the pyridyl ring; these shifts are similar to other photocaged complexes we have characterized in the past.23,64,65 Mass spectra of the photocaged complexes show major peaks with suitable isotope patterns with m/z values consistent with that expected for parent molecular dications [Ru(tpy)(Me2bpy)(4)]2+ (7, m/z = 474) and [Ru(tpy)(Me2bpy)(6)]2+(8, m/z = 526) and the monocations ([Ru(tpy)(Me2dppn)(4)]Cl)+ (9, m/z = 1159), ([Ru(tpy)(Me2dppn)(6)]Cl)+ (10, m/z = 1263), and ([Ru(tpy)(Ph2Me2dppn)(6)]Cl)+ (11, m/z = 1415). Taken together, these data are consistent with the structural assignments shown in Scheme 1.</p><!><p>The irradiation of 7 effectively liberates 4, resulting in ligand exchange with a solvent molecule, generating the corresponding [Ru(tpy)(Me2bpy)(L)]2+ (L = H2O or CH3CN) product in H2O or CH3CN, respectively, under N2 atmosphere. Photoactivated ligand exchange (λirr = 500 nm) of 7, with absorption maximum at 474 nm, results in a blue shift to 450 nm in CH3CN (Figure 2A) and a red shift to 495 nm in H2O (Figure 2B). The resulting absorption maxima are consistent with the formation of the corresponding product with a coordinated CH3CN or H2O molecule.23,60,70 Similarly, the irradiation of 8 with 500 nm light in CH3CN resulted in a decrease in 470 nm absorption and a concomitant increase at 455 nm. This hypsochromic shift in the metal-to-ligand charge transfer (MLCT) band is consistent with the substitution of 6 coordinated to the Ru(II) metal through a pyridine unit for a CH3CN solvent molecule (Figure S8).23,70 The presence of an isosbestic point at 463 nm indicates the formation of a single photoproduct, [Ru(tpy)(Me2bpy)(CH3CN)]2+. Comparable changes in the electronic absorption spectra of 9–11 are observed under similar experimental conditions (Figures S9–S11).</p><p>The quantum yields (ΦLE) for the ligand exchange with a solvent molecule for 7–11 are listed in Table 1. For 7, ΦLE values of 0.15(1) in H2O and 0.31(1) in CH3CN were measured upon 500 nm irradiation (Table 1). The value in H2O is lower than that observed for [Ru(tpy)(Me2bpy)(py)]2+, ΦLE = 0.41(2), but similar in CH3CN, ΦLE = 0.33(1).60 The lower quantum yield observed for 7 vs Ru(tpy)(Me2bpy)(py)]2+ in H2O can be attributed to the lower solubility of CYP3A4 inhibitor 4 in water as compared to pyridine, which reduces the ability of the former to escape the solvent cage upon release from Ru(II). Similarly, Table 1 reveals a ΦLE value lower for 8 relative to 7 in CH3CN, which likely arises from the larger size and poorer solubility of inhibitor 6 as compared to 4. Following the same trend as 7 and 8, complex 9 containing the CYP3A4 inhibitor 4 showed ~2-fold more efficient photorelease than its analog 10 containing the bulky inhibitor 6. Complex 11 showed the most efficient photorelease in the 9–11 subseries, which is consistent with our earlier observations showing that complexes containing arylated Me2dppn derivatives, such as Ph2Me2dppn, undergo more efficient photorelease than Me2dppn derivatives.68 Complex 7 exhibits the highest ligand exchange quantum yield of the five complexes. It is hypothesized that the initially populated 1MLCT excited state intersystem crosses to the triplet manifold, populating both the lowest-energy dppn 3ππ* state and the 3LF states in 9–11, and the population of the latter results in ligand dissociation. The absence of a lowest-energy long-lived dppn-centered 3ππ* excited state in 7 and 8 precludes the bifurcation of intersystem crossing, resulting in an increased population of the 3LF state and, consequently, greater photoinduced ligand exchange quantum yield as compared to 9–11.69,71</p><p>In addition to photosubstitution of the monodentate ligand, 9–11 produce cytotoxic 1O2 through the population of the lowest-energy, long-lived 3ππ* excited state upon irradiation. The quantum yields for 1O2 production, ΦΔ, by 9–11 of 0.59(6), 0.57(6), and 0.80(7), respectively, are comparable to those of other dual-activity complexes possessing dppn ligands, such as [Ru(tpy)(Ph2Me2dppn)(py)](PF6)268 and [Ru(tpy)(Me2dppn)(imatinib)]2+ (Table 1).70,72 Our prior studies established that the Ru(II) photocaging group [Ru(tpy)(Me2bpy)] found in 7 and 8 does not generate 1O2 either before or after photorelease because its excited state lifetime is too short to undergo bimolecular reactions, as is the case with other Ru(II) complexes containing the tpy ligand or those that undergo facile ligand photodissociation.69,70</p><p>The stability of 7–11 was assessed in cell growth medium at 37 °C as previously described.73,74 No spectral changes were observed for 7–9 in the dark (Figures S18–S24) over a course of 24 h, consistent with the exceptional stability of Ru(II)-caged aromatic heterocycles. Complexes 10 and 11 did show some spectral changes over the 24 h period that are consistent with compound precipitation from solution and/or thermal ligand dissociation.</p><!><p>After establishing that CYP3A4 inhibitor 4 is photochemically released from its Ru(II) cage 7, the complex was evaluated against the purified CYP3A4 enzyme under dark conditions and upon irradiation. Stock solutions of 7 were left in the dark or exposed to light (λirr = 400–700 nm, tirr = 40 min) before titrating against soluble CYP3A4 (residues 3–22 deleted). Heme binding to the iron center in CYP3A4 was monitored via electronic absorption spectroscopy. Data indicated that the caged inhibitor 7 effectively released 4 from the ruthenium center upon irradiation with visible light, allowing the pyridine functional group of 4 to bind to CYP3A4 via a type II heme ligation. The difference spectra were similar to those obtained for the free inhibitor 4 and showed an increase in intensity at 427 nm and a decrease at 407 nm, consistent with type II binding (Figure 3A), where the water ligand is substituted with pyridine, converting the heme center to a low spin ferric state. Hyperbolic fitting to the titration plot resulted in Kd = 340 nM for 7 under irradiation (Figure 3B). In contrast, no spectral evidence for type II binding was observed during titration of CYP3A4 with 7 under dark conditions. Minor perturbations to the absorption spectra were attributed to the Ru(II) complex, strongly absorbing at 400–500 nm, rather than type II binding (Figure 3C). Similarly, the titration of CYP3A4 with a control compound, [Ru(tpy)(Me2bpy)(Cl)]Cl, led to minor spectral changes. Taken together, these data indicate that type II heme binding is effectively blocked by Ru(II) caging and that irradiation with visible light triggers the release of inhibitor 4, enabling its ligation to the CYP3A4 heme.</p><p>Next, compound 7 was evaluated for its ability to inhibit CYP3A4 activity under light and dark conditions. The free inhibitor 4 and [Ru(tpy)(Me2bpy)(Cl)]Cl were included as controls. IC50 values were determined using a fluorogenic assay that monitors the O-debenzylation of 7-benzyloxy-4-trifluoromethylcoumarin (BFC), with 100% activity set at vehicle (DMSO) only. After treatment with visible light (λirr =400–700 nm, tirr = 40 min), 7 inhibited CYP3A4 nearly as well as free inhibitor 4 (IC50 of 2.2 μM and 1.5 μM, respectively), which agrees well with the spectral data (Figures 2 and 3) showing that 4 is released from 7 upon irradiation. However, to our surprise, the intact 7 was more potent in the dark (IC50 of 0.9 μM; Figure 4), suggesting that the Ru(II) complex could bind to CYP3A4 more strongly than free 4. Control experiments with [Ru(tpy)(Me2bpy)(Cl)]Cl (IC50 > 50 μM) showed that CYP3A4 inhibition was not due to just the Ru(II) fragment. Taken together, these data indicate that 7 is a stronger inhibitor when kept in the dark as compared to under irradiation.</p><p>To confirm that the intact 7 is able to access the active site, we crystallized the CYP3A4-7 complex and solved the structure to 2.5 Å resolution. Indeed, 7 was bound in a well-defined manner within the active site (Figure 5). The inhibitor tail curls above the heme without direct binding to the iron center, while the bulky Ru(II) cage stacks inside the substrate channel. Protein–ligand interactions are predominantly hydrophobic. The inhibitor tail is surrounded by Phe241, Ile301, Phe304, and Ile369, whereas the ligands of the [Ru(tpy)(Me2bpy)] cage fragment stack with Phe108, Phe215, and Phe220 and are in close contacts with Phe57, Leu217, Met371, and Leu482. The anionic residues Asp76, Asp217, and Glu374 may also help to strengthen the inhibitory complex by creating favorable electrostatic interactions with the dicationic Ru(II) fragment. Importantly, the 7 N-pyridine does not bind to the heme iron because it is stably coordinated to Ru(II). This structure is highly valuable because it demonstrates that strong CYP3A4 inhibition by the intact, nonirradiated chimeric compound does not require Fe–N ligation.</p><p>On the basis of the finding that 7 potently inhibits CYP3A4 in the dark, the inhibitory assays for complexes 8–11 were conducted under both dark and light conditions (λirr = 400–700 nm, tirr = 40 min). IC50 values for the BFC activity of CYP3A4 are presented in Table 2. Complex 8, which contains inhibitor 6, inhibits CYP3A4 nearly to the same extent under dark and light conditions, giving a phototherapy index (PI) of 1.1. Interestingly, under dark conditions, 8 inhibits CYP3A4 roughly twice as potently as 7 (IC50 of ~400 nM). Since 8 willingly cocrystallized with CYP3A4, we also determined the CYP3A4-8 complex structure. Despite the fact that resolution was similar, 2.5 Å, 8 was poorly defined and the electron density around the ligand was discontinuous, which can be attributed to multiple binding modes. Nonetheless, the Ru-center and the core of the inhibitor tail could be located, allowing ligand fitting. As shown in Figure S12, the [Ru(tpy)(Me2bpy)] cage binds within the same pocket in the substrate channel. The inhibitor end-portion, in turn, similarly curls above the heme. Again, the complex is largely stabilized by aromatic stacking and hydrophobic interactions mediated by Phe57, Phe108, Phe220, Phe221, Phe241, and Phe304.</p><p>Similar to 7, the Me2dppn complexes 9 and 10 inhibited CYP3A4 more potently under dark than light conditions but at lower concentrations than 7. Dark IC50 values for 9 and 10 were in the 250–280 nM range, with PI values of 0.30 and 0.61, respectively. Attempts to cocrystallize 9 and 10 with CYP3A4 were unsuccessful. Examination of inhibitors' solutions showed that both 9 and 10 have a tendency to aggregate. Compound aggregation in solution can lead to false positives for enzyme inhibition, e.g., by trapping active enzyme within colloidal particles that block access of substrates.75 One way to distinguish between specific and nonspecific inhibition is to add detergents or other solubilizing agents to enzymatic assays. Therefore, we screened several detergents known to break up aggregates, including CHAPS, CYMAL-5, octylglucoside, and cyclodextrin. CYP3A4 was highly sensitive to detergents, with most detergents abolishing the BFC activity even in the absence of inhibitors. However, CYP3A4 preserved ~80% activity in the presence of 2% cyclodextrin. The latter agent was used for re-evaluation of 10 and, as we found, reversed the trend: dark IC50 = 1.02 μM, light IC50 = 0.44 μM, giving a PI of 2.3. Thus, aggregation was at least partially responsible for CYP3A4 inhibition by 10 in the dark. Importantly, the higher PI with 2% cyclodextrin was due to a higher IC50 for 10 in the dark; light data with and without 2% cyclodextrin were virtually the same and agreed well with those for free 6. Finally, the bulky Ph2Me2dppn-containing complex 11 showed an improved PI value, 1.90, as compared to PI = 0.61 for 10, implying that the larger caging group [Ru(tpy)(Ph2Me2dppn)] disfavors binding to the CYP3A4 active site.</p><p>In order to characterize the scope of CYP3A4 inhibition, we screened a library of 15 compounds, consisting of a diverse set of mono- and dicationic Ru(II) complexes (12–26, Figure 6; see Figure S25 for structures), against the purified enzyme. All complexes were screened against CYP3A4 under dark conditions at a concentration of 1 μM. Activities were determined using BFC as a substrate and expressed as percentage vs vehicle (DMSO) control. Thirteen complexes failed to decrease CYP3A4 activity below 75% at 1 μM concentration, confirming that potent CYP3A4 inhibition is not a general property of Ru(II) complexes. Only two compounds, [Ru(bpy)2(dppn)](PF6)2 (18) and [Ru(dppz)2(bpy)]Cl2 (19) reduced CYP3A4 activity below 75% at 1 μM concentration. Collectively, these data reveal complex structure–activity relationships for inhibition of CYP3A4 by Ru(II) complexes that warrant further investigation.</p><p>To gain further insight into the potential biological applications of Ru(II)-based CYP inhibitors, we determined IC50 values for 4 and complexes 7 and 9 against microsomal CYP3A4 and two other major drug metabolizing enzymes, CYP1A2 and CYP2C9,76 using commercially available inhibitor screening kits (BioVision, Table 3). It should be pointed out that protein concentration and CYP:reductase ratios in the BioVision kits and our soluble reconstituted system were different, owing to which data in Tables 2 and 3 cannot be directly compared. For microsomal CYP3A4, inhibitor 4 was active in the nM range, with the IC50 values being nearly the same (~200 nM) under dark and light conditions. Complex 7 also inhibited CYP3A4 at nanomolar concentrations but more potently under dark vs light conditions, following the same trend as data for 7 presented in Table 2. Importantly, both 4 and 7 were much weaker inhibitors of CYP1A2 and CYP2C9. The selectivity of 4 for CYP3A4 was ~500-fold higher, whereas 7 inhibited CYP3A4 ~70-to-130-fold and 60-to-76-fold more strongly than the other CYPs under dark and light conditions (460–470 nm; 20 min), respectively. A multifold difference in IC50 measured for 4 and 7 under dark conditions suggests some influence of the released Ru(II) cage in the inhibition. The respective data were also collected for Ru(II) complex 9, which contains the same inhibitor 4 linked to the bulky and more hydrophobic photocaging group [Ru(tpy)(Me2dppn)]. Compared to 7, the inhibitory potency of 9 for microsomal CYP3A4 was ~8-and 5-fold lower under dark and light conditions, respectively, and its selectivity for other isoforms could not be accurately measured due to solubility problems. Even so, there was a common trend, as all three compounds displayed higher specificity for CYP3A4 albeit to a different extent.</p><!><p>Studies on the interaction of 7–11 with isolated CYP3A4 showed that inhibition can be achieved via blockage of the active site by the intact caged compounds, light-activated release of the inhibitory fragment and its subsequent heme ligation, and efficient 1O2 generation. However, questions remained regarding the role of aggregation vs direct inhibition of CYP3A4 in the dark, due to sensitivity of the recombinant enzyme to common detergents. These challenges prompted us to utilize an in vitro cell-based assay to probe for CYP3A4 inhibition by our compounds. Importantly, prior studies demonstrated that CYP3A4 inhibitors work synergistically with microtubule-destabilizing drugs in cancer cells.77,78 We chose to evaluate our compounds in DU-145 prostate cancer cells because (i) they have high levels of CYP3A4 expression, (ii) prior studies showed synergism between the CYP3A4 inhibitor ketoconazole and vinblastine,77 a drug commonly used in combination therapies for various cancers, and (iii) utilization of a validated protocol for in vitro detection of synergism between a chemotherapeutic drug and CYP3A4 inhibitors would provide a reliable cell-based assay for evaluation of our compounds. Vinblastine binds to tubulin and stops production of microtubules, leading to M-phase specific cell cycle arrest. Synergism between vinblastine and ketoconazole was previously achieved by blocking CYP3A4-dependent vinblastine metabolism in several prostate cancer cell lines.77 On the basis of this knowledge, we designed experiments with DU-145 cells and our panel of compounds. First, free CYP3A4 inhibitors 4 and 6 (5 μM) were evaluated against DU-145 cells in the presence or absence of vinblastine (5 nM). Cells were treated with 4 or 6 and vinblastine or vehicle, and viability was assessed after 72 h by the 3-(4,5-dimethylthiazol-2-yl)-2,5-diphenyltetrazolium bromide (MTT) assay. It was found that both 4 and 6 reduce viability of the DU-145 cells by up to ~40% in the presence of 5 nM vinblastine. This reduction of viability is similar to that observed with ketoconazole77 and suggests the synergy between CYP3A4 inhibition and the microtubule-destabilizing drug (Figure 7A).</p><p>Next, inhibitors 4 and 6 were evaluated alongside the photocaged inhibitors 7–11 in the dark or with irradiation in the presence of vinblastine (5 nM). In these experiments, cells were treated with 4, 6, or 7–11 (5 μM) and kept in the dark for 1 h, then the medium was replaced with medium containing 5 nM vinblastine, and cells were irradiated with blue light (460–470 nm, 20 min) or left in the dark for 20 min. Viabilities were determined 72 h after light treatment by the MTT assay. As expected, results with the free inhibitors 4 and 6 were virtually identical under dark and light conditions, ruling out synergy between 4 or 6 and light. In these experiments, synergy with 4 and 6 was less pronounced compared to incubations where 4 or 6 was left with cells for the full 72 h without medium replacement (Figure S13), which may indicate slower uptake of these inhibitors by DU-145 cells. Among the investigated compounds, complex 9, which not only releases the CYP3A4 inhibitor 4 (PCT) but also generates 1O2 (PDT), showed the strongest response by reducing viability to ~10% in the light compared to ~90% in the dark (Figure 7B). Again, results were less pronounced when 7–11 were left with DU-145 cells for the full 72 h without replacement of medium (Figure S13), further indicating that cell uptake is slower for some of the Ru(II) complexes. Nonetheless, complex 9 showed a strong response with replacement of medium after only 1 h, which supports the ability of 9 to penetrate DU-145 cells within that time frame.</p><p>Next, we probed the impact of CYP3A4 inhibition in cell-induced toxicity with vinblastine. Complex 9 was compared side-by-side with the [Ru(tpy)(Me2dppn)(py)](PF6)2 complex (27), which generates 1O2 just as efficiently69 but serves as a control by releasing pyridine rather than the CYP3A4 inhibitor 4. Experiments with 27 were important to carry out because prior studies demonstrated that PDT can work synergistically with microtubule-targeting drugs.79 The results in Figure 7C show that 9 (5 μM) produces a strong, dose-dependent synergy with vinblastine (0–5 nM), whereas less toxic 27 (5 μM) does not. These data suggest that CYP3A4 inhibition is synergistic with PDT and vinblastine.</p><p>In order to quantify drug synergy, the Chou–Talalay method was applied, which is the field standard for assessing the synergy of a drug combination.80–83 DU-145 cells were treated with either 4, 7, 9, 27, or vinblastine alone over a range of concentrations to determine EC50 (Table 4). For only 9, 27, and vinblastine, the EC50 values were <25 μM under light conditions (tirr = 20 min, λirr = 460–470 nm, 72 h MTT); in the dark, EC50 for all compounds was >25 μM. Next, DU-145 cells were treated with a combination of single concentrations of 9 or 27 and vinblastine over a range of concentrations spanning the EC50 values under light conditions (tirr = 20 min, λirr = 460–470 nm, 56 J/cm2, 72 h MTT), resulting in a panel of 16 distinct combinations (Figure 8 for 9, Figure S14 for 27). Viabilities for the single drug and combination treatments were compared against the vehicle control to measure the % effectiveness as the proportion between live and dead cells in a given treatment. By use of the dose and effect for each monotreatment and each combination, the combination index (CI) values for each treatment pair were calculated using Compusyn software (Figure 8).80 CI values less than 1 indicate synergy, equal to 1 indicate an additive effect, and greater than 1 indicate antagonism. For compound 9, 12 out of 16 combinations surveyed showed CI values <1, with the other four combinations showing CI values near 1, indicating high synergism between 9 and vinblastine under light conditions. In contrast, 27 showed weaker synergism under all concentrations surveyed under light conditions (Figure S14). Taken together, these data suggest that (i) 9 blocks intracellular metabolism of vinblastine via CYP3A4 inhibition and (ii) the CYP3A4 inhibition, PDT, and vinblastine act together and produce a stronger cytotoxic response in the DU-145 cells than the combination of PDT and vinblastine. Because microtubule-destabilizing drugs have deleterious side effects and narrow therapeutic indexes, the combination of localized CYP3A4 inhibition and PDT may prove to be a promising approach to achieve synergy and lower the doses of chemotherapeutic drugs like vinblastine.</p><!><p>This is the first report on the synthesis and biological evaluation of metal-based inhibitors of the major human drug metabolizing enzyme CYP3A4. Using two analogs of ritonavir, we synthesized and characterized five Ru(II)-caged CYP3A4 inhibitors (7–11) that showed either single action PCT or dual action PCT/PDT behavior. Serendipitously, we demonstrated that CYP inhibition can be enhanced through inhibitor metalation, as the caged complexes can tightly and selectively bind to the CYP3A4 active site without heme ligation. Moreover, compound 9 was identified as a dual-action PCT/PDT lead compound, which effectively generates 1O2 and releases the CYP3A4 inhibitor to act synergistically with the common chemotherapeutic drug vinblastine in DU-145 adenocarcinoma cells. These findings warrant further studies on photoactive CYP inhibitory compounds to determine their potential use for clinical applications, such as enhancement of therapeutic efficacy of chemotherapeutic drugs.</p><!><p>Some reactions were performed under ambient atmosphere unless otherwise noted. Anaerobic reactions were performed by purging the reaction solutions with Ar or nitrogen. Complexes 12 and 13 were purchased (Strem Chemicals). Complexes 14,84 15,85 16,86 17,86 18,87 19,88 20–22,89 23 and 24,90 25,70 and 2691 were prepared following literature protocols. For synthesis of 7–11, a solution of [Ru(tpy)(L1)Cl]Cl in EtOH was treated with pyridine. Water was added, and the mixture was deoxygenated by bubbling Ar through a submerged needle for 20 min. The pressure tube was sealed and heated to 80 °C for 16 h. The reaction mixture was cooled to room temperature, concentrated, and the residue was purified by column chromatography on neutral alumina to give [Ru(tpy)(L1)(L2)](Cl)2 complexes.</p><!><p>[Ru(tpy)(Me2bpy)Cl]Cl67 (19.0 mg, 0.0300 mmol) was added to a solution of 4 (28 mg, 0.070 mmol) in a 1:1 mixture of EtOH and H2O (3.0 mL each) under inert atmosphere in a pressure flask. The pressure flask was wrapped with aluminum foil. The solution was purged with argon for 10 min at room temperature. The pressure flask was sealed, and the reaction mixture was refluxed at 80 °C for 16 h under an inert atmosphere. The color of the reaction mixture turned from purple to brown. The reaction mixture was cooled to room temperature and concentrated under reduced pressure. The crude product was purified over neutral alumina (5% MeOH/DCM) in the dark to give 7 as a brown solid (25 mg, 75%): 1H NMR (400 MHz CD3OD) δ 8.75–8.72 (m, 1H), 8.69 (d, 2H, J = 8.4 Hz), 8.63 (d, 2H, J = 8.0 Hz), 8.48–8.45 (m, 1H), 8.29 (t, 1H, J = 8.0 Hz), 8.24–8.18 (m, 3H), 8.14 (t, 2H, J = 8.0 Hz), 7.81–7.72 (m, 2H), 7.66–7.60 (m, 4H), 7.55 (d, 1H, J = 5.6 Hz), 7.28–7.17 (m, 5H), 7.08 (t, 1H, J = 7.6 Hz), 7.01 (d, 1H, J = 7.6 Hz), 4.04–3.93 (m, 2H), 3.86–3.79 (m, 1H), 2.94–2.88 (m, 1H), 2.72 (t, 2H, J = 7.2 Hz), 2.68–2.65 (m, 1H), 2.60–2.55 (m, 2H), 2.34 (t, 2H, J = 7.2 Hz), 2.03 (s, 3H), 1.50 (s, 3H), 1.36–1.29 (m, 9H); IR νmax (cm−1) 3372, 2926, 2830, 1740, 1711, 1536, 1447, 1371, 1223, 1022, 519; ESMS calcd for C50H54N8O3RuS (M+2) 474, found 474; UV–vis λmax = 474 nm (ε = 7700 M−1 cm−1).</p><!><p>Compound 8 was prepared by following the general procedure by treating [Ru(tpy)(Me2bpy)Cl]Cl67 (13 mg, 0.022 mmol) with 6 (23 mg, 0.044 mmol) in EtOH (3 mL) and water (3 mL). The residue was purified by column chromatography on neutral alumina (4–6% MeOH in DCM) to give a red solid (15 mg, 63%). 1H NMR (400 MHz, methanol-d4) δ 8.76 (d, J = 8.0 Hz, 1H), 8.73–8.62 (m, 3H), 8.57 (t, J = 7.3 Hz, 1H), 8.50 (d, J = 8.1 Hz, 1H), 8.27 (dd, J = 9.2, 6.9 Hz, 3H), 8.18 (ddd, J = 12.3, 6.1, 3.6 Hz, 2H), 8.07 (ddt, J = 7.9, 3.9, 2.0 Hz, 1H), 7.77 (dt, J = 13.1, 7.7 Hz, 2H), 7.68–7.60 (m, 1H), 7.60–7.54 (m, 1H), 7.48 (dq, J = 8.7, 6.0, 4.8 Hz, 3H), 7.28–7.15 (m, 8H), 7.15–6.99 (m, 4H), 3.82 (ttd, J = 8.9, 6.7, 3.0 Hz, 1H), 3.55 (dd, J = 9.7, 5.7 Hz, 1H), 3.08 (ddd, J = 13.7, 9.7, 6.6 Hz, 1H), 2.97–2.81 (m, 2H), 2.78–2.60 (m, 5H), 2.10 (t, J = 8.1 Hz, 5H), 1.51 (d, J = 3.1 Hz, 3H), 1.39–1.25 (m, 9H), 1.25–1.18 (m, 2H); IR (KBr) 3395, 3242, 3058, 3027, 2974, 2927, 2859, 1698, 1660, 1602, 1542, 1523, 1496, 1447, 1388, 1364, 1282, 1248, 1168, 1119, 1078, 1016, 916, 778, 748, 701, 672, 646; UV–vis λmax = 470 nm (ε = 9700 M−1 cm−1); ESMS calculated for C58H62N8O3RuS [M2+] 526, found 526.</p><!><p>Compound 9 was prepared by following the general procedure by treating [Ru(tpy)(Me2dppn)Cl]Cl65 (22 mg, 0.029 mmol) with 4 (25 mg, 0.058 mmol) in EtOH (4 mL) and water (4 mL). The residue was purified by column chromatography on neutral alumina (3–4% MeOH in DCM) to give a red solid (20 mg, 57%). 1H NMR (400 MHz, methanol-d4) δ 9.93 (dd, J = 8.4, 1.6 Hz, 1H), 9.33 (d, J = 8.2 Hz, 1H), 8.98 (d, J = 12.1 Hz, 2H), 8.78 (d, J = 8.1 Hz, 1H), 8.78–8.68 (m, 2H), 8.65 (d, J = 8.2 Hz, 1H), 8.33–8.23 (m, 3H), 8.26–8.17 (m, 1H), 8.16 (t, J = 7.8 Hz, 2H), 8.11 (h, J = 3.8 Hz, 2H), 7.75–7.68 (m, 2H), 7.68–7.58 (m, 3H), 7.61–7.50 (m, 1H), 7.50 (t, J = 6.7 Hz, 1H), 7.42 (d, J = 8.3 Hz, 1H), 7.26–7.09 (m, 7H), 4.03 (d, J = 4.6 Hz, 2H), 3.85–3.74 (m, 1H), 3.34 (d, J = 3.6 Hz, 1H), 2.86 (ddd, J = 15.9, 10.9, 5.5 Hz, 1H), 2.74 (t, J = 7.1 Hz, 2H), 2.66–2.58 (m, 1H), 2.58 (q, J = 2.7, 2.3 Hz, 2H), 2.37 (t, J = 7.0 Hz, 2H), 2.34 (s, 3H), 1.79 (s, 3H), 1.31 (d, J = 7.6 Hz, 9H); IR (KBr) 3394, 3056, 3027, 2924, 2853, 1966, 1697, 1662, 1542, 1520, 1446, 1363, 1247, 1167, 1056, 1017, 880, 776, 703; UV–vis λmax = 485 nm (ε = 13 500 M−1 cm−1); ESMS calculated for C62H58ClN10O3RuS [M+] 1159, found 1159.</p><!><p>Compound 10 was prepared by following the general procedure by treating [Ru(tpy)(Me2dppn)Cl]Cl65 (17 mg, 0.022 mmol) with 6 (23 mg, 0.044 mmol) in EtOH (5 mL) and water (5 mL). The residue was purified by column chromatography on neutral alumina (3–4% MeOH in DCM) to give a red solid (17 mg, 61%). 1H NMR (400 MHz, methanol-d4) δ 9.96 (dd, J = 8.3, 6.4 Hz, 1H), 9.45 (dd, J = 8.2, 3.8 Hz, 1H), 9.14–9.02 (m, 2H), 8.83–8.72 (m, 2H), 8.68 (d, J = 8.0 Hz, 1H), 8.59–8.50 (m, 1H), 8.35 (t, J = 6.3 Hz, 1H), 8.26 (dtd, J = 15.0, 9.2, 5.4 Hz, 4H), 8.15 (dd, J = 18.4, 6.0 Hz, 2H), 8.05–7.98 (m, 1H), 7.77–7.66 (m, 3H), 7.61–7.53 (m, 3H), 7.57–7.30 (m, 2H), 7.18 (s, 4H), 7.27–7.12 (m, 1H), 7.16–7.04 (m, 5H), 3.76–3.46 (m, 1H), 3.47 (dt, J = 9.1, 6.0 Hz, 1H), 3.13–2.98 (m, 1H), 2.80 (dtd, J = 43.5, 15.3, 6.5 Hz, 1H), 2.67 (s, 2H), 2.67–2.58 (m, 1H), 2.56 (s, 1H), 2.36 (dd, J = 11.6, 3.3 Hz, 3H), 2.16 (dp, J = 21.1, 7.0 Hz, 1H), 1.80 (s, 3H), 1.29 (d, J = 8.1 Hz, 11H); IR (KBr) 3365, 3256, 3056, 3025, 2974, 2922, 2852, 2360, 2342, 1868, 1792, 1760, 1733, 1698, 1653, 1558, 1542, 1522, 1447, 1388, 1362, 1243, 1161, 1056, 881, 841, 775, 752, 700, 669; UV–vis λmax = 480 nm (ε = 12 000 M−1 cm−1); ESMS calculated for C70H66ClN10O3RuS [M+] 1263, found 1263.</p><!><p>Compound 11 was prepared by following the general procedure by treating [Ru(tpy)(Ph2Me2dppn)Cl]Cl68 (19 mg, 0.021 mmol) with 6 (22 mg, 0.042 mmol) in EtOH (3 mL) and water (3 mL). The residue was purified by column chromatography on neutral alumina (3–4% MeOH in DCM) to give a red solid (14 mg, 48%). 1H NMR (400 MHz, methanol-d4) δ 9.42–9.30 (m, 1H), 8.88–8.58 (m, 4H), 8.50 (dt, J = 6.6, 3.3 Hz, 1H), 8.32–7.89 (m, 9H), 7.83–7.55 (m, 14H), 7.46–7.29 (m, 2H), 7.29–6.96 (m, 12H), 3.83–3.68 (m, 1H), 3.46 (td, J = 9.2, 6.7 Hz, 1H), 2.99 (dd, J = 13.7, 9.6 Hz, 1H), 2.93–2.47 (m, 7H), 2.28 (d, J = 10.6 Hz, 3H), 2.14 (tdd, J = 16.6, 8.0, 3.7 Hz, 2H), 1.73 (s, 3H), 1.33 (d, J = 4.0 Hz, 1H), 1.27 (d, J = 2.2 Hz, 10H); IR (KBr) 3255, 3057, 3025, 2973, 2924, 2853, 2360, 2330, 1698, 1684, 1653, 1558, 1542, 1522, 1496, 1490, 1448, 1420, 1387, 1362, 1246, 1166, 1073, 1014, 839, 773, 701, 670; UV–vis λmax = 491 nm (ε = 13 500 M−1 cm−1); ESMS calculated for C82H74ClN10O3RuS [M+] 1415, found 1415.</p><!><p>NMR spectra were recorded on a Varian FT-NMR Mercury 400 MHz spectrometer. UV–vis spectra were recorded on a Varian Cary 60 spectrophotometer. Steady state electronic absorption spectra were collected using an Agilent Cary 8453 diode array spectrometer, and emission data were collected using a Horiba FluoroMax-4 fluorimeter. All experiments involving DU-145 cells were carried out in Dulbecco's modified Eagle's medium containing 10% FBS and 1000 units/mL penicillin/streptomycin. The irradiation source for quantum yield measurements was a 150 W Xe arc lamp (USHIO) in a MilliArc lamp housing unit, powered by an LPS-220 power supply and an LPS-221 igniter (PTI). The emission wavelengths were selected using a CVI Melles Griot long-pass filter, and the appropriate irradiation wavelengths for photolysis experiments were selected with a bandpass filter (Thorlabs) and long-pass filter (CVI Melles Griot). The quantum yields (Φ) for ligand dissociation were determined in CH3CN with an irradiation wavelength of 500 nm. The rate of moles reacted at early irradiation times was determined by monitoring the decrease in the MLCT absorption maximum as a function of time. The photon flux of the lamp with a 435 nm long-pass filter and a 500 nm bandpass filter was determined using ferrioxalate actinometry as previously described in detail.92 Singlet oxygen quantum yields were performed using [Ru(bpy)3]2+ as a standard (ΦΔ = 0.81 in MeOH) and 1,3-diphenylisobenzofuran (DPBF) as a 1O2 trapping agent and following a previously established procedure.93</p><!><p>Full-length and truncated (Δ3–22) human CYP3A4 was expressed and purified as described previously.94</p><!><p>Equilibrium ligand binding to Δ3–22 CYP3A4 was monitored in a Cary 300 spectrophotometer at ambient temperature in 0.1 M phosphate buffer, pH 7.4, supplemented with 20% glycerol and 1 mM dithiothreitol. Inhibitors and caged compounds, with or without visible light irradiation (λirr = 400–700 nm, tirr = 40 min), were dissolved in DMSO and added to a 2 μM protein solution in small aliquots, with the final solvent concentration of <2%. Spectral dissociation constants (Kd) were determined from hyperbolic fits to titration plots.</p><!><p>Inhibitory potency for the 7-benzyloxy-4-(trifluoromethyl)coumarin (BFC) O-debenzylase activity of CYP3A4 was evaluated fluorometrically in a soluble reconstituted system. Full-length CYP3A4 and rat cytochrome P450 reductase (40 μM and 60 μM, respectively) were preincubated at room temperature for 1 h before 10-fold dilution with the reaction buffer consisting of 0.1 M potassium phosphate, pH 7.4, catalase, and superoxide dismutase (2 units/mL each). Prior to measurements, 85 μL of the reaction buffer was mixed with 10 μL of the NADPH-regenerating system (10 mM glucose, 0.2 mM NADP+, and 2 units/mL glucose 6-phosphate dehydrogenase), 5 μL of the protein mixture (0.2 μM final CYP3A4 concentration), and 2 μL of the cage/inhibitor solution or DMSO. The mixture was incubated for 2 min, after which 1 μL of 2 mM BFC and 1 μL of 7 mM NADPH were added to initiate the reaction. Accumulation of the fluorescent product, 7-hydroxy-4-(trifluoromethyl)coumarin, was monitored for 2 min at room temperature in a Hitachi F400 fluorimeter (λex = 404 nm; λem = 500 nm). Within this time interval, fluorescence changes were linear. The average of three measurements was used to calculate the remaining activity, with the DMSO-containing sample used as a control (100% activity). The IC50 values were derived from four-parameter logistic fittings to the [% activity] vs [inhibitor] plots.</p><!><p>Both complexes were crystallized using a microbatch method under oil. Prior to crystallization setup, Δ3–22 CYP3A4 (50–60 mg/mL in 75–100 mM phosphate, pH 7.4) was incubated with a 2-fold ligand excess for 15 min and centrifuged to remove the precipitate. The supernatant (0.4 μL) was mixed with 0.4–0.5 μL of the crystallization solution containing 10% PEG 3350 and 80 mM tribasic ammonium citrate, pH 7.0, for 7 and 8% PEG 3350 and 70 mM dl-malate, pH 7.0, for 8. Crystals were grown at room temperature for 2–3 days and cryoprotected with Paratone-N before freezing in liquid nitrogen.</p><!><p>X-ray diffraction data were collected at the Stanford Synchrotron Radiation Lightsource beam-lines 9-2 and 12-2. Crystal structures were solved by molecular replacement with PHASER95 and 5VCC as a search model. Ligands were built with eLBOW96 and manually fit into the density with COOT.97 The initial models were rebuilt and refined with COOT and PHENIX.96 Polder omit electron density maps were calculated with PHENIX. Data collection and refinement statistics are summarized in Table S1. The atomic coordinates and structure factors for the 7- and 8-bound CYP3A4 were deposited to the Protein Data Bank with the identifier codes 7KS8 and 7KSA, respectively.</p><!><p>Cytochrome P450 inhibitor screening kits for CYP3A4, CYP1A2, and CYP2C9 were obtained from BioVision. Stock solutions of compounds 4, 7, and 9 were prepared at 5× concentrations in the provided assay buffer. Stock solutions were dispensed into triplicate wells of a 96-well plate and irradiated (tirr = 20 min, λirr = 460–470 nm) or left in the dark. Compounds 4 and 7 (100 μM to 100 nM) and compound 9 (100 μM to 100 nM) were evaluated following the manufacturer's protocols. Percentage of enzyme activities was calculated from the initial linear slopes of the fluorescence vs time plots (first 5 min), using solvent control (no inhibitor, 1% MeCN in assay buffer) as 100% activity. The slope of the blank plot (no enzyme, 1% MeCN in assay buffer) was subtracted from each experimental slope value. Percent inhibition was expressed as the quotient of the blank subtracted experimental slopes over the blank subtracted solvent control slope. Igor Pro graphing software was used to produce % activity vs log(molarity) dose–response plots (Figures S15–S17), from which IC50 values were determined.</p><!><p>DU-145 cells were seeded in a 96-well plate at a density of 7000 cells per well in 100 μL of Dulbecco's modified Eagle's medium (DMEM) containing 10% FBS and 1000 units/mL penicillin/streptomycin. Each plate was incubated in a 37 °C humidified incubator ventilated with 5% CO2 overnight (16 h). The medium was aspirated from each well, and octuplicate wells were treated with medium containing 4 or 6–12 (5 μM) in DMEM medium with 1% DMSO. Plates also contained blank wells with no cells and control wells with DMEM medium containing 1% DMSO (vehicle). After 1 h of incubation at 37 °C, plates were removed from the incubator and the medium was aspirated and replaced with either vehicle or medium containing vinblastine (2.5–5 nM). The plates were then irradiated using a blue LED light source (tirr = 20 min, λirr = 460–470 nm) or left in the dark and incubated for 72 h in a 37 °C humidified incubator ventilated with 5% CO2. After incubation, MTT reagent (10 μL, 5 mg/mL in PBS) was added to each well, and plates were kept at 37 °C and 5% CO2 for 2 h. The medium was aspirated from each well, and DMSO (100 μL) was added. The wells were shaken for 30 min to allow solvation of the formazan crystals. Absorbance at 570 nm was measured in each well. Average absorbance values for the blank wells were subtracted from absorbance values for each sample to eliminate the background. Viability data were obtained by averaging blank-normalized absorbance values for control cells and expressing average absorbance for the treated samples as percent control.</p><!><p>DU-145 human prostate cancer cells were seeded in a 96-well plate at a density of 7000 cells per well in 100 μL of Dulbecco's modified Eagle's medium (DMEM) containing 10% FBS and 1000 units/mL penicillin/streptomycin. Each plate was incubated in a 37 °C humidified incubator ventilated with 5% CO2 overnight (16 h). The medium was aspirated from each well, and quadruplicate wells were treated with medium containing 4, 7, 9, 12 (25 μM–0.5 μM) or vinblastine (10 nM–0.5 nM) in 1% DMSO. Plates also contained blank wells with no cells and control wells with medium containing 1% DMSO. After 1 h of incubation at 37 °C, wells containing 4,7, 9, or 12 were aspirated and replaced with fresh medium. Wells with vinblastine were left alone. Plates were then irradiated with blue light (460–470 nm; 20 min) or left in the dark and incubated for 72 h in a 37 °C humidified incubator ventilated with 5% CO2. After incubation, MTT reagent (10 μL, 5 mg/mL in PBS) was added to each well, and plates were kept at 37 °C and 5% CO2 for 2 h. The medium was aspirated from each well, and DMSO (100 μL) was added. The wells were shaken for 30 min to allow for the solvation of the formazan crystals. Absorbance at 570 nm was measured in each well. Average absorbance values for the blank wells were subtracted from absorbance values for each sample to eliminate the background. Viability data were obtained by averaging normalized absorbance values for untreated cells and expressing absorbance for the treated samples as percent control. EC50 values were determined using Igor Pro graphing software or Compusyn software.</p><!><p>DU-145 human prostate cancer cells were seeded in a 96-well plate at a density of 7000 cells per well in 100 μL of Dulbecco's modified Eagle's medium (DMEM) containing 10% FBS and 1000 units/mL penicillin/streptomycin. Each plate was incubated in a 37 °C humidified incubator ventilated with 5% CO2 overnight (16 h). The medium was aspirated from each well and replaced with treatment medium containing compound 9 (10−1 μM) or vehicle (medium with 1% DMSO). Plates were than incubated for 1 h. After incubation the medium from each well was aspirated and replaced with medium containing vinblastine (10−0.5 nM) or vehicle, resulting in vinblastine alone and 9 alone monotreatments as well as combination treatments at each compound concentration, all in quadruplicate. Plates were then irradiated with blue light (460–470 nm; 20 min). After irradiation the plates were incubated in a 37 °C humidified incubator ventilated with 5% CO2 for 72 h. After incubation, MTT reagent (10 μL, 5 mg/mL in PBS) was added to each well, and plates were kept at 37 °C and 5% CO2 for 2 h. The medium was aspirated from each well, and DMSO (100 μL) was added. The wells were shaken for 30 min to allow for the solvation of the formazan crystals. Absorbance at 570 nm was measured in each well. Average absorbance values for the blank wells were subtracted from absorbance values for each sample to eliminate the background. Viability data were obtained by averaging normalized absorbance values for untreated cells and expressing absorbance for the treated samples as percent effect. Dose and effect data points were then inserted into the Compusyn software, which solved for the EC50 for both the monotreatments and the combination as well as the CI values for each treatment combination (Figure 8).</p>

PubMed Author Manuscript

Reconciling Intermediates in Mechanical Unfolding Experiments With Two-State Protein Folding in Bulk

Most experimentally well-characterized single domain proteins of less than 100 residues have been found to be two-state folders. That is, only two distinct populations can explain both equilibrium and kinetic measurements. Results from single molecule force spectroscopy, where a protein is unfolded by applying a mechanical pulling force to its ends, have largely confirmed this description for proteins found to be two state in ensemble experiments. Recently, however, stable intermediates have been reported in mechanical unfolding experiments on a cold-shock protein previously found to be a prototypical two-state folder. Here, we tackle this discrepancy using free energy landscapes and Markov state models derived from coarse-grained molecular simulations. We show that protein folding intermediates can be selectively stabilized by the pulling force and that the populations of these intermediates vary in a force-dependent manner. Our model qualitatively captures the experimental results and suggests a possible origin of the apparent discrepancy.

reconciling_intermediates_in_mechanical_unfolding_experiments_with_two-state_protein_folding_in_bulk

<p>The development of single molecule technologies over the past two decades has allowed the observation of details of protein free energy landscapes that had been inaccessible to ensemble experiments.1 In single molecule force spectroscopy experiments a pulling force is applied directly to the termini of the protein, resulting in unfolding. Evidence from both experiment and computer simulations indicates that the folding mechanism may change as a function of force, including the population of intermediates2–6 as well as changes in folding transition state (via ϕ-value analysis in experiment).6–12 However, for most proteins which have been shown to be two-state13 from bulk experiments in the absence of force, only a folded and an extended unfolded state become observable in the pulling experiments, confirming the existing description. In other cases, in which more than two stable states were identified in force experiments, the unfolding mechanism was generally consistent with previous equilibrium and kinetic measurements.14 The recent increase in stability of the instruments and the use of different measurement modes such as force clamp15,16 and force ramp17 have further helped to reveal discrepancies between mechanical and ensemble unfolding mechanisms.18,19</p><p>Here we focus in the thermostable protein CspB from Thermotoga maritima, CspTm (see Fig. 1a). For this system, equilibrium fluorescence measurements and stopped flow kinetic experiments have universally indicated two-state folding,20 a finding later confirmed at the single molecule level using FRET.21,22 AFM experiments in the constant speed mode also fit with a two-state description.23 However, recent work has clearly identified multiple metastable states which are populated when pulling the protein from its ends in the constant force mode.24 Intermediates in single molecule force experiments had previously been identified, perhaps the best known example being Titin I27.2–5 What is remarkable in this case is the longevity (milliseconds in the presence of a pulling force) and heterogeneity of the intermediates, considering the small size of CspTm (66 residues), and the fact that no intermediates were resolved in ensemble experiments. To tackle this problem we run molecular simulations of a coarse-grained topology based model of CspTm and derive a force-dependent Markov state model (MSM) that captures the kinetics of the system. In accord with experiment, we find that while in the absence of force the system is a two-state folder by thermodynamic and kinetic criteria, in the presence of a constant biasing force, certain intermediate species are stabilized. The kinetic model we present recapitulates the shift from two-state to multi-state folding upon application of force recently discovered experimentally.</p><p>Using the experimental NMR structure of the protein25 as a starting point (see Fig. 1a) we have used well established methods to build a topology-based coarse-grained (i.e. Gō) model26 (see SI for details). Such models, in which only native contacts are considered attractive, are justified in the context of a funneled folding energy landscape.27 First we check whether this simple model is able to recapitulate the two-state description in the absence of force. We have run simulations without a pulling force at multiple temperatures and combined them using WHAM.28 The projection on the folding coordinate Q (fraction of native contacts)29 of one of these trajectories, corresponding to a temperature where we observe reversible folding (T=310 K) shows how the protein inter-converts in sharp, cooperative, transitions between the folded and unfolded states (Fig. 1b). The calculated heat capacity thermogram and potential of mean force (PMF) on Q both show unequivocal two-state features (see Fig. 1c–d). The former presents a single, well defined, narrow peak as a function of the temperature, with the simulation midpoint temperature located around Tm≃ 315 K (Fig. 1c). This is low compared to the experimental value (Thermotoga maritima grows up to 90°C20) but enough to yield a folding barrier in native conditions. In the same way, from the PMF we can also diagnose two-stateness for the CspTm model: there are two well defined wells, albeit with a small shoulder at an intermediate value of the fraction of native contacts (Q = 0.6). The two-state description is consistent with the simulation work carried out with a similar native-centric coarse-grained model where the denaturation effects from chemicals, instead of temperature, are explicitly accounted for.30</p><p>To analyze the effects of an external force in the system we have run multiple simulations pulling from the protein termini at different forces (see Table S1). The situation that we encounter is quite different from that at zero force. In Fig. 2a we show time series data for the projection on the pulling coordinate (d, the end to end distance) for simulations carried out at different pulling forces and 300 K. While at zero force the protein remains in the folded state but for transient excursions to higher distance conformations (i.e. the shoulder in the PMF from Fig. 1d), at non-zero forces intermediate species are clearly stabilized in the 2–6 nm range. At forces between 10 and 15 pN we see an additional shorter-lived intermediate at ∼7–8 nm (see traces in Fig. S1). Our projected data resembles some of the experimental traces obtained with the AFM at constant force.24 In the experiments, up to 5 different intermediates could be observed, although in the majority of individual pulls the number of intermediates was either zero or just one. Also, considering an approximate value of Lo≃1.3 Å for the native state, three of the experimental intermediates (C1 at ΔL =2.8±0.7 nm, C2 at ΔL =5.3±0.7 nm and C3 at ΔL =8.6±0.9 nm24) have total extensions within the range of d identified for metastable species in our pulling simulations (Fig. 2a).</p><p>The emergence of intermediates can be rationalized from the changes in the folding free energy landscape upon the application of force. Using the umbrella sampling technique we have run simulations at different pulling forces and calculated 2D PMFs for the folding and extension order parameters (Q and d, see Fig. 2b and SI for details). At low forces the dominant free energy barrier separates the unfolded well from a broad native state, with no barrier between regions with intermediate values of the order parameters and the folded state. However at 5–10 pN a free energy barrier emerges between intermediate states that now separate from the native well. At the highest forces used in our simulations (15–20 pN), the free energy well with Q ≃ 0.6 and d ≃5 nm becomes even more stable than the native state. Hence, an external force can tilt the free energy landscape, stabilizing more extended conformations, and modulating the population of the stable forms of the protein and the barriers between them. Extended intermediates which are only transiently visited at zero force become significantly populated as the force is increased.</p><p>In order to better understand the information present in the pulling simulations and define the dominant intermediate species more systematically we integrate the data at different forces in a minimal Markovian kinetic model (i.e. an MSM31–33). In this type of model the dynamics of the system are encapsulated in a master equation, dP(t)/dt=KP(t), where P(t) is a vector with the populations of the states in the model and K is the rate matrix, containing the rates kji for all the microscopic transitions i ⟶ j. To derive this model from the simulations we have built on established procedures,34,35 with the additional development that in this case the MSM was estimated at the different pulling forces (see full details in SI Methods). We have used a contact map-based clustering34,36 to identify a small number of states (N = 4), two of which are intermediates (I1 and I2, see Fig. 2c) plus the unfolded (U) and native (N) states. On the PMFs in Fig. 2b we show the mean values of Q and d of the four states in the MSM. The two intermediates I1 and I2 differ in the fraction of native contacts for β-strand 1 that are broken, with only the N-terminal residues being unfolded in I2, and the whole of strand 1 being unfolded in I1 (see contact maps in Fig. S5). The intermediate states have extensions of 2.8 and 4.8 nm, respectively, which are close to the extensions of the experimental intermediates C1 and C2,24 and agree well with the locations of the free energy minima in the PMF, which were not used in the clustering process. We recolour the time series in Fig. 2a based on the MSM state to show how the assignment procedure captures the important slow dynamics (see also Fig. S4). After assigning the trajectory frames to the MSM states, we calculate the microscopic rate coefficients, kji (from state i to state j), at the different pulling forces using a lifetime based method.33</p><p>The force dependence of the rate coefficients shows the expected behaviour (see Fig. 3a–b), with refolding processes being more sensitive to the pulling force than unfolding processes, due to the larger changes in molecular extension associated with refolding, and the extensible nature of the unfolded state.37 Naturally, in some cases transitions are not observed due to the finite time of the simulations. This is the case, for example, of refolding events at high pulling forces (e.g. U ⟶ I1 or I1⟶ F), where the bias is too strong for the refolding process to be observed. To overcome this difficulty, we fit the calculated rates to the Bell or Dudko-Hummer-Szabo models for barrier crossing processes induced by force38,39 (in the case of kUI1 and kUI2 an additional intrinsic rate was also included to model the variation of folding pathway with pulling force,10 see Supplementary Methods). We find that in all cases the microscopic rate coefficients can be well captured by the functional form of the model. From the fits we can recover a rate matrix at any desired pulling force, Kfit(F), even at forces where some transitions are not observed directly. Although in some cases this requires an extrapolation, it is based on the fit to a physical model.</p><p>Having assembled this model, we can attempt to recover the phenomenology observed in different conditions. We use the solution to the master equation, P(t)=exp(Kfitt)P(0) to propagate the dynamics and relax from a given initial condition, P(0). In native conditions we relax from the unfolded state, mimicking a stopped flow refolding experiment or the average behaviour of many refolding trajectories by quenching the force in the AFM. We find that the system relaxes to the equilibrium distribution with most of the amplitude (99%) corresponding to the slowest exponential phase (see Fig. 3b, top). This slowest mode corresponds to the first eigenvector of the rate matrix (ψ1), which indicates exchange of density between the unfolded state and the folded and intermediate states (i.e. the folding process, see Fig. 3c). This observation is the kinetic counterpart of the dominant barrier separating the unfolded and folded wells, the latter of which here incorporates the MSM states F, I1 and I2 (Fig. 1b). The intermediates equilibrate rapidly with the folded state as dictated by fast processes corresponding to eigenvectors ψ2 and ψ3 (see Fig. 3c, bottom). In ensemble experiments, this situation, involving a single exponential phase, would inequivocally correspond to two-state folding.20</p><p>As we increase the force the model yields a substantially different scenario (see Fig. 3b). Propagating the rate matrix from the folded state we find that, particularly for forces > 10 pN an intermediate state accumulates as the protein unfolds, reaching up to ∼ 80% of the population. When we propagate the rate matrix from the folded configuration, we find that, in a first kinetic phase, the population of the intermediates increases substantially (mostly for I2, as dictated by the eigenvector corresponding to the slowest process), and only later do they interconvert with the unfolded conformation. Again, the eigenvectors help us understand how the slowest process at higher forces involves the interconversion between I1 and U (see ψ1 in Fig. 3) while equilibration between the folded state and the intermediates is faster (ψ2 and ψ3). This explains the long-lived formation of intermediates when unfolding upon application of a pulling force, as seen experimentally.24</p><p>A natural question is why the application of force is able to expose intermediates hidden at equilibrium, and why other perturbations such as chemical denaturants do not have a similar effect: CspTm has always appeared two-state in ensemble kinetics experiments using chemical denaturants to destabilize the protein.20 A possible explanation lies in the relative sensitivity of the intermediates to force or denaturant. Since the effect of denaturant is, to a first approximation, proportional to the solvent-accessible surface area (SASA) of a given state,40 its effect can be estimated from a free energy landscape projected onto (Q, SASA), analogous to the (Q, d) projections for force (Fig. S6). The obvious difference between the two PMFs is the good correlation between SASA and Q for the metastable free energy basins, compared with the situation for d and Q. The reason for that difference is most likely that SASA is more closely related to the global degree of folding, captured by the fraction of native contacts. On the other hand, larger extensions can be sampled by intermediates via only local unfolding (as indeed evident from the structures in Fig. 2). This poor correlation between extension d and Q means that force can selectively stabilize intermediates which have a larger extension. For example intermediate I1 has a similar extension to the unfolded state, and so at low forces is similarly stabilized; as force increases, the softer unfolded state shifts to larger extensions and becomes more stable.</p><p>Recent efforts using highly stable atomic force microscopes are starting to resolve interesting features of energy landscapes that were not observable before, as in the case of the protein CspTm.24 This work has resulted in an apparent discrepancy between conventional (i.e. two-state) descriptions of protein folding and more intricate descriptions where alternative paths10,11,19 and novel intermediate states24 need to be accounted for. Here we have shown how a simple coarse-grained simulation model can explain many of the experimental observations. Using a force-dependent MSM derived from the simulation data in combination with models for force-induced transitions38,39 we find that low pulling forces can stabilize intermediates. A molecular explanation for the stability of the I1 and I2 intermediates may lie in the orientation of the force-bearing beta-sheet relative to the applied force. An analysis of each of the stable states shows that while the hydrogen bonds in the force-bearing sheet are aligned with the force in the case of the native state (corresponding to an unzipping mechanism), for I1 and I2, the corresponding hydrogen bonds are closer to perpendicular to the applied force (a shearing mechanism, see Fig. S7). Since proteins whose force-bearing beta sheets are oriented perpendicular to the force (such as titin I27 or the B1 domain of protein G) are known to be mechanically strong, this may help to explain the relative stability of intermediates I1 and I2. Consistent with this, the force-dependent unfolding rates for the two intermediates I1⟶ U and I2 ⟶ U show an initial insensitivity to the pulling force, one of the signatures of a shearing mechanism,10,11 in which the mechanism only shifts at sufficiently high force from the intrinsic unfolding pathway to that which is accelerated by pulling. As a result, describing the force dependence of these unfolding rates required the sum of a force independent intrinsic rate and a force-dependent rate from a one-dimensional (1D) rate theory to fit the simulation data, while all of the other unfolding rates (including N⟶I1 and N⟶I2) could be described by 1D rate theory alone. At zeo force these intermediate states may be only marginally populated and remain on the folded side of the dominant barrier, hence becoming invisible in the bulk.</p><p>We anticipate that, with further improvements in instruments, intermediates may in future be observed in force experiments on other proteins in future. Evidence of intermediates with populations that can be tuned by the combination of temperature and force has also been reported in simulations using a similar structure-based model for immunoglobulin-like ddFLN.41,42 We have shown that apparent discrepancies between the results of bulk and single-molecule experiments on mechanical unfolding of a small single-domain protein can be reconciled within the context of a coarse-grained folding model. Application of an alternative bias (force, rather than denaturant) helps to reveal more details of the energy landscape by exposing previously hidden intermediate states which are more easily perturbed by the application of a biasing force than by chemical denaturant or other perturbations.</p>

PubMed Author Manuscript

Synthesis of nitrile-bearing quaternary centers via an equilibriumdriven transnitrilation and anion-relay strategy

The efficient preparation of nitrile-containing building blocks is of interest due to their utility as synthetic intermediates and their prevalence in pharmaceuticals. As a result, significant efforts have been made to develop methods to access these motifs which rely on safer and non-toxic sources of CN. Herein, we report that 2methyl-2-phenylpropanenitrile is an efficient, non-toxic, electrophilic CN source for the synthesis of nitrile-bearing quaternary centers via a thermodynamic transnitrilation and anion-relay strategy. This one-pot process leads to nitrile products resulting from the gemdifunctionalization of alkyl lithium reagents.

synthesis_of_nitrile-bearing_quaternary_centers_via_an_equilibriumdriven_transnitrilation_and_anion-

<p>The rapid generation of molecular complexity is a long-standing objective in organic chemistry. [1] One-pot transformations, which avoid the purification of synthetic intermediates, not only accelerate the synthesis of complex molecules but also minimize the amount of waste generated in the process. Complex nitrilecontaining molecules are of interest to the community due to their versatility as synthetic intermediates for the generation of amines, ketones, carboxylic acids and aldehydes. [2] They are also found in numerous pharmaceuticals (Scheme 1a). [3] As a result, a wide variety of methods for the preparation of nitrile-containing building blocks have been developed with recent efforts focusing on the use of safer and non-toxic sources of CN. [4] Of particular interest are reactions that avoid the use or generation of cyanide ions. [5] Herein, we report a method for the generation of all-carbon quaternary centers bearing a nitrile group from secondary alkyl lithiums (or halides) using a one-pot transnitrilation, anion-relay [6] and electrophile-trapping strategy (Scheme 1c). This method uses 2-methyl-2-phenylpropanenitrile as a non-toxic electrophilic CN source, avoiding the use of cyanide salts that are typically involved in the synthesis of alkylnitriles from alkyl halides.</p><p>We were inspired by a recent report by Reeves and coworkers who have demonstrated that dimethylmalononitrile is an efficient, non-toxic, carbon-bound electrophilic CN source for the transnitrilation of aryl Grignard reagents and aryllithiums (Scheme 1b). [5a, b] The reaction is proposed to occur via addition of the organometallic reagent to a nitrile group and retro-Thorpe fragmentation to yield the corresponding aryl nitriles. Given our interest in the synthesis of nitrile-containing molecules, [7] we wondered if a similar strategy could be applied for the gemdifunctionalization of alkyl lithium reagents via an equilibriumdriven transnitrilation and anion-relay process (Scheme 1c). Addition of the organometallic reagent to an appropriately functionalized electrophilic CN source would yield lithium imine A, which fragments to generate nitrile intermediate B along with tertiary organolithium intermediate C. [8] It should be noted that this transnitrilation process is under thermodynamic control (i.e. reversible) and that tuning the basicity of the leaving group in the electrophilic CN source is crucial for pushing the equilibrium towards transnitrilated organolithium intermediate D, which can be trapped with an electrophile to generate the gemdifunctionalized product. With a more reactive transnitrilation reagent such as dimethylmalononitrile, [5b] the leaving group is not basic enough for complete conversion to D, and subsequent electrophile-trapping leads to mixtures of products.</p><p>Scheme 1. Transnitrilation strategies for the synthesis of pharmaceuticallyrelevant nitrile-containing building blocks.</p><p>We initiated our study by evaluating the transnitrilation of s-BuLi (1a) with various electrophilic nitrile sources (2). Benzyl bromide was chosen as a terminal electrophile to trap the anion resulting from fragmentation of the lithium imine intermediate (Scheme 2). We began by examining a range of nitrile sources (2a-2e) with varying electronic properties. Trimethylacetonitrile 2a is not an efficient transnitrilation reagent since none of the desired product 3a was observed in this transformation. Only the imine, resulting from addition of 1a to 2a, was obtained. We hypothesized that fragmentation of the lithium imine A to release t-BuLi was too challenging and that a transnitrilation reagent containing a better leaving group would be more efficient for this transformation. While dimethylmalononitrile would be efficient for the transnitrilation process, as reported by Reeves and</p><p>Li NC thermodynamic transnitrilation and anion-relay</p><p>coworkers, [5b] the anion resulting from the retro-Thorpe fragmentation would not be basic enough to drive the anion-relay process. With these considerations in mind, we examined the use of benzylic nitrile 2b as an electrophilic CN source in our reaction. Product 3a was obtained in 85% from 2b after a brief optimization of the reaction conditions, which revealed that addition of s-BuLi to the electrophilic nitrile source and subsequent fragmentation of the lithium imine intermediate (A) was efficient in THF at room temperature in 1 hour. Transnitrilation reagents 2c and 2d afforded the desired product but in low yields. In both cases, formation of the lithium imine (A) was efficient, however fragmentation of this intermediate was inefficient for 2d while anion-relay and alkylation by benzyl bromide was low yielding for the reaction using 2c. This demonstrates the need to subtly tune the electronic properties of the transnitrilation reagent to observe the desired thermodynamic gem-difunctionalized product. Finally, the reaction of 1a with 2e led to a complex mixture of products, possibly due to the poor stability of the highly electron-deficient lithium imine intermediate. It is interesting to note that Grignard reagents did not react with 2b at room temperature. High temperatures (refluxing in THF) were required for metal imine formation, but the product resulting from the nitrile transfer process was not obtained.</p><p>Scheme 2. Evaluation of electrophilic CN sources. Yields determined by 1 H NMR using CH2Br2 as an internal standard.</p><p>To better understand the conditions required for fragmentation of the metal imine intermediate, imine 4 was treated with various bases and subsequently treated with benzyl bromide to yield 3a (Table 1). n-BuLi, NaHMDS and KHMDS (entries 1-3) efficiently enabled the formation of 3a in 69%, 65% and 75% yield, respectively, using THF as the solvent mixture. Treating imine 4 with Et2Zn or MeMgBr did not result in fragmentation, and 4 was recovered (entries 4-5). Other ethereal solvents were also examined in the fragmentation of 4 with n-BuLi since the alkyllithium reagents that will be used for this transnitrilation and anion-relay process are generally prepared in Et2O or in Et2O/hydrocarbon solvent mixtures. [9] The imine fragmentation process appears to be highly solvent-dependent. Product 3a was not observed in reactions performed in Et2O or 1,4-dioxane, as well as in other ethereal solvents (entries 6-7 and Table S1). Thus, a solvent switch to THF may be necessary when exploring the reaction scope with respect to various alkyllithium reagents (see Scheme 4). [a] Determined by 1 H NMR using CH2Br2 as an internal standard.</p><p>Having determined that 2b is an optimal reagent for transnitrilation and anion-relay functionalization of alkyllithiums, we next explored the scope of electrophiles that can be used in this transformation. [10] A commercially available solution of s-BuLi (1.4 M in cyclohexane) was used as the model substrate (Scheme 3). Alkyl halides were efficient electrophiles in this transformation, affording products 3a, 3b and 3f in good yields. Carbonyl-based electrophiles, such as benzoyl chloride, N-tosyl imine, an aldehyde and a ketone, provided gem-difunctionalized products 3c, 3d, 3e and 3i, respectively, in good yields. Tertiary epoxide 3j was obtained in a 58% yield using 2-bromoacetophenone as the electrophile. Products 3g and 3h were obtained via nucleophilic aromatic substitution of the corresponding 2-chlorobenzoxazole and 2-chlorobenzothioazole in 75% and 73% yield, respectively. Finally, trapping the transnitrilated organolithium intermediate with methyl benzenesulfinate or phenyl disulfide yielded products 3k and 3l in 61% and 72%, respectively. The scope of this reaction with respect to the alkyllithium reagent is shown in Scheme 4. Starting from alkyl iodides, various substituted organolithium reagents were prepared via lithiumhalogen exchange. [9] Using Knochel's procedure, [11] alkyllithiums were prepared in situ and trapped with the electrophilic CN source 2b. Since the alkyllithium reagents in these reactions are generated in a mixture of Et2O and n-hexane, the resulting lithium imines A do not fragment to initiate the anion-relay process. Removing the solvent and dissolving A in THF leads to fragmentation and deprotonation to generate the transnitrilated organolithium intermediates (D in Scheme 1), which can be trapped with various electrophiles. Various secondary alkyl iodides were successfully converted to the corresponding functionalized nitrile derivatives 6a-6j in moderate to good yields (Scheme 4). Various electrophiles, including aldehydes (6c, 6j) and 2-chlorobenzothioazole (6g), were also used in these reactions, generating diverse products. Functional groups including an acetal (6a, 6h), allyl (6c, 6d), benzyl-protected alcohol (6i) and aryl bromide (6j) provide handles for further product diversification.</p><p>Overall, these reactions lead to the formation of two new C-C bonds and the generation of an all-carbon quaternary center bearing a nitrile functional group. This 4-step, one-pot sequence occurs in good yields from readily available secondary alkyl halide starting materials. Most notably, in comparison with traditional synthetic sequences involving the nucleophilic cyanation of alkyl halides, this transformation avoids the use of hazardous cyanide anions and is not plagued by elimination side-reactions as is often observed with secondary alkyl halides. [12], [13] Using our transnitrilation and anion relay strategy, the direct cyanation and functionalization of iodocyclohexyl derivatives was achieved with good yields. Simple cyclic building block 6a or natural product analogue 6h were obtained in this one-pot protocol in 61% and 60% yield, respectively. The formation of products 6c and 6d was more challenging due to low conversion of the alkyl iodide to the corresponding organolithium during the first step of the reaction. Side-products resulting from elimination and reduction reactions were also observed. [11,14] A slight excess of iodoalkane (2.0 equiv) and t-BuLi relative to 2b was required to obtain products 6c and 6d in 54% and 41%, respectively. The synthesis of nitrile-bearing tertiary centers was achieved using primary alkyl lithium reagents (Scheme 4, 6k-6n). Using n-BuLi (6k), we found that the fragmentation of imine A (where R 1 = C3H7 and R 2 = H) required higher dilution in THF (0.25 M) and a longer time reaction (2 h). [15] Primary alkyl iodides also participated in the one-pot lithium-halogen exchange, transnitrilation, anion-relay and electrophile trapping reaction. Nitriles 6l-6n were obtained in good yields using this sequence. Diverse α-arylnitriles, which are not only versatile synthetic intermediates but are also prevalent in pharmaceuticals, [3] could be prepared using our transnitrilation strategy (Scheme 5). Two strategies for the generation of benzylithium reagents were investigated: carbolithiation of styrene (Scheme 5a) [16] and selective deprotonation of toluene derivatives (Scheme 5b). [17] Using the carbolithiation protocol, three new C-C bonds can be formed in a one-pot process to generate functionalized α-arylnitriles from styrene and 2b as the electrophilic CN source. As a proof of concept, products 8a-8c were prepared in 69-80% using n-BuLi, s-BuLi and t-BuLi as nucleophiles. A solvent switch to THF was required here too to initiate fragmentation of the lithium imine intermediate.</p><p>Selective benzylic deprotonation using a superbase, generated by the addition of t-BuOK to an alkyllithium (n-BuLi or t-BuLi), [17a] was also explored as a means to access α-arylnitriles from toluene derivatives (Scheme 5b). The addition of t-BuOK increases the basicity of the organolithium species, which can efficiently deprotonate benzylic positions. We chose to use this metalation protocol for our one-pot transnitrilation and anion-relay reaction, as the deprotonation could take place in THF, which avoids a solvent switch step in the process. Using this strategy, α-arylnitriles 10a-10e were prepared from very simple starting materials. For example, p-xylene was converted to product 10a in 70% yield, and 10b was obtained selectively from p-butyltoluene in 61% yield. Benzyl cyanide derivatives are generally obtained from the corresponding benzyl halides using nucleophilic cyanide salts. These benzyl halides are typically prepared from the corresponding toluene derivatives via radical halogenation, which can lead to regioselectivity challenges in more substituted derivatives. Using our method, complete selectivity for the transnitrilated products 10a and 10b is observed. Products resulting from dicyanation or regioisomers (in the case of 10b) are not observed. For the synthesis of 10c and 10d, TMPH was used as an additive to generate a potassium/lithium amide. This helps to avoid the formation of side-products resulting from directed ortho metalation by the OMe group. [17b, d] Complete selectivity for deprotonation of the methyl group was observed and the corresponding secondary benzylic nitriles were synthesized in excellent yields. Product 10d is an intermediate in the synthesis of verapamil, a drug on the WHO's List of Essential medicines (Scheme 1a). Finally, using a method reported by Strohmann and coworkers for the direct metalation of sensitive benzyl derivatives such as (2-phenylethyl)-dimethylamine, [17c] product 10e was obtained in 46% yield. While the secondary benzyllithium intermediate in this reaction is known to undergo β-elimination reactions, [17c] this side reaction can be avoided by using a mixture of t-BuOK and t-BuLi in THF at -78 °C.</p><p>In summary, we have developed a one-pot method for the synthesis of various nitrile-containing building blocks based on an equilibrium driven transnitrilation and anion-relay strategy. A broad range of electrophiles and alkyl lithium reagents, generated via lithium-halogen exchange, carbolithiation or benzylic deprotonation, participate in this reaction. The nature of the transnitrilation reagent is crucial for the success of this one-pot gem-difunctionalization process; not only must it readily generate lithium imine adducts (A) under mild conditions, but it must also fragment to generate a base responsible for anion-relay. Other applications of transnitrilation strategies for the synthesis of nitrilecontaining building blocks are currently being explored in our laboratory.</p>

ChemRxiv

Separation of CH4/N2 of Low Concentrations From Coal Bed Gas by Sodium-Modified Clinoptilolite

Clinoptilolite is a widely distributed tectosilicate, mainly composed of Al2O3, SiO2 with exchangeable cations such as Ca, K, Mg, and Na. In this research, raw clinoptilolite was ground, gravimetrically concentrated and ion-exchanged using different concentrations of NaCl solution. Then the modified clinoptilolite powder was formulated into particles as adsorbents. The adsorbents were applied to CH4 separation in coal bed gas. The raw and modified clinoptilolites were characterized by X-ray diffraction (XRD), scanning electron microscope (SEM), transmission electron microscope (TEM), atomic emission spectrometer (ICP-AES), Fourier transform infrared spectrometer (FTIR), and Brunauer Emmett Teller (BET) specific surface area. The CH4 absorptivity by raw and modified clinoptilolites was evaluated using pressure swing adsorption (PSA) to assess the CH4 separation ability. The results indicated that the ion-exchanged clinoptilolite using 0.2 mol/L NaCl solution was found to be promising for the kinetic PSA separation of CH4/N2, giving a better absorptivity for CH4 separation under different influence factors. Based on the simulated static experiments, it was indicated that both CH4 and N2 were capable of diffusing into clinoptilolite while N2 adsorption by clinoptilolite was excellent. The experiment results also indicated that ion-exchanged clinoptilolite using a 0.2 mol/L NaCl solution was the optimal adsorbent for separating CH4/N2 at the low pressure condition. From the simulated dynamic experiments, the ion-exchanged clinoptilolite using a 0.2 mol/L NaCl solution as a potential sorbent in kinetic PSA processes for N2/CH4 separation, exhibited the best performance at 648 K under 0.2 MPa within 28 min, in comparison to the raw clinoptilolite and clinoptilolite under other modification conditions. In the next phase of research, the modified clinoptilolite will be tested for CH4 separation in real coal bed gas.

separation_of_ch4/n2_of_low_concentrations_from_coal_bed_gas_by_sodium-modified_clinoptilolite

Introduction<!>Clinoptilolite and All Agents<!>Clinoptilolite Pretreatment<!>Preparation of Modified Clinoptilolite<!>Separation Experiment With CH4/N2<!><!>Pressure/Vacuum Swing Adsorption Experimental Program<!>Analytical Methods<!><!>Analytical Methods<!>Characterization of Clinoptilolites<!><!>Characterization of Clinoptilolites<!><!>Characterization of Clinoptilolites<!><!>Characterization of Clinoptilolites<!><!>Characterization of Clinoptilolites<!><!>Simulated Experiments<!><!>Simulated Experiments<!><!>Simulated Experiments<!>Conclusion<!>Author Contributions<!>Conflict of Interest Statement<!>

<p>The coalbed gas is found in coal bed with a main composition of methane (CH4), which was absorbed on the surface of the coal particles. Part of coal bed gas was dissociated or dissolved in the hydrocarbon gas in the coal pore and the water of coal bed, which is automatically stored up in the coal bed as the powerful complement to raw gas. CH4 in the coal bed is a high quality gas fuel. Meanwhile, it is also one of detrimental gases influencing mining underground coal and an important harmful source leading to atmospheric greenhouse effect. In China, there is up to 13 billion m3 CH4 under the process of coal mine a year, which accounts for around one third of its emissions globally. On the other hand, the utilization ratio of CH4 in the coal bed gas was only 35%, resulting in a huge CH4 resource loss. We know that the greenhouse effect of CH4 is 21 times to the CO2 and power of CH4 for damaging ozone (O3) is 7 times to the CO2. Thus, recycling coal bed gas is of great significance on both energy development and environmental protection. With improving consciousness of human on the coal mine safety and environmental protection, the exploitation of CH4 in the coal bed has been attached great importance to the world in recent years.</p><p>The separation technology of CH4 in the coal bed is not effective, which is one of the main reasons for low the utilization ratio. In the separation process of low concentration of coal bed gas, the physicochemical property of N2 and CH4 was similar (Perry et al., 1999; Johnson III, 2015). It led the recycle and separation technology to be a key common technology challenge. It was also one of the most important technological obstacles on gas development, energy saving and emission reduction (Bomberger et al., 1999; Cavenati et al., 2006; Tagliabue et al., 2009). At present, the common technologies were cryogenic distillation, pressure swing adsorption (PSA), membrane separation, hydration technology and dissolution-absorption technology. The PSA separation method has become the mainstream technology for the purification of coal bed gas at small and medium scales, due to its advantages of low energy consumption, less investment equipment and high degree of automation (Arya et al., 2014; Yin et al., 2015). Its key challenge is the selection of adsorbents. The main adsorbents currently used are activated carbon (AC) (Zhou et al., 2002; Gu et al., 2015; Gao et al., 2017), carbon molecular sieve (CMS) (Fatehi et al., 1995; Cavenati et al., 2005; Grande et al., 2005), natural clinoptilolite (Aguilar-Armenta et al., 2001; Jayaraman et al., 2004, 2005), titanium silicon molecular sieve (Aguilar-Armenta et al., 2001; Jayaraman et al., 2004, 2005; Faghihian et al., 2008). The equilibrium adsorption capacity of CH4 is higher than that of N2 for AC. Although the separation coefficient is higher and the effect is better based on results from laboratory studies, it is still far away from industrial application. The main reason is that the preparation process of AC is complex and the cost is relatively high. And it obtains CH4 product in vacuum desorption stage, the subsequent operation needs to be compressed, so the power cost is increased, and the economic effect is not obvious. The separation of CH4 and N2 by CMS is based on the kinetic effect. The diffusion rate of N2 in the micropore is higher than CH4. A large amount of N2 is adsorbed into the pore and CH4 remains outside the pore in a relatively short time. Therefore, the product CH4 is obtained by the adsorption or sequestration of the PSA process, instead of the vacuum step. With the increase of adsorption time, the kinetically the effect becomes weaker, and the equilibrium effect will dominate, making CH4 and N2 separation difficult. Although CMS has achieved good results in the laboratory, it is mainly aimed at the coal bed gas with high concentration of CH4 (CH4 content >70%). However, there are few reports on the study of coal bed gas with low concentrations. The natural clinoptilolite as a kind of PSA adsorbents is of great potential for application with the advantage of acid resistance, heat resistance, alkali resistance, stable structure, rich resources, and low price. They can show both equilibrium and kinetic effects. However, the adsorbent prepared from natural clinoptilolite are of different sodium contents and its application in low concentration coal bed gas (CH4 < 30%) has not been reported.</p><p>Here we present a study of the adsorption isotherms of four adsorbents made from natural clinoptilolite with different sodium contents at 298 K. And the corresponding adsorption kinetics were measured at the same and different pressure using the feed gas containing 20% CH4 and 80% N2 at 298 K. This study will provide technical support for the implementation of industrialization.</p><!><p>The raw clinoptilolite was collected from the south of the Liaoxi metallogenic belt in China. The adsorbents used were CH4 (99.95%), N2 (99.95%). The purging gas for adsorbent activation/regeneration was He (99.999%, pre-purified). All gases were provided by Praxair. The reagents, including NaCl, used in this research were of analytically pure and bought from Sinopharm Group Chemical Reagent Co., Ltd.</p><!><p>The raw clinoptilolite was ground by a ball grinder to a granular size <70 μm. Then the milled pulp with clinoptilolite powders was poured into a Falcon centrifuge to remove some heavy impurities. The purified clinoptilolite was dried at 105°C and stored in a desiccator. It was used as a raw material for the preparation of adsorbents.</p><!><p>The processed clinoptilolite powders were mixed NaCl solutions at concentrations of 0.1, 0.2, 0.4, and 0.6 mol/L at a solid to liquid ratio of 1:20 for 2.5 h in Erlenmeyer flasks separately and covered with sealing films and maintained in a 90°C water bath. The mixture was centrifuged to separate the solids, then washed using deionized water until no Cl−. All ion-exchanged clinoptilolite samples were pressed into a round cake and calcined at 200°C (to dry the samples) for 2 h. And then they were crushed and sieved. Particles of 0.5–1.5 mm sizes were used as adsorbents.</p><!><p>The gas mixture of CH4 and N2 was prepared by high-pure standard gas, and the ratio of CH4/N2 was 20/80%. The experimental device for adsorption was a single absorbing tower filled with raw and modified clinoptilolites (Figure 1). At first, the device was pressurized using high-pure standard He until the adsorption pressure was up to the setting pressure. Then the intake valve of He was closed and the intake valve of mixture gas was opened (it was the start time of data recording). In order to keep the pressure of absorbing tower reaching the experimental value, it was adjusted by control valves (the flow value of gas was set to 50 mL/min). The outlet discharge was set using mass flow controller before the test. In the process of absorption, the change of concentration of CH4 was tested and recorded by a gas analyzer. The test was continued until the concentration of CH4 was the same as the initial concentration of CH4 in the mixture. The activation and regeneration of modified clinoptilolite was not begun until inverse vacuum was pumped for 10 min.</p><!><p>The dynamic test device. 1. He cylinder; 2. Mixture cylinder of CH4/N2; 3, 4. Pressure reducing valve; 5, 6, 7, 8, 9 and 12. Needle-type valve; 10. Absorbing tower(16mm*320mm); 11. Pressure probe; 13. Mass flow controller; 14. Gas analyzer; 15. Vacuum pump; 16. Windows control center (Copyright©1994–1997 SIEMENS, AG).</p><!><p>A vacuum pressure swing adsorption (VPSA) cycle was devised to be experimented in the pilot-scale unit, which was the purpose of catching CH4 from a dynamical mix of CH4/N2 simulated coal bed gas. The device, single-tower adsorption layer, was shown in Figure 1. Before test, the absorbents were modified clinoptilolites which were activated in the vacuum rotation activation furnace for 6 h in 648 K. When the test began, the pressure of He and pressure of adsorption were both set at the certain pressure, and flow value of tower top was 60 mL/min. It needed to be emphasized that vacuum pumping treatment using vacuum pumps was initiated before the experiments.</p><!><p>The morphologies of purified clinoptilolite and modified clinoptilolite were observed with TEM (Tecnai G2 TF30) and SEM (Hitachi S-4800).</p><p>X-ray diffraction (Rigaku MiniFlex600) measurements were proceeded with copper CuKα1 radiation (λ = 1.5406 Å), utilizing a voltage of 40 kV and a current of 15 mA. The divergence slit was 0.3 mm and data was gathered for 2θ scanned from 3° to 80° at 10°/min.</p><p>The chemical components of purified clinoptilolite and modified clinoptilolite samples were analyzed using an inductively coupled plasma atomic emission spectroscopy (ICP-AES, ICAP7400 THERMO Fisher). To obtain a more representative chemical composition of a sample, the analysis was done in triplicates per sample, and the element contents were averaged (Table 1).</p><!><p>The chemical composition of clinoptilolites (wt%).</p><!><p>Mid-infrared spectra were recorded using a Fourier transform infrared (FT-IR) spectrometer (Nicolet IS50) with a Smart Endurance™ single bounce diamond ATR cell. Spectra were acquired of 4,000–400 cm−1 by the average of 64 scans with a resolution ratio of 4 cm−1. A mirror speed of 0.6 cm/s was used.</p><p>The BET surface area was 57.84 ± 0.20 m2/g measured by an ASAP 2020 instrument (Micromeritics, USA). N2 (at 298 K) and CH4 (at 298 K) adsorption were measured to determine the BET surface area and micropore size distribution.</p><!><p>Figure 2 showed SEM photographs of unmodified and modified clinoptilolites, They didn't change significantly among them. In Figure 2, C-0, unmodified clinoptilolite, was purified clinoptilolite with no NaCl treatment. The modified clinoptilolites were depicted by C-1, C-2, C-3, and C-4, which were treated by 0.1, 0.2, 0.4, and 0.6 mol/L NaCl solution, respectively. These crystals are flaggy or schistose, which appears as parallel conjunctive aggregate. Individual crystals span is from hundreds of nanometers to several microns. A number of ordered small particles were also found on the crystals from Figure 2a*.</p><!><p>SEM photographs of clinoptilolites: (a,a*). C-0, (b). C-1, (c). C-2, (d). C-3, (e). C-4.</p><!><p>In order to observe the internal microstructure of samples, TEM photographs of modified clinoptilolites were shown in Figure 3. From Figure 3, inside of C-0 was large lamella stacking. However, inside of C-1, C-2, C-3, and C-4 were stacked with small lamella, in which the number of fissures produced by flake particles. This is because they are stirred on thin sheets during ion exchange, so more lamellar clinoptilolite is stripped, mainly concentrated at 100 × 300 nm, which also makes the exchange easier. The 131 faces of clinoptilolite can be seen in C-2 of photograph c*, whose crystal plane spacing is 3.98 Å, which have not seen in other published papers.</p><!><p>TEM photographs of clinoptilolites: (a). C-0, (b). C-1, (c,c*). C-2, (d). C-3, (e). C-4.</p><!><p>ICP-AES was adopted to analyze the ion change before and after modification (table 1). It is obviously that with the increasing of nacl concentrations, the Na+ component increased from 0.65 to 3.68 wt% gradually. Meanwhile, the Ca2+ component decreased from 3.8 to 1.89 wt% apparently. In addition, other elements showed a random change with increased nacl concentrations (such as Si, Al, K, Mg, Fe, and Ti). The phenomenon indicated that the ion exchange of Na+ for Ca2+ were carried out in clinoptilolite samples with NaCl solutions bath.</p><p>The XRD patterns were mainly used for confirmation the difference in all clinoptilolites without and with treatment using NaCl solution (Figure 4). The major mineral in all samples was clinoptilolite with minor amounts of quartz. All the lattice parameters were not observably influenced by the different contents of NaCl. We couldn't see the characteristic peak of NaCl in modified clinoptilolites, It showed that the Na+ entered the lattice, and we combined the chemical composition analysis in Table 2. So we know it was replaced with Ca2+. The obvious difference was the D value of the characteristic peak was smaller than modified clinoptilolites. It was because the exchange of different ions leaded to the change of lattice spacing.</p><!><p>XRD of clinoptilolites.</p><p>BET of modified clinoptilolites treated by NaCl of different concentrations with N2 at 77 K.</p><!><p>The broad band at 620 cm−1 in the clinoptilolites spectra were attributed to stretching vibrations related to Si-O tetrahedron structure (Figure 5). The characteristic peak of clinoptilolite without ion-exchange was weak. Furthermore, the vibration absorption peak of Si-O-Si and Al-O-Si appears at 795 cm−1. Compared with clinoptilolite without ion-exchange, some evident changes in characteristic absorption peaks were discovered. The characteristic peak at 985.46 cm−1 deriving from Al-O vibrations shifted to 1043.32 cm−1, caused a small amount of Al-O losing after ion-exchange. A small account of non-framework Al stuck in the unit cell was shifted by ion-exchange. And a small account of Al was shifted from framework because of Al-O-Si hydrolyzing in the clinoptilolites. Then cavities were formed leading to pore volume increasing. For modified clinoptilolites, the Si-O and/or Al-O out-of-plane bend occurred at 485 and 1,622 cm−1. Thus, the intensity of Al-O out-of-plane bend in modified clinoptilolites was stronger than that of clinoptilolite. It was because ion-exchange caused a small amount of Al-O losing. The broad band at 2,328 cm−1 mainly resulted from stretching vibrations of OH− groups on modified clinoptilolites after ion-exchange. The broad band at 3,743 cm−1 mainly resulted from stretching vibrations of O-H on modified clinoptilolites after ion-exchange.</p><!><p>FT-IR spectra of modified clinoptilolites.</p><!><p>The N2 adsorption-desorption isotherms at 77 K with the four adsorbents were shown in Figure 6A. According to the IUPAC classification, all curves were identified as type IV. The surface has mesopore and macropore, The curve of p/p0 region of low relative pressure is convex up, in the higher p/p0 region, the adsorbed material is condensed by capillary, the isotherm obtained by desorption does not coincide with the isotherm obtained by adsorption, and the desorption isotherm lags over the adsorption isotherm. So they present a hysteresis loop. It belongs to the class D loop. It is mainly due to the slit holes formed by sloping sheet stacking. In Figure 6B, we know most of the pores are between 5 and 80 nanometers, They have different pore structure by loading different contents of sodium ions. According to Table 2, we know that the internal surface area of C-2 adsorbents is the largest. With the increase of the loading of sodium ions, the micropore surface area of adsorbent increased first and then decreased. So the micropores are adjusted. At the beginning, the calcium ion was replaced by sodium ion, and the pore channel of clinoptilolite became smaller. When loading a certain amount, a large number of sodium ions loaded the surface of clinoptilolite and blocked some channels.</p><!><p>(A) N2 adsorption-desorption isotherms at 77 K; (B) pore size distributions.</p><!><p>To explore the separation capability of CH4/N2 on the modified clinoptilolites, the simulated static experiments were conducted using BET equipment, and the simulated dynamic experiments were conducted using the device of adsorption of single tower (Figure 1), in which pure N2 and CH4 gas were chosen as gas-supply.</p><p>CH4 adsorption and N2 adsorption of the simulated static experiments were shown in Figure 7. It should be noted that adsorption of the two molecules is competitive and thus the gas-supply in the simulated static experiments is high purity N2 and high purity CH4 at 298 K, separately. Before the adsorption experiment begun, The absorbents were vacuum activated for 8 h at 370°. From Figure 7, there was obvious difference between quantity adsorbed of CH4 and N2 using clinoptilolites. At the same relative pressure condition, quantity adsorbed of N2 on clinoptilolites was much more than that of CH4. From Figure 7A, the quantity adsorbed of N2 using clinoptilolites followed: C-3 = C-2> C-4> C-1> C-0. From Figure 7B, the quantity adsorbed of CH4 using clinoptilolites was as follows: C-3> C-4> C-0> C-1> C-2. Considering the contradiction between adsorption capability of clinoptilolites for CH4 and N2, it is obvious that the C-2 adsorbents static equilibrium separation coefficient is the larger than other three. It is the most potential adsorbent for separating CH4/N2.</p><!><p>(a) N2 adsorption of the simulated static experiments at 298 K; (b) CH4 adsorption of the simulated static experiments at 298 K.</p><!><p>Dynamic experiments had been done. The CH4 volume concentration of product is obtained at 298 K on certain pressure when the feed gas is a mixture of CH4 (20%) and N2 gas (80%), as shown in Figure 8. Concentrated CH4 could be obtained directly by using these adsorbents of C-1, C-2, C-3, C-4. In this experiment, before testing, the adsorbent was vacuum activated for 8 h at 648 K, and then the package was sealed for use. When the concentration of CH4 of the top of the tower is 20%, it is put back to normal pressure, and a 30 min vacuum is activated and regenerated. The experiment was repeated three times, and the data are recorded in the third experiment. Figure 8A showed the different adsorbents breakthrough curve of nitrogen adsorption at 298 K on 0.2 Mpa. In the dynamic adsorption curve, it showed that CH4 concentration can be increased.C-1 is from 20 to 65.2%, C-2 is from 20 to 70.0%, C-3 is from 20 to 66.1%, C-4 is from 20 to 63.1%. Moreover, they can be continuously regenerated. So the adsorbent of C-2 is the best among these adsorbents, which is consistent with the static adsorption results. Figure 8B shows the C-2 breakthrough curve of nitrogen adsorption at 298 K on the different pressure. We can control residence time of raw gas in adsorbent by adjusting the pressure of carrier He gas. The residence time corresponding to 0.1, 0.2, 0.3, and 0.5 Mpa is 5, 7.5, 11, and 14 min, respectively. The peak value of CH4 reaches 70.0% when the residence time is 7.5 min. The results show that the N2 adsorption was bigger than CH4 when the mixture of CH4/N2 were in the absorbing tower. The bigger the pressure, the longer the residence time. CH4 adsorption was bigger than N2 adsorption when they were adsorbed. The separation factors would decrease. So, It is very important to choose the suitable residence time.</p><!><p>(A) The different adsorbents breakthrough curve of nitrogen adsorption at 298 K on 0.2 Mpa, (B) The C-2 breakthrough curve of nitrogen adsorption at 298 K on 0.1, 0.2, 0.3, 0.5 Mpa.</p><!><p>From the above, it seemed that the C-2 shows the greatest performance at 648 K under 0.2 MPa within 50 min, in comparison to the other modified clinoptilolites, as a underlying sorbent in kinetic PSA processes for the N2/CH4 separation. The different concentrations of Na+ that were existed in its porous network as well as their distribution were the primary influence factor that specifies the adsorption and kinetic properties of the clinoptilolites. Thus, the ion-exchange with differences in the concentration of Na+ disturbed the Na+ distribution as well as the electrostatic field inside the clinoptilolite's pores affecting the adsorption property.</p><!><p>The material structure and CH4/N2 adsorbability of raw and Na+ ion-exchanged clinoptilolites have been examined in detail using PSA. The effect of adsorbent prepared by clinoptilolite with different sodium ion content on methane nitrogen separation is very different. The clinoptilolite adsorbents can be adjusted for their pore channel by controlling the loaded amount of sodium ions.</p><p>The C-2 adsorbent prepared using 0.2 mol/L NaCl solutions was the most promising for the kinetic PSA separation of CH4/N2, giving the better adsorptivity and influence factors concerning the CH4 separation.</p><p>From the simulated static experiments, it indicated that N2 and CH4 are both competent in diffusing into the clinoptilolites while N2 adsorptions of clinoptilolites are more excellent. The pertinent results also indicated that adsorption capability of ion-exchanged clinoptilolite using 0.2 mol/L NaCl solutions was the optimal adsorbent for separating CH4/N2 at low pressure, considering the contradiction between adsorption capability of clinoptilolites for CH4 and N2.</p><p>According to the simulated dynamic experiments, the ion-exchanged clinoptilolite using 0.2 mol/L NaCl solutions exhibits the best performance at 648 K under 0.2 MPa within 50 min, in comparison to raw and other modified clinoptilolites, as a underlying sorbent in kinetic PSA processes for the N2/CH4 separation.</p><p>The ion-exchange with differences in the concentration of Na+ as well as the electrostatic field inside the clinoptilolite's pores affecting the adsorption property. Finally, further manipulation for CH4 separation of the clinoptilolite is underway with coal bed gas as gas-supply.</p><!><p>XH and ZL conceived and designed the project. XH performed the experiments and wrote the manuscript. HH, XL, and YH analyzed the data.</p><!><p>The authors declare that the research was conducted in the absence of any commercial or financial relationships that could be construed as a potential conflict of interest.</p><!><p>Funding. The work was financially supported by the Special Fund for National/International Science and Technology Cooperation in China (Grant no. 2015DFR60640) and the Science and Technology Plan of Henan Province (No. 182102310031).</p>

PubMed Open Access

Detecting intracellular translocation of native proteins quantitatively at the single cell level

The intracellular localization and movement (i.e. translocation) of proteins are critically correlated with the functions and activation states of these proteins. Simple and accessible detection methods that can rapidly screen a large cell population with single cell resolution have been seriously lacking. In this report, we demonstrate a simple protocol for detecting translocation of native proteins using a common flow cytometer which detects fluorescence intensity without imaging. We sequentially conducted chemical release of cytosolic proteins and fluorescence immunostaining of a targeted protein. The detected fluorescence intensity of cells was shown to be quantitatively correlated to the cytosolic/nuclear localization of the protein. We used our approach to detect the translocation of native NF-kB (an important transcription factor) at its native expression level and examine the temporal dynamics in the process. The incorporation of fluorescence immunostaining makes our approach compatible with the analysis of cell samples from lab animals and patients. Our method will dramatically lower the technological hurdle for studying subcellular localization of proteins.

detecting_intracellular_translocation_of_native_proteins_quantitatively_at_the_single_cell_level

Introduction<!>Results and discussion<!>Cell sample preparation<!>Standard uorescence immunostaining<!>Selective release of cytosolic proteins followed by uorescence immunostaining<!>Fluorescence microscopy<!>Western blotting<!>Flow cytometry analysis

<p>Within eukaryotic cells, proteins efficiently and selectively transit between functionally distinct subcellular compartments including plasma membrane, cytosol, nucleus and other membrane-enclosed organelles. The subcellular localization of an intracellular protein or the change of it (i.e. intracellular translocation) is highly signicant for several reasons. First, intracellular translocation can be a prerequisite for proteins to carry out their intended functions. For example, a transcription factor needs to move from the cytosol into the nucleus in order to regulate gene transcription and such events occur typically as a consequence of outside stimuli to the cell. Second, since translocation is oen associated with modication and activation at the molecular level (e.g. phosphorylation and methylation), the subcellular localization of the protein molecule is oen indicative of its state. In most cases, the proteins are only active at their intended subcellular location. Finally, subcellular mislocalization of proteins leads to diseases ranging from metabolic disorders to cancers. 1,2 Mislocalization of Akt, NF-kB, FOXO, p27, and p53 have been well-documented as key features in a variety of cancers. 2,3 Modulation of protein translocation is practiced as an important therapeutic approach for cancer treatment. 1,2 The subcellular location of a target protein can also serve as a useful read-out for high-content screening of cancer drugs. 2 Conventionally intracellular translocations such as nucleocytoplasmic transport (i.e. the translocation between the nucleus and the cytosol 4,5 ) have been evaluated using uorescence microscopy or subcellular fractionation. [6][7][8][9][10][11][12][13] However, there are important limitations with these approaches. Fluorescence microscopy (including total internal reection uorescence microscopy, or TIRFM 13 ) typically analyzes a limited number of cells and does not provide information on the distribution of the cell population. Subcellular fractionation involves lysis and homogenization of cells and then separation of the materials from various subcellular compartments by centrifugation. The data obtained by subcellular fractionation reect only the average properties of the cell population without revealing the heterogeneity that is oen critically involved in cell signaling networks. [14][15][16][17][18][19] For example, when cells show an all-ornone response to a particular stimulus (bistability), only a subset of cells respond to the signal. 15,[20][21][22][23] Thus high-throughput methods are desired to generate information on the translocation of a large number of cells with single cell resolution.</p><p>Laser scanning cytometry (LSC) 24,25 and imaging ow cytometry 26 which permit rapid acquisition of uorescence images of solid phase or owing cells, have been used to detect the subcellular localization of intracellular proteins. However, these instruments are very expensive and typically have only limited accessibility through large central facilities. More importantly, these imaging-based instruments require sufficient exposure time in order to generate enough spatial resolution for the analysis. This determines that their throughputs are no more than several hundred cells per second (compared to 10 4 -10 5 cells per second by a conventional ow cytometer which detects only uorescence intensity) and the complex algorithms used for image analysis oen introduce errors and bias. [27][28][29] In our previous work, we developed electroporative ow cytometry to examine protein translocations by adding electroporationbased protein release and ow cytometric screening. 30,31 Unfortunately, electroporative ow cytometry requires special apparatus for combing electroporation and laser-induced uorescence detection. More importantly, the approach requires proteins of interest tagged by uorescent protein markers and does not allow examination of native proteins and primary cells isolated from animals and patients. We also explored using TIRF-based ow cytometry for probing protein translocation. 32 However, its potential for high resolution recognition of nucleocytoplasmic translocation was limited.</p><p>In this work, we demonstrate a simple method that combines selective chemical release of cytosolic proteins and standard uorescence immunostaining for detecting the translocation of native proteins at the native expression level with single cell resolution. We demonstrate the proof-of-principle by detecting nucleocytoplasmic transport of an important transcription factor NF-kB. NF-kB undergoes nucleocytoplasmic transport from the cytosol to the nucleus in order to regulate transcription and gene expression, upon extracellular stimuli (e.g. by TNFa). 33,34 Briey, we used saponin (i.e. a class of amphipathic glycosides) to selectively release cytosolic fraction of intracellular proteins by dissolving the cholesterol content and permeabilizing the plasma membrane. 35 Such treatment was followed by uorescence immunostaining of residual NF-kB. The cell population was then screened by a common ow cytometer for uorescence intensity of each cell. We showed that the uorescence intensity of a cell could be correlated to the subcellular localization of the protein. Taking advantage of common ow cytometry which is widely accessible, our approach detects the translocation of native proteins without imaging and with high throughput. Our approach is readily compatible with analysis of primary samples from animals and patients.</p><!><p>In Fig. 1, we outline the procedure and working principle of our protocol (i.e. selective release and immunostaining), in comparison to those of the standard immunostaining protocol, and how these procedures generate different results for single cell screening with standard ow cytometry.</p><p>In our experiments, we applied TNFa to stimulate cells and produce NF-kB translocation from the cytosol to the nucleus. With standard uorescence immunostaining, HeLa cells with NF-kB translocation (i.e. +TNFa) and those without NF-kB translocation (i.e. untreated) are crosslinked by paraformaldehyde and permeabilized by Triton X-100 before all intracellular NF-kB is uorescently labeled by antibodies that specically target NF-kB. The crosslinking by paraformaldehyde conserves all proteins in the cells and the permeabilization ensures the full access of the targeted protein (i.e. NF-kB in this study) by the antibodies for labeling. When these cells are subsequently screened by ow cytometry, which detects uorescence from the entire cell, the translocation does not create difference in the uorescence intensity detected because it only varies the subcellular localization of the protein, not the overall expression level (Fig. 1A). In comparison, in our method, we add a selective release step before immunostaining. During this step, saponin is used to treat the cells to dissolve cholesterol in the plasma membrane and make the membrane leaky. [35][36][37] Saponin is known to primarily permeabilize the plasma membrane while keeping the cholesterol-poor internal membranes (e.g. the mitochondrial membrane and nuclear envelope) intact. 35,38 Gentle treatment by saponin has also been shown to have minimal effects on cellular functions such as protein synthesis. 39,40 Cytosolic proteins are preferentially released out of the cells and the nucleic proteins are largely unaffected in their amounts. Thus cells with NF-kB translocation (with NF-kB mostly in the nucleus) have far less decrease in the NF-kB amount due to the release, compared to the cells without NF-kB translocation (with NF-kB primarily in the cytosol). The selective release step is immediately followed by immunostaining that xes the cells and labels all the remaining intracellular NF-kB. The ow cytometry results now are signicantly different for cells with translocation and those without translocation, with the latter showing much smaller uorescence intensity. Thus we are able to link the subcellular localization of the protein with the detected uorescence intensity of a cell treated by our protocol.</p><p>Fig. 2 shows the ow cytometry data obtained aer standard immunostaining (Fig. 2A and B) and combined selective release and immunostaining (Fig. 2C and D).</p><p>Fig. 2A shows that as expected, with standard immunostaining the uorescence intensity histogram generated by a cell population without NF-kB translocation overlaps with that generated by a cell population with the translocation (stimulated by TNFa for 30 min). In comparison, the two cell populations exhibit marked difference in the uorescence intensity histogram when we had selective release of cytosolic proteins by saponin (0.05% saponin for 10 min) before the immunostaining (Fig. 2C). Furthermore, we also discovered that the two cell populations (processed with the selective release protocol) were even more distinct when two-dimensional dot plots were used to include information on the cell size (via detecting the forward scatter signal) (Fig. 2D). The two cell populations were entirely separated from each another in Fig. 2D. This improvement is attributed to the differentiation of large cells without translocation (NF-kB in the cytosol) and small cells with translocation (NF-kB in the nucleus). These two subpopulations may have similar uorescence intensities aer the selective release protocol but are very different in the cell size. The uorescence images of cells aer standard immunostaining and selective release/immunostaining were also collected (ESI Fig. S1 †). The images conrm the proposed mechanism in Fig. 1. Our technique renders the uorescence intensity different for cells with translocation and those without translocation, whereas standard immunostaining reveals the different localizations of the protein via imaging with the overall uorescence intensity from whole cells being the same for the two types of cells.</p><p>We also optimized the selective release protocol in order to create the maximum differentiation of cells based on NF-kB subcellular localization. We varied the concentration and duration of saponin treatment for two cell populations (untreated cells and cells stimulated by 50 ng ml À1 TNFa for 30 min) and observed the difference in their uorescence intensity histograms (ESI Fig. S2 †). Fig. S2A † shows that with 10 min treatment time, the saponin concentration of 0.05% yielded the best separation between the two cell populations. The decreased differentiation at higher saponin concentrations (0.2-0.5%) presumably resulted from the release of both cytosolic and nucleic proteins. Fig. S2B † shows that with a xed concentration of 0.05% for saponin, the optimal treatment time was in between 1 and 10 min. Exceedingly long treatment times (>10 min) also led to decreased separation, due to release of both cytosolic and nucleic proteins. Thus we determine that saponin treatment of 0.05% concentration and 1-10 min works the best as the selective release step for differentiation of cells with NF-kB in the nucleus and those with the same protein in the cytosol.</p><p>Using the optimized selective release protocol (in combination with immunostaining), we show that our method is effective for revealing various degrees of NF-kB translocation in the cell population. In Fig. 3, various concentrations of TNFa were used to stimulate the cell population for 30 min. With low concentrations (0.1-1.0 ng ml À1 ) of TNFa, the cell populations appear to have lower degree of NF-kB translocation overall and a subpopulation of cells have no translocation at all (based on the broad peak shape which is not Gaussian).</p><p>With high concentrations of TNFa (10-100 ng ml À1 ), the translocation occurs more completely for the cell population and the separation between the stimulated population and the control is increasingly complete.</p><p>Finally, we used our approach to examine temporal dynamics in the cell population during NF-kB translocation. In Fig. 4A, the shi in the uorescence intensity histogram suggests the movement of NF-kB from the cytosol to the nucleus aer TNFa stimulation. The increase in the uorescence intensity of the histogram indicates the increased occupation of Fig. 2 The detection of NF-kB translocation using the selective release/immunostaining protocol and conventional flow cytometry. The cell populations with and without TNFa stimulation were examined. The difference between the two populations was not revealed in either the fluorescence intensity histograms (A) or 2D dot plots (B) involving both fluorescence intensity and forward scatter (i.e. FSC), when standard immunostaining was used. In comparison, there was pronounced difference between the two populations in both the fluorescence intensity histograms (C) and 2D dot plots (D) when selective release followed by immunostaining was conducted. TNFa stimulation was conducted at 37 C with 50 ng ml À1 TNFa for 30 min. Selective release was performed using 0.05% saponin for 10 min at room temperature.</p><p>the nucleic localization over time. The data show that the translocation occurs substantially within the rst 5 min aer simulation. There is increased translocation until 30 min aer stimulation. Interestingly, translocation in the reverse direction (from the nucleus to the cytosol) occurs between 40-60 min. Such reverse translocation was previously reported in the literature 41 and is due to re-inhibition of newly synthesized repressor IkB. 8 The western blotting analysis (Fig. 4B) also corroborates these ndings by our technique.</p><p>The resolution of the technology depends on several factors. First, the amount of the protein translocation between the nucleus and the cytosol affects the resolution. Based on the comparison of Fig. 4A and B, the translocation of $13% of the total NF-kB in the entire cell can be clearly resolved by the ow cytometry data. Second, the completeness and selectiveness of cytosolic release by saponin are critical for high resolution. As demonstrated in Fig. S2, † optimal treatment conditions need to be obtained for a particular cell/protein system in order to reach the best resolution. Third, the immunostaining aer the selective release needs to be complete and yields strong uorescence signal. This facilitates obtaining high quality ow cytometry data.</p><p>To conclude, by combining selective release of cytosolic proteins via chemical permeabilization with uorescence immunostaining, we develop a protocol that links the uorescence intensity of a single cell with the subcellular localization of a targeted protein. By screening the uorescence emitted by single cells using common ow cytometry, we are able to detect the translocation quantitatively with single cell resolution.</p><p>Because uorescence immunostaining is ideally suited for studying cell samples from animals and patients, our approach provides a very simple route for examining protein translocation at the single cell level with direct biomedical relevance. We expect that this approach can be extended to a wide range of cell types and proteins.</p><!><p>HeLa (CCL-2) cells were grown at 37 C with 5% CO 2 in Dulbecco's modied Eagle's medium (DMEM) (Mediatech, Herndon, VA) supplemented with 10% (v/v) fetal bovine serum (Sigma) and 1% penicillin (100 mg ml À1 , Sigma). The cells were trypsinized and diluted at a ratio of 1 : 5-1 : 8 every 2 days to maintain the cells in the exponential growth phase. The Fig. 3 The dose dependence of TNFa stimulation analyzed by our approach. TNFa stimulation was conducted at 37 C for 30 min. Selective release was performed by incubating cells with 0.05% saponin for 10 min at room temperature.</p><p>Fig. 4 The temporal dynamics of NF-kB translocation detected by our approach (A) and verified by western blotting analysis of the nuclear fraction (B). Cells were stimulated by 50 ng ml À1 TNFa at 37 C for various durations (0-60 min). Selective release was performed by 0.05% saponin for 10 min at room temperature.</p><p>This journal is © The Royal Society of Chemistry 2014 Chem. Sci., 2014, 5, 2530-2535 | 2533 harvested cells were centrifuged at 300g for 5 min and resuspended in DMEM culture medium at a nal concentration of 1 Â 10 6 cells per ml before experiments. To stimulate cells, cells (1 Â 10 6 ml À1 ) were suspended in the culture medium with various concentrations of TNF-a (AbD Serotec, Raleigh, NC, USA) at 37 C for various periods.</p><!><p>Fluorescence immunostaining was conducted following the literature with minor changes. 42 The cell sample ($1 Â 10 6 cells) was xed in 100 ml pre-warmed (at 37 C) xation buffer (4% paraformaldehyde in PBS buffer) for 10 min. Subsequently, the xed cells were washed with a blocking buffer (1% BSA in PBS buffer) and permeabilized with 100 ml of a permeabilization buffer (0.2% Triton X-100 in PBS buffer). Aer 20 min incubation with the permeabilization buffer, the cells were centrifuged at 300g for 5 min to remove the permeabilization buffer and washed with the blocking buffer. The cells were then incubated with the blocking buffer containing a primary antibody [1 : 100 dilution of NF-kB p65 (sc-8008, Santa Cruz Biotechnology, Dallas, Texas, USA)] for 1 h at room temperature. Aer incubation with the primary antibody, the cell sample was pelleted by centrifugation (300g, 5 min) and washed twice with the blocking buffer. Then, the cells were incubated (protected from light) in the blocking buffer containing uorophore-conjugated secondary antibody [1 : 150 dilution of DyLight™ 488 Goat antimouse IgG1 antibody (409102, Biolegend, San Diego, CA)], which bound to the primary antibody, for 1 h at room temperature. The staining solution was then aspirated out and the labeled cells were washed twice with the blocking buffer to remove nonspe-cic binding. The cells were stored in PBS with 0.1% sodium azide at 4 C, if not immediately analyzed by ow cytometry.</p><!><p>The cell sample ($1 Â 10 6 cells) was incubated in 100 ml of the releasing buffer [0.05% (w/v) saponin (ID# 419-25A, Chem Service, West Chester, PA, USA) in DMEM] for 5 min. The processed cells were then pelleted by centrifugation at 300g for 5 min to remove excessive releasing buffer and then immediately xed by the xation buffer for 30 min. Subsequently, the xed cells were permeabilized in 100 ml of the permeabilization buffer. Aer 5 min incubation in the permeabilization buffer, the cells were centrifuged at 300g for 5 min to remove excessive permeabilization buffer and washed once with the blocking buffer. The rest of the procedure involving labeling using primary and secondary antibodies was the same as that in "Standard uorescence immunostaining".</p><!><p>Immunostained cells were transferred to a 96 well plate and then centrifuged for 5 min at 300g to settle the cells to the bottom. Fluorescence images were taken by an inverted uorescence microscope (IX-71, Olympus, Melville, NY) with a 20Â dry objective (0.5 NA). The uorescence excitation was provided by a 100 W mercury lamp. The excitation and emission were ltered by a uorescence lter cube (exciter HQ480/40, emitter HQ535/50, and beam splitter Q505lp, Chroma Technology) for observing green uorescence.</p><!><p>Cellular protein samples were made using NE-PER Nuclear and Cytoplasmic Extraction Reagents (Pierce Biotechnology, Rockford, IL, USA) following the manufacturer's recommendations. The total protein concentration for each sample was measured by Pierce BCA protein assay kit (Pierce Biotechnology, Rockford, IL, USA), and then equal amount of denatured proteins from each sample was separated by standard SDS-PAGE and analyzed by Western blotting. Briey, samples were loaded into polyacrylamide gel for SDS-PAGE followed by transferring to a polyvinylidene uoride (PVDF) membrane, which was blocked by 5% milk in TBST (Tris-Buffered Saline, 0.1% Tween-20) buffer. The membrane was stained with 1 : 1000 diluted primary antibody (SC-8008, Santa Cruz Biotechnology) and secondary antibody [Rabbit anti-mouse horseradish peroxidase (HRP) (Pierce, Rockford, IL)] for 1 h each. Membranes were visualized by LAS-3000 luminescent image analyzer (Fujilm, Hanover Park, IL, USA) aer chemiluminescence treatment with Pierce ECL Western Blotting Substrate (Pierce Biotechnology, Rockford, IL, USA). The intensity of each band was quantied using ImageJ soware.</p><!><p>Fluorescently stained cell samples were analyzed at medium ow rate by a FACS Canto II cytometer (BD, San Jose, CA, USA) equipped with 488 nm laser/lters for FITC and forward scatter (FSC) detection. For each histogram, 10 000-20 000 events were collected. The cytometer was routinely calibrated with Calibrite beads (BD). Stained cell samples may be stored for up to 24 h at 4 C before analysis without a signicant loss in uorescence intensity. The data were processed by FlowJo and Origin 9.0.</p>

Royal Society of Chemistry (RSC)

Recent Advances in Organic Light-Emitting Diodes Based on Pure Organic Room Temperature Phosphorescence Materials

Pure organic room temperature phosphorescence (RTP) materials have attracted extensive attention in recent years due to their unique characteristics, such as flexible design method, low toxicity, low cost, as well as the ease of production at scale. The involvement of triplet state and direct radiative transition from the triplet state show that RTP materials have great potential as a new generation emitter in organic light-emitting diodes (OLEDs). Based on the mechanism of phosphorescence, various methods have been developed to achieve RTP emissions in the crystal state. However, the observation of RTP in the thin film state is much more difficult to achieve because of the lower degree of rigidity and suppression of the non-radiative transition. In this mini-review, molecular design strategies developed to achieve RTP emissions and their application in OLEDs are summarized and discussed. The conclusion and outlook point to great potential as well as the challenges for the continued study of pure organic RTP materials-based OLEDs.

recent_advances_in_organic_light-emitting_diodes_based_on_pure_organic_room_temperature_phosphoresce

Introduction<!>Basic Principles for RTP<!><!>Basic Principles for RTP<!>Brief Introduction to OLEDs<!>OLEDs Based on Pure Organic RTP Materials<!><!>OLEDs Based on Pure Organic RTP Materials<!>Other Pure Organic RTP Materials Potential for OLEDs<!>Conclusion and Outlook<!>Author Contributions<!>Conflict of Interest Statement

<p>The early stage of organic light-emitting diodes (OLEDs) are based on fluorescent materials (Tang and Vanslyke, 1987), which could not utilize the triplet excitons that accounted for 75% of the total excitons (Baldo et al., 1999), and caused incomplete energy utilization and low device efficiency. In 1998, Ma et al. (1998) and Baldo et al. (1998) introduced osmium complex and platinum complex as luminescent materials into OLEDs, which increased the theoretical maximum internal quantum efficiency (IQE) of the device from 25% of the fluorescent material to 100% of the phosphorescent material. So far, phosphorescent OLEDs have achieved great success and has even been applied in commercial devices, such as mobile phones, televisions, and so on.</p><p>However, the noble metals contained in phosphorescent complexes are expensive, low in abundance and toxic, which restricts the further development and popularization of OLEDs. Therefore, metal-free luminescent materials have attracted increased interests in OLEDs, among which thermally activated delayed fluorescent (TADF) materials and pure organic room temperature phosphorescence (RTP) materials are successively introduced into OLEDs as emitters, while the OLEDs also showed a theoretical maximum IQE up to 100%.</p><p>Different from TADF materials, which have been demonstrated great success in OLEDs, the application of pure organic RTP materials in OLEDs is still in its initial stage, because high efficiency and short-lived RTP molecules suitable for OLEDs are rare. Nevertheless, RTP materials will provide more possibilities for high performance OLEDs and deserve to be explored further. This mini-review starts with an introduction to basic concepts such as RTP and OLEDs, and then discusses representative work on the electroluminescence study of pure organic RTP materials as well as the reported pure organic RTP materials potentially using in fabricating OLEDs. Finally, the potential and challenges of the study of electroluminescence on pure organic RTP materials are summarized.</p><!><p>In general, the production of phosphorescence in pure organics involves two necessary processes: (i) intersystem crossing (ISC) from the lowest excited singlet state (S1) to a triplet state (Tn) and (ii) radiative transition from the lowest excited triplet state (T1) to the ground state (S0) (Figure 1A). However, the excited triplet state can only be generated by ISC process from an excited singlet state (Reineke and Baldo, 2014). Therefore, kISC > 0 is a necessary condition for generating phosphorescence emission, where kISC is the ISC rate (Hirata, 2017). Only the energy level and electronic configuration determine kISC. The ISC process can be accelerated by a small energy gap between S1 and T1 (ΔEST). Experiments have shown that the ISC and reverse ISC (RISC) process are both accelerated when ΔEST is extremely small (< 100 meV), and TADF emission can be obtained under this condition (Uoyama et al., 2012). However, TADF showed a different photophysical process to phosphorescence, as TADF contains both a prompt and delayed radiative transition from S1 to S0, while phosphorescence is a radiative transition from T1 to S0. The effect of electronic configuration on ISC has been confirmed by El-Sayed (Kalyanasundaram et al., 1977). He found that the spin-orbit coupling could be promoted by mixing different electronic configuration singlet and triplet states, such as (π, π*) and (n, π*). In addition, the heavy atom effect is also widely used to accelerate the kISC process (Plummer et al., 1993). Therefore, introducing n electrons containing atoms such as O and N, and heavy atoms like Br and I, are strategies frequently used to design efficient RTP materials.</p><!><p>Schematic diagram of OLEDs based on RTP materials. (A) Schematic Jablonski diagram of photoluminescence for RTP materials. (B) The typical structure of three-layer OLEDs. (C) The schematic injection, transport, and recombination process of holes (black circles) and electrons (black dots) in OLEDs.</p><!><p>Due to a longer lifetime, the excited triplet state can be easily quenched under ambient conditions (Schulman and Parker, 1977). The second necessary condition for obtaining high efficiency phosphorescence is kP > knr, where kP and knr are a radiative and non-radiative transition rate from T1 to S0, respectively. The non-radiative process can be divided into external losses caused by the interaction with environmental conditions and intramolecular losses (Liu et al., 2016). At room temperature, kP is generally less than knr in pure organic compounds, which is the main reason for the low photoluminescence quantum yield (PLQY) of pure organic RTP materials. Therefore, suppressing non-radiative transitions may be the most important and challenging part of achieving effective RTP in pure organic materials.</p><!><p>A typical OLEDs (Figure 1B) includes a hole transport layer (HTL), an emitting layer (EML), an electron transport layer (ETL), an anode, and a cathode. In addition, a hole injection layer (HIL) and an electron injection layer (EIL) are widely used to reduce the carrier injection barrier from the electrode to the organic layer, while a hole blocking layer (HBL) and an electron blocking layer (EBL) are usually used to effectively confine the hole and electron within the EML. Based on the working mechanism (Figure 1C), the external quantum efficiency (EQE) of OLEDs could be deduced as EQE = ηe•h × ηPL × ηexciton × ηout, where ηe•h is the recombination efficiency of injected holes and electrons, ηPL is the intrinsic photoluminescence efficiency, i.e., PLQY of the EML, ηexciton is the radiative exciton ratio, and ηout is the light out-coupling efficiency. Since the theoretical ηexciton and ηout are 100% and ca. 20% respectively, the challenge for RTP based OLEDs is to achieve high PLQY in a thin film.</p><!><p>Though pure organic RTP materials have great potential as emitters in OLEDs, there are only a few examples of the study of electroluminescence in pure organic RTP materials, since most reported pure organic RTP materials showed low PLQY and a long excited lifetime (Kabe and Adachi, 2017), which is not suitable for fabricating high efficiency OLEDs.</p><p>In 2013, Bergamini et al. synthesized RTP-1 (Figure 2) consisting of a hexathio-benzene core and peripheral tolyl substituents (Bergamini et al., 2013). The compound showed outstanding phosphorescence in solid state at room temperature, while no luminescence was observed in solution. The authors attributed the luminescence behavior to a rigid environment and limited conformational migration of the tolyl substituent, which suppresses the non-radiative deactivation process from T1. Since the compound showed a high PLQY up to 80% in solid powder, the authors applied RTP-1 in OLEDs as an emitter and fabricated OLEDs with a structure of indium tin oxide (ITO)/poly(3,4-ethylenedioxythiophene)-poly-(styrenesulfonic acid) (PEDOT:PSS)/polyvinylcarbazole (PVK):2-(4-biphenylyl)-5-(4-tert-butylphenyl)-1,3,4-oxadiazole (PBD):RTP-1/Ba/Al. The device showed EQE and current efficiency of 0.1% and 0.5 cd A−1 at 11 V, respectively. The performance was not satisfactory because the device architecture was not optimized and the PLQY measured in the same blend film (PVK:PBD:RTP-1) was only 2%, which is much lower than that measured in the powder. The result demonstrated that the low rigidity of blend film cannot effectively suppress the competitive deactivation process of the excited triplet state. Although the device performance was poor, this work is the first attempt at using pure organic RTP materials as an emitter in OLEDs. Pure organic RTP materials then became a new choice for OLEDs after fluorescence materials, phosphorescence materials and TADF materials. It should be noted that the luminescence mechanism of RTP-1 in the film has not been well-studied, and the device exhibited different electroluminescence spectra under various voltages, which indicates that the device performance could be further improved by using proper host matrix and device structures.</p><!><p>Chemical structure of RTP materials used or potential to be used in OLEDs.</p><!><p>Later in 2013, Chaudhuri et al. developed two pure organic RTP materials RTP-2 and RTP-3 (Figure 2) and their OLEDs were fabricated with a structure of ITO/PEDOT:PSS/N,N'-bis(3-methylphenyl)-N,N'-diphenylbenzidine (TPD)/PVK:RTP-2 (or RTP-3)/PBD/CsF/Al (Chaudhuri et al., 2013). The electroluminescence spectra showed two distinct peaks, 560 and 690 nm for RTP-2, and 630 and 760 nm for RTP-3. Such dual emissions contain both singlet and triplet transitions of the molecule under electrical excitation and reveal a long-lived triplet afterglow. However, RTP-2 and RTP-3 showed low PLQY of 4.6 and 1.3% and their corresponding device also showed very low EQE of 2.54 × 10−4 and 5.58 × 10−5, respectively. In this work and subsequent work, the authors studied the photoluminescence properties of these RTP materials in doped polymer films in detail (Ratzke et al., 2016). Through characterization and attribution of different emission peaks in the spectra, the authors provided a new understanding on the behavior of RTP materials in film, which is useful for RTP materials to be applied as emitters in OLEDs.</p><p>In 2016, Kabe et al. dispersed RTP-4 (Figure 2) into the host molecule of 3-(N-carbazolyl)-androst-2-ene (CzSte) and observed dual emissions of blue fluorescent and green phosphorescent under photoexcitation (Kabe et al., 2016). By using RTP-4 doped CzSte as the EML, they fabricated an OLEDs with a device structure of ITO/4,4;-bis[N-(1-naphthyl)-N-phenyl-amino]biphenyl (α-NPD) (30 nm)/1,3-bis(N-carbazolyl)benzene (mCP) (10 nm)/ RTP-4:CzSte (1 %, 30 nm)/1,3,5-tris(N-phenylbenzimidazol-2-yl)benzene (TPBi) (60 nm)/LiF (0.8 nm)/Al (80 nm). The device emitted a blue emission with an EQE around 1% under the external electric current excitation. When the external electric current excitation was turned off, the device could still emit a green permanent emission with a lifetime of 0.39 s. The significant improvement of the device performance compared with the aforementioned two examples may arise from a better device fabrication method, i.e., thermal evaporation vs. solution spin coating, which provides more choices for host materials and other functional layers.</p><!><p>The requirements for luminescent materials in OLEDs are often combined with the advantages of high PLQY and a short-excited state lifetime, in order to produce devices with a high efficiency and low efficiency roll-off. Screening by these requirements, there are a few pure organic RTP materials reported in the literature, based on which OLEDs are expected to have good performance.</p><p>In 2014, Koch et al. designed and synthesized a series of boron-based RTP materials (Koch et al., 2014). The photophysical results showed that the compounds RTP-5–RTP-11 (Figure 2) have a high PLQY (29-104%) and a short-excited state lifetime (1233-5413 ns) that is suitable for OLEDs. In particular, the compound RTP-7 showed a high PLQY up to 104% and a short-excited state lifetime of 1,382 ns in a diluted dichloromethane solution, as well as a high PLQY up to 118% in poly(methyl methacrylate) (PMMA). This indicates that the non-radiative transition in these compounds can be well-suppressed even in a less rigid atmosphere, leading to a high PLQY and short excited state lifetime comparable to phosphorescent complexes (Endo et al., 2008). However, OLEDs based on these RTP materials are yet to be fabricated and explored.</p><p>In 2016, Shimizu et al. reported RTP-12–RTP-16 (Figure 2) as a new class of RTP materials (Shimizu et al., 2016). The crystals of these compounds showed photoluminescence under ambient conditions. Intermolecular interactions were observed in each crystal, which contributed to the restriction of intramolecular motion and suppressed non-radiative transitions. The excited state lifetimes were dozens of microseconds and the PLQY was 14 and 8% for RTP-13 and RTP-14, respectively. Though these compounds were not emissive either in solution or in a doped polymer film, the short-excited state lifetime in microseconds is attractive for the high efficiency OLEDs, which deserves to be explored in more detail.</p><p>Later in 2016, Gutierrez et al. reported red phosphorescence from RTP-17–RTP-20 (Figure 2) at room temperature (Gutierrez et al., 2016). The photophysical properties of these compounds in deoxygenated cyclohexane are presented. The excited state lifetime of 2.8–5.4 μs is similar to that of traditional phosphorescent complexes, making them suitable for OLEDs. However, the PLQYs of this class of compound in a solution are very low (<1%). The reason is probably due to the presence of long alkyl chains. In the low-rigid solution, the twisting of the alkyl chain may lead to an increase in the non-radiative transition. Reducing the length of the alkyl chain or incorporating the luminescent molecules into more rigid host materials could upgrade the PLQY of these compounds.</p><p>In 2018, Huang et al. proposed a series of donor-acceptor-donor (D–A–D) compounds RTP-21–RTP-24 (Figure 2) (Huang et al., 2018). The unsubstituted compound RTP-21 exists in both equatorial and axial forms in the ground state, but the equatorial conformer prevails in the excited state. The changing in conformers lead to enhancement of RTP emissions with a high PLQY up to 71% in zeonex solid films. The excited state lifetime of 63.3 μs also indicates that the molecule is suitable for application in OLEDs. The phosphorescence quantum yield of RTP-21 is the highest among all current reported RTP molecules, and highly efficient OLEDs are expected when using RTP-21 as the emitter.</p><!><p>In recent years, a variety of RTP materials have been designed and synthesized, the color of which can cover the entire visible region (Li et al., 2018). However, the application of pure organic RTP materials in OLEDs is still in its infancy. The reasons are as follows: (i) The PLQYs of pure organic RTP materials in the thin film state tend to be very low. Compared with the phosphorescent complexes or TADF materials commonly used in OLEDs, the PLQYs of pure organic RTP materials are still at low levels, and the effects of non-radiative transitions are significant. Most pure organic RTP materials tend to exhibit high PLQY in crystals or at low temperatures, while their PLQYs decrease significantly in a solution or a thin film at room temperature. The reason is that the rotation and vibration of the molecules are suppressed in a rigid environment or at a low temperature, where the rate of non-radiative transition is greatly reduced; (ii) The pure organic RTP materials have low radiative transition rates and long excited state lifetimes. Most of the reported pure organic RTP materials have lifetimes in the order of milliseconds or even seconds, and RTP materials with short lifetimes of several microseconds have hardly been reported. The long-excited state lifetime may result in serious triplet-triplet annihilation, leading to obvious efficiency roll-off in OLEDs.</p><p>Though the current performance of pure organic RTP based OLEDs is poor, it can theoretically achieve 100% IQE like noble metal complexes and TADF materials, which provide more possibilities for high performance OLEDs. The design and synthesis of pure organic RTP materials with a high PLQY and a short-excited state lifetime, especially in the thin film state, is the key goal in this field, since the EML in OLEDs is often a spin-coated or vapor-deposited thin film. To achieve this, it is conceivable to introduce n electrons containing atoms and heavy atoms (Saigusa and Azumi, 1979) into the luminescent molecule to accelerate kISC and kp, to improve the PLQY of the RTP materials. It is also possible to introduce a more rigid host material or a full-deuterated host to suppress non-radiative transitions. Moreover, OLEDs architecture optimization, using proper functional materials for balanced charge transport, favorable exciton confinement, and efficient energy transfer is also critical in improving the performance of RTP based OLEDs.</p><!><p>GZ wrote the manuscript. ZL, ZB, and CH helped to revise the manuscript.</p><!><p>The authors declare that the research was conducted in the absence of any commercial or financial relationships that could be construed as a potential conflict of interest.</p>

PubMed Open Access

Sterically controlled reductive oligomerisations of CO by activated magnesium(<scp>i</scp>) compounds: deltate <i>vs.</i> ethenediolate formation

An extremely bulky, symmetrical three-coordinate magnesium(I) complex, [{( TCHP Nacnac)Mg} 2 ] ( TCHP Nacnac ¼ [{(TCHP)NCMe} 2 CH] À , TCHP ¼ 2,4,6-tricyclohexylphenyl) has been prepared and shown to have an extremely long Mg-Mg bond (3.021(1) A) for such a complex. It was shown not to react with either DMAP (4-dimethylaminopyridine) or CO. Three unsymmetrical 1 : 1 DMAP adducts of less bulky Mg-Mg bonded species have been prepared, viz. [( Ar Nacnac)Mg-Mg(DMAP)( Ar Nacnac)] ( Ar Nacnac ¼ [(ArNCMe) 2 CH] À Ar ¼ 2,6-xylyl (Xyl), mesityl (Mes) or 2,6-diethylphenyl (Dep)), and their reactivity toward CO explored. Like the previously reported bulkier complex, [( Dip Nacnac)Mg-Mg(DMAP)( Dip Nacnac)] (Dip ¼ 2,6-diisopropylphenyl), [( Dep Nacnac)Mg-Mg(DMAP)( Dep Nacnac)] reductively trimerises CO to give a rare example of a deltate complex, [{( Dep Nacnac)Mg(m-C 3 O 3 )Mg(DMAP)( Dep Nacnac)} 2 ]. In contrast, the two smaller adduct complexes react with only two CO molecules, ultimately giving unusual ethenediolate complexes [{( Ar Nacnac)Mg{m-OC(H)]C(DMAP ÀH )O}Mg( Ar Nacnac)} 2 ] (Ar ¼ Xyl or Mes). DFT calculations show the latter reactions to proceed via reductive dimerizations of CO, and subsequent intramolecular C-H activation of Mg-ligated DMAP by "zig-zag" [C 2 O 2 ] 2À fragments of reaction intermediates. Calculations also suggest that magnesium deltate complexes are kinetic products in these reactions, while the magnesium ethenediolates are thermodynamic products. This study shows that subtle changes to the bulk of the reacting 1 : 1 DMAP-magnesium(I) adduct complexes can lead to fine steric control over the products arising from their CO reductive oligomerisations. Furthermore, it is found that the more activated nature of the adduct complexes, relative to their symmetrical, threecoordinate counterparts, [{( Ar Nacnac)Mg} 2 ], likely derives more from the polarisation of the Mg-Mg bonds of the former, than the elongated nature of those bonds.

sterically_controlled_reductive_oligomerisations_of_co_by_activated_magnesium(<scp>i</scp>)_compound

Introduction<!>Results and discussion<!>Conclusions<!>Conflicts of interest

<p>Carbon monoxide is a readily available C 1 feedstock gas, that is used in many industrial processes for the production of hundreds of millions of tonnes of commodity chemicals each year. One of the most important of these processes is Fischer-Tropsch (F-T), which typically utilises heterogeneous transition metal catalysts to transform synthesis gas (CO/H 2 ) into liquid hydrocarbons and oxygenates on a massive scale. 1 Considering the importance of F-T, a great deal of effort has focussed on investigating the mechanisms by which it operates. These investigations have increasingly involved the use of low-valent organometallic complexes as soluble models to shed light on the fundamental steps, e.g. C-C bond formations, that are central to the process. 2 Such studies have the potential to enhance the selectivity and energy efficiency of F-T, and to aid the development of new homogeneous catalysts of commercial importance.</p><p>In the realm of homogeneous organometallic models for the F-T process, considerable recent interest has been directed towards the activation and reductive homologation of CO, a normally inert gas which possesses one of the strongest bonds known (BDE ¼ 257 kcal mol À1 (ref.</p><p>3)). For example, this work has led to the discovery that low-valent metal and non-metal compounds, from across the periodic table, can reductively oligomerise CO to ethynediolate, [OC^CO] 2À , aromatic oxocarbon dianions, [C n O n ] 2À (n ¼ 2-6), and related species, under mild conditions. [4][5][6] From a historical perspective, it should be noted that alkali metals have been known to reductively a School of Chemistry, Monash University, PO Box 23, VIC, 3800, Australia. E-mail: cameron.jones@monash.edu; Web: http://www.monash.edu/science/research-groups/ chemistry/jonesgroup oligomerise CO to salts of oxocarbon dianions since the early 19th century, though those salts have been poorly characterised. 7 In order to access well-dened s-block metal complexes of oxocarbon anions derived from CO, we have recently reported on reactions of this gas with reducing b-diketiminate coordinated magnesium(I) compounds. 8,9 Initially, it was found that three-coordinate examples of these Mg-Mg bonded species, viz. were shown to selectively reductively trimerise CO to give the deltate complexes 3 and 4 under ambient conditions (Scheme 1). 8 The only prior example of a structurally authenticated deltate complex was reported by Cloke and co-workers to arise from reductive trimerisation of CO by an organo-uranium(III) complex. 6a In light of our preparations of 3 and 4, we were interested in investigating the effect that the steric prole of a magnesium(I) reductant has on the outcome of its reaction with CO. Here, we show that magnesium deltate or ethenediolate complexes can be selectively prepared, simply by altering the b-diketiminate N-aryl substituent. Computational studies have been used to probe the mechanisms of the observed reactions, and the origins of reaction product selectivity.</p><!><p>At the outset, a magnesium(I) adduct complex bearing substantially bulkier b-diketiminate ligands than the Dip Nacnac substituents in 1 and 2, was targeted. The reasoning here stemmed from recent work by Harder and co-workers, who showed that the extremely bulky magnesium(I) compound, [( DiPep Nacnac)Mg-Mg( DiPep Nacnac)] (DiPep ¼ 2,6-diisopentylphenyl), has an Mg-Mg bond (3.0513(8) A) that is ca. 0.2 A longer than in any other symmetrical three-coordinate complex, and similar in length to those in the activated adduct complexes, 1 and 2. 11 If a 1 : 1 adduct of such a bulky magnesium(I) compound could be formed, its Mg-Mg bond would likely be even longer, and more activated.</p><p>To test this hypothesis the very hindered magnesium(I) complex [{( TCHP Nacnac)Mg} 2 ] ( TCHP Nacnac ¼ [{(TCHP)NCMe} 2 -CH] À , TCHP ¼ 2,4,6-tricyclohexylphenyl) 5 was prepared in good isolated yield, by reduction of a toluene/diethyl ether solution of [( TCHP Nacnac)MgI(OEt 2 )] over a sodium mirror (see ESI † for full details). 12 However, subsequent treatment of 5 with one equivalent of either DMAP or the NHC, :C{(MeNCMe) 2 }, led to no reaction, presumably due to steric inaccessibility of its Mg centres, as is evident from its molecular structure (Fig. 1a). This is in contrast to Harder's similarly bulky magnesium(I) dimer, which did show evidence of adduct formation when treated with DMAP. 11 Despite compound 5 possessing an Mg-Mg bond similar in length (3.021(1) A) to that in Harder's compound, and to those in activated 1 and 2, it proved unreactive towards CO, even when the mixture was heated at 60 C for hours. Considering the size of the CO molecule, it seems unlikely that this lack of reactivity derives solely from the steric bulk of 5, and perhaps indicates that unsymmetrical adduct complexes, [( Ar Nacnac)Mg-Mg(L)( Ar Nacnac)], are required to enable CO reduction (see below).</p><p>So as to explore this in more detail, attention turned to the preparation of magnesium(I) compounds, related to 1 and 2, but in this case, bearing smaller b-diketiminate ligands. To this end the magnesium(I) compounds, [{( Ar Nacnac)Mg} 2 ] (Ar ¼ Xyl, Mes or Dep), 10 were all treated with one equivalent of DMAP, in two cases affording the 1 : 1 adduct complexes, [( Ar Nacnac)Mg-Mg(DMAP)( Ar Nacnac)] (Ar ¼ Xyl 6, or Dep 8), in moderate to good isolated yields as red or red-orange crystalline solids. 13 While the mesityl substituted analogue of these compounds, [( Mes Nacnac)Mg-Mg(DMAP)( Mes Nacnac)] 7, could not be isolated, it could be generated in situ and used for further reactions. Compounds 6 and 8 are oxygen sensitive, but stable in the solid state and solution for days at room temperature. Similar to the situation for solutions of 1 and 2, 8 variable temperature NMR spectroscopic studies revealed uxional behaviour for 6 and 8, which is believed to originate from rapid "hopping" of the Lewis base donor between the two Mg centres of the adduct complexes at room temperature (see ESI † for further discussion).</p><p>The solid-state molecular structures of the adducts (see Fig. 1b for the molecular structure of 8) are also reminiscent of those for 1 and 2, and show both to possess one trigonal planar, and one distorted tetrahedral, magnesium centre. Interestingly, their Mg-Mg bond lengths (6: 2.8925(9) A; 8: 2.9336(7) A), while longer than those typically seen for symmetrical, threecoordinate magnesium(I) compounds (e.g. 2.875(1) A for [{( Dep-Nacnac)Mg} 2 ] 10 ), are not as long as the metal-metal interaction in 1 (viz. 3.0886(6) A). Indeed, the Mg-Mg distances in this trio of compounds are loosely proportional to the size of their bdiketiminate ligands, and cover a range of more than 0.2 A. This observation is fully consistent with the previously computed shallow potential energy surface for the elongation of Mg-Mg bonds in compounds such as [{( Ar Nacnac)Mg} 2 ]. 13a With 6 and 8 in hand, toluene solutions of the compounds, and of in situ generated 7, were stirred under atmospheres of CO, in order to investigate if their steric differences had an inuence on the outcomes of these reactions. This seemed to be the case, as the most hindered adduct, 8, behaved similarly to 1, in that it reductively trimerised CO to give a low isolated yield of the magnesium deltate complex, 9, as a colourless crystalline solid (Scheme 2). In contrast, the two smaller adduct complexes, 6 and 7, only consumed two equivalents of CO to give moderate isolated yields of the unusual, thermally stable ethenediolate complexes, 10 and 11. These are presumably formed via an initial reductive dimerisation of CO, followed by activation of one of the ortho-C-H bonds of the coordinating DMAP molecule by the generated [C 2 O 2 ] 2À fragment (see below). It is noteworthy that, when the progress of all of these reactions was monitored by 1 H NMR spectroscopy, no intermediates in the formations of the CO coupled products were observed, and there was no evidence for mixtures of deltate or ethenediolate products in any case. Furthermore, treating the adduct complexes, 6-8, with 1 : 1 mixtures of CO/H 2 did not lead to involvement of dihydrogen in the reactions, which instead returned 9-11 in yields similar to those in its absence. 14 Once crystallised, 9-11 are poorly soluble in most commonly used organic solvents. With that said, compounds 9 and 11 had sufficient solubility in d 8 -THF for their 1 H and 13 C NMR spectra to be recorded. The spectra for deltate complex, 9, are consistent with its proposed formulation, and comparable to the spectra for 3, 8 The molecular structures of 9 and 11 are depicted in Fig. 2, while that for 10 can be found in the ESI. † As compound 9 is essentially isostructural to 3, 8 little comment will be passed on it here, except to point out that its deltate dianions are close to planar, with nearly equivalent C-C and C-O bond lengths, that lie between those for localised single and double bonds. 15 It is apparent, therefore, that there is a signicant degree of electronic delocalisation over the compound's aromatic deltate dianions. 16 In contrast, the only other previously structurally characterised deltate complex of a non s-block metal, It seems likely that the mechanism of formation of 9 is similar to that previously calculated for the closely related NHC coordinated deltate complex, 4. 8 In that case, there were two sequential insertions of CO into the Mg-Mg bond of the magnesium(I) starting material, yielding an intermediate with a trans-bent ("zig-zag") [C 2 O 2 ] 2À dianion, bridging two [( Dip-Nacnac)(DMAP) 0 or 1 Mg] + fragments. This reacts with a third CO molecule, ultimately leading to deltate complex 4. Interestingly, Fig. 3 Computed (B3PW91) enthalpy profile at 298 K for the formation of ethenediolate complex 11, or deltate complex 12, from magnesium(I)adduct complex 7, and two or three molecules of CO, respectively. a very similar mechanism, via a "zig-zag" intermediate, was computed by Cloke and co-workers for the formation of 18 They also showed that controlling the stoichiometry of the reaction between [U(COT † )(Cp*)(THF)] and CO led to the linear ethynediolate system, [{U(COT † )(Cp*)} 2 (m-OC^CO)], which did not react with CO to give [{U(COT † )(Cp*)} 2 (m-C 3 O 3 )]. 6f This result gave credence to the importance of the "zig-zag" intermediate in the formation of the latter complex. That is not to say that uranium ethynediolate complexes are unreactive, as Arnold and co-workers showed when they heated solutions of [{[(Me 3 Si) 2 N] 2 U} 2 (m-OC^CO)]. This led to an intramolecular C-H activation of one of its methyl substituents, yielding an ethenediolate species, related to 10 and 11. 6g This raised the question as to whether the mechanism of formation of 10 and 11 proceeds via reactive ethynediolate intermediates, [{( Ar Nacnac)Mg} 2 (m-OC^CO)], or by another process.</p><p>To explore these possibilities, DFT calculations (B3PW91) were carried out to determine the reaction prole that led to 11 (Fig. 3). The initial stages of the reaction were found to be similar to that calculated for the formation of 4. 8 That is, the rst step involves nucleophilic attack of the three-coordinate Mg 2 centre of polarized 7 (Natural Bond Orbitals, NBO, charges: Mg 1 0.45; Mg 2 0.19) at one of the p*-orbitals of CO (see ESI † for further details), giving adduct TS1 (10.5 kcal mol À1 ). The coordinated CO then inserts into the Mg-Mg bond affording stable intermediate INT1 (À3.6 kcal mol À1 ). From INT1, a second CO insertion leads to a "zig-zag" intermediate INT2 (À18.7 kcal mol À1 ) via a low kinetic barrier (5.4 kcal mol À1 ). It is of note that an alternative pathway was explored, whereby the DMAP C-H activation process occurred from INT1, but this was found not to be kinetically viable (barrier ¼ 21.4 kcal mol À1 ). Instead, the favoured intermediate INT2 readily isomerised to the more stable (by 10.9 kcal mol À1 ) "zig-zag" intermediate INT2 0 . Interestingly, this isomer appears more amenable to DMAP C-H activation, as one DMAP ortho-proton is pointing in the direction of the [C 2 O 2 ] 2À moiety. From INT2 0 , two kinetically reasonable pathways were examined. Firstly, isomerization back to INT2 and reaction with a third molecule of CO led to the deltate complex 12 (À61.1 kcal mol À1 ) via a pathway similar to that calculated for the formation of 4 (blue pathway). Secondly, INT2 0 undergoes an intramolecular DMAP C-H activation via a number of kinetically accessible steps, ultimately giving the experimentally observed product, 11 (black pathway). While compound 11 is signicantly more stable (by 31.4 kcal mol À1 ) than the alternative deltate product, 12, the overall kinetic barrier to its formation is higher.</p><p>Taken as a whole, the experimental and computational studies indicate that 11 is the thermodynamic product of the reaction of 7 with excess CO, while deltate complex 12, is the kinetic product. Despite this, in the experimental situation, compound 11 is formed in preference to the deltate complex, 12, even when the reaction that afforded it is carried out at low temperature. Moreover, it is clear that 11 is not formed via an ethynediolate intermediate, [{( Mes Nacnac)Mg} 2 (m-OC^CO)] (which was never experimentally observed), in contrast to Arnold's aforementioned report on intramolecular C-H activation of a uranium ethynediolate complex. 6g It can additionally be speculated from the results of these and prior calculations, that ethenediolate complexes, 10 and 11, result when less hindered activated magnesium(I) adduct complexes, 6 and 7, are treated with CO, because an ortho-C-H bond of the ligating DMAP molecule can approach the [C 2 O 2 ] 2À fragment in the "zig-zag" transition state TS2, more readily than in reactions of 1 and 8 with CO. If so, in those latter cases, the kinetic barrier to ethenediolate formation should be raised sufficiently to favour formation of the experimentally observed deltate complexes, 3 and 9. In the case of the extremely bulky magnesium(I) compound 5, no DMAP adduct can be formed, but its Mg-Mg bond is very long, yet it does not react with CO under ambient conditions. This suggests that the enhanced reactivity of 1 : 1 adduct complexes 1 and 2, and 6-8, towards CO arises more from the polarised nature of their Mg-Mg bonds, than the elongation of those bonds.</p><!><p>In summary, an extremely bulky, symmetrical three-coordinate magnesium(I) complex has been prepared and shown to have a very long Mg-Mg bond for a such a species. This does not react with either DMAP or CO. Three 1 : 1 DMAP adducts of less bulky Mg-Mg bonded species have been prepared (one in situ), and their enhanced reactivity toward CO explored. It was found that when the compounds incorporate bulkier b-diketiminate ligands, they reductively trimerise CO to give magnesium deltate complexes. When substituted with smaller b-diketiminates, the magnesium(I) adducts react with only two CO molecules, ultimately giving unusual ethenediolate complexes. DFT calculations show that these reactions proceed via reductive dimerization of CO, and subsequent intramolecular C-H activation of Mg-ligated DMAP by "zig-zag" [C 2 O 2 ] 2À fragments of reaction intermediates. It is apparent that magnesium deltate complexes are kinetic products in these reactions, while magnesium ethenediolates are thermodynamic products. As a result, subtle changes to the bulk of the 1 : 1 DMAP-magnesium(I) adducts can lead to ne steric control over the products arising from their CO reductive oligomerisations. We continue to investigate the reactivity of activated magnesium(I) compounds towards CO and other small molecules, and how selectivity in the products of those reactions can be achieved.</p><!><p>There are no conicts to declare.</p><p>LM is a senior member of the Institut Universitaire de France. CalMip is also acknowledged for a generous grant of computing time.</p>

Royal Society of Chemistry (RSC)

Role of P-glycoprotein inhibitors in ceramide-based therapeutics for treatment of cancer

The anticancer properties of ceramide, a sphingolipid with potent tumor-suppressor properties, can be dampened via glycosylation, notably in multidrug resistance wherein ceramide glycosylation is characteristically elevated. Earlier works using the ceramide analog, C6-ceramide, demonstrated that the antiestrogen tamoxifen, a first generation P-glycoprotein (P-gp) inhibitor, blocked C6-ceramide glycosylation and magnified apoptotic responses. The present investigation was undertaken with the goal of discovering non-anti-estrogenic alternatives to tamoxifen that could be employed as adjuvants for improving the efficacy of ceramide-centric therapeutics in treatment of cancer. Herein we demonstrate that the tamoxifen metabolites, desmethyltamoxifen and didesmethyltamoxifen, and specific, high-affinity P-gp inhibitors, tariquidar and zosuquidar, synergistically enhanced C6-ceramide cytotoxicity in multidrug resistant HL-60/VCR acute myelogenous leukemia (AML) cells, whereas the selective estrogen receptor antagonist, fulvestrant, was ineffective. Active C6-ceramide-adjuvant combinations elicited mitochondrial ROS production and cytochrome c release, and induced apoptosis. Cytotoxicity was mitigated by introduction of antioxidant. Effective adjuvants markedly inhibited C6-ceramide glycosylation as well as conversion to sphingomyelin. Active regimens were also effective in KG-1a cells, a leukemia stem cell-like line, and in LoVo human colorectal cancer cells, a solid tumor model. In summary, our work details discovery of the link between P-gp inhibitors and the regulation and potentiation of ceramide metabolism in a pro-apoptotic direction in cancer cells. Given the active properties of these adjuvants in synergizing with C6-ceramide, independent of drug resistance status, stemness, or cancer type, our results suggest that the C6-ceramide-containing regimens could provide alternative, promising therapeutic direction, in addition to finding novel, off-label applications for P-gp inhibitors.

role_of_p-glycoprotein_inhibitors_in_ceramide-based_therapeutics_for_treatment_of_cancer

1. Introduction<!>2.1. Materials<!>2.2. Cell culture<!>2.3. Synthesis of triphenylbutene and didesmethyltamoxifen<!>2.4. Cell viability assays<!>2.5. Apoptosis assays<!>2.6. Measurement of mitochondrial reactive oxygen species<!>2.7. Cytochrome c release<!>2.8. Vitamin E experiments<!>2.9. C6-ceramide metabolism assays in intact cells<!>2.10. Data Analysis<!>3.1. Chemical structures of agents investigated with C6-ceramide<!>3.2. The effect of tamoxifen, tamoxifen metabolites, and other P-gp inhibitors on cytotoxic response to C6-ceramide in HL-60/VCR and KG-1a cells<!>3.3. Influence of C6-ceramide-adjuvant on mitochondrial ROS, impact of antioxidant, and effects on cytochrome c<!>3.4. The influence of P-gp inhibitors on C6-ceramide metabolism in intact cells<!>3.5. Effect of C6-ceramide-P-gp inhibitor combinations on apoptosis<!>3.6. Efficacy of C6-ceamide-zosuquidar in colorectal cancer cells<!>4. Discussion

<p>Ceramide, a potent tumor-suppressor sphingolipid [1–4], can be generated in situ by an array of anticancer drugs or administered exogenously, most prominently in the form of a short-chain ceramide, C6-ceramide [5, 6]. Whereas both avenues of enhancing ceramide levels are utilized, the sphingolipid-metabolizing machinery of cancer cells can function to dampen the tumor-censoring impact of this lipid. For example, metabolism of ceramide to glucosylceramide (GC) by glucosylceramide synthase (GCS) is a main route utilized by cancer cells to diminish ceramide-driven apoptosis- and autophagy-inducing responses [7, 8] . In addition, ceramide hydrolysis by ceramidases is an effective mode of ceramide elimination; however, this avenue can be problematic as sphingosine, produced via ceramidase activity, can be phosphorylated by sphingosine kinase (SK) to yield sphingosine 1-phosphate (S1-P), a mitogenic sphingolipid with an important role of its own in cancer biology [9, 10]. Maintaining a balance between ceramide and S1-P is thought paramount in maintaining the tumor-suppressor properties of ceramide. To this end, a number of pharmacologic and molecular approaches have been explored to improve ceramide's anticancer properties, approaches that encompass use of antisense oligonucleotides [11] as well as inhibitors of ceramide glycosylation and hydrolysis [12–16] . Of further importance, ceramide can be phosphorylated by intracellular ceramide kinase yielding ceramide 1-phosphate. This sphingolipid is also mitogenic and anti-apoptotic [17–19] , properties that would as well limit the tumor-suppressor actions of ceramide.</p><p>In several prominent studies of ceramide metabolism, GCS inhibitors have demonstrated efficacy and supported the idea that inhibition of ceramide glycosylation is an effective means to drive ceramide-orchestrated cancer cell death [1]. These inhibitors, often referred to as "P-drugs" include agents like D-threo-1-phenyl-2-decanoylamino-3-morpholino-propanol (PPMP), 1-phenyl-2-palmitoylamino-3-pyrrolidino-1-propanol (PPPP), and derivatives thereof [20]. One distinct agent, structurally and functionally divorced from the P-drugs that blocks GC synthesis in cancer cells is tamoxifen, a front-line breast cancer drug that functions as an estrogen receptor antagonist. In addition to inhibition of ceramide glycosylation [21], tamoxifen also exhibits a number of estrogen receptor-independent actions, including circumvention of multidrug resistance, downregulation of survivin, inhibition of Acyl-CoA: cholesterol acyl transferase (ACAT) [22], and downregulation of acid ceramidase [15]. The capacity to block ceramide glycosylation has made tamoxifen an object of myriad investigations into its use as an adjuvant with ceramide-centric therapies, including 4-HPR [23], short-chain ceramides [24], and short-chain ceramides in combination with paclitaxel [25]. Although tamoxifen is not a direct inhibitor of GCS, it limits intracellular production of GC by blocking GC transport into the Golgi, a process that requires Golgi-resident P-gp [22]. This interesting action well complements the long, enduring history of tamoxifen as a first generation P-gp inhibitor and modulator of multidrug resistance in cancer; tamoxifen interacts directly with P-gp but itself is not a substrate transport [26, 27].</p><p>Although tamoxifen and desmethyltamoxifen (DMT) have been shown effective in combination with C6-cermide in acute myeloid leukemia (AML) [28, 29], herein our aim was to discover alternatives to tamoxifen that would be void in antiestrogen activities. Additionally, having effective alternatives to tamoxifen would broaden the utility of ceramide as a cancer therapeutic.</p><p>The present work relates the discovery of a number of agents that are effective in combination with C6-ceramide and reveals commonalities in structure-function and in mechanism of action. Specifically, the most efficacious C6-ceramide-adjuvant-containing regimens blocked the metabolism of C6-ceramide via the glycosylation route and elicited the generation of reactive oxygen species (ROS). Importantly, these data suggest that specific P-gp inhibitors such as zosuquidar and tariquidar may find new utility when paired with ceramide-centric therapies as opposed to combining with standard, cytotoxic chemotherapies such as daunorubicin and vinblastine. In addition, that DMT is effective in combination with C6-ceramide is noteworthy, as this predominant tamoxifen metabolite in humans exerts < 1% of the antiestrogenic activity of parent tamoxifen [30], indicating that traditional anti-estrogen pathways are not involved in cellular responses. Of clinical relevance, we have previously shown that the C6-ceramide-tamoxifen combination is non-toxic in peripheral blood mononuclear cells, indicative of a cancer-selective action [28].</p><!><p>C6-ceramide (N-hexanoyl-D-erythro-sphingosine) and C6-NBD-lactosylceramide (N-[6-[(7-nitro-2-1,3-benzoxadiazol-4-yl)amino]hexanoyl]-D-lactosyl-β1-1′-sphingosine) were obtained from Avanti Polar Lipids, Alabaster, AL. C6-NBD-ceramide (N-[7-(4-nitrobenzo-2-oxa-1,3-diazole)]-6-aminocaproyl-D-erythro-sphingosine) was from Cayman Chemical Company, Ann Arbor, MI. N-Hexanoyl-NBD-glucosylceramide and N-hexanoyl-NBD-galactosylceramide were purchased from Matreya, State College, PA. NBD-X (6-(N-(7-nitrobenz-2-oxa-1,3-diazol-4-yl)amino) hexanoic acid) and NBD-C6-ceramide complexed to bovine serum albumin (BSA) were from Invitrogen, Carlsbad, CA. NBD-C6-sphingomyelin [N-(6-(7-nitrobenz-2-oxa-1,3-diazol-4-yl)amino)hexanoyl)sphingosine phosphocholine] was from Setareh Biotech, Eugene, OR. High performance TLC (HPTLC) glass plates, 10 cm × 10 cm, silica gel 60 matrix polymeric binder with fluorescent indicator (product number Z740223) were from Millipore-Sigma, Darmstadt, Germany. Calibrated micropipets were from Drummond Science Company, Broomall, PA. Tamoxifen-HCl, N-Desmethyltamoxifen-HCl, propidium iodide solution (PI), and Vitamin E (alpha-tocopherol) were purchased from Sigma, St. Louis, MO. Hank's Balanced Salt Solution (HBSS) was purchased from Thermo Scientific, Waltham, MA. MitoSOX™ Red Mitochondrial Superoxide Indicator for measuring ROS was purchased from Life Technologies, Carlsbad, CA. The pan-caspase inhibitor Z-VAD-fmk was a product R&D Systems, Minneapolis, MN. The Cell Titer 96 Aqueous One Solution Assay Kit (known as MTS) for determining cell viability was purchased from Promega (Madison, WI). LY335979 (Zosuquidar-3HCL), was from ApexBio, Houston, TX. Tariquidar (XR9576) and Fulvestrant (Faslodex) were purchased from AdooQ BioScience, Irvine, CA. Cyclosporin A and verapamil-HCl were purchased from Enzo Life Sciences, Farmingdale, NY. All experimental drugs were dissolved in 100% DMSO (Life Technologies, Carlsbad, CA) and stored as stock solutions (10 mM) at −20 °C. Plastic tissue culture items were from Falcon and Corning, and purchased from various suppliers.</p><p>Microwave (MW) irradiation experiments were carried out in a CEM Corporation (Matthews, NC) Discover monomode microwave operating system at a frequency of 2.45 GHz. The reactions were carried out in 10 mL glass tubes, sealed with a Teflon septum. 1H NMR spectra were recorded on a 500 MHz Bruker spectrometer (Boston, MA).</p><!><p>The human, vincristine-resistant (multidrug resistant) AML cell line, HL-60/VCR, was provided by A.R. Safa (Indiana University School of Medicine, Indianapolis, IN); cells were grown in medium containing 1.0 μg/mL vincristine sulfate (LC Laboratories, Woburn, MA). The KG-1a human AML cell line and the human colorectal cancer (CRC) cell line, LoVo, were obtained from the American Type Culture Collection (ATCC), Manassas, VA. Cells were cultured in RPMI-1640 medium (Life Technologies, Carlsbad, CA), supplemented with 10% fetal bovine serum (FBS), (Atlanta Biologicals, Atlanta, GA), and 100 units/mL penicillin and 100 μg/mL streptomycin (Life Technologies, Carlsbad, CA). The cell lines were not tested or authenticated over and above documentation provided by the ATCC, which includes antigen expression, DNA profile, short tandem repeat profiling, and cytogenic analysis. Cells were grown in humidified conditions in a tissue culture incubator with 95% air and 5% CO2, at 37 °C. Confluent LoVo cells were subcultured using 0.05% trypsin/0.53 mM EDTA solution (Invitrogen Corp, Carlsbad, CA). For experiments with HL-60/VCR cells, vincristine was removed from the medium.</p><!><p>But-1-ene-1,1,2-triyltribenzene (triphenylbutene, TØb, 3) was prepared as reported by Pathe and Ahmed [31] in good yields (Fig. 1A). Briefly, to a freshly prepared Zn-SnCl4 complex under N2 atmosphere, a mixture of benzophenone (1) and propiophenone (2) in THF was added slowly at same room temperature. Progress of the reaction was monitored by TLC and the reaction mixture was quenched with 10% aqueous NaHCO3 solution and extracted with ethyl acetate. After usual work up and column chromatography, the desired product 3 was obtained as yellow semi solid in 70% Yield, 1H NMR (CDCl3, 500 MHz) δ 7.38 (t, 2H, J=8.0Hz), 7.32-7.26 9m, 3H), 7.21-7.18 (m, 5H), 7.06-6.98 (m, 3H), 6.93-6.89 (m, 2H), 2.51 (q, 2H, CH2, J=7.5 Hz), 0.97 (t, 3H, CH3, J= 7.5Hz) (E/Z) N,N-Didesmethyltamoxifen (DiDMT, 6) was synthesized as reported [32] with minor modification (Fig. 1B). The key intermediate in the synthesis (E,Z)-1-(4-hydroxyphenyl)-1,2-diphenylbut-1-ene (5) was prepared from 4-hydroxybenzophenone (4) using super-base metalated propylbenzene [33]. Powdered potassium hydroxide (2.5 mmol) was added to a stirred solution of (E,Z)-1-(4-hydroxy)-1,2-diphenyl-1-ene (5) (0.5mmol) in dry toluene/dioxane (6:1, 7 ml) and mixture was irradiated in microwave at 90 °C (maximum power 250W) for 10 min. 2-Chloroethylamine hydrochloride (1.0 mmol) was then added and the mixture was irradiated in microwave at 90°C (maximum power 250W) for additional 30 min. After the usual work up as reported [29], the desired compound 6 was obtained in 68% yield as a mixture of E & Z isomers (30:70); 1H NMR (CDCl3, 500 MHz) 7.37 (t, 1H, J=7.5 Hz), 7.31-7.24 (m, 3H), 7.23-7.09 (m, 10H), 7.05-6.98 (m, 3H), 6.94-6.86 (m, 4H), 6.80 (d, 1H, J=8.5Hz), 6.58 (d, 1H, J=9.0Hz), 4.052 (t, 1.38H, Z-CH2-NH2, J=5.0 Hz), 3.89 (t, 0.62H, E-CH2-NH2, J=5.0 Hz), 3.15 (br s, 1.38H, Z-NH2), 3.05 (br s, 0.62H, E-NH2), 2.53 (q, 1.40H, Z-CH2-CH3), 2.47 (q, 0.60H, E-CH2-CH3), 0.97 (t, 3H, CH3, J=9.0 Hz).</p><!><p>For viability by propidium iodide (PI), HL-60/VCR cells were seeded in black-wall 96-well plates at 100,000 cells/well and treated with indicated drugs in 0.2 mL 5% FBS RPMI-1640 medium. After addition of agents, cells were incubated at 37 °C, 5% CO2 for 24 h, and viability was determined using PI as follows. A positive cell control was first permeabilized by addition of 10 μL of 1 mg/mL digitonin and incubated at 37 °C, 5% CO2 for 20 min, followed by the addition of 0.1 mL of a 3X PI solution in 1X PBS for a final well concentration of 5 μM PI. The plate was then incubated for an additional 20 min, and viability was calculated as the mean (n = 4 or n = 6) fluorescence (minus permeabilized vehicle control) at 530 nm excitation and 620 nm emission, using a Bio-Tek Synergy H1 microplate reader, BIO-TEK Instruments (Winooski, VT). The effect of the pan-caspase inhibitor Z-VAD-fmk on HL-60/VCR cell viability in response to C6-ceramide-P-gp inhibitor regimens was evaluated by pre-incubating cells for 2 h with the inhibitor ( 20, 40, 50 μM) prior to addition of drugs.</p><p>LoVo cell viability was assessed using the CellTiter 96 One Solution Cell Proliferation Assay Kit (MTS) following manufacturer instructions. Cells were seeded in 96-well plates at 10,000 cells/well in 0.1 mL complete medium and allowed to attach at 37 °C, 5% CO2 for 24 h before adding drugs. Drugs were diluted freshly into culture medium containing 1% FBS and added to wells to a total volume of 0.2 mL, thus the final concentration of FBS during treatment of LoVo cells was 5.5%. Viability was calculated as the mean (n = 4 or n = 6) absorbance (minus vehicle control) at 490 nm, using a Bio-Tek Synergy H1 microplate reader.</p><!><p>Apoptosis was detected by flow cytometry using the ApoDETECT Annexin V-FITC Kit (Life Technologies, Carlsbad, CA), following the manufacturers protocol. Briefly, cells were seeded in 6-well plates at 5 × 105 cells/mL in 5 mL of RPMI-1640 medium containing 5% FBS. Cells were treated with the indicated drugs for 18 h, collected by centrifugation and washed with PBS and stained with Annexin V using 1X Annexin binding buffer (provided in the kit), and the percent of Annexin V-positive cells was determined by flow cytometry. Cell acquisition was performed on a Becton Dickinson FACSCalibur. Analysis was performed using FCS Express 4 from De Novo Software (Glendale, CA). Apoptosis was also determined by flow cytometric analysis of DNA fragmentation following our previously published protocol [34].</p><!><p>Mitochondrial superoxide was assayed using MitoSOX™ Red. Cells (5 × 105 cells/mL RPMI-1640, 5% FBS medium) seeded in 6-well plates, were treated with the indicated drugs for 18 or 24 h and then collected (adherent cells were collected using trypsin) and washed in HBSS and incubated in 0.25 mL staining buffer (HBSS containing 5 μM MitoSOX) for 15 min at 37°C, protected from light. Cells were washed again in HBSS, resuspended at 1 × 106 viable cells/mL in HBSS, and a 0.1 mL aliquot was added to the wells of black-walled 96-well plates. Fluorescence was measured at 510 nm excitation and 580 nm emission, using a Bio-Tek Synergy H1 microplate reader. For photomicrographs, a 0.1 mL aliquot of cells was added to black-walled 96-well plates, and after centrifugation images were captured using fluorescence microscopy.</p><!><p>Cytochrome c release from mitochondria was assessed as described [28, 35, 36]. Briefly, 2 × 106 cells/2 mL RPMI-1640 medium, 2.5% FBS medium, were seeded in 6-well plates and treated with selected agents for 18 h, after which 1 × 106 cells were removed and placed on ice in 0.1 mL of digitonin (Sigma, St. Louis, MO) (100 μg/mL in PBS, 100 mM KCl) for 3–5 min, until 95% of the cells were permeabilized (stain positive with 0.2% trypan blue). At this point, 0.1 mL of 4% paraformaldehyde was immediately added to the cells. After centrifugation, cells were then fixed at room temperature in 4% paraformaldehyde in PBS for 20 min, washed with PBS and resuspended in blocking buffer (PBS, 3% BSA, 0.05% saponin); the saponin (Sigma, St. Louis, MO) was freshly prepared. Cells were then incubated for 30 min at 4°C in a 1 : 100 dilution of FITC conjugate (6H2) anti-cytochrome c antibody (Life Technologies, Carlsbad, CA) in blocking buffer, washed, and levels of cytochrome c were determined by flow cytometry as described in the above sections.</p><!><p>Vitamin E was prepared fresh before each experiment. A 250 mM stock solution of vitamin E in 100% ethanol was diluted to 25 mM in 1.0 M HEPES buffer, pH 7.3; this working solution was used in experiments. To assess the effect of vitamin E, cells were preincubated in culture medium containing 250 μM vitamin E for 2 h prior to addition of test agents.</p><!><p>A modified method of a previous procedure was followed [37]. Briefly, control and 18 h inhibitor-pretreated HL-60/VCR cells (1 × 106 cells/mL RPMI-1640 medium containing 5% FBS, 6-well plates) were assessed for viability using trypan blue and seeded into 96-well strip wells at 100,000 viable cells/45 μL serum-free RPMI-1640 containing 1% BSA. LoVo cells were seeded at 100,000/well, 96-well plate, and allowed to attach overnight before pretreatment with inhibitor for 18 h. LoVo cells were also seeded in parallel to assess viability after pretreatment with the inhibitors. The enzyme reaction (inhibitors were present during the assay) was initiated by addition of 5 μL NBD-C6-ceramide complexed to BSA (25 μM final substrate concentration) and placed in a tissue culture incubator for 2 h. Samples were then placed on ice, and the cells were transferred to 1-dram glass vials for lipid extraction [38]. The lower, lipid-containing chloroform phase was evaporated to dryness under a stream of nitrogen. Total lipids were dissolved by addition of 40 μL chloroform/methanol (5:1, v/v), vortex mixed, and 5 μL was applied to the origin of a HPTLC plate. Commercial lipid standards were spotted in lateral lanes. Lipids were resolved in a solvent system containing chloroform/methanol/ammonium hydroxide (80:20:2, v/v/v). Products were analyzed directly on the HPTLC plates using the BioRad ChemiDoc Touch and quantified with Image Lab software by BioRad (Hercules, CA).</p><!><p>Results are expressed as means ± SEM and were analyzed by ANOVA. Differences among the treatment groups were assessed by Tukey post hoc test. Differences were considered significant at *P ≤ 0.05, **P ≤ 0.01, ***P ≤ 0.001. Drug-induced cytotoxic synergy was analyzed by CalcuSyn® software from Biosoft (Great Shelford, Cambridge, United Kingdom). Each condition (single agent and combination) was tested in replicates of six and repeated at least twice. The mean proliferative index for each compound at the indicated concentrations was entered into the CalcuSyn program for dose–effect analysis. By this method, a combination index (CI) is determined based on the Chou–Talalay method [39] where a CI of 0.9–1.10 indicates an additive effect, and CI values of 0.3–0.7 and 0.1–0.3 indicate synergism and strong synergism, respectively.</p><!><p>The chemical structures of tamoxifen, tamoxifen metabolites, and the other agents utilized in this study are shown in Fig. 2. Tamoxifen contains a TØb nucleus and an N-dimethylethanolamine function. The corresponding desmethylated metabolites, DMT and Di-DMT contain one and no methyl groups, respectively. TØb, the aromatic nucleus, is devoid of the N-dimethylethanolamine moiety. Although an antiestrogen, tamoxifen has been widely utilized as a P-gp inhibitor in clinical studies [22] and is also a component of the Dartmouth regimen for treatment of melanoma [40]. It is termed a first generation P-gp inhibitor. In addition to tamoxifen, we studied two other first generation P-gp inhibitors, the nonpolar, cyclic oligopeptide immunosuppressive, cyclosporin A, and the vasodilator and calcium channel blocker, verapamil (Fig. 2). Zosuquidar, a cyclopropyldibenzosuberane, and the anthranilamide derivative, tariquidar (Fig. 2), are third generation P-gp inhibitors and despite the diverse chemical structures and origins, these agents demonstrate high potency and specificity for the P-gp transporter, and typically third generation inhibitors do not inhibit other ABC transporters, such as the multidrug resistance protein, MRP1 (ABCC1). Fulvestrant (trade name Faslodex), a synthetic estrogen receptor antagonist, like tamoxifen, is composed of a steroid nucleus and is thus chemically distinct from the other agents under evaluation.</p><!><p>As shown in Fig. 3A, after a 24 h exposure, HL-60/VCR cells were refractory to C6-ceramide. In turn, tamoxifen and metabolites DMT, DiDMT, and the tamoxifen nucleus, TØb, imposed limited cytotoxicity when administered singly. However, sensitivity to C6-ceramide was strongly enhanced by addition of either tamoxifen or the tamoxifen metabolites. For example, combination C6-ceramide-tamoxifen, -DMT, and -DiDMT exposure resulted in 60%, 70% and 75% cell death, respectively, whereas the tamoxifen nucleus, TØb, was completely devoid of C6-ceramide-enhancing activity, and thus served as an appropriate negative control. Tariquidar, cyclosporin A, and zosuquidar each demonstrated efficacy when co-administered with C6-ceramide (Fig. 3B); when administered singly however, these agents were only moderately cytotoxic. For example, whereas tariquidar elicited approximately 20% cell death over control, the C6-ceramide-tariquidar regimen elicited >50% cell death. C6-ceramide-cyclosporine A and C6-ceramide-zosuquidar combinations produced like cytotoxic responses in HL-60/VCR cells. Much in contrast, verapamil and fulvestrant were devoid of C6-ceramide-enhancing cytotoxicity (Fig. 3C). To determine possible synergy, we investigated one of the nontamoxifen-related compounds, zosuquidar. Based on the Chou-Talalay method [39], combination indexes (CI) were determined. By this method, a CI of 1.0 indicates an additive effect, and CI's of 0.7–0.85, 0.3–0.7, and 0.1–0.3 indicate moderate synergism, synergism, and strong synergism, respectively. As shown in Fig. 3D, (left, isobologram; right, in tabulated data) all combinations, with the exception of number 1, demonstrated synergy as denoted by the CI's. Combination 6, yielding a CI of 0.3, was strongly synergistic. We next sought to determine the efficacy of C6-ceramide-adjuvant regimens in KG-1a, a human AML cell line that exhibits characteristics of CD34+CD38− leukemia stem cells [41–43], which are considered to be the principal reason for relapse. As shown in Fig. 3E, tamoxifen and zosuquidar were effective in reducing KG-1a cell viability when administered with C6-ceramide, albeit these assays were conducted over 72 h as opposed to the 24 h exposures used in the HL-60/VCR experiments.</p><!><p>Mitochondria play an important role in cancer cell biology. In addition, generation of mitochondrial ROS can have a potential therapeutic role in AML [44]. Thus, we next sought to investigate the effects of the C6-ceramide-adjuvant on the production of mitochondrial ROS. The data in Fig. 4A demonstrate a positive correlation between effective cytotoxic regimens and ROS generation. For example, whereas the C6-ceramide-verapamil regimen did not influence ROS levels, over control, the tamoxifen- , DMT-, and zosuquidar-containing combinations imparted approximate 1.7-, 3.7-, and 2.5-fold increases over control in ROS levels. Thus generation of ROS appears to be a common denominator in the cytotoxic action of the effective regimens. Noteworthy, short-time, 12 h exposure to combination C6-ceramide-DMT (10 + 10 μM), promoted a 3-fold increase in ROS levels, while cell viability remained at 80% of control (data not shown). Thus, temporally, ROS generation precedes the full brunt of this regimen's cytotoxic impact. To determine whether antioxidant could protect HL-60/VCR cells from C6-ceramideadjuvant-induced insult and thus implicate ROS in the underlying cytotoxic mechanism, we evaluated the effects of vitamin E. The data in Fig. 4B show that cytotoxicity imparted by adjuvants that were effective in combination with C6-ceramide, tamoxifen, DMT, DiDMT, cyclosporin A, and zosuquidar, was mitigated by pre-exposure to vitamin E. For example, pretreatment of cells with Vitamin E diminished C6-ceramide-DiDMT-induced cytotoxicity from approximately 65 to 20% cell death, compared to drug-free control. Vitamin E protection from combination C6-ceramide-adjuvant (-tamoxifen, -DMT, -zosuquidar) was verified using the MTS cell viability assay method (data not shown). Further mitochondrial targeting was evidenced by evaluating cytochrome c release, a measure of mitochondrial injury and apoptotic signaling; both tamoxifen- and zosuquidar-containing C6-ceramide regimens promoted release of cytochrome c (Fig. 4C, flow cytometry histogram, left; quantitation, right).</p><!><p>Tamoxifen has previously been shown an effective inhibitor of ceramide glycosylation in cancer cells [21]. As several of the P-gp inhibitors tested coactively enhanced C6-ceramide cytotoxicity, it was of interest to determine whether there would be a correlation between this enhancement and the effects of the inhibitors on C6-ceramide metabolism. Firstly, the thin-layer chromatogram in Fig. 5A, control lane, shows that glycosylation (generation of NBD-C6-GC) and conversion to sphingomyelin (NBD-C6-SM) are the major pathways of NBD-C6-ceramide metabolism in HL-60/VCR cells, whereas lower levels of metabolism via hydrolysis were observed (NBD-hex, hexanoic acid, indicative of ceramidase activity). It is important to note that the NBD-C6-ceramide substrate spot is prominent because it consists of both extracellular and intracellular NBD-C6-ceramide; in these experiments the total incubation mixture, cells plus media, was extracted. Figure 5A also clearly demonstrates the impact of various P-gp inhibitors on NBD-C6-ceramide metabolism. Noteworthy, tamoxifen, DMT, and zosuquidar, all of which were synergistic with C6-ceramide (see Fig. 3), inhibited NBD-C6-GC production by 75, 63, and 83%, respectively (Fig. 5B, quantitation of chromatogram in Fig. 5A), whereas Fulvestrant was less effective, 28% inhibition. Verapamil had no significant inhibitory effects, whereas TØb, the tamoxifen nucleus, was slightly stimulatory. Additionally, tamoxifen, DMT, and verapamil caused only a modest reduction in the conversion of NBD-C6-cermide to NBD-C6-SM, 24, 21, and 22% inhibition, respectively; however, zosuquidar blocked synthesis by 67% (Fig. 5A, B). Fulvestrant also inhibited NBD-C6-SM synthesis by 53%, similar with zosuquidar. Of note, zosuquidar, a specific, high affinity P-gp inhibitor (Ki 60 – 80 nM), at 0.1 μM significantly depressed NBD-C6-GC synthesis; the inhibition was dose-dependent (Fig. 5C). Zosuquidar, which is cytotoxic in combination with C6-ceramide in KG-1a cells (see Fig. 3E), also strongly inhibited NBD-C6-ceramide glycosylation and NBD-C6-ceramiade conversion to NBD-C6-SM by 85 and 100%, respectively, in KG-1a cells (Fig. 5D).</p><!><p>We next investigated several of the agents that were co-actively effective with C6-ceamide to determine whether apoptosis was a factor underlying cytotoxic responses. The data in Fig. 6A, flow cytometry histograms, show that combination C6-ceramide-tamoxifen, -DMT, and–zosuquidar effectively induced apoptosis that was well above control values. All combinations, with the exception of the verapamil-containing regimen, elicited apoptosis that measured 55–65%, a significant increase above the 10% in control cultures (Fig. 6B). With the C6-ceramideverapamil combination, apoptosis measured approximately 20% over control values. To confirm apoptotic responses, we utilized flow cytometry to evaluate DNA fragmentation. Results by this method verified that the C6-ceramide-P-gp inhibitor regimens elicited apoptotic cell death (data not shown). We then employed the pan-caspase inhibitor, Z-VAD-fmk to determine whether apoptosis was caspase-dependent. Results demonstrated that inclusion of caspase inhibitor at 20, 40, 50 μM failed to reverse cytotoxic responses, suggesting that these drug regimens induce caspase-independent apoptosis, a topic of relevance in cancer therapy [45–47] .</p><!><p>The effect of zosuquidar on cytotoxic response to C6-ceramide in LoVo cells was investigated in order to determine whether this ceramide-based pharmacological approach could be of broader utility. Firstly, the thin-layer chromatogram in Fig. 7A, left, and quantitation, Fig. 7, right, demonstrate that zosuquidar inhibited NBD-C6-ceramide glycosylation in a dose-dependent manner in LoVo cells. For example, exposure to 1 μM zosuquidar inhibited glycosylation by >20%. Also noteworthy, zosuquidar inhibited the synthesis of NBD-C6-SM, dose-dependently (Fig, 7A). Accordingly, combination C6-ceramide-zosuquidar sturdily reduced LoVo cell viability, whereas exposure to the single agents had limited impact (Fig. 7B). Closer examination reveals that the combination regimens, at the concentrations shown (Fa, fraction affected), affected 60, 80, 87, and 79% of the cell populations, as detected by isobologram analysis (Fig. 7C, left); combinations 1, 2, 3, and 4 produced in CI's ranging from 0.390 to 0.265, indicating synergism and strong synergism (Fig. 7C, right). Analysis of NBD-C6-ceramide levels and metabolites in LoVo cells, after cells had been washed free of extracellular substrate, provided a better assessment of intracellular only events. These experiments revealed that approximately 50% of the ceramide that had been taken up was metabolized to GC and sphingomyelin (SM) (Fig. 7D, control lane), effectively lowering the intracellular concentration of NBD-C6-ceramide (control lane, top spot); however, in the presence of either tamoxifen or zosuquidar, this metabolic conversion was largely blocked, as shown by the reduction in intensity of NBD-C6-GC and NBD-C6-SM spots (Fig. 7D). This resulted in markedly higher levels of free, intracellular NBD-C6-ceramide, 77 and 79% of total metabolites in tamoxifen-and zosuquidar-treated cells, respectively (Fig. 7D). Thus, in cells with P-gp inhibitors, higher levels of free ceramide (NBD-form) are maintained. Lastly, in LoVo as in HL-60/VCR cells, C6-ceramide-tamoxifen and C6-ceramide-zosuquidar combinations were effective generators of ROS (Fig. 7E, photomicrographs, left; bar graph quantitation, right).</p><!><p>Although tamoxifen has previously been shown to enhance ceramide-driven cancer cell death [24], little is known regarding the structure-activity relationship (SAR) supporting this response, nor, for purposes of broadening therapeutic application, have alternatives to tamoxifen been explored. Knowledge of the SAR underlying these responses would provide a useful guide for the rational engineering of potent ligands based on the chemical structure of tamoxifen. Results herein illustrate the close relationship of P-gp-interacting drugs with enhancement of C6-ceramide-elicited cytotoxicity and with inhibition of C6-ceramide metabolism. Although it is possible that the effects of P-gp inhibitors on C6-ceramide cytotoxicity are related to inhibition of efflux, we have previously demonstrated that C6-ceramide is not a substrate for P-gp and that P-gp inhibitors do not influence retention of C6-ceramide [48]. This work documents discovery of alternatives to tamoxifen that synergistically enhance ceramide-driven apoptosis and reveals insight into the mechanism of action underlying the cytotoxic responses.</p><p>Golgi-resident P-gp functions as a SM and GC transmembrane flippase [49, 50]. This finding spawned the idea that P-gp played a role in sphingolipid metabolism, and thus could be involved in regulating ceramide levels and therefore, ceramide sensitivity. Studies of Shabbits and Mayer [51] and Smyth et al [52] bolstered this notion. The idea that P-gp protects cells from ceramide cytotoxicity was supported by studies in HeLa cells that conditionally express P-gp. Those works showed that P-gp-expressing cells were resistant to ceramide, whereas the P-gp-devoid counterpart was ceramide-sensitive [48]</p><p>With regard to tamoxifen and metabolites, that DMT also enhanced response to C6-ceramide in HL-60/VCR cells is of added clinical relevance, as this tamoxifen metabolite is a poor antiestrogen [30] and would therefore not impact estrogen receptor-related biology. The efficacy of DMT also demonstrates that responses are not linked to traditional antiestrogen pathways. Further, that fulvestrant, a specific estrogen receptor antagonist, was without effect, also underscores that the C6-ceramide-enhancing activities of tamoxifen and metabolites are divorced from estrogen receptor jurisdiction. DiDMT is also a metabolite of tamoxifen in humans [53, 54] ; however, its efficacy as a modulator of MDR is not known. Our work is the first to demonstrate that DiDMT enhances ceramide's pro-apoptotic effects. The tamoxifen nucleus, TØb, was devoid of C6-ceramide-enchancing activity; this highlights structural specificity and requirement for the dimethylethanolamine moiety, although the methyl groups do not appear to be a requirement. Perhaps the slight increases in activity noted with DMT and DiDMT (see Fig. 3A, tamoxifen, DMT, DiDMT), compared to tamoxifen, result from enhanced uptake of these desmethylated tamoxifen metabolites.</p><p>The non-tamoxifen-related drugs, cyclosporin A, tariquidar, and zosuquidar, were effective enhancers of C6-ceramide cytotoxicity. Cyclosporin A, a substrate for P-gp, blocks the pumping of drugs in a competitive manner [55] and inhibits drug-activated and basal ATPase activity of P-gp [56]. Tariquidar, a potent inhibitor of P-gp [57, 58], but also a substrate and inhibitor for breast cancer resistance protein (BCRP/ABCG2) [59], shows a noncompetitive interaction with P-gp substrates and inhibits the ATPase activity of P-gp [60]; it could thus be considered to have an allosteric effect on substrate recognition or ATP hydrolysis. There is also a report showing that tariquidar inhibits P-gp drug efflux by blocking transition to the open state during the catalytic cycle [61]. Tariquidar has been evaluated in Phase I and in Phase II studies [62, 63]. Zosuquidar, a high affinity (Ki = 60–80 nM) P-gp competitive inhibitor does not inhibit other members of the ATP-drug binding transporter family, such as MRP and BCRP [64], and lacks pharmacokinetic interactions often seen with other MDR inhibitors that alter the plasma concentration of co-administered oncolytic agents. Zosuquidar restores drug sensitivity in P-gp-expressing AML cells [65], and is generally well tolerated, as evaluated in Phase I trials in patients with advanced malignancies [66, 67]. The agent can be given safely to patients with AML in combination with induction doses of conventional cytotoxic drugs [68]; however, zosuquidar did not improve outcome in older AML in part because of the presence of P-gp-independent mechanisms of resistance [69]. Of note, however, results in a study by Lancet et al [70] in P-gp positive patients indicated that pre-administration of zosuquidar followed by continuous infusion, prior to daunorubicin administration was well tolerated and able to completely inhibit P-gp function.</p><p>Verapamil and fulvestrant were ineffective in combination with C6-ceramide. Verapamil, a calcium channel blocker and a well-known, much utilized first generation P-gp modulator, is a P-gp substrate that inhibits drug transport in a competitive manner; it also stimulates P-gp ATPase activity [29, 71]. Fulvestrant is a pure estrogen receptor antagonist, which unlike tamoxifen, works both by down-regulating and by degrading the estrogen receptor; it binds competitively to the estrogen receptor in cells. The agent has been shown to inhibit P-gp function and subsequently reverse P-gp-mediated drug resistance in a breast cancer model [72]; however, its history as MDR modulator is truly limited.</p><p>The agents demonstrating efficacy with C6-ceramide are inhibitors of P-gp function; verapamil, at the concentration employed, was the exception. It is perhaps noteworthy that the C6-ceramide-verapamil regimen also failed to activate ROS generation (see Fig. 4A) and weakly induced apoptosis when compared to the other combination regimens (see Fig. 6B). Another commonality among agents that demonstrated synergy with C6-ceramide was the capacity to inhibit C6-ceramide glycosylation (see Fig. 5), in this instance tamoxifen, DMT, and zosuquidar were potent inhibitors. Thus, we propose that the P-gp inhibitors amplify the ceramide effect in part by contributing to preserve high levels of intracellular C6-ceramide, via metabolic blockade. This effect is clearly illustrated in LoVo cells (see Fig. 7E) where both conversion to GC and SM were severely compromised, a maneuver that may implement and enforce ROS generation and cellular demise. That cytotoxicity was reversed by exposure to a ROS scavenger, vitamin E, demonstrates that oxidative injury may play a vital role in the cytotoxic response.</p><p>We have recently demonstrated that combination C6-ceramide-tamoxifen promotes a decrease in mitochondrial membrane potential and inhibits complex I respiration in KG-1 AML cells [73]. As damaged mitochondria stimulate increased ROS production [74], we propose that the active combinations presented herein share mitochondria as a common target and employ ROS. The capacity of ROS to drive tumor cell death has been exploited as an avenue in cancer therapy. Along these lines, leukemia stem cells (LSC) [75], thought to play a pivotal role in relapse and in the refractory nature of AML, become an attractive target. Taking into account the unique nature of self-renewing LSCs, the ceramide-containing-P-gp-inhibitor-containing regimens may be of unique utility. Firstly, LSCs express high levels of P-gp, and these cells usually reside in a quiescent state, making agents that target cell cycle ineffective. As well apropos in the current context, oxidative stress inhibits self-renewal in LSCs [76]. Thus, agents that simultaneously enhance the apoptotic impact of ceramide via P-gp modulation and induce oxidative stress should target LSCs effectively. Our results in KG-1a cells indicate that the C6-ceramide-containing combinations could be effective against LSC. That the C6-ceramide-P-gp inhibitor regimens were also effective in LoVo, a colon cancer model, broadens the potential therapeutic utility of such an approach.</p>

PubMed Author Manuscript

N-Arylphenothiazines as strong donors for photoredox catalysis – pushing the frontiers of nucleophilic addition of alcohols to alkenes

A new range of N-phenylphenothiazine derivatives was synthesized as potential photoredox catalysts to broaden the substrate scope for the nucleophilic addition of methanol to styrenes through photoredox catalysis. These N-phenylphenothiazines differ by their electron-donating and electron-withdrawing substituents at the phenyl group, covering both, σ and π-type groups, in order to modulate their absorbance and electrochemical characteristics. Among the synthesized compounds, alkylaminylated N-phenylphenothiazines were identified to be highly suitable for photoredox catalysis. The dialkylamino substituents of these N-phenylphenothiazines shift the estimated excited state reduction potential up to −3.0 V (vs SCE). These highly reducing properties allow the addition of methanol to α-methylstyrene as less-activated substrate for this type of reaction. Without the help of an additive, the reaction conditions were optimized to achieve a quantitative yield for the Markovnivkov-type addition product after 20 h of irradiation.

n-arylphenothiazines_as_strong_donors_for_photoredox_catalysis_–_pushing_the_frontiers_of_nucleophil

<!>Introduction<!>Results and Discussion<!><!>Results and Discussion<!><!>Results and Discussion<!><!>Results and Discussion<!><!>Results and Discussion<!><!>Results and Discussion<!>Conclusion<!>

<p>This article is part of the thematic issue "Photoredox catalysis for novel organic reactions".</p><!><p>Visible-light photoredox catalysis has become a precious tool in modern synthetic organic chemistry and experiences a continuously growing interest in industrial applications. The access to electronically excited states of organic molecules allows unlocking new and sometimes complementary chemical reactivities that cannot be tackled by using thermally driven chemical reactions [1]. This complementarity allows for the development of so far unknown transformations [2]. The photochemical reactivity can be tuned by the absorption and excited state characteristics of the photocatalyst. In this context, organic dyes represent a perfectly suited class of photocatalysts as they can easily be modified by the introduction of functional groups that allow fine-tuning the optoelectronic properties of the molecules. Photochemical methods have allowed to overcome some of the current limitations in thermally driven chemistry and to substitute conventional energy demanding chemistry by highly sustainable photochemical methods [3–12].</p><p>Phenothiazines have become a precious class of organic molecules, not only due to their widespread use in medicinal chemistry [13] but also because of their fascinating electronic properties. Recently their use in photoredox catalysis allowed for the development of some novel transformations, namely dehalogenation [14] as well as the first pentafluorosulfanylation method starting from sulfur hexafluoride [2]. We are convinced that the value of phenothiazine derivatives in photoredox catalysis is still underestimated. While these compounds found widespread use in ATRA (atom transfer radical addition) polymerization [15–16] the interest of using this class of catalysts only gained limited interest during the last years. The advantage of using N-phenylphenothiazine catalysts in photoredox chemistry is attributed to their beneficial redox properties. Moreover, a modification of the core is rather simple and allows fast access to a wide variety of catalysts. Recently it was shown that the radical cation of the photoredox catalyst can play a key role in photoinduced oxidation chemistry [16]. This is rather unusual due to the usually short lifetime of radical cations in solution attributed to their low-lying excited states. Normally, this is the reason why photochemical processes can hardly compete with photophysical decay processes. However, a pre-coordination of the substrate may facilitate electron transfer under non-diffusional controlled conditions. Very recently, the fast (picosecond) excited state dynamics of the radical cation of N-phenylphenothiazine was investigated by Wasielewski et al. This radical cation had a high reduction potential of about +2.1 V (vs SCE) [17] allowing the reduction of poorly oxidizing agents. The combination of both properties in one system is a remarkable feature for chemical redox dynamics between −2.1 V up to +2.1 V. One of the key problems in photochemistry, which was recently addressed by the development of the consecutive photoelectron transfer process (conPET) [5], is the need to push the frontiers by accessing high reduction potentials. While the classical photoredox concept is based on the photophysical properties of the excited photoredox catalyst, the idea of the conPET concept mimics nature's light collection system and consecutively collects the energy of two photons stored in the excited state of the initially pre-promoted photoredox catalyst's radical ion. This was one of the features to use N-phenylphenothiazine for the photoactivation of SF6 for the pentafluorosulfanylation of styrenes [2]. This two photon concept can further be extended to the photoredox catalytically generation of hydrated electrons as very powerful reductants (E = −2.8 V (vs NHE)) for organic reactions [18].</p><p>During the last years, we investigated the photoredox chemistry of new classes of catalysts like perylene bisimides, for their suitability in these types of processes [19] and evaluated the addition of methanol to alkenes as a simple model system. Due to the insufficient reduction potential of the photoredox catalyst, the Markovnikov addition of alcohols through oxidative quenching is yet limited to highly activated, aromatic alkenes. To the best of our knowledge no methods are known today that allow the addition of alcohols to α-methyl-substituted styrenes through photoredox catalysis. The currently available methods are based on a two-step procedure involving an iodoalkoxylation with NIS followed by the reduction of the formed alkyl iodide generating the product in moderate yields [20], or through the direct addition of MeOH catalyzed by either acidic conditions or heated ion exchange resin [21–22]. These methods are therefore not suitable for the alkoxylation of acid or base-labile substrates. To overcome the current limitations of reduction potentials of single electron transfer processes in photoredox catalysis we present herein a range of new N-phenylphenothiazine derivatives 1–11 as photoredox catalysts. Three of them were identified to be highly suitable for the addition of methanol to alkenes affording the corresponding Markovnikov products.</p><!><p>Activated aromatic olefins, such as 1,1-diphenylethylene (12), have reduction potentials Ered(S/S−·) in the range of −2.2 to −2.3 V [23–24], α- and β-methylstyrene (13a and 13b) have an Ered(S/S−·) of −2.5 to −2.7 V [25], and styrene (14) an Ered(S/S−·) of −2.6 V (Figure 1) [25–26]. For non-aromatic, alkylated olefins, like 1-methylcyclohex-1-ene (15), the reduction potentials are estimated to values of Ered(S/S−·) = −3.0 V [25]. In our initial photoredox catalyst screening [26], we identified 1-(N,N-dimethylamino)pyrene (16) having an excited-state reduction potential E*ox(P+·/P*) of −2.4 V (determined by cyclic voltammetry and E00). Thus, we are able to photoreduce 1,1-diphenylethylene (12), but not yet α-methylstyrene (13), and clearly not non-aromatic (alkylated) olefins, such as methylated cyclohexene 15, as basic structures. The absorption of N-phenylphenothiazine (1) disappears at around 390 nm. This feature requires the excitation of the molecule using UV light sources and contradicts the use of visible light irradiation due to vanishing extinction coefficients at the edge to the visible region. To reach for high excited state reduction potentials and excite at rather long irradiation wavelengths an energetically high lying electronic groundstate potential has to be connected with a small S0–S1 gap for the development of strongly reducing photoredox catalysts. Thus, we first focused our strategy in catalyst development on the synthesis of some highly electron-rich phenothiazines 2–5 as well as some electron-deficient phenothiazines 6–9 to analyze the influences of modifications of the core and the aryl moiety (Figure 2). The observed trends allowed us to come up with a set of strongly reducing photoredox catalysts that operate under UV-A conditions close to the visible range. In order to extend the scope of available reduction potentials we expected that the N-phenylphenothiazine core having installed additional electron-donating groups, like NR2 in 2, reaches very low reduction potentials in the range of E(P+·/P*) = −2.5 to −3.0 V which is in the range of solid sodium [27], that would be able to attack low-substituted styrenes like 13a and 13b.</p><!><p>Reduction potentials (vs SCE) of common photoredox catalysts, pyrene 16 and phenothiazine 2, in comparison to addressable substrate scope 12–15. Bottom: photoredox catalytic addition of MeOH to α-methylstyrene (13a).</p><p>Acceptor or donor-modified phenothiazines 1–11 as potential photoredox catalysts.</p><!><p>Firstly, the absorbance characteristics of the derivatives 1–9 were analyzed and compared (Figure 3). The parent compound N-phenylphenothiazine (1) shows an absorption maximum at 320 nm. Substitution of the arene moiety results in a shift of the absorption maxima due to a change in the HOMO–LUMO gaps. It turned out that the introduction of the π-donating dimethylamino substituent in 2 induces a hypsochromic shift of the absorption maximum by 7 nm to 313 nm, while the mesityl group present in 5 as a σ-donor causes a bathochromic shift of about 8 nm. The detailed structure of the alkyl group attached to the amino part in the phenothiazines 2, 10 and 11 showed no significant change in the absorption maximum of the S1 transition (10 in comparison to 2), but replacement of the methyl groups by branched isobutyl groups in 11 resulted in a hypsochromic shift of the bathochromic features of absorption. The nitro compound 6 turned out to show a distinct long wavelength absorption that is apart from the region of absorption of all other catalysts by a shift of about 40 nm which is probably due to a charge transfer state. Interestingly, the spectrum of the methylpyridine derivative 9 showed a rather short absorption maximum at 302 nm.</p><!><p>Normalized UV–vis absorption spectra above 290 nm of N-phenylphenothiazines 1–11 (left) and representative cyclic voltammogram of 2 (right).</p><!><p>All N-phenylphenothiazines 1–11 were additionally characterized by cyclic voltammetry (Figure 3 and Table 1) [28]. We found the first oxidation half wave of the unsubstituted N-phenylphenothiazine (1) as a fully reversible process as it was described in literature recently [18]. The second oxidation process of 1 is almost reversible but induces to some extent an irreversible oxidation process that shows up as further reduction half wave in the cyclic voltammogram. This is true for almost all synthesized derivatives 3–9. The radical dication is known to undergo disproportionation reactions [27], which potentially explain the results. Only the amino derivatives 2, 10 and 11 managed to undergo a second completely reversible oxidation process. To exclude interference with water and oxygen the measurements were carried under strict exclusion of any contaminants. The first oxidation of the lead structure 1 was found to occur at E(1+·/1) = 0.75 V (vs SCE). The substitution of the arene moiety by one (see 7) or two fluorine substituents (see 8) only leads to a shift in the reduction potential of about 0.06 V. This trend was expected due to the lower electron density of these two N-phenylphenothiazines at the arene moiety. However, the effect by the pure σ-acceptor fluorine is not very pronounced. In the case of the 4-NO2 substituted derivative 6 the pronounced influence of the π-acceptor shifts the reduction potential to a value of up to E(6+·/6) = 0.89 V (vs SCE). Substitution of the N-aryl moiety by electron-donating substituents shifts the potentials correspondingly towards lower reduction potentials, as expected. By introducing the thioether substituent (see 4) to the arene the potential drops to about E(4+·/4) = 0.71 V (vs SCE). If the steric bulk is enhanced by a mesityl substituent (see 5) the reduction potential interestingly is higher than in the parent compound 1 although there are electron-donating alkyl groups present in the molecular structure. This can be explained by a twist of the arene moieties due to steric bulk causing an interruption of the delocalization. The electron transfer is found to occur at E(5+·/5) = 0.67 V (vs SCE). However, the introduction of the π-donating dimethylamino substituent dramatically shifts the reduction potential to up to E(2+·/2) = 0.57 V (vs SCE). We hypothesized that the introduction of even more donating substituents could reduce the reduction potential further. Therefore, we synthesized the modified alkylated compounds 10 and 11, respectively. Indeed, both compounds showed a lower potential with 11 having an Ered(11+·/11) = 0.49 V (vs SCE). This made us curious about the excited state potential of the catalysts. Using the Rehm–Weller equation (without considering the solvent term) [29] we estimated the excited state potential of these catalysts to be as low as Ered(X+·/X*) ≈ −2.9 to −3.0 V (vs SCE).</p><!><p>Reduction potentials Ered(X+·/X) of N-phenylphenothiazines 1–11 (determined by cyclic voltammetry using ferrocene as standard).</p><p>aConverted from the ferrocene scale to the scale vs SCE: +0.38 V [30]. bE00 was estimated by using the method of determination of the intersection of the normalized absorption and fluorescence. cIrreversible. dFluorescence in the UV-A range, see Figure S27 (Supporting Information File 1).</p><!><p>The proposed photoredox catalytic mechanism (Figure 4) for the nucleophilic addition of methanol to olefins starts with photoinduced electron transfer from the N-phenylphenothiazine (1) as photocatalyst to 13a as substrate. The resulting substrate radical anion 13a−· is instantaneously protonated to radical 13a· and back-electron transfer to the intermediate phenothiazine radical cation 1+· yields the substrate cation 13a+. The latter is attacked by methanol as nucleophile and finally deprotonation gives rise to the product 17 (see Figure 4). The principal problem of this type of photoredox catalytic cycle is that the back-electron transfer cannot compete with the initial electron transfer because both components, 1+· and 13−·, are formed only in stationary low concentrations. In the past, we used electron mediators as additives (triethylamine) [19,26] or peptides with substrate-binding sites [31–32] to overcome this problem. For the current work, we propose a radical ion pair in a solvent cage that undergoes an extremely fast proton transfer followed by the intracage back-electron transfer, since triethylamine is no longer needed (vide infra).</p><!><p>Proposed mechanism for the photoredox-catalyzed addition of methanol to α-methylstyrene (13a). (ET = electron transfer, BET = back-electron transfer).</p><!><p>The evaluation of both the optical and electrochemical properties of the prepared phenothiazine derivatives 1–11 leads to the conclusion that only the dialkylamino derivatives 2, 10 and 11 come up with an estimated excited state reduction potential capable of reducing α-methylstyrene (13a). The optoelectronic properties and the excited state reduction potential of Ered(2+·/2*) = −2.5 V of dimethylamino compound 2 that is close to the reduction potential of the substrate 13a encouraged us to approach the so far not yet observed addition of methanol to this less activated substrate promoted by catalyst 2. After optimizing our catalytic system with catalyst 2, we could also confirm the expected reactivity of the branched dialkylamino-substituted derivatives 10 and 11 (entry 12 and 13, Table 2).</p><!><p>Screening of reaction conditions for the methanol addition to α-methylstyrene (13a).a</p><p>aConditions: 30 °C, 65 h, 365 nm LEDs. b10 equiv. cNo light. dNo catalyst. e20 h.</p><!><p>The initial conditions included irradiation of substrate 13a in the presence of the catalyst (5 mol %) in methanol and triethylamine (10% (v/v)) as the additive according to our previously reported photoredox catalysis with pyrene 16 [18]. Under these conditions the product 17 was formed in a yield of 31%. It turned out that omitting the additive as electron shuttle enhanced the catalytic efficiency and the yield increased up to 84%. Obviously, this is a major difference between the photoredox catalysis with pyrene 16, where triethylamine was absolutely crucial to obtaining good product yields, and N-phenylphenothiazine 1. Having this electron shuttle (ca. 1 M) in the reaction mixture efficiently leads to silent or non-silent quenching of the reactive species due to the following modes of quenching. While the back-electron transfer under generation of the triethylamine radical cation unproductively consumes electrons while oxidizing triethylamine, the hydrogen abstraction pathway generates the reduced phenylethane, which is observed in small concentrations in the reaction mixture. The analysis of the reaction mixture still showed some unreacted starting material. Assuming the first electron transfer as the rate-determining step the substrate concentration was reduced to 42 mM and the catalyst concentration was increased to 10 mol %. This change in the reaction conditions led to a quantitative product formation after 65 h. Finally, the rather long reaction times were addressed by speeding up the reaction simply by raising the concentration of all components to 170 mM. This reduced the reaction time to 20 h irradiation producing the product 17 in quantitative yield. However, a further increase of substrate concentration slowed down the reaction again by speeding up silent electron transfer processes.</p><!><p>One of the major current challenges in photoredox catalysis is the design of chromophores suitable for the most reductive chemical reactions, like for instance reductions by alkaline metals, affording reaction conditions that are easier to handle. While solid sodium comes up with a reduction potential of −3.0 V (vs SCE) the present novel N-phenylphenothiazine-based photoredox catalysts reach impressive excited state reduction potentials with up to −3.0 V (vs SCE) in case of catalyst 10. We applied the strongly reducing N-phenylphenothiazines 2, 10 and 11 for the photoredox catalytic reduction of α-methylstyrene (13a) as a less activated styrene that could not be addressed before. After optimization, the photoredox catalytic addition of methanol proceeded in quantitative yield within 20 h without any further additive, like triethylamine as electron shuttle. We could speed up the reaction by using increased concentrations of the substrate and the catalyst affording the product in quantitative yield after 20 h reaction time. We believe that photoredox catalysis with synthetically easily accessible N-phenylphenothiazines will lead to the development of new photoredox catalytic approaches based on their strongly reducing excited states.</p><!><p>Copies of 1H and 13C NMR spectra, mass spectra, absorption and emission spectra and cyclic voltammetry data of 1–11 and 17.</p>

PubMed Open Access

BIGL: Biochemically Intuitive Generalized Loewe null model for prediction of the expected combined effect compatible with partial agonism and antagonism

Clinical efficacy regularly requires the combination of drugs. For an early estimation of the clinical value of (potentially many) combinations of pharmacologic compounds during discovery, the observed combination effect is typically compared to that expected under a null model. Mechanistic accuracy of that null model is not aspired to; to the contrary, combinations that deviate favorably from the model (and thereby disprove its accuracy) are prioritized. Arguably the most popular null model is the Loewe Additivity model, which conceptually maps any assay under study to a (virtual) single-step enzymatic reaction. It is easy-to-interpret and requires no other information than the concentration-response curves of the individual compounds. However, the original Loewe model cannot accommodate concentration-response curves with different maximal responses and, by consequence, combinations of an agonist with a partial or inverse agonist. We propose an extension, named Biochemically Intuitive Generalized Loewe (BIGL), that can address different maximal responses, while preserving the biochemical underpinning and interpretability of the original Loewe model. In addition, we formulate statistical tests for detecting synergy and antagonism, which allow for detecting statistically significant greater/lesser observed combined effects than expected from the null model. Finally, we demonstrate the novel method through application to several publicly available datasets.

bigl:_biochemically_intuitive_generalized_loewe_null_model_for_prediction_of_the_expected_combined_e

<!>Results<!>Inverse Hill equation: concentration as function of occupancy or readout.<!>Biochemically Intuitive Generalized Loewe (BIGL).<!>Discussion<!>Methods<!>Estimation details.

<p>The treatment of heterogeneous disorders such as cancer, infectious diseases and autoimmune diseases is often complicated by adverse effects that limit the tolerable dose of a single agent, or by the gradual development of resistance. Combination therapy, i.e. the simultaneous treatment with multiple therapeutic agents, presents an approach to mitigate such complications. Individual agents are selected that differ in terms of adverse activities and hence can be administered at tolerable doses, yet that all converge on a key disease-causing process, thus maximizing the treatment's efficacy. A combination of multiple effective drugs also encumbers the evolution of resistance, which then requires the simultaneous evasion of multiple mechanisms of action on the same process. Therefore, combination therapies are increasingly adopted as the standard of care for various diseases 1 .</p><p>Consequently, combinations of pharmacological compounds are routinely tested and their activities assessed during early drug discovery. Generating all the data needed for complete mechanistic understanding and modeling of these compound interactions with differential equations is labor intensive and generally not feasible or too costly at this early stage. However, reading out activities in concentration-response, e.g. from a serial dilution of the single agents in the combinations, is typically straightforward and affordable. Conveniently, these single agent results enable the formulation of a plausible expectation of their combined effect under a null model. Comparison of the observed effect of a combination of compounds to that predicted by the null model, then allows an assessment of the potential of that combination. The most promising combinations are the ones with an observed effect that significantly exceeds the expected effect; these are called synergistic.</p><p>Notably, the null model does not aim to be mechanistically accurate. In fact, typically the most attractive combinations are the ones that show the largest favorable deviation from the null model. By its very definition, a synergy call rejects the null model, i.e. it disproves the mechanism underlying it. The costly and labor-intensive modelling of the actual mechanism underlying the combination effect of interest is reserved for those few combinations that are the most attractive and that have their favorable effect confirmed. It often involves elaborating systems of differential equations and custom data generation.</p><p>The choice of a null model for the early (and scalable) evaluation of compound combinations is mostly inspired by convenience. Can base-line effect expectations be computed leveraging only the limited information that is routinely available during early discovery, i.e. on concentration response curves?</p><p>Even though several null models exist, such as Bliss Independence 2 , arguably the most frequently used null model is the one proposed by Loewe 3 . The Loewe model, also known as Concentration Additivity, is a mechanistically inspired and biochemically interpretable model that conceptualizes the assay under study as a (virtual) single-step enzymatic reaction, even if the underlying biology is in fact much more complicated (for instance if the assay relies on entire cells or even organisms), and hence the model is obviously incomplete at best. The conceptual biochemical mechanism posits that the effective binding of an enzyme molecule by any compound switches its activity state (turning it on or off, depending on the activity state of unbound enzyme). Moreover, compound binding to an enzyme molecule is mutually exclusive, i.e. compounds compete for binding. Under this model (under the two classical Loewe constraints), the fraction of bound and thereby identically affected enzyme molecules increases with the concentration of any compound. Once all enzyme molecules are occupied by compound, a maximal response is achieved which is inherently the same for all effective compounds and their combination. However, the common maximal response implied by the classical Loewe model regularly proves untenable. Thus, in cellular assays it occurs that different compounds elicit maximal responses in the same assay that differ in magnitude (partial agonism) or even in direction (agonism/antagonism, for instance in receptor signaling). The classical Loewe model cannot accommodate these cases.</p><p>Although several methods have been described that can cope with partial agonism or agonism/antagonism, these are no longer easily biochemically interpretable (see Discussion). Therefore, we propose a novel direct generalization of the Loewe model, called BIGL (Biochemically Intuitive Generalized Loewe), which maintains the biochemical interpretability of the classical Loewe model, even in cases of partial agonism and antagonism.</p><p>The BIGL model together with two bootstrap statistical tests for synergy calling (MeanR/MaxR) have been implemented in R as the R package BIGL (https://cran.r-project.org/web/packages/BIGL). We demonstrate the method on some publicly available datasets.</p><!><p>Hill equation: occupancy or readout as a function of concentration. The Hill equation is the established model to describe the concentration-response of a biochemical reaction with cooperativity 4 ; it expresses the occupancy o j , i.e. the fraction of enzyme bound to compound j, at a given compound concentration c (equation (1A)). Under the constraint that the binding of a compound j to a molecule of enzyme affects the enzymatic activity in a binary way, the occupancy o j equals the fractional (enzymatic) effect E j (c), i.e. the observed effect relative to the maximum enzymatic effect.</p><p>where i j and h j are the inflection point (half maximal effective concentration, EC 50 ) and the Hill or cooperativity coefficient, respectively. The equation was derived to describe a single-step enzymatic reaction with positive (h j > 1) or negative (h j < 1) cooperativity, but is often used to describe other, more complicated assays, for instance a cellular assay. In practice a four-parameter log-logistic function (4PLL), which amounts to a linearly transformed version of the occupancy o j , is used to directly fit the assay readout values R j (c) (equation (1B)). For a recent mechanistic derivation of this equation, see Mager 2003 5 .</p><p>The above equation contains an intercept b that denotes the activity when no compound is bound to enzyme (at c = 0) and a scaling factor (m j − b), where m j is the maximal effect of the compound. Note that the baseline activity b is the same for all compounds, because it refers to o = c = 0, which reflects absence of any compound. Equations (1A) and (1B) describe occupancy and readout as a function of concentration of a given compound, respectively. When used in the context of compound combination studies, equations (1A) and (1B), which apply to single compounds, are called the marginal (or mono-therapy) concentration response functions.</p><!><p>By simple rearrangement an inverse marginal function can be derived that describes concentration as a function of occupancy (equation (2A)) or readout (equation (2B)), respectively. describes the occupancy of enzyme exposed to a combination of two or more compounds. More specifically, occupancy is defined as the fraction of enzyme bound to any of the compounds in the mix. Under the first classical Loewe constraint that binding of compound to molecules of enzyme affects their activity in a binary (on/off) way, occupancy equals the fractional enzymatic effect of the mix of compounds.</p><p>In equation (3A), c j is the concentration of compound j in the combination, and C j (o) is the concentration of compound j that would result in occupancy o when the compound would be administered as mono-therapy. Upon using equation (2A), equation (3A) becomes</p><p>By numerical solution of equation (3A.1), the occupancy can be computed under the assumption of Loewe Additivity 6 . Note that only the parameters of the Hill equation for the mono-therapies are required.</p><p>Analogously, equation (3B) describes the classical Loewe Additivity (LA) model as a function of readout.</p><p>which is the unnormalized form of the null model as derived by Greco 7 .</p><!><p>In practice, it sometimes occurs that the maximal responses of individual compounds in the same assay differ. Such cases cannot be reconciled with classical Loewe's first constraint, which stipulates that the binding of a molecule of enzyme to any compound in the mixture results in an identical switch of binary activity state of that molecule. Indeed, this first constraint implies a single maximal response, reflecting saturation of all enzyme with compound. However, we propose a generalization of the Loewe model by allowing for varying and fractional compound effects at the enzyme molecule level. The classical Loewe model (equation (3A)) remains valid to describe enzyme occupancy, i.e. the fraction of enzyme bound to compound, because this equation does not refer to the (possibly compound-specific) maximal readouts m j . However, the linear transformation that maps occupancy to readout (equation (1B)) can be compound-specific: the scaling factors (m j − b) then also account for variable and fractional compound effects. Yet, these scaling factors cannot be applied directly to the readout of combination activities. Our solution arises from the inherent properties of the classical Loewe model in that the readout of a combination is the sum of the readouts that are caused by the individual compounds in the combination. Hence, the readout of a combination can be written as a weighted average of the compound-specific readouts of equation (1B); this gives equation (3C)</p><p>where weight f j (o) is the fraction of enzyme that are bound to compound j among those bound to any compound in the combination (with occupancy o). This fraction is exactly the interpretation that is given to the term</p><p>In summary, the BIGL model is applicable for combinations of compounds with different maximal responses. It is described by equations (3A) and (3D). The former expresses Concentration Additivity (classical Loewe) and is used for the calculation of the occupancy. The latter is subsequently used for the prediction of the readout r for the combination experiment under Loewe. Note that the BIGL model only requires parameter estimates from the marginal concentration response curves to predict the null readout of compound combinations.</p><p>Two statistical tests for synergy. In a synergy experiment compounds are tested in several concentration combinations. If the assay readout values deviate too strongly from what is predicted under BIGL model (here referred to as the null model), the combination effect is called synergistic or antagonistic. Statistical tests are designed to measure the evidence in the data against the null model in the presence of variability. Two hypothesis tests are proposed here. Both tests basically contrast the observed readouts of the combination experiments with those predicted from the null model (equations (3A) and (3D)). The only unknown parameters in the null model originate from the Hill equations (equation (1B)), and these are estimated from the mono-therapy data. The first test assesses the average deviation from the null model, and is referred to as the MeanR test. The second test makes use of the largest absolute deviation between observed and predicted readout among all concentration combinations, and allows for the identification of concentration combinations at which synergy or antagonism is present. This test is referred to as the MaxR test. Details about the construction, assumptions and theory of the statistical tests are presented in Supplementary Methods. Results from a simulation study demonstrate the validity of the two testing procedures; see Supplementary Results (Supplementary Figures S1 and S2).</p><p>Evaluation of the method. First, we evaluated the BIGL methodology on some sham (i.e. self) combinations, which by definition should be additive. We used the sham combinations of 25 compounds as presented in Cokol et al. 8 . After filtering, based on the variance on the estimated log(EC50) to remove poor quality data (see Methods), 14 compounds were left and their sham combinations were analyzed both at 8 hours as well as 12 hours. Of the 28 sham combinations tested, 4 (14%) were called non-additive, which is slightly higher than expected at the 5% significance level (Table 1). The null was rejected for all 4 combinations in favor of an antagonism call. However, the combination data of both BRO (8 h) and MET (8 h, 12 h) are showing lower effects than expected under the null at higher dose ranges (Fig. 1a). This implies either deviation from additivity or data quality issues (e.g. due to toxicity). The TAM combination is also called antagonistic, but this call is clearly driven by one extreme outlier (Fig. 1b). In summary, after accounting for outliers and data quality issues, the BIGL call rate for deviation of additivity on sham combinations is in line with expectations.</p><p>Next, we evaluated the performance of the BIGL methodology on a larger combination dataset, namely the OncoPolyPharmacology Screen published by O'Neil 9 . We successfully executed BIGL analyses on all combination data provided by O'Neil (Supplementary Figures S5 and S6 present an overview by heatmap). As a quality control prior to further analysis, we applied a similar filtering as used in the Cokol dataset (see Methods) which led us to invalidate 72% of the initial combinations (583 pairs of compounds tested in a varying subset of 39 cell lines). 438 (75%) compound pairs were found synergistic in at least one cell line under the BIGL model, which is more than the 50% in the O'Neil study, which reports analyses for the Highest Single Agent (HSA) and Bliss null models. Figure 2 visualizes the fraction of cell lines with synergy calls for each compound pair. For convenience, the ordering of the compounds is kept consistent with the heatmap in the O'Neil study. 181 combinations (31%) were called synergistic in half of the analyzed cell lines. The O'Neil study highlighted the combination of Wee1 inhibitor (AZD1775) with MK-8776 as a consistently strong synergistic drug combination, a finding confirmed in Guertin et al. 10 . The combination data for this compound pair in 15 out of 39 cell lines passed data quality criteria; the BIGL evaluation returned a synergy call for 13 of those, in line with the O'Neil conclusions. The O'Neil study also reports synergy of the pairing of the same Wee1 inhibitor (AZD1775) with an mTOR inhibitor (MK-8669) in roughly 30% of the cell lines. The combination data for this compound pair in 18 out of 39 cell lines passed data quality criteria, and the BIGL approach called synergy for 6 (33%) of those.</p><p>Additionally, the BIGL approach also flagged some combinations that were reported as synergistic elsewhere but that were not picked up in the O'Neil study. Figure 3 (marginal dose-response curves) and Fig. 4 (combination dose-response surface) illustrate the analysis of one such case, namely the combination of Gemcitabine and Dasatinib in the LOVO cell line. These figures also visualize the hallmark ability of the BIGL methodology to accommodate the partial agonism of Dasatinib (Fig. 3), which classical Loewe handles poorly. In line with findings elsewhere 11 , BIGL flags the combination of Gemcitabine and Dasatinib as synergistic in 14 out of the 18 cell lines with quality controlled datasets.</p><!><p>With the advent of combination therapy (for a recent review, see e.g. He et al. 12 ), the treatment options for many diseases have drastically improved. The Loewe or Concentration Additivity is often used as a biochemically interpretable null model for an early assessment of the potential of combinations. Even though the very simple biochemical reaction mechanism that underlies Loewe is in most cases (e.g. in a cellular or organism context) incomplete at best, it still enables the straightforward computation of an easy-to-interpret baseline expectation. This computation only requires routinely available concentration-response curves of the individual compounds, and no mechanistic information that is typically unavailable during early drug discovery. Used in a synergy call context, the Loewe null model enables the prioritization of compound combination that deviate favorably from the expectation. These per definition reject Loewe as an accurate mechanistic model. For the most interesting combinations, a mechanistic model can be elaborated that explains the deviation from the Loewe null, but this typically requires considerably more data generation.</p><p>The classical Loewe reference model is not applicable whenever the concentration-response curves of the compounds in the combination have different maximal effects, like in pairings of an agonist with a partial or an inverse agonist. Whereas several methods that build on classical Loewe have been developed to address these issues (Table 2), to the best of our knowledge, BIGL is the first method to handle combinations with partial and inverse agonists (see Supplementary Figure S3 and Flaveny et al. 13 ) that preserves the biochemical interpretability of the original null model. In this paper, we have derived the model equations of the BIGL method and developed statistical tests for synergy testing. Moreover, we demonstrated the amenability of the BIGL approach to a real world large oncology screen of combinations which contains clear examples of combinations with partial agonists 9 . The BIGL obtained synergy assessment results on this screen were generally consistent with literature. Additionally, we showed through simulation (Supplementary Results) and performance on self combinations that the type I error rate is controlled in general.</p><p>Like classical Loewe, BIGL posits that compounds compete for the same enzymatic binding site (otherwise its results would deviate from classical Loewe where this applies). The combination response reflects the sum of the difference between maximal and baseline response for each compound, weighted by the fraction of (virtual) enzyme bound to that compound. Hence, the most extreme maximal response of any single compound in a combination also defines the most extreme maximal response of that combination, which is achieved by saturation of the enzyme with that compound 14,15 . The response of combinations that include the same concentration of a full agonist (i.e. the compound with the most extreme maximal response) will be reduced with increasing concentrations of a partial agonist (i.e. the compound with a less extreme maximal response). Both in clinical applications and in discovery, an agonist compound can be combined with a neutral antagonist (which competitively blocks the agonist without being enzymatically active) or an inverse agonist (which triggers an enzymatic response in the other direction, for instance in assays with a constitutively active receptor, channel or transporter, an inverse agonist would reduce such activity). Known clinical compound combinations target the μ opioid receptor and the α1 adrenergic receptor 16,17 . To illustrate the flexibility of BIGL, we simulated the expected response of a combination of a partial opioid receptor agonist with a neutral antagonist with dose response taken from 18 (Supplementary Figure S4). The case of a neutral antagonist presents the extreme case where one of the combination compounds show a maximal response equal to baseline. Notably, if a compound combination is specified as a combination of individually active compounds, then it could be argued to opt for the alternative null model of the single active compound. A straightforward test of a null hypothesis of non-activity for both compounds would enable to automate the identification of such cases.</p><p>As stated above, BIGL is not the first method that generalizes the classical Loewe model. The currently existing methods that generalize the classical Loewe model to deal with partial and/or inverse agonisms are listed in Table 2. Some of these methods overextend conventional methods beyond their justified application domain by simply ignoring inconvenient response curve or compound combination characteristics, even if these are very real. Thus, the Generalized Concentration Addition (GCA) method of Howard 19 imposes Hill coefficients of 1 for all compounds and thereby ignores possible cooperativity. The extrapolation method described in Scholze 20 deals with partial agonism by rescaling any maximal effects to the overall most prominent maximal effect, hereby ignoring observable differences in maximal effect. A piecewise approach is described in 21,22 , where contributions of compounds are accounted for only within their individual effect range, but completely ignored outside of it. As a general guideline, it is advised to deploy methods that correctly consider observed data characteristics.</p><p>Another group of methods was designed to address different maximal responses properly. The FLM model 23 , for example, is similar to BIGL in that its function is constructed as a weighted average of terms that are scaled to the original marginal dose-response functions. However, it conceptually hinges on effect equivalence concepts that implicitly assume parallel concentration-response curves (with identical Hill coefficients and maximal responses). Under those constraints and only then, the FLM model is a special case of the LA model, and by extension the BIGL model. In all other cases, the model is mathematically convenient, but biochemically uninterpretable. The BRAID model of Twarog et al. 24 is another method that adheres to the LA constraints, thus building on the Hill dose response for the individual compounds and satisfying the Loewe self-combination additivity property. However, the BRAID additivity surface is not Loewe additive 24 . Therefore, unlike BIGL, it never reduces to LA conceptually which limits interpretability. Moreover, as the BRAID model presents a mathematical solution to deal with partial agonists, the modeled concentration addition can involve negative concentrations which is biochemically unintuitive. Besides different ways of generalizing the classical Loewe model, the methods listed in Table 2 also differ in the way they assess synergy. They can be divided in two types: response surface modeling and lack of fit testing of the null model. Response surface models capture the deviation of the observed response surface from the null model relying on the parameters of the marginal dose response functions and one additional parameter that enables synergy antagonism calling. For example, the BRAID model 24 comprises a single interaction parameter κ, which is statistically equivalent to 0 under additivity, but deviates under synergy or antagonism. In addition to enabling compound interaction calls, well fitted response surface models can be used for interpolating the effect of unobserved compound combinations. However, while a single interpretation parameter supports straightforward compound interaction calling, the resulting models often lack the flexibility to accurately adjust the observed response surface. For instance, they impose that over the entire concentration range the effects of the compounds in the mixture occur in the same direction -either consistently positive or consistently negative 6 . A better fit and hence interpolation performance can be achieved by exploring a bigger modeling space. Thus, multiple reference models can be evaluated 22 , and the compound interaction call based on the best fitting one. However, this may encumber the comparison of combinations, if optimal fitting selects different models as reference for the compound interaction assessment. In practice, response surface model methods often face a trade-off of interpolation performance versus consistency of interaction assessment. Notably, optimal predictivity typically benefits from more extensive model flexibility, as illustrated by mechanism based models (outlined in 5 ). However, these models are not designed to support a comparison to a null model, which is different from the second group of methods: lack of fit models where compound interaction evaluation only aims to assess the deviation from the null model, informally or formally (i.e. statistically). Early examples include various isobologram approaches, which evaluate deviation of the null model for isoeffective sections through the dose response surface. Isobologram analyses have been extended to the partial agonist case 25 .</p><p>The proposed statistical tests within the BIGL framework are lack of fit methods. They rely on a null model that, like the Loewe model, requires no other parameterization than the parameters of the concentration-response curves of the individual compounds and as assumptions the ones underpinning the standard practice of fitting a four-parameter log-logistic dose-response function or Hill equation (which generalizes the Michaelis-Menten equation to accommodate cooperativity as Hill coefficients deviating from one). Additionally, the BIGL lack of fit tests assume equality of variance of outcomes -however this could be relaxed in future developments. BIGL naturally generalizes Loewe by allowing for different maximal responses. Even though the Loewe and consequently BIGL models are inspired by a specific and simple enzymatic mechanism, when used in the context of synergy or antagonism call, they are not assumed to be mechanistically accurate. Like Loewe, BIGL is straightforward to compute with minimal and routinely available information (concentration-response curves), and enables to select combinations of interest. Importantly, the most attractive compound combinations would typically be the ones that deviate favorable from the BIGL expectation, and thereby reject BIGL as the mechanism of action.</p><p>In summary, we propose BIGL as a tractable and flexible approach to assess compound interaction with minimal and routinely available data requirements, that naturally generalizes the widely used Loewe model and retains Loewe's biochemical interpretability.</p><!><p>Datasets. The BIGL methodology was applied on two different publicly available datasets. The first dataset, which was described in Cokol et al. 8 , contains 25 sham (self) combinations provided for each compound combination as raw cell growth measurements in a full, 8 concentration checkerboard design (8 × 8) in 24 hour time course with 15 minute intervals. In the current study data for the 8 and 12 hour timepoints were analysed.</p><p>The second dataset came from the OncoPolyPharmacology screen as described in O'Neil et al. 9 , which describes the assessment of 38 drugs (22 experimental and 16 approved drugs) in a panel of 39 cancer cell lines for a total of 583 pairs. The mono-therapy data are expressed as cell growth rate relative to the growth rate of DMSO-treated controls.</p><!><p>For the mono-therapy data for each compound pair, a 4-parameter log-logistic function was fitted, subject to the constraint that both compounds share the same baseline response. Additionally, to stabilize residual variance during the marginal parameter estimation procedure, the readouts were log-transformed to instantaneous growth reads 26 . Parameters were estimated by a non-linear least squares approach. A model was considered of sufficient quality if the standard deviation of log-transformed half-maximally effective concentration (log(EC50)) estimates for both compounds in a pair did not exceed 10.</p><p>The marginal parameters were subsequently used to construct the response surface as predicted by the generalized Loewe model and MeanR and MaxR statistics were computed from the predicted and observed off-axis data. Null distributions for these statistics were obtained assuming normally distributed off-axis residuals for all compound pairs in the O'Neil et al. dataset (see Supplementary Methods). In contrast, a bootstrapping method with 1000 runs was used for the Cokol et al. 8 dataset to approximate null distributions of the MeanR and MaxR statistics. Since no replicates were available in the latter dataset, the readout errors for off-axis points during bootstrapping were sampled from a centered Gaussian distribution with variance equal to the residual variance estimate from the marginal model. In all of the above cases, the C p matrix (see Supplementary Methods) was calculated using 100 bootstrap runs and all statistical tests were performed at a significance level of 5%.</p>

Scientific Reports - Nature

Rapid Identification and Systematic Mechanism of Flavonoids from Potentilla freyniana Bornm. Based on UHPLC-Q-Exactive Orbitrap Mass Spectrometry and Network Pharmacology

Potentilla freyniana Bornm. (P. freyniana), belonging to the family Rosaceae, has been used as a folk medicine in China. However, as we know, the constituents and the systematic elucidation of the mechanism were not fully investigated. Therefore, it is necessary to develop a rapid method using LC-MS and network pharmacology for the detection and identification of constituents and the systematic mechanism of P. freyniana. Firstly, the flavonoids were detected and identified based on ultra-high-performance liquid chromatography coupled with Quadrupole-Exactive Focus Orbitrap MS (UHPLC-Q-Exactive Orbitrap MS). After that, the potential targets of those constituents were obtained by database mining. Then, the core targets were predicted by protein-protein interaction network and network analysis. Finally, Gene Ontology (GO) and Kyoto Encyclopedia of Genes and Genomes (KEGG) pathway enrichment analysis were carried out via DAVID. This finding revealed that P. freyniana possessed 43 flavonoids (40 of them were first reported) with 23 core target genes, which are associated with PI3K-Akt, MAPK, TNF signaling pathway, and pathway in cancer. This study demonstrated the multicompound, multitarget, and multimechanism of P. freyniana, which are very beneficial to develop the further study and utilization of this plant including the material basis and quality control research.

rapid_identification_and_systematic_mechanism_of_flavonoids_from_potentilla_freyniana_bornm._based_o

1. Introduction<!>2.1. Chemicals and Materials<!>2.2. Sample Preparation<!>2.3. Instruments and Conditions<!>2.4. Data Processing and Analysis<!>2.5. Target Identification of Flavonoids<!>2.6. Protein-Protein Interaction Network<!>2.7. GeneMANIA Analysis<!>2.8. GO and Pathway Analysis<!>3.1. Analytical Strategy<!>3.2. Identification of Flavonoids<!>3.3. Target Identification of Flavonoids<!>3.4. Protein-Protein Interaction Network<!>3.5. GeneMANIA Analysis<!>3.6. GO and Pathway Analysis<!>3.7. Network Analysis<!>4. Conclusion

<p>Potentilla freyniana Bornm. (P. freyniana), a genus Potentilla of the family Rosaceae, named Difengzi, is a perennial plant with branched and tufted roots widely distributed and cultivated all-around of China, especially in Hunan, Hubei, Jiangxi. Their roots have been used as a folk medicine in clearing away heat and toxic materials for treating canker, bone tuberculosis, external bleeding [1–3]. Previous investigations on P. freyniana showed the presence of different compounds including terpenes and flavonoids [4–7] and possessed a variety of activities such as anti-inflammatory, analgesic effects [8, 9]. However, as we know, the constituents and systematic pharmacological mechanism were not fully investigated. For instance, 14 compounds, including eriodictyol, phlorizin, were separated from the roots of P. freyniana [7]. Therefore, it is worthwhile to establish a highly sensitive method for characterizing their chemical constituents and elucidating systematic pharmacological mechanism of P. freyniana.</p><p>The complexity of chemicals in Traditional Chinese Medicine (TCM) including P. freyniana has presented a significant challenge in the rapid identification and characterization of components. During the past decades, HPLC-MS, as a new technique has been used to profile and identify the chemical in TCM due to its validity, sensitivity, and specialness [10, 11]. Especially, UHPLC-HRMS such as UHPLC-Q-Exactive Orbitrap MS, UHPLC-Q-TOF MS, and UHPLC–LTQ-Orbitrap-MS, affording a higher and faster separation and higher resolution of mass, was a much more powerful equipment in the identification of TCM compared to traditional HPLC-MS [12–14].</p><p>Network pharmacology is an impressive methodology for investigating the systematic pharmacological mechanism through the constructing and analyzing biological networks such as protein-protein interaction, chemical-target-pathway network, which could provide direction for the further discovery of new drug without enormous time, money, and effort [15–17].</p><p>Therefore, this current study was designed to develop a fast and effective method for the chemical characterization and systematic pharmacological mechanism of P. freyniana using UHPLC-Q-Exactive Orbitrap MS and network pharmacology. P. freyniana possessed 43 flavonoids (40 of them was first reported) with 23 core target genes, which are associated with PI3K-Akt, MAPK, TNF signaling pathway, and pathway in cancer. This study demonstrated the multicompound, multitarget, and multimechanism of P. freyniana, which are very beneficial for the further study and utilization of this plant including the material basis and quality control research.</p><!><p>The chemical reference standards isoquercitrin, luteolin, naringenin, and kaempferol were provided by Chengdu Herbpurify biotechnology CO., LTD (Chengdu, China); Phlorizin, Phloretin, and Trilobatin were purchased from Sichuan Wei Keqi Biotechnology Co., Ltd (Sichuan, China); eriodictyol and hyperoside were provided by Chengdu Push biotechnology CO., LTD (Chengdu, China); quercetin and apigenin were provided by Chengdu Alfa biotechnology CO., LTD (Chengdu, China); baicalein and wogonin were obtained from Chengdu Desite biotechnology CO., LTD (Chengdu, China). The purities of all chemical reference standards were above 98% by HPLC-DAD.</p><p>Acetonitrile and methanol of chromatography grade were provided by MERCK (Darmstadt, Germany); The ultrapure water was produced by a milli-Q water purification system (Millipore, Milford, MA, United States); formic acid of LC-MS grade and all other reagents of analytical grade were purchased from Aladdin Industrial Corporation.</p><p>P. freyniana was collected from Tong-Dao country of Huaihua (109.86 longitude, 26.03 latitude), Hunan province and were identified by Professor Wei Cai (Hunan Provincial Key Laboratory of Dong Medicine, Hunan University of Medicine). The voucher specimen was deposited at School of Pharmaceutical Sciences, Hunan University of Medicine.</p><!><p>A total of 10 g powdered root of P. freyniana was ultrasonically extracted with 50 mL of 70% aqueous methanol for 1 h, and then the extracted solution was filtered for further UHPLC Q-Exactive Focus Orbitrap MS analysis.</p><p>The reference standards including hyperoside, isoquercitrin, phlorizin, eriodictyol, trilobatin, quercetin, luteolin, naringenin, apigenin, phloretin, kaempferol, baicalein, and wogonin were weighed and dissolved in methanol to obtain the reference solution with the final concentrations of 10.2, 9.8, 10.0, 9.8, 9.9, 10.5, 10.2, 10.8, 9.3, 10.1, 10.6, 9, 4, and 10.5 ug/mL, respectively, and then these solutions were stored in −4°C before analysis.</p><!><p>Chromatography analysis was performed on an Ultimate 3000 focused system (Dionex, Sunnyvale, CA, USA) consisting of an online vacuum degasser, a binary pump, and an autosampler. The sample separation was carried out on the Hypersil GOLD C18 column (100 × 2.1 mm, 1.9 μm) at 40°C. The mobile phase consisted of acetonitrile as solvent A and water with 0.1% formic acid as solvent B using the following gradient elution of 5–15 % A at 0–2 min; isocratic 15 % A at 2–5 min; 15–18 % A at 5–10 min; 18–50 % A at 10–15 min; 50–100 % A at 15–16 min; 100–5 % A at 16–17 min; isocratic 5 % A at 17–20 min. The flow rate was 0.3 mL/min.</p><p>All LC-MSn analyses were performed on the Q-Exactive Focus Orbitrap MS connected to the UHPLC system via a heated electrospray ionization source (Thermo Electron, Bremen, Germany). The optimized tune operating parameters in negative ion mode were listed as follows: sheath gas and auxiliary gas flow rate of 30 and 10 arbitrary, respectively; the capillary and auxiliary gas heater temperatures of 320°C and 350°C, respectively; spray voltage of 3.0 kV; RF lens of 50; High-resolution MS analysis was performed at full scan MS1 with the mass range of m/z 100–1000 at a resolution of 35000 and targeted MS2 at a resolution of 17500 triggered by parallel reaction monitoring mode; nitrogen was set as sheath, auxiliary, and collision gas; the isolation widow was 2 amu, and the normalized collision energy (NCE) was 30%.</p><!><p>All high-resolution MS data were acquired and processed using the Xcalibur version (2.0 software, Thermo Fisher Scientific, San Jose, CA, USA). The compounds were detected by the Compound Discover version 3 using the metabolism workflow templates by the expected compounds predicted method [18]. The detailed parameters of the workflow template were set as follows: the minimum peak intensity was set as 10000; the maximum element counts were C30 H60 O20; the mass tolerance of MS and MS2 was within 5 and 10 ppm, respectively; baicalein and phloretin were set as the carbon skeleton; reduction, oxidation was set as Phase I transformation; glucoside conjugation, glucuronide conjugation, pentoside conjugation, methylation was set as Phase II transformation.</p><!><p>TCMSP database is a free and online database for potential target identification of small molecules, especially TCM. The target genes were converted to the official gene symbol by STRING (https://string-db.org) or Uniport (https://www.uniprot.org).</p><!><p>STRING was a free tool, which can construct the PPI network by uploading the potential targets. The species was set as "Homo sapiens" with a confidence score >0.4. The network analysis was performed at Cytoscape to obtain the core targets.</p><!><p>GeneMANIA (http://genemania.org) is an online free and friendly tool for investigating gene function and gene interaction. The species was set as "Homo sapiens."</p><!><p>The GO and KEGG pathway analysis was performed on the DAVID (https://david.ncifcrf.gov, Version 6.8). The specific species in the list and background was set as "Homo sapiens." The entire compounds, targets, and pathway network were visualized by Cytoscape.</p><!><p>In order to identify flavonoids fully, an analytical strategy based on UHPLC Q-Exactive Focus Orbitrap MS was established in this study. First, the sample was prepared and injected into the UHPLC Q-Exactive Focus Orbitrap MS to gain the full scan high-resolution MS data. Then, those data were processed using Compound Discover software with metabolism workflow to predict and detect the molecule of flavonoids. Third, the MS2 of the predicted molecule were acquired using UHPLC Q-Exactive Focus Orbitrap MS by parallel reaction monitoring mode. Finally, the compounds were identified based on the full scan MS, MS2 data, retention time, and bibliography.</p><!><p>The total content of flavonoids was measured by NaNO2-Al(NO3)3-NaOH spectrophotometric colorimetry [19]. The calibration curve obtained by the rutin standard of absorbance concentrations(mg/mL) using five dilutions was y = 7.15x − 0.001, with the corresponding determination coefficient at 0.9999. Finally, the content of flavonoids is 32.17 ± 0.26%. A total of 43 constituents were unanimously and tentatively characterized based on UHPLC Q-Exactive Focus Orbitrap MS combined with the expected compounds predicted method. 40 excluded eriodictyol, phloretin, and hyperoside were reported from P. freyniana for the first time. The detailed information of those compounds is listed in Table 1. The high-resolution extracted ion chromatography is shown in Figure 1.</p><p>Peaks 15, 17, 27, 33, 35–43 were unanimously identified as hyperoside, isoquercitrin, phlorizin, eriodictyol, trilobatin, quercetin, luteolin, naringenin, apigenin, phloretin, kaempferol, baicalein, and wogonin, respectively, by comparing the retention time, high-resolution mass measurement, and MS2 spectrum with those reference standards.</p><p>Peak 13 was eluted at 6.57 min and possessed the deprotonated ion [M−H]− at m/z 303.0507 (−0.99 ppm, C15H11O7). The fragment ions at m/z 125.0234 (−8.14 ppm, C6H5O3) and 285.0407 (0.84 ppm, C15H9O6) were detected in the MS2 spectrum, which is consistent with the MS data of taxifolin in bibliography [20]. Thus, peak 13 was tentatively identified as taxifolin. Peaks 1–3, and 8 possessed the deprotonated ion [M−H]− at m/z 465.1039 (0.21 ppm, C21H21O12), m/z 465.1033 (−1.08 ppm, C21H21O12), m/z 465.1042 (0.86 ppm, C21H21O12), and m/z 465.1042 (0.86 ppm, C21H21O12), respectively, 162 Da(C6H10O5, glucose moiety) more than that of taxifolin (peak 13). The fragmentation ions at m/z 285.041(C15H9O6), 125.023(C6H5O3), 303.051(C15H11O7) in the MS2 spectrum were matched to those attributed to taxifolin. Therefore, Peaks 1–3, and 8 were tentatively characterized as taxifolin-glucoside.</p><p>Peaks 4 and 26 were eluted at 4.13 and 11.18 min, respectively. All of them showed the same deprotonated ion [M−H]− at m/z 463.088 (C21H19O12), 176.032 Da(C6H8O6, glucuronide moiety) more than that of eriodictyol, suggesting they are eriodictyol-glucuronide, which were further identified by the presence of fragmentation ion at m/z 287.056 (C15H11O6). In a similar way, peaks 9, 28, and 30 were tentatively identified as taxifolin-glucuronide, quercetin-glucuronide, and quercetin-glucuronide, respectively.</p><p>Peaks 5, 7, 10, 16, 19, 22, and 34 were eluted at 4.19, 4.56, 5.88, 6.90, 7.99, 9.61, and 13.57 min, with the same deprotonated ion [M−H]− at m/z 449.109 (C21H21O11). Peaks 19 and 34 possessed the fragment ions at m/z 167.034 (C8H7O4) and m/z 123.044 (C7H7O2), which are the diagnosis fragmentation ions of phloretin, suggesting they were phloretin derivatives. Thus, Peaks 19 and 34 were tentatively inferred as phloretin-glucuronide. Peaks 5, 7, 10, 16, and 22 yielded the same fragmentation ion at m/z 287.056 (C15H11O6), suggesting they were eriodictyol derivatives. The ion at m/z 287.056 was yielded by the neutral loss of 162.053 (C6H10O5, glucose moiety), suggesting the presence of glucose moiety. Therefore, they were tentatively characterized as eriodictyol-glucoside.</p><p>Peak 6 with the deprotonated ion [M−H]− at m/z 593.1535 (3.88 ppm, C27H29O15) was eluted at 4.41 min. It yielded fragment ions at m/z 353.0667 (0.07 ppm, C19H13O7), 383.0774 (0.42 ppm, C20H15O8), 473.1092(0.56 ppm, C23H21O11), and 413.0874 (−0.98 ppm, C21H17O9), resulting from the loss of C4H8O4 + C4H8O4, C4H8O4 + C3H6O3, C4H8O4, and C3H6O3 + C3H6O3, respectively, suggesting the presence of two carbon-glucoside. According to the published paper [21, 22], peak 6 was tentatively identified as Vicenin II. In a similar way, peak 18 was tentatively identified as Phloretin-C-diglucoside.</p><p>Peaks 11, 14, 24, and 31 generated the same quasimolecular ion [M−H]− at m/z 433.114 (C21H21O10), 162 Da(C6H10O5, glucose moiety) more than that of naringenin (peak 38), suggesting they were naringenin-glucoside, which were further confirmed by the presence of m/z 271.061 and 151.003 in MS2 spectrum.</p><p>Peak 12 eluted at 6.22 min and showed a pseudomolecular ion at m/z 625.1408 (0.00 ppm, C27H29O17), 176.032 Da(C6H8O6, glucuronide moiety) more than that of eriodictyol-glucoside, suggesting it is eriodictyol-glucoside-glucuronide, which was confirmed by the presence of the base peak at m/z 287.0558 (eriodictyol).</p><p>Peaks 20, 21, 25, and 29 presented the same deprotonated ion [M−H]− at m/z 431.099 (C21H19O10) and generated the same fragment ions at m/z 269.044 (C15H9O5) by loss of the glucose moiety (C6H10O5), which suggested the presence of glucose moiety. The base peak at m/z 268.037 [Y0–H] ions in the MS2 spectrum of peaks 25 and 29 was a characteristic of apigenin aglycone. According to the published paper [23, 24], they were tentatively inferred as apigenin-7-glucoside and apigenin-4′-glucoside, respectively. Meantime, peaks 20 and 21 were tentatively characterized as baicalein-glucoside.</p><p>Peak 32 was detected at 13.39 min. It presented a pseudomolecular ion at m/z 567.1730 (1.94 ppm, C26H32O14) and exhibited the MS2 fragmentation ions at m/z 273.0771 (1.10 ppm, C15H13O5), resulting from the loss of glucose moiety and pentoside moiety (294.096). Thus, peak 32 was tentatively characterized as phloretin-pentoside-glucoside. In a similar way, peak 23 was tentatively identified as naringenin-pentoside-glucoside.</p><!><p>212 putative targets of flavonoids were obtained from the TCMSP database. A visual compounds-targets network with 224 nodes and 440 edges was built by Cytoscape Version 3.7.2 (Figure S1). Compounds quercetin, apigenin, kaempferol, luteolin, and wogonin are the top 5 compounds with a maximum degree and betweenness in the compound-targets network. The detailed information of putative targets linked to compounds was provided in Supplementary Table S1.</p><!><p>In order to find the key targets of flavonoids, a total of 212 putative targets were imported into the STRING to obtain the protein-protein interaction (PPI) data. The PPI network with 206 nodes and 3980 edges was established by Cytoscape (Figure S2). A total of 23 targets, including AKT1, INS, TP53, IL6, HSP90AA1, EGFR, VEGFA, JUN, EGF, CASP3, MAPK1, ESR1, ERBB2, PTGS2, MYC, MAPK8, MMP9, FN1, FOS, PPARG, CXCL8, CYCS, and CCND1, were selected as the core targets for GO and KEGG pathway analysis by setting the parameters as follows: the degree ≥50; betweenness centrality ≥0.01; closeness centrality ≥0.6.</p><!><p>Among the 23 key target genes and their interacting genes, it was found that 42.75 % had coexpression characteristics, 41.10 % displayed physical interactions characteristic. Other characteristics, including pathway, genetic interactions, colocalization, and shared protein domains, are displayed in Figure 2.</p><!><p>In order to further study the 23 core target genes, GO and KEGG pathway analysis were performed by DAVID. GO term enrichment analysis results were divided into the biological process (BP, 23/23), cell compound (CC, 23/23), and molecular function (MF, 23/23). A total of 158 BP, 16 CC, and 33 MF has a p-value less than 0.05 (Table S2). In GO term enrichment analysis, the BP might be related to positive regulation of transcription from RNA polymerase II promoter (10/23), response to drug (9/23), negative regulation of apoptotic process (9/23), positive regulation of transcription, DNA-templated (9/23), signal transduction (9/23), positive regulation of gene expression (8/23), and positive regulation of cell proliferation (8/23), and so on. The top 4 of CC are nucleus (16/23), nucleoplasm (12/23), cytosol (12/23), and cytoplasm (12/23). The MF are protein binding on 100%, identical protein binding on 56.5%, enzyme binding on 39.1%, and transcription factor binding on 34.8%. The top 10 enriched terms in BP, CC, and MF are displayed in Figure 3. In addition, 83 KEGG pathways (Table S3) were enriched as p-value less than 0.05. The result showed that the pathway was mainly related to the signaling pathway including PI3K-Akt (12/23), MAPK (10/23), TNF (9/23), ErbB (8/23), HIF-1 (8/23), Estrogen (8/23), FoxO (8/23), and cancer in the pathway. The top 20 KEGG pathways are shown in Figure 4.</p><!><p>Based on the target and KEGG pathway analysis, the entire compounds, targets, and pathway network were established by Cytoscape. The network with 122 nodes and 595 edges is shown in Figure 5. The red diamond, green ellipse, and blue triangle represent compounds, genes, and pathways, respectively.</p><!><p>In the present investigation, this finding revealed that P. freyniana possessed 43 flavonoids (40 of them was first reported) with 23 core target genes, which were associated with PI3K-Akt, MAPK, TNF signaling pathway, and pathway in cancer. This study demonstrated the multicompound, multitarget, and multimechanism of P. freyniana, which are very beneficial for the further study and utilization of this plant including the material basis and quality control research.</p>

PubMed Open Access

Gas-Purged Headspace Liquid Phase Microextraction System for Determination of Volatile and Semivolatile Analytes

In order to achieve rapid, automatic, and efficient extraction for trace chemicals from samples, a system of gas-purged headspace liquid phase microextraction (GP-HS-LPME) has been researched and developed based on the original HS-LPME technique. In this system, semiconductor condenser and heater, whose refrigerating and heating temperatures were controlled by microcontroller, were designed to cool the extraction solvent and to heat the sample, respectively. Besides, inert gas, whose gas flow rate was adjusted by mass flow controller, was continuously introduced into and discharged from the system. Under optimized parameters, extraction experiments were performed, respectively, using GP-HS-LPME system and original HS-LPME technique for enriching volatile and semivolatile target compounds from the same kind of sample of 15 PAHs standard mixture. GC-MS analysis results for the two experiments indicated that a higher enrichment factor was obtained from GP-HS-LPME. The enrichment results demonstrate that GP-HS-LPME system is potential in determination of volatile and semivolatile analytes from various kinds of samples.

gas-purged_headspace_liquid_phase_microextraction_system_for_determination_of_volatile_and_semivolat

1. Introduction<!>2.1. Fabrication of the Semiconductor Condenser and Heater in GP-HS-LPME System<!>2.2. Constitution and Working Principle of GP-HS-LPME System<!>2.3. Experimental Procedure<!>3.1. Optimization of GP-HS-LPME<!>3.2. Evaluation of GP-HS-LPME System<!>3.3. Sensitivity of GP-HS-LPME for Volatile and Semivolatile<!>4. Conclusion

<p>As is known, sample treatment is a very important stage of any analytical procedure. However, it takes much time to prepare the samples. It is reported that up to 80% of the total time is spent in preparing samples in a complete sample analysis process. So as to shorten time of the whole process of sample analysis, there is a trend towards integration and automation in modern sample treatment techniques. Among various sample treatment techniques, headspace liquid phase microextraction (HS-LPME), which integrates extraction, cleanup, and concentration, is a highly integrated sample treatment technique developed in recent years. In this technique, target compounds are evaporated from the sample matrix into the gas phase and then enriched by the solvent microdrop hanging on the tip of the microsyringe needle [1–3]. As a result of its characteristics of integration, simplicity, and rapidness, HS-LPME technique has been widely used for enriching volatile and semivolatile analytes from various kinds of sample matrixes [4–9].</p><p>For HS-LPME technique, the temperatures of sample matrix and extraction solvent are key factors that affect enrichment effect. Generally, a high temperature promotes release of target compounds from the sample and accelerates the diffusion of them into the gas phase, which speeds up the entry of target compounds into extraction solvent. Moreover, a low temperature of extraction solvent is beneficial to enrichment, because a low temperature is advantageous for target compounds in the headspace gas phase to dissolve in extraction solvent since the extraction of target compounds into the extraction solvent is an exothermic process [10–13]. Furthermore, in HS-LPME technique, the volume of headspace gas phase is limited because HS-LPME is performed in a closed system. Based on the theory of ideal gases, it is concluded that the amount of chemicals in the gas phase increases with increasing gas volume under a given temperature and pressure [14], so the amount of target compounds entering the headspace in HS-LPME is constant due to limited gas phase volume.</p><p>In practice, in order to obtain a high temperature of sample matrix and a low temperature of extraction solvent, various kinds of heating and cooling methods have been proposed. Recently, main heating methods such as hydrodistillation method, microwave heating, and recycling hot water have been used to increase the temperature of sample matrix. In addition, circulating cold water, CO2 cooling technique, and thermoelectric cooler have been used to cool the extraction solvent [15–17]. However, these methods have shortcomings including large device volume, inconvenience of operation, and difficulty in controlling the temperature and achieving online enrichment. Besides, much electrical energy is consumed due to use of microwave heating.</p><p>To achieve rapid, automatic, and efficient extraction, a gas purge headspace liquid phase microextraction (GP-HS-LPME) system was researched and developed here, in which semiconductor condenser and heater were designed, respectively, to cool the extraction solvent and to heat the sample (the ranges of refrigerating and heating temperatures are from −5°C to room temperature and from room temperature to 125°C, resp.). Besides, inert gas was constantly led into and discharged from the system for the purpose of accelerating the movement of target compounds and increasing the gas phase volume in GP-HS-LPME system by prolonging extraction time. Thus, the amount of target compounds in the headspace gas phase is raised and the enrichment factor is improved. In order to evaluate the enrichment ability of this system, extraction experiments were performed with 15 polycyclic aromatic hydrocarbons (PAHs) standard mixture using original HS-LPME technique and GP-HS-LPME system, respectively, and the extraction solvents were analyzed by GC-MS. The results indicate that a higher enrichment factor can be obtained by GP-HS-LPME system compared with the one obtained by original HS-LPME technique. Moreover, the GP-HS-LPME system represents advantages of simplicity of the operation, automation, and accuracy of the control in extraction conditions, and rapidness of the extraction.</p><!><p>A semiconductor condenser used for GP-HS-LPME, which was an application of Peltier effect [18, 19] of semiconductor to refrigeration, was constructed according to the principle of thermoelectric refrigeration [20]. As is shown in the upper part of Figure 1, the semiconductor condenser is composed of the aluminum box, temperature sensor, copper rod, refrigeration piece, insulation cover, heat sink, cooling fan, and the brackets for microsyringe.</p><p>As is shown in Figure 1, the hot side of the refrigeration piece was attached to the heat sink and cooling fan combination to dissipate the generated heat. An aluminum box inside which a column was machined to fix a hollow copper rod (external and internal diameter of the copper rod are 1.4 mm and 0.5 mm, resp.) was attached to the cold side of the refrigeration piece. In order to detect the current temperature of the condenser, a hole was machined at the side of the aluminum box to embed the temperature sensor. To protect the extraction process from the influence of environmental temperature, a perspex insulation cover was mounted covering the aluminum box and refrigeration piece and it was fixed tightly to the heat sink by screws. Two small holes were machined separately in the upper and lower parts of the insulation cover to pass through the copper rod. On one side of the heat sink, brackets for microsyringe were mounted to hold the microsyringe at a proper settled height. After passing the microsyringe needle continuously through the brackets and the hollow copper rod, the microsyringe was finally stabilized at the brackets.</p><p>To increase the temperature of sample matrix, a metal-oxide ceramic heater (MCH) was designed in GP-HS-LPME system. Besides, for enhancing enrichment factor, inert gas was introduced into the system. As is shown in the lower part of Figure 1, the heater consists of the copper bed, heater band, two PTFE caps, glass tube, sample tube, glass wool layer, temperature sensor, and the gas pipe. A cylinder was machined inside the copper bed for placing the sample tube and a glass wool layer on which sample was placed was set inside the sample tube. Two PTFE caps were used to cover the top and bottom of the sample tube, respectively. A glass tube (external and internal diameter are 3.7 mm and 1.8 mm, resp.), which was inserted into the top PTFE cap to serve as the gas outlet channel for the inert gas, made the GP-HS-LPME system become an open system. A gas pipe bringing in the inert gas was inserted into the sample tube by sticking it in the bottom PTFE cap. The gas pipe was connected to the mass flow controller (not shown in Figure 1), which was used for measuring and adjusting the gas flow of inert gas led into the system. To monitor the current temperature of the heater, a temperature sensor was embedded in the copper bed. A MCH heater band was attached on the surface of the copper bed so as to generate heat to heat the sample tube when electric current was applied to it.</p><p>In order to achieve a good cooling effect of the extraction solvent, the copper rod described above in the condenser was plunged into the glass tube (gas outlet channel) of the heater, which was close to the microsyringe needle during the extraction process.</p><!><p>The GP-HS-LPME system is mainly composed of the semiconductor condenser, heater, switch power supply, microcontroller, keyboard, and LCD. Switch power supply is employed for supplying power to the system, keyboard is equipped to set refrigerating temperature of the semiconductor condenser and to set the heating temperature of the heater and gas flow rate of inert gas introduced, and LCD is used to display the above parameter values accordingly. Figure 2 illustrates the electrical schematic diagram of the total system.</p><p>The semiconductor condenser consists of AT89C51 microcontroller [21], DS18B20 1-wire digital temperature sensor [22], FPH1-3120NC semiconductor refrigeration piece, drive circuit for the refrigeration piece, switch power supply, LCM141 LCD module [23], and 4 × 4 keyboard. The heater is composed of AT89C51 microcontroller, DS18B20 temperature sensor, MCH heater band, drive circuit for the heater band, PCF8591 8-bit A/D and D/A converter [24], S49-32B/MT mass flow controller [25], switch power supply, LCM141, and 4 × 4 keyboard.</p><p>After setting refrigerating and heating temperature, gas flow rate of inert gas and timing time by keyboard and starting the GP-HS-LPME system, by control of the microcontroller, the whole system can work automatically according to the set value and the working process is as follows.</p><p>Figure 3 illustrates the control circuit diagram of GP-HS-LPME system. First, two DS18B20 temperature sensors are used to measure the current temperatures of the semiconductor condenser and the heater, respectively; they convert the two temperatures directly into two digital electric signals with 12-bit reading each (for each DS18B20, default 12-bit resolution is adopted and the converted thermal data is stored in the scratchpad memory in a 16-bit, sign-extended two's complement format, sign bits indicate whether the temperature is positive or negative). Next, the two digital signals are transferred, respectively, over the 1-wire interface of DS18B20 to AT89C51 microcontroller by issuing Read Scratchpad [BEh] commands when the temperature conversions have been performed. Then, the microcontroller calculates error amounts by comparing the temperatures measured by two DS18B20 sensors with the ones set by keyboard, and PID algorithm is employed to figure out controlled variables. Two PWM (Pulse Width Modulation) signals are generated according to the controlled variables and are used to drive the refrigeration piece and MCH heater band to work by connecting the PWM signals to drive circuits for the refrigeration piece and the heater band, respectively. If the measured temperature value by DS12B20 sensor does not correspond to the set value, the microsyringe needle and sample tube will be cooled and heated by the refrigeration piece and MCH heater band, respectively, and ultimately reach and retain the set value until the timing time is over.</p><p>At the same time, S49-32B/MT mass flow controller (MFC) is adopted in GP-HS-LPME system to measure and control the gas flow of inert gas introduced into the sample tube through the gas pipe. The detected gas flow analog signal by MFC is transferred to a piece of PCF8591 to be A/D-converted to 8-bit digital signal, and then the signal is input into AT89C51 microcontroller through I2C interface. The microcontroller calculates error amount by comparing the measured gas flow rate with the one set by keyboard and PID algorithm is adopted to figure out controlled variable, which is converted into digital control signal and exported by microcontroller. To be input into drive circuit for gas flow solenoid valve (inside the mass flow controller), the exported digital control signal should be connected to the piece of PCF8591 to be converted into analog control signal, which will be connected to the drive circuit and drive the gas flow solenoid valve to turn up or turn down so as to adjust gas flow of the inert gas. Thus, inert gas introduced into the system will finally reach and remain the set gas flow rate until the timing time is over.</p><p>In conclusion, through the control of the AT89C51 microcontroller, the whole system can achieve the required refrigerating and heating temperature and gas flow of inert gas rapidly and automatically.</p><!><p>To perform an extraction experiment using GP-HS-LPME system, the operation process is as follows. (1) The real or standard sample was put on the glass wool layer inside the sample tube; the sample tube was put into the copper bed and covered by two PTFE caps both on the top and the bottom. (2) The glass tube was inserted into the top PTFE cap and the copper rod was plunged into the glass tube. The gas pipe, which brings in inert gas, was inserted through the bottom PTFE cap. (3) Suitable extraction solvent was added into the microsyringe, and then the microsyringe was inserted continuously through the brackets for microsyringe and the copper rod into the glass tube. (4) The microsyringe plunger was depressed and solvent microdrop formed on the tip of the microsyringe needle, and the height of glass tube was adjusted so as to make the microdrop of extraction solvent locate where it just fills the glass tube. (5) The system was applied to set suitable values of gas flow rate, refrigerating and heating temperatures and timing time, and then the extraction started. (6) After the set extraction time is time out, the solvent microdrop was retracted back to the microsyringe and the microsyringe was removed from the system. Finally, the extraction solvent within the microsyringe was injected to the GC-MS for composition analysis.</p><!><p>The gas-purged headspace liquid phase microextraction system was applied in the determination of volatile and semivolatile chemicals. The phenanthrene, anthracene, fluoranthene, and pyrene were used as typical chemicals. In order to obtain high enrichment efficiency for volatile and semivolatile analytes from various kinds of samples, the parameters that affect enrichment factor in GP-HS-LPME system, such as the gas flow rate, the position of the extraction solvent microdrop, the diameter of the glass tube, the temperatures of the extraction solvent and the sample, and the extraction time, were systematically optimized. The optimized conditions were 2.7 mL min−1 for the gas flow rate of the inert gas, the extraction solvent microdrop filling the glass tube, 1.8 mm for the internal diameter of the glass tube, −6°C and 80°C for the temperatures of the extraction solvent and of the sample, respectively, and 20 min for the extraction time. Of the various parameters, higher temperature of the sample and lower temperature of the extraction solvent are greatly favorable for high enrichment factor and reducibility of the GF-HS-LPME technique as reported by Yang et al. in 2009 [14]; the desirable values of the parameters are easily obtained and accurately controlled using semiconductor condenser and a metal-oxide ceramic heater developed here.</p><!><p>Under the optimization conditions, 15 PAHs standard mixture samples were extracted using the GP-HS-LPME system for evaluation of enrichment factor and reproducibility of the technique. In addition, contrastive experiment was done under the identical optimized parameters for the same kind of sample using the HS-LPME technique. The extraction solvent chosen for the two extraction experiments was dodecane and the amount of extraction solvent used for each was controlled at 2 μL for the subsequent GC-MS analysis. In this study, the 15 PAHs standard mixture was also used to evaluate the reproducibility of GP-HS-LPME system and original HS-LPME technique. The reproducibility was represented by the relative standard deviation (RSD). It can be seen from Table 1 that enrichment efficiency of GP-HS-LPME was enhanced by 4 times higher than this of HS-LPME, and the RSD value of the GP-HS-LPME system ranged from 4.56% to 9.45% and this value ranged from 8.96% to 19.38% for the HS-LPME technique, which demonstrated that GP-HS-LPME revealed better reproducibility.</p><!><p>In order to evaluate sensitivity of the GP-HS-LPME system on the volatile and semivolatile chemicals, the 15 PAHs standard mixture which covered wide range of boiling point (from 298.9°C to 561.1°C at 760 mmHg) was used as target chemicals, and the results were compared with this obtained by the HS-LPME. As is shown in Figures 4(a) and 4(b) denote the chromatograms of target compounds analyzed by GC-MS after using HS-LPME and GP-HS-LPME technique. Total of 11 and 9 compounds were, respectively, detected by the GP-HS-LPME and HS-LPME. In Figure 4, Numbers 1–11 indicate 11 kinds of target compounds enriched by the two kinds of extraction techniques. Although 1–9 compounds were detected from both GP-HS-LPME and HS-LPME, the intensity obtained by the GP-HS-LPME was about 3-4 times as high as those in the HS-LPME. Furthermore, 10-11 compounds were hardly found in the HS-LPME case, while they were easily found in the GP-HS-LPME case. Comparing the two intensity values of each target compound enriched by HS-LPME and GP-HS-LPME, it could be concluded that higher enrichment efficiency could be obtained by using GP-HS-LPME system developed here for determination of volatile and semivolatile chemicals in contrast with the usage of HS-LPME technique.</p><!><p>With the use of semiconductor condenser and heater, the volume of the system was reduced and the stability was improved. The advantages of simplicity of operation, automatic control of experimental parameters (conditions), and high efficiency of extraction enable GP-HS-LPME system to be used in enriching both volatile and semivolatile target compounds from various kinds of samples. The enrichment factor of the GP-HS-LPME is 3 or 4 times as high as this value of HS-LPME. In addition, because of its ability of low-voltage power supply (e.g., a car battery) and miniaturization of the device, GP-HS-LPME system enables on-site and online field enrichment of analytes from different samples. Thus, GP-HS-LPME system has a wide prospect that was applied in the enrichment of different kinds of target compounds from various sample matrixes in fields such as food chemistry, biochemistry, and environmental chemistry.</p>

PubMed Open Access

Insight into CaO2-based Fenton and Fenton-like systems: strategy for CaO2-based oxidation of organic contaminants

This study conducted a comparison of the CaO2-based Fenton (CaO2/Fe(II)) and Fenton-like (CaO2/Fe(III)) systems on their benzene degradation performance. The H2O2, Fe(II), Fe(III), and HO\xe2\x97\x8f variations were investigated during the benzene degradation. Although benzene has been totally removed in the two systems, the variation patterns of the investigated parameters were different, leading to the different benzene degradation patterns. In terms of the Fe(II)/Fe(III) conversion, the CaO2/Fe(II) and CaO2/Fe(III) systems were actually inseparable and had the inherent mechanism relationships. For the CaO2/Fe(III) system, the initial Fe(III) must be converted to Fe(II), and then the consequent Fenton reaction could be later developed with the regenerated Fe(II). Moreover, some benzene degradation intermediates could have the ability to facilitate the transformation of the Fe(III) to Fe(II) without the classic H2O2-associated propagation reactions. By varying the Fe(II) dosing method, an effective degradation strategy has been developed to take advantage of the two CaO2-based oxidation systems. The proposed strategy was further successfully tested in TCE degradation, therefore extending the potential for the application of this technique.

insight_into_cao2-based_fenton_and_fenton-like_systems:_strategy_for_cao2-based_oxidation_of_organic

Introduction<!>Experiment procedure<!>Analytical method<!>Benzene degradation<!>Mechanisms of the CaO2/Fe(II) and CaO2/Fe(III) systems in benzene degradation<!>Fe(II)/Fe(III) conversion in the CaO2/Fe(II) and CaO2/Fe(III) systems<!>H2O2 variation in the CaO2/Fe(II) and CaO2/Fe(III) systems<!>HO\xe2\x97\x8f generation in the CaO2/Fe(II) and CaO2/Fe(III) systems<!>Intermediates analysis and benzene degradation pathway in the CaO2/Fe(II) and CaO2/Fe(III) systems<!>An efficient strategy for employing the CaO2-based oxidation system<!>Implications for the application of the CaO2-based oxidation system<!>Conclusions

<p>Calcium peroxide (CaO2) is a solid oxidant and has been attracting more and more attention, due to its versatile chemical properties. Based on the Eq. (1), CaO2 usually has been considered as an oxygen releasing compound (ORC) and is therefore applied to oxygen-demanding environmental remediation for many years [1–5]. In terms of the Eq. (2), CaO2 has the potential to act as a solid hydrogen peroxide (H2O2) source, and in recent years a growing number of studies are developing various CaO2 based chemical oxidation systems [6]. It is reported that CaO2 alone can degrade toluene [7] and 2,4-dichlorophenol [8], and can work as an additive in the permeable reactive barrier [9]. Moreover, ozone [10], persulfate [11,12], ultraviolet [13], and other activators have also been studied to promote the CaO2 performance in contaminants remediation [14,15]. Among these emerging oxidation techniques, the CaO2-based Fenton (CaO2/Fe(II)) and Fenton-like (CaO2/Fe(III)) systems are the promising prototypes of the developed CaO2 based oxidation systems, because of their effective decontamination performance [16–23].</p><p>The CaO2/Fe(II) and CaO2/Fe(III) systems have been used to treat various contaminated soil and water, and achieved satisfied decontamination performances. Table 1 summarizes the target pollutants which were treated both systems. It is clear that the CaO2/Fe(II) and CaO2/Fe(III) systems were usually considered and studied separately. Most of the testes just concentrated on the decontamination performance under their operational conditions and investigated the influence of some environmental parameters, such as the solution pH, solution matrix, and other common factors [16–23]. Although some studies mentioned the comparison of the CaO2/Fe(II) and CaO2/Fe(III) systems, the aim of these comparisons was not to clarify the relationships among each other. Zhang et al. compared the pollutant degradation performance of the CaO2/Fe(II) and CaO2/Fe(III) systems, but only highlighted the characteristics of their technique and then optimized their experimental conditions [22]. The comparison of the CaO2/Fe(II) and CaO2/Fe(III) systems, so far, are still not thoroughly studied, and the inherent relationships between the two systems are lacking sufficient researches.</p><p>According to the literature reviews on the CaO2/Fe(II) and CaO2/Fe(III) systems, the well-accepted mechanisms for the two systems are concluded (Table 1). Although HO● in the CaO2/Fe(II) and CaO2/Fe(III) systems both originated from the reaction involving Fe(II) and H2O2, the two systems differ in the specific mechanisms. In the CaO2/Fe(II) system, due to the initial Fe(II) presence, HO● can be directly generated from the reaction between Fe(II) and the released H2O2. Previous studies also reported that most of the initial Fe(II) could be quickly converted to Fe(III), and the CaO2/Fe(II) system then actually performed as the CaO2/Fe(III) system for the rest of the experimental time [20,22]. In contrary, for the CaO2/Fe(III) system, the reaction involving Fe(II) and H2O2 is still the dominant HO● source, but the Fe(II) here is from the reduction of the initial Fe(III), and it is the regenerated Fe(II) that causes the Fenton reaction in the CaO2/Fe(III) system. Therefore, the CaO2/Fe(III) system also contains the CaO2/Fe(II) system [26,27]. Although the CaO2/Fe(II) and CaO2/Fe(III) systems were independently studied, the HO● formation is the widely recognized mechanism for both systems [6,24–26]. Moreover, the highlights mentioned above show that that the CaO2/Fe(II) and CaO2/Fe(III) systems are actually related and might have the intrinsic relationships, thus it is necessary to conduct further comparable investigations into the two systems, clarifying the connections between the two systems.</p><p>Based on the literature review, trichloroethene (TCE) [17,19,22], BTEX (benzene, toluene, ethylbenzene, and xylenes) [20,23,28], PAHs [13], and other refractory organic pollutants [9–12,16,29–31] have been treated with the CaO2-based oxidation techniques. Among these pollutants, benzene is a typical one in many contaminated sites and also listed as a toxic organic compound [32,33]. Since the reactivity of benzene and HO● is very high (kHO● = 7.8 × 109 M−1 s−1) [34], benzene prefers to react with HO● in the CaO2-based oxidation systems. Thus, benzene, as a target pollutant, was selected in this study, which can provide more convincing explanations of the mechanisms in both systems without other potential debates. Based on the mechanism analysis, an efficient strategy was proposed to optimize the CaO2-based oxidation systems. This strategy was then further tested in TCE treatment, since TCE is another common toxic environmental pollutant that is reported to possess a high reactivity with HO● (kHO● = 3.5 × 108 M−1 s−1) [17,34–36]. This makes TCE suitable to check the proposed strategy, which could further extend the adaptability of our technique in the application of groundwater remediation.</p><p>Therefore, the objectives of this study were to (1) compare and evaluate the performances of the CaO2/Fe(II) and CaO2/Fe(III) systems on benzene removal; (2) compare and clarify the intermediates generated in the CaO2/Fe(II) and CaO2/Fe(III) systems; (3) compare and elucidate the benzene degradation mechanisms of the CaO2/Fe(II) and CaO2/Fe(III) systems; and (4) combine the advantages of the CaO2/Fe(II) and CaO2/Fe(III) systems extending the potential application to different pollutants.</p><!><p>The reactor of 250 mL glass vessel with a water jacket was used in this study and the temperature for all experiments was controlled at 20 ± 2°C. The equilibrated benzene solution was added into the reactor and then diluted to the desired concentration, sealing the reactor and mixing solution with a 600 rpm stirring speed. The initial benzene concentration was set at 1.0 mM, and experiments showed that the volatilization and absorption of benzene caused limited influence in this study (Supplementary Material, Fig. S1). The initial solution pH in this study was adjusted to 3 before dosing Fe(II) or Fe(III) reagent, and the reaction started as soon as adding the predetermined dose of CaO2. At the desired time, 2.5 mL sample was collected in headspace vials containing 1.0 mL methanol solution to stop the reaction. The vial was sealed immediately and then analyzed by headspace-gas chromatography (HS-GC). The tests were conducted duplicate, and the mean values were displayed. The results derived from benzene degradation were further tested in the treatment of TCE, which is another frequently detected toxic pollutant in groundwater and also has a high reactivity with HO● [34–36]. The initial concentration of TCE was set at 0.15 mM, and 1 mL sample was collected in vials containing 1.0 mL hexane at the predetermined time intervals. After the extraction procedure, the extracts were immediately analyzed by GC.</p><p>For the HO● measurement, 5,5-Dimethyl-1-Pyrroline N-oxide (DMPO, 8.84 mM) was applied to capture the generated HO● in the CaO2/Fe(II) and CaO2/Fe(III) systems. At the determined time, a 1.0 mL sample was collected from the reactor and then well mixed with 1.0 mL DMPO solution, and then the sample was analyzed using the electron paramagnetic resonance (EPR).</p><p>All the chemicals used in this study are listed in Supplementary Material (Text S1). The detailed procedures for the detection of the intermediates could be found in Supplementary Material (Text S2).</p><!><p>An Agilent HS-GC (Agilent 7890B, Palo Alto, CA, USA) has been used to analyze benzene samples, which coupled with a flame ionization detector (FID), an HP-5 column (30 m length, 0.32 mm I.D., 0.25 μm thickness), and an auto-sampler (Agilent 7967A, Palo Alto, CA, USA). The HO● was captured by DMPO and then tested using the EPR (EMX-8/2.7C, Bruker, Germany) instrument. TCE was measured using a GC (Agilent 7890A, Palo Alto, CA, USA) equipped with an autosampler (Agilent 7693), an electron capture detector (ECD), and a DB-VRX column (250 μm i.d., 1.4 μm thickness, and 60 m length). The intermediates produced in the CaO2/Fe(II) and CaO2/Fe(III) systems were identified by GC/MS (Agilent 6890/5973N, Palo Alto, CA, USA) and high performance liquid chromatography (HPLC, LC-20AT, Shimadzu, Japan). The specific conditions for the analyses can be found in Supplementary Material (Text S2).</p><p>For the analysis of H2O2, the filtered sample was analyzed after the reaction with TiSO4 [37]. The concentration of the available Fe(II) and total Fe was determined based on the 1,10-phenanthroline method [38], and the total Fe concentration was determined following Fe(II) procedure after dosing hydroxylamine. The difference between the total Fe and Fe(II) was used to quantify the Fe(III) concentration [39]. The solution pH was measured by a pH meter (Sartorius, PB-10, Germany) equipped with a pH/ATC electrode (Sartorius, Germany).</p><!><p>The benzene treatment performance of the CaO2/Fe(II) and CaO2/Fe(III) systems were compared and the results have been shown in Fig. 1. Benzene was completely removed in the two systems, but had differing benzene degradation patterns.</p><p>In the CaO2/Fe(II) system, benzene degradation could be fiinshed in 10 min. When varying the molar ratio of CaO2/Fe(II) from 5/5 to 15/15, benzene removal efficiencies were only enhanced 10% (Fig. 1a), while the observed apparent kinetic reaction rates significantly increased from 0.42 to 4.62 M−1 s−1 according to the second-order kinetic model (Table 2). The results indicating that the molar ratio of CaO2/Fe(II) just slightly influenced the benzene removal efficiency and the benzene degradation rate was regulated by the molar ratio of CaO2/Fe(II). For the CaO2/Fe(III) system, the benzene degradation patterns were different from that in the CaO2/Fe(II) system. When the molar ratio of CaO2/Fe(III) was 5/5, the benzene degradation exhibited a two-stage degradation. During the first 200 min, the benzene removal efficiency was less than 10%, indicating a slow degradation stage; in the following 100 mins, an obvious increase in benzene degradation efficiency leading to a complete removal of benzene by 300 min, indicates a switch to a fast degradation stage. When increasing the molar ratio of CaO2/Fe(III) to 10/10 and 15/15, the benzene degradation finished within 120 and 80 min, respectively (Fig. 1b), while the apparent kinetic reaction rate increased from 5.50 × 10−5 to 1.29× 10−3 M−1 s−1 (Table 2). The results suggested that the benzene degradation performance of the CaO2/Fe(III) system greatly relies on the molar ratio of CaO2/Fe(III) and thus the slow degradation stage could be shortened by increasing the molar ratio CaO2/Fe(III).</p><p>The above results indicated that the CaO2/Fe(II) system could achieve a faster benzene degradation than that of the CaO2/Fe(III) system. Although the nano-scale CaO2 alone has the ability to remove some pollutants [6–8], our preliminary study showed that the CaO2 alone only removed little benzene (Supplementary Material, Fig. S1). Hence, it was the Fe(II) or Fe(III) that enhanced benzene degradation in both systems, and the observed different benzene degradation patterns could be ascribed to the different mechanisms of the two systems.</p><!><p>The varied benzene degradation patterns indicated that the benzene degradation mechanisms of the CaO2/Fe(II) and CaO2/Fe(III) were different, so the investigation on the mechanisms would reveal the underlying cause of the disparity between the two systems. The preliminary EPR analysis confirmed that HO● was the dominant reactive oxygen species in the CaO2/Fe(II) and CaO2/Fe(III) systems (Supplementary Material, Fig. S2). Since the dose of H2O2, Fe(II), and Fe(III) could influence HO● generation in both investigated systems, the temporal variations of H2O2, Fe(III), Fe(II), and HO● were studied to clarify the benzene degradation mechanisms of both systems. The mechanism investigation was studied using 1 mM benzene, 10 mM CaO2, and 10 mM Fe(II) or Fe(III).</p><!><p>The Fe(II)/Fe(III) conversion is a key factor in the conventional Fenton and Fenton-like systems, thus it is necessary to investigate the Fe(II)/Fe(III) conversion in the CaO2/Fe(II) and CaO2/Fe(III) systems, and the results were presented in Fig. 2.</p><p>The Fe(II)/Fe(III) speciation of the CaO2/Fe(II) system was shown in Fig. 2a. In the blank experiment (without benzene), it was clear that Fe(II) declined to a low concentration in the first few minutes and maintained in this level throughout the experiment, which behaved like a conventional Fenton reaction (Eq. 3). In contrary, for the CaO2/Fe(III) system, it was observed that only trace Fe(II) was detected during the experiment and the Fe(II)/Fe(III) conversion was not clear in this condition. However, the addition of 1 mM benzene led to the different Fe(II)/Fe(III) conversion in both systems. For the CaO2/Fe(II) system, it was still observed a quick decline in the initial Fe(II) concentration, but a clear Fe(II) recovery then appeared in the first few minutes. The recovered Fe(II) obviously declined to a low level at 60 min, accompanied with a slight Fe(II) increase along the rest reaction period. For the CaO2/Fe(III) system, there was still a small amount of Fe(II) in the first 90 min, then the detected Fe(II) concentration increased to the maximum concentration and dropped to a low level accompanied by a gradual increase. This observed Fe(II)/Fe(III) conversion is the typical iron variation in Fenton-like systems and was also reported in previous studies [39,40].</p><p>As concluded in Table 1 [41,42], the reaction involved Fe(II) and H2O2 is the main HO● source regardless in Fenton or Fenton-like systems. In the CaO2/Fe(II) system, HO● was directly generated from the reaction between Fe(II) and H2O2, which then resulted in a quick benzene degradation. However, for the CaO2/Fe(III) system, the reaction between Fe(III) and H2O2 cannot directly produce HO●, it is the regenerated Fe(II) which reacted with H2O2 to produce HO●. Since the Fe(II) regeneration reaction rate (Eq. 4) was slower than that of the Eq. 3 (kFe(II), H2O2 = 7.6 M−1 s−1, kFe(III), H2O2 = 2.0 × 10−3 M−1 s−1) [26,27], it was observed a clear lag period before the obvious benzene degradation in the CaO2/Fe(III) system. When increasing the Fe(III) dose, the reaction between Fe(III) and H2O2 would be also facilitated and consequently accelerated the Fe(II) regeneration and HO● formation, leading to a shorter lag period (Fig. 1b). Although many studies reported that the H2O2-associated propagation reactions in classic Fenton or Fenton-like could cause the Fe(III) reduction [41–45], some reports [46–48] also mentioned that the HO2●− could be generated in the CaO2-based systems. This could have the potential to reduce Fe(III) to Fe(II) [48], but the results of the blank experiments indicated that the observed Fe(II) recovery was not owed to these pathways. With the addition of benzene, the clear Fe(II) recovery indicated that the benzene degradation reactions could bring additional reactions participated in Fe(III) reduction in both systems [45]. It is possible that some intermediate products during the benzene degradation could be responsible for Fe(II) recovery (Eq. 6), which were more efficient than the reduction of Fe(III) by H2O2-associated propagation reactions (Eqs. 4 ~ 5). The roles of intermediate products will be discussed in the later section to clarify their effects on the benzene degradation.</p><!><p>The variation of H2O2 is an important factor in the studied systems and the temporal H2O2 changes in the CaO2/Fe(II) and CaO2/Fe(III) systems were shown in Fig.3.</p><p>As shown in Fig. 3a, for the CaO2/Fe(II) system, in the absence of benzene (the blank experiment), the H2O2 concentration peaked at 5 min and then dropped to a low concentration by 60 min. In the presence of benzene, the detected H2O2 would gradually increase to the maximum concentration by 60 min and then H2O2 would slowly and continuously decline in the remaining test period. As for the CaO2/Fe(III) system, regardless of the presence of benzene, high H2O2 concentrations were measured within the first 120 min, but the H2O2 concentration would steadily decline in the following experiment period in the presence of benzene (Fig. 3b).</p><p>The reaction with Fe(II) is the dominant H2O2 sink in the two systems because of the high reaction constant (kFe(II), H2O2 = 7.6 M−1 s−1, kFe(III), H2O2 = 2.0 × 10−3 M−1 s−1) [26,27], and the maximum H2O2 depletion should be observed when the obvious Fe(II) recovery was obtained. Therefore, the H2O2 variation, to some extent, was relevant to Fe(II) variation in the two systems. According to the Fe(II)/Fe(III) variations (Fig. 2), it was clear that the consumption of H2O2 was much faster in the CaO2/Fe(II) system, while H2O2 concentration was maintained at a high level for a much longer time in the CaO2/Fe(III) system. In addition, the presence of benzene could also promote the H2O2 consumption in both systems, which could be mainly ascribed to its influence to the Fe(II)/Fe(III) variations (Fig. 2). Based on the results, it could be concluded that the rapid H2O2 consumption in the CaO2/Fe(II) system could lead to a rapid pollutant degradation; while the CaO2/Fe(III) system could maintain its oxidation capacity for a longer time, which can be an advantage for minimizing the unexpected reagents loss during the injection and distribution process for subsurface remediation application.</p><!><p>Since the Fe(II)/Fe(III) conversion and H2O2 decomposition patterns were different in the CaO2/Fe(II) and CaO2/Fe(III) systems, it could be inferred that the HO● generation pathways could be also different. The HO● variation was analyzed by the EPR instrument and the results were shown in Fig. 4.</p><p>Fig. 4a showed HO● variation during the benzene degradation in the CaO2/Fe(II) system. At the beginning, the HO● intensity was very high, and then the intensity decreased obviously. For the CaO2/Fe(II) system, due to the adaquate Fe(II), the reaction between Fe(II) and the released H2O2 could quickly occurred and produce excessive HO● in a short time leading to a high HO● intensity at 0.5 and 1 min. Due to the consequent Fe(II) exhaustion and lack of H2O2 (Eqs. 3 ~ 5), a notable decline in the HO● intensity was measured in the following experiment period (Fig. 4a). Conversely, owing to the gradual increase of H2O2 (Fig. 3a), the HO● intensity increased, presenting an enhancement at 30 min (Fig. 4a).</p><p>As for the CaO2/Fe(III) system, the reaction between Fe(III) and the released H2O2 could not directly produce HO●: Fe(III) should be reduced to Fe(II) and then it was the regenerated Fe(II) reacted with the released H2O2 producing HO● (Figs. 2b & 4b) [26,27,49]. Hence, in the initial 90 min, due to the low Fe(II) concentration and slow Fe(II) regeneration (Fig. 2b), the measured HO● intensity was weak at 30 and 90 min (Fig. 4b). Then with a notable increase of Fe(II) recovery, the HO● intensity was obviously promoted at 110 min (Figs. 2b & 4b). After the significant intensity enhancement, as a result of Fe(II) and H2O2 exhaustion, the detected HO● intensity at 130 min was similar to that detected before the enhancement. This significant increase of HO● intensity, together with the Fe(II)/Fe(III) conversion and H2O2 variation, reveals that some degradation intermediates could facilitate the HO● production.</p><!><p>The intermediate products in the CaO2/Fe(II) and CaO2/Fe(III) systems were analyzed with the HPLC and GC/MS. The analytical results showed that phenol, catechol, and benzoquinone were the major intermediates (Supplementary Material, Fig. S3), but their generation patterns were different in the investigated systems. It was clear that phenol was the primary degradation product in the the CaO2/Fe(II) and CaO2/Fe(III) systems, but the observed phenol variations were different. In the CaO2/Fe(II) system, it was observed that phenol sharply increased and declined within the first 10 min (Fig. 5a); in contrary, in the CaO2/Fe(III) system, phenol displayed a gradual increase and reached to its peak around 120 min (Fig. 5b). Along with the increase of phenol, benzoquinone and catechol also reached to their peaks in the two systems. Moreover, all the intermediates peak values were recorded during the fast benzene degradation period, and the benzene degradation pathways are then proposed based on the above experimental results (Fig. 6).</p><p>As for benzene degradation in the CaO2/Fe(II) system, the released H2O2 would quickly react with Fe(II), producing excessive HO● in a short time. The produced HO● then immediately attacked benzene resulting in a rapid, significant decline in benzene while increase in phenol, which then would react with the surrounding HO● generating catechol and hydroquinone. Meanwhile, the generated catechol and hydroquinone could simultaneously reduce Fe(III) to Fe(II) (Eqs. 7 & 8), which could result in the notable Fe(II) increase and benzoquinone formation (Figs. 2a & 5a) [40,50–52]. All the intermediate products would competed for HO● with benzene, presenting clear declines (Fig. 5a). In regards to the CaO2/Fe(III) system, due to the low initial Fe(II) concentration and slow Fe(II) regeneration, the HO● was gradually generated (Figs. 2b & 4b), which limited the benzene degradation rate and resulted in a gentle phenol accumulation (Fig. 5b). With the increas of phenol, in addition to the initial benzene, phenol became another target compound for HO●, and resulted in the increase of catechol and hydroquinone [40,51]. Since catechol and hydroquinone have the ability to reduce the initial Fe(III) to Fe(II), Fe(II) recovery (Fig. 2b) and the subsequent HO● burst (Fig. 4b) along with the intermediates accumulation were then observed [52–54]. When the benzene degradation was completed, the intermediates variations also settled down. In addition, it was found that the HPLC spectrums of the degradation products in the CaO2/Fe(II) system could present the higher levels than that in the CaO2/Fe(III) system. The TOC analysis further confirmed that the CaO2/Fe(II) system produced less benzene mineralization than the CaO2/Fe(III) system (Supplementary Material, Fig. S4).</p><!><p>By comparing the benzene degradation performance in the two systems, it was easy to conclude that the CaO2/Fe(II) system can degrade benzene in a short time, while the CaO2/Fe(III) system can maintain its oxidation capacity for a relatively long time. In comparison to regulate the released H2O2, to manipulate the Fe(II)/Fe(III) conversion could be easier which also has a significant influence on the benzene degradation performance. Thus, we tested different combinations of Fe(II) and Fe(III), attempting to take the advantages of both systems and develop an efficient strategy for applying the CaO2-based oxidation system to groundwater remediation. Fig. 7a showed the benzene degradation results with different combinations of Fe(II) and Fe(III). With Fe(II) addition, it was observed that benzene was completely removed within 5 min regardless of Fe(II) and Fe(III) dosage. According to the previous analysis, the CaO2/Fe(III) system could hold the advantage in maintaining its stable oxidation capacity for a long-term and the lag time before finishing the benzene degradation could be very long. Hence, the addition of Fe(II) could be able to responsible for fast benzene degradation. Moreover, the result suggested that Fe(II) could possibly act as an accelerator to reduce the lag time in the CaO2/Fe(III) system and further experiments were conducted to clarify this hypothesis. 5.0 mM Fe(III) and 5.0 mM CaO2 were added in the reactor, then 1.0 mM Fe(II) was introduced at different times to test whether the benzene degradation could be accelerated, and the results have been shown in Fig. 7b. Before dosing Fe(II), the CaO2/Fe(III) system experienced very slow benzene degradation, however, after the introduction of Fe(II), the benzene degradation was promoted and quickly finished. The significant enhancements clearly indicated that Fe(II) succeeded in accelerating the benzene degradation and Fe(II) can be used as a complementary stimulant in the CaO2/Fe(III) system.</p><p>The combination of Fe(II) and Fe(III) took the advantages of the CaO2/Fe(II) and CaO2/Fe(III) systems and resulted in a better benzene remediation. TCE is a typical chloride solvent and has been widely used in industry for several decades. Due to its high toxicity and widespread occurrence in groundwater, TCE can bring a significant threat to human health and the natural environment, and has been concerned as priority pollutant in many countries [35,36]. Based on the benzene degradation results, we tested the above strategy in TCE remediation to verify its adaptability. The TCE treatment results with the above strategy was presented in Fig. 7c. TCE degradation was very slow with the CaO2/Fe(III) system alone, however, after dosing Fe(II), it was observed that TCE degradation was greatly enhanced and the degradation finishied as soon as Fe(II) was added. The different TCE degradation patterns indicated that the Fe(II) addition also succeeded in regulating the TCE degradation. The observed TCE decontamination trends were similar to that of the benzene degradation, which demonstrated that the proposed strategy has a good adaptability.</p><!><p>Although the efficiency of the CaO2-based oxidation system in groundwater treatment have been tested with the laboratory-scale experiments, there are still some problems that should be noticed. Unlike the laboratory conditions, the environmental conditions are hard to control and vary significantly in the actual sites. Among the condition parameters, pH is one key factor that should concern us when using the CaO2-based oxidation system. According to the Eqs. 1 ~ 2, the CaO2 dissolution process will unavoidably bring additional alkaline into the solution. Although the natural aquifer has a certain buffer capacity, the CaO2 dissolution can still elevate the solution pH [55,56]. Previous studies reported that the solution pH could be over 10 (even over 11) with the inappropriate molar ratios of CaO2/Fe(II) or CaO2/Fe(III) [20,22,55]. However, it was reported that the Fenton and Fenton-like reactions prefer the pH around 3 [24–27,39], and many documents reported that the elevated solution pH could constrain the performance of the CaO2-based oxidation system in pollutant remediation [17,20,55]. The iron could easily precipitate with the high solution pH, which could inhibit the iron recycle [26,27,39,57]. Particularly, for the CaO2-based oxidation system, the high solution pH would drive the CaO2 dissolution reaction from Eq. 2 to Eq. 1. O2 would replace H2O2 as the dominant product when pH > 10 [55,56], suppressing the subsequent H2O2-based decontamination performance. Besides, recent studies demonstrated that the CaO2-based oxidation system could produce O2●- in neutral pH [48,58], and the HO● yield in the would be greatly inhibited in the alkaline solution [58]. Therefore, the actual remediation process requires some proper conditioning reagents, such as the sulfuric acid [58], chelates [21–23], to attune the site condition, after which the CaO2-based oxidation reagents could be injected into the contaminated sites. Based on the above discussion, a model using the CaO2-based oxidation technique is proposed to remediate subsurface contamination (Supplementary Material, Fig. S5).</p><p>Since the CaO2/Fe(III) reagents can maintain their oxidation capacity for a relatively long time, its greater persistence compared with the CaO2/Fe(II) system means that migration distance from the injection well is anticipated to be longer, which results in the extent of larger treatment zones. This, in turn, reduces the number of wells and injection rounds required (thereby reducing costs), thus the CaO2/Fe(III) reagents can be used as an effective injection reagent for the subsurface remediation. Once the CaO2/Fe(III) reagents are injected and well distributed in the contaminated zone, the degradation process can be greatly accelerated by injecting a small amount of Fe(II). During the entire remediation process, the environmental conditions of the contaminated zone are required to be monitored, and the operating parameters need to be adjusted according to the real-time feedback. Moreover, the potential influence of the CaO2-based oxidation system on the subsurface ecosystem is another concern, which needs more investigations when applying this technique to the actual remediation.</p><!><p>The results of this study showed the CaO2/Fe(II) and CaO2/Fe(III) systems have the intrinsic relationship. The CaO2/Fe(II) would quickly convert to the CaO2/Fe(III) system during the remediation, while the CaO2/Fe(III) system need convert to the CaO2/Fe(II) system to promote the decontamination. The benzene degradation in the CaO2/Fe(II) system was much faster than that in the CaO2/Fe(III) system, whereas the CaO2/Fe(III) system maintained stable oxidation capacity for a longer time. To take the advantages of both systems, we proposed an effective strategy to treat benzene and TCE using the CaO2-based oxidation system. These encouraging results could provide new knowledge of more effective CaO2-based oxidation technique for subsurface remediation.</p>

PubMed Author Manuscript

Pancreatic Acinar Cells: Molecular insight from studies of signal-transduction using transgenic animals

Pancreatic acinar cells are classical exocrine gland cells. The apical regions of clusters of coupled acinar cells collectively form a lumen which constitutes the blind end of a tube created by ductal cells-a structure reminiscent of a \xe2\x80\x9cbunch of grapes\xe2\x80\x9d. When activated by neural or hormonal secretagogues, pancreatic acinar cells are stimulated to secrete a variety of proteins. These proteins are predominately inactive digestive enzyme precursors called \xe2\x80\x9czymogens\xe2\x80\x9d. Acinar cell secretion is absolutely dependent on secretagogue-induced increases in intracellular free Ca2+. The increase in [Ca2+]i has precise temporal and spatial characteristics as a result of the exquisite regulation of the proteins responsible for Ca2+ release, Ca2+ influx and Ca2+ clearance in the acinar cell. This brief review discusses recent studies in which transgenic animal models have been utilized to define in molecular detail the components of the Ca2+ signaling machinery which contribute to these characteristics.

pancreatic_acinar_cells:_molecular_insight_from_studies_of_signal-transduction_using_transgenic_anim

1. Introduction<!>2. Cellular Origins<!>3. Functional significance of the expression of multiple InsP3R family members in acinar cells<!>4. The molecular basis of Ca2+ influx in pancreatic acinar cells<!>5. Dysregulated Ca2+ signaling is associated with pancreatitis<!>6. Concluding Remarks<!>

<p>The pancreas is a mixed exocrine and endocrine tissue, physically situated as an accessory organ connected to the gut tube at the level of the duodenum. The exocrine pancreas primarily consists of two major cell types; acinar and ductal cells. Together these cells are responsible for the production of a HCO− rich solution containing a plethora of proteins which are responsible for the efficient assimilation of nutrients. While the ductal cells account for the production of the vast majority of alkaline fluid, pancreatic acinar cells are classical polarized epithelial cells and are morphologically and functionally specialized for the exocrine secretion of protein containing vesicles across the apical membrane (Figure 1).</p><p>The exocytosis of secretory granules is absolutely dependent on secretagogue induced elevations in cytosolic free Ca2+ ([Ca2+]i) (Futatsugi et al., 2005). The important event leading to the increase in [Ca2+]i following binding of the neurotransmitter acetylcholine or gastrointestinal hormone cholecystokinin on the basolateral aspects of the acini (Figure 1) is the Gαq-stimulated increase in activity of phospholipase C, leading to the generation of the second-messenger inositol 1,4,5 trisphosphate (InsP3) (Williams & Yule, 2006). Binding of InsP3 to its receptors (InsP3R), which are calcium permeable channels localized in the endoplasmic reticulum (ER) results in Ca2+ release into the cytoplasm (Streb et al., 1983). Secretagogue stimulation also triggers Ca2+ influx from the extracellular space across the plasma membrane (PM). Contributions from both Ca2+ release and Ca2+ influx are necessary for continued, sustained exocytosis of granules (Tsunoda et al., 1990). Activation of this machinery results in intracellular calcium signals with complex spatial and temporal characteristics. For example, although production of InsP3 occurs predominately at the lateral and basal membranes, Ca2+ release invariable is initiated in the extreme apical pole of the acinus where the vast majority of InsP3R are located (Figure 2) (Kasai & Augustine, 1990; Thorn et al., 1993; Nathanson et al., 1994; Yule et al., 1997). At threshold levels of stimulus, Ca2+ signals can be confined to this domain, allowing standing gradients of [Ca2+]i to be established (Kasai et al., 1993; Thorn et al., 1993). It is thought that Ca2+ uptake by a peri-granular belt of mitochondria function as a "fire-wall" to confine signals to the apical domain (Tinel et al., 1999; Straub et al., 2000). Physiological levels of agonist result in the propagation of a Ca2+ wave traveling from the initiation sites in the so called apical "trigger zone" towards the basal region of the cell (Kasai et al., 1993; Thorn et al., 1993) (Figure 2A). The Ca2+ wave relies on Ca2+-induced Ca2+ release via InsP3R and ryanodine receptors distributed on ER throughout the cytoplasm (Straub et al., 2000). In the continued presence of secretagogue, extrusion of Ca2+ across the PM and sequestration into mitochondria and the ER serves to reduce [Ca2+]i prior to the initiation of a further apical to basal Ca2+ wave (Figure 2B) (Lee et al., 1997a; Tinel et al., 1999; Straub et al., 2000). These spatial characteristics result temporally in the generation of Ca2+ spikes or oscillations (Yule & Gallacher, 1988; Yule et al., 1991). Together these stereotyped features of the Ca2+ signals are thought to be fundamentally important in the appropriate activation of effectors necessary for secretion (Maruyama et al., 1993).</p><p>These remarkable spatial and temporal characteristics are attributed to the dynamic interplay between the proteins responsible for Ca2+ release, influx and clearance from the cytoplasm (Figure 2B). While the general class, or in some cases, the gene family of the proteins responsible has been indicated, it is common that multiple individual members of gene families are expressed. With little discriminating conventional pharmacology available, the particular proteins which are responsible for a specific attribute of the Ca2+ signal has been difficult to establish. Additionally, whether expression of multiple protein family members represents redundancy or a means of permitting subtype specific regulation is not clearly established. Recently, studies utilizing transgenic gene targeting of particular signaling proteins have begun to address these gaps in our knowledge. In this review, selected studies will be highlighted which have provided insight into the particular proteins contributing to Ca2+ homeostasis in pancreatic acinar cells in both physiological and pathological conditions.</p><!><p>In mice at embryonic day 9.5, the pancreas begins to develop from buds that form as the endodermal gut tube envaginates into the overlying mesenchyme (Gittes, 2009). The bud elongates and undergoes extensive branching morphogenesis in response to complex signals to the epithelial cells from the surrounding mesenchyme and matrix. At around E14 cellular differentiation and lineage selection is initiated. A major amplification in the numbers of both endocrine cells expressing insulin and concurrent activation of acinar cell specific gene programs occurs. The latter event results in cells with substantial rough endoplasmic reticulum and zymogen granules. An increase in the zymogen content and the size of the granules in the acinar cells occurs until birth.</p><!><p>All three members of the InsP3R gene family are expressed in acinar cells with largely the same sub-cellular distribution, albeit with differing relative abundances (Wojcikiewicz, 1995; Lee et al., 1997b; Yule et al., 1997). The majority, ~90% of the InsP3R complement comprises roughly equal numbers of InsP3R-2 and InsP3R-3 with the remainder InsP3R-1 (Wojcikiewicz, 1995). This raises the question as to whether this is reflective of redundancy or do particular sub-types make specific contribution to the signals? Transgenic knock-out of individual or a combination of InsP3R genes has provided significant insight into these issues. Futatsugi and colleagues (Futatsugi et al., 2005) reported that targeted ablation of either the type-2 or type-3 InsP3R individually, had no significant effects on muscarinic- receptor stimulated digestive enzyme secretion and these animals had no overt phenotype. Consistent with these observations, the peak secretatagogue-stimulated increases in [Ca2+]i were not altered (Futatsugi et al., 2005) by the removal of InsP3R-3 and only modestly impacted by the loss of the InsP3R-2 (Futatsugi et al., 2005; Park et al., 2008) In InsP3R-2 null acini, marked changes were only seen at low secretagogue or InsP3 concentrations (Park et al., 2008). In both cases the spatial aspects of the signals were largely unaffected. Taken together these data indicate that the complement of InsP3R-2 or InsP3R-3 in isolation, or perhaps in combination with InsP3R-1, is sufficient to maintain signaling and preservation of stimulated exocytosis.</p><p>Analysis of the compound InsP3R-2/InsP3R-3 null animal revealed a much more striking phenotype resulting from the widespread general disruption of exocrine function (Futatsugi et al., 2005). Although animals were born normally, they failed to survive past weaning, largely due to a failure to ingest and subsequently assimilate food. Even when fed wet mash food to overcome the salivary deficit, the animals failed to thrive as a result of diminished pancreatic secretory function. The immediate cause being, somewhat surprisingly, the complete absence of any measurable secretagogue-induced Ca2+ signal-even at supramaximal concentrations of agonist (Futatsugi et al., 2005). The conclusion from these data is that while the residual expression of InsP3R-1 is not sufficient to mount a Ca2+ signal, either the InsP3R-2 or InsP3R-3 in isolation will suffice.</p><p>InsP3R are subject to diverse regulatory input which is thought to markedly influence the properties of the Ca2+ signal (Patel et al., 1999). As a result of considerable diversity in the primary sequence of the family members, modulation often occurs in a subtype specific manner. Given the previous data, the important modes of regulation of Ca2+ release in acinar cells would be predicted to occur predominately through either modulation of InsP3R-2 and InsP3R-3. It is however difficult to predict if one particular receptor's properties would dominate over the other. Ca2+ release via InsP3R is markedly influenced by the levels of cellular ATP, potentially linking the extent of Ca2+ release to the metabolic status of the cell. In acini, ATP (~Kd 40 µM) markedly enhances Ca2+ release at low levels of stimulation but appears not to influence release at high InsP3 levels (Park et al., 2008). In contrast in InsP3R-2 knock out animals, presumably dominated by InsP3R-3, ATP was shown to modulate release at all InsP3 levels, however the Kd for this effect was 10 fold higher (~450 µM). Interestingly, the properties of the wild–type animal are essentially identical to those shown for the InsP3R-2 in isolation, while the InsP3R-2 KO mirror those of the InsP3R-3 (Betzenhauser et al., 2008). These data indicate that while for Ca2+ release per se, InsP3R-2 and InsP3R-3 are interchangeable, in terms of the fine regulation of Ca2+ release, the individual InsP3R are not redundant. Further, when expressed together the properties of InsP3R-2 dominate over InsP3R-3. Moreover, the high ATP sensitivity of InsP3R-2 would be consistent with acinar cells being relatively resistant to the deleterious effects of ATP depletion.</p><!><p>Secretagogue stimulated Ca2+ influx from the extracellular space is absolutely required for sustained exocytosis (Tsunoda et al., 1990). In acinar cells, like other exocrine cells, voltage-operated Ca2+ channels, common to nerves and muscles, are not expressed. Instead, following the InsP3-mediated depletion of Ca2+ pools, so called "store-operated channels" (SOCs) are activated (Kim et al., 2009). In addition, at lower concentrations of secretagogues, a pathway dependent on the arachidonic- acid activation of a Ca2+ selective channel is engaged (Mignen et al., 2005). In other cells, recent studies have indicated that the channels constituting the depletion-operated pathway are from either the Orai or TRPC family (Feske et al., 2006; Liu et al., 2007). In both cases, an additional protein, STIM-1, functions as the luminal ER Ca2+ sensor responsible for relaying the state of store depletion to the actual channel protein (Roos et al., 2005). Pancreatic acinar cells express TRPC1, 3 and 6. Using a TRPC3 null animal, Kim and colleagues demonstrated that the magnitude of Ca2+ influx initiated by high-concentrations of agonist was significantly reduced. In addition, the frequency of Ca2+ oscillations stimulated by physiological concentrations of secretagogues was also markedly reduced. These data suggest that TRPC3 is a constituent of the native SOC in pancreatic acinar cells (Kim et al., 2009). Pancreas also express Orai 1 (Gwack et al., 2007) and this is consistent with the presence of an additional SOC, Calcium-Release Activated Current (ICRAC) and the Arachidonate-Activated Current (IARC) both of which have Orai-1 as a component of their native channel (Feske et al., 2006; Mignen et al., 2008).</p><!><p>To safeguard the organ from auto-digestion, digestive enzymes are largely secreted as inactive precursors termed zymogens and under physiological conditions only become activated when they reach the duodenum. The inappropriate activation of zymogens occurs as an early event in the pathogenesis of acute pancreatitis- a serious and often life threatening malady which can result in a wide-spread immune response and severe necrosis of the pancreas (Pandol et al., 2007). Disruption of physiological Ca2+ signaling and premature, intracellular activation of proteases has been strongly implicated in the etiology of acute experimental pancreatitis in animal models (Petersen & Sutton, 2006). Aberrant Ca2+ release and Ca2+ influx have both been reported to contribute to this phenomenon. For example, exposure of isolated acini to pancreatic toxins such as bile acids or ethanol metabolites results in Ca2+ release from InsP3-sensitive stores (Gerasimenko et al., 2009). In turn, this results in intracellular trypsin activation and the cellular hallmarks of pancreatitis. It has been reported that this occurs at least in part through activation of InsP3R directly. Consistent with the dominant expression of InsP3R-2 and InsP3R-3, toxin-induced Ca2+ release and trypsin activation are markedly attenuated in either the InsP3R-2 or InsP3R-3 null animal, unaffected in the InsP3R-1 animal, and absent in the compound double knock-out (Gerasimenko et al., 2009). A question arises as to why Ca2+ release via InsP3R stimulated by toxins is detrimental to the acinar cell, while physiologically InsP3-induced release is essential for normal exocytosis? The answer may be related to the identity of the store as both bile acids and ethanol metabolites appear to activate insP3R in a compartment defined as an "acidic store" distinct from the ER (Gerasimenko et al., 2009). Prolonged Ca2+ entry through SOCs has also been demonstrated to contribute to experimental pancreatitis. Consistent with the role of TRPC-3 as a component of native SOC, the severity of experimental pancreatitis is markedly reduced in TRPC-3 null animals (Kim et al., 2009).</p><!><p>Analysis of Ca2+ signaling in transgenic animal models has provided a wealth of information regarding stimulus-secretion coupling in pancreatic acini in both physiological and pathological situations. The availability of additional animal models, particularly pancreas-specific knock-downs of other signaling proteins will similarly be insightful in the future. Obvious potential targets for investigation include other Ca2+ release channels such as ryanodine receptors and Two-Pore Channels (TPC); Ca2+ influx channels including the Orai channels and clearance proteins such as Ca2+ ATPases and mitochondrial Ca2+ transport proteins.</p><!><p>This is a PDF file of an unedited manuscript that has been accepted for publication. As a service to our customers we are providing this early version of the manuscript. The manuscript will undergo copyediting, typesetting, and review of the resulting proof before it is published in its final citable form. Please note that during the production process errors may be discovered which could affect the content, and all legal disclaimers that apply to the journal pertain.</p>

PubMed Author Manuscript

Stereospecific Formation of E- and Z- Disubstituted Double Bonds by Dehydratase Domains from Modules 1 and 2 of the Fostriecin Polyketide Synthase

The dehydratase domain FosDH1 from module 1 of the fostriecin polyketide synthase (PKS) catalyzed the stereospecific interconversion of (3R)-3-hydroxybutyryl-FosACP1 (5) and (E)-2-butenoyl-FosACP1 (11), as established by a combination of direct LC-MS/MS and chiral GC-MS. FosDH1 did not act on either (3S)-3-hydroxybutyryl-FosACP1 (6) or (Z)-2-butenoyl-FosACP1 (12). FosKR2, the ketoreductase from module 2 of the fostriecin PKS that normally provides the natural substrate for FosDH2, was shown to catalyze the NADPH-dependent stereospecific reduction of 3-ketobutyryl-FosACP2 (23) to (3S)-3-hydroxybutyryl-FosACP2 (8). Consistent with this finding, FosDH2 catalyzed the interconversion of the corresponding triketide substrates (3R,4E)-3-hydroxy-4-hexenoyl-FosACP2 (18) and (2Z,4E)-2,4-hexadienoyl-FosACP2 (21). FosDH2 also catalyzed the stereospecific hydration of (Z)-2-butenoyl-FosACP2 (14) to (3S)-3-hydroxybutyryl-FosACP2 (8). Although incubation of FosDH2 with (3S)-3-hydroxybutyryl-FosACP2 (8) did not result in detectable accumulation of (Z)-2-butenoyl-FosACP2 (14), FosDH2 catalyzed the slow exchange of the 3-hydroxy group of 8 with [18O]-water. FosDH2 unexpectedly could also support the stereospecific interconversion of (3R)-3-hydroxybutyryl-FosACP2 (7) and (E)-2-butenoyl-FosACP2 (13).

stereospecific_formation_of_e-_and_z-_disubstituted_double_bonds_by_dehydratase_domains_from_modules

INTRODUCTION<!>Recombinant Fostriecin PKS Domains<!>Chemoenzymatic synthesis of C4 and C6 acyl-FosACP substrates<!>Fostriecin Synthase Module 1. Substrate and Product Specificity of FosDH1<!>Fostriecin Synthase Module 2. Substrate and Product Specificity of FosKR2<!>FosDH2<!>DISCUSSION<!>Materials<!>Methods<!>Expression and Purification of FosDH1 and FosDH2<!>Expression and Purification of His6-tag-NusA-FosACP1 and His6-tag-NusA-FosACP2 and Preparation of apo-FosACP1 and apo-FosACP2<!>Expression and Purification of FosKR2<!>(2Z,4E)-2,4-Hexadienoic Acid (18)<!>(3R,4E)-3-Hydroxy-4-hexenoic Acid (22) and (3S,4E)-3-Hydroxy-4-hexenoic Acid<!>(3R)- and (3S)-Hydroxybutyryl-CoA<!>(E)- and (Z)-2-Butenoyl-CoA ((E)- and (Z)-Crotonyl-CoA<!>L-(3R,4E)-3-Hydroxy-4-hexenoyl-CoA<!>(2Z,4E)- and (2E,4E)-2,4-Hexadienoyl-CoA<!>Monitoring of FosDH1 and FosDH2 Activity Using Acyl-CoA Substrates<!>Small scale acylation reactions for LC-MS analysis<!>FosDH1 and FosDH2-Catalyzed Dehydration and Hydration of C4 Acyl-FosACP Substrates. LC-ESI(+)-MS and LC-ESI(+)-MS/MS Analysis<!>Larger scale acylation reaction for GC-MS analysis of FosDH-catalyzed dehydration or dehydration<!>FosDH1-Catalyzed Dehydration and Hydration of C4 Acyl-FosACP Substrates. Chiral GC-MS Analysis<!>FosKR2-Catalyzed Reduction of 3-Ketobutyryl-FosACP2 (23)<!>FosDH2-Catalyzed Dehydration and Hydration of C6 Acyl-FosACP2 Substrates, (3R,4E)-18 and (2Z,4E)-21<!>FosDH2-Catalyzed Dehydration and Hydration of C4 Acyl-FosACP Substrates. Chiral GC-MS Analysis<!>FosDH2-Catalyzed Isotope Exchange of 3-Hydroxybutyryl-FosACP2

<p>Fostriecin (1), a polyketide phosphate monoester isolated from Streptomyces pulveraceus,1 belongs to a class of selective protein phosphatase inhibitors showing anti-metastatic and antitumor activity that includes the closely related natural products cytostatin (2) and phoslactomycin B (3) (Figure 1).2 Fostriecin also exhibits antimycotic activity.3 In the biosynthesis of fostriecin (1), the parent nonaketide 4 is assembled by a polyketide synthase (PKS) consisting of 8 modules, with a loading tri-domain at the N-terminus of module 1 responsible for priming by the acetyl starter unit, while a thioesterase (TE) domain at the C-terminus of module 8 mediates polyketide chain release and lactonization (Figure 2).4a A series of tailoring reactions, including elimination of malonate to introduce the 2,3-double bond of the dihydropyrone moiety, then converts 4 to the mature product fostriecin (1).4</p><p>The acyclic chain of fostriecin harbors four disubstituted double bonds. The Δ16,17 and Δ6,7 E (trans) double bonds of 1 are predicted to be generated by the dehydratase domains of fostriecin modules 1 and 6, FosDH1 and FosDH6 respectively,4c while the Δ14,15 and Δ12,13 Z (cis) double bonds are attributed to the action of the corresponding FosDH2 and FosDH3 domains from fostriecin modules 2 and 3.4a Although the majority of the thousands of known polyketides harbor E double bonds,5 isomeric Z double bonds, while considerably less common, are nonetheless well represented, being found in not only 1–3, but also in the microtubule stabilizer epothilone,6 the microtubule polymerization inhibitor curacin,7 the anti-angiogenic agent borrelidin,8 and the antitubercular rifamycins.9 Surprisingly, all reported attempts to demonstrate in vitro formation of the relevant Z-double bonds by PKS-derived DH domains have been unsuccessful, resulting only in the generation of products containing the isomeric E double bonds.8b,c,9d, While PlmKR1, the ketoreductase from module 1 of the phoslactomycin PKS, has been shown to generate a (3S)-hydroxyacyl thioester that likely serves as the precursor of the Z-3-cyclohexylpropenoate produced by phoslactomycin module 1, direct biochemical evidence for the function of the paired PlmDH1 domain is still lacking.10 We now report the elucidation of the biochemical function of both FosDH1 and FosDH2, thereby confirming the predicted role of each dehydratase domain in the stereospecific formation of their respective E- and Z-disubstituted enoyl-ACP products.</p><!><p>FosDH1 and FosDH2 as well as FosKR2 from module 2 of the fostriecin PKS were each expressed in Escherichia coli as N-terminal His6-tagged proteins using codon-optimized synthetic genes, based on well-precedented PKS domain boundaries (Figures S1–S4). The purity and molecular mass of each recombinant protein was verified by SDS-PAGE and LC-ESI(+)-MS (Figures S6–S8). Although direct expression in E. coli of synthetic genes for FosACP1 and FosACP2 gave predominantly insoluble protein inclusion bodies, the corresponding NusA-FosACP1 and NusA-FosACP2 fusion proteins were obtained in soluble form (Figures S5, S9, and S10).9d Cleavage of the N-terminal NusA with HRV 3C protease gave apo-FosACP1 and apo-FosACP2, which were each confirmed to have the expected mass by LC-ESI(+)-MS (Figures S9 and S10).</p><!><p>(3R)- and (3S)-3-Hydroxybutyryl-FosACP1 (5 and 6) as well as (3R)- and (3S)-3-hydroxybutyryl-FosACP2 (7 and 8) were chemo-enzymatically prepared from (3R)- and (3S)-hydroxybutyric acid (9 and 10), respectively, via the corresponding -SCoA esters, using apo-FosACP1 or apo-FosACP2 and the surfactin phosphopantetheinyl transferase Sfp (Scheme 1). Similar procedures were also used to prepare the individual unsaturated (E)- and (Z)-2-butenoyl-FosACP1 (11 and 12) as well as (E)- and (Z)-2-butenoyl-FosACP2 (13 and 14) from (E)- and (Z)-2-butenoic acid (15 and 16). For the synthesis of ACP-bound triketides, synthetic (3R,4E)-3-hydroxy-4-hexenoic acid (17, 3R/3S 90:10) was chemo-enzymatically converted via its -SCoA thioester to the corresponding (3R,4E)-3-hydroxy-4-hexenoyl-FosACP2 (18, 80% d.e.) (Scheme 2).11,12 In like manner, (2Z,4E)-2,4-hexadienoic acid (19), prepared as previously described by ring-closing metathesis–base-induced ring opening,13 and (2E,4E)-2,4-hexadienoic acid (20) were each converted via their -SCoA esters to the corresponding (2Z,4E)-2,4-hexadienoyl-FosACP2 (21) and (2E,4E)-2,4-hexadienoyl-FosACP2 (22).</p><!><p>Module 1 of the fostriecin PKS is predicted to produce (E)-2-butenoyl-FosACP1 (11), integrated into the parent module, as inferred from the E (trans) geometry of the derived Δ16,17 double bond of fostriecin (1) (Figure 1). The substrate for FosDH1 is expected to be (3R)-3-hydroxybutyryl-ACP1, as suggested by the presence of the characteristic highly conserved Leu-Asp-Asp motif in the paired ketoreductase domain FosKR1 (Figure S1).14 Fully consistent with these predictions, incubation of FosDH1 with (3R)-3-hydroxybutyryl-FosACP1 (5) resulted in dehydration to give exclusively (E)-2-butenoyl-FosACP1 (11) (Scheme 3). Thus treatment of the incubation mixture with PICS TE, the thioesterase from the picromycin PKS, so as to hydrolyze the ACP-bound diketide substrate and products, followed by treatment with N,Obis( trimethylsilyl)trifluoroacetamide (N,O-bis(TMS)-TFA) gave the corresponding (E)-2-butenoyl-TMS derivative whose geometry was confirmed by GC-MS analysis and direct comparison with authentic standards derived from both (E)-15 and (Z)-16. No dehydration product from 5 was detected in the absence of added FosDH1 (Figure S11).</p><p>In a complementary set of incubations, FosDH1 catalyzed the reverse hydration reaction by stereospecifically converting (E)-2-butenoyl-FosACP1 (11) to (3R)-3-hydroxybutyryl-FosACP1 (5) (Scheme 3). LC-ESI(+)-MS analysis of the reaction mixture showed the expected addition of water (M+H2O) for the parent ion while LC-ESI(+)-MS/MS confirmed the predicted increase of M+18 in the mass of the derived pantetheinyl ejection fragments (329 to 347 Da) (Figures S12 and S13).15 Chiral GC-MS analysis after hydrolysis of the incubation products with PICS TE and derivatization by (N,O-bis(TMS)-TFA established the exclusive formation of bis(TMS)-(3R)-3-hydroxybutyrate (bis(TMS)-(3R)-9) (rt 9.61 min, identical with an authentic reference sample) (Figure S14). Hydration of 11 was not detected in the absence of FosDH1. Neither the stereoisomeric (3S)-3-hydroxybutyryl-FosACP1 (6) nor (Z)-2-butenoyl-FosACP1 (12) underwent any detectable reaction in the presence of FosDH1.</p><!><p>Module 2 of the fostriecin PKS is predicted to produce (2Z,4E)-2,4-hexadienoyl-FosACP2 (21), integrated into the parent module, as inferred from the Z (cis) geometry of the derived Δ14,15 double bond of fostriecin (1) (Figure 1). Within fostriecin synthase module 2, FosKR2 is responsible for generating the bound 3-hydroxyacyl-FosACP2 triketide that serves as the native substrate of the paired FosDH2 domain in intact fostriecin module 2. The absence of the characteristic Leu-Asp-Asp triad typical of 3R-hydroxy (D)-specific ketoreductases suggested that FosKR2 should generate the (L)-hydroxy product typical of an A-type KR domain.14 To establish the stereochemistry of this reduction, 3-ketobutyryl-FosACP2 (23), chemoenzymatically prepared by treatment of acetoacetyl-SCoA with Sfp and apo-FosACP2, was reduced with FosKR2 in the presence of NADPH (Scheme 4 and Figure S15). The exclusive product was (3S)-3-hydroxybutyryl-FosACP2 (8), as established by chiral GC-MS analysis of the corresponding bis(TMS)-derivative.</p><!><p>Incubation of FosDH2 with the chemoenzymatically prepared form of the natural substrate stereoisomer, (3R,4E)-3-hydroxy-4-hexenoyl-FosACP2 (18) (~80% e.e.), resulted in stereospecific dehydration to give (2Z,4E)-2,4-hexadienoyl-FosACP2 (21), as determined by GC-MS analysis of the derived methyl ester (2Z,4E)-21-Me following PICS TE-catalyzed hydrolysis and treatment with TMS-CHN2, (Scheme 5, Figures S16 and S17).16 Complementary incubation of (2Z,4E)-2,4-hexadienoyl-FosACP2 (21) with FosDH2 resulted in the reverse hydration reaction to give (3R,4E)-3-hydroxy-4-hexenoyl-FosACP2 (18) as the exclusive product of DH-catalyzed hydration (Scheme 5). This result was established by chiral GC-MS analysis of the derived methyl ester, including direct comparison with reference standards of synthetic methyl (3R,4E)-17-Me and the enantiomeric methyl (3S,4E)-3-hydroxy-4-hexenoate (Figures S18 and S19). Hydration of 21 did not occur in the absence of added FosDH2, although (2Z,4E)-21 did undergo 10–15% buffer-catalyzed isomerization to (2E,4E)-22 after 2 h incubation under the same conditions.</p><p>We also carried out the analogous series of reactions with FosDH2 and the corresponding FosACP2-bound C4 substrate analogs. Thus (Z)-2-butenoyl-FosACP2 (14) underwent time-dependent, stereospecific hydration to give (3S)-3-hydroxybutyryl-FosACP2 (8) when incubated with FosDH2 (Scheme 6, Figures 3 and S20–S22). On the other hand, when FosDH2 was incubated with 8 (Figures S23 and S24) the expected dehydration product, (Z)-14, could not be directly detected, presumably due to the thermodynamically unfavorable Keq ~10−3 for dehydration to the (Z)-enoyl-ACP product.17 Evidence for the transient formation of 14 was obtained, however, by the observation that incubation of 8 with FosDH2 in [18O]-H2O for 90 min resulted in ~10% net exchange of the 3-hydroxyl oxygen of 8, as revealed by the enzyme- and time-dependent increase in the relative intensity of the [M+2] peak of the derived pantetheinate ejection fragment (349.23 Da) observed by LC-MS/MS analysis of recovered 8 (Figure S26). Isotope exchange was not detectable in the absence of FosDH2. FosDH2 was found to be unexpectedly permissive in also being able to interconvert the unnatural pair of stereoisomeric C4 analogs, (3R)-3-hydroxybutyryl-FosACP2 (7) and (E)-2-butenoyl-FosACP2 (13) (Figures S22-1 and S27–S30). Consistent with these observations, incubation of (3R)-7 with FosDH2 in [18O]-H2O resulted in essentially complete isotope exchange of the 3-hydroxyl oxygen of recovered 7 within 30 min (Figure S31).</p><!><p>Dehydratases of both Type I and Type II fatty acid synthases (FASs) catalyze the exclusive syn dehydration of a (3R)-3-hydroxyacyl-ACP thioester to the corresponding (E)-enoyl-ACP.18 The structure of the E. coli dehydratase (FabA) displays a characteristic hotdog fold that has been observed in all other DH structures from both FAS and PKS systems.7a,19,20 The active site of each DH harbors a universally conserved pair of His and Asp residues. Schwab, in a critical review of research on FabA and the closely related dehydratase-isomerase FabZ, has discussed a one-base, two-step mechanistic model in which the active site imidazole residue first catalyzes the stereospecific removal of 2-Hsi of the 3-hydroxyacyl-ACP substrate following which the transiently-generated imidazolium species donates its proton to the 3-hydroxyl group to promote C–O bond cleavage. 18a,21,22 Although the distinct enoyl-CoA hydratase of fatty acid oxidation differs significantly from DH enzymes in both protein fold and the presence of two essential active site Glu residues in place of the His-Asp dyad of DH domains, it catalyzes an analogous net syn hydration of (E)-2-enoyl-CoA thioesters to yield the corresponding (3S)-3-hydroxyacyl-CoA products.17a Both protons and the oxygen of the nucleophilic water are incorporated into the product, consistent with a single-base mechanism for enoyl-CoA hydratase in which one Glu residue acts sequentially, first as base and then as active site acid, while the second Glu side chain positions the active site water by an essential H-bond.23 Interestingly, while enoyl-CoA hydratase can also add water to the isomeric (Z)-2-butenoyl-CoA to give the (3R)-3-hydroxybutyryl-CoA product, the reverse dehydration could not be observed, an observation that was attributed to the calculated unfavorable Keq <10−3 for formation of the (Z)-isomer.17a</p><p>EryDH4 (from module 4 of the erythromycin PKS) and NanDH2 (from module 2 of the nanchangmycin PKS) both catalyze the syn elimination of water from a (2R,3R)-2-methyl-3-hydroxyacyl-ACP substrate to the corresponding (E)-2-enoyl-ACP product,24 while RifDH10 (from module 10 of the rifamycin PKS) catalyzes the syn dehydration of the diastereomeric (2S,3S)-2-methyl-3-hydroxyacyl-RifACP10 substrate to the (E)-2-enoyl-ACP product.9d The structures of both EryDH4 and RifDH10 display the characteristic DH double hotdog fold as well as the conserved active His and Asp residues.9d,20 While one or two of the four DH domains from the curacin PKS are thought to be responsible for formation of unsaturated intermediates possessing (Z) double bonds, all four proteins exhibit the common double hotdog fold and high levels of mutual structural homology, while the actual enzymatic formation of (Z)-enoyl-ACP products has not yet been reported.7 FosDH2 shows 71.0% mutual sequence identity (88.8% similarity) over 276 aa to the closely related PlmDH1 from module 1 of the phoslactomycin PKS, which has also been implicated in the formation of a cis double bond (Figure S32). Interestingly, FosDH1 and FosDH2 themselves show a more modest 40.8% mutual sequence identity (63.4% similarity), while retaining each of the key conserved amino acid motifs typified by the structurally characterized dehydratases EryDH4 and RifDH10.</p><p>In spite of the frequent occurrence of Z double bonds in complex polyketides, the experimental demonstration that FosDH2 can catalyze the interconversion of the L-3-hydroxyacyl-ACP and (Z)-2-enoylacyl-ACP thioesters is the first documented in vitro confirmation of this nominally straightforward reaction directly catalyzed by a PKS DH domain. Earlier unsuccessful attempts to observe DH-catalyzed formation of conjugated Z-enoyl thioester double bonds have been summarized above. One recent report has described the highly unusual dehydration of (3S,4S)-3,4-dihydroxypentanoyl-N-acetylcysteamine thioester to the (Z)-2-enoyl-SNAC catalyzed by a modular PKS domain formally classified as a TE on the basis of phylogenetic sequence comparisons.25 Curiously, the TE-catalyzed dehydration reaction could not be detected with the corresponding ACP thioester. The well-studied FabZ-catalyzed formation of the non-conjugated Z-3,4-decenoyl-ACP from (3R)-3-hydroxydecanoyl-ACP involves the allylic isomerization of the initially-formed (E)-2-decenoyl-ACP.18,21 Although there is molecular genetic evidence that the characteristic Z-double bond of epothilone is introduced by the DH domain from the proximal downstream module acting in trans, this transformation has not been verified at the enzyme level.26</p><p>The finding that FosKR2 stereospecifically reduces 3-ketobutyryl-FosACP2 (23) to L-(3S)-3-hydroxybutyryl-FosACP2 (8) establishes that FosDH2 never encounters the unnatural epimer, D-(3S,4E)-3-hydroxy-4-hexenoyl-FosACP2, in its native modular context. Indeed, we have now shown that FosDH2 stereospecifically interconverts L-(3R,4E)-3-hydroxy-4-hexenoyl-FosACP2 (18) and (2Z,4E)-hexadienoyl-FosACP2 (21), the predicted product of fostriecin synthase module 2. Interestingly, under the incubation conditions tested, dehydration of the C4-analog L-(3S)-8 to (Z)-14 could be detected only indirectly by FosDH2-catalyzed isotopic exchange of the 3-hydroxyl group with [18O]-water. This apparent discrepancy between the results with C6 and the C4 substrates is likely the consequence of the use of a 1.5:1 stoichiometric excess of FosDH2 to triketide alcohol 18, compared to the 5-fold lower 0.3–0.5:1 ratio that was used for the incubation of FosDH2 with the shorter chain diketide analog (3S)-8. Thus although FosDH2 cannot alter the Keq ~10−3 for free substrates and products, enzymes can differentially bind substrates and products so as to shift the ratio of bound substrates species closer to 1:1.27 For the intact fostriecin synthase module 2, the unfavorable equilibrium of the FosDH2-catalyzed dehydration reaction is overcome by metabolic coupling to the thermodynamically favorable, metabolically irreversible chain elongation reaction catalyzed by the KS domain of the fostriecin synthase module 3.</p><p>Finally, the demonstration that FosDH2 catalyzes the dehydration of L-(3R,4E)-18 to (2Z,4E)-21 does not validate the common assumption that there is a requisite correlation between an L-hydroxy configuration of the substrate and the cis geometry of the olefinic product of DH-catalyzed dehydration.14a There are already sufficient exceptions7b,8b,c,9d,23,25 to this superficially appealing generalization to establish that it does not have reliable predictive value.</p><!><p>IPTG and kanamycin were purchased from Thermo Fisher Scientific. (E)-2-Butenoyl-CoA ((E)-Crotonyl-CoA), 3-ketobutyryl-CoA (acetoacetyl-CoA), N,O-BSTFA, TCEP, (3R)- and (3S)-3-hydroxybutyric acids, 2-butenoic acid (crotonic acid), and (2E,4E)-2,4-hexadienoic acid were purchased from Sigma-Aldrich and utilized without further purification. (Z)-2-Butenoic acid was prepared as previously described.17a,28 [18O]-H2O was purchased from Cambridge Isotope Laboratories. HRV 3C protease was obtained from Pierce. Pseudomonas fluorescens Amano lipase P was from Sigma-Aldrich. Sfp and PICS TE were each expressed and purified as previously described.9d,24,29 DNA primers were synthesized by Integrated DNA Technologies. Competent E. coli DH5α, DH10β and BL21(DE3) cloning and expression strains were purchased from New England Biolabs (NEB). Restriction enzymes, T4 DNA ligase and Phusion High-Fidelity PCR Master Mix with HF Buffer were purchased from NEB. Amicon Ultra Centrifugal Filter Units (Amicon Ultra-15 and Amicon Ultra-0.5, 3000, 10000 and 30000 MWCO) were obtained from Millipore. The 20 mL HisPrep FF 16/10 column, prepacked with precharged Ni Sepharose 6 Fast Flow, was purchased from GE Healthcare.</p><!><p>General methods were as previously described. 30 Growth media and conditions used for E. coli strains and standard methods for handling E. coli in vivo and in vitro were those described previously, unless otherwise noted.31 All DNA manipulations were performed following standard procedures.31 Plasmid DNA was purified using a Thermo Scientific GeneJET Plasmid mini-prep kit. DNA sequencing was carried out by Genewiz, South Plainfield, NJ. Synthetic genes, optimized for expression in E. coli, were prepared by DNA 2.0, Newark, California. All proteins were handled at 4 °C unless otherwise stated. Protein concentrations were determined according to the method of Bradford, using a Tecan Infinite M200 Microplate Reader with bovine serum albumin as standard. 32 Protein purity and size were estimated using SDS-PAGE, visualized using Coomassie Blue stain, and analyzed with a Bio-Rad ChemiDoc MP System. Gas chromatography-mass spectrometry (GC–MS) analyses were performed using an Agilent 5977A Series GC/MSD instrument (70 eV, electron impact) with a 3 min solvent delay. Protein accurate molecular weight was determined on an Agilent 6530 Accurate-Mass Q-TOF LC-MS. A Thermo LXQ equipped with Surveyor HPLC system and a Phenomenex Jupiter C4 column (150 mm × 2 mm, 5.0 μm) was utilized for analysis of acyl-ACP compounds. HPLC-ESI-MS/MS analysis was carried out in positive ion mode for analysis of pantetheinate ejection fragments, as previously described.15,33</p><!><p>A synthetic gene for FosDH1, optimized for expression in E. coli and corresponding to the region from A1992 to G2294 of Fos Module 1 (Figures S1 and S2), was sub-cloned into pET28a between the NdeI and XhoI restriction sites. The FosDH2 expression plasmid corresponding to the region from A947 to A1232 of Fos Module 2 (Figure S2 and S3) was generated by subcloning the synthetic gene optimized for expression in E. coli into pET28a between the NdeI and XhoI restriction sites. Single colonies of E. coli BL21(DE3) cells that had been transformed with the individual FosDH1 or FosDH2 expression vectors were inoculated into 10 mL LB media containing 50 mg/L kanamycin and incubated overnight at 37 °C. This starter culture was then inoculated into 500 mL Super Broth (SB) containing 50 mg/L kanamycin. The culture was then grown at 37 °C at 225 rpm until OD600 = 0.5. At this point, the culture was cooled to 18 °C and then induced with 0.2 mM IPTG. Cells were harvested after 20 h by centrifugation at 4000g for 40 min. The purification was then carried out at 4 °C unless mentioned otherwise. Harvested cells were re-suspended in 50 mL start buffer (50 mM sodium phosphate, 500 mM NaCl, pH 7.5). The cells were lysed by sonication and cell debris was removed by centrifugation at 20000g for 50 min. The supernatant was loaded onto a HisPrep FF 16/10 (GE Healthcare Life Science) column pre-equilibrated with start buffer. The column was washed with 150 mL of wash buffer (50 mM sodium phosphate, 500 mM NaCl, 10 mM imidazole, pH 7.5) to elute contaminating proteins. FosDH1 or FosDH2 were then eluted with a linear gradient from 10 mM – 500 mM imidazole in the same buffer. The fractions containing FosDH1 or FosDH2 were pooled, buffer-exchanged, and concentrated to final buffer (50 mM sodium phosphate, 250 mM NaCl, 10% glycerol, pH 7.5) using an Amicon Ultra-15 (30000 MWCO) centrifugal filter. The purity and MW of FosDH1 and FosDH2 were analyzed by SDS-PAGE and LC-ESI(+)-MS (Figures S6 and S7) Aliquots of the purified proteins were stored at −80 °C.</p><!><p>Synthetic genes for FosACP1 (V3130 to T3220 of Fos Module 1 (Figures S1 and S2) and FosACP2 (region from R1696 to T1820 of Fos Module 2 (Figures S3 and S4) were each subcloned into pET28a between the NdeI and XhoI restriction sites. Since the resulting recombinant ACP proteins were produced predominantly as insoluble inclusion bodies when expressed in E. coli BL21(DE3), the corresponding NusA-FosACP1 and NusA-FosACP2 fusion proteins were then generated. In brief, the DNA regions encoding FosACP1 and FosACP2 were each amplified by PCR from the above-described pET28a constructs using the primers FosACP1/2-HRV-FP and pET-28a-RP which is complementary to the region just 3′ of the native XhoI site (Figure S5). The resultant amplified DNA harboring FosACP1 or FosACP2 was digested with NheI and XhoI before ligation into the corresponding sites of an NheI/XhoI-digested vector, immediately downstream of DNA encoding His6-NusA-HRV3C housed in a pET28 vector. The resultant expression plasmids encoded the individual NusA-FosACP1 and NusA-FosACP2 fusion proteins, each carrying an N-terminal His6-tag and an HRV 3C protease site between the N-terminal NusA and either FosACP1 or FosACP2.</p><p>NusA-FosACP1 or NusA-FosACP2 were expressed in E. coli BL21(DE3) as described above for FosDH1 and FosDH2. Cells were harvested after 20 h by centrifugation at 4000g for 40 min. The recovered cells were resuspended in 50 mL start buffer and lysed by passage three times through a French Press at 10000 psi, then centrifuged at 20000g for 50 min. The pellet was discarded and the supernatant was loaded on to a HisPrep FF 16/10 (GE Healthcare Life Science) column preequilibrated with start buffer. The column was washed with 150 mL of wash buffer to elute contaminating proteins. The respective His6-NusA-FosACP1 and His6-NusA-FosACP2 proteins were then eluted with a gradient from 10 mM – 500 mM imidazole in the same buffer. The fractions containing His6-NusA-FosACP1or His6-NusA-FosACP2 were pooled. Protein was concentrated to 2–3 mL using an Amicon Ultra-15 (30000 MWCO) centrifugal filter before further purification on a Hiload 16/600, Superdex 200 pg size exclusion column pre-equilibrated with 50 mM sodium phosphate and 250 mM NaCl, pH 7.5. Fractions containing His6-NusA-FosACP1 or His6-NusA-FosACP2 were pooled and treated with 1 μL of HRV-3C protease (1U/μL) per 200 μg of purified protein. After overnight incubation at 4 °C with continuous shaking, the mixture was loaded on to a HisPrep FF 16/10 column pre-equilibrated with start buffer. Both His6-NusA up to the HRV 3C protease cleavage site, as well as the HRV-3C protease, which also carries a His6-tag, along with any undigested His6-Nus-AFosACP were retained by the column while the cleaved FosACP1 and FosACP2 eluted during the column wash step. The fractions containing cleaved FosACP1 or FosACP2 were pooled and buffer-exchanged with final buffer (50 mM sodium phosphate, 250 mM NaCl, 10% glycerol, pH 7.5) using an Amicon Ultra-15 (30000 MWCO) centrifugal filter. The purity and MW of FosACP1 and FosACP2 were analyzed by SDS-PAGE and LC-ESI(+)-MS (Figures S9 and S10). Aliquots of purified proteins were stored at −80 °C.</p><!><p>A synthetic gene for FosKR2, optimized for expression in E. coli and corresponding to the region from F1239 to L1683 of Fos Module 2 (Figures S3 and S4) was subcloned into pET28a between the NdeI and XhoI restriction sites. FosKR2 was expressed and purified by the same procedures described for FosDH1 and FosDH2 (Figure S8).</p><!><p>(2Z,4E)-2,4-Hexadienoic acid (18) was prepared as previously described. 13 The cis-olefin geometry was verified by 1H NMR and chiral GC-MS by which the (2Z,4E)-2,4-hexadienoic acid (18) was readily separated from commercially available (2E,4E)-17 either as the carboxylic acids or derived methyl esters 17-Me and 18-Me. The carboxylic acid forms were analyzed with an Agilent ChiraSil-Dex capillary GC column, (0.32 mm ID × 25 m length × 0.25 μm film) using a temperature program (GC Method A) with a 1 min hold at 50 °C, followed by a 7.5 °C/min increment to 200 °C. Separation of the methyl esters 17-Me and 18-Me forms used a temperature program (GC Method B) with a 1 min hold at 50 °C, followed by a 1.00 °C/min increase to 90 °C, a 2 min hold at 90 °C, and then a further increment of 20.00 °C/min to 200 °C (Figure S16). The 1H NMR data for 18 matched the literature values:13 1H NMR (400 MHz, CDCl3): δ 7.35 (ddd, 1 H), 6.63 (dd, 1 H), 6.04 (dq, 1 H), 5.57 (d, 1 H), 1.90 (d, 3 H).</p><!><p>(±)-Ethyl (3RS,4E)-3-Hydroxy-4-hexenoic acid was synthesized as previously described.34 The enantiomers were kinetically resolved using Amano lipase P (Scheme S1).12b,c In brief, (±)-ethyl (4E,3RS)-3-hydroxy-4-hexenoate (1.58 g, 10 mmol) was dissolved in 20 mL of hexane. Amano lipase P (1 g) was added with vinyl acetate (20 mmol) and the reaction mixture was stirred at room temperature for 24 h with monitoring by GC-MS. The lipase was removed by filtration and the solvent was evaporated. Ethyl (3S,4E)-3-hydroxy-4-hexenoate and ethyl (3R,4E)-3-acetoxy-4-hexenoate were separated by SiO2 flash column chromatography. The ethyl-3-hydroxy-4-hexenoate was eluted with 10% ethylacetate/90% hexane and then (3R,4E)-3-acetoxy-4-hexenoate was eluted by 20% ethylacetate/80% hexane. If the resolution was incomplete and the acetate ester contained more than 10% (3S,4E) isomer, a second round of Amano lipase P reaction was performed. Hydrolysis of the recovered ethyl (3S,4E)-3-hydroxy-4-hexenoate with LiOH (1.0 eq) at 0 °C for 1–2 h generated (3S,4E)-3-hydroxy-4-hexenoic acid. Methanolysis of (3R,4E)-3-acetoxy-4-hexenoate (700 mg, 3.5 mmol) was effected by treatment with K2CO3 (966 mg, 2.0 eq) in 12 mL MeOH with vigorous stirring at room temperature for 30 min. The solution was diluted with EtOAc (240 mL) and washed with 0.1 M aq. NaOH (240 mL). The aqueous layer was back-extracted with EtOAc (3 × 240 mL). The combined organic fractions were dried over MgSO4 and concentrated to give methyl (3R,4E)-3-hydroxy-4-hexenoate (22-Me) which was purified by flash column chromatography on silica gel (20% ethylacetate/80% hexane). Hydrolysis with LiOH (1.0 eq) at 0 °C for 1–2 h generated (3R,4E)-3-hydroxy-4-hexenoic acid (22). By chiral GC-MS analysis (Method B) of the methyl esters, the major component (~90%) of the (3R,4E)-22-Me eluted ca. 0.9 min earlier than the (3S,4E)-methyl ester (~10%), corresponding to ~80% ee (Figure S18). Chiral GC-MS analysis also established that the preparation of the enantiomeric methyl (3S,4E)-3-hydroxy-4-hexenoate was obtained in ~90% ee.</p><!><p>Using the previously described method,24a 25 mg (0.24 mmol) of (3R)- or (3S)-3-hydroxybutyric acid in 1 mL THF were treated by dropwise addition of 60 mg (0.36 mmol, 1.5 eq) of 1,1′-carbonyldiimidazole (CDI) in 0.5 mL of THF. After reaction at 0 °C for 60–90 min, a solution of CoASH (20 mg in 1.5–2.0 mL H2O, 0.024 mmol, 0.1 eq) was added dropwise. The reaction was continued at room temperature and under nitrogen for 4 h. Organic solvent was removed by rotary evaporation. The aqueous phase, after extraction with ethyl acetate to remove byproducts, was purified through by HPLC (Agilent) using a Phenomenex Gemini semi-preparative C18 column (150 × 10 mm, 10 μm) equilibrated with 2% CH3CN/H2O. Elution was carried out with a linear gradient from 2% to 90% CH3CN/H2O. Collected peaks were checked for purity by LC-ESI(+)-MS using an Agilent Zorbax Extend C18 column (100 × 2.1 mm, 3.5 μm) with a linear gradient from 2% to 70% CH3CN/H2O. Peaks containing product of the desired mass, LC-ESI(+)-MS [M+H]+ 854, were collected, lyophilized and stored at −80 °C.</p><!><p>Using the previously described method,24a 20 mg (0.23 mmol) of (E)- or (Z)-2-butenoyl-CoA was dissolved in 2 mL anhydrous CH2Cl2 under nitrogen, then treated with 65 μL of triethylamine (47 mg, 0.47 mmol, 2 eq) followed after 10 min by 44 μL of ethylchloroformate (50 mg, 0.46 mmol, 2 eq) . The reaction mixture was stirred at 0 °C for 2 h. After removal of the organic solvent by rotary evaporation, the residue was dissolved in 2 mL THF. Insoluble salts were removed by centrifugation and the mixed anhydride was added to a round bottom flask containing 20 mg of CoASH dissolved in 2 mL of 50 mM aq. NaHCO3 (pH 8.0) and the resultant mixture was stirred for 1–3 h at room temperature under N2. After removal of excess starting materials by extraction with ethyl acetate, the aqueous phase was purified by HPLC (Agilent) using the Phenomenex Gemini semi-preparative C18 column (150 × 10 mm, 10 μm) equilibrated with 2% CH3CN/H2O. Elution was carried out with a linear gradient from 2% to 10% CH3CN/H2O. Collected peaks were checked for purity by LC-ESI(+)-MS as described above for 3-hydroxybutyryl-CoA. Peaks exhibiting the desired mass, LC-ESI(+)-MS [M+H]+ 836 were collected, lyophilized and stored at −80 °C.</p><!><p>L-(3R,4E)-3-Hydroxy-4-hexenoic acid (17, 26 mg, 0.2 mmol) in 1 mL anhydrous CH2Cl2 was treated with 80 μL of 2M oxalyl chloride solution (0.16 mmol, 0.8 eq) and 4 drops of anhydrous DMF in a flask fitted with a glass funnel filled with Drierite. (Note: It is important not to use an excess of oxalyl chloride in order to avoid unwanted reaction of the allylic alcohol group.) After vigorous stirring at room temperature for 2–3 h, the solvent was evaporated after dilution with ethyl acetate, resulting in co-evaporation of any unreacted oxalyl chloride. The residue was dissolved in 2 mL THF and the solution was added to 20 mg of CoASH dissolved in 1 mL of 0.4 M NaHCO3, pH 8.0. After stirring for 1–2 h at room temperature, excess starting materials were removed by extraction with ethyl acetate. The resulted aqueous crude acyl-CoA mixture was purified by HPLC using a Phenomenex Gemini semi-preparative C18 column, 150 × 10 mm, equilibrated with 10% CH3CN/H2O. The sample was eluted with a linear gradient from 10% to 100% of CH3CN/H2O. HPLC peaks were collected and lyophilized. Each fraction was analyzed by HPLC-ESI(+)-MS using an Agilent Zorbax C18 column (2.1 × 50 mm, 3.5 μm) and a linear gradient from 10% to 100% of CH3CN/H2O.</p><!><p>(2Z,4E)-2,4-Hexadienoic acid (19) or (2E,4E)-2,4-hexadienoic acid (20) (0.2 mmol) was dissolved in 1 mL of anhydrous CH2Cl2 under N2. Triethylamine (70 μL, 47 mg, 4.0 eq) was added, followed after 10 min at 0 °C by 50 μL of ethylchloroformate (45.8 mg, 3.0 eq). The reaction mixture was stirred for 2 h at 0 °C. After evaporation of the solvent, the residue was dissolved in 2 mL THF and the insoluble salts were removed by centrifugation. The mixed anhydride was then slowly added to a separate round bottom flask containing 20 mg of CoASH in 1 mL of NaHCO3 buffer (pH 8.0). The reaction mixture was stirred 1–3 h at room temperature with monitoring by LC-MS. After removal of excess starting material by extraction with EtOAc, the aqueous crude acyl-CoA mixture was purified by HPLC using the Phenomenex Gemini semi-preparative C18 column, 150 × 10 mm, equilibrated with 10% CH3CN/H2O. The sample was eluted with a linear gradient from 10% to 100% of CH3CN/H2O. HPLC peaks were collected and lyophilized. Each fraction was analyzed by HPLC-ESI(+)-MS using an Agilent Zorbax C18 column (2.1 × 50 mm, 3.5 μm) and a linear gradient from 10% to 100% of CH3CN/H2O.</p><!><p>The activity of various protein preparations of FosDH1 and FosDH2 was conveniently checked using the surrogate –SCoA substrates (E)-2-butenoyl-CoA and 3-hydroxybutyryl-CoA. Assay mixtures contained 1 mM 3-hydroxybutyryl-CoA or (E)-2-butenoylCoA in 50 mM sodium phosphate, pH 7.5 Buffer in a total volume of 100 μL. The assay mixture was divided into 50-μL portions. To one, 100 μM FosDH1 or FosDH2 was added while the blank used an equivalent volume of 50 mM sodium phosphate, pH 7.5 buffer. The assay mixtures were incubated at room temperature for 2 h, then diluted with 200 μL of H2O and passed through a Millipore 30 kDa MWCO 500 μL filter and centrifuged at 14000g to remove FosDH1 or FosDH2 by buffer exchange. The assay mixtures were analyzed by LC-ESI(+)-MS using an Agilent Zorbax Extend C18 column (100 × 2.1 mm, 3.5 μm) with a linear gradient from 2% to 70% CH3CN/H2O to monitor for the expected increase or decrease of 18 amu in the mass of the acyl-CoA components as a result of hydration or dehydration.</p><!><p>For preparation of acyl-FosACP substrates, each reaction contained 100–150 μM apo-FosACP1 or apo-FosACP2, and 250–300 μM (3R)- or (3S)-3-hydroxybutyryl-CoA or (E)-or (Z)-2-butenoyl-CoA, plus 2 μM Sfp, 10 mM MgCl2, and 1 mM TCEP in 50 mM sodium phosphate, pH 7.5, in a total volume of 100 μL. The reactions were incubated at room temperature for 10 min to form the corresponding acyl-FosACP1 (5–8) or acyl-FosACP2 (11–14). The reaction mixture was then concentrated using a Amicon Ultra-0.5 (3000 MWCO) centrifugal filter and centrifuged at 14000g to remove unreacted acyl-CoA by buffer exchange, with recovery of the acyl-FosACP retentate which was diluted to a total volume of 200–300 μL.</p><!><p>The above-described samples of acyl-FosACP were divided into 2 equal 100–150 μL portions. To one, FosDH1 or FosDH2 was added to a final concentration of 50 μM while the blank was supplemented with an equivalent volume of 50 mM sodium phosphate, pH 7.5. After parallel 60-min incubations at room temperature, each reaction mixture was diluted with formic acid/H2O and centrifuged for 5 min at 14000g. These samples were then analyzed by LC-ESI(+)-MS and LC-ESI(+)-MS/MS using an analytical Aeris widepore-C4 column (3.6 μm, 2.1 × 150 mm) from Phenomenex using a linear gradient from 30% to 70% CH3CN/H2O on a Thermo-LXQ mass spectrometer. For LC-ESI(+)-MS/MS analysis15 the M11+ ion was selected for MS/MSs such that, both the hydrated and dehydrated pPant ejection fragments could be observed together (Figures S12, S13, S23, S24, S20, S21, and S27–S30).</p><!><p>Each reaction consisted of 100 μM apo-FosACP1 or apo-FosACP2 and 250–300 μM (3R)- or (3S)-3-hydroxybutyryl-CoA E)- or (Z)-2-butenoyl-CoA, plus 2 μM Sfp, 10 mM MgCl2, and 1 mM TCEP in 50 mM sodium phosphate, pH 7.5, in a total volume of 500 μL. (TCEP was omitted from the reaction with (Z)-2-butenoyl-CoA.) Each reaction was incubated at room temperature for 10 min to form the corresponding acyl-FosACP product. The reaction mixture was then concentrated using an Amicon Ultra-0.5 (3000 MWCO) centrifugal filter with centrifugation at 14000g to remove unreacted acyl-CoA by buffer exchange, with recovery of the acyl-FosACP retentate, which was diluted to a total volume of 500 μL.</p><!><p>The above-described samples of acyl-FosACP1 (5, 6, 11, and 12 were divided into 2 equal 250 μL portions. To one, FosDH1 was added to a final concentration of 40 μM while the blank was supplemented with an equivalent volume of 50 mM sodium phosphate, pH 7.5. Both mixtures, with and without FosDH1, were incubated at room temperature for 30 min before addition to each of 200 μM of PICS TE. After hydrolysis by PICS TE for 15 min, the reaction was quenched and the pH was adjusted to 3.0–3.5 by addition of ~9 μL of 1 M HCl. After centrifugation at 13000g for 5 min to remove precipitated protein, the supernatant was extracted with 4 × 800 μL of CH2Cl2. After removal of solvent by rotary evaporation, the residue was dissolved in 130 μL CH2Cl2 and then treated with 20 μL BSTFA for derivatization of the organic acids (Total 150 μl). The derivatized samples were directly analyzed by chiral GC-MS (HP GCD system) on a Varian CP ChiraSil_DEX column (25 m length, 0.25 μm diameter) using a temperature program (GC Method C) with a 1 min hold at 50 °C, followed by an increment of 8 °C/min to 150 °C, a 1 min hold, and then an increase of 20 °C/min to 210 °C, and a final hold at this temperature for 3 min (Figures S11 and S14).</p><!><p>3-Ketobutyryl-CoA (400 μM) was reacted for 10 min at room temperature with 150 μM apo-FosACP2 in the presence of 2 μM Sfp in 50 mM sodium phosphate, pH 7.2 containing 10 mM MgCl2 and 1 mM TCEP (total vol 400 μL) to form 3-ketobutyryl-FosACP2 (23). The reaction mixture was passed through a Millipore 3 KDa MWCO 500 μL filter by centrifugation at 14000g to remove unreacted acyl-CoA by buffer exchange. The retentate containing 3-ketobutyryl-FosACP2 (23) was divided into two separate 200 μL aliquots. To each of these, 1 mM NADPH and 40 μM FosKR2 was added. One reaction was quenched immediately by treatment with 150 mM NaOH (18 μL of 2 M NaOH) at 65 °C for 20 min. The other portion was incubated for 30 min before being quenched in the same manner with aq.with 150 mM of NaOH. After hydrolysis, each sample was cooled on ice and then acidified to pH 3.0–3.5 by addition of 54 μL of 1 M HCl. Each sample was centrifuged at 14000g for 5 min to remove precipitated protein. The supernatant was then extracted with 4 × 800 μL of CH2Cl2. The solvent was removed by rotary evaporation. The residue dissolved in 100 μL of CH2Cl2 was derivatized by treatment with 10 μL BSTFA. The samples were directly analyzed by chiral GC-MS on a ChiraSil_Dex capillary GC column using a temperature program (GC Method D) with a 1 min hold at an initial temperature of 50 °C followed by an increment of 8 °C/min to 150 °C, a 1 min hold at 150 °C, an increase of 20 °C/min to 210 °C, and a final 3 min hold at 210 °C (Figure S15).</p><!><p>(2Z,4E)-2,4-Hexadienoyl-CoA or (3R,4E)-3-hydroxy-4-hexenoyl-CoA (200–500 μM) was incubated with 50 μM apo-FosACP2 and 40 μM Sfp, 2 mM DTT, and 15 mM MgCl2 in reaction buffer (350 mM NaCl 50 mM phosphate pH 7.5, total vol 2.5 mL) for 30 min at 30 °C. The residual CoA substrate, MgCl2, and DTT were removed by passage of 2.5 mL of the incubation mixture through a PD-10 column that had been equilibrated with 25 mL of the reaction buffer. FosACP2-bound product, 18 or 21, was eluted by 3.5 mL of the reaction buffer. The PD-10 eluate was concentrated using an Amicon centrifuge unit 10,000 MWCO (30 min at 1,600g at 4 °C reduced the volume to 200–300 μL. The acylated FosACP2 was analyzed by LC-MS The low concentration of FosACP2 was used to avoid precipitation during the procedure. FosDH2 was added to a final concentraion of 300 μM along with 2 mM DTT and the reaction buffer was added to adjust the total volume to 500 μL. The reaction mixture was incubated at room temperature for 1–2 h. At the end of the reaction, the product was released from the FosACP2 by PICS TE (20 μL per reaction, 20 min at room temperature). The reaction mixture was acidified to pH <3 by addtion of 1 M HCl. The C6 acids were extracted with 2 × 800 μL ethyl acetate. The solvent was removed by rotary evaporation. The product was taken up in 80 μL of MeOH, 20 μL of TMS-CHN2 was added, and the solution incubated at room temperature for 5 min. The resulting methyl esters were analyzed by chiral GC-MS on a ChiraSil-Dex capillary GC column using GC Method B (Figures S16–S19).</p><!><p>The above-described samples (1 mL) of acyl-FosACP2 (7, 8, 13, and 14 were mixed with FosDH2 (final concentration 50 μM). Aliquots of 250 μL were collected at 0, 15, 30 and 45 min incubation time and added to Eppendorf tubes containing 200 μM of PICS TE and incubated for 15 min at room temperature. After hydrolysis by PICS TE, the reaction was quenched and the pH was adjusted to 3.0–3.5 by addition of ~9 μL of 1 M HCl. Samples were centrifuged at 13000g for 5 min to remove precipitated protein. The supernatant was extracted with 4 × 800 μL of CH2Cl2. The solvent was then removed by rotary evaporation and the residue was dissolved in 100 μL CH2Cl2 to which 10 μL BSTFA was added for derivatization of acids (Total 110 μl). These samples were directly analyzed by chiral GC-MS (HP GCD system) with Varian CP ChiraSil_DEX column (25 m length, 0.25 μm diameter) using a temperature program (GC Method D) with a 1 min hold at 55 °C, followed by an increment of 0.5 °C/min to 65 °C, a 1 min hold, and then an increase of 15 °C/min to 90 °C, a 1-min hold at 90 °C, an increase of 20 °C/min to 200 °C, and a final 1 min hold at this temperature (Figures S22, and S25).</p><!><p>(3R)- and (3S)-3-Hydroxybutyryl-FosACP2 (7 and 8) were generated by incubation of 800–1000 μM (3R)- or (3S)-3-hydroxybutyryl-CoA and 200 μM apo-FosACP2 with 5 μM Sfp in the presence of 10 mM MgCl2 and 1 mM TCEP in 50 mM sodium phosphate, pH 7.5 in a total volume of 100 μL. After 15 min incubation at room temperature to form (3R)- or (3S)-3-hydroxybutyryl-FosACP2 (7 or 8), the reaction mixture was passed through a Millipore 3 KDa MWCO 500 μL filter with centrifugation at 14000g to remove unreacted acyl-CoA by buffer exchange. The retentate (100 μL) containing 7 or 8 was mixed with 300 μL of [18O]-H2O-based buffer (final 18O enrichment 75 atom%.) To this solution, 50 μM (final concentration) FosDH2 was added. Samples of 95 μL were withdrawn after 0, 30 and 90 min for (3S)-3-hydroxybutyryl-FosACP2 (8) and 0, 30, 90 and 300 min for (3R)-3-hydroxybutyryl-FosACP2 (7). Each sample was added to an Eppendorf tube containing 3.5% formic acid (final concentration) and then directly analyzed by LC-ESI(+)-MS/MS with monitoring of the pPant ejection fragment from only the 3-hydroxybutyryl-FosACP and measurement of the relative abundance of the 347 Da (16O) and 349 Da (18O) species (Figure S26 and S31.</p>

PubMed Author Manuscript

Towards an atomistic understanding of polymorphism in molecular solids

Understanding and controlling polymorphism in molecular solids is a major unsolved problem in crystal engineering. While the ability to calculate accurate lattice energies with atomistic modelling provides valuable insight into the associated energy scales, existing methods cannot connect energy differences to the delicate balances of intra-and intermolecular forces that ultimately determine polymorph stability ordering. We report herein a protocol for applying Quantum Chemical Topology (QCT) to study the key intra-and intermolecular interactions in molecular solids, which we use to compare the three known polymorphs of succinic acid including the recently-discovered 𝛾 form. QCT provides a rigorous partitioning of the total energy into contributions associated with topological atoms, and a quantitative and chemically intuitive description of the intra-and intermolecular interactions. The newly-proposed Relative Energy Gradient (REG) method ranks atomistic energy terms (steric, electrostatic and exchange) by their importance in constructing the total energy profile for a chemical process. We find that the conformation of the succinic acid molecule is governed by a balance of large and opposing electrostatic interactions, while the H-bond dimerisation is governed by a combination of electrostatics and sterics. In the solids, an atomistic energy balance emerges that governs the contraction, towards the equilibrium geometry, of a molecular cluster representing the bulk crystal. The protocol we put forward is as general as the capabilities of the underlying quantum-mechanical model and it can provide novel perspectives on polymorphism in a wide range of chemical systems.

towards_an_atomistic_understanding_of_polymorphism_in_molecular_solids

Introduction<!>a. Periodic calculations<!>b. Molecular calculations<!>c. Quantum-chemical topology calculations<!>Results and Discussion<!>b. The IQA energy decomposition scheme<!>c. The Relative Energy Gradient method<!>d. Conformation of the succinic acid monomer<!>e. H-bond dimerisation<!>Monomer<!>𝛼<!>𝛼-SA<!>Conclusions

<p>Antibiotic resistance is fast becoming a major public health concern 1 . The development of new drugs has been drastically held back by the time and costs involved in research, with no new classes of antibiotic having been discovered since 1987 2 . Among the most difficult steps in taking a new drug molecule to a marketable formulation is identifying and controlling the resulting solid form. This solid form dictates key physical properties including the compressibility and the dissolution rate, which in turn determine processability and bioavailability, respectively 3,4 . The conformational flexibility and broad spectrum of intermolecular interactions often enables molecules to crystallise into multiple polymorphs and/or solvates under different crystallisation conditions. These polymorphs may subsequently transform into different forms under processing and storage conditions 5 . Polymorphs and solvates often display significant differences in physicochemical properties, introducing an extra level of complexity to drug design and manufacturing 6 .</p><p>A number of well-documented cases highlighting the impact of polymorphism on the pharmaceutical industry have made this an important contemporary research area 7,8 . In 1998, the capsule form of the HIV drug Ritonavir had to be temporarily removed from the market because the original Form I converted to a more stable and less soluble form, Form II, in the final formulation 9 . Although Form II was not discovered in the four years from initial development to marketing, once production lines became contaminated, the supply of the drug was drastically reduced while a new formulation was developed. Another example is the 1991 patent dispute over the anti-ulcer drug ranitidine hydrochloride. Ranitidine hydrochloride has two polymorphic forms, Form I and Form II, with very similar solubility and bioavailability, but Form II is easier to prepare than Form I. After discovering Form II, GSK obtained a new patent and, given the difficulty of preparing phase-pure Form I, the company was able to limit competition from generic manufacturers once the original patent on Form I expired 10 .</p><p>A recent statistical analysis of molecular crystals in the Cambridge Structural Database found that as many as 50 % of known molecules display polymorphism, and that differences in lattice energy very often lie within the chemical accuracy threshold of 1 kcal mol -1 (4 kJ mol -1 ) 8 . Despite the inherent challenge these circumstances pose to theoretical methods, crystal-structure prediction (CSP) 11 is a highly active research field.</p><p>In a typical CSP study, 10 3 -10 4 candidate crystal structures are generated and their lattice energies evaluated using a parameterised force field model, a first-principles electronic-structure method such as densityfunctional theory (DFT), or a combination of the two. Depending on the system, CSP may find a single (global) energy minimum or a set of energetically similar metastable polymorphs 11 . CSP has evolved with computing capability to become a useful counterpart to experiment, for example, to screen for unidentified polymorphs of new drugs 12 . Nevertheless, even moderately complex systems can challenge current state-of-the-art methods 13 .</p><p>A key disadvantage common to most contemporary CSP methods is the difficulty of ascribing the subtle energy differences between competing structures to specific chemical interactions. The implications of this situation are twofold. Firstly, it limits the insight available from CSP studies, which may otherwise point to predictive rules or "smarter" screening approaches. Secondly, in the cases where CSP fails to predict -Page 3 -experimental outcomes, it is difficult to identify the classes of interactions that the underlying total-energy methods cannot describe appropriately.</p><p>The current state-of-the-art for analysing total energies is to use quantum-chemical topology (QCT) methods 14,15 such as the Interacting Quantum Atoms (IQA) energy decomposition scheme 16 . IQA examines the molecular quantum-mechanical wavefunction (or the corresponding electron density function in DFT) to rigorously partition the total energies obtained from electronic-structure calculations into a sum of intra-and interatomic terms with intuitive chemical interpretations. We and others have successfully applied IQA, in conjunction with tools such as the Relative Energy Gradients (REG) method 17 , to examine many different phenomena. A few examples include hydrogen 17 and halogen 18 bonding, the fluorine gauche effect 19 , the biphenyl torsional angle energy barrier 20 , and the reaction mechanism of the peptide hydrolysis of HIV-1 protease 21 . In this work, we apply these methods to elucidate the chemical origin of the polymorphism in succinic acid (SA). By combining complementary periodic electronic-structure calculations with IQA analyses of SA monomers, dimers and clusters, we explore the delicate energetic balances that ultimately determine the structure and stability of the three known SA polymorphs. This method is general and provides a foundation for future studies to improve our fundamental understanding of polymorphism and to devise and improve novel CSP methods.</p><!><p>Periodic plane-wave DFT calculations were performed on the crystal structures of the 𝛼, 𝛽 and 𝛾 polymorphs of SA using the Vienna Ab initio Simulation Package (VASP) code 22 . A plane wave basis with a kinetic-energy cut off of 850 eV was used with Projector Augmented-Wave (PAW) pseudopotentials 23,24 including the H 1s and C and O 2s/2p electrons in the valence shell. Calculations were performed with six different exchange-correlation functionals: (i) the PBE generalised-gradient approximation (GGA) functional 25 , (ii) the PBE0 26 and (iii) B3LYP 27 hybrid functionals, (iv) PBE with the DFT-D2 correction 28 , (v) PBE with the DFT-D3 correction 29 , and (vi) PBE with the Tkatchenko-Scheffler (TS) dispersion correction 30 .</p><p>Γ-centered Monkhorst-Pack k-point meshes 31 with 2 2 3 (𝛼-SA), 2 1 3 (𝛽-SA) and 2 1 1 subdivisions (𝛾-SA) were used for the Brillouin-zone integrations. A series of gas-phase calculations were also performed as follows. SA molecules and H-bonded dimers were extracted from the experimental structures and placed at the centre of a large periodic box with an initial distance of 15 Å between images. These were then optimised with the same technical parameters as used for the crystal structures but with Γ-point Brillouin zone sampling.</p><p>In all calculations, the PAW projection was performed in the reciprocal space and non-spherical contributions to the gradient corrections inside the PAW spheres were accounted for. A tolerance of 10 -8 eV on the total energy was applied when optimising the Kohn-Sham orbitals. Geometry relaxations were performed with the atomic positions, and the lattice parameters and cell volumes in the periodic structures, allowed to vary until the magnitude of the forces on the ions fell below 10 -2 eV Å -1 .</p><p>-Page 4 -</p><!><p>Gas-phase electronic-structure optimisations and single-point calculations were performed on SA monomer, dimer and multi-molecule cluster models using B3LYP 27 and the 6-31+G(d,p) split-valence basis set 32 with the GAUSSIAN09 33 software. The Kohn-Sham orbitals were optimised with tolerances of 10 -6 and 10 -8 a.u. on the maximum and root-mean-square (RMS) changes in the density matrix, respectively. Geometry optimisations were performed to tolerances of 4.5 10 -4 and 3 10 -4 a.u. on the maximum and RMS force, and 1.8 10 -3 and 1.2 10 -3 a.u. on the maximum and RMS displacements, respectively. For the larger cluster calculations, we used the recommended SuperFineGrid setting for computing integrals.</p><!><p>The Kohn-Sham electron densities obtained from the molecular calculations were analysed using the IQA partitioning scheme as implemented in the AIMAll package 34 . The largest value of L(Ω) was 1.5 10 -3 a.u for the carboxylic carbon atoms in the γ conformation. The integration strategy was carefully and successfully optimised to reduce the absolute recovery error, defined as the difference between the calculated total energy and the sum of the IQA energy terms, to below 1 kJ mol -1 for all SA monomers and dimers. As outlined in the Results and Discussion section, series of IQA calculations for configurations along carefullyselected "control coordinates" were analysed using the Relative Energy Gradient (REG) method implemented in our in-house ANANKE software 17 .</p><!><p>a. Solid-state calculations Succinic acid has three reported polymorphs: a triclinic P-1 (𝛼) phase and the two monoclinic P2 1 /c (𝛽) and C2/c (𝛾) phases (Fig. 1). All three structures are built from chains of SA molecules formed by strong directional H-bonded carboxylic acid dimers, which pack parallel with weaker intermolecular interactions between adjacent chains. 𝛽-SA crystallises from solution and is stable under ambient conditions, 35,36 while 𝛼-SA is obtained by rapid quenching of a melt above ~135 °C37 and can also be prepared by sublimation 36 .</p><p>The 𝛼 → 𝛽 phase transition is slow, and once prepared, 𝛼-SA remains stable for long periods of time. 𝛾-SA was isolated 38 serendipitously in 2018 and differs markedly from 𝛼and 𝛽-SA in that the SA molecules adopt a "folded" or "twisted" rather than planar geometry. SA adopts the planar geometry in 89 % of reported multicomponent crystals (i.e. those formed from two or more different molecules), making the twisted configuration comparatively rare. geometry, and 𝛾 (e), which is based on the twisted geometry. These images were generated using the VESTA software 39 .</p><p>PBE predicts an energetic ordering of 𝛼 < 𝛾 < 𝛽, with energy differences of 1.2 and 1.7 kJ mol -1 per SA molecule between the 𝛼 and 𝛾 polymorphs, and the 𝛾 and 𝛽 polymorphs, respectively, which we denote E( -) and E( -). These energy differences are both well within the 4 kJ mol -1 chemical accuracy threshold.</p><p>The two hybrid functionals, PBE0 and B3LYP, predict an ordering of 𝛾 < 𝛽 < 𝛼, and these XC-functionals notably predict the 𝛾 polymorph to be 7.2 and 5.9 kJ mol -1 lower in energy than the 𝛽 phase, respectively (Table S1 of the Electronic Supplementary Information (ESI)). When one of the three dispersion corrections is applied to PBE, the energy differences between polymorphs are reduced to within 1-2 kJ mol -1 per SA molecule. Our PBE-D2 and PBE-TS results are in line with the calculations carried out by Lucaioli et al. 38 , and a full set of energy differences calculated with the six functionals is given in Table S1. Overall, there are six possible energy orderings, of which four are recovered by the six levels of theory tested, and none of the orderings is predicted by more than two of the functionals.</p><p>By comparing the optimised and experimental structures, we find that PBE and B3LYP overpredict the unit cell volumes by 13-19 %, PBE0 overpredicts by 8-11 %, and the three dispersion-corrected functionals predict smaller volume changes ranging from a 1 % expansion to a 6 % contraction (Tables S2-S4 in the ESI).</p><p>-Page 6 -All six functionals predict similar molecular conformations, with RMSD values from 8.1 10 -3 Å to 2.8 10 -2 Å compared to the PBE structure (overlay plots in Fig. S1-S3, Table S5, all in the ESI). The six functionals also predict similar SA dimer H-bond distances with a maximum difference of ~0.1 Å across the three polymorphs (Table S6, ESI).</p><p>Gas-phase calculations on the planar and twisted SA conformation of the single molecule consistently predict the twisted form to be lower in energy than the planar conformer, with energy differences ranging from 0.7 kJ mol -1 with PBE to 1.9 kJ mol -1 with PBE-D2 (Table S7, ESI). Calculations on gas-phase dimers in both conformations indicate formation energies (E F ) from 65 to 84 kJ mol -1 , and all six functionals predict the twisted dimer to be more stable than the planar dimer by 0.8-1.4 kJ mol -1 (Tables S7 and S8 in the ESI).</p><p>The gas-phase calculations consistently show the twisted monomer and dimer to be the lowest in energy, and the solid-state calculations predict similar molecular geometries and H-bond distances. We therefore suggest that the large differences in the cell volumes, and the variability in the energetic ordering predicted by the six functionals, is due to differences in how these functionals describe the weaker intermolecular forces between the SA chains.</p><!><p>Having established a baseline for our calculations, we next applied the IQA method to determine the origin of the predicted energetic differences between the SA conformations and the three solid-state polymorphs. The IQA scheme emerges from the Quantum Theory of Atom in Molecules (QTAIM) approach as a rigorous decomposition of the total energy into a sum of intra-and interatomic energy terms, 16,40 and provides detailed and quantitative descriptions of the underlying chemical interactions.</p><p>The total energy is decomposed into a sum of the IQA energies 𝐸 of the 𝑁 topological atoms in the molecular system (single molecule or molecular aggregate) according to:</p><p>The 𝐸 (i) energy of the i-th topological atom can be expanded as a sum of intra-and inter-atomic contributions:</p><p>The energetic contribution 𝐸 comprises a sum of the (intra-)atomic kinetic energy 𝑇 𝑖 and the electron- where 𝑉 𝑒 𝑛 𝑖, 𝑗 and 𝑉 𝑛 𝑒 𝑖, 𝑗 are, respectively, the potential energy contributions due to interaction of the electrons associated with atom i and the nucleus of atom j, and vice versa, i.e. the order of the subscripts is significant 𝑉 𝐶𝑜𝑢𝑙 𝑖, 𝑗 is the Coulombic interaction between the electrons in atoms i and j while 𝑉 𝑥𝑐 𝑖, 𝑗 primarily reflects the degree of covalent bonding between the two atoms 43 . Finally, it is often convenient to group the "classical"</p><p>("cl") terms in Eqs. ( 3) and (4), viz. V n-n , V e-n , V n-e and the purely electrostatic part of V e-e , i.e. V Coul , into a single term V cl (i,j).This new term allows Eq. ( 3) to be rewritten:</p><p>The interatomic terms arising from electron-electron interactions are calculated from a six-dimensional integration of the appropriate densities over the volumes of the two topological atoms involved. The classical electrostatic terms capture the electrostatic energies along with charge transfer effects, while the exchangecorrelation interactions capture, at DFT level, only covalency and (hyper)conjugation.</p><p>A limitation of all current IQA implementations is that they only work with molecular (and thus aperiodic) systems. We therefore identified and analysed the three main interactions involved in the SA crystal packing using appropriate molecular models, viz. the planar and twisted conformers of the SA molecule, dimers of SA molecules in the two conformations, and H-bonded chains packed to form larger clusters representative of the extended crystal structure. We chose to perform our calculations with the B3LYP hybrid functional, as this is a typical choice for molecular quantum chemistry, but we note that IQA can in principle be applied to the wavefunctions (or electronic densities in the case of DFT) obtained from any electronic-structure method.</p><p>-Page 8 -</p><!><p>The number of individual energy terms in an IQA decomposition rapidly becomes large as the size of the system increases, making manual analysis of the data impractical. As a result, it becomes hard to answer a crucial chemical question: which individual energy terms are most responsible for the energetic behaviour of the total system? This question is at the heart of any chemical phenomenon, such as hydrogen bonding, the gauche effect, the anomeric effect, and rotational energy barriers, to name a few. In the current study, we aim to identify which atoms play a pivotal role in the crystallisation of SA into the three polymorphs, and which type of energy (i.e. steric, electrostatic or exchange) controls the relevant interactions. The Relative Energy Gradient (REG) method is designed to answer this question, and to do so by unbiased computation. REG operates on a dynamic change, i.e. it requires a sequence of molecular geometries and the corresponding energies that represent the chemical phenomenon being studied. For example, a REG analysis of a rotational energy barrier requires a series of geometries generated by varying the relevant torsion angle, termed the "control coordinate". In the case of a REG analysis of a hydrogen bond, the control coordinate is typically the H-acceptor distance. In the current study we perform three different REG analyses, each with its own control coordinate: (i) the central C-C torsional angle within one SA molecule, or both molecules in a SA dimer; (ii) the hydrogen bond distance between two SA molecules; and (iii) the unit cell volume in the crystal structures.</p><p>As the name suggests, the REG compares two energy gradients by calculating their (dimensionless) ratio, which is termed the REG coefficient. The gradient of each energy contribution is compared to the gradient of the total energy, both of which vary along the control coordinate. These ratios are then ranked to identify the most significant energy components in terms of their impact on the overall change in total energy.</p><p>The key idea is to identify the largest positive REG coefficients, corresponding to the atomic energy contributions that most support the total energy change, and the most negative REG coefficients, which identify the energy terms that most oppose the total energy change.</p><p>The control coordinate is divided into segments whose extremes are at critical points of the potential energy surface (PES) as a function of the control coordinate (i.e. minima, maxima and/or saddle points). The behaviour over each segment is analysed separately, and both IQA and total energies are calculated over a number of geometries determined by the control coordinate. The REG coefficient (R k ) for the k-th IQA term is calculated as follows:</p><p>where 𝑅 denotes the coefficient of the linear regression used to fit (over all geometries of a given segment)</p><p>the IQA energy term 𝜖 𝑘 and the IQA energy of the total system 𝐸 , and c k are constants without physical meaning. The REG coefficients 𝑅 measure how large the changes in 𝜖 𝑘 are compared to the change in the total energy within each segment. Note that the sign of R k and its interpretation (i.e. whether a term supports or opposes the change in total energy) is independent of the direction in which the analysis is performed (i.e.</p><p>from minimum to maximum or vice versa).</p><p>-Page 9 -</p><!><p>To explore the energetic differences between the planar 𝛼/𝛽 and twisted 𝛾 conformations of SA, we performed a scan of the PES associated with the C-C torsion angle (Fig. 2). The PES has two unique minima at dihedral angles of 70 and 180° corresponding to the twisted ( 𝛾 ) and planar ( 𝛼 / 𝛽 ) conformations, respectively. These calculations predict the twisted conformation to be lower in energy than its planar counterpart by 0.3 kJ mol -1 with a rotation barrier of 5 kJ mol -1 . The 0.3 kJ mol -1 energy difference between the twisted and planar forms is of the same order of magnitude as the 0.7 kJ mol -1 computed with B3LYP plane-wave calculations. These segments correspond to the paths (i) from the twisted 𝛾 conformation to the energetic maximum, and</p><p>(ii) from the maximum to the planar 𝛼/𝛽 conformer. Table 1 identifies the largest REG coefficients (in absolute value), which are the most important to understand the chemical origin of the rotational barrier.</p><p>Table 1. REG analysis of the IQA energies along the paths linking (i) the twisted (𝛾) SA conformer with the energy maximum (Segment 1), and (ii) the maximum and the planar (𝛼/𝛽) conformation (Segment 2). These segments are marked on the PES as a function of the C-C torsion angle in SA shown in Fig. 2(a)/(b), which also shows the atom labelling scheme. The notation 𝑉 (i,j) denotes a classical electrostatic interaction between the pair of atoms in parentheses. For each segment, the terms with the largest absolute REG coefficient 𝑅 𝑘 are shown along with the (Pearson) correlation (R)</p><p>to the total energy. -Page 11 -</p><p>In Segment 2, which corresponds to torsion from the local maximum to the minimum at the planar conformation, these same repulsive interactions support the decrease in total energy, i.e. the repulsion energy decreases and thereby stabilises the planar minimum. However, new electrostatic interactions become significant, viz. those between the carbonyl atoms and the methylenic hydrogens in both 1,3 and 1,4 relationships. Finally, the attractive interaction between the two opposing carboxyl groups now emerge as the most dominant negative R k values, indicating that these interactions continue to strengthen on approach to the planar minimum.</p><p>In conclusion, the above analysis shows that the rotational barrier is governed by classical electrostatic interactions, in particular those of the two carboxylic acid groups. Indeed, a comparison of the two REG analyses (one for each Segment) identifies the most important terms with the largest absolute 𝑅 to be (i) the attractive interactions between opposing carbonyl C and acceptor O atoms, 𝑉 𝑐𝑙 (C,O a '), (ii) the repulsive contacts between carbonyl C atoms, 𝑉 𝑐𝑙 (C,C') , and (iii) the repulsive interactions between acceptor O atoms, 𝑉 𝑐𝑙 (O a ',O a ). As shown in Fig. 2(d)/2(e), the rotation from the twisted to the planar conformer, via the PES maximum, leads to a continuous weakening of all three interactions. The twisted conformer therefore maximises the attractive interaction relative to the two repulsive terms, making it slightly more stable. We note that the changes in energy associated with these electrostatic terms are some two orders of magnitude larger than the barrier height itself. The energy difference between the two conformers, and thus the PES, arises from a balance of energetically large, but opposing, chemical interactions.</p><!><p>We next examined the formation of H-bonded dimers of both SA conformers, as the H-bond between carboxylic acid groups likely represents the strongest single intermolecular interaction in all three SA polymorphs. Here the H-bond distance was taken as the control coordinate in REG analyses of the twisted and planar dimers. The coordinate was adjusted from the calculated equilibrium distance of ~1.65 Å to values between 1.15 Å (compression), and 4.55 Å (extension) in steps of 0.1 Å (Fig. 3). The resulting potential energy curves predict dimer formation energies of -66.8 and -68.0 kJ mol -1 for the planar and the twisted dimers, respectively, which are once again very similar to those computed from the plane-wave calculations (-65.1 and -65.5 kJ mol -1 ). The twisted dimer is predicted to be 1.4 kJ mol -1 per SA molecule more stable than the planar dimer, which is a fivefold increase on the energy difference between the monomers, although the plane-wave calculations predict a much smaller stabilisation of 0.2 kJ mol -1 per molecule.</p><p>In order to gain further insight into the selective stabilisation of the dimer, the curves in Fig. 3 were divided into two segments corresponding to the repulsive and attractive part of the potential energy curve, respectively at small and large monomer separations. The IQA decomposition of the total energies of the configurations in each segment was analysed using REG, in order to identify the most important terms summarised in Table 2.</p><p>-Page 12 - Table 2. REG analysis of the partitioned IQA energies along the H-bond compression (Segment 1) and lengthening (Segment 2) regions of the H-bond potential energy curves for the planar and twisted succinic acid dimers shown in Fig. 3. The two monomers in the dimer are indicated by the subscripts "1" and "2". The parameters R k and R have the same meaning as in Table 1. The 𝐸 refer to the intra-atomic energies of the atoms in parentheses, and the 𝑉 refer to the classical electrostatic interactions between the pairs of atoms in parentheses. The atom labelling is shown in Fig. 3 We find that electrostatic interactions between atoms in the two carboxyl groups involved in the H bond play the largest role in the formation of the dimer (Segment 2). Attractions between the carbonyl carbon and -Page 14 -donor/acceptor oxygen atoms on the opposing group make a substantial supporting contribution, as does the attraction between the acceptor oxygen and donor acidic proton. The latter phenomenon is expected and has been seen in REG analyses of other H-bonded systems, and confirms the H-bond in the SA dimer to be predominantly electrostatic in nature. We emphasise that while the two O⋯H hydrogen bonds feature much in stabilising the SA dimer, they do so alongside non-bonded C⋯O contacts between the two adjacent carboxyl groups.</p><p>Energy terms with negative REG coefficients identify destabilising interactions that oppose the H-bond formation. In principle, the positive REG coefficients suffice to explain the nature of the attraction between the monomers when forming the dimer. However, the negative REG coefficients provide an alternative narrative, which is again identical for both the planar and twisted dimers. The dominant negative REG coefficients again involve atoms from opposing carboxyls. This time all electrostatic interactions are repulsive in nature, starting with the most dominant one, which is the repulsion between the carbonyl carbons. As expected, all possible O⋯O interactions across the carboxyls play a dominant role. More surprising, however, is the strong repulsion between the acidic protons and the carbonyl carbons.</p><p>We now explain the nature of Segment 1, starting with the most positive REG coefficients. As for Segment 2, the analysis is qualitatively the same for the planar and the twisted dimer. As the dimer is compressed beyond its equilibrium geometry, the intra-atomic energy 𝐸 of the donor O atoms increases most, compared to other types of local energy. This indicates a steric effect where the atom's kinetic energy is combined with the potential energy of the deforming electron cloud to strengthen the energy barrier to compression. The next three most dominant energy contributions are all electrostatic, and by deduction repulsive, because they help in constructing the compression energy barrier. Perhaps unexpectedly, the interaction between the carbonyl C and the acidic proton of the opposite COOH plays a leading role. The interaction between the two carbonyl carbon atoms follows closely, as does that between the donor and acceptor oxygen atoms. Finally, the alternative narrative associated with the negative REG coefficients shows that increased electrostatic attraction between the carboxyl groups play the most important role in counteracting the energy barrier. This assertion reinforces the role of the electrostatic interaction between the carboxyl groups over the whole energy profile, throughout the two segments.</p><p>The REG coefficients for the twisted and planar dimers are similar, and thus do not highlight any clear differences in H-bond strength that might explain the higher stability of the twisted dimer. We therefore investigated the hypothesis that this higher stability is instead due to differences in the intramolecular interactions within the SA monomers. A set of calculations analogous to those performed on the SA monomer in Fig. 2, but where both molecules in the dimer are rotated from the twisted to the planar form, were therefore run, as shown in Fig. 4(a). This procedure yields a rotational barrier of 11.1 kJ mol -1 (5.5 kJ mol -1 per SA molecule), which is ~10 % higher than in the monomer. A REG analysis taking as the control coordinate the C-C torsion angle -again in both monomers -confirms that the same terms govern the rotational barrier in the monomer and dimer (Table 3). Table 3. Comparison of the REG coefficients 𝑅 𝑘 for the three major electrostatic interactions determining the variation in energy along the rotational PES between the twisted (𝛾) conformations of the succinic acid monomer and dimer and the local energy maximum (Segment 1), and the maximum and the planar (𝛼/𝛽) monomer and dimer (Segment 2). These segments are marked on the twist potential energy surfaces in Figs. 2 and 4. The atom labelling scheme follows that used in Figs. 2 and 4. Note that the two monomers in the dimers are equivalent. The notation 𝑉 (A,B) denotes the classical electrostatic interaction between the pairs of atoms in parentheses. Further insight can be obtained by comparing the QTAIM atomic charges in the SA molecules in the monomers and dimers in the planar and twisted conformations (Table 4). Importantly, these charges are obtained directly from the same type of volume integral as the IQA energies, a uniformity not found in other common partitioning schemes. There is little difference between the charges in the twisted and planar conformations, whether in the monomer or the dimer. However, the dimerisation leads to a clear redistribution of charge, on the order of tens of milli-electrons. In both the planar and twisted dimers, there is a quantitatively similar charge transfer within the carboxyl group involved in the H-bonding. Upon dimerisation the H and acceptor O both become more positive, while the carboxyl C atom becomes more negative, which can be interpreted as an internal charge transfer. The increase in positive charge of the hydrogen-bonded H atoms is well known and can be observed through enhanced infrared activity in H-bonded systems [44][45][46] .</p><p>Table 4. Atomic net charges q (a.u.) on each atom in the planar and twisted conformations of the succinic acid monomers and dimers found in 𝛼-/𝛽-SA and 𝛾-SA, respectively (note that the two monomers in the dimers are equivalent). The charge difference (Δ) for atom A is defined as q A (dimer)q A (monomer). The atom labels are shown in Figs. 2 and 4. In these analyses, the volume of the unit cell in the periodic calculation provides a natural control coordinate for REG analyses because the expansion and contraction of the volume about the computed equilibrium (i.e. the energy/volume equation of state (EoS) curve) probes the full range of energetic interactions that determine the equilibrium structure. We therefore performed a set of periodic calculations in which each of the three SA structures was re-optimised with the cell volume fixed to 5 % of the calculated equilibrium value in steps of 1 %. Due to the significant computational overhead of hybrid functionals in the periodic electronic-structure calculations, it was not possible to compute the EoS curves using B3LYP. We therefore used PBE instead, as this functional predicts the most similar equilibrium volume to B3LYP, and we performed a series of single-point energy calculations on the PBE-optimised structures. This procedure is equivalent to the rapid volume optimisation method outlined by Jackson et al. 47 . The resulting energy/volume curves are shown in Fig. 5. A fit of the Birch-Murnaghan equation of state 48 to the PBE 𝐸/𝑉 curve yields equilibrium energies, 𝐸 0 , within 0.5 kJ mol -1 per molecule and equilibrium volumes, V 0 , within 4-6 % of the values obtained by variable-cell optimisation. Fitting the 𝐸/𝑉 curve obtained with the B3LYP single-point energies computed with PBE structures yields a similar error in the predicted V 0 but a rather larger ~3.5 kJ mol -1 error in the 𝐸 0 . Nevertheless, the computed energies predict the same stability ordering of 𝛾-SA < 𝛽-SA < 𝛼-SA.</p><!><p>-Page 18 - (extracted with the IQA energy partitioning) in the centre of the clusters, represented in the images using balls and sticks rather than lines. The images were produced with the VMD software 49 .</p><p>-Page 19 -</p><p>To examine how well the cluster models reproduce the solid-state 𝐸/𝑉 curves, we compared singlepoint energy calculations on the clusters, using B3LYP, with single-point B3LYP calculations on the periodic structures. The gas-phase computations show a reasonable overlap with the solid-state calculations at expanded volumes but the calculations on 𝛼-SA and 𝛽-SA deviate significantly at compressed volumes. This is likely because the outer shell of molecules in the clusters are in a very different chemical environment to those inside the periodic structure. Partitioning the total energies using the IQA and extracting the energy of the reference "bulk like" molecule largely corrects this discrepancy, which suggests that the central molecules in these clusters are representative of the monomers within the corresponding crystal structures. However, we note that the cluster and IQA calculations both predict a different energetic ordering to the periodic calculations, viz. 𝛼 < 𝛽 < 𝛾 (Fig. S4(b)) and 𝛼 < 𝛾 < 𝛽 (Fig. S4(d)), respectively. The comparison of the full energy/volume curves (Fig. S4) shows that this effect is not due to the noise in the energies. Instead, we attribute the discrepancy between the periodic and molecular B3LYP calculations to implementation differences in the periodic and aperiodic codes used for the solid-state and molecular cluster models. Given the small energy differences between the polymorphs predicted by the initial periodic calculations, the differences in qualitative stability ordering are perhaps inevitable.</p><p>Nonetheless, we proceed to analyse the energy differences based on the partitioned energies of the different types of atoms in the reference molecules (Table 5). Comparison of the IQA contributions in 𝛼-SA and 𝛽-SA, for which the reference SA molecule is in the planar conformation, shows that the higher energy of the 𝛽-SA phase is almost entirely by virtue of the destabilisation of the donor O atoms. The same is true when comparing the 𝛼-SA and 𝛾-SA reference molecules, for which the higher electronic energy of the latter occurs through a balance of (i) stabilisation of the acidic H and both C atoms, and (ii) destabilisation of the two O atoms and the methylene (𝛼) H atoms. The respective stabilisation and destabilisation of the C and donor O atoms in the carboxylic acid groups are particularly significant.</p><p>-Page 20 - To better understand these effects, each E/V curve in Fig. 5 was separated into two segments bounded by the volume with the lowest energy, resulting in two segments corresponding to volume compression and expansion. We found that these changes in volume have a minimal effect on the conformations of the SA monomers and the H-bond distances. We observed a maximum RMSD of 2.5 10 -2 Å in the atomic positions of the SA monomers and a maximum change in the H-bond distance of 5.7 10 -2 Å across the full set of expansions and compression or all three structures (Tables S9-S11 in the ESI). Thus, the differences in cell volume are almost entirely due to changes in the distances between the SA chains. Therefore, the region of the EoS curve from the most expanded volume to equilibrium mimics the process of the SA chains coming together to form the crystals, and the analysis of this section of the EoS curve gives insight relevant to crystal growth. Likewise, examination of the compression region would be relevant to explain changes to the crystal structure under pressure, which is in itself an interesting topic but which we do not pursue here. We therefore analysed the IQA energy curves only over the expansion region using the REG method. It is natural to analyse this energy segment from the expanded configuration to the equilibrium, i.e. in the direction corresponding to forming the equilibrium crystal. Thus energy terms with positive (negative) 𝑅 coefficients correspond to terms that stabilise (destabilise) the crystal formation.</p><p>Due to the size of the clusters, we restricted the number of energy terms calculated in the IQA decompositions by using two complementary analysis modes, viz. 𝐴𝐵 and 𝐴𝐴'. The 𝐴𝐵 analysis considers for each atom in the reference molecule an intra-atomic (𝐴) energy and a series of pairwise interactions with the other atoms in the reference molecule (𝐴𝐵). This procedure yields a total of 14(14-1)/2+14=105 energy terms and describes how the atoms in a single SA molecule interact with each other in the bulk environment of the -Page 21 -crystal. This AB analysis does not take into account explicit interactions with the other molecules in the crystal but it does consider the influence of the environment on the intra-and inter-atomic energies with respect to the gas-phase monomer the molecules in the gas-phase dimer. On the other hand, the 𝐴𝐴' analysis returns only a single energy term for each of the 14 atoms in the reference molecule, but these energies include both the intraatomic energy and the interaction energies with all the other atoms in the cluster. In other words, the 𝐴𝐴' analysis adds a description of how the energies of the atoms in the reference molecule are influenced by explicit interactions with the other neighbouring molecules in the crystal. The comparison of these two analyses allows the separation of the energetic contributions due to (i) the conformation of the molecule, and (ii) the intermolecular interactions associated with the crystal packing.</p><p>REG analyses show that the dominant energetic terms governing the packing in the SA crystal structure are again predominantly electrostatic in nature (Table 6), except for the weak steric stabilisation (E ntra )</p><p>of the carbonyl C atoms in the 𝛽-SA and 𝛾-SA polymorphs. The majority of the energy contributions are attractive electrostatic interactions between the acidic H, methylene (𝛼) H and carbonyl C on one hand, and the donor and acceptor O atoms on the other hand. Furthermore, the positive 𝑅 values indicate that the conformations of the monomers adapt to the crystal environment in order to optimise these attractive contacts.</p><p>In the planar 𝛼 and 𝛽 polymorphs, the equilibrium conformation also reduces the repulsion between the acidic H and carbonyl C atoms, whereas in 𝛾-SA the repulsion between the two carbonyl C atoms is reduced.</p><p>However, these reduced repulsions are counteracted by 𝑉 𝑥 terms between the methylene C and H atoms, as reflected by their negative R k values, indicating that they oppose the change in total energy. The adapted conformation thus weakens the covalent bonding between these atoms. Finally, the electrostatic stabilisation is strongly counteracted by the steric destabilisation of the methylenic H and acceptor O atoms in all three polymorphs, and by the steric destabilisation of the donor O in the 𝛼 and 𝛽-SA polymorphs. Thus the crystal packing also leads to destabilising deformation of the electron densities of the atoms in the monomers.</p><p>-Page 22 -Table 6. REG analysis of the partitioned IQA energies of the central reference molecules in cluster models of the 𝛼-SA, 𝛽-SA and 𝛾-SA crystal structures, computed with the 𝐴𝐵 analysis, as the unit-cell volume is adjusted from the expanded to the equilibrium volume. 𝐸 𝐼𝑛𝑡𝑟𝑎 denote intra-atomic energies modified by deformation of the atomic densities, while V cl (A,B) and V x (A,B) denote, respectively, classical electrostatic and exchange interactions between the pairs of atoms in parentheses. The R k are only shown for polymorphs where the corresponding energetic terms are significant. Positive (negative) R k values correspond to energy terms that stabilise (destabilise) the crystal as unit-cell volume is adjusted from an expanded volume to the equilibrium. The atom labelling is shown in Fig. 3. The complementary 𝐴𝐴' analysis shows that the interactions with neighbouring molecules include a variety of classical electrostatic, exchange interactions and steric influences (Table 7). All three polymorphs show stabilising exchange interactions at the donor O atoms. The 𝛼-SA and 𝛽-SA forms both show strong electrostatic stabilisation of the donor O atoms, together with weaker exchange stabilisation of the acceptor O atoms. All three polymorphs also show weak exchange stabilisation of the α H. In 𝛼-SA and 𝛽-SA the strongest destabilisation is in the intra-atomic energy of the donor O atoms, while a similar steric destabilisation of the acceptor O and α H atoms is present in all three polymorphs. We note that the overall steric destabilisation on adjusting the volume to the equilibrium is consistent with the AB analysis.</p><!><p>Table 7. REG analysis of the partitioned IQA energies of the central reference molecules in cluster models of the 𝛼-, 𝛽and 𝛾-SA crystal structures, computed with the 𝐴𝐴 analysis, as the unit cell-volume is adjusted from the expanded to the equilibrium volume. 𝐸 𝐼𝑛𝑡𝑟𝑎 denote intra-atomic energies modified by deformation of the atomic densities, while 𝑉 (A)</p><p>and 𝑉 (A) denote respectively classical electrostatic and exchange interactions associated with the atom in parentheses.</p><p>The R k are only shown for polymorphs where the corresponding energetic terms are significant. Positive (negative) R k values correspond to energy terms that stabilise (destabilise) the crystal as unit-cell volume is adjusted from an expanded volume to the equilibrium. The atom labelling is shown in Fig. 3. By taking these analyses together, we can extract the following general trends. In the bulk crystal environments, the monomers optimise the intra-molecular electrostatic interactions between atoms at the expense of steric destabilisation of some atoms. The interaction with neighbouring molecules produces additional stabilisation through a mix of electrostatic and covalent interactions associated mainly with the O atoms and the methylene groups. Within the IQA analysis, the 𝑉 𝑥 and 𝑉 𝑐𝑙 terms reflect covalent and polar interactions respectively, and their importance in the AA' analysis can be attributed to the formation of strong H-bonds with neighbouring molecules. This indicates that the dominant interaction in the SA crystals is the formation of hydrogen bonds with neighbouring chains in the cluster.</p><!><p>The REG analyses also provide additional insight into the origin of the (predicted) energy differences between the polymorphs. The 𝐴𝐵 analysis shows that the intra-molecular electrostatic interaction between the donor O and C atoms in the carboxylic acid group has a larger 𝑅 value for the 𝛼-SA than for 𝛽-SA. On the other hand, the 𝐴𝐴' analysis shows that electrostatic stabilisation of the donor O atoms is more important in lowering the energy of the 𝛼 phase as the crystal is formed. This suggests that the difference between the two planar SA polymorphs is primarily due to differences in the electrostatics. On the other hand, the AB analysis shows that some of the intra-molecular electrostatic interactions that stabilise the 𝛼 and 𝛽 polymorphs are not important in 𝛾-SA, while the 𝐴𝐴' analysis shows a reduced significance of 𝑉 𝑥 terms, in particular interactions with the acceptor O atoms, in 𝛾-SA. However, both analyses notably show that the increased 𝐸 of the donor O atoms, which constitutes a significant destabilising effect in the 𝛼 and 𝛽 polymorphs, is not important in the formation of the 𝛾-SA crystal. This observation is consistent with the comparison of the atomic energies in Table 5, but provides greater insight into the chemical interactions responsible for the differences. Thus, as for the SA monomer, the differences in energy between the twisted and planar polymorphs may be a balance of energetically large, but opposing effects, which partially explains the differences in qualitative stability ordering obtained with different functionals.</p><p>Before moving on to the general conclusions, two remarks on future developments are useful. Firstly, the small energy differences, on the order of kJ mol -1 , between the three succinic acid polymorphs is fairly typical of molecular solids and challenges the accuracy of theoretical methods. In particular, it is possible that an accurate description of dispersion forces may be important to account for the correct energetic ordering between polymorphs. The IQA can be used with more accurate electronic-structure methods such as MP2.</p><p>This approximation should provide an improved description of electron correlation and it would more accurately model dispersion. However, calculations on the large cluster models used here to represent the bulk crystal structure are likely to be prohibitively expensive. Nonetheless, analyses of the type outlined here may provide useful quantitative information on why different DFT functionals predict different energetic ordering, which may inform future development of new electronic-structure methods.</p><p>Secondly, current implementations of IQA are restricted to non-periodic systems. While our cluster model obtained from a solid-state energy/volume curve appears to work reasonably well in this case, adapting IQA for periodic systems would likely be both more accurate and more efficient. On the other hand, many -Page 25 -molecular solids have unit cells containing hundreds of atoms, and periodic plane-wave DFT calculations on such systems with hybrid functionals or post-DFT methods are likely to be prohibitively expensive. This problem may be partially mitigated by periodic DFT implementations with local orbitals. On the other hand, the development of improved functionals is an active development area, and advances in software efficiency and computing power are steadily enabling more accurate calculations to be performed on larger systems. We would therefore expect that the protocol we put forward here will be applicable to a wide variety of interesting and topical polymorphism problems in the near future.</p><!><p>The present case study of succinic acid demonstrates that detailed information from quantum-chemical topology calculations can provide atomistic chemical insight into polymorphism in molecular solids. The REG method, when combined with intra-atomic and interatomic energies from Quantum Chmiacl Topology, identifies the energy terms that best represent and thereby govern the energetic behaviour of the total system.</p><p>We studied all three known polymorphs of succinic acid (𝛼, 𝛽 and 𝛾), for which the twisted conformer, corresponding to γ, is consistently the lowest in energy in the gas phase, at any level of theory used. Three REG analyses were performed on the monomer, the dimer and clusters of succinic acid molecules, representing the different interactions in the solid-state.</p><p>Firstly, the relative energies and rotational barrier between the twisted and planar forms of succinic acid result from a balance of large and opposing electrostatic interactions that are 2 to 3 orders of magnitude larger than the corresponding energy differences. The rotation barrier between the twisted (𝛾) and planar (𝛼, 𝛽)</p><p>conformers is electrostatic in nature, and governed by atoms from the two COOH groups at opposite ends of succinic acid. More precisely, we find that repulsive C⋯C interactions and attractive C⋯O(=C) interactions dominate the rotation barrier.</p><p>Secondly, the assembly of a H-bonded dimer of either planar or twisted succinic acid molecules is again determined predominantly by electrostatic interactions between the two COOH groups involved in the hydrogen bond. Remarkably, the four non-bonded C⋯O contacts between the two adjacent COOH groups are slightly more important in determining the equilibrium H-bond distance than the O ⋯ H interactions themselves. As the dimer is compressed beyond its equilibrium geometry, the intra-atomic energy of the donor O atoms explains the increase in total energy and thus the barrier to compression, followed closely in importance by repulsive interactions between the carbons, and between the carbon and acidic proton. Furthermore, the dominant energetic terms governing the packing in the crystal structure are again predominantly electrostatic in nature, except for the weak steric stabilisation of the carbonyl C atoms in the 𝛽 and 𝛾 polymorphs. A further analysis was also performed focusing on a single energy term for each of the 14 atoms in a reference succinic acid molecule in a bulk-like chemical environment, which includes both the intra-atomic energy and the interaction energies with all the other atoms in the cluster. This revealed that all three polymorphs show stabilising exchange interactions of the donor O atoms. However, only the 𝛼 and 𝛽 forms show strong electrostatic stabilisation of the donor O atoms.</p>

ChemRxiv

A Concise Synthesis of a Methyl Ester 2-Resorcinarene: A chair-conformation macrocycle

Anions are important hydrogen bond acceptors in a range of biological, chemical, environmental and medical molecular recognition processes. These interactions have been exploited for the design and synthesis of ditopic resorcinarenes as the hydrogen bond strength can be tuned through the modification of the substituent at the 2-position. However, many potentially useful compounds, especially those incorporating electron-withdrawing functionalities, have not been prepared due to the challenge of their synthesis: their incorporation slows resorcinarene formation that is accessed by electrophic aromatic substitution. As part of our broader campaign to employ resorcinarenes as selective recognition elements, we need access to these specialized materials, and in this article we report a straightforward synthetic pathway for obtaining a 2-(carboxymethyl)-resorcinarene, and resorcinarene esters in general. We discuss the unusual conformation it adopts, and propose that this arises from the electron-withdrawing nature of the ester substituents that renders them better hydrogen bond acceptors than the phenols, ensuring that each of those acts as a donor only. DFT calculations show that this conformation arises as a consequence of the unusual configurational isomerism of this compound and interruption of the archetypal hydrogen bonding by the ester functionality.

a_concise_synthesis_of_a_methyl_ester_2-resorcinarene:_a_chair-conformation_macrocycle

Introduction<!>Results and Discussion<!>Conclusion:<!>Conflicts of Interest:<!>Supporting Materials:

<p>Resorcinarenes are (usually)-bowl-shaped macrocyclic compounds stabilised by a circular network of intramolecular O•••H−O hydrogen bonds. [1,2] These compounds represent a unique family of host compounds which have been extensively studied in supramolecular host-guest chemistry because they display several sites for non-covalent interactions, excellent pKa tunability, and an electron rich bowl-shaped cavity in the C4v symmetric conformation, among a myriad of other interesting properties. [2][3][4] Their cavity can accommodate a wide range of guest molecules through non-covalent interactions including (but not limited to) hydrogen bonding, halogen bonding, cation•••π, C−H•••π as well as π•••π interactions depending on both the size and charge distribution of the respective guest molecules and the functionalization of the resorcinarene. 4 In addition to their structural role enforcing the upper rim of the macrocycle, the hydroxyl groups at the 1 and 3 positions on the aromatic subunits can participate extensively in hydrogen bonding with hydrogen bond accepting guest molecules. [5][6][7][8][9] As a direct result of these hydrogen bonded supramolecular networks, resorcinarenes have been extensively exploited as appropriate hosts to accommodate a myriad of guests ranging from alcohols, [10][11][12][13][14] to sugars, [15][16][17][18] steroids [19][20][21] and even heterocyclic five-and six-membered ring compounds as guest molecules. [22][23][24][25][26] On resorcinarenes themselves, reaction at C2 is selective over C4 and C6 positions, as these are blocked by the lower rim linkages of the resorcinarene ring. The hydrogen bonded network of hydroxyl groups enhances the acidity of the phenol while increasing π-basicity inside the cavity. [27] Attenuation or cleavage of the O−H bonds, exo to the upper rim, by bases results in increased electron density on the oxygen, effectively strengthening the hydrogen bonding. [26,[28][29][30] Figure 1: A generic resorcinarene illustrating the numbering convention Functionalization of resorcinarenes at the 2-position tunes the relative acidity of the phenolic hydrogens allowing for selective reactions with certain guests. Deprotonation of the phenolic hydrogens with amine bases creates protonated ammonium cations which form interesting supramolecular complexes with the anionic resorcinarenes. These assemblies may have enhanced crystallinity that can then be studied both in the solid state and solution state by single crystal X-ray diffraction and 1 H NMR respectively, as well as in the gas phase by mass spectrometry. The challenge is to access a wide enough variety of resorcinarenes to take advantage of these potential specific interactions. As part of our campaign to access a greater variety of these molecules, we wish to report the synthesis of a simple ester resorcinarene, and its very unresorcinarene like conformation.</p><!><p>A resorcin [4]arene with an ester functionality in the 2-position has not been reported; this moiety would act as an electron-withdrawing functionality that would increase the acidity of the phenols.</p><p>The formation of resorcinarene macrocycles as crystalline solids with high melting points through the acid-catalyzed condensation of resorcinol (or functionalized resorcinols) with aldehydes is well established.[1] Högberg was one of the first to discover the synthesis of resorcinarenes using formaldehyde and resorcinol in acidic conditions. [31] This approach works extremely well for simple 2-haloresorcinarenes and we have found success employing it for other functionalities, so it was the starting point for our synthesis. [32] To obtain 2-substituted resorcinarenes functionalization can take place either before or after cyclisation. As macrocycle formation blocks the 4 and 6 positions the post-cyclisation strategy can have advantages in terms of regioselectivity, although as four functional group transformations must occur in every step, incomplete substitution can lead to complex mixtures, difficult purification and low yields. Pre-cyclization methods, by contrast, introduces regioselectivity issues, but the use of a purified monomer ensures uniform substitution in the macrocycle. In this case we pursued a pre-cyclisation derivatization protocol because of the ready availability of a suitable precursor; the monomer unit was readily obtainable via a slow Fischer esterification of commercially available 2,6-dihydroxybenzoic acid using sulfuric acid in methanol. Following removal of the solvent in vacuo, the residue was dissolved in dichloromethane and washed with saturated sodium bicarbonate, which removed any unreacted starting material along with the sulfuric acid catalyst. Pure methyl 2,6-dihydroxy benzoate was obtained as a pinkish solid in 51% yield (Scheme 1). Scheme 1: Synthesis of the resorcin [4]arene from 2,6-dihydroxybenzoic acid.</p><p>With the functionalized resorcinol in hand, several approaches toward macrocyclization were attempted using isovaleraldehyde, as the tetra isobutyl resorcinarenes are typically highly crystalline in our experience. Initially, an acid catalyzed cyclization using the methodology from Högberg, [31] a 2:2:1 v/v mixture of methanol, water and concentrated hydrochloric acid at reflux over several days, was attempted. However, upon workup a complex mixture was observed that included partial hydrolysis of the methyl ester moieties, and a significant amount of acyclic oligomeric and polymeric material, presumably as a consequence of the high water content in the mixture; none of the tetra ester product could be detected. Attempts to re-esterify this complex mixture resulted in decomposition of the material. This suggests that it might be difficult to access the ester from the known 2-carboxy resorcinarene. [33] As an alternative, we know that the electron-withdrawing ester group decreases the pKa of the phenol groups relative to the unsubstituted homologue making them easy to deprotonate; the resulting phenolates would increase the nucleophilicity of the resorcinol, and any undesired oxygen-centered nucleophilic attack on the aldehyde electrophiles would be non-productive.</p><p>Consequently, we attempted a base-catalyzed approach. Bourgeois has effected the macrocyclization of 2-nitroresorcin [4]arene using sodium hydroxide in water; [34] instead we used a solution of sodium methoxide in methanol so as not to hydrolyze the methyl ester. Upon quenching with acid and filtration, a complex mixture was observed that showed 1 H NMR signals at approximately 6.3 ppm. This suggested that the aldehyde component had undergone an aldol self-condensation as the major reaction. It should be noted that any of the resorcinarenes reported to date that were cyclized under basic conditions were only done so using formaldehyde as the aldehyde component. Lacking any α-protons, enolate formation (and hence aldol reaction) is impossible in these cases, explaining why this complication hasn't been reported, though it has likely been encountered. However, no hydrolysis of the ester was observed. Combining these insights, we were able to effect the desired macrocyclization by employing concentrated sulfuric acid in methanol, providing the desired resorcinarene as a white solid in a poor 1-4.4% yield over repeated trials, with a great majority of the lost mass balance attributed to the formation of polymer and oligomer (Scheme 1). Curiously the NMR was not as we expected and gave us grave concern (Figure 2). Generally, resorcinarenes are, as we have emphasized, found in a C4v symmetric bowlshaped conformation. In this form, the protons on each of the subunits is magnetically equivalent with its congeners on the others. Consequently, one only observes a single aromatic signal, a single benzylic signal, and a single set of peaks for the lower rim alkyl chain. This is not what we found. Instead, our spectrum was consistent with a pair of isomers. We spent a significant amount of time examining this challenge but the apparent mixture behaves as a single compound on TLC, and HPLC, and we could never separate these signals. Seeking clarification on this issue, we attempted to recrystallize, but this also did not change the ratio of the signals or enrich our sample in either compound. However, it did provide us with material of sufficient quality for X-Ray analysis. Crystals suitable for single-crystal X-ray diffraction could be obtained from the white powder by slow evaporation of a chloroform solution (Figure 3). The crystal structure of the obtained compound revealed two unusual features. Firstly, the configuration of the isobutyl groups around the lower rim of the resorcinarene is reminiscent of C2 symmetry (this can be seen in the 2D representation in Figure 4a), in contrast to the more commonly observed C4v isomer. Secondly, in the majority of resorcinarene crystal structures, the observed conformation is the archetypal bowl shape. This crystal instead exhibited a "chair" conformer with a pseudo-C2 rotation axis where two of the resorcinol subunits (2 and 4, see Figure 2 inset for numbering) are coplanar with one another while the other two (1 and 3) sit orthogonal to the plane and antiperiplanar to one another. This result also clearly contextualizes the doubling of the resonances in the NMR spectra: this conformation is not an artifact of crystallization but appears to persist in solution and not rapidly interconvert or "flip" the pseudochair, in which case we would observe a single set of peaks as the average of the two chemical environments. The NMR spectrum can now be understood in terms of this conformational preference, where the two 6H singlets at 4.10 and 3.98 ppm correspond to the methyl esters in two different environments. This is consistent with the pseudochair conformation, where rings 1 & 3 are related by a C2 rotation axis (through the C2/C5 atoms of rings 2 & 4) and 2 & 4 are related by a mirror plane that bisects rings 1 & 3. We are currently developing a model to explain this unusual conformation, and stereoisomeric product, and are also preparing additional electron-poor members that might show similar behaviour. However, we speculate that without the phenols working together to form the hydrogen bond network and template the forming resorcinarene, the typical C4v conformation might not be favoured. To investigate the unusual conformational preference of the resorcinarenes, DFT calculations were performed at the ωB97XD/6-311G(d,p) level of theory in the gas phase and using the polarized continuum solvation model (PCM) to consider solvent effects. Geometric optimizations of the "chair" conformation and a theoretical "bowl" geometry were performed. The initial geometry for the "chair" conformer was obtained from the solid-state molecular structure, whereas the "bowl" conformer was based on solid state molecular structure of known resorcinarenes. The energies and structures of the solvent-corrected optimized conformations are provided in Figure 4a; the optimised structures (as .mol2 files), gas phase energies and all thermodynamic parameters can be found as Supporting Information. These calculations showed that the classic resorcinarene bowl conformation was disfavoured by 20.5 kcal/mol using the solvent correction (25.7 kcal/mol in the gas phase), an enormous preference for the observed conformer. This large preference has two possible contributing factors: first, the ester functional groups are Lewis basic and therefore have the ability to act as hydrogen-bond acceptors to the phenolic hydrogen bond donors when coplanar to the benzene ring. Second, most resorcinarenes are found as the C4v configurational isomer; the steric hinderance of the isobutyl groups in the isomer obtained in this case may also impact the conformational preference of the macrocycle. To investigate the impact of these factors, chair and bowl structures of a C4v resorcinarene were also calculated (Figure 4b). The chair conformer is still preferred in these structures, although the preference is much reduced compared to the C2 isomer; (4.1 kcal/mol using the chloroform solvent correction, 9.8 kcal/mol in the gas phase). This shows that the configuration at the carbons bridging the resorcinol subunits can have a significant effect on the conformational preference of the macrocycle, but is not the most important factor in this case. The upper rim of a resorcinarene bowl is formed by a hydrogen bond network; this is interrupted by the presence of the esters as hydrogen bond acceptors. The esters in the crystal structure are all coplanar with the benzene rings; this maximises the delocalisation of electron density from the electron-rich ring into the carbonyl of the ester, which enhances its Lewis basicity. This likely has a synergistic effect with the hydrogen bond-donor phenols, which will hold the ester coplanar. It should be noted that delocalisation is not the only reason for coplanarity; in each of the computed bowl structures two of the ester groups rotate out of the plane in the absence of this hydrogen bonding. We can therefore conclude that the ester acting as a hydrogen bond acceptor therefore has the most significant effect upon the conformational preference of this macrocycle. Investigations into the generality of this phenomenon are underway in our laboratory.</p><!><p>We have successfully synthesized a novel 2-methyl ester resorcin [4]arene under simple acid catalyzed conditions, if in poor yield, and the structure and solid-state conformation were determined by single crystal X-ray diffraction and NMR spectra. Computational investigations of this system revealed a significant preference for the observed pseudochair conformation and shed light upon the interplay of configurational and hydrogen bonding effects that are in operation in resorcinarene structures. Our studies in this area will further investigate the conformational preference of this and related systems, and methods to rationalise and control this aspect of supramolecular architecture will be developed.</p><!><p>The authors claim no conflicts of interest.</p><!><p>The structure of 2-(methylcarboxyl)resorcin [4]arene has been deposited in the Cambridge Crystal Databank (#2070167). The geometry-optimized structures (as .mol2 files), video surveys of the crystal and calculated structures, including a superimposition of the crystal structure with the minimum energy calculated C2 pseudochair configuration, complete methods and materials are found in the accompanying supporting information.</p><p>CREDIT: Conceptualization, JFT; Funding acquisition, JFT; Investigation, MRR, JJH; Methodology, MRR, JJH, FP; Visualization and crystal analysis, FP; Project administration, JFT; Supervision, JFT, JJH; Writing original draft, MRR; Writing-review and editing, all authors.</p>

ChemRxiv

Kraft Process—Formation of Secoisolariciresinol Structures and Incorporation of Fatty Acids in Kraft Lignin

The complex chemical structure and the fact that many areas in pulping and lignin chemistry still remain unresolved are challenges associated with exploiting lignin. In this study, we address questions regarding the formation and chemical nature of the insoluble residual lignin, the presence of fatty acids in kraft lignin, and the origin of secoisolariciresinol structures. A mild thermal treatment of lignin at maximum kraft-cooking temperatures (∼170 °C) with tall oil fatty acids (TOFA) or in an inert solvent (decane) produced highly insoluble products. However, acetylation of these samples enabled detailed chemical characterization by nuclear magnetic resonance (NMR) spectroscopy. The results show that the secoisolariciresinol (β–β) structure in kraft lignin is formed by rearrangement of the β-aryl ether structure. Furthermore, fatty acids bind covalently to kraft lignin by reacting with the stilbene structures present. It is highly probable that these reactions also occur during kraft pulping, and this phenomenon has an impact on controlling the present kraft pulping process along with the development of new products from kraft lignin.

kraft_process—formation_of_secoisolariciresinol_structures_and_incorporation_of_fatty_acids_in_kraft

Introduction<!>Materials and Reagents<!>Procedure for Lignin Heat Treatment<!>Acetylation of Samples for NMR Analysis and SEC<!>Chemical Characterization by FT-IR Spectroscopy<!>Thermogravimetric Analysis<!>Differential Scanning Calorimetry<!>Detailed Characterization of the Chemical Structure by NMR Spectroscopy<!>Pyrolysis-Gas Chromatography/Mass Spectrometry<!>Analysis of Molar Masses by SEC<!>Results and Discussion<!>Structural Characterization of Lignin Before and After Heating in Decane (Lignin–Decane) or TOFA (Lignin–TOFA) by NMR Spectroscopy<!><!>Structural Characterization of Lignin Before and After Heating in Decane (Lignin–Decane) or TOFA (Lignin–TOFA) by NMR Spectroscopy<!>Chemical Characterization and Compositional Evaluation of Lignin–Decane and Lignin–TOFA by FT-IR Spectroscopy after Lignin Heat Treatment<!><!>Chemical Characterization and Compositional Evaluation of Lignin–Decane and Lignin–TOFA by FT-IR Spectroscopy after Lignin Heat Treatment<!>Thermal Characterization of Heat-Treated Lignin–TOFA by TGA and DSC<!><!>Thermal Characterization of Heat-Treated Lignin–TOFA by TGA and DSC<!><!>Thermal Characterization of Heat-Treated Lignin–TOFA by TGA and DSC<!>Pyrolysis-Gas Chromatography/Mass Spectrometry<!><!>Pyrolysis-Gas Chromatography/Mass Spectrometry<!>Analysis of Molar Masses of Starting Lignin and Lignin–TOFA<!>Impact of Results on the Kraft Pulping Process and Development of New Products by Heat Treatment<!>

<p>Material production from sustainable resources is an ongoing global challenge. Kraft pulping is currently the main processing technique to produce renewable cellulosic fibers used in various products.1,2 During the kraft pulping process, the most abundant linkage of lignin, namely, the β-O-4 linkage (45–50% in softwood),1,2 is mainly cleaved to form soluble products in the presence of hydroxyl and hydrogen sulfide anions. However, part of the β-O-4 linkages, 3–7% according to quantitative NMR experiments,3,4 remain intact during the process. The main side product formed from the kraft process, that is, sulfur-containing kraft lignin, is still usually burned as energy but could potentially be used to replace many applications based on scarce oil sources.</p><p>Kraft pulping delignification can be divided into three stages, namely initial, bulk, and residual delignification.1 After the fast bulk delignification stage, about 90–95% of the original lignin has been dissolved, depending on the processing conditions.1,5 The rate of delignification is much slower at the final residual delignification stage. Compared to the conventional kraft pulping process, processing parameters have been optimized, for example, by lowering the processing temperatures from 170 °C, and by tuning the concentrations of hydroxyl and hydrogen sulfide ions for more efficient delignification (i.e., extended cooking).5</p><p>Although the kraft lignin process is currently highly efficient, all aspects are not totally understood. Examples include the residual lignin in kraft pulping, which is highly insoluble and difficult to remove, and the occurrence of fatty acids in kraft lignin. From earlier studies, residual lignin is known to contain β-O-4 structures (3–7%) and "condensed aromatic structures".3,4,6−8 According to the present view, the aromatic condensed structures are accumulated, rather than formed by chemical reactions during the process.8 Secoisolariciresinol is also one of the structures found in residual kraft lignin (2%), in addition to resinol (i.e., β–β, 2%) and phenyl coumaran (i.e., β-5, 5%).9 Processing conditions (conventional or extended cooking) affect the structure of residual lignin, in terms of the amount of phenolic groups, β-O-4 linkages, and condensed lignin units.10 Lignin–carbohydrate complexes have also been suggested to play some role in the more difficult delignification of residual lignin.3,11,12 Earlier studies by Gellerstedt et al. (1987, 2004) suggested that residual lignin contains also material resulting from the formation of aliphatic C–C bonds because corresponding signals appear in the NMR spectra of kraft lignins and insoluble material that remains in kraft lignin cooked with 2,6-xylenol after thioacidolysis.9,13 Furthermore, the involvement of radical reactions during the formation of residual lignin has been suggested, as fatty acids are incorporated into lignin during kraft pulping.9</p><p>The resinol structure is the major type β–β structure of lignin, whereas the secoisolariciresinol structure presents another, a minor type β–β structure. The secoisolariciresinol structure is probably not, as initially believed, formed through the resinol structure, that is, the more typical β–β structure in lignin.14,15 The amount of secoisolariciresinol structure in kraft lignin is 3%.4 Secoisolariciresinol is also one of the structures found in residual kraft lignin (2%).9 However, comparison of lignins in the same study showed that the amount of secoisolariciresinol structure, along with fatty acids, is higher in kraft and residual lignins as compared to milled wood lignin.8</p><p>In addition to the reactions occurring during kraft pulping, understanding the thermal reactions of lignins is important in the development of new processes based on pyrolysis and milder heat treatments.16 For example, kraft lignin has been suggested to be suitable for preparing pyrolysis oils or as a starting material for carbon fibers.6,17,18 Torrefaction is a milder thermal pretreatment for biomass taking place at 200–300, °C, which aims for improved fuel production or consumption.19 In all of these processes, lignin is thermally treated and information on the reactions involved is highly beneficial for their efficient development.</p><p>In order to investigate mild thermal heat treatments of lignin, the reactions were performed in tall oil fatty acids (TOFA) or in an inert solvent decane. Our initial aim was to study the reactions of kraft lignin and TOFA during mild thermal treatment at moderate temperatures (∼170 °C) to assess the incorporation of fatty acids into kraft lignin. The kraft lignin used was from the Lignoboost process, where lignin is precipitated using carbon dioxide (CO2) as the acid.20 TOFAs, obtained by distillation, were used in the reaction as a model compound representing the fatty acids released during the kraft pulping process. TOFA was also used as a reaction medium for heat transfer. The use of alkaline water solution containing hydrogen sulfide as the reaction medium was omitted, because the aim was to investigate radical reactions instead of ionic reactions. The material formed as a result of the heat treatment of lignin and fatty acids was hard, but very brittle, and furthermore dissolved very poorly in all of the tested organic solvents. According to initial chemical analysis, the material was composed of lignin and fatty acids, suggesting that the starting materials were attached through covalent bonding (it was impossible to "extract" any of the components with any solvents or their combinations). To understand the compositions of the materials and chemical reactions involved, a set of further experiments and analyses was performed. To define lignin–lignin reactions, a similar mild heat treatment was performed using the chemically inert decane instead of TOFA as the solvent. Chemical and thermal properties of the formed products were characterized thoroughly using various complementary techniques.</p><!><p>Kraft softwood lignin (Metso, LignoBoost lignin), received in the powdered form, was dried further in an air-circulating oven at 40 °C before use. Distilled TOFA (product name: FOR2) was received from Forchem (Rauma, Finland), and typically comprises of certain rosin acids (total 2%), and the following fatty acids (total 96%) as the main components: linoleic acid (18:2, 42%), oleic acid (18:1, 32%), and pinolenic acid (18:3, 7%). Decane was purchased from Sigma-Aldrich (St. Louis, MO, USA). The solvent used for NMR (d6-dimethyl sulfoxide, DMSO-d6) was purchased from Eurisotop (Saint-Aubin, France).</p><!><p>To obtain solutions with 5 wt % of lignin with a similar level of volume, lignin (5.0 g in TOFA and 4.0 g in decane) and TOFA (95 g, d = 0.91 g mL–1) or decane (100 mL, d = 0.73 g mL–1) were placed in an inner PTFE cup of a 450 mL stainless-steel Parr reactor (Parr Instrument Company, Moline, IL, USA) equipped with a heating mantle, temperature control, and a pressure monitor (controller model 4843). The reactor was flushed, to remove air, by filling and emptying it three times with N2 (100 psi, 0.69 MPa). After flushing, N2 pressure was set at 70 psi (0.48 MPa), which was maintained during heating by releasing N2 through a valve. The reaction mixture was heated to 170 °C and maintained at that set temperature for 2 h. The product started to form already after 15 min, but to confirm that all lignin reacted, and for the purpose of comparison, a reaction time of 2 h was used. The monitored heating temperature was approximately 170–180 °C in decane and 170–195 °C in TOFA. After cooling, the floating reaction product (lignin–decane or lignin–TOFA) was collected from the surface of the liquid, washed with acetone, and dried in an air-circulation oven at 40 °C. Part of the product could not be easily removed from the temperature sensor of the Parr reactor, and therefore reliable mass balance calculations could not be performed.</p><!><p>Starting lignin or the heat-treated sample (100–150 mg) was placed in a round bottom flask. Pyridine (2 mL) and acetic anhydride (2 mL) were added. The mixture was allowed to react at room temperature without stirring for 20 h, which resulted in dissolution of the starting lignin. The polymerized samples were further refluxed at 80 °C with stirring for 7 h, resulting in mainly soluble products. The acetic anhydride was quenched by adding and evaporating ethanol twice (20 mL), the trace of pyridine was evaporated with toluene (four times, 20 mL), and finally the trace of toluene was evaporated with chloroform (twice, 20 mL).</p><!><p>The FT-IR spectra of the starting lignin, TOFA, lignin–decane, and lignin–TOFA were recorded with a Bruker ALPHA (Bruker Corporation, Billerica, MA, USA) attenuated total reflectance (ATR)-FT-IR spectrometer in a spectral range of 375–4000 cm–1.</p><!><p>Thermal decomposition was analyzed by thermogravimetric analysis (TGA) using a Mettler Toledo TGA/SDTA 851e (Mettler-Toledo International Inc.). The samples (5–10 mg) were analyzed in alumina crucibles with pinhole lids. A heating rate of 10 °C min–1 and an N2 flow rate of 50 mL min–1 were used.</p><!><p>Differential scanning calorimetry (DSC) measurements were performed using a DSC Q200 (TA Instruments, NJ, USA). The samples (5–10 mg) were analyzed in aluminum pans with lids. A heat–cool–heat cycle was used with a heating rate of 10 °C min–1 and a cooling rate of 5 °C min–1, and the measurements were performed under nitrogen. The maximum temperature used was below the degradation temperature of the material to be analyzed. A second heating stage was used to analyze the data of samples containing lignin.</p><!><p>The starting lignin, lignin–decane, and lignin–TOFA were characterized using NMR spectroscopy. All samples were analyzed in DMSO-d6 as the solvent. The thermally treated acetylated samples were mainly soluble in DMSO-d6 (1.6 mL). After filtration through a cotton wool plug using a Pasteur pipette, the sample was concentrated under vacuum to a volume of 0.7 mL.</p><p>The 2D 1H–13C heteronuclear single-quantum coherence (HSQC) NMR spectrum of the starting lignin was acquired using a Bruker 600 MHz spectrometer, and the 2D HSQC NMR spectra of lignin–decane and lignin–TOFA using a Varian Unity Inova 600 MHz spectrometer. The temperature used during measurement was 25 °C for the starting lignin and 40 °C for heat-treated lignins. The pulse sequence used was hsqcetgpsisp2.2 (Bruker) or gHSQC (Varian). The 90° pulse width was determined separately for all samples. The number of increments used in the measurements was 256 (Varian) or 300 (Bruker), and the number of scans was 64 (Varian) or 512 (Bruker).</p><p>The TOFA used for the reaction was analyzed by 1H NMR before and after heat treatment. Spectra are presented in the Supporting Information. The fatty acid contents of the starting lignin and lignin–TOFA were evaluated from the 1H NMR spectra (Supporting Information) by comparing the integrals of fatty acid Fω (CH3, ∼0.9 ppm) and lignin Ar–OCH3 (∼3.8 ppm, including overlapping signals from the side-chain structures of lignin).</p><!><p>Measurements were performed using a Pyrolab2000 pyrolyzer connected to a Bruker Scion SQ 456 GC–MS. The pyrolysis chamber temperature was kept at 150 °C and the samples were pyrolyzed by heating the platinum filament in 8 ms to 200 °C and keeping the temperature elevated for 2 s before rapid cooling of the filament. The second heating to 580 °C was also 2 s long. Helium was used as a carrier gas at a flow rate of 1 mL min–1. Injector temperature was 250 °C and a 1:20 split ratio was used. Pyrolysis products were separated in an Agilent DB-5MS UI [(5%-phenyl)-methylpolysiloxane, 30 m × 0.250 mm × 0.25 μm film] capillary column. Column oven temperature was kept at 50 °C for 2 min after which the temperature was increased at a rate of 10 °C min–1 to 280 °C and kept at 280 °C for 5 min resulting in a 30 min overall run time. Ion source temperature was 250 °C with an electron ionization of 70 eV. The MS scan range was 40–400 m/z. Compounds were identified by comparing them to the National Institute of Standards and Technology (NIST) database.</p><!><p>Molar masses of acetylated lignin–TOFA and starting lignin acetylated in the presence of TOFA (roughly 50 wt %) were analyzed using SEC. The Waters Acquity APC system was used for the analysis (Waters Corporation, Milford, MA, USA), equipped with Acquity APC XT columns 45, 200, and 450 Å, and using 30 °C as the column oven temperature. Sample concentration was 1 mg mL–1 in tetrahydrofuran (THF). Samples were filtered through a 0.2 μm syringe filter (GHP Acrodisc 13 mm, Pall Corp., Ann Arbor, MI, USA) before injection (50 μL). Total run time was 12 min, with a flow rate of 0.8 mL min–1. Molar masses were evaluated using data from the UV range (254 nm), and the following polystyrene standards were used for calibration: 474, 840, 2500, 5010, 13,300, 32,300, 43,400, 76,000, and 139,400 Da (Polystyrene Calibration Kit, Scientific Polymer Products, Inc.) along with 321,000 and 526,000 Da (Fluka).</p><!><p>The starting kraft lignin was heat-treated in decane or TOFA to discriminate the internal reactions of lignin from cross-reactions of lignin with TOFA. The treatments were performed at ca. 175 °C, which is the maximum temperature during the conventional kraft pulping process.1 The heat treatment in decane provided information on thermal reactions of kraft lignin alone in the selected temperature, whereas the participation of fatty acids in reactions was evaluated in TOFA. The first difference between treatments was observed when monitoring the reaction temperature: the temperature set at 170 °C remained fairly constant in decane (170–180 °C), whereas it increased spontaneously (170–195 °C) in TOFA, indicating that exothermic reactions were occurring in the reaction vessel.</p><!><p>To gain insights into the reactions occurring under thermal treatments, the chemical structures of the starting and heat-treated lignins were analyzed using nuclear magnetic resonance (NMR) spectroscopy. The samples were acetylated because the very low solubility of nonacetylated samples prevented analysis. The underivatized products were not soluble in many of the typical organic solvents tested, and only slightly soluble in aqueous NaOH-solutions. The main part of the acetylated products was soluble in the NMR solvent used (DMSO-d6), and it therefore should be kept in mind that the minor insoluble fraction may have some structural differences compared to the observed results.</p><p>In practice, all of the lignin was floating on the surface of the liquid used for reaction media after the reaction in TOFA or decane. However, part of the material was tightly attached to the temperature sensor of the reaction vessel jeopardizing determination of accurate mass fractions. To confirm that lignin was not dissolved during reaction, the used TOFA was analyzed before and after the reaction by 1H NMR analysis (Supporting Information). According to visual inspection, decane did not contain lignin-based materials.</p><p>The 2D HSQC NMR spectrum of the starting lignin and the structures identified in NMR spectra are presented in Figure 1. The 2D HSQC NMR spectrum of lignin–decane, with expansion of the aromatic area, is presented in Figure 2 and the spectrum of lignin–TOFA, with expansion of the aliphatic area, is presented in Figure 3. The NMR assignments were based on the previously published data: acetylated lignin samples,21,22 structures of acetylated secoisolariciresinol and non-acetylated dihydrocinnamyl alcohol,14,23 stilbene structures formed from phenylcoumaran (β-5) and spirodienone structures (β-1),24 and fatty acids of various chain lengths and degrees of unsaturation.25 It is worth noting that some of the published data are for nonacetylated samples and, therefore, some of the values may differ compared to acetylated samples used here. Furthermore, there might be some variation as a result of different solvents used for NMR analysis. Lists of identified cross peaks are presented in Table 1 for fatty acid-derived structures and in Table 2 for lignin-derived structures. From the cross-coupling patterns, the starting lignin (Figure 1) was composed mainly of β-aryl ether type (A) and resinol type (B) structures. The dihydrocinnamyl alcohol structure (D) was also present along with stilbene-type structures (S1 and S5). Some signals originating from fatty acids (F) were detected and those have also been observed earlier in samples of kraft lignin.8,9,24 As a rough estimation based on the 1H NMR spectrum, the fatty acid content of starting lignin was around 4 mol % (Supporting Information).</p><!><p>2D HSQC NMR spectrum of starting lignin (acetylated) and identified chemical structures.</p><p>2D HSQC NMR spectrum of lignin heated in decane (top), and expansion of the aromatic area (bottom).</p><p>2D HSQC NMR spectrum of lignin heated in TOFA (top), and expansion of the aliphatic area.</p><!><p>Based on the NMR spectra of both lignin–decane (Figure 2) and lignin–TOFA (Figure 3), all the β-aryl ether type (A) linkages disappeared during heat treatment and secoisolariciresinol structures (C) were formed. The stilbene structures (S1 and S5) of lignin were unreactive in the inert solvent decane, but new, unknown signals appeared in the same area (125–130 ppm), indicating the possible formation of new unsaturated structures (Figure 2, expansion). The signals originating from stilbenes (S1 and S5) disappeared also in lignin–TOFA case, while the intensities and number of signals representing fatty acids (F) clearly increased (Figure 3, and the expansion). According to an estimation based on the 1H NMR spectrum, the fatty acid content of lignin–TOFA was around 7 mol %, almost twice as much compared to starting lignin (Supporting Information).</p><p>According to the results from the NMR analysis presented above, two sites seem reactive during heat treatment, also depending on the reaction media. In decane, β-aryl ether structures (A) are reacting but stilbenes (S1 and S5) are not, whereas both the β-aryl ether structures (A) and stilbenes (S1 and S5) react in TOFA. The formation of secoisolariciresinol (C) seems to be associated with reactions of β-aryl ether (A) because secoisolariciresinol (C) is formed and β-aryl ether (A) is reacting in both heat treatments, that is, decane and TOFA. The study by Zhang et al. 2003 concluded that secoisolariciresinol (C) is not formed from resinol (B) but instead is somehow connected to the β-aryl ether structure.14 Our findings support the earlier conclusions, but instead of a separate pathway in lignin biosynthesis, secoisolairciresinol (C) may form directly from β-aryl ether (A).</p><p>Stilbenes (S1 and S5) also react during heat treatment in TOFA, whereas these structures are stable in decane. Stilbenes and fatty acids, therefore, possibly form certain adducts or condensation products, although the site of connection could not be identified from the NMR spectrum of lignin–TOFA. The reaction pathway must be clearly more complex as more reacting components are present. The very obvious hypothetical product between lignin and TOFA, formation of an ester in a reaction with fatty acid and lignin could not be observed neither, in alignment with previous NMR work with kraft lignin, which contained fatty acids.24 Apparently, another experimental setup with some simpler model compounds is required for the identification of the formed products, which would be beyond the scope of the experiments presented in this paper.</p><!><p>To further support the NMR analysis findings and to evaluate the compositions of the materials, we analyzed starting lignin, TOFA, and lignin heated in TOFA (lignin–TOFA) or decane (lignin–decane) using FT-IR (Figure 4). A list of the assigned absorption bands is presented in the Supporting Information. The FT-IR assignments for lignin are based on the previously published results for lignin.26 The FT-IR assignments for TOFA, a mixture of fatty acids produced as side products of the forest industry, are based on earlier studies on vegetable oils.27,28</p><!><p>FT-IR transmission spectra of starting lignin and TOFA, and heat-treated lignin–decane and lignin–TOFA. Spectral range 500–4000 cm–1is shown. The spectra were obtained by means of ATR mode.</p><!><p>The lignin samples were measured in earlier studies with KBr,26,29,30 and the values measured in this work, using FT-IR with ATR accessory with no need for separate sample preparation, may therefore vary slightly. For example, the spectrum of starting lignin has a weak absorption band at 1361 cm–1, which can be assigned to the aliphatic C–H stretch in methyl (not in −OMe) and to the phenolic hydroxyl group. Although the value slightly differs from the literature (1365–1370 cm–1) the position and shape of the absorption band are comparable.26 Similarly, the absorption band of lignin at a value of 1208 cm–1 may be assigned arising from C–C plus C–O plus C=O stretching (G condensed > G etherified), although the literature value is 1221–1230 cm–1).26</p><p>As seen in Figure 4, both heat-treated samples (lignin–decane and lignin–TOFA) have similar broad absorption bands for −OH stretching as the starting lignin (3500–3000 cm–1). Similarly, all the typical absorption bands found in the starting lignin, from absorption band 1703 cm–1 (C=O group) to 816 cm–1 (C–H out-of-plane in positions 2, 5, and 6 of G units), are also found in the heat-treated samples. These results also confirm that the material after heat treatment comprises of lignin, rather than totally charred components.</p><p>In Figure 4, the absorption band of the C=O group for lignin-decane (1705 cm–1) has a similar strength compared to starting lignin, whereas the strength of the C=O absorption band (1708 cm–1) for lignin–TOFA is similar compared to TOFA. Furthermore, the absorption band for C–H stretching in methine (3006 cm–1) is present in the spectrum of the lignin–TOFA sample, providing further proof that lignin–TOFA contains a notable amount of additional fatty acids.</p><p>For both heat-treated samples, the absorption bands at the area corresponding to either methyl or methylene C–H stretching (2960–2853 cm–1) were more intense compared to starting lignin, which would also be compatible with the formation of the secoisolariciresinol (C) structure with a relatively increasing amount of −CH2–.</p><!><p>To answer the question raised during the chemical analysis, that is, whether TOFA adsorbed into the lignin material during heat treatment in a way that would prevent washing of the unreacted fatty acids, thermal gravimetric analyses were performed on starting lignin, TOFA, and lignin–TOFA (Figure 5). Both samples containing lignin were clearly degraded and evaporated in a similar manner, and approximately 40 wt % of the material remained after heating to 800 °C. The mass loss of TOFA occurred at clearly lower temperatures, and less than 5 wt % of the material remained after heating to 400 °C.</p><!><p>Thermogravimetric analysis of starting lignin, heat-treated lignin–TOFA, and TOFA.</p><!><p>Thermal properties of starting lignin, TOFA, and heat-treated lignin–TOFA were also determined using DSC, presented for lignin–TOFA in Figure 6, and for starting lignin and TOFA in Supporting Information. According to our results, the glass transition temperature (Tg) of the starting lignin was 156 °C, which is in the same range with previously reported values for kraft lignin (120–174 °C).31,32 For TOFA, transitions for crystallization and melting were observed at temperatures of −23 °C (Tc) and −19 °C (Tm), respectively. These values are reasonable compared to the DSC analysis of fatty acids in the literature, considering that the aim of the method used in this study was to compare the materials containing lignin and TOFA.33 The observed Tg was 142 °C for lignin–TOFA (Figure 6), which was slightly lower than that of the starting lignin. Another melting transition (Tm) for lignin–TOFA was observed at −8 °C, which was in a similar range compared to transitions observed for TOFA.</p><!><p>DSC curve of lignin–TOFA. For clarity, only the areas representing thermal changes are shown (the upper graph for Tm and lower for Tg).</p><!><p>The results from the thermal analyses also suggest that lignin and TOFA were connected by a covalent bond, because no separate evaporation of fatty acid was observed in the TGA of heat-treated lignin–TOFA, while this material contains some fatty acids according to chemical analyses (∼7 mol % by comparing 1H NMR integrals of lignin Ar–OCH3 and fatty acid Fω). Furthermore, the glass transition temperature of lignin–TOFA was slightly decreased compared to starting lignin and an additional melting transition was also observed.</p><!><p>Pyrolysis-gas chromatography/mass spectrometry (Pyr-GC/MS) was used to identify volatiles in order to understand the underlying mechanisms occurring during heat treatments. The Pyr-GC/MS analyses were performed by first heating the samples to 200 °C, mimicking the heat treatment performed earlier, and then bringing the temperature up to 580 °C. The results of the main peaks, each representing more than 3% peak area of the total amount of all peaks (100%), except for silylated compounds most probably originating from other sources, are shown in Table 3. All results are presented in the Supporting Information.</p><!><p>First pyrolysis was performed at 200 °C and the second one at 580 °C. Results are presented as peak areas (%) of the total area of all peaks. Results for peaks more than 3% peak area are presented here; all results are found in the Supporting Information.</p><!><p>Pyrolysis at higher temperature (580 °C) produced a typical fragmentation pattern of kraft softwood lignin, and no significant difference was observed between the samples, lignin, and lignin–TOFA mixture. However, only two products, that is, dimethyl disulfide and guaiacol, were formed from the starting lignin at lower pyrolysis temperature (200 °C). In the lignin–TOFA mixture, vanillin was the main lignin-based compound released followed by guaiacol and dimethyl disulfide, and fatty acid-based 2,4-decadienal, 2-decen-1-ol, and hexanal were concurrently released.</p><p>The products formed from the pyrolysis of starting lignin were consistent compared to previous studies in both applied temperatures.6,34,35 The significant release of guaiacol at lower pyrolysis temperatures (200–300 °C) has also been observed earlier.6,35 Based on the structural analysis by the NMR presented above, the release of guaiacol also supports the hypothesis that secoisolariciresinol (C) is formed from β-aryl ether (A). Secoisolariciresinol (C) has previously been suggested to form during the biosynthesis of lignin from two coniferyl alcohol radicals and this structure would then be attached to the β-aryl ether structure (A).14 Previously, (C) has also been observed in DFRC (derivatization followed by reductive cleavage) analysis of lignin dimers, resulting from the DFRC procedure, in which β-aryl ether bonds are cleaved.15 Our results suggest that the Cβ–O bond of β-aryl ether structure is homolytically cleaved at mildly elevated temperatures (ca. 170–200 °C) or in other suitable conditions for prompting radical reactions, and the resulting Cβ-radicals couple to form the β–β bond (Scheme 1). The formation of the β–β bond would thus be similar to the suggested coupling of two coniferyl alcohol radicals.14 The formation of the secoisolariciresinol (C) structure then takes place after release of H2O2, which can react with either lignin or TOFA. The mechanism for the release H2O2 is not clear at this point, but the required reducing component could be present in a reaction media containing lignin, fatty acids, and sulfur compounds. Therefore, (C) formation does probably not occur during lignin biosynthesis from coniferyl alcohol, as suggested previously,14,36 but instead, the rearrangement of β-aryl ether, prompted by radical formation, may create a more chemically resistant, and possibly more insoluble structure.</p><p>The presence of TOFA clearly affected pyrolysis at 200 °C, but evaluation of plausible reaction pathways involved is more complex because of the higher number of reacting components. The similar formation of secoisolariciresinol (C), as suggested above, would be reasonable, because guaiacol was also released in the presence of TOFA. However, vanillin was the major lignin-based component released, suggesting cleavage of the Cα–Cβ bond. Results from the NMR analysis showed that stilbenes were concurrently consumed (disappearing). The formation of aldehydes from stilbenes is a known reaction under certain conditions, but based on the results presented here, it seems to require the presence of fatty acids, which also attach to lignin. The reaction pathway leading to the covalent bonding of TOFA under heat treatment is thus not evident from the present results, and requires further investigation.</p><!><p>The molar masses of acetylated starting lignin and heat-treated lignin–TOFA were evaluated by SEC. The lignin sample was acetylated in the presence of TOFA for improved comparison of the samples (the SEC chromatograms can be found in Supporting Information). By visual inspection, the acetylated samples seemed to be soluble in the used solvent (THF). According to the results, presented in Table 4 for the signal with the largest area (i.e., the first signal of the chromatogram, see Supporting Information), the molar masses before and after heat treatment were very similar, and in fact, the molar mass of lignin–TOFA was slightly lower. Based on the low solubility of the product, this initially seemed surprising. However, this is well in line with the reaction pathway suggested for the formation of a secoisolariciresinol structure (C). This result is also consistent with the results of Balakshin et al. (2003), in that aromatic condensation products, that is, formed via coupling to the 5-position, are not produced during kraft pulping.8 Instead, cleavage of the C–O bond and rearrangement of β-aryl ether structure and formation of a new aliphatic C–C bond would result in a more hydrophobic and chemically resistant structure (Scheme 1). Notably, comparison of empirical formulas of two β-radicals formed from β-aryl ether (2 × C10H12O4H = C20H24O8H2) and the secoisolariciresinol dimer (C20H22O6H2) show that the relative reduction (−9%) in molar masses (from 394 to 360) is very similar compared to the molar masses acquired from SEC.</p><!><p>Cleavage of the β-O-4 linkage by the action of hydroxyl and hydrogen sulfide ions is the major ionic reaction pathway during kraft delignification. In addition, according to the results presented here, minor radical reaction pathways are involved in the kraft process. The existence of fatty acids and secoisolariciresinol structures in the kraft and residual lignins is known from earlier studies, and a radical pathway leading to these products has been proposed previously.9 However, the structural patterns of lignin taking part in these radical reactions is novel information, in addition to the reaction mechanism for the formation of secoisolariciresinol structure.</p><p>In this study, we have shown that temperature of 170–175 °C is one of the essential reaction conditions that induce the formation of these products through a radical pathway. Considering this temperature with the fact that delignification improves by lowering the temperature of conventional kraft pulping process, the formation of these products could be one of the reasons for the slow residual delignification phase. Unless derivatized by acetylation using harsh conditions, the products are highly insoluble in any common solvents, and could, therefore, form a protective layer, which is not easily accessible by the pulping chemicals anymore. Furthermore, the formed C–C bonds are more stable, less reactive, and more hydrophobic compared to the functional structures reacting in the starting materials. Understanding all underlying mechanisms of the kraft delignification process, even the minor ones, is useful for improved control of the whole process.</p><p>In addition to kraft pulping process, it seems that radical reactions of lignin are also most probably taking place during all heat treatment processes of lignin, and knowledge of the possible reaction pathways is useful in developing possible applications for lignin. However, further studies, in addition to the kraft lignin, are required to evaluate if the rearrangement of β-O-4 linkage to secoisolarisiresinol is taking place with other types of lignins as well.</p><p>Concluding the results presented here, β-aryl ether structures of lignin can rearrange to form secoisolariciresinol structures during mild thermal heat treatment. In addition, the formed structure with new C–C bonds and less hydroxyl groups is chemically more stable and at least partially responsible for the observed lower solubility of the formed product. On the other hand, if fatty acids are present, the stilbene structures of lignin react and form covalent bonds with fatty acids. Whereas the rearrangement of β-aryl ethers to secoisolariciresinol is likely to occur through radical reactions, the reaction pathway with fatty acids remains unclear. Finally, the knowledge of the reacting sites under mild thermal treatment of kraft lignin is valuable in the development of new applications based on heat treatment, along with controlling and optimizing the existing kraft pulping process.</p><!><p>List of selected FT-IR assignments; Pyr-GC/MS results, pyrograms, and pictures of materials after heat treatment; 1H NMR spectra of acetylated starting lignin and acetylated mixture of lignin and TOFA without heat treatment, TOFA before and after heat treatment, and acetylated lignin–TOFA after heat treatment; DSC curves of lignin and TOFA; and SEC curves (PDF)</p><p>jf1c00705_si_001.pdf</p><p>The authors declare no competing financial interest.</p>

PubMed Open Access

Microfluidic System for In-Flow Reversible Photoswitching of Near-Infrared Fluorescent Proteins

We have developed a microfluidic flow cytometry system to screen reversibly photoswitchable fluorescent proteins for contrast and stability of reversible photoconversion between high- and low-fluorescent states. A two-color array of 20 excitation and deactivation beams generated with diffractive optics was combined with a serpentine microfluidic channel geometry designed to provide five cycles of photoswitching with real-time calculation of photoconversion fluorescence contrast. The characteristics of photoswitching in-flow as a function of excitation and deactivation beam fluence, flow speed, and protein concentration were studied with droplets of the bacterial phytochrome from Deinococcus radiodurans (DrBphP), which is weakly fluorescent in the near-infrared (NIR) spectral range. In agreement with measurements on stationary droplets and HeLa S3 mammalian cells expressing DrBphP, optimized operation of the flow system provided up to 50% photoconversion contrast in-flow at a droplet rate of few hertz and a coefficient of variation (CV) of up to 2% over 10 000 events. The methods for calibrating the brightness and photoswitching measurements in microfluidic flow established here provide a basis for screening of cell-based libraries of reversibly switchable NIR fluorescent proteins.

microfluidic_system_for_in-flow_reversible_photoswitching_of_near-infrared_fluorescent_proteins

<!>Protein Expression, Purification, and Characterization.<!>Protein Expression in Mammalian Cells.<!>Single-Point Photoswitching Measurements.<!>Design of the Microfluidic Chip.<!>Design of the Optical System.<!>Characterization of DrBphP-PCM Photoswitching in Ensemble.<!>Off-Chip Measurements of DrBphP-PCM Photoswitching.<!>Optical and Microfluidic System Design.<!>Constitutively Fluorescent NIR FP.<!>Reversibly Switchable NIR FP Variant.<!>DISCUSSION<!>CONCLUSION

<p>Since its invention in the 1990s, fluorescence microscopy-based super-resolution imaging techniques1–4 have been developed by many groups and have found their place in numerous applications. These advances have been reviewed recently.5–7 Regardless of the particular technique (pointwise scanning with shaped beams in STED/GSD1,2,8 or stochastic excitation/emission under full-field illumination in STORM/PALM9,10), characteristics of the fluorescent probes critically influence imaging performance.7,11–13 The importance of the photophysical factors for required fluorescent probes were first recognized in a generalization of STED – scanning GSD microscopy2 and then in reversible saturable optical fluorescence transitions (RESOLFT) imaging.8</p><p>In contrast to STED and GSD, RESOLFT implements reversibly photoswitchable fluorescent proteins (rsFPs) as imaging probes.14–17 The photoswitching of rsFPs between high (ON) and low (OFF) fluorescent states permits a >100 000-fold decrease in switching the OFF intensity (in contrast to depletion in STED), compared to conventional STED/GSD techniques, employing constitutively fluorescent probes. Important characteristics that determine the successful use of rsFPs in RESOLFT include their photoswitching contrast, photoswitching half-times, and photoswitching fatigue resistance (Figure 1). The photoswitching contrast (defined as the ratio of fluorescence intensities, ION/IOFF) determines the overall performance of an rsFP. Ideally, rsFPs should have switching kinetics suitable for both high-speed and low-intensity imaging, as the latter case requires longer exposures, thus increasing pixel dwell time. Finally, resistance to photoswitching fatigue permits execution of a large number of switching cycles, thus improving the signal-to-noise ratio of an image.</p><p>All currently available rsFPs belong to the green fluorescent protein (GFP)-like16,17 family of proteins. They have been used in wide-field diffraction-limited intracellular photolabeling and tracking, as well as in super-resolution RESOLFT microscopy. However, their application to deep-tissue imaging are limited, because none of them have excitation and emission maxima in the near-infrared (NIR) tissue transparency window (~650–900 nm).18</p><p>Recently, several constitutively fluorescent NIR FPs, (i.e., iRFPs), were developed from bacterial phytochromes (BphPs).16,19 The BphP-derived family of iRFPs employs the mammalian-cell-abundant linear tetrapyrrole biliverdin (BV), as a chromophore. BphPs contain a photosensory core module (PCM), which is comprised by PAS, GAF, and PHY domains, and an output effector module (Figure 1C).20,21 The BV chromophore is covalently bound to a cysteine residue at the N-terminus of the PAS domain and located within a pocket of the GAF domain. The PCM module is mainly responsible for the photophysical properties of BphPs, which can exist in two stable interconvertible spectral states: far-red absorbing (Pr) and NIR absorbing (Pfr), associated with cis and trans photoisomerization of BV, respectively (Figure 1D). The PCM is the minimal BphP fragment that can efficiently and reversibly photoconvert between Pr and Pfr states, but NIR rsFPs based on PCMs are not yet available, despite the high potential for this class of fluorophores. Engineering of the PCM of AtBphP2 from Agrobacterium tumefaciens to photoactivatable proteins, PAiRFPs, which photoconvert only once, represents a first step in this direction.22 PAiRFPs have been selectively photoactivated within small regions of tissue in mice, but several drawbacks impede their further use in super-resolution imaging.22</p><p>To date, the properties of GFP-like rsFP have been optimized by multiple rounds of screening of large cell-based libraries. There are three major techniques available for high-throughput screening and sorting of cells: fluorescence-activated cell sorters (FACSs), robotic microtiter assay, and custom microfluidic-based systems. Manipulation of switching contrast and photostability requires complex approaches where excitation wavelength, irradiance, pulse sequence, and exposure times are tailored to properly extract the correct photophysical parameters.23–25 For example, screening for photoswitching requires a dedicated set of excitation beams to realize at least one cycle of in-flow photoconversion, whereas screening for low fatigue requires additional beams and delay lines for tracking fluorescence changes under controllable photobleaching. Because of the relatively slow photoconversion kinetics of rsFPs and BphPs,22,26,27 a time window from tens to hundreds of milliseconds is required to complete all measurements, which then must be analyzed for each cell in real time while it moves in the flow path. These factors make it impossible to use a conventional FACS, because of its high speed and virtually nonconfigurable arrangement of excitation beams. Photoswitching of single MTLn3 carcinoma cells labeled with the rsFP Dendra2 has been observed with a custom-built optical setup both on glass slides and within blood vessels in mice.28 Although this approach represents a key step in the evolution of the instrumentation with some of the necessary functionality for screening rsFP libraries, such as the appropriate multipulse excitation and timing, it does not enable selection of single cells. In contrast, microfluidics systems provide excellent flexibility: they allow for an arbitrary number of excitation beams, varying the time and order of excitation/read-out beams, long time delays,29,30 and accurate cell sorting.31</p><p>Here, we describe a microfluidic in-flow multiple cycle photoswitching assay for NIR rsFPs. Diffractive optics are used to generate a two-color multiple-beam illumination pattern, which is matched to a fluidic channel geometry appropriate for the pulse sequence probing photoswitching. We used droplets of purified constitutively fluorescent iRFP713 as a calibration sample to compensate for beam-to-beam intensity variations.32 To assess the effects of operating parameters on measurements of photoswitching in flow, we employed DrBphP-PCM, which is the BphP from Deinococcus radiodurans. This construct is a potential template for the development of new NIR rsFPs, because it has been successfully used to engineer constitutively fluorescent iRFPs.18 To our knowledge, this is the first flow system that permits screening of NIR rsFPs on the basis of their most critical photophysical properties. The methods reported here represent a foundation for the design of high throughput screening systems enabling the development of rsFPs for super-resolution imaging, and other biophotonic functionalities for visualizing cellular phenomena.</p><!><p>DrBphP-PCM and iRFP713 with N-terminal polyhistidine tags were expressed and purified as described.32 To determine fluorescence quantum yield of DrBphP-PCM, we compared fluorescence intensity of the purified protein to that of an equally absorbing iRFP713.32 Photoconversion of purified protein solution (10 μM) or suspension of mammalian cells stably expressing DrBphP-PCM was measured with 5 mW/cm2 of 780/20 nm and 636/20 nm LED sources.</p><!><p>We obtained a HeLa S3 (ATCC) stable preclonal mixture by transfecting cells with pDrBphP-PCM mammalian expression plasmid. Transfected HeLa S3 cells were selected for DrBphP-PCM expression with 700 μg/mL of G418 antibiotic and further enriched using a FACSAria sorter (BD Biosciences). The brightest cells expressing DrBphP-PCM were collected in the 710/20 nm fluorescence channel.</p><!><p>For single droplet measurements, purified protein samples were placed in a microwell 8 μm in diameter and 10 μm deep, closed with a coverslip. For single cell measurements, cell suspension was placed between two coverslips and sealed. The samples were illuminated with a pulse sequence consisting of a NIR (783 nm) pulse of 4 kW/cm2, followed by three equally spaced red (643 nm) laser pulses of 1.2 kW/cm2. This illumination corresponds to a single switching cycle, which starts by switching the protein on with the 783 nm laser and continues with stepwise switching off by the 643 nm laser. Each measurement consisted of 250 switching cycles.</p><!><p>Microfluidic chips were fabricated with polydimethylsiloxane (PDMS), using soft-lithography techniques. We used a flow-focusing junction for droplet generation as this geometry offers robust performance over a wide range of operating conditions.33 Calculations of the chip geometry and its hydrodynamic properties used the model and algorithms described earlier.34,35 The microfluidic chip has channels with 40 μm square cross-section, and both aqueous and oil phase channels are 50 mm long. The channels were treated with 1% perfluorododecyltrichlorosilane solution. This configuration produces ~2–3 50 pL droplets/s, moving with a speed that is adjustable in the range of 10–30 mm/s at input pressures of 1–2 psi. Driving pressure was applied to the microfluidic chip through 116-in.-diameter PFA microtubings connected to precision dual-stage pressure regulators with an accuracy of 0.01 psi at the output stage. We used two separate regulators to adjust input pressures in aqueous- and oil-phase channels independently.</p><!><p>In the optical setup (Figure 2), red and pink areas show traces of one (extreme) of the beams in the 643 and 783 nm illumination arms. The illumination optics generate 40-μm-diameter beams separated by a center-to-center distance of 130 μm, which was chosen to minimize beam overlap and provide dark relaxation times comparable to or longer than exposure times. This particular pattern gives dark relaxation times equal to twice the exposure time. The length of the beam waist (Rayleigh range) is ~1 mm, which greatly exceeds the 40 μm channel depth, thus ensuring that defocusing (axial displacement of the chip) minimally impacts the measured signal. All software required for acquisition and processing of optical signals was written in NI LabVIEW, using the NI DAQmx library to access the data acquisition board. Operational software allowed acquisition of the signal triggered by the first pulse in series, finding amplitude and position of pulses in the acquired signal, real-time calibration of amplitudes, and calculation of the switching contrast for every droplet.</p><!><p>In the Pr (ON) state, purified DrBphP-PCM exhibited an absorbance maximum at 700 nm (Figure 1E), an emission maximum at 720 nm (Figure 1F), and a fluorescence quantum yield of 2.9%. Upon 636 nm illumination, it photoconverted to the Pfr (OFF) state with an absorbance maximum at 750 nm and a switching contrast of 230% for fluorescence intensity in the spectral range of 690–740 nm (Figure 1F). The absorbance at 700 nm also decreased almost 2-fold in the Pfr state. Similar fluorescence behavior was observed in HeLa S3 cells stably expressing DrBphP-PCM (Figure 1G). We next detected repetitive reversible photo-switching of DrBphP-PCM fluorescence in cell suspension. Under these conditions, DrBphP-PCM was photostable over several photoswitching cycles, with a switching contrast of 150% (Figure 1H).</p><!><p>To determine microfluidic chip operating parameters such as flow speed (i.e., exposure and dark times) and excitation/switching laser energies, we measured photoswitching on stationary single droplets of purified DrBphP-PCM and single HeLa S3 cells expressing DrBphP-PCM under similar conditions. Microsecond laser pulses were used to simulate the passage of droplets through CW beams in a microchannel. The pulse sequence and duration for one illumination cycle are shown in Figure 3A, where the black curve represents a 783 nm laser pulse, and blue, orange, and green lines are successive 643 nm laser pulses. Although the 643 nm pulses are all of equal intensity, this sequence can be thought of as "on-probe–off-probe." The fluorescence response of the sample was quantified from the amplitudes of the signal pulses. The data shown in Figures 3B–E, as well as Figures S2–S4 in the Supporting Information were processed with a 30–50-point moving average filter. Photoswitching is evident from changes in signal amplitudes from the 643 nm pulses within a cycle. When the 783 nm laser is off, responses from the 643 nm excitation pulses within a cycle are equal (Figures 3B and S2C). However, when the 783 nm pulse is on, the fluorescence signal from the first 643 nm pulses is substantially higher, but the signal then decreases for the next two pulses, because the 643 nm pulses switch off the fluorescence. Figures 3B–E show dependence of signal pulse amplitude on the number of cycles, where line color corresponds to the pulse colors in Figure 3A. The decrease of pulse amplitude with increasing number of cycles is caused by photobleaching. We varied the time delay between 783 nm pulse and 643 nm pulses (τ783), length of the switching cycle (Δcycle), and pulse duration (Δ643 and Δ783). In droplets of purified DrBphP-PCM, we discovered that the magnitude of photoswitching (vertical shift between blue and green points in the plots) is dependent on the time delay (τ783). The highest switching response (illustrated in Figure 3C) was observed for τ783 = 1 ms. Longer τ783 caused a decrease in switching response (Figures S3A–C in the Supporting Information). This effect is caused by photoswitching upon exposure to low-intensity ambient light, because of the high light sensitivity of DrBphP-PCM. To reduce the influence of this ambient light as much as possible, all measurements were made in the dark room with sample that had been additionally covered from the back side. Further light insulation did not cause considerable improvements in switching response.</p><p>Next, we varied the dark time of a single switching cycle, i.e. cycle length Δcycle (Figure 3A). This parameter influences the transition of the protein into the ON (Pr) state due to exposure to ambient light between consecutive pulse cycles. This effect is easily revealed when the 783 nm laser is off. In this case, all three fluorescence pulses within a cycle should ideally have equal amplitudes, as seen in Figure 3B, where Δcycle = 50 ms. However, Figures S3D–F show that a small but measurable amount of switching occurs at Δcycle = 100 ms and Δcycle = 200 ms. Therefore, single-cell measurements were made using τ783 = 1 ms to ensure highest switching response and Δcycle = 50 ms to reduce the impact of uncontrollable transition of DrBphP-PCM into the Pr state.</p><p>In single-cell experiments, we varied the total energy exposure by varying pulse durations Δ643 and Δ783 from 15 μs to 100 μs, with fixed irradiances at 643 and 783 nm. These exposures correspond to fluences of 0.018–0.119 J/cm2 at 643 nm and 0.060–0.398 J/cm2 at 783 nm. Photoswitching was observed, even at the lowest exposures (Figure 3D). The highest switching response and largest amplitude signal were achieved at Δ643 = Δ783 = 60 μs (Figure 3E). Longer pulses did not significantly enhance either the switching response or signal amplitude (see Figures S4A–D in the Supporting Information). Based on this observation, we therefore determine the optimal conditions to observe in-flow switching of DrBphP-PCM in the microfluidic chip are 0.240 J/cm2 at 783 nm and 0.072 J/cm2 at 643 nm. Assuming a beam diameter of 40 μm and droplet speed of 10 mm/s, we obtain an exposure time of 4 ms and irradiances required in the microfluidic chip are 60 W/cm2 CW at 783 nm, and 18 W/cm2 CW at 643 nm. Using these intensities ensures identical switching conditions in both stationary off-chip experiments and in-flow measurements in the microfluidic chip.</p><!><p>The use of identical DOEs vastly simplifies management of the number and order of beams in the final diffraction pattern, thus allowing customization of the pulse sequence in the cytometry system, although at the cost of laser power approximately proportional to the number of beams that are blocked. Thus, for example, we used 15 of 25 643 nm beams and 5 of 25 783 nm beams in the initial diffraction pattern. Figures 2D and 2E show intensity distributions for the 643 nm laser beams measured in the object plane, where the microfluidic channels were placed (783 nm laser is turned off and spatial filter SF 643 nm is removed). The variation in beam intensity over the entire pattern is 10%, excluding the zero-order beam, which has an intensity that is 30% higher than the mean value. To account for the effect of these variations on flow measurements, we performed calibrations with constitutively fluorescent iRFP713 (see below). As shown below, the 783 nm beam intensity only weakly impacts the switching contrast, and therefore, we neglected calibration for the 783 nm laser beams.</p><p>We folded the fluidic channel multiple times into a geometry that matches the two-dimensional array of beams from the DOE (Figure 2C) and provides a sufficient length within the field of view of the imaging system. Generally, the number of cycles chosen is a compromise between complexity of the signal to be analyzed and consistency of the analysis. Our current design permits up to five cycles of switching, and any cycle can be excluded from analysis individually using a spatial filter.</p><p>Chip design requires optimization of the channel geometry using a set of initial parameters, such as the viscosity and density of the liquids, a usable range of driving pressures, and a set of desired properties of the droplets, such as volume, speed, and spacing. Droplet frequency and speed are the most important operating parameters, because they must take into account the photoswitching kinetics of DrBphP-PCM.22,26,27 Given the intensities and diameters of the laser beams and distances between them, low speeds ensure sufficient exposure times for droplets to absorb sufficient energy for photo-switching and for conformational changes in the protein to occur. At the same time, the speed and frequency of the droplets should be high enough for the system throughput to be at a reasonable level. The compromise between these factors resulted in the droplet generator used in our study, which gave a droplet volume of ~50 pL, a droplet frequency of 2–3 droplets/s, and a speed of 10–30 mm/s at 1–2 psi driving pressure.</p><!><p>To compensate for fluctuations in the fluorescence signal due to a nonuniform distribution of the 643 nm beam intensities in the diffraction pattern, we performed calibrations with droplets of purified constitutively fluorescent iRFP713, which has an excitation peak at 690 nm and emission maximum at 713 nm.32 Figure S5A in the Supporting Information shows typical signals from a droplet of iRFP713 in flow. To ensure that photobleaching of iRFP713 does not affect the accuracy of the calibration, we performed time-lapse measurements with a single beam exciting a single motionless droplet filled with iRFP713. Figures S6A and S6B in the Supporting Information show that bleaching does not exceed 1% within the first 200 ms. In flow, droplets move at a speed of 10–30 mm/s and pass through a 40-μm-diameter laser beam in 2–4 ms. The time needed for a droplet to pass the entire interaction region is ~200 ms (Figure S5A). Assuming that all pulses from a single droplet should be of equal amplitude, one can calculate an array of calibration coefficients (i.e., a lookup table, LUT).</p><!><p>We studied DrBphP-PCM in flow under various conditions, such as droplet speed and continuous wave intensities of 643 and 783 nm lasers (Table S1 in the Supporting Information). Figures 4A and 4B illustrate the fluorescence responses of DrBphP-PCM in flow. The switching contrast was calculated as a ratio of amplitudes of the first-pulse responses to third-pulse responses within a cycle (blue bars in the plot). This ratio was calculated for every cycle of switching and averaged over all cycles. Amplitude of the second pulse (light blue bar) served as an additional parameter sensitive to the kinetics of switching. Bar plots in Figures 4A and 4B show amplitudes averaged over 200 droplets. To verify that photoswitching occurs, we examined a signal from DrBphP-PCM when 783 nm laser is off. Figure 4A shows that amplitudes of the pulses are equal within 6% error. In contrast, when the 783 nm laser is on (Figure 4B), switching is clearly visible in every cycle.</p><p>We found that the 643 nm excitation (switching-off) laser irradiance has a considerable and complex effect on the switching contrast. It was expected that, with decreasing droplet speeds, the switching contrast should increase, as longer exposure ensures complete photoswitching. However, near the maximum values of 643 nm irradiance examined here (178 W/cm2), we find an inverse relationship. With decreasing irradiance, the influence of exposure time is weak (near 84 W/cm2) and then directly proportional to irradiance in the range of 20–45 W/cm2. Even though PDMS is transparent at these wavelengths, laser light scatters from impurities and reflects from interfaces within the chip, leading to switching between excitation pulses. At low droplet speeds, the impact of this incidental switching increases and, at an extreme irradiance of 178 W/cm2, results in a greatly decreased photoswitching contrast.</p><p>We observed a difference in the photoswitching response of 15 μM and 30 μM droplets of DrBphP-PCM. The most significant differences are seen at the minimal and maximal values of 643 nm irradiance, revealing an inverse relationship between switching contrast and droplet speed. This behavior can be understood by considering that lower concentrations of DrBphP-PCM requires less energy to be switched in both directions, while higher concentrations of droplets demonstrate substantial inertia in photoswitching.</p><p>The influence of the NIR laser irradiance on switching was much weaker (Figure 4E). At half of the maximum 783 nm laser irradiance, a relatively high 40% switching contrast was observed, whereas 45% was achieved with maximum irradiance, while adjustment of the droplet speed and irradiance of 643 nm laser beams produced changes of the switching contrast in the range of 15%–50%. Most likely, the 783 nm laser did not promote considerable switching between beams, because of the lower scattering of long-wavelength irradiation, at least within the available laser power.</p><p>Dotplots of switching contrast versus brightness were made for droplets of 1.8, 7.5, 15, and 30 μM DrBphP-PCM in flow. For example, Figure 5A shows a dotplot for 10 000 droplets of 15 μM DrBphP-PCM measured both with the 783 nm laser off and on, and averaged over all five cycles. Dotplots for droplets filled with 1.8, 7.5, and 30 μM DrBphP-PCM are shown in Figure S7 in the Supporting Information. Values to the left and below each cluster give the coefficients of variation (CV) of the switching contrast and brightness, respectively. The cluster of points near the bottom left is collected with the 783 nm switching laser off (highlighted green area), whereas the other pair, near 20%–40%, are measured with the laser on (highlighted red area).</p><p>For the points collected with the 783 nm laser on, the cluster at the highest contrast value represents switching contrast versus brightness for the droplet from the first 643 nm beam and the other represents measurements from the second 643 nm beam. In both cases, the signal (brightness) from the third pulse is used to calculate the contrast ratio, i.e., for beam 1contrast ratio = 〈brightness in the first pulse〉〈brightness in the third pulse〉 and for beam 2contrast ratio = 〈brightness in the second pulse〉〈brightness in the third pulse〉 The rationale for this choice is that, after the second 643 nm switching-off beam, it is assumed that the entire population is now in the OFF (Pfr) state. This can be seen in that the shift between clusters of points along the horizontal axis is larger, because the droplets are initially turned on by the 783 nm beam in each cycle and, therefore, the first 643 nm beams have the largest effect on switching the fluorescence off. Figures 5B and 5C, as well as Figure S7 in the Supporting Information, show the dependence of the CVs on concentration of the protein solution for the first pulse (highest switching contrast and highest brightness). Interestingly, even though dispersion of the brightness is relatively high (and we can assume that it is not dependent on concentration), because of the droplet-to-droplet variation of the brightness, the CV of the switching contrast values is very small. Changes in brightness are obviously caused by changes in droplet size and speed; however, these changes do not influence the performance of the system and, therefore, the switching contrast can be calculated with very high precision. Even for the lowest concentration of 1.8 μM, the CV of the switching coefficient is only 4%, which means deviation of the switching coefficient is an order of magnitude lower than its mean value. Droplet-to-droplet variation of the brightness caused by changes in droplet size can be reduced further by introducing droplet size sensor based on either detection of scattered light or microscopic image recognition. Although other representations of these rather high-dimensional data can be considered, we expect the dotplots shown to be of significant value in quantifying the photoswitching performance of rsFPs (for example, in the screening of libraries).</p><!><p>The innovative features of our microfluidic system arise from a combination of two factors: (i) simplicity of the system (i.e., simple design of the microfluidic chip, diffractive-optics-based multiple beam illumination, rigid optical scheme) and (ii) complex analysis possible with this simple scheme (i.e., multiple cycles of switching, multiple variations in the beam arrangement). Applications of microfluidics to single cell, particularly fluorescence-activated, analysis have expanded considerably through the past decade. An important category of these applications concerns systems that reside in two distinct states. These states may, for example, correspond to ligand bound/unbound states of a sensor for some cellular analyte, or activated/deactivated states of an optogenetic element controlling a signaling pathway. Although many reports describe cell manipulation within single-cell traps and cell arrays36,37 few publications describe multiple in-flow operations on single droplets or cells23–25,38 as would be necessary for profiling and sorting populations.</p><p>For example, Dolega et al.39 have proposed the most direct implementation of such a system using multiple valves for iterative routing of a droplet through a chip. Using multiple valves requires precise and complex synchronization, especially at higher speeds of operation. Their system incorporates long droplet residence times in a looped microfluidic channel to study crystallization and precipitation processes in droplets. Screening for photoswitching efficiency in such a system would be impractical, because of the long residence times, relative to the photoswitching kinetics of rsFPs. Another report describes an interesting idea,40 where environmental conditions within the microfluidic chip were reversibly changed by varying solution flow rates, similar to the sheath flow in the flow-focusing junction, and optical tweezers were used for precise positioning of a cell in the flow. Although this scheme allows iterative single-cell analysis, realization of a high-throughput screening system on this basis is unrealistic. Therefore, the work reported here represents a substantial advance, because it is the first to describe in-flow triggering, probing, and quantification of multiple reversible photoswitching of DrBphP-PCM both in single-point and in-flow experiments.</p><p>For measurements of switching contrast in a cell population, accuracy of estimation and the influence of exogenous factors are crucial. One of these factors is spontaneous or uncontrollable switching, which can be systematic and cause bias of estimation. For example, single-point measurements showed that switching contrast was dependent on time delay between illumination of the sample with the 783 nm laser (switching protein ON) and illumination with the 643 nm laser (excitation/switching protein OFF). Switching contrast decreased as the time delay increased (see Figures S3A–C). This effect is due to the uncontrollable transition of the DrBphPPCM protein into the Pfr state under the influence of weak ambient light. Excessive time delay between consecutive switching cycles caused early transition of the protein into the ground Pr state (Figure S3D–F). Competitive influence of these two opposite processes in the microfluidic chip overlapped with other factors, such as protein concentration. In agreement with measurements on stationary droplets, we found that the low droplet speed/low intensity of the 643 nm laser provide optimal conditions for observing high photo-switching contrast in flow (Figure 4). However, single-point experiments showed that, although it is possible to use higher excitation intensity at 643 nm (Figure 3), switching contrast does not increase, although the magnitude of the signal does increase. This flexibility will facilitate switching measurements on rsFPs in cells over a range of dimensions and expression levels. Note that the rather low (2.9%) fluorescence quantum yield of DrBphP-PCM represents the lower bound of what will be of interest in efforts to develop new rsFPs.</p><p>Both types of measurement described here, single-point and in-flow, demonstrate the ability of DrBphP-PCM to be reversibly and repeatedly photoswitched and used as a calibration standard for in-flow measurements. Moreover, single-point measurements provided clear evidence that the protein withstands at least 250 cycles of consecutive switching without a substantial reduction in photoswitching contrast. During photoswitching, DrBphP-PCM undergoes structural rearrangements. Upon illumination with far-red light, the C15=C16 double bond in the biliverdin isomerizes from cis to trans conformation. It leads to a concomitant reorientation of the hydrogen bond network around D-ring and adjacent amino acids residues of the protein. These combined motions are followed by complex structural remodeling of the PAS-GAF-PHY triad domains.41 The observed decrease in photo-switching contrast after the increase in delay time between illumination with red and NIR lasers could be explained by the fast reverse BV isomerization caused by exposure to ambient light.</p><p>The developed system operates on principles similar to those of conventional flow cytometers but provides multidimensional data with substantially more information content. Because of the combination of the robust optical system, a simple, flexible, and cost-effective droplet-based microfluidics technique, and a specialized calibration procedure for both intensity and switching our system proposes screening based on the unique property of the rsFPs, such as the efficiency of switching. Measurement of the switching contrast has been achieved with a hardware-limited CV of the system of as low as 1.7%, because droplet-to-droplet fluctuations in the fluorescent signal are suppressed by the ratiometric quantity. Since this system operates under steady-state flow conditions, we expect that actuation methods such as dielectrophoresis can easily be implemented to subsequently sort the droplets with criteria based on the brightness, photoswitching contrast, switching fatigue, or any changes in spectral properties (e.g., absorbance) of the cells/proteins/particles contained within them.</p><!><p>We developed the first system for in-flow switching and screening of near-infrared (NIR) reversibly photoswitchable fluorescent proteins (rsFPs), which consits of diffractive optics for a multiple-beam illumination pattern and droplet-based microfluidics to interrogate cell-sized volumes of the protein. This system enabled observation of several consecutive cycles of reversible photoswitching of the weakly fluorescent protein bacterial phytochrome photosensory core module from Deinococcus radiodurans (DrBphP-PCM) in droplets flowing through a microfluidic channel. We determined the optimal conditions necessary to observe the highest contrast of in-flow switching. Finally, we tested the system in flow-cytometry mode with droplets filled with purified protein solution to characterize and to visualize population of droplets in industry-accepted form. The real-time analysis capability of this system makes it straightforward to incorporate cell-sorting functionality, thereby providing a solid basis for microfluidics-based directed evolution of NIR rsFPs.</p>

PubMed Author Manuscript

Identification of inhibitors that target dual-specificity phosphatase 5 provide new insights into the binding requirements for the two phosphate pockets

BackgroundDual-specificity phosphatase-5 (DUSP5) plays a central role in vascular development and disease. We present a p-nitrophenol phosphate (pNPP) based enzymatic assay to screen for inhibitors of the phosphatase domain of DUSP5.MethodspNPP is a mimic of the phosphorylated tyrosine on the ERK2 substrate (pERK2) and binds the DUSP5 phosphatase domain with a Km of 7.6 ± 0.4 mM. Docking followed by inhibitor verification using the pNPP assay identified a series of polysulfonated aromatic inhibitors that occupy the DUSP5 active site in the region that is likely occupied by the dual-phosphorylated ERK2 substrate tripeptide (pThr-Glu-pTyr). Secondary assays were performed with full length DUSP5 with ERK2 as substrate.ResultsThe most potent inhibitor has a naphthalene trisulfonate (NTS) core. A search for similar compounds in a drug database identified suramin, a dimerized form of NTS. While suramin appears to be a potent and competitive inhibitor (25 ± 5 μM), binding to the DUSP5 phosphatase domain more tightly than the monomeric ligands of which it is comprised, it also aggregates. Further ligand-based screening, based on a pharmacophore derived from the 7 Å separation of sulfonates on inhibitors and on sulfates present in the DUSP5 crystal structure, identified a disulfonated and phenolic naphthalene inhibitor (CSD3_2320) with IC50 of 33 μM that is similar to NTS and does not aggregate.ConclusionsThe new DUSP5 inhibitors we identify in this study typically have sulfonates 7 Å apart, likely positioning them where the two phosphates of the substrate peptide (pThr-Glu-pTyr) bind, with one inhibitor also positioning a phenolic hydroxyl where the water nucleophile may reside. Polysulfonated aromatic compounds do not commonly appear in drugs and have a tendency to aggregate. One FDA-approved polysulfonated drug, suramin, inhibits DUSP5 and also aggregates. Docking and modeling studies presented herein identify polysulfonated aromatic inhibitors that do not aggregate, and provide insights to guide future design of mimics of the dual-phosphate loops of the ERK substrates for DUSPs.Electronic supplementary materialThe online version of this article (doi:10.1186/s12858-015-0048-3) contains supplementary material, which is available to authorized users.

identification_of_inhibitors_that_target_dual-specificity_phosphatase_5_provide_new_insights_into_th

Background<!><!>Molecular docking<!>Ligand-based searching<!>Synthesis of RR505 and RR506. Synthesis of Carbazole-1,3,6-trisulfonic acid, trisodium salt (RR505)<!>Synthesis of Carbazole-1,3,6,8-tetrasulfonic acid, tetrasodium salt (RR506)<!>Alternative synthesis of RR506<!>Protein production<!>p-nitrophenol phosphate (pNPP) activity assay<!>Validation of pNPP assay for high throughput screening (HTS)<!>IC50 measurements<!>Nephelometry<!>NMR binding assay<!>ERK dephosphorylation assay<!>Docking and ligand-based in silico searches yield candidate small molecules that target the DUSP5 PD domain<!><!>Expression, purification and assay of the DUSP5 PD domain<!><!>Suramin, an FDA approved analog, was identified via lead hopping<!><!>Docking and HTS assay to identify polysulfated lead molecules<!>Lead-hopping to an FDA approved drug and beyond<!>DUSP5 inhibition vs. activation: implications for vascular anomalies<!>Charge separation vs. distance hypothesis<!>Conclusion<!>Availability of supporting data<!>

<p>Mitogen-activated protein kinases (MAPKs), such as extracellular regulated kinase (ERK) [1], are activated by phosphorylation of tyrosine and serine/threonine residues in their activation loops. MAPKs can then be deactivated by phosphatases that remove these phosphate groups from their activation loop. One such class of phosphatases, dual-specificity phosphatases (DUSPs), is unique in that it can dephosphorylate both serine/threonine and tyrosine residues. The Ramchandran lab has shown that DUSP5 is necessary for early vascular patterning in vertebrates, and is mutated in patients with vascular anomalies [2]. DUSP5 plays a regulatory role in vascular development based on its ability to specifically interact with and dephosphorylate phosphorylated ERK (pERK) [3–7]. Indeed, we identified a clinically relevant serine to proline mutation (S147P) that is associated with vascular defects [2]. This mutation has been shown previously by our group to interfere with the dephosphorylating activity of DUSP5 protein [8], and makes the protein hypoactive. However, the direct causal role of S147P in vascular anomaly progression is yet to be established. Nevertheless, DUSP5 is a critical drug target for vascular-related diseases, and more broadly, MAPKs and their DUSP partners are involved in cell signaling that is directly involved in a wide range of diseases, including cancer, diabetes, and autoimmune disorders [6, 9–11]. Recently, DUSP5 has gained increased attention in the scientific literature [12–14] especially as it relates to loss or gain of expression of DUSP5 in murine models, and its associated phenotypic changes in both the immune and cancer biology systems. DUSP5 knockout (KO) mice are alive, and display no overt phenotype, indicating that it is dispensable for embryonic development. However, Holmes et al did report that these mice showed increased function and survival of eosinophils, a key player in the immune system's ability to clear parasitic infections [12]. Further, Rushworth et al reported increased sensitivity to a skin cancer model in their murine model [14]. In terms of the vasculature, the DUSP5 KO rat displays enhanced myogenic response and autoregulation of cerebral blood flow [15]. Taken together, these studies demonstrate that inhibition of DUSP5 will result in biologically relevant changes in vivo.</p><p>From a conformation perspective, DUSP5 is comprised of two domains, an N-terminal ERK binding domain (EBD) and a C-terminal phosphatase domain (PD) [5, 16]. While there is no structure available for intact DUSP5, there is a crystal structure of the PD [162]. The DUSP5 PD structure has two anionic sulfate groups bound in the active site near the catalytic Cys263 (mutated to serine in the structure), and separated by 7.2 Å. These sulfates had been proposed to occupy the same binding pockets that are occupied by the phosphate groups on the substrate [16]. For the ERK2 substrate, the pThr-Glu-pTyr tripeptide region of the ERK2 activation loop presumably occupies this region in the DUSP5 PD [17, 18].</p><!><p>DUSP5 and ERK2 Models. a Model depicting the two domains of DUSP5. This model is comprised of two domains, the ERK binding domain (EBD) and phosphatase domain (PD), and illustrates the relative location of the domains and their connection via a 30 amino acid linker of unknown structure. The homology model of EBD was constructed using the solution structure (21 % identity and 35 % homology) of human MKP-3 protein (PDB:1HZM) as a template [35]. The phosphatase domain is the previously reported crystal structure (PDB:2G6Z) [16]. The 30 amino acid linker region connecting the two domains was prepared manually, and is of unknown structure. The S147P mutation present in patients with vascular anomalies is shown in green, and arginine-rich basic regions have been identified. b DUSP5 and ERK2 binding model. DUSP5 (blue) is positioned similarly in respects to panel a with the EBD to the left and PD to the right, wrapping around human ERK2 in yellow. Model was prepared as described in our previous paper [8]. The linker region may have the first 11 amino acids as helical based secondary structure predictions [46–48], although this was only found to be loosely helical after molecular dynamics simulations. The ERK2 (yellow) structure (PDB:3I60) [18] is shown between the DUSP5 domains to illustrate relative shape and size complementarity; and, relative orientation of ERK2 and DUSP5 is based on the molecular dynamics simulation and associated analysis presented in our previous paper [8]</p><!><p>The Center for Structure-based Drug Design and Development (CSD3) chemical library, consisting of 11,500 drug-like chemicals, was prepared in electronic format as two-dimensional (2D) SDF files. Using Pipeline Pilot [19], the protonation state of all compounds was adjusted to reflect the most prevalent form at a pH of 7.4. CORINA [20] was used to convert these files to three-dimensional (3D) PDB coordinate files, which resulted in energy-minimized 3D structures. The files were then processed with the python script prepare_ligand4.py, which comes with the Autodock Tools Suite [21]. This script generates a pdbqt file and adds partial charges to the ligand, sets all torsions in the ligand to active (to permit rotation), and merges all non-polar hydrogen atoms.</p><p>The DUSP5 PD structure (PDB:2G6Z) [16] was prepared for docking using the Autodock Tools Suite [21]. Grid maps were used in the energy calculations performed by Autodock. Partial charges were added and all non-polar hydrogen atoms were merged, resulting in a pdbqt file. The 13 different grid maps, one for each of the different atoms present in the chemical library of compounds (ex. C, H, F, Cl, etc.), were generated using Autogrid4 [21]. A grid box, the site used to dock the ligands, was positioned to cover the entire protein in a blind docking experiment to ensure unbiased identification of binding location and orientation.</p><p>The docking parameter file (dpf), which contains the parameters that Autodock4 uses to dock ligands into the protein, was prepared using the python script prepare_dpf4.py, and default docking parameters were used, except that 50 separate docking calculations were performed with each calculation consisting of 1,750,000 energy evaluations, and a root mean square deviation (rmsd) tolerance set to 2.0 angstroms (to define entry of structure into a given cluster). The dpf files were then automatically docked using the MUGrid Cluster (Marquette University) with HTCondor [22, 23] and AutoDock4 [21, 24] using the Lamarckian genetic algorithm local search method to perform the optimization of docking poses. The docking poses were then clustered on the basis of the rmsd between the coordinates of the atoms in a given ligand, and were ranked on the basis of calculated free energy of binding. The docking log files were then analyzed using the python script summarize_results4.py contained in the shell script sumresults_4.py [21], which rank orders all the dockings by binding energy. The results were then analyzed to find the best-clustered compounds with lowest free energy of binding as determined by Autodock4.2. Additional docking of all experimentally tested chemicals was performed as described above, but with 100 dockings trials.</p><!><p>As previously described, the CSD3 chemical library was electronically prepared and protonation state adjusted using Pipeline Pilot [19]. Using OpenEye Scientific Software's Omega2 [25, 26], three dimensional coordinates were calculated and stored in OpenEye Scientific Software's preferred file format, .oeb.gz, for subsequent molecular overlay evaluation. OpenEye Scientific Software's Rapid Overlay of Chemical Structures (ROCS) [25] software was used to search for molecules with similar shape and electronic properties to a lead molecule. Lead molecules identified from DUSP5 docking and inhibition studies were used as chemical queries to search a database of Food and Drug Administration (FDA) approved drugs, to identify FDA approved drugs that might also be DUSP5 inhibitors.</p><p>The ZINC library [27] of 13 million commercially available chemicals was obtained as 2D SDF files and prepared similarly to the CSD3 chemical library for use with OpenEye Scientific Software's ROCS. This further expanded the availability of chemical analogs available for experimental screening. ROCS calculations were also performed against the DrugBank [28, 29] database of FDA approved drugs.</p><!><p>Solid carbazole (3.0 g, 17.9 mmol) was placed in a 50-mL round-bottom flask and 67 % H2SO4 (12 mL) was added drop-wise at 22 °C and a slurry thus obtained was stirred and heated at 115 ± 5 °C for 6 h. The resulting dark solution was cooled to room temperature and poured into a saturated NaCl solution (100 mL) containing NaOH (2.4 g, 60 mmol) to afford an ash-colored precipitate, which was filtered, washed with saturated NaCl solution (50 mL) and dried at 90 °C for 10 h to get 7.5 g of the crude product.</p><p>The crude solid was dissolved in distilled/deionized water (150 mL), treated with activated charcoal Norit (1.1 g) and the resulting mixture was refluxed for 15 min. The solution was filtered hot through a pad of Celite®, and evaporated slowly to afford a white powder of RR505 (5.5 g, 65 % yield). 1H-NMR (400 MHz, D2O): 7.60 (1H, d, J = 8.8 Hz), 7.79 (1H, d, J = 8.8 Hz), 8.08 (1H, s), 8.53 (1H, s), 8.64 (1H, s), 1H-NMR (400 MHz, DMSO-d6): 7.58-7.66 (2H, m), 7.96 (1H, s), 8.24 (1H, s), 8.26 (1H, s), 10.81 (1H, s). 13C-NMR (100 MHz, D2O): 112.3, 118.8, 121.6, 121.64, 121.7, 124.5, 124.7, 125.5, 133.7, 134.7, 136.9, 142.1.</p><!><p>Solid carbazole (3.0 g, 17.9 mmol) was placed in a 50-mL round-bottom flask and chlorosulfonic acid (41.7 g, 358 mmol) was added in small portions with vigorous shaking at 22 °C, after which the mixture was stirred and heated at 100 ± 5 °C for 1 h. The resulting dark solution was cooled to room temperature and then poured slowly onto crushed ice (~100 g). The resulting precipitate was filtered by gravity filtration and was dried by placing between paper towels. The resulting semi-dried solid was dissolved in ethyl acetate (150 mL), treated with Norit (1.5 g), and refluxed for 15 min and filtered hot through a pad of silica gel (~1x1.8 inch). The filtrate was concentrated in vacuo and recrystallized from a 1:9 mixture of ethyl acetate and hexanes to afford a yellow solid, which was filtered and dried in vacuo.</p><p>The dried solid was dissolved in a mixture of dioxane (20 mL) and distilled/deionized water (20 mL) and heated under reflux for 12 h. The resulting solution was cooled to room temperature and was extracted with diethyl ether (2 x 50 mL) to remove nonpolar impurities. The aqueous layer was neutralized by a dropwise addition of NaOH solution (1 M) with continuous monitoring of pH using pH paper. The resulting solution was concentrated to ~10 mL and acetone was added to afford a white powder of RR506 (3.5 g, 22 % yield, average yield from 3 runs). 1H-NMR (400 MHz, D2O): 8.12 (2H, s), 8.68 (2H, s), 1H-NMR (400 MHz, DMSO-d6): 7.97 (2H, s), 8.26 (2H, s), 10.61 (1H, s); 13C-NMR (100 MHz, D2O): 121.9, 122.1, 123.8, 125.7, 134.7, 136.7.</p><!><p>Solid carbazole (1.0 g, 6 mmol) and nitrobenzene (20 mL) were placed in a 50-mL round-bottom flask and chlorosulfonic acid (14 g, 120 mmol) was added in small portions at 22 °C, after which the mixture was stirred at 22 °C for 72 h. The resulting solution was poured into aqueous saturated NaCl solution (100 mL) containing NaOH (0.96 g, 24 mmol) which resulted in a fluffy precipitate. The precipitate thus formed was filtered and dried. The solid was dissolved in distilled/deionized water (100 mL) and refluxed with 1.5 g of Norit for 15 min and filtered hot through a pad of celite. The filtrate was concentrated to ~25 mL and RR506 was precipitated by addition of acetone. The precipitate was filtered and dried to afford RR506 as a white solid (2.9 g, 84 % yield). 1H-NMR (400 MHz, D2O): 8.12 (2H, s), 8.68 (2H, s), 1H-NMR (400 MHz, DMSO-d6): 7.97 (2H, s), 8.26 (2H, s), 10.60 (1H, s).</p><!><p>The DUSP5 PD gene was synthesized by Blue Heron (Bothell, WA) in both an active wild type form (DUSP5 PD(WT)) and an inactive form, where the catalytic cysteine was mutated to a serine (DUSP5 PD(C263S)). The genes were inserted into Origene pEX plasmids with ampicillin resistance and an N-terminal hexa-histidine tag to facilitate protein purification. Plasmids were transformed into BL21(DE3) cells (Invitrogen) for expression.</p><p>For unlabeled DUSP5 PD(WT) preparation, an overnight culture was used to inoculate 2 L of LB (Luria-Bertani) media, containing 50 μg/mL of ampicillin. Cells were grown at 37 °C to an OD600 of 0.7 and then induced with 0.6 mM isopropyl β-D-1-thiogalactopyranoside (IPTG) for 4 h at 37 °C, then for 14 h at 16 °C. Cells were harvested using centrifugation and frozen prior to purification. Thawed cells were lysed in a buffer containing 50 mM Tris, 300 mM NaCl, 5 mM imidazole, and 10 % glycerol at pH 7.8. Lysate was centrifuged at 15,000 rpm for 1 h. The supernatant was loaded on to Ni-Sepharose Fast Flow resin (GE Healthcare) and washed three times successively with five column volumes of lysis buffer containing 25 mM imidazole. Protein was eluted with lysis buffer containing 305 mM imidazole. Protein was then dialyzed in a buffer containing 50 mM potassium phosphate and 2 mM dithiothreitol (DTT) at pH 6.8.</p><p>For 15 N-labeled DUSP5 PD(C263S) preparation (for NMR titrations), an overnight culture was used to inoculate 2 L of LB media supplemented with 50 μg/mL of ampicillin. Cells were grown to an OD600 of 0.7 at 37 °C, then harvested and washed with M9 minimal media (pH 7.0) [26]. Cells were resuspended in 500 mL M9 minimal media containing 0.5 g 15NH4Cl, 2 g D-glucose, 5 mL Basal Medium Eagle with Earle's salts and sodium bicarbonate (Sigma Aldrich), 0.146 g L-glutamine (Sigma Aldrich) 1.0 mL 1 M MgSO4, and 0.5 mL 1 M CaCl2 (pH 7.2) [30]. Additionally, a metal mix containing Zn, Mn, Cu, Co, B, and Mo salts was added to supply cells with necessary micronutrients [30]. Cells were allowed to acclimate for 30 min at 37 °C, then induced with 1 mM IPTG for an additional 4 h at 37 °C. Cells were harvested and 15 N-labeled protein was purified as described before with the addition of 2 mM DTT during all purifications steps.</p><!><p>To measure enzymatic activity of the DUSP5 PD and the inhibitory capacity of selected molecules, an in vitro phosphatase assay was developed based on previous studies [31]. In this assay, DUSP5 PD will dephosphorylate the substrate p-nitrophenol phosphate (pNPP, Sigma Aldrich), yielding p-nitrophenolate, which absorbs at 405 nm with an extinction coefficient of 18,000 M−1 cm−1.</p><p>Thus, an increase in absorbance at 405 nm corresponds to the turnover of pNPP to p-nitrophenolate. The assay was initially optimized in 1 mL quartz cuvettes, then was subsequently optimized for and validated in a 96-well plate format. All IC50 values were obtained using the 96-well plate assay format (see below). The assay buffer contained 100 mM Tris, 100 mM sodium chloride, 5 mM magnesium chloride, and 1 mM DTT at pH 7.5. The pNPP substrate was prepared as a 50 mM stock by dissolving the solid substrate in assay buffer. The DUSP5 PD and pNPP were assayed initially in a cuvette (1 mL total volume) and initial velocities were fitted to the Michaelis-Menten equation:1\documentclass[12pt]{minimal} sepackage{amsmath} sepackage{wasysym} sepackage{amsfonts} sepackage{amssymb} sepackage{amsbsy} sepackage{mathrsfs} sepackage{upgreek} \setlength{\oddsidemargin}{-69pt} egin{document}$$ v=rac{V_{\max}\left[S ight]}{K_m+\left[S ight]} $$\end{document}v=VmaxSKm+Swhere v is the initial velocity, Vmax is the maximum velocity, Km is the Michaelis constant, and [S] is the concentration of pNPP. Data were fitted using a nonlinear least squares fit to eq. 1, with GraphPad Prism 6 software.</p><!><p>For the 96-well plate validation assay, sodium orthovanadate (Sigma Aldrich) was utilized as a positive control for inhibition [32] at a final concentration of 10 μM, to completely block DUSP5(WT) enzymatic activity. All plate assays were performed in standard 96-well clear bottom plates (Thermo Scientific Nunc) with a total assay volume of 200 μL, using a SpectraMax M5 Microplate Reader (Molecular Devices). The plate validation assay was performed with replicate columns of positive control wells, negative control wells and blank wells. Blank columns contained only buffer and pNPP. Negative control (uninhibited) contained buffer, pNPP, and DUSP5 PD(WT); and, positive control contained the same components, but also contained 10 μM sodium orthovanadate. The plate was then shaken and allowed to equilibrate in the spectrophotometer at 25 °C for 30 min. After incubation, 4 μL of a 50 μM enzyme stock was dispensed into appropriate wells utilizing a single-channel pipette. This produced a final enzyme concentration of 1 μM. Before a read was taken, the plate was shaken for five seconds. The initial rate for the DUSP5 PD(WT) reaction was linear for approximately 90 min; and, the plate was kept in the spectrophotometer at 25 °C for an additional 80 min after the kinetic read. The endpoint reading was subsequently taken at 90 min after initiation of reaction.</p><p>Slopes from the kinetic read, as well as single-point absorbance values at the 90-minute endpoint read, were then averaged. For blank wells and positive control wells, both slope values (continuous assay) and single point absorbance values (fixed time assay) were approximately zero, as expected (Table 2). Standard deviations were calculated and a Z' value [33] subsequently determined using the following equation:2\documentclass[12pt]{minimal} sepackage{amsmath} sepackage{wasysym} sepackage{amsfonts} sepackage{amssymb} sepackage{amsbsy} sepackage{mathrsfs} sepackage{upgreek} \setlength{\oddsidemargin}{-69pt} egin{document}$$ Z\hbox{'}=1-rac{3\left({\sigma}_p+{\sigma}_n ight)}{\left|{\mu}_p-{\mu}_n ight|} $$\end{document}Z'=1−3σp+σnμp−μnwhere σp is the standard deviation for the positive control, σn is the standard deviation for the negative control, μp is the mean for the positive control, and μn is the mean for the negative control. The Z' value is a coefficient denoting the quality of a high throughput screening assay, reflecting both the variation in data and dynamic range for the assay. A good assay exhibits a high signal to background ratio. A Z'-factor of 1.0 reflects an ideal assay; and, for an assay to be considered reliable, must exceed 0.5 [33].</p><!><p>IC50 values were obtained using the assay described above, in 96 well plates. The maximum inhibitor concentration screened in any plate was 300 mM and the minimum screened concentration was 1 μM. The IC50 plate was designed so that the first column of wells served as blanks, with wells containing only buffer and substrate. The second column of wells functioned as the plate negative control, with each well containing buffer, substrate and enzyme. The remaining wells in the plate contained buffer, substrate, enzyme, and varying amounts of inhibitor, with inhibitor concentration increasing from left to right across the plate. Data points were collected at a minimum in triplicate, and inhibitor concentrations were chosen to provide data equally spread on a logarithmic scale. The composition of buffer and the concentrations of substrate and enzyme utilized were identical to those in the plate validation assay. After initiation and shaking, a ten-minute kinetic read was taken.</p><p>For each plate assayed, the slope values for all negative control wells were averaged and the measured value considered representative of full enzymatic activity. Fractional activity was then calculated by dividing the slope of each inhibitor well by this value, determining the relative amount of enzyme activity observed at each concentration of inhibitor. Values were then plotted as percent activity versus the log of the concentration of inhibitor, and fitted to the following equation:3\documentclass[12pt]{minimal} sepackage{amsmath} sepackage{wasysym} sepackage{amsfonts} sepackage{amssymb} sepackage{amsbsy} sepackage{mathrsfs} sepackage{upgreek} \setlength{\oddsidemargin}{-69pt} egin{document}$$ y= Bottom+rac{\left( Top- Bottom ight)}{1+{10}^{x- \log I{C}_{50}}} $$\end{document}y=Bottom+Top−Bottom1+10x−logIC50where Top and Bottom are plateaus for the values of initial velocity when uninhibited and fully inhibited, respectively.</p><!><p>Nephelometry is a technique for measuring the relative aggregation of particles in solution, based on the light-scattering properties of molecular aggregates [34]. We performed nephelometry to explore the ability of the chemicals studied herein to form aggregates, which can lead to artifactual inhibition effects. Compounds were tested for aggregation in 96-well plates using a buffer containing 100 mM Tris base, 100 mM sodium chloride, and 5 mM magnesium chloride at pH 7.5. Each compound analyzed in these experiments contained concentrations of compound ranging from 10-100 μM, recorded in quadruplet. Each plate was analyzed at two separate gain values of 52 and 72. Data were collected using a BMG NEPHELOstar Plus, equipped with a 635 nm laser.</p><!><p>NMR samples of DUSP5 PD(C263S) were prepared for 2D 1H-15N HSQC (heteronuclear single quantum coherence) spectral titration studies. The 15 N-labeled DUSP5 PD(C263S) protein was concentrated using an Amicon Ultra-4 centrifugal device (Millipore) to 600 μM. NMR samples were prepared with the following conditions for RR505: 250 μM RR505, 250 μM DUSP5 PD(C263S), 10 % D2O, 50 mM potassium phosphate, 100 mM KCl, and 2 mM DTT at pH 6.8 and for CSD3-2320: 0 or 500 μM CSD3-2320, 500 μM DUSP5 PD(C263S), 10 % D2O, 50 mM potassium phosphate, 100 mM KCl, and 2 mM DTT at pH 6.8. NMR experiments were performed on a 500 MHz Varian NMR System using a triple resonance probe with z-axis gradients at 25 °C.</p><!><p>For this assay, 10 ng of GST-tagged recombinant phosphorylated ERK2 (R&D Systems, 1230-KS) was incubated with and without the indicated DUSP5 proteins (0.5 nM final concentration) for 15 min at room temperature, with or without the indicated drugs. The reactions were halted with 2x Laemmli sample buffer and subjected to SDS-PAGE. The proteins were transferred to polyvinylidene difluoride (PVDF) and immunoblotted using antibodies to pERK (Cell Signaling Tech., #9106) and total ERK, which includes both phosphorylated and unphosphorylated ERK1 and ERK2 (Cell Signaling Tech., #9102). Bound antibodies were visualized using horseradish peroxidase-linked anti-mouse IgG (Cell Signaling Tech, #7076S) and anti-rabbit IgG (Cell Signaling Tech, #7074S), respectively, and ECL reagents (Pierce, #34708) according to the manufacturer's protocol. For calculating IC50 values, gel bands were imaged by chemiluminescence with either film or digital image capture by a FluorChem HD2 imager (Alpha Innotech). Density of each band was quantified with ImageJ software by using the gel analysis tool. Relative values of phosphorylated ERK present for each drug concentration treatment compared to pERK only controls were calculated. These relative values were then used to obtain IC50 values with GraphPad Prism 6 software. Each experiment was repeated at least three independent times, and IC50 values provided as a range.</p><!><p>In this study, we were interested in identifying inhibitors that could selectively target dual-specificity phosphatase 5 (DUSP5), which we have shown previously to be mutated in patients with vascular anomalies. As shown in Fig. 1a, DUSP5 contains two domains namely an ERK-binding domain (EBD) and a phosphatase domain (PD) that are fused together by an unstructured linker region. The X-ray structure of PD of human DUSP5 was previously reported (PDB:2G6Z) [16], while the structure of EBD was constructed using homology modeling based on the solution structure (21 % identity and 35 % homology) of human MKP-3 protein (PDB:1HZM) as a template [35]. The 30 amino acid linker region connecting the two domains, which is of unknown structure, was prepared manually. A model of the human DUSP5-ERK2 complex (Fig. 1b) illustrates how DUSP5 (blue) wraps around ERK2 (yellow), its natural substrate, with the EB and PD DUSP5 domains located on opposite sides of ERK2. The model was prepared as described in our previous paper [8], and the relative orientation of ERK2 and DUSP5 is based on molecular dynamics simulations described previously [8].</p><!><p>Structures, Docking Energies, and IC50 Values of DUSP5 PD Inhibitors</p><p>aObtained in absence of Triton X-100</p><p>Docking Results. a Predicted docking pose of SM1842/RR505 (gold) in DUSP5 PD (blue), using Autodock 4.2. The inset image shows predicted binding position relative to the rest of the protein. The side chains around the bound ligand (mostly arginine guanido groups) are delineated in light turquoise and the catalytic cysteine is displayed in yellow. Three arginine residues are observed around one sulfonate group of SM1842/RR505. The calculated binding energy for this pose was -9.69 kcal/mol and had a cluster population of 10. b Optimal overlay of SM1842/RR505 (gold) and naphthalene trisulfonate (NTS, moss green), using OpenEye Scientific Software ROCS v. 3.0 [26]. c Lowest energy binding pose for NTS (moss green) in DUSP5 PD (blue), with a calculated binding energy of -8.48 kcal/mol with a cluster population of 7. d Second lowest energy binding pose for NTS (seafoam), with a calculated binding energy of -8.21 kcal/mol. e Ligplot drawing of SM1842/RR505 in the DUSP5 PD binding pocket, showing key interactions</p><p>Michaelis-Menten Kinetics. a Michaelis-Menten plot of DUSP5 PD(WT) initial velocity versus substrate (pNPP) concentration, monitoring production of p-nitrophenolate at 405 nm. Reaction was in 100 mM Tris-HCl (pH 7.5), 100 mM NaCl, 5 mM MgCl2 and 1 mM DTT, and was initiated with enzyme. The line represents a nonlinear least squares fit to equation 1. b Enzymatic rate as a function of DMSO concentration (% v/v), and at a fixed level of pNPP (5 mM), with other conditions as in panel (a). Relative enzyme activation represents the rate normalized to that obtained at 0 % DMSO</p><!><p>To measure IC50 values, a control inhibitor was used, and the assay was performed with the addition of a known broad-spectrum inhibitor of phosphatases, sodium orthovanadate (vanadate) [29]. Initial experiments were performed in 1 mL cuvettes and concentrations of vanadate were varied. Initial velocities from kinetic reads were plotted as a function of the log of vanadate concentration and fitted to equation 3, to obtain the IC50 of 88 ± 8 nM (Additional file 1: Figure S4).</p><!><p>DUSP5 PD(WT) pNPP enzymatic assay data for Z' calculation (96 well plates)</p><p>IC 50 Measurements. a DUSP5 PD(WT) initial velocity versus inhibitor concentration, and fitted to equation 3 to obtain IC50 values (Table 1). Conditions were as described for Fig. 3. (b) Same as panel (a), but comparing suramin and NTS, demonstrating the affinity increase that is obtained due to tethering the NTS fragments</p><p>Effect of Detergent on Suramin Inhibition. a NTS IC50 measurement in the presence and absence of 0.5 % Triton, showing no detergent-effect on inhibition of DUSP5 PD(WT) pNPP phosphatase activity. b Suramin IC50 measurement in the presence and absence of 0.5 % Triton X-100 shows a loss of some inhibitory capability in the presence of detergent. c Effect of increasing detergent levels (Triton X-100) on rate of DUSP5 PD(WT) in the presence of a fixed concentrations of inhibitor and substrate (at 1 μM suramin and 5 mM pNPP). Detergent removes some, but not all, of suramin's inhibitory effect, showing a plateau level 30 % inhibition</p><!><p>We additionally confirmed the potential aggregating effects of suramin and other compounds using nephelometry methods. When an inflection occurs in the Relative Nephelometry Units, measured as a function of compound concentration, this indicates that the particle size in solution is increasing, due to aggregating effects. RR505, RR506, NTS, and suramin were subjected to nephelometry measurements (Additional file 1: Figure S6). Suramin appears to start forming aggregates around 25 μM, while RR505, RR506 and NTS do not appear to form aggregates. This is consistent with the Triton X-100 studies that suggest that suramin inhibition, unlike NTS, is at least partially due to aggregation effects.</p><!><p>Pharmacophore-based Identification of DUSP5 PD Inhibitors. a Crystal structure of DUSP5 PD(C263S) [16], showing the two bound sulfate ions in the two anion-binding pockets postulated to be occupied by the two phosphate groups of the ERK2 activation loop (pThr-Glu-pTyr) [16–18]. The anion pocket closest to the catalytic nucleophile (Cys263) is labeled S1, and the distal anion pocket is labeled S2. The S2 anion (sulfate) is stabilized be several arginine residues, while the S1 anion may derive some helix dipole stabilization by virtue of its location at the N-terminal end of a long central helix. The sulfur to sulfur distance of 7.2 Å defines the DUSP PD pharmacophore as two anionic groups separated by ~7 Å. Overlay of the S1-S2 pharmacophore (two sulfates, shown as purple) on RR505 indicates a poor match, while (b) overlay on NTS (c) in one of two possible orientations (related by a 180° rotation) is better. d A ligand-based search using this pharmacophore identified CSD 3 _2320, which also matched the S1-S2 sulfate positions well. The overlay in panel (d), as in panel (c), is shown in one of the two possible orientations that optimally align active site sulfate and ligand sulfonate groups. e Flow chart summarizing the docking and ROCS alignment procedures used to identify lead molecules. Once SM1842/RR505 was identified from the CSD3 Library, it was used as a ROCS query and searched against the CSD3 Library and ZINC Library. NTS was identified from the ROCS search. NTS was used as a ROCS query to search Drugbank, which led to identification of Suramin</p><p>CSD 3 _2320 binding to DUSP5 PD. a Dose response curve for CSD 3 _2320 as an inhibitor of the DUSP5 PD(WT) phosphatase activity, using pNPP as substrate. Experimental conditions as in Figs. 3 and 4. Chemical structure of CSD 3 _2320 in the insert. b Dose response curve for CSD 3 _2320 as an inhibitor of the DUSP5 (full-length protein) phosphatase activity, using pERK2 as a substrate. c DUSP5 PD(C263S) 1H-15N HSQC spectrum of DUSP5 PD( C263S) in pH 6.8, 50 mM potassium phosphate, 100 mM potassium chloride buffer. Overlay is of 500 μM 15 N-labeled DUSP5 PD alone (black), and in the presence of 500 μM CSD 3 _2320 (red). Potentially important chemical shift perturbations due to binding are indicated using arrows. d The model from Fig. 1, with CSD 3 _2320 positioned such that its two sulfonate groups are optimally overlaid with the two phosphate groups on the ERK2 pThr-Glu-pTyr peptide. This overlay results in the phenolic ring of the CSD 3 _2320 naphthalene core being superimposed directly on the tyrosine phenol ring of the pThr-Glu-pTyr peptide</p><!><p>Docking into the phosphatase domain (PD) of the full-length DUSP protein (Fig. 1) and ROCS alignment calculations have identified various polysulfonated aromatic compounds, with both carbazole and naphthalene scaffolds (Fig. 2; Table 1; Fig. 6e). In order to determine affinity of compounds identified using docking studies, an enzyme inhibition (IC50) assay was developed whereby dephosphorylation of pNPP is monitored. pNPP was found to be a substrate for DUSP PD(WT) with a Km of 7.6 mM (Fig. 3), apparently serving as a mimic of the natural substrate, the phospho-tyrosine of ERK2. The DUSP5 PD(WT) IC50 assay using pNPP as substrate (Fig. 4) was adapted, optimized and validated as a high throughput screening (HTS) assay, and was found to be suitable for HTS with a Z' value > 0.7 (Table 2). Enzymatic screening of compounds identified by docking identified a number of weak-binding polysulfonated inhibitors that could be used as drug lead scaffolds (Table 1; Fig. 6), off of which more potent lead molecules could be developed by rational drug design or by fragment-based drug design techniques, if proximal binding pockets can be identified.</p><!><p>We have also employed a novel approach, to transition from initial lead molecules (SM1842/RR505 and RR506; Figs. 2 and 4) to an advanced clinical candidate, by screening for FDA approved compounds that match the shape and electronic properties of a lead molecule, using ROCS. Based on this ROCS overlay, naphthalene trisulfonate (NTS) was found to match the shape and electronic properties of RR505 (Fig. 2b); and, NTS was found to be present in the suramin (Table 1), an FDA approved drug available from Centers for Disease Control for treating African sleeping sickness [40]. Suramin is a competitive inhibitor versus pNPP, binding with a Ki of 25 μM (Additional file 1: Figure S5).</p><p>While initially promising, suramin does not exhibit properties of a good drug lead molecule, even though it is FDA approved. In particular, while suramin is a reasonably potent competitive inhibitor, it also causes non-specific aggregation. Based on DUSP5 PD(WT) IC50 assays performed with and without detergent (Fig. 5), along with nephelometry studies (Additional file 1: Figure S6), we conclude that while suramin does inhibit by direct binding to the phosphatase domain (Figs. 5b and c), it also forms aggregates in vitro which can lead to additional non-specific protein inhibition effects. This aggregation phenomenon raises more global concerns regarding the current clinical use of suramin, and may in part explain some of the known toxicity associated with suramin [40]. Indeed, literature on suramin [35, 41–44] indicates that it can bind to many protein targets, so may lack specificity in its mechanism of inhibition.</p><!><p>We identified the S147P mutation in DUSP5 in patients with vascular anomalies [2], which results in a mutant hypoactive protein [8]. This mutation thus presumably results in increased pERK levels in the "putative causative cell," whose identity is unknown for now. It is presumptive to imply that this mutation is causative because: (a) most diseases are not the result of a single aberration in a gene product, (b) single gene knockouts in mice and its subsequent phenotype does not necessarily imply causative role in disease, but perhaps the potential functions of the gene product in different tissues, and (c) finally, the etiology of disease, and the context of the mutation in the disease needs to reconciled, which is often not considered. For example, in vascular anomalies such as hemangiomas, which are thought to be inborn errors during embryonic development, there are two phases: the first phase is the increased proliferative phase or the rapid growth phase, and the second phase is the involution or the regression phase. The cellular dynamics, behavior and local milieu in the two phases are likely to be distinct. Whether DUSP5 functions in the early or later phase is not known. Because the proliferative phase is the initial phase, and p-ERK is involved in cell proliferation [5], therefore the natural presumption is that DUSP5 is involved in the first phase. Therefore, our attempts to inhibit DUSP5 could stop the disease in the first phase. However, if inhibiting DUSP5 accelerates the disease in the first phase as the putative tumor suppressor role of DUSP5 would suggest, then, perhaps the involution second phase of hemangiomas could be triggered earlier assuming that the two phases are linked by a common mechanism involving DUSP5. Therefore, the benefit of inhibiting a "putative tumor suppressor," such as DUSP5, and in turn accelerating the disease etiology to a phase where the disease regresses is counterintuitive. It is noteworthy that loss of DUSP5 does increase apoptosis of endothelial cells [2, 45], suggesting that DUSP5 as a survival factor for ECs. This perhaps occurs in the regression phase of the hemangioma disease. The debate as to whether to develop activators or inhibitors of DUSP5 is therefore context dependent, and probably both have benefits in specific stages of disease. Irrespective of the strategy, phosphatases as targets for drug discovery present their unique challenges as highlighted in the findings in this manuscript. Although we rationalized on developing DUSP5 inhibitors for vascular anomalies, it is becoming increasingly clear that DUSP5 inhibitors could be viable for other conditions especially those associated with immune system. Recent publications [6, 7] have demonstrated a role for DUSP5 in the immune system. Our unpublished work (Kutty, R, Ramchandran, R. et al.) also supports these findings. These studies together underscore the importance of DUSP5 in a wide array of phenotypes in different tissue types, with likely more to be discovered in the future.</p><!><p>While protein-based methods (i.e. docking) have identified a series of weak binding polysulfonated lead molecules (Table 1) for DUSP5, and lead-hopping with ROCS has identified the FDA-approved drug suramin, none of these are viable drug leads without further modification. Thus, more lead molecules and analogs are needed. An interesting feature of all these weak-binding lead molecules is the presence of at least two charged sulfonates, separated by 6–9 Å (Table 1). This led us to hypothesize that this trend is occurring because the active site pocket of DUSP5 PD binds a peptide loop from ERK2 containing two phosphates, so is designed to accommodate two negatively charged functionalities separated by this approximate distance. Indeed, DUSP5 PD was found to crystallize with two sulfate anions bound, at an S-S distance of 7.2 Å (Fig. 6). These observations led us to conclude that the key pharmacophore feature for DUSP5 PD binding is two negatively charged groups (such as sulfates or sufonates, tethered by a core scaffold (carbazole and naphthalenes have been identified herein). Negatively charged functional groups are commonly observed on phosphatase inhibitors, but are also associated with poor ability to penetrate cell membranes. Indeed, the polysulfonate compounds identified herein did not show activity in our preliminary assays using human umbilical vein endothelial (HUVEC) cells, which we speculate is due to their inability to penetrate cell membranes. Thus, future studies will be directed to substituting the sulfonates with functional groups that are more likely to penetrate into cells, such as carboxylates, tetrazoles or sulfonamides. Using this pharmacophore feature of two negatively charged groups separated by 7.2 Å in a ligand-based screen, a naphthalene-based disulfonate compound, CSD3_2320, was identified (Fig. 7). CSD3_2320 has an IC50 of only 33 mM if assayed using the phosphatase domain alone (Fig. 7a), but 33 μM if assayed using the full-length DUSP5 with ERK2 as substrate (Fig. 7b). CSD3_2320 is unique, in that it is the only compound tested that showed such a dramatic difference in IC50 values when measured in the two assays, indicating that it is especially sensitive to conformational differences that may exist in the binding site pocket in the full length versus the isolated phosphatase domain. Supporting this argument is the fact that the full length DUSP5 protein also contains an ERK binding domain, tethered via a flexible linker (Fig. 1). Also, the native substrate for DUSP5, namely the ERK2 protein, is much larger and capable of a wider range of inter-molecular interactions than the pNPP substrate, which is intended only to mimic the phosphotyrosine of pERK. Thus, while the DUSP5 PD(WT)/pNPP assay is a useful preliminary screen, a subsequent assay using full-length DUSP5 and ERK2 substrate provides the in vitro "physiologically relevant" assessment of potency for a lead molecule. Importantly, CSD3_2320 shows no tendency to aggregate. Thus, the 7.0–7.5 Å—separated disulfonate is a consistent pharmacophore feature for inhibition of the DUSP5 PD (Table 1), which shows some dependence on the presence of intact DUSP5 protein versus use of just the phosphatase domain. These and other features are part of ongoing studies to further improve the potency of CSD3_2320.</p><!><p>This study illustrates the challenges associated with structure-based drug design applied to dual-specificity phosphatases, which have a preference for highly charged ligands. Screening results presented herein typically yielded polysulfonated aromatic compounds with charged groups separated by ~7.2 Å, and included the FDA-approved drug suramin (Fig. 1). While polysulfonated aromatic compounds often aggregate like suramin, careful secondary screens using nephelometry and detergent have allowed for the identification of authentic competitive inhibitors, such as CSD3_2320. The potency of CSD3_2320 under the more biologically relevant conditions of the DUSP5 (full-length)/pERK2 assay is 33 μM. CSD3_2320's sulfonate groups are positioned 7 Å apart, to mimic the two phosphates on the ERK2 tripeptide substrate (pThr-Glu-pTyr). CSD3_2320 is a suitable scaffold upon which to build more potent and selective DUSP5 inhibitors; but, in any such inhibitor optimization effort, it will be crucial to perform secondary assays using the full-length DUSP5 protein, and using nephelometry and detergent screens to eliminate compounds that show the nonspecific aggregation effects common to sulfonates.</p><!><p>The data supporting results in this article are included in the article, and in supplementary materials.</p><!><p>Supplementary Figures. (DOCX 401 kb)</p><p>Center for Structure-based Drug Design and Development</p><p>Dimethyl sulfoxide</p><p>Dithiothreitol</p><p>Dual-specificity phosphatase</p><p>Extracellular regulated kinase</p><p>ERK binding domain</p><p>Heteronuclear single quantum coherence</p><p>Isopropyl β-D-1-thiogalactopyranoside</p><p>Luria-Bertani</p><p>Mitogen-activated protein kinase</p><p>Nuclear magnetic resonance</p><p>Naphthalene trisulfonate</p><p>Phosphatase domain</p><p>Phospho-ERK</p><p>p-nitrophenol phosphate</p><p>Root mean square deviation</p><p>Rapid Overlay of Chemical Structures</p><p>Wild type</p><p>2-dimensional</p><p>Rajendra Rathore, Ramani Ramchandran and Daniel S. Sem contributed equally to this work.</p><p>Competing interests</p><p>The authors confirm that they have no competing interests.</p><p>Authors' contributions</p><p>TN: Intellectual contribution to project design and interpretation of results; conducted docking studies, ROCS calculations and NMR titrations; aided in figure development and manuscript preparation. EAS: Intellectual contribution to project design and interpretation of results; aided in the design and development of the pNPP assay; implemented the high throughput screen and Z' measurements; conducted PD protein purification. KK: Conducted pNPP and nephelometry assays; performed PD protein purification. RB: Intellectual contribution to project design and interpretation of results; conducted screening assays, including IC50 measurements; conducted PD protein expression and purification; Aided in figure and manuscript preparation. AJG: Conducted enzymatic characterization, pNPP assays and western blotting; aided in figure and manuscript preparation. ML: Conducted the Western assay using full-length DUSP5 protein; intellectual contribution to project design and interpretation of results; aided in figure preparation. RGK: Contributed to the development and optimization of the Western assay using full-length DUSP5 protein; intellectual contribution to project design and interpretation of results. JN: Conducted full-length DUSP5 protein purification and optimization. CB: Conducted DUSP5 PD protein purification and 15 N labeling of PD for NMR titrations. RL: Implemented high throughput screen using automation, and performed Z' measurements. MIS: Performed organic synthesis, purification and characterization of RR compounds. MRT: Conducted molecular dynamic simulations; intellectual contribution to protein. Modeling studies and interpretation of results; aided in manuscript preparation. RR: Key role in guiding group members in the organic synthesis and characterization of DUSP5/ERK inhibitors; also in molecular dynamic simulations; aided in manuscript preparation. RR: Key role in guiding group members in the characterization of DUSP5 activity and purification; aided in manuscript preparation. DSS: Key role in guiding group members in the characterization of DUSP5 inhibition as well as docking and ROCS calculations; aided in all aspects of manuscript preparation. All authors read and approved the final manuscript.</p>

PubMed Open Access

Direct introduction of nitrogen and oxygen functionality with spatial control using copper catalysis

Synthetic chemists have spent considerable effort optimizing the synthesis of nitrogen and oxygen containing compounds through a number of methods; however, direct introduction of N-and O-functionality remains challenging. Presented herein is a general method to allow for the simultaneous installation of N-and O-functionality to construct unexplored N-O heterocyclic and amino-alcohol scaffolds. This transformation uses earth abundant copper salts to facilitate the formation of a carboncentered radical and subsequent carbon-nitrogen bond formation. The intermediate aminoxyl radical is terminated by an intramolecularly appended carbon-centered radical. We have exploited this methodology to also access amino-alcohols with a range of aliphatic and aromatic linkers.Scheme 1 Examples of strategies that enable the direct installation of nitrogen and oxygen heteroatoms and examples of biologically active products that relied on these methods.

direct_introduction_of_nitrogen_and_oxygen_functionality_with_spatial_control_using_copper_catalysis

<!>Conclusions

<p>Historically, cycloaddition reactions have provided an efficient strategy to build molecular complexity into heterocycles by taking advantage of multiple bond formations in a single step. Given the ubiquity of C-N and C-O bonds in biologically active molecules, the ability of nitrones and nitroso compounds to directly install nitrogen and oxygen heteroatoms is of particular importance. 1 Moreover, due to the labile nature of the N-O bond, these transformations have also served as strategic approaches for the synthesis of amino alcohols bearing a 1,3 or 1,4-relationship. 2 Despite the prevalence of these transformations in organic synthesis, most methods have been restricted to the construction of isoxazoline (Scheme 1A) 3 and 1,2-oxazine (Scheme 1B) 4 heterocyclic scaffolds and the corresponding amino alcohols upon reduction. To date, there is no unied method for the synthesis of N-O heterocycles with varying ring sizes (small to large) or a direct approach to construct amino-alcohols with spatial control without independently installing the N-and O-functionality in sequential steps.</p><p>Previously, in an effort to provide alternatives to cycloaddition reactions or electrophilic functionalization of carbonyls using nitroso compounds, we 5 and others 6 described the use of radical transformations with nitroso compounds to construct sterically hindered amines. This process employs earth abundant copper salts, tolerates a range of functional groups and employs widely available radical precursors. Here we report the development of a generalized method to construct N-O heterocycles and amino-alcohols of any size and distribution. These studies demonstrate that bi-molecular reactions are not necessary to trap the in situ generated aminoxyl radical, despite the well-known challenges of forming larger macrocyclic rings. 7 In addition and in contrast to our previous work, this method also increases the atom economy of the nitroso additions, accessing products that incorporate both heteroatoms. Previously, the C-O bond constructed during the radical transformation was treated as part of the waste stream and discarded upon N-O bond cleavage. Combined, this radical-based process provides efficient entry into many unexplored scaffolds (Scheme 1C).</p><p>To begin our investigations, we examined the intramolecular reaction of the 1,3-dibromide scaffold. Initially, we were able to identify conditions inspired by our previous work and others 8 (5 equiv. of both Cu I and Cu 0 , 2.5 equiv. of PMDTA, 2 equiv. of nitrosobenzene, THF, 40 C) that afforded the desired N-O heterocycle 1 in 70% yield. We were further encouraged to nd that, in a two-step-one-pot approach, the heterocycle could be reduced to the corresponding amino-alcohol 15 (65% overall yield) by simply adding additional Cu I and ascorbic acid. Furthermore, the heterocycle was formed in a 2 : 1 ratio of diastereomers (dr), favouring the cis isomer over the trans, and the N-O bond reduction did not erode the selectivity. Through optimization of the reaction parameters, we found that Cu 0 could be removed entirely, Cu I loading could be reduced to 2 equivalents, and nitrosobenzene loading could be lowered to 1.5 equivalents (see ESI, Table S2 †). These modications increased the yield of the desired product (1) to 85%. Unfortunately, we discovered that reduction of the N-O heterocycle with Cu I and ascorbic acid was only useful for ve-membered ring heterocycles, with incomplete reduction occurring when larger rings were investigated. A screen of various reducing conditions revealed that stronger reducing agents such as zinc in HCl and sodium-naphthalenide afforded the desired amino alcohol 15 in higher yield (67% isolated yield over two steps using Zn/HCl conditions) and these methods proved general. Notably during optimization studies, we discovered that increasing the reaction temperature to 50 C increased the dr of this transformation to 5 : 1 favoring the cis-isomer. Finally, a copper ligand screen was investigated; reactions run with the more activating ligands such as Me 6 TREN provided yields very similar to those run with PMDTA. However, using a less activating ligand, such as 2,2 0bipyridyl, resulted in limited or no conversion of the starting material.</p><p>With optimized conditions established, we initially explored the generality of this method to construct N-O heterocycles with varying ring sizes (Fig. 1A). Five (1) and six (2) membered rings were synthesized in good yields with the optimized conditions. The seven-membered ring (3) required more dilute reaction conditions, as increasing amounts of oligomers were observed by 1 H-NMR spectroscopy, presumably formed via a competitive intermolecular radical termination. The larger 8-12 membered heterocycles (4-7) required the same dilute reaction conditions, as well as the addition of 5 mol% copper(II) bromide relative to the copper(I) bromide. Cu II is known to have a strong effect on the kinetics of atom transfer radical polymerization (ATRP) systems 9 and we hypothesize that the addition of Cu II decreases the concentration of carbon-centered radicals, leading to a more controlled reaction. As expected when forming larger macrocyclic ring systems, the stereoselectivity of the transformation decreases as the spacer length increases. The vemembered ring 1 demonstrated a relatively high dr of 5 : 1 cis:trans, while the six-and seven-membered rings 2 and 3 demonstrated dr's of 1.8 : 1 and 1.5 : 1, respectively. 10 Rings eight-membered and greater demonstrated no selectivity. Alkylnitroso compounds were used to create heterocycles with yields similar to their aromatic counterparts; compound 8 was synthesized using the commercially available 2-methyl-2-nitrosopropane dimer. We were pleased to nd that the intramolecular reaction could be extended to readily available a-bromo carbonyl-based scaffolds. Impressively, as shown in Fig. 1B, these scaffolds were found to cyclize very efficiently, creating up to 19-membered heterocycles in great yield (9-14). Overall, these results open the door for efficient access to a series of unexplored N-O based heterocyclic scaffolds.</p><p>We were intrigued by the large discrepancy in yields between the glycol-linked 10-14 and the alkyl-linked 1-5 substrates, and considered that a Cu II templating effect was responsible. Cu II has been employed advantageously in a number of similar cyclizations. 11 To test this hypothesis, we synthesized an alkyllinked 18-membered heterocycle that cannot benet from templating and subjected it to optimized conditions (see ESI, S25 †). Compared to the closest derivatives, compounds 13 and 14, the yield dropped from greater than 60% to 32%. This direct comparison indicates that a templating effect might be responsible for the increased yields observed with substrates 10-14.</p><p>Aer demonstrating the construction of N-O heterocycles with spatial control, we were now set to examine the scope of our two-step-one-pot approach to construct amino-alcohols of various distributions (Fig. 2). We were pleased to nd that many of the yields are actually higher for the amino-alcohols using the two-step-one-pot approach than those of the corresponding N-O heterocycle. For example, synthesis of an amino-alcohol (21) bearing a 1-10 relationship, which represents the direct installation of both N-and O-functionality over 12 angstroms of space, afforded the desired product in 48% overall yield. Notably, this is 20% higher than that of the corresponding N-O heterocycle (7, 20% yield). We speculate this is due to the in situ reduction of oligomers that also afford the desired aminoalcohol product 21. Previously, the oligomers were removed during the heterocycle purication and isolation. For the in situ reduction, the ve-membered heterocycle affording aminoalcohol 15 and 22 was reduced using Zn/HCl conditions, but all others were reduced using sodium-naphthalenide.</p><p>Next, we explored how structural modications to the nitrosoarene and the dibromide architecture were tolerated. Given the higher yields of amino-alcohol synthesis, the twostep-one-pot approach was used for these studies. A small library of both electron-rich and decient nitrosoarenes was synthesized and subjected to the optimized conditions (Fig. 3A). With respect to the nitrosoarene coupling partner, the reaction was tolerant of electronic changes. Not surprisingly, the reaction tolerates halogenated compounds 26 and 28 that allow for facile downstream functionalization. Of note, the amine functional group (NH 2 ) group in substrate 27 is derived from the corresponding nitro group and was generated in situ upon treatment with zinc and HCl conditions. Moreover, structural changes can be made to the dibromide scaffold, either the methylene linker or the aromatic rings, affording the anticipated product in moderate to good yields (40% to 66%) (Fig. 3B). Notably, the reaction efficiency decreased slightly when gem-dimethyl groups are introduced alpha to the dibromide (30). This is not surprising considering the costly steric interactions of forming C-N and C-O bonds adjacent to a quaternary center. Interestingly, while most of the modications to the scaffold had limited effect on the diastereomeric ratio of the products (3 : 1 dr ratio was observed for 29-31, 33), compound 32 was formed in a 10 : 1 dr, suggesting that diastereoselectivity can be enhanced using substitution at the meta-position.</p><p>To demonstrate the synthetic utility of this methodology beyond symmetrical substrates, we investigated strategies to construct N-and O-bonds on unsymmetrical scaffolds with regioselective control. A common feature of radical reactions with nitroso compounds is that the initial carbon centered radical reacts with nitrogen. Consequently, we hypothesized that radical initiation rates could be leveraged to control the regioselectivity. The success of this approach would also require a second intramolecular radical reaction with the intermediate aminoxyl radical to outcompete the intermolecular reaction. Despite the challenges of balancing the reaction rates of these highly reactive radical intermediates, we were encouraged by the wealth of literature on activation rates for various initiators used for ATRP. 12 Guided by these activation studies, we designed a mixed-initiator scaffold containing both an a-bromoester and a benzyl-bromide radical precursor which could be synthesized in one step from styrene and ethyl dibromopropanoate (Fig. 4). The k act of the a-bromoester moiety is roughly an order of magnitude greater than that of the benzyl bromide under standard ATRP conditions. 13 Given this difference, we predicted that the initial radical would predominately form at the a-bromoester, leading to carbon-nitrogen bond formation a to the ester and carbon-oxygen bond formation at the less active benzyl site. To our gratication, subjection of the unsymmetrical scaffold to the optimized reaction conditions resulted in the N-O heterocycle with a 10 : 1 ratio of products 35 to 36 favouring the predicted major isomer. This result indicates that the major regioselectivity can be predicted through the relative k act of each radical precursor; moreover, the approximate ratio of the regioisomers can be predicted from the ratio of the k act of the initiators. Further studies are underway to elucidate these factors in more detail and explore the scope of unsymmetrical scaffolds.</p><!><p>In summary, we have developed a new method for the direct installation of nitrogen and oxygen functionality where N-O heterocycles and amino-alcohol scaffold size are unencumbered by traditional olen coupling reactions. The described method is general in terms of scope and provides an efficient method capable of construction macrocycles up to 19-members in size and amino-alcohols with up to 12 Å separating the N-and Oheteroatoms. The reaction is catalysed by copper salts and leverages readily available radical precursors and nitroso compounds to generate a new C-N bond and an intermediate aminoxyl radical, which is subsequently terminated with a second intramolecularly appended radical. Moreover, we have shown that the regioselectivity of the installation of nitrogen and oxygen functionality can be predicted using well-documented ATRP rate constants for radical formation. The method reported herein provides a new versatile platform for the development of N-O heterocycles and the corresponding amino-alcohols, all with high atom economy and earth-abundant catalysts.</p>

Royal Society of Chemistry (RSC)

Nucleated polymerization with secondary pathways II. Determination of\nself-consistent solutions to growth processes described by non-linear master\nequations

Nucleated polymerisation processes are involved in many growth phenomena in nature, including the formation of cytoskeletal filaments and the assembly of sickle hemoglobin and amyloid fibrils. Closed form rate equations have, however, been challenging to derive for these growth phenomena in cases where secondary nucleation processes are active, a difficulty exemplified by the highly non-linear nature of the equation systems that describe monomer dependent secondary nucleation pathways. We explore here the use of fixed point analysis to provide self-consistent solutions to such growth problems. We present iterative solutions and discuss their convergence behaviour. We establish a range of closed form results for linear growth processes, including the scaling behaviours of the maximum growth rate and of the reaction end-point. We further show that a self-consistent approach applied to the master equation of filamentous growth allows the determination of the evolution of the shape of the length distribution including the mean, the standard deviation and the mode. Our results demonstrate the power of fixed-point approaches in finding closed form self-consistent solutions to growth problems characterised by highly non-linear master equations.

nucleated_polymerization_with_secondary_pathways_ii._determination_of\nself-consistent_solutions_to_

Introduction<!>Second Order Self-Consistent Solutions<!>Analysis of Limiting Cases<!>Long time limit<!>Absence of seed material<!>Infrangible filaments<!>Lag Time Scaling<!>Higher Order Iterations<!>Monomer Dependent Secondary Nucleation<!>Solution prior to the inflection point<!>Solution post-inflection point<!>End Point Scaling<!>Self-Consistent Solution for the Length Distribution<!>Conclusion<!>Appendix<!>

<p>There has been recently considerable renewed interest in the development of theoretical models that describe the general problem of growth of filamentous protein structures[1–7]. The theoretical analysis of such polymerising protein systems was initiated in the 1960s in the context of a range of functional biological assembly phenomena[8–13], including the growth of actin and tubulin filaments. Furthermore, much experimental and theoretical attention has focused on aberrant biological assembly, starting with sickle hemoglobin assembly[14–16], and more recently the processes leading to the formation of amyloid fibrils often observed in association with neurodegenerative and other diseases[17–24]. In the overall kinetics of growth, secondary nucleation processes, including fiament fragmentation, have emerged as dominant factors for amyloid formation[1, 3, 25–27] and prion assembly[1, 28, 29]. Such secondary events lead to a description of the growth problem through kinetic equations that are commonly highly non-linear.</p><p>In order to address the difficulty in obtaining analytical results for the full time course of the reaction, we proposed[30, 31] a self-consistent analysis scheme and showed that this approach yields closed form solutions to the growth problem. The focus of the present paper is to extend this approach to highly non-linear master equations by exploring the nature of the higher order solutions that emerge from the repeated application of the fixed-point iteration to the moment equations (Sections II–VII) and also by exploring self-consistent solutions to the full length distribution (Section VIII). We show that closed form solutions of high accuracy can be generated using this general and powerful approach. In particular, we derive analytical results, valid for the full duration of the reaction, which describe in closed form the time evolution of lower principal moments of filament systems which grow through primary nucleation, filament elongation and either monomer independent or monomer dependent secondary nucleation. These higher order iterations also allow access to important corrections to the scaling laws that emerge from the analysis of lower order solutions, as well as to information on the shape of the length distribution.</p><p>The differential equations that describe monomer-dependent secondary pathways involve very strong non-linearities in addition to those originating from the primary nucleation. These types of equations are not readily amenable to perturbative treatments as all of the non-linear terms contribute significantly to the overall reaction as shown in Fig. 2. The main result of this paper is an approach to treat such highly non-linear growth problems, and we provide an iterative scheme to obtain a closed form expression of high accuracy for the integrated rate laws (c.f. Appendix), the result of which is illustrated in Fig. 2.</p><!><p>In the first part of this paper, we focus on higher order iterative solutions for the case where the secondary pathway is filament fragmentation, and is hence monomer independent. This type of secondary pathway does not introduce non-linear terms into the master equation, and therefore the only source of non-linearity is the elongation process [terms proportional to M(t)P(t)] and the primary nucleation [terms proportional to a polynomial of degree nc in M(t)][31]. The evolution of the principal moments, the polymer number concentration P(t) and the polymer mass concentration M(t), has been shown to obey the differential equations[5, 9, 29, 34]: (1)dP(t)dt=k−[M(t)−(2nc−1)P(t)]+knm(t)nc(2)dM(t)dt=2[m(t)k+−koff−k−nc(nc−1)/2]P(t)+ncknm(t)nc where m(t) = mtot − M(t) is the free monomer concentration and k+, koff, k− and kn are the rate constants for elongation, depolymerisation, fragmentation and primary nucleation, respectively. In order to recover the detailed scaling exhibited by solutions to the problem of filament growth under conditions where filament fragmentation enhances the number of free filament ends, we require self-consistent solutions that go beyond those previously obtained as the first order corrections restoring mass conservation[30, 31] to the linearized solutions[5, 16, 32, 33] which do not include mass conservation. We base our discussion on a fixed-point analysis. The kinetic problem is reformulated as a fixed-point equation[31]. Subsequent applications of the fixed-point operator yield increasingly accurate iterative self-consistent solutions. The first improvement beyond the low order result [30, 31] is obtained by establishing the second order self-consistent solution to the filament growth problem.</p><p>The self-consistent result obtained after one iteration has been shown to emerge as the simple closed form expression[31] for the polymer mass concentration: (3)M1(t)=M(∞)[1−exp(−C+eκt+C−e−κt+D)] where the filament multiplication rate is defined as: (4)κ=2[m(0)k+−koff]k− for initial monomer concentration m(0). The final polymer mass concentration M (∞) is given approximately by[31, 35]: (5)M(∞)=2k+mtot−2koff−nc(nc−1)k−2k+ The constants are functions of the initial conditions and of the rate constants kn, k+, koff and k−: (6)C±=k+P(0)κ±M(0)k+2[m(0)k+−koff]±λ022κ2(7)D=λ02κ2−M(0)M(∞)+k+M(0)m(0)k+−koff where λ0=2k+knm(0)nc is the effective rate constant derived by Oosawa[8, 9, 31] for nucleated polymerization without secondary pathways. The aim of the present section of this paper is to generalise the equation (3) to higher order approximations and recover the corresponding corrections to the scaling behaviour in the system kinetics. The two components of the fixed-point operator 𝒜, appropriate for this system, are derived by formal integration of Eqs. (1) and (2) to yield [30, 31]: (8)M(t)=M(∞)(1−e−M(0)M(∞)−2k+∫0tP(τ)dτ)(9)P(t)=k−e−(2nc−1)k−t(P(0)+∫0te(2nc−1)k−τM(τ)dτ) The convergence of the fixed-point scheme is most effective when the operator acts on a solution which is close to the final fixed point. In our first order solutions, this condition is better satisfied for early times, and therefore the accuracy of the first order self-consistent solution is less good for times corresponding to the reaction end point. The accuracy of the expression for M(t) in Eq. (3) may be improved with higher order corrections from further fixed point iterations. This improvement corresponds to substituting the first order expression P1(t) into the integral operator for M(t) in Eq. 8 to find a more accurate solution in this self-consistent scheme.</p><p>By substituting Eq. (3) into Eq. (9), the first order result for P1(t) has been shown to obey[31]: (10)P1(t)=k−e−(2nc−1)k−t{P(0)+∫0te(2nc−1)k−τM(∞)[1−exp(−C+eκτ+C−e−κτ+D)]dτ} For sufficiently large breakage rates (C+C− ≪ 1) and at early times (t ≪ κ−1), the exponential term may be approximated as: (11)exp(−C+eκτ+C−e−κτ+D)≈exp(−C+eκτ)+exp(C−e−κτ)−1 On rearrangement and expansion for early times t≪κ−1≪k−−1 Eq. (10) yields: (12)P1e(t)=k−M(∞)e−(2nc−1)k−t{P(0)k−M(∞)−∫0t[exp⁡(−C+eκτ)+exp⁡(C−e−κτ)+2t]+𝒪(k−t)dτ} This integral has a closed form expression valid for early times: (13)P1e(t)=e−(2nc−1)k−t(P(0)+k−M(∞)κ[E1(C+eκt)−E1(C+)−E1(−C−e−κt)+E1(−C−)+2κt]) This result is accurate for significantly longer times than the linearized solution[31, 32] used to generate M1(t), and would, therefore, be expected to produce an improved expression for M(t) when operated on by Eq. (8). Indeed the correction for the first moment can now be obtained from Eq. (13) and Eq. (8) as M2(t). In this case, we obtain: (14)M2tM(∞)=1−exp(−2k+∫0t𝒪(k−t)2+k−M(∞)κ[E1(C+eκt)−E1(C+)+κt]−k−M(∞)κ[E1(−C−e−κt)−E1(−C+)−κt]dt−M(0)M(∞)) In order to carry out the integration in Eq. (14), the integrand is rewritten as: (15)I=Iy+Iz=k−M(∞)κ[log(y)+E1(C+y)−E1(C+)]−k−M(∞)κ[log(z)+E1(−C−z)−E1(−C−)] with y = eκt and z = e−κt. The terms Iy and Iz may be integrated using an identical approach; we carry out the integration for Iy. Expansions around y = 1 (t = 0) yield: (16)log(y)=∑p=1∞(−1)p+1(y−1)pp(17)E1(C+y)−E1(C+)=∑p=1∞(−1)pC+pE1−p(C+)(y−1)pp! and expanding the exponential integral about C+ = 0 gives: (18)E1−p(C+)=C+−p(p−1)!+∑q=1∞(−1)qC+q−1(q−1)!(q−1+p) Due to the convergence of the series, Iy may be obtained exactly: (19)Iy=k−M(∞)κ∑p=1∞∑q=1∞(−1)p+qC+p+q−1(y−1)pp!(q−1)!(q−1+p) Transforming the rest of the integral into to be a function of the variable y, and writing (y − 1)p as a binomial sum, the integration can be carried out exactly term by term. For small depolymerization rates[31, 36], koff ≪ k+mtot, this operation results in: (20)∫0tIydt≈−12k+∑p=1∞∑q=1∞(−1)p+q+1C+p+q−1(−1)pκt+∑r=0p−1(−1)rp!r!(p−r)!(p−r)(e(p−r)κt−1)p!(q−1)!(q−1+p) The major contribution to the summation over r in Eq. (20) originates in the leading order exponentials where r = 0 and, similarly, for C+ ≪ 1 the leading order terms in the summation over q are given from q = 1, resulting in: (21)∫0tIydt=12k+∑p=1∞(−C+)pepκtp2p! The result for Iz may be found through the replacement C+ → −C−, with the minus sign before the integral cancelling with that from the change of variable dt → dz in the integration. This yields a compact result for M2(t): (22)M2(t)M(∞)=1−exp⁡(∑p=1∞(−C+)pepκtp2p!+C−pe−pκtp2p!) We now use this solution obtained in the early time limit t ≪ κ and sufficiently large k− (C−C+ ≪ 1) to construct the full time-course solution. To this effect, we first note that the full second order solution must match in the early time limit the first order iteration result, which is itself exact in this limit[31]. Furthermore, we also know the limit that should be obtained in the absence of fragmentation and depolymerisation (Oosawa limit[9, 31]). Using these two conditions to complete the missing terms in Eq. (22) by comparison with the two special cases, the full result for the second order iteration is obtained as: (23)M2(t)M(∞)=1−exp⁡(∑p=1∞[(−C+)pp2p!(epκt−∑q=02p−2(pκt)qq!)+C−pp2p!(e−pκt−∑q=02p−2(−pκt)qq!)]−M(0)M(∞)) This equation has an analogous form to the first order solution discussed in [30, 31], but we note the presence of additional terms in the exponential. As in Part I[31], a linear term ncknmtotnc−1t has been neglected since the breakage rate is small such that most bonds in the system do not break over the reaction time.</p><p>Fig. 3 shows the first and second order iterations, given by Eqs. (3) and (23). From this figure, it is evident that the second order iteration offers a significant improvement in accuracy over the first order result in matching the exact solution - which it does almost exactly - and in particular offers significant improvement over the first order result towards the end of the growth phase.</p><p>In considering after how many terms the summation in Eq. (23) can be terminated while still offering a good approximation for M(t), first note that the summation is absolutely convergent for any given t, with the absolute ratio between the (p + 1)th and pth terms for long times: (24)up+1un=p2(p+1)3C+eκt Using the parameters of Fig. 3, for t = 14 hours the ratio of the p = 3 to p = 2 terms is ~ 0.01. This same ratio is ~ 18 at t = 60 hours. Generally, the series converges in a small number of terms for times such that C+eκt ≪ 1, but for longer times than this, a very large number of terms is needed. Due to the alternating nature of the series, if the sum is terminated after an odd number of terms then in the limit t → ∞ it will yield a large negative number; whereas for termination after an even number of terms the result gives a large positive number in this limit. Since the summation is within an exponential function in Eq. (23), for the case of termination after an odd number of terms the exponential term tends to zero and M(t) → M(∞) in the long time limit, as required. Conversely, if the sum is terminated after an even number of terms then in the limit t → ∞, M(t) → −∞. As such, to achieve a good approximation of M(t), enough terms must be included before terminating the summation such that the ratio between the last included and first omitted term, given by Eq. (24), is small for times before which M(t) ~ M(∞), and the summation must also be terminated after an odd number of terms.</p><p>Using the parameters of Fig. 3, these conditions are fully satisfied terminating the summation after 25 terms. The result including just 3 terms, offering the first improvement over the first order iteration M1(t) [31], is given explicitly by: (25)M2(t)M(∞)=1−exp(−(eκt−1)C++(e−κt−1)C−+(e2κt−1−2κt−2κ2t2)C+28+(e−2κt−1+2κt−2κ2t2)C−28−(e3κt−1−3κt−9κ2t22−9κ3t32−27κ4t48)C+354+(e−3κt−1+3κt−9κ2t22+9κ3t32−27κ4t48)C−354−M(0)M(∞)) We note that compared with the first order result M1(t)=M(∞)[1−exp(−C+eκt+C−e−κt+D)] there are new terms ~ e2κt, e3κt and associated polynomials in κt in the exponent.</p><!><p>The equation Eq. (23) is a general result valid for all initial conditions. For many special cases, however, the growth kinetics are described by expressions which are considerably simpler, and can be obtained from (23) in the appropriate limit. We discuss three examples of practical importance below.</p><!><p>The result for long time limits, for t ≫ κ−1, is given from Eq. (23): (26)M2(t)M(∞)=1−exp(∑p=1∞(−C+)pepκtp2p!) as the decaying exponentials e−pκt and polynomial terms in κt can be neglected in front of the growing exponential terms epκt.</p><!><p>When M(0) = P(0) = 0, Eq. (23) reduces to: (27)M2(t)M(∞)=1−exp⁡(∑p=1∞[(−C+)pp2p!(4sinh⁡2(pκt2)−2∑q=1p−1(pκt)2m(2m)!)])(28)C+=λ2κ2 for λ=2k+k−mtot since C+ = −C− and m(0) = mtot, and the resulting symmetry in the expression Eq. (23) allows the regrouping of the exponential terms using the double angle formula.</p><p>The time evolution of the polymer mass in this case depends only upon two combination of the kinetics parameters, κ and λ; for comparison, in the case derived by Oosawa[8, 9] for nucleated polymerization without secondary pathways, the rate law depended primarily only upon λ in the absence of seed material.</p><!><p>For a system of infrangible filaments, k− = 0, undergoing irreversible growth, koff = 0, the limit of Eq. (27) is given as: (29)M2(t)mtot=1−exp⁡(∑p=1∞(−C+)pp2p!(knk+mtotnc)p(2p)!(pt)2p) where the first few terms are given by: (30)M2(t)mtot=1−exp⁡(− t2k+knmtotnc+16t4k+2kn2mtot2nc−380t6k+3kn3mtot3nc+𝒪(t8)) In particular, higher powers of t are now present in comparison with the first order result[31]. It is interesting to note that the exact Oosawa result for irreversible growth of infrangible fiaments[9] in the absence of seeds and depolymerisation, M(t)=mtot(1−sech2/nc[ncknk+mtotnc]), admits the series expansion: (31)M(t)mtot=1−exp⁡(−t2k+knmtotnc+16t4k+2kn2mtot2nc−245t6k+3kn3mtot3nc+𝒪(t8)) It can be seen that whilst the first order result is correct to 𝒪(t2)[31], the second order iteration now reproduces the correct limit to 𝒪(t4). The limits given by the first and second order analytical results are compared with the exact result in Fig. 4, where the second order result is seen to provide a significant improvement.</p><!><p>The availability of a highly accurate analytical solution to the growth kinetics over the full time course of the polymerisation reaction provides us with the opportunity to analyse the corrections to the scaling behaviour in the lag time that has emerged from lower order solutions[30–32]. In this section, we show that the numerical values of the coefficients in the scaling laws from the first order solutions are significantly improved when considering the corrections obtained from the second order results. We recall that in the case of the first iteration, the following results hold, where we label the time at which the maximum growth rate occurs tmax: (32)tmax⁡=log⁡(1/C+)κ;rmax⁡=M(∞)κe The lag time is found to scale as: (33)M(tmax⁡)/(tmax⁡−τlag)=rmax⁡⇒τlag=tmax⁡−M(tmax⁡)rmax⁡(34)τlag=[log⁡(1/C+)−e+1]κ−1≈[log⁡(1/C+)−1.718]κ−1 The exact numerical result for this scaling from the master equation has been shown to be[37]: (35)τlag=[log⁡(1/C+)−1.825]κ−1</p><p>We now derive the equivalent result for the scaling of the lag time predicted by the second order analytical iteration, Eq. (23); as shown below, the result from the first order result is only semi-quatitatively correct, whereas the scaling laws obtained from the analysis of the second order result are highly accurate when compared with the numerical results.</p><p>In order to obtain the expressions for tmax, rmax and M (tmax) we first perform a substitution to introduce the dimensionless quantity y = C+eκt in Eq. (26) and obtain: (36)M2(y)=M(∞)[1−exp⁡(∑p=1∞(−1)pypp2p!)] The value of y that occurs when t = tmax is ymax: (37)ymax⁡=C+eκtmax⁡ whereby the explicit dependencies on C+ and κ have been subsumed into the change of variable. Explicitly, tmax is given by: (38)tmax⁡=log⁡(ymax⁡C+)κ=log⁡(1C+)+log⁡(ymax⁡)κ Remarkably, this form maintains the inverse correlation between the multiplication rate κ and the lag time, but introduces a correction to the proportionality constant. Note that ymax = 1 recovers the first iteration result.</p><p>The determination of ymax requires the evaluation of the inflection point of M2(t) Eq. (36) Under the substitution dy/dt = κy: (39)d2M2dt2=k2y(yd2M2dy2+dM2dy) Hence the inflection point is found as: (40)(d2M2dy2+1ydM2dy)y=ymax⁡=0 After substituting for the derivatives, Eq. 40 becomes a polynomial equation for ymax: (41)0=(∑q=1∞(−1)qymax⁡q−1qq!)2+∑q=1∞(−1)qymax⁡q−2q!</p><p>This equation must be solved numerically and results in: (42)ymax⁡=0.99616 This value determines the maximum growth rate, rmax = (dM/dt)t=tmax = κy(dM/dy)y=ymax, as: (43)rmax⁡=0.3267M(∞)κ which may be compared with the first order and exact numerical results to show a significant improvement over the first order: (44)rmax⁡={0.3679M(∞)κFirstiteration0.3267M(∞)κSeconditeration0.3182M(∞)κExactnumerical</p><p>Substituting ymax into Eq. 33 yields the lag time scaling: (45)τlag=[log⁡(1/C+)+s]κ−1 where the constant s is defined by: (46)s=log⁡(ymax⁡)+1−exp⁡(∑p=1∞(−1)pymax⁡pp2p!)(∑p=1∞(−1)pymax⁡ppp!)exp⁡(∑p=1∞(−1)pymax⁡pp2p!)=−1.8053 Therefore: (47)τlag=[log⁡(1/C+)−1.805]κ−1</p><p>As a consistency check we can verify that setting the upper limits of the summations to unity and setting ymax = 1 recovers s = −e + 1 ≈ −1.718, which is the first iteration result[31], as expected. Summarising the three results: (48)τlag={[log⁡(1/C+)−1.718]κ−1Firstiteration[log⁡(1/C+)−1.805]κ−1Seconditeration[log⁡(1/C+)−1.825]κ−1Exactnumerical The second order result is a substantial improvement over the first with the numerical constant now correct to within 1% of the exact result.</p><!><p>In this section, we discuss briefly the convergence behaviour of the fixed-point scheme beyond the second order solutions for the case of fragmenting filaments. In the fixed-point analysis presented so far, the fixed-point operator neglects terms 𝒪(kn). This is an accurate approximation for the first two self-consistent solutions in the case of fragmenting filaments which we have presented, since the influence of the primary nucleation comes mainly from its effect on the linear solutions used as the starting point of the fixed point scheme. Whilst the rate of creation of filaments from fragmentation increases as the reaction proceeds, the rate of creation of filaments from primary nucleation slows as monomer is depleted. Hence primary nucleation is most important for earlier times, and its relative important compared to secondary nucleation decreases monotonically.</p><p>Repeated iteration, however, using our fixed point scheme, would result in higher order solutions that ultimately converge towards the solution of the master equation neglecting 𝒪(kn); in other words, further iterations will converge towards the solution of our model with kn set equal to zero, as if there were no primary nucleation.</p><p>A strategy to correct for this convergence behavior in higher order iterations is to develop, or seed, the system using the linearized early time solutions Me(t) and Pe(t) derived in Part I[31], which include 𝒪(kn), for as long as the early time solution is a good approximation to the exact solution. If the time for which the early solution remains a good approximation is denoted te, for times greater than te the fixed point iteration scheme is then used to find M(t), but now using Me(te), Pe(te) from the early time solutions as the initial conditions. The final solution, which converges to the solution including 𝒪(kn), is then given by the early time limit result continued piecewise to the higher order fixed point iteration at te.</p><p>This completes our discussion of the polymerisation when the secondary pathway is concentration independent and given by filament fragmentation.</p><!><p>In the second part of this paper, we focus on polymerisation processes where the secondary pathway is concentration dependent[4, 32]. The analysis of this type of growth process was pionered by Eaton and Ferrone[16, 32] in their analysis of sickle hemoglobin gelation. Linear solutions can be obtained much in the same way as for fragmentation, but this type of process introduces more non-linear terms into the master equations which become relevant when mass conservation is enforced. Consequently, such processes are more challenging to treat than the fragmentation case, and we anticipate less rapid convergence of the fixed point scheme. In addition, since the secondary nucleation term is now also affected by monomer depletion, the relative importance of primary and secondary nucleation may vary in a more complicated way than in the fragmentation case where the relative importance of primary to secondary nucleation decreases monotonically as the reaction proceeds. In the case of monomer-dependent secondary nucleation, the inclusion of primary nucleation only in the early time limit is expected to be insufficient in describing accurately the full time course of the reaction.</p><p>For the case of monomer dependent secondary nucleation, the equations for the moments become[4, 16, 32]: (49)dP(t)dt=k2M(t)m(t)n2+knm(t)nc(50)dM(t)dt=2(m(t)k+−koff)P(t)+ncknm(t)nc+n2k2M(t)m(t)n2 The first order result for the kinetics of this process was shown to be[30, 31]: (51)M1(t)≈M(∞)[1−exp(−C+eκt+C−e−κt+D)] for the polymer mass concentration M1 with κ=2m(0)n2[m(0)k+−koff]k2 and C±≈k+P(0)κ±k+M(0)2[m(0)k+−koff]±λ022κ2, D=λ02κ2−M(0)/M(∞)+k+M(0)m(0)k+−koff. The long-time limit is given by M(∞) = M(∞) ≈ mtot − koff/k+.</p><p>Whilst this first order result is a significant improvement over the linear results previously known, it becomes increasingly inaccurate as n2 is increased as shown in Fig. 2. The strategy for finding a more accurate solution is first to perform a second order iteration, following the fragmentation case, neglecting 𝒪(kn). This result will be accurate for early and intermediate times, but would be expected to be inaccurate for later times, in particular for high values of n2 and intermediate values of kn. As a general strategy, the accuracy of the solution at later times can be improved by using the solution itself to provide a new linearization as an input for a subsequent fixed-point scheme that will be accurate at later times. An additional advantage of this two part approach is the possibility of including explicitly the primary nucleation terms in the fixed point operator for later times. Our two solutions, each requiring only a small number of terms to remain convergent over a smaller time range, can then be continued piecewise together at an appropriate time given in closed form; in this paper we will chose the point of inflection of P(t) for this purpose.</p><!><p>In analogy with the fragmentation case, the second order solution in our original fixed point scheme can be obtained from: (52)M(t)=M(∞)[1−exp⁡(−M(0)M(∞)−2k+∫0tP(τ)dτ)] Following the fragmentation case, we first derive a result for P1(t) for early times. To obtain the first iteration result for P(t), we substitute Eq. (51) into Eq. (49). Neglecting terms 𝒪(kn), the equation to be solved for P1(t) becomes: (53)dP1e(t)dt=k2M1(t)m1(t)n2 We substitute M1(t) = mtot – m1(t) using conservation of mass. This reduces the moment equation to: (54)dP1e(t)dt=k2mtotm1(t)n2−k2m1(t)n2+1 where the monomer concentration m(t) can be approximated for small depolymerisation rates by making use of Eq. (51): (55)m1(t)≈m(0)exp⁡(−C+eκt+C−e−κt+D) Performing an equivalent approximation to that of Eq. (11) results in the expression: (56)dP1e(t)dt≈k2m(0)n2+1[exp⁡(−n2C+eκt)−exp⁡(n2C−e−κt)−exp⁡(−(n2+1)C+eκt)+exp⁡((n2+1)C−e−κt)] Integrating both sides with respect to time yields the self-consistent solution for the polymer number concentration: (57)P1e(t)=P(0)+k2m(0)n2+1κ[E1((n2+1)C+eκt)−E1((n2+1)C+)−E1(−(n2+1)C−e−κt)+E1(−(n2+1)C−)−E1(n2C+eκt)+E1(n2C+)+E1(−n2C−e−κt)−E1(−n2C−)] which is analogous to Eq. (13) from our treatment of fragmenting filaments.</p><p>We substitute y = eκt, z = e−κt so that the integrand from Eq. (52) can be written as: (58)I=k2m(0)n2+1κ[E1(n2C+)−E1(n2C+y)−E1(−n2C−)+E1(−n2C−z)−E1((n2+1)C+)+E1((n2+1)C+y)+E1(−(n2+1)C−)−E1(−(n2+1)C−z)] This expression is the difference between two integrands of the same form as in the fragmentation case, Eq. (15), for which we already know the solutions. Thus, splitting the integral in two, the solution is given by the difference between these two known results Eq. (21). Making the same asymptotic analysis as in the fragmentation case Eq. (23) to acquire the full result gives immediately the closed form expression: (59)M2series(t)M(∞)=1−exp(∑p=1∞[(−C+)p((1+n2)p−n2p)p2p!(epκt−∑q=02p(pκt)qq!)+C−p((1+n2)p−n2p)p2p!(e−pκt−∑q=02p(−pκt)qq!)+(−C+)pp2p!(((2p−1)κt)2p−1(2p−1)!+(2pκt)2p(2p)!)+(C−)pp2p!((−(2p−1)κt)2p−1(2p−1)!+(−2pκt)2p(2p)!)]−M(0)M(∞)) This equation, which is accurate for early times, has an analogous form to the result obtained for the fragmentation case, Eq. (23), except for the presence of the multiplying factor (1+n2)p−n2p in each term, and also additional polynomial terms. Furthermore, as expected, Eq. (59) reduces to Eq. (23) for n2 = 0[31].</p><p>The convergence discussion from the fragmenting growth section applies to this result as well, except that here the additional multiplicative factor, which increases exponentially with the index of summation, will reduce the convergence rate of the series. Specifically, the rate of convergence in Eq. (59), for the same timescale, is approximately a factor of (1+n2) slower than that observed for the analogous series in the fragmentation case, Eq. (24), for large p. Crucially, however, whilst the first order iteration result predicts a similar timescale for the completion of the reaction both for the fragmentation and monomer-dependent secondary nucleation cases, the more accurate second order iteration result shows that the kinetic profile in the latter case is extended in time, with the end point being severely underestimated by the first iteration. A consequence of this observation is that the truncation of the summation must be valid in the monomer dependent secondary nucleation case over a much longer timescale than analagous curves in the fragmentation case. This implies that, although only (1 + n2) times more terms are required to give convergence over the same timescale, many more terms are required to provide a valid truncation over the extended timescale present in the monomer-dependent case.</p><p>The convergence behaviour is demonstrated in Fig. 5, where the first two analytical iterations and the corresponding numerical results (including 𝒪(kn)), are shown. For comparison we also consider the analogous fragmentation induced growth process obtained by identifying k− = k2m(0)n2 and resulting in a secondary process which for t = 0 has the same effect as that of the monomer-dependent case. In this analogous fragmentation case, 25 terms in the summation are sufficient to prevent any significant errors from being introduced in to the summation; this success is due to the fact that the divergence in the series occurs after the reaction end point when M(t) ≈ mtot and is therefore inconsequential to the description of the growth phenomenon. In agreement with the convergence discussion above, it is observed in Fig. 5 that with n2 + 1 = 3 times as many terms, i.e. 75, the summation is valid to the same time in the monomer dependent secondary nucleation case as in the fragmentation case, as shown by the fact that the step indicating the divergence in the series Eq. (59) occurs precisely at the end-point of the first order result. Due to the nature of the monomer-dependent secondary processes, the overall growth profile is extended in time in comparison to fragmentation driven polymerisation reactions; this behavior originates from the fact that the rate of monomer-dependent secondary nucleation decreases monotonically as the reaction proceeds, whereas the fragmentation rate remains constant in time. The black dotted in Fig. 5 shows the exact solution in the fragmentation case; it is clear that the first order iteration is significantly better for this class of growth phenomena.</p><p>The change in the shape of the curve relative to the fragmentation case can also be seen in the scaling of the lag time. Although this has the same form as the fragmentation case, the constant term in the scaling law is changed, reflecting the change in the shape of the trace. The scaling law from the second iteration result can be found using the same approach as in the fragmentation case, and the results for different vales of n2 are shown below. We note that n2 = 0 recovers the scaling characteristic of the growth of fragmenting filaments. (60)τlag={[log⁡(1/C+)−1.718]κ−1Firstiteration,alln⁡2[log⁡(1/C+)−1.805]κ−1Seconditeration,n2=0[log⁡(1/C+)−2.202]κ−1Seconditeration,n2=1[log⁡(1/C+)−2.225]κ−1Seconditeration,n2=2[log⁡(1/C+)−2.684]κ−1Seconditeration,n2=5[log⁡(1/C+)−3.167]κ−1Seconditeration,n2=10</p><p>As n2 is increased, the first iteration result becomes increasingly poor. In particular, the end time is increasingly underestimated for increasing n2, and so more and more terms are needed in the summation for the second order iteration using Eq. (59).</p><p>We note that the integrand from Eq. 58 may also be integrated by re-writing: (61)I=k2m(0)n2+1κ(g+(n2+1)−g+(n2)−g−(n2+1)+g−(n2)+E1(n2C+)−E1((n2+1)C+)−E1(−n2C−)+E1(−(n2+1)C−)) with the functions g+(x) = E1(xC+eκt) and g−(x) = E1(−xC−e−κt). Since the integrand contains the same functions g± evaluted with different arguments, the integrand can be rewritten exactly as: (62)I=k2m(0)n2+1κ(∑i=1∞1i!dig+(x)dxi|x=n2−∑i=1∞1i!dig−(x)dxi|x=n2+E1(n2C+)−E1((n2+1)C+)−E1(−n2C−)+E1(−(n2+1)C−))</p><p>This expression can be evaluated and integrated analytically term by term. For n2 ≥ 2, an expansion to the fifth derivative is sufficient to give an excellent approximation to the exact result, with the result becoming more accurate as n2 increases. For n2 = 1, however, the expansion in Eq. (62) is poorly convergent.</p><!><p>The result Eq. (59) is accurate for all initial conditions and early and intermediate times and gives in closed form the time evolution of a filamentous system that grows through monomer dependent secondary nucleation and filament elongation. In order to extend the applicability of the solution to later times, we will use Eq. (59) to provide the appropriate initial conditions at the characteristic inflection point in the sigmoidal growth curve to initiate a new fixed point iteration providing a more accurate description past the inflection point. Crucially, this approach will also enable us to include exactly the primary nucleation terms into the fixed-point operator, which is vital as the relative balance of primary and secondary nucleation varies as the reaction proceeds.</p><p>We consider as the initial input to the late-time fixed point scheme the straight line P0tmaxP(t) that matches P(t) and its gradient at its point of inflection in the reaction profile; the time corresponding to the inflection point is denoted tmaxP. This choice for the separation between the early and late time solutions is made since at the point of inflection the second derivative of P(t) is zero and therefore a linear approximation is valid here up to the third derivative. The concentration of free monomer m* = m(tmaxP) at the time corresponding to the point of inflection is given through differentiation of Eq. (49) and enforcing the condition d2P/dt2 = 0 to yield: (63)0=k2n2m*n2−1mtot−k2(n2+1)m*n2+ncknm*nc−1 First note that in the case kn = 0 has the simple solution: (64)m*=n2n2+1mtot and to lowest order the nucleation terms can be included through the use of a Newton-Raphson correction to yield: (65)m*=n21+n2mtot−knηnc−1ncknnc(nc−1)ηnc−2−k2n2ηn2−2mtot where η = mtotn2/(1 + n2). From the first order iteration result, Eq. (51), the time to reach m* is approximately given by: (66)tmaxP={κ−1log(−b+b2+4C+C−2C+)P″(0)/m′(0)<00P″(0)/m′(0)>0 for b = log(m*/m(0)) − D with m* from Eq. (65). The condition on the second derivative accounts explicitly for cases where no point of inflection exists, where: (67)P″(0)/m′(0)=k2n2m(0)n2−1mtot−k2(n2+1)m(0)n2+ncknm(0)nc−1 Hence, our new initial linearization of M(t) is given as: (68)M0tmaxP(t)=α+β(t−tmaxP) To ensure that our solution is also valid for cases without a point of inflection (tmaxP = 0), the definitions of α and β are given by: (69)α={M2(tmaxP)tmaxP>0M(0)tmaxP=0 (70)β={M2′(tmaxP)tmaxP>02(m(0)k+−koff)P(0)+ncknm(0)nc+n2k2M(0)m(0)n2tmaxP=0 where M(0), m(0) and P(0) are initial conditions and M2(t) is given by Eq. (59); as the we only require the pre-inflection point solution Eq. (59) to be valid until tmaxP typically only terms in the sum until p = 3 are required for adequate convergence.</p><p>The corresponding initial solution for P(t) may be found from the differential equations Eqs. (49): (71)P0tmaxP(t)=γ+δ(t−tmaxP) where: (72)γ={M2′(tmaxP)−ncknm2(tmaxP)nc−n2k2M2(tmaxP)m2(tmaxP)n22(m2(tmaxP)k+−koff)tmaxP>0P(0)tmaxP=0 (73)δ={k2M2(tmaxP)m2(tmaxP)n2+knm2(tmaxP)nctmaxP>0knm(0)nc+k2M(0)m(0)n2tmaxP=0 and m2(t) = mtot − M2(t). Inserting Eq. (71) into Eq. (50), the first self-consistent solution for M is found by direct integration neglecting 𝒪(kn). With the condition M(tmaxP) = α we obtain: (74)M1tmaxP(t)M(∞)=1−(M(∞)−α)M(∞)exp⁡(−2k+[γ(t−tmaxP)−δtmaxP(t−tmaxP)+(1/2)δ(t2−tmaxP2)]) The first post-inflection point self-consistent solution for the moment P(t) is found similarly by inserting M1tmaxP⁡ into Eq. (49); crucially, we include here exactly the term 𝒪(kn). Together with the condition P(tmaxP) = γ this yields: (75)P1tmaxP(t)=γ+k2mtoth(n2,t)−k2h(n2+1,t)+knh(nc,t) where (76)h(x,t)=eγδk+γxπ(er(x,t)−er(x,tmaxP))(M(∞)−α)x2k+δx and (77)er(z,t)=Erf(γδk+δz+(t−tmaxP)k+δz) for the error function Erf(z)=2π∫0ze−t2dt. If an expression only in terms of elementary functions is desired, the error function can be approximated as</p><p>Erf(z)≈1−exp⁡(−z2(4/π+0.147z2)/(1+0.147z2)) [38]. A further iteration can be performed from Eq. (50) upto 𝒪(kn) by direct integration, yielding the final result for the post-inflection point solution M2tmaxP: (78)M2tmaxP(t)M(∞)=1−(M(∞)−α)M(∞)exp⁡[−2k+γ(t−tmax⁡)−κ02v(n2,t)+κ02v(n2+1,t)−λ02v(nc,t)] where (79)v(x,t)=(γδ+(t−tmaxP))h(x,t)m(0)x+(e−k+δx(t−tmaxP)2−2k+γx(t−tmaxP)−1)2k+δx(M(∞)−α)xm(0)x for κ0=κ|koff=0=2m(0)n2+1k+k−. Apart from the influence of the boundary conditions defined by α and β, these new results for M, Eqs. (74) and (78), depend on the rate constants only through the combinations k+γ, k+δ and their ratio γ/δ. For cases with a lag phase, these combinations are given approximately as: (80)k+γ≈k+M2′(tmaxP)2(k+m2(tmaxP)−koff) (81)k+δ=M2(tmaxP)m2(tmaxP)n22m(0)n2+1κ02+m2(tmaxP)nc2m(0)ncλ02</p><p>We note that due to the explicit inclusion of primary nucleation terms, we recover the combination λ0 which is the single parameter that defines growth of filaments in the absence of secondary nucleation[9, 31]. These terms governed by λ0 in Eq. (78) become important for kinetic parameters that result in primary nucleation being more significant that secondary nucleation for creating seeds for times after tmaxP. In many cases, particularly for nc > n2, these terms are not significant.</p><p>The result Eq. (78) can be continued piecewise at tmaxP with the result for M2series from Eq. (59) to give a final result which is continuous by construction. The result is illustrated in Fig. 6 and written out fully up to third order terms in the pre-inflection point solution in the Appendix.</p><p>Although these closed form solutions contain more terms than in the case where the secondary pathway is fragmentation, Eq. (23), it is interesting to note that the dependencies on the kinetic parameters are analogous. In particular, in the case of irreversible growth, koff = 0, Eq. (78) continued piecewise with Eq. (59) depends only upon the three kinetic parameters κ, λ0 and k+. Additionally, in the absence of seed material, M(0) = P(0) = 0, the closed-form result depends only upon the two combinations of kinetic parameters κ and λ, similarly to Eq. (27).</p><!><p>The polymerization reaction comes to completion when the free monomer has been partially depleted, and an equilibrium is established between the aggregated and soluble phases. The time at which this equilibrium is reached depends on the initial concentration of monomer at the beginning of the reaction. We show here that this dependence is essentially a power-law of a form analogous to that governing the lag-time, but with a different proportionality constant. We consider here the case where the overall reaction is driven by secondary nucleation processes, and the primary nucleation terms described by λ can be neglected.</p><p>For times t ≫ κ−1, Eqs. (23) and (59) show that (for any given n2) the fractional polymer mass concentration is approximately a function only of the combination C+eκt. As such, the time needed to reach each given fractional polymer mass concentration in this time limit must correspond to a constant value of C+eκt. In particular, if the end point of the reaction is defined as τend such that M(tend)/M(∞) = s, for some constant s → 1, then it must be the case that C+eκτend is constant. Explicitly, for any given n2 there is a constant c such that: (82)C+eκτend=c If τend is determined from any one curve (choosing some s to define the end point) and used to determine c, it is then possible to rearrange the above result and predict the end time τend for any other set of parameters via the allometric law: (83)τend=[log⁡(1/C+)+log⁡(c)]κ−1 For s = 0.999, approximate scaling results for some specific cases are found to be: (84)τend={[log⁡(1/C+)+3.0]κ−1 n2=0 (fragmentation)[log⁡(1/C+)+6.9]κ−1 n2=1[log⁡(1/C+)+11]κ−1 n2=2[log⁡(1/C+)+25]κ−1 n2=5[log⁡(1/C+)+47]κ−1 n2=10 which are similar in form to the lag time scaling Eq. (60). This scaling is also of practical interest as it allows the prediction of the relevant timescale of a reaction at the design stage of experiments. In particular, the scaling law reveals approximately τend∼m(0)−n2+12, which is useful in predicting the relevant timescales of reactions at different total monomer concentrations.</p><!><p>We have derived earlier[31] the time evolution of the mean and standard deviation of the length distribution of fragmenting filaments in the case of a constant monomer concentration[31]. In the case of the standard deviation, these results were obtained under the assumption that the distribution is not heavily skewed. The availability of expressions for both the mean and standard deviation makes it possible to evaluate the time evolution of the full filament length distribution within a self-consistent framework and determine the skewness and higher central moments of the length distribution. To this effect, we demonstrate the wider applicability and power of our self-consistent approach by finding the first correction to the symmetric Gaussian approximation to the length distribution using a self-consistent scheme. We use this result to determine the correction, relative to the mean, to the time evolution of the mode of the distribution, which is not available simply through knowledge of the first three moments of the distribution.</p><p>Formally integrating the master equation, the solution for the full length distribution can be written as: (85)∂f(t,j)∂t=2mtotk+f(t,j−1)−2mtotk+f(t,j)−k−(j−1)f(t,j)+2k−∑i=j+1∞f(t,i)+knmtotncδj,nc Formally integrating, the solution can be written as: (86)f(t,j)=2mtotk+e−2mtotk+t∫0te2mtotk+τf(τ,j−1)dτ−k−e−2mtotk+t∫0te2mtotk+τ[(j−1)f(t,j)−2∑i=j+1∞f(t,i)]dτ+knmtotncδj,nce−2mtotk+t∫0te2mtotk+τdτ We note that it is possible to solve this equation exactly for any f(j, t) recursively beginning with f(t, nc) and integrating j − nc times. However, given the considerable coupling of equations in this system through the summation term relating to filament breakage, this approach becomes impractical beyond small j. Instead, it is possible to remove the coupling originating in the fragmentation related terms using a self-consistent approach.</p><p>Consider initially the master equation without filement fragmentation, k− = 0. In this case each equation is only coupled to its nearest neighbours and the solution fk− =0(j, t), found by repeated integration, is given as: (87)fk−=0(j,t)P(t)=knmtotnc−12k+(1−e−2mtotk+t)−knmtotnc−12k+e−2mtotk+t∑i=1j−n(2mtotk+t)ii! In this case the master equation is a discretization of an advection equation and so the solution describes a travelling wave in j moving to encompass larger filament sizes at velocity 2k+mtot.</p><p>We now account in a self-consistent manner for the terms in the master equation describing fragmentation by introducing for the filament length distribution in these terms the Gaussian approximation with mean L(t) and standard deviation L(t)/3: (88)f0(j,t)=32πe−3(j−L(t))22L(t)2L(t)P(t)E∞ where E∞ is a normalization constant for the distribution between j = nc and j = ∞, E∞≈1/2(1+Erf(3/2))≈0.96, where the approximation is valid since L(t) ≫ nc. Note that the distribution f0(j, t), due to the cut-off at nc, has a mean length, L0(t), different to the exact value, L(t): (89)L0(t)=∫nc∞sf0(s,t)ds=L(t)+13L(t)2f0(nc,t)</p><p>We now consider the breakage related terms in Eq. (86). Introducing the initial Gaussian approximation and replacing the summation term in Eq. (86) with a continuum approximation leaves: (90)−k−e−2mtotk+t∫0te2mtotk+τ[(j−1)f0(t,j)−2P(t)+2∫i=ncjf0(t,i)di]dτ In addition, we must account for the coupling of these breakage related terms to nearest neighbours through the first term in Eq. (86), which introduces a summation over j. This allows us to write within this self-consistent framework the full length distribution as a function of time: (91)f1(j,t)=fk−=0−k−e−2mtotk+t∫ncj∫0te2mtotk+τ[(s−1)f0(t,j)−2P(t)+2∫i=ncsf0(t,i)di]dτds Performing the integration with respect to time using the general expansion for large α, e−αt∫0teαty(t)=y(t)/α+𝒪(y′(t)/α2), allows us to evaluate these integrals to first-order approximation to give a new expression for the filament length distribution: (92)f1(j,t)=fk−=0+k−2mtotk+(2P(t)(j−nc)−2F1(j,t)−F2(j,t)) where the functions E(j, t), F1(j, t) and F2(j, t) are given in terms of error functions: (93)E(j,t)=∫ncjf0(s,t)ds=P(t)2E∞[−Erf(32(L(t)−j)L(t))+Erf(32)] (94)F1(j,t)=∫ncjE(s,t)ds=∫ncj∫ncsf0(r,t)drds=[j−L(t)]E(j,t)+13L(t)2[f0(j,t)−f0(nc,t)] (95)F2(j,t)=∫ncj(s−1)f0(s,t)ds=[L(t)−1]E(j,t)−13L(t)2[f0(j,t)−f0(nc,t)] For large j, the errors 𝒪(α−2) accumulate due to the integral over j in Eq. (91), resulting in excellent accuracy for small and intermediate j but an incorrect limit for j → ∞. This limit, however, can be evaluated exactly, yielding L0(t) − 2nc + 1. Removing the associated terms yields: (96)f1(j,t)=fk−=0+k−2mtotk+[2[1−E(j,t)][j−nc]−13L(t)2[f0(j,t)−[1−E(j,t)]f0(nc,t)]]</p><p>The result f1(j, t), Eq. (96), is expected to be highly accurate for small filament sizes, since the coupled integration over j is most accurate in this regime. Due to the contribution from the solution of the advection equation, Eq. (87), we would expect the result to be accurate up to lengths 2L(t), or equivalently 3≈1.7 standard deviations above the mean, thus being accurate for all but the tail of the distribution. The improvement over the initial Gaussian solution obtained using the information from the first three principal moments is shown in Fig. 7.</p><p>The mode of the distribution can be obtained from the condition: (97)0=f0(jmode,t)(2−L0(t)−j)+2−2E(jmode,t) Expanding the Gaussian f0(j, t) to lowest order about the mean results in: (98)0=32πL(t)E∞[2−L0(t)−jmode]+2−2(12E∞Erf(32L(t)−ncL(t))+32πL(t)E∞[jmode−L(t)])+𝒪(jmode2) Using L(t) ≫ nc and Eq. (89) yields the simple result: (99)jmode=13(1+2π3−1332πe−32E∞)L(t)≈0.8L(t) As expected, the mode of the distribution is below the mean, and we recover the interesting result that the ratio of the mode to the mean, and hence also to the standard deviation, is approximately a constant in time for this system; the mode evolves with the same functional form as the mean and variance. The improvement in the value of the mode of the filament distribution in the distribution f1(j, t) compared with the original Gaussian f0(j, t) is shown in Fig. 8.</p><!><p>We have used the power of iterative fixed point schemes to provide self-consistent solutions to growth processes described by highly non-linear master equations. We have shown that corrections to the scaling behaviour emerging from first order results can be well captured by second order self-consistent solutions, which yield coefficients for the characteristic scaling laws which are very close to ones obtained from numerical evaluation. More generally, these results illustrate the value of fixed-point analysis strategies in order to provide self-consistent solutions to complex growth problems beyond perturbative treatments.</p><!><p>The second-order self-consistent solution to the non-linear equation system describing filamentous growth:(100)dP(t)dt=k2M(t)m(t)n2+knm(t)nc (101)dM(t)dt=2(m(t)k+−koff)P(t)+ncknm(t)nc+n2k2M(t)m(t)n2 with m(t) = mtot − M(t) derived in this paper is written out in full in terms of only elementary functions as:(102)M(t)M(∞)={1−exp⁡[κ2t2+4κ4t43+81κ6t640+(e−κt−1+κt−κ2t22)C−−(eκt−1−κt−κ2t22)C++18(e−2tκ−1+2tκ−2t2κ2+4t3κ33−2t4κ43)C−2((1+n2)2−n22)+18(e2tκ−1−2tκ−2t2κ2−4t3κ33−2t4κ43)C+2((1+n2)2−n22)t≤tmaxP+154(e−3tκ−1+3tκ−9t2κ22+9t3κ32−27t4κ48+81t5κ540−81t6κ680)C−3((1+n2)3−n23)−154(e3tκ−1−3tκ−9t2κ22−9t3κ32−27t4κ48−81t5κ540−81t6κ680)C+3((1+n2)3−n23)−M(0)M(∞)]1−(M(∞)−α)M(∞)exp⁡[−2k+γ(t−tmax⁡)−κ02v(n2,t)+κ02v(n2+1,t)−λ02v(nc,t)]t>tmaxP where the dependencies on the rate constants and initial conditions are defined through: (103)κ=2(k+m(0)−koff)m(0)n2k2κ0=2m(0)n2+1k+k2λ0=2m(0)nck+kn (104)C±=k+P(0)κ±k+M(0)2[m(0)k+−koff]±λ022κ2tmaxP={κ−1log(−b+b2+4C+C−2C+)P″(0)/m′(0)<00P″(0)/m′(0)>0 (105)M(∞)=mtot−koff/k+α=M(tmaxP)b=log(m*m(0))−λ02κ2+M(0)M(∞)−k+M(0)k+m(0)−koff (106)m*=n21+n2mtot−kn(mtotn21+n2)nc−1ncknnc(nc−1)(mtotn21+n2)nc−2−k2mtotn2(mtotn21+n2)n2−2 (107)h(x,t)=eγδk+γxπ(er(x,t)−er(x,tmaxP))(M(∞)−α)x2k+δx (108)er(z,t)=Erf(γδk+δz+(t−tmaxP)k+δz)≈1−exp⁡(−(γδk+δz+(t−tmaxP)k+δz)2(4/π+0.147(γδk+δz+(t−tmaxP)k+δz)2)(1+0.147(γδk+δz+(t−tmaxP)k+δz)2)) (109)v(x,t)=(γδ+(t−tmaxP))h(x,t)m(0)x+(e−k+δx(t−tmaxP)2−2k+γx(t−tmaxP)−1)2k+δx(M(∞)−α)xm(0)x (110)k+γ≈k+M2′(tmaxP)2(k+m2(tmaxP)−koff)k+δ=M2(tmaxP)m2(tmaxP)n22m(0)n2+1κ02+m2(tmaxP)nc2m(0)ncλ02 The accuracy of this solution even for high exponents n2, nc is demonstrated in Fig. 2.</p><!><p>Highly non-linear processes in protein polymerisation. The primary nucleation rate depends on the monomer concentration m(t) with a power of the critical nucleus size nc, and secondary nucleation processes depend on both the polymer concentration M(t) and the monomer concentration to a power n2; values of n2 up to n2 = 30 have been reported[14, 16]. The special case of filament fragmentation[1] is monomer independent n2 = 0.</p><p>Illustration of the nature of the solutions to highly non-linear growth problems derived using the fixed point scheme (insert). An analytical solution to filament growth through primary nucleation, monomer dependent secondary nucleation, filament elongation and depolymerization [Eqs. (49), (50)] is shown in red and the complete closed form self-consistent solution is provided in the Appendix. The black line is the exact numerical result. For comparison, the green dashed line is the numerical result neglecting primary nucleation, the purple dashed line is the numerical result neglecting secondary nucleation. The blue dotted line is the first order result[31] and the orange dot-dashed line is the linearized solution[5, 16, 32, 33]. The parameters are: nc = 35, n2 = 20, k+ = 5 · 104 M−1 s−1, koff = k+mtot/30, k2mtotn2=2 · 10−8s−1,knmtotnc−1=5 · 10−10s−1, mtot = 5 · 10−5M, M(0) = 5 · 10−11M, P(0) = M(0)/5000.</p><p>Convergence of the first two fixed point analytical iterations towards the exact solution. The blue dotted line is the first order result, the red dashed line is the second order result and the solid black line is the exact result. The orange dot-dashed line is the early time limit linearized solution. The parameters are: k+ = 5 · 104 M−1 s−1, koff = k+mtot/100, k− = 2 · 10−8 s−1, kn = 5 · 10−5 M−1 s−1, mtot = 1 · 10−6M, nc = 2, M(0) = 1 · 10−8M, P (0) = M(0)/5000. The summation in the second order iteration is truncated after 25 terms.</p><p>Convergence of the first two fixed-point analytical iterations towards the exact solution in the absence of breakage. The blue dotted line is the first order result, the red dashed line is the second order result and the solid black line is the numerical result. The parameters are: k+ = 5·104 M−1 s−1, koff = k− = 0 s−1,kn = 5 · 10−4 M−1s−1, mtot = 5 · 10−6M, nc = 2, M(0) = P(0) = 0.</p><p>Convergence of the pth fixed point iteration, Mn(t), using the early time linearization. M1 (blue dashed, Eq. (3)), M2series (red dashed, Eq. (59) with 75 terms), exact solution including 𝒪(kn) (black). The lighter solid lines show the corresponding numerical results. The black dotted line is the exact result for the analogous fragmentation case for comparison. The orange dot-dashed line is the early time linearization. The parameters are M(0) = P(0) = 0M, k+ = 5 · 104 M−1 s−1, koff = 0, k2m(0)n2 = 2 · 10−8/s−1, kn = 2 · 10−5 M−1s−1, mtot = 5 · 10−6M, nc = 2, n2 = 2.</p><p>The solution found using a second linearization about the point of inflection, tmaxP (indicated by the vertical black dotted line), is continued piecewise to M2series. The cyan dot-dashed line is the new initial approximation, Eq. (68). For t < tmaxP, the red dashed line is M2series (Eq. (59) with 25 terms). For t > tmax, the blue dashed line is the first order iteration given by the new expansion, M1tmax⁡ (Eq. (74)), and the red dashed line is the corresponding second order iteration, M2tmax⁡ (Eq. (78)). The orange dot-dashed line is the early time linearization. The parameters are M(0) = 5 · 10−8M, P(0) = M(0)/5000, k+ = 2 · 104 M−1 s−1, koff = k+mtot/20, k2mtotn2=5 · 10−8s−1, kn = 1 · 10−4 M−1s−1, mtot = 5 · 10−6M, nc = 2, n2 = 1.</p><p>The solution derived for the time evolution of the full filament length distribution in our self-consistent scheme, Eq. (96), is shown in red (dashed). The numerical result is shown in black, and the initial Gaussian approximation, Eq. (88), is shown in blue (dotted).</p><p>The correction to the mode in relation to the mean, Eq. (99), exhibited by our solution to the full length distribution, Eq. (96), is shown in red (dashed). The numerical result is shown in black, and the result from the initial Gaussian approximation is shown in blue (dotted).</p><p>Comparison of the first and second order self-consistent solutions for irreversible filament growth with fragmentation with exact results calculated numerically</p>

PubMed Author Manuscript

High ambient temperature increases the toxicity and lethality of 3,4-methylenedioxymethamphetamine and methcathinone

Rationale: Methylenedioxymethamphetamine (MDMA) and methcathinone (MCAT) are abused psychostimulant drugs that produce adverse effects in human users that include hepatotoxicity and death. Recent work has suggested a connection between hepatotoxicity, elevations in plasma ammonia, and brain glutamate function for methamphetamine (METH)-induced neurotoxicity. Objectives: These experiments investigated the effect of ambient temperature on the toxicity and lethality produced by MDMA and MCAT in mice, and whether these effects might involve similar mechanisms to those described for METH neurotoxicity. Results: Under low (room temperature) ambient temperature conditions, MDMA induced hepatotoxicity, elevated plasma ammonia levels, and induced lethality. Under the same conditions, even a very high dose of MCAT produced limited toxic or lethal effects. High ambient temperature conditions potentiated the toxic and lethal effects of both MDMA and MCAT. Conclusion: These studies suggest that hepatotoxicity, plasma ammonia, and brain glutamate function are involved in MDMA-induced lethality, as has been shown for METH neurotoxicity. The toxicity and lethality of both MDMA and MCAT were potentiated by high ambient temperatures. Although an initial mouse study reported that several cathinones were much less toxic than METH or MDMA, the present results suggest that it will be essential to assess the potential dangers posed by these drugs under high ambient temperatures.

high_ambient_temperature_increases_the_toxicity_and_lethality_of_3,4-methylenedioxymethamphetamine_a

Introduction:<!>Subjects:<!>Drugs:<!>Experimental Procedure:<!>Determination of serum ammonia:<!>Statistical analysis:<!>MDMA- and MCAT-induced lethality at different ambient temperatures<!>The effects of ambient temperature on MDMA- and MCAT-induced temperature changes:<!>Plasma ammonia:<!>Organ weight changes:<!>Discussion:<!>MDMA-induced toxicity and lethality<!>Cathinone-induced toxicity and lethality<!>Conclusions

<p>Synthetic psychoactive cathinones (SPCs) are drugs with similar chemical and subjective properties to amphetamines (Baumann et al., 2012), for which there are significant public health concerns (Baumann, 2014). Widespread use of SPCs emerged at the beginning of this century. This group of abused drugs include numerous and varied analogues with actions similar to the common illicit drugs methamphetamine (MA), methylenedioxymethamphetamine (MDMA), and cocaine, which prompt their use among illicit users (Ashrafioun et al., 2016; Johnson and Johnson, 2014). Consequently, there is concern about the potential of these drugs for abuse and addiction (e.g. (Watterson et al., 2012; Watterson et al., 2014), and see comprehensive reviews by Watterson and Olive (2017) and Riley et al. (2020)), as well as acute overdose that is similar in presentation to other stimulant drugs. Symptoms of human SPC overdose include neurologic, cardiac and other physiological symptoms: hyperthermia, tachycardia, heart palpitations, chest pain, hypertension, agitation, violent behavior, hallucinations, paranoia, confusion, mydriasis, vomiting, myoclonus, seizures, organ toxicity, and lethality, according to clinical case reports (Marusich et al., 2014; Spiller et al., 2011). Such reports are small and uncontrolled due to the nature of the situations they describe, often without comprehensive analysis of the drugs consumed, with uncertainties about the time since the exposure, and involving consumption of multiple drugs at the same time, among other complicating factors (for example see Lee et al. (2015)). These uncertainties about clinical reports make the suggestions that SPCs may be more toxic than other amphetamine-like drugs (American Association of Poison Control Centers, 2011, Wood et al., 2015, 2013) difficult to evaluate. The adverse consequences of SPCs (like amphetamines), involve many different bodily systems, as is clear from the symptom profile described above. The psychiatric consequences of SPC use (and overdose) are certainly consistent with the neural effects of these drugs, but there is also evidence of cardiac (Beck et al., 2018; Marinetti and Antonides, 2013; Spiller et al., 2011), renal (Benzer et al., 2013; Borek and Holstege, 2012; Eiden et al., 2013), and hepatic (Carvalho, Márcia et al., 2012; Marinetti and Antonides, 2013) toxicity. Although not widely explored, SPCs have been shown to produce toxicity in hepatic cells through the production of reactive metabolites, mitochondrial impairment, (local and systemic) hyperthermia, and apoptosis (for review see Carvalho, et al. (2012)), but the importance of these hepatic effects for lethality, compared to effects on other bodily systems is uncertain. The uncertainty of illicit drug composition contributes to adverse effects (Araújo et al., 2015), especially for SPCs (Krabseth et al., 2016). Indeed, many self-identifying "MDMA" users have been found to be unknowing users of SPCs (Caudevilla-Gálligo et al., 2013; Palamar et al., 2016). The risk of adverse effects for SPCs is much less well-known than for MDMA, but many SPCs have higher potency, which may increase the risk of overdose, as it has for synthetic opioids (Hall and Miczek, 2019).</p><p>The extent to which users or suppliers drive the use of higher potency drugs is a matter of debate (Mars et al., 2018), but to understand the potential of these drugs for human overdose, it is first necessary to develop a clearer understanding of the lethal and toxic potential of these drugs. Preclinical studies are needed to answer these questions, but few investigations of SPC lethality in animals have been reported to date. One study reported a slightly increased lethality (e.g. slightly reduced LD50 value) for methylone compared to MA or MDMA (Piao et al., 2015). Methylone produced only modest hyperthermia, and lethality was independent of hyperthermia – dopamine transporter (DAT) gene deletion reduced lethality, but not hyperthermia. A similar result was previously obtained for MA (Numachi et al., 2007). This does not mean that hyperthermia is not involved in amphetamine-induced lethality as there is certainly plenty of evidence for that relationship for MA- and MDMA-induced neurotoxicity (for review see Krasnova and Cadet (2009) and Kiyatkin and Sharma (2012)). Hyperthermia after MA or MDMA is most often observed using a "binge-dosing" regimen, mimicking the human pattern of use of these drugs that involves repeated moderate dosing cumulatively equivalent to about an LD50 dose (for review see Cadet et al. (2007) and Krasnova and Cadet (2009)). Within a single binge of MA, MDMA or methylone, the first dose can induce hypothermia, while subsequent doses progressively induce greater hyperthermia (Granado et al., 2010; Miner et al., 2017). SPCs commonly produce hypothermia (Aarde et al., 2013; Aarde et al., 2015; Anneken et al., 2017a; Anneken et al., 2017b; Shortall et al., 2013). Indeed, cathinone (CAT), methcathinone (METH), methylenedioxypyrovalerone (MDPV), and mephedrone (Muskiewicz et al., this issue) produce temperature reductions of as much as 7 °C. Importantly, that study found that the lethality of those drugs was much lower than was observed for MA and MDMA, and their LD50values could only be calculated for one of the SPCs. These studies were conducted at standard room temperature, so it remains possible that SPCs may be more toxic at higher ambient temperatures.</p><p>Ambient temperature has been previously shown to affect MDMA-induced neurotoxicity, and hepatotoxicity (Carvalho et al., 2001; Fantegrossi et al., 2003; Gordon et al., 1991; Sanchez et al., 2004), as well as lethality (Fantegrossi et al., 2003). It should be noted that lethality is not a focus of most of these studies, often being just an incidental endpoint. These studies tend to study sub-lethal, toxic doses, and consequently the lethal effects of these drugs are less well characterized than lower-dose toxic effects, including the role of ambient temperature in lethality. At a high ambient temperature, MDMA causes blood clots, bleeding, and liver vacuolation in vivo, and hepatotoxicity in vitro (Carvalho et al., 2002). Neurotoxicity associated with binge MA administration also involves hepatotoxicity that subsequently elevates plasma ammonia levels and extracellular brain glutamate levels (Halpin et al., 2013; Halpin et al., 2014; Halpin and Yamamoto, 2012). It is unknown whether this proposed mechanism contributes to MA-induced lethality, or to the toxicity/lethality of other cathinones and amphetamines, but it is interesting to note that in our other studies (Muskiewicz et al., in this volume), death from MA, MDMA, and cathinones, when it was observed, was associated with seizure, perhaps indicative of elevated glutamate release. High plasma ammonia levels associated with hepatic disease, or experimentally-induced elevations in ammonia, produce neurotoxicity involving brain glutamate (for review see Dabrowska et al. (2018)), the mechanism underlying hepatic encephalopathy. Many SPCs are also hepatotoxic (Araújo et al., 2015), and this hepatotoxicity also appears to be temperature-dependent (Valente et al., 2016).</p><p>The present studies were conducted to determine if: (1) hepatotoxicity and increased plasma ammonia occur under conditions that produce MDMA-induced lethality, and (2) if ambient temperature influences this process. A dose of MCAT that is non-lethal under standard room temperatures was also examined under high ambient temperature conditions.</p><!><p>Adult male and female C57Bl/6J mice were obtained from the Jackson Laboratory (Bar Harbor, ME, USA) at 5 weeks of age. Subjects were housed in groups of 3-4 mice for at least one week in standard plastic cages (7.5 x 11.75 x 5") prior to testing to habituate to animal facility conditions. Animals were tested at 6-8 weeks of age (N = 10 all experimental conditions except MCAT, N= 5 males and N=5 females; for MCAT the total N = 5; 3 males and 2 females). Food (Teklad rodent diet 2916) and water were available ad libitum. Temperature was maintained at 22-23 °C and humidity was maintained at 40-45%.</p><!><p>S(+)−3,4-Methylenedioxymethamphetamine hydrochloride (MDMA) and methcathinone (MCAT) was obtained from the National Institute on Drug Abuse Drug Supply Program. All doses are expressed as the salt form. For injection, all drugs were dissolved in sterile isotonic saline and injected in a volume of 10 mL/kg. The doses of MDMA used were 103.8 mg/kg, 120.8 mg/kg and 140.6 mg/kg (IP), which were approximately the LD25, LD50, and LD75 doses from a previous study in our laboratory (see Muskiewicz et al., in this volume). In that previous study MCAT produced limited lethality, less than 10%, even at the highest dose tested. Therefore, the highest dose from that study (160 mg/kg) was used. Doses are expressed as the weight of the salt.</p><!><p>On the experimental day, mice were placed in individual cages and habituated for two hours before injection under either high ambient temperature (32 ± 1 °C) or a low ambient temperature (20 ± 1 °C) conditions. Rectal temperature of each mouse was measured every 30 minutes beginning 120 min before the injection until 120 min after the injection. Mice were injected with either saline (SAL), MCAT, or MDMA (denoted as MDMA 25, MDMA 50, or MDMA 75) IP. Half of the mice tested in each condition were tested at a high ambient temperature (HT) and half at room temperature (LT). Mice were continuously observed, but behavior was recorded by an observer every 30 min, along with rectal temperature. The time of death, if it occurred, was noted as the latency to die. Mice that died during the experiment were immediately dissected, and major organs (brain, heart, lungs, liver, and kidney) removed and weighed. Although the latency to die was noted, the primary outcome measure was percent lethality under each treatment condition. Any mice that survived until 120 min after the injection were euthanized by rapid decapitation. Blood samples from each mouse were collected for determination of plasma ammonia. All testing was conducted between 10 AM and 4 PM.</p><!><p>Blood samples were placed at room temperature for at least 30 min for coagulation. Serum samples were obtained by centrifugation at 3000 rpm for 15 min at 4°C. The ammonia concentration in the serum was determined using a Sigma-Aldrich Ammonia Assay Kit (AA0100). The procedure was performed according to the manufacturer's instructions.</p><!><p>Statistical analysis was performed using GraphPad Prism 7.0 (San Diego, California, USA). MDMA- and MCAT-induced lethality at different ambient temperatures was evaluated by determining the proportion of animals that died during the experiment and evaluated using a χ2 analysis. As an additional measure of lethality, the latency to die was determined for all groups in which death occurred and this data subjected to a combined analysis of both MDMA and MCAT groups in which death occurred (e.g. LT MDMA 25, LT MDMA 50, LT MDMA 75, HT MDMA 25, HT MDMA 50, HT MDMA 75, and HT MCAT). For this data, all mice that did not die within 120 min were given that value as the latency for purposes of analysis, but only groups where death was observed were used in the analysis. MDMA and MCAT data for the latency to die were analyzed together for all groups where death occurred using 1-way analysis of variance (ANOVA), followed by Tukey's post hoc comparisons between individual means. Temperature data for MDMA and MCAT were analyzed separately compared to the saline-treated group. MDMA data were evaluated using analysis of variance (ANOVA) with the factors of MDMA Dose (SAL vs. MDMA 25, MDMA 50, and MDMA 75), Temperature (high temperature vs. low (room) temperature; HT vs. LT), and time as a repeated measure (−60 to +120 min, with the drug injection being given after the 0 min time bin). MCAT data were analyzed with the factors of MCAT Dose (SAL vs. MCAT) and Temperature (HT vs. LT), and time as a repeated measure (−60 to +120 min, with the drug injection being given after the 0 min time bin). Because most subjects died in the HT groups, temperature data for MDMA and MCAT were evaluated by ANOVA only for the LT groups. Because this data might have underestimated temperature changes contributing to lethality, since it did not include subjects that died, temperature was also measured upon death and analyzed separately. This data was analyzed as percent change from untreated (saline) mice and analyzed separately by 1-way ANOVA of the combined MDMA and MCAT groups in which death occurred (e.g. LT MDMA 50, LT MDMA 75, HT MDMA 25, HT MDMA 50, HT MDMA 75, and HT MCAT). Naturally this only examined groups in which a substantial portion of the mice died, so this does not represent all of the groups studied. Data for plasma ammonia were evaluated by one-way ANOVA including data for both MDMA and MCAT together since not all of the samples could be analyzed (see below). Data for organ weights were evaluated by ANOVA, separately for MDMA and MCAT, with the same factors as the temperature data. Although both sexes were examined, since there were no apparent differences observed between male and female mice for lethality, sex was not used as a factor in any of the analyses. Nonetheless, it should be noted that although no sex differences were observed for lethality, the number of subjects examined, particularly for MCAT, would not have had sufficient statistical power to resolve smaller effects. Post hoc comparisons were made using Tukey's post hoc test. The significance level for all analyses was set to p<0.05.</p><!><p>The mortality rate after MDMA and MCAT administration are shown in Fig. 1. Saline injection (SAL) did not induce any lethality at either temperature. At the low ambient temperature, the mortality rate for MDMA 25, MDMA 50, MDMA 75 were 30%, 64%, and 90%, quite close to the estimated values. The percent lethality for MCAT under the low temperature conditions was 20%, just slightly more than our previous study, but still lower that the LD50. χ2 tests indicated that the high ambient temperature significantly increased MDMA- and MCAT-induced lethality (χ2 (2, N =80) = 68.7, p < 0.0001; χ2 (2, N = 20) = 31.3, p < 0.0001); indeed, 100% of mice tested under the high ambient temperature conditions that were treated with either MDMA or MCAT died.</p><p>The latency of death is shown in Fig 2. Obviously in groups were no (SAL) or limited (LT MCAT) lethality was observed the latency of death was near the maximum value (120 min; data not shown). For the LT MDMA 25 and LT MDMA 50 groups the latency was much higher than the other groups, reflecting a number of subjects which did not die. Among those mice that did die under these conditions the average latency was less than 20 min, but more variable than the other groups. The latency of death decreased with MDMA dose and was lowest for MDMA 75. The latency of death was as low for MCAT under the high temperature condition as it was for MDMA at the highest concentrations at the high temperature.</p><!><p>Figs. 3A and 3B show the rectal temperatures of mice that survived during the experiment, while Fig. 3C shows the temperature changes in animals that died. Consistent with our previous studies that examined single injections of MDMA under similar conditions, no significant changes in rectal temperature were observed after administration of MDMA compared to saline under low ambient temperature conditions. Neither the main effect of MDMA treatment (F(3, 176)=0.4 , NS) nor the MDMA treatment x time interaction (F(18, 176)=0.3, NS) were significant. Note however that this excludes mice that died at some point during this experiment. MCAT produced a large decrease in rectal temperature under room temperature conditions compared to saline-treated mice. Both the main effect of MCAT treatment (F(1, 80)=132.4, p<0.0001) and the MCAT treatment x time interaction (F(6, 80)=28.0, p<0.0001) were significant. This hypothermia reached a nadir at 60 min after MCAT administration, 31.3 ± 0.8 °C at 60 min after injection. Temperature was significantly reduced at 30, 60 and 90 minutes after the injection. Fig. 3C shows the change in rectal temperature at death for the LT MDMA 50 and LT MDMA 75 groups, and the HT MDMA 25, HT MDMA 50, HT MDMA 75, and HT MCAT groups. In mice treated with MDMA in the LT condition, decreases in rectal temperature of more than 2 °C were observed upon death. Under high ambient temperature conditions, MDMA produced increases in temperature of more than 2 °C in the HT MDMA 25 and HT MDMA 50 groups. These increases in temperature were slightly lower in the HT MDMA 75 group, perhaps because the latency to die was so short (see Fig. 2). Although under low temperature conditions MCAT induced hypothermia, under high temperature conditions, hyperthermia was observed (Fig. 3C). The extent of HT MCAT-induced hyperthermia was similar to HT MDMA 75, perhaps also because the latency of death was short. Rectal temperatures at death for HT MDMA 25, HT MDMA 50, HT MDMA 75 and HT MCAT were 39.7 ± 0.5°C, 40.6 ± 0.4 °C, 38.5 ± 0.3 °C, and 39.2 ± 0.4°C. These patterns were confirmed by a significant overall effect of treatment in the 1-way ANOVA (F(5, 19)=24.6, p<0.0001).</p><!><p>Fig. 4 shows plasma ammonia levels after administration of MDMA or MCAT under the low ambient temperature condition. Only samples from subjects alive at the conclusion of the experiments were analyzed for plasma ammonia levels because the assay might have been confounded by hemolysis occurring post mortem. Consequently, only samples under the low ambient temperature condition were analyzed, and all samples were analyzed together. At low ambient temperatures significant changes were observed in ammonia levels between groups (F(4, 24)=10.6, p<0.0001). The MDMA 25 dose produced a significant increase in plasma ammonia levels compared with the saline group as confirmed by a post hoc comparison (p<0.05 vs. saline). Neither MDMA 50 nor MDMA 75 treatments increased plasma ammonia levels significantly. As for some other measures already mentioned, this might be because death occurred so quickly in these groups. In contrast, MCAT caused a decrease in plasma ammonia levels compared with the saline-treated group (p<0.05, Tukey's post hoc test).</p><!><p>The organs examined were brain, heart, lung, liver and kidney. All data were normalized by dividing the organ weight by total body weight. No changes in normalized weight were observed for brain or lung samples for either MDMA or MCAT (data not shown). Normalized weight changes for the other organs are shown in Figs. 5-7 for both MDMA and MCAT.</p><p>For the MDMA data, initial analysis of normalized heart weight by 2-way ANOVA found significant effects of MDMA dose (F(3,72)=4.3, p<0.001) and a significant interaction between MDMA dose and ambient temperature (F(3,72)=2.87, p<0.05), as can be seen in Fig. 5A. 1-way ANOVA conducted separately for each temperature condition found a significant effect of MDMA dose on normalized heart weight in the LT condition (F(3,36)=3.3, p<0.02), and the HT condition (F(3,36)=3.3, p<0.04). In the LT condition, normalized heart rate was not decreased compared to the saline condition after MDMA treatment, but normalized heart weights were reduced in the LT LD50 MDMA and LT 75 MDMA conditions compared to the LT LD25 MDMA treatment condition, which was slightly, but not significantly, elevated compared to the saline condition. In the HT condition, there was a 21% decrease in normalized heart weight in the HT LD75 MDMA group compared to the HT SAL group. MCAT did not affect normalized heart weight (Fig. 5B).</p><p>MDMA and MCAT had pronounced effects upon liver weights that were affected by temperature. For the MDMA normalized liver weight data (Fig. 6A), 2-way ANOVA revealed significant effects of MDMA dose (F(3,72)=4.3, p<0.01) and ambient temperature (F(1,72)=9.7, p<0.003), as well as a significant interaction between MDMA dose and ambient temperature (F(3,72)=3.6, p<0.02). Separate analysis of the data from the LT condition found a significant effect of MDMA dose (F(3,36)=3.5, p<0.03). However, no significant changes were observed for normalized liver weight after MDMA versus the saline condition. Normalized liver weight was significantly increased in the LT MDMA 75 condition compared to the LT MDMA 25 condition. In the HT condition, normalized liver weights were increased by MDMA (F(3,36)=4.3, p<0.02). Indeed, in the HT condition increased normalized liver weights were observed at all MDMA doses, including the MDMA 25, MDMA 50, and MDMA 75 groups compared to SAL (p<0.05, Tukey's post hoc tests). For the MCAT normalized liver weight data, 2-way ANOVA found a significant effect of ambient temperature (Fig. 6B; F(1,24)=9.0, p<0.001) and a significant interaction of ambient temperature with MCAT treatment (F(1,24)=8.9, p<0.001). In the LT condition, post hoc comparisons found no significant effect of MCAT treatment on normalized liver weight, but normalized liver weight was increased by MCAT in the high ambient temperature condition (p<0.05, Tukey's post hoc test).</p><p>The kidneys were only affected by drugs in the HT condition, by either MDMA or MCAT, as is clear in Figs. 7A and 7B. 2-way ANOVA of the MDMA data revealed a significant effect of MDMA treatment (F(3,72)=5.23, p<0.003), but not ambient temperature or the interaction of ambient temperature with MDMA treatment. However, 1-way analyses found that normalized kidney weights were not significantly altered in the LT condition by MDMA (F(3,36)=1.2, NS), but were significantly increased by MDMA in the HT condition (F(3,36)=6.5, p<0.002), which was shown to be significantly different from SAL in the MDMA 50 and MDMA 75 groups in post hoc comparisons (p<0.05). A similar overall pattern was seen for MCAT (Fig. 6B), which was confirmed by a significant effect of MCAT treatment in the 2-way ANOVA (F(1,24)=13.7, p<0.002, as well as a significant interaction between MCAT treatment and ambient temperature (F(1,24)=5.5, p<0.03). Normalized kidney weights were not significantly altered in the LT condition by MCAT either (Tukey's post hoc comparison, p>0.05), but in the HT condition normalized kidney weights were significantly increased by MCAT (Tukey's post hoc comparison vs. SAL, p<0.05).</p><!><p>The studies presented here addressed a specific hypothesis about the causes of amphetamine and cathinone lethality: that ambient temperature affects drug-induced hepatic toxicity and elevates plasma ammonia levels, which may potentially elevate brain glutamate levels. This idea is based in part on a series of studies by Yamamoto, Halpin, and colleagues demonstrating that MA-induced neurotoxicity involves these mechanisms (Halpin et al., 2013; Halpin et al., 2014; Halpin and Yamamoto, 2012). More broadly speaking, stimulant overdose has been associated with hepatic encephalopathy (Kramer et al., 2003). The current data shows that similar mechanisms contribute to acute lethality produced by MDMA, as well as MCAT, and that an important additional factor contributing to lethality is ambient temperature.</p><!><p>Under room temperature conditions, LD25, LD50, and LD75 doses of MDMA induced lethality closely matched those values established in a previous study. Temperature was not elevated by MDMA under these conditions, either in mice that survived the entire 2-hr test period, or mice that died during this time. Moreover, for the LD50 and LD75 doses of MDMA, the temperature at death was decreased by over 2 °C. This certainly suggests that hyperthermia is not necessary for the observation of MDMA-induced lethality. In these same subjects, plasma ammonia increases were observed. Glutamate was not directly measured in the current studies, but the observation of seizure (see Muskiewicz et al., in this volume) is certainly consistent with elevated brain glutamate function. These increases in plasma ammonia were only significant at the lowest dose, but the estimates of ammonia changes at the highest doses may be inaccurate since the animals that died were not included in the analysis. In those mice, ammonia may have risen more than in the mice that survived, but that cannot be determined from the present data. In any case, these data clearly demonstrate that MDMA, at near lethal doses, does increase plasma ammonia levels, as was previously observed for MA (Halpin et al., 2014; Halpin and Yamamoto, 2012). Additionally, there was evidence of hepatic injury, or impairment, based on increases in normalized organ weight under these conditions, and no changes in the other major organs examined.</p><p>Although the room temperature data suggest that hyperthermia is not necessary for MDMA-induced lethality, this is not to say that hyperthermia is not involved. Indeed, hyperthermia was shown to be associated with greatly increased MDMA lethality under HT conditions. Indeed, all MDMA doses induced 100% lethality at high ambient temperatures, and this was associated with hyperthermia as assessed at death, in sharp contrast to observations at room temperatures. This would seem to indicate that the effect of MDMA on temperature homeostasis is not necessarily to induce hypothermia or hyperthermia, but rather to impair the ability to maintain temperature under different ambient temperature conditions. Most preclinical work investigating the effects of MDMA on temperature homeostasis has emphasized hyperthermia because hyperthermia is most often observed in cases of human MDMA overdose. This likely relates to the conditions in which the drug is usually taken (e.g. high ambient temperatures, in a binge-like fashion). Although a human study reported that MDMA produced hyperthermia under both warm and "cold" conditions (Freedman et al., 2005), the warm condition was 30 °C while the "cold" condition was 18 °C. Given that a standard room temperature is often 20 °C this is not a good test of the effects of MDMA under cold conditions, or whether MDMA primarily induces hyperthermia or more generally impairs thermoregulation. It is also difficult to compare warm and cold conditions between humans and rodents, as rodents often prefer temperatures of about 30 °C (Goodrich and Wilk, 1981). In any case, the hyperthermia and increased lethality observed here was also associated with other signs of organ toxicity. Indeed, unlike the LT conditions, increased normalized liver weight was observed for all MDMA doses under HT conditions. Additionally, changes were observed in other organs under HT as well: increased normalized kidney weight at all doses of MDMA, and reduced normalized heart weight at the highest dose of MDMA.</p><p>The data presented here for temperature are consistent with data showing that MDMA can produce both hypothermia and hyperthermia (Miner et al., 2017) and the ability of ambient temperature to affect MDMA-induced neurotoxicity, hepatotoxicity, and lethality (Carvalho et al., 2001; Fantegrossi et al., 2003; Gordon et al., 1991; Sanchez et al., 2004). Hepatotoxicity appears to be one of the main causes of MDMA overdose, although often the time-course of such effects is longer than those observed here. Indeed, there was evidence here that at very high MDMA doses other toxic effects also occur. There were more widespread changes in organ weights under these conditions, a rough indicator of organ dysfunction, and death occurred very rapidly, perhaps even before ammonia levels rose in the blood. Additional studies are needed to clarify these issues. There was certainly evidence of hepatic toxicity at all MDMA doses under high ambient temperature conditions. At these temperatures MDMA has been shown to cause blood clots, bleeding, and liver vacuolation in vivo, and hepatotoxicity in vitro (Carvalho et al., 2002). High plasma ammonia levels have been associated with hepatic disease, and the effects of high plasma ammonia levels on brain glutamate are associated with hepatic encephalopathy (for review see Dabrowska et al. (2018)).</p><p>The quick deaths occurring after administration of MDMA in the HT condition may involve many factors, but one possible explanation may be factors contributing to drug-induced hepatic encephalopathy (HE) (Kramer et al., 2003). In overdoses with these types of complications it is difficult to identify a single cause of death, and certainly multiple treatments are pursued to address the observed physiological symptoms. It is interesting to note that the primary symptoms in such reports include fulminant hepatic failure, rhabdomyolysis, myocardial infarction, and hyperammonemia – a pattern quite similar to that observed here, as well as after experimentally-induced hyperammonemia (Satpute et al., 2014). That the drug-induced-hyperammonemia was central to lethality was shown by the success of an ammonia detoxification treatment (Kramer et al., 2003). Indeed, similar reductions in hyperammonemia after MA treatments reduce neurotoxicity in mice (Halpin and Yamamoto, 2012). That study also showed that glutamate receptor antagonism also reduced neurotoxicity. Seizures, such as were found in the present study after MDMA administration, is commonly found in patients with ALF, which is associated with hyperammonemia (Capocaccia and Angelico, 1991; Sellinger et al., 1968), increased brain glutamate (Takahashi et al., 1991), and microglial activation and astrocyte swelling (Rao et al., 2013).</p><!><p>Cathinones can produce stimulant-like drug overdoses (Marusich et al., 2014; Spiller et al., 2011), and may even have a greater potential for overdose than amphetamines (American Association of Poison Control Centers, 2011; Wood et al., 2013, 2015). However, few animal studies, in which dose and other extraneous factors can be controlled, have addressed this question. Based upon an initial study showing that methylone had a slightly lower LD50 than MA or MDMA (Piao et al., 2015), we predicted that the LD50 values for other cathinones might also be lower than MA or MDMA. However, this was not the case (Muskiewicz et al. this volume). Indeed, at the doses tested, an LD50 values could not even be calculated for many of the cathinones studied. The current study confirmed this lack of lethality for MCAT, under standard room temperature conditions. Moreover, in contrast to the effects of MDMA under similar temperature conditions, MCAT produced hypothermia (~ 5 °C) and reduced plasma ammonia levels. The magnitude of this hypothermia was similar to the decrease in temperature observed in our previous study after MCAT administration. Indeed, hypothermia is observed under standard temperature conditions for a number of cathinone analogues (Aarde et al., 2013; Aarde et al., 2015; Anneken et al., 2017a; Anneken et al., 2017b; Shortall et al., 2013), and certainly much more often than is typically observed for amphetamines.</p><p>In sharp contrast to the effects of MCAT under LT conditions, MCAT was found to be highly lethal, killing 100% of the mice tested, producing hyperthermia and increasing normalized liver and kidney weights under HT conditions. All of these effects were quite similar to MDMA under these conditions, and fit many other clinical and experimental observations for MDMA discussed in the previous section. Unfortunately, this study did not examine wider ranges of MCAT doses, but it nonetheless demonstrates that examining MCAT under standard temperature conditions does not provide a very good indication of its lethal potential. This indicates that SPCs will need to be assessed under HT conditions to give an accurate estimate of the potential of these drugs to affect humans, matching the conditions in which these drugs are commonly consumed by humans. These data additionally suggest that hepatotoxicity may be an important factor in SPC toxicity, when it is observed, although this remains to be determined for a wider range of SPC analogues. However, both clinical reports (Carvalho, Márcia et al., 2012; Marinetti and Antonides, 2013), and in vitro studies (Araújo et al., 2015; Carvalho, M. et al., 2012), are consistent with this suggestion. Moreover, in vitro studies have also shown that SPC hepatotoxicity is temperature-dependent (Valente et al., 2016), again emphasizing the importance of this factor for evaluating the toxicity and lethality of SPCs.</p><!><p>The current findings showed that the effects of MDMA and MCAT on core body temperature and lethality were highly dependent on ambient temperature. Although the initial findings presented here are correlational in nature, they do provide potential hypotheses to be exploited in studies of amphetamine and cathinone overdose, with regard to the fundamental importance of organ damage, elevated plasma ammonia levels and brain glutamate. The lethality of MDMA was potentiated by a high ambient temperature and the limited lethality observed for MCAT under low ambient temperatures was converted into complete lethality at high temperatures. Ambient temperature also had substantial effects on temperature homeostasis. Mice that died after MDMA under low ambient temperature conditions were hypothermic, while those that died under high ambient temperature conditions were hyperthermic. Similarly, MCAT reduced temperature under low ambient temperature conditions, but hyperthermia was observed in mice that died after MCAT under high ambient temperature conditions. Regardless of effects on temperature, there was evidence of hepatic impairments, including increased plasma ammonia under some conditions, which was substantially exacerbated by the high temperature conditions.</p>

PubMed Author Manuscript

Synthesis of trifluoromethyl-substituted pyrazolo[4,3-c]pyridines – sequential versus multicomponent reaction approach

A straightforward synthesis of 6-substituted 1-phenyl-3-trifluoromethyl-1H-pyrazolo[4,3-c]pyridines and the corresponding 5-oxides is presented. Hence, microwave-assisted treatment of 5-chloro-1-phenyl-3-trifluoromethylpyrazole-4-carbaldehyde with various terminal alkynes in the presence of tert-butylamine under Sonogashira-type cross-coupling conditions affords the former title compounds in a one-pot multicomponent procedure. Oximes derived from (intermediate) 5-alkynyl-1-phenyl-3-trifluoromethyl-1H-pyrazole-4-carbaldehydes were transformed into the corresponding 1H-pyrazolo[4,3-c]pyridine 5-oxides by silver triflate-catalyzed cyclization. Detailed NMR spectroscopic investigations (1H, 13C, 15N and 19F) were undertaken with all obtained products.

synthesis_of_trifluoromethyl-substituted_pyrazolo[4,3-c]pyridines_–_sequential_versus_multicomponent

<!>Introduction<!><!>Introduction<!>Chemistry<!><!>Chemistry<!>NMR spectroscopic investigations<!><!>NMR spectroscopic investigations<!>Conclusion<!>

<p>This article is part of the Thematic Series "Multicomponent reactions II".</p><!><p>Fluorine-containing compounds play an important role in medicinal and pharmaceutical chemistry as well as in agrochemistry [1–4]. A popular approach for the modulation of activity consists in the introduction of one or more fluorine atoms into the structure of a bioactive compound. This variation frequently leads to a higher metabolic stability and can modulate some physicochemical properties such as basicity or lipophilicity [1–2]. Moreover, incorporation of fluorine often results in an increase of the binding affinity of drug molecules to the target protein [1–2]. As a consequence, a considerable amount – approximately 20% – of all the pharmaceuticals being currently on the market contain at least one fluorine substituent, including important drug molecules in different pharmaceutical classes [5]. Keeping in mind the above facts, the synthesis of fluorinated heterocyclic compounds, which can act as building blocks for the construction of biologically active fluorine-containing molecules, is of eminent interest. In the field of pyrazoles, pyridines and condensed systems thereof trifluoromethyl-substituted congeners can be found as partial structures in several pharmacologically active compounds. In the pyridine series the HIV protease inhibitor Tipranavir (Aptivus®) [6] may serve as an example, within the pyrazole-derived compounds the COX-2 inhibitor Celecoxib (Celebrex®) is an important representative (Figure 1) [7].</p><!><p>Important drug molecules containing a trifluoromethylpyridine, respectively a trifluoromethylpyrazole moiety.</p><!><p>In continuation of our program regarding the synthesis of fluoro- and trifluoromethyl-substituted pyrazoles and annulated pyrazoles [8–9] we here present the synthesis of trifluoromethyl-substituted pyrazolo[4,3-c]pyridines. The latter heterocyclic system represents the core of several biologically active compounds, acting, for instance, as SSAO inhibitors [10], or inhibitors of different kinases (LRRK2 [11–12], TYK2 [13], JAK [14–15]).</p><!><p>The construction of the pyrazolo[4,3-c]pyridine system can be mainly achieved through two different approaches. One strategy involves the annelation of a pyrazole ring onto an existing, suitable pyridine derivative [16]. Alternatively, the bicyclic system can be accessed by pyridine-ring formation with an accordant pyrazole precursor. Employing the latter approach we recently presented a novel method for the synthesis of the pyrazolo[4,3-c]pyridine system by Sonogashira-type cross-coupling reaction of easily obtainable 5-chloro-1-phenyl-1H-pyrazole-4-carbaldehydes with various alkynes and subsequent ring-closure reaction of the thus obtained 5-alkynyl-1H-pyrazole-4-carbaldehydes in the presence of tert-butylamine [17]. Furthermore, we showed that the oximes derived from the before mentioned 5-alkynylpyrazole-4-carbaldehydes can be transformed into the corresponding 1-phenylpyrazolo[4,3-c]pyridine 5-oxides [17].</p><p>For the synthesis of the title compounds a similar approach was envisaged. As starting material the commercially available 1-phenyl-3-trifluoromethyl-1H-pyrazol-5-ol (1) was employed which, after Vilsmaier formylation [18] and concomitant transformation of the hydroxy function into a chloro substituent by treatment with excessive POCl3, gave the chloroaldehyde 2 [19] (Scheme 1). Although Sonogashira-type cross-coupling reactions are preferably accomplished with iodo(hetero)arenes – considering the general reactivity I > Br/OTf >> Cl [20] – from related examples it was known that the chloro atom in 5-chloropyrazole-4-aldehydes is sufficiently activated to act as the leaving group in such kind of C–C linkages [17]. Indeed, reaction of chloroaldehyde 2 with different alkynes 3a–c under typical Sonogashira reaction conditions afforded the corresponding cross-coupling products 4a–c in good yields (Scheme 1). In some runs compounds of type 8 were determined as byproducts in differing yields, but mostly below 10%, obviously resulting from addition of water to the triple bond of 4 under the reaction conditions (or during work-up) and subsequent tautomerization of the thus formed enoles into the corresponding ketones. The hydration of C–C triple bonds under the influence of various catalytic systems, including also Pd-based catalysts, is a well-known reaction [21–22]. It should be emphasized that NMR investigations with compounds 8a,c unambiguously revealed the methylene group adjacent to the pyrazole nucleus and the carbonyl moiety attached to the substituent R originating from the employed alkyne.</p><!><p>Synthesis of the title compounds.</p><!><p>In the next reaction step, alkynylaldehydes 4a,b were cyclized into the target pyrazolo[4,3-c]pyridines 5a,b in 71%, resp. 52% yield by reaction with tert-butylamine under microwave assistance [17]. In view of the fact, that the two-step conversion 2→4a,b→5a,b was characterized by only moderate overall yields (59%, resp. 43%) it was considered to merge these two steps into a one-pot multicomponent reaction. The latter type of reaction attracts increasing attention in organic chemistry due to its preeminent synthetic efficiency, also in the construction of heterocyclic and condensed heterocyclic systems [23–27]. After testing different reaction conditions we found that microwave heating of the chloroaldehyde 2 with tert-butylamine in the presence of 6 mol % of Pd(PPh3)2Cl2 afforded the desired pyrazolopyridines 5a–c in high (5a: 89%, 5c: 92%) respectively acceptable yields (5b: 51%) in a single one-pot and copper-free reaction step (Scheme 1). It should be mentioned that compounds 5 are also accessible by heating of ketones 8 with ammonium acetate in acetic acid according to a procedure described in [28]. Following this way, 5a and 5c were obtained in 70% yield from the corresponding ketones 8a and 8c. Although ketones 8 were only obtained as byproducts, the latter transformation allowed increasing the overall yield of compounds 5 through this 'bypass'.</p><p>In order to gain access to the corresponding N-oxides of type 7, aldehydes 4a–c were transformed into the corresponding oximes 6a–c by reaction with hydroxylamine hydrochloride in ethanol in the presence of sodium acetate (Scheme 1). Subsequent treatment of the oximes with AgOTf in dichloromethane [29] finally afforded the corresponding pyrazolo[4,3-c]pyridine 5-oxides 7a–c by a regioselective 6-endo-dig cyclisation [30] in high yields. Moreover, we tested an alternative approach to access compounds 7 through multicomponent reactions (MCR). Attempts to react chloroaldehyde 2 with hydroxylamine hydrochloride and an alkyne 3 in the presence of a suitable catalytic system were not successful. However, after conversion of 2 into the corresponding aldoxime 9 the latter could be transformed into the N-oxides 7a and 7b by reaction with alkynes 3a and 3b, respectively, employing Pd(OAc)2 as the catalyst and under microwave irradiation (Scheme 1). Although a number of different reaction conditions were tested, we were not able to increase the yields in excess of 30%. Thus, with respect to the overall yields the successive approach 2→4→6→7 (overall yields: 6a: 62%, 6b: 43%) here is still advantageous compared to the multicomponent reaction following the path 2→9→7. Azine N-oxides of type 7 are estimated to be of particular interest due to the possibility of further functionalization adjacent to the nitrogen atom (position 4), for instance by palladium-catalyzed direct arylation reactions [31].</p><!><p>In Supporting Information File 1 the NMR spectroscopic data of all compounds treated within this study are indicated. Full and unambiguous assignment of all 1H, 13C, 15N and 19F NMR resonances was achieved by combining standard NMR techniques [32], such as fully 1H-coupled 13C NMR spectra, APT, HMQC, gs-HSQC, gs-HMBC, COSY, TOCSY, NOESY and NOE-difference spectroscopy.</p><p>In compounds 4–7 the trifluoromethyl group exhibits very consistent chemical shifts, ranging from δ(F) −60.8 to −61.9 ppm. The fluorine resonance is split into a doublet by a small coupling (0.5–0.9 Hz) due to a through-space (or possibly 5J) interaction with spatially close protons (4: CHO; 6: CH=N; 5 and 7: H-4). Reversely, the signals of the latter protons are split into a quartet (not always well resolved). The corresponding carbon resonance of CF3 is located between 120.2 and 121.2 ppm with the relevant 1J(C,F) coupling constants being approximately 270 Hz (269.6–270.6 Hz). As well, the signal of C-3 is always split into a quartet (J ~ 40 Hz) due to the 2J(C,F3) coupling.</p><p>As the 15N NMR chemical shifts were determined by 15N,1H HMBC experiments the resonance of (pyrazole) N-2 was not captured owing to the fact that this nitrogen atom lacks of sufficient couplings to protons, thus disabling the necessary coherence transfer (19F,15N HMBC spectra were not possible with the equipment at hand). For N-1, with pyrazole derivatives 4 and 6 remarkably larger 15N chemical shifts were detected (−158.8 to −160.2 ppm) compared to the corresponding signals for pyrazolopyridines 5 and 7 (−182.2 to −185.9 ppm). When switching from an azine to an azine oxide partial structure (5→7) the N-5 resonance exhibits an explicit upfield shift (15.6–18.3 ppm), being typical for the changeover from pyridine to pyridine N-oxide [33].</p><p>NMR experiments also allowed the determination of the stereochemistry of oximes 6: considering the size of 1J(N=C-H) which is strongly dependent on lone-pair effects [34] as well as the comparison of chemical shifts with those of related, unambiguously assigned oximes [17] reveals E-configuration at the C=N double bond.</p><p>With byproduct 8a the position of the carbonyl group unequivocally follows from the correlations between phenyl protons and the carbonyl C-atom and, reversely, from those between the methylene protons with pyrazole C-4 and pyrazole C-5 (determined by 13C,1H HMBC).</p><p>In Figure 2 essential NMR data for the complete series of type c (4c, 5c, 6c, 7c) are displayed, which easily enables to compare the notable chemical shifts and allows following the trends described above.</p><!><p>1H (in italics, red), 13C (black), 15N (in blue) and 19F NMR (green) chemical shifts of compounds 4c, 5c, 6c and 7c (in CDCl3).</p><!><p>Full experimental details as well as spectral and microanalytical data of the obtained compounds are presented in Supporting Information File 1.</p><!><p>To sum up, the presented approach represents a simple method for the synthesis of 6-substituted 1-phenyl-3-trifluoromethyl-1H-pyrazolo[4,3-c]pyridines 5 and the analogous 5-oxides 7 starting from commercially available 1-phenyl-3-trifluoromethyl-1H-pyrazol-5-ol (1). In the case of the former (5) the described multicomponent reaction approach is superior compared to the sequential one, whereas the step-by-step synthesis of N-oxides 7 is still characterized by higher overall yields. In addition, in-depth NMR studies with all synthesized compounds were performed, affording full and unambiguous assignment of 1H, 13C, 15N and 19F resonances and the designation of ascertained heteronuclear spin-coupling constants.</p><!><p>Experimental details and characterization data.</p>

PubMed Open Access

\xe2\x80\x98New Trends for Metal Complexes with Anticancer Activity\xe2\x80\x99

Summary Medicinal inorganic chemistry can exploit the unique properties of metal ions for the design of new drugs. This has, for instance, led to the clinical application of chemotherapeutic agents for cancer treatment, such as cisplatin. The use of cisplatin is, however, severely limited by its toxic side effects. This has spurred chemists to employ different strategies in the development of new metal-based anticancer agents with different mechanisms of action. Recent trends in the field are discussed in this review. These include the more selected delivery and/or activation of cisplatin-related prodrugs and the discovery of new non-covalent interactions with the classical target, DNA. The use of the metal as scaffold rather than reactive centre and the departure from the cisplatin paradigm of activity towards a more targeted, cancer cell-specific approach, a major trend, are discussed as well. All this, together with the observation that some of the new drugs are organometallic complexes, illustrates that exciting times lie ahead for those interested in \xe2\x80\x98metals in medicine\xe2\x80\x99.

\xe2\x80\x98new_trends_for_metal_complexes_with_anticancer_activity\xe2\x80\x99

Introduction<!>New modes of interaction with the classical target, DNA<!>Non-covalent interactions with DNA<!>The metal as scaffold<!>Proteins and enzymes as non-classical targets<!>From inorganic to bioorganometallic drugs<!>Conclusions

<p>Medicinal inorganic chemistry [••1-3] is a field of increasing prominence as metal-based compounds offer possibilities for the design of therapeutic agents not readily available to organic compounds. The wide range of coordination numbers and geometries, accessible redox states, thermodynamic and kinetic characteristics, and the intrinsic properties of the cationic metal ion and ligand itself offer the medicinal chemist a wide spectrum of reactivities that can be exploited. Although metals have long been used for medicinal purposes in a more or less empirical fashion [4], the potential of metal-based anticancer agents has only been fully realised and explored since the landmark discovery of the biological activity of cisplatin [5]. To date, this prototypical anticancer drug remains one of the most effective chemotherapeutic agents in clinical use. It is particularly active against testicular cancer and, if tumours are discovered early, an impressive cure rate of nearly 100% is achieved. The clinical use of cisplatin against this and other malignancies is, however, severely limited by dose-limiting side-effects such as neuro-, hepato- and nephrotoxicity [5]. In addition to the high systemic toxicity, inherent or acquired resistance is a second problem often associated with platinum-based drugs, with further limits their clinical use. Much effort has been devoted to the development of new platinum drugs and the elucidation of cellular responses to them to alleviate these limitations [5,6]. These problems have also prompted chemists to develop alternative strategies based on different metals and aimed at different targets. We summarize here recent activities in the field of metal-based anticancer drugs. Space limitations mean that this overview is not comprehensive, but aims to highlight significant advances and illustrate emerging trends.</p><!><p>In classical chemotherapy, anticancer agents target DNA directly according to the cisplatin paradigm to generate lesions which ultimately trigger cell death. Much effort has been directed towards combatting the high systemic toxicity of traditional platinum anticancer agents by designing drug delivery systems capable of delivering platinum to tumour cells only.</p><p>A recent example of the latter strategy is the encapsulation of cisplatin and carboplatin in the hollow protein cage of the iron storage protein ferritin, which can be internalized by some tumour tissues. Indeed, the drug-loaded protein showed cytotoxic activity against the rat pheochromocytoma cell line (PC12) [7]. In a different approach, minicells, bacterially-derived 400 nm anucleate particles, have been packed with chemotherapeutics, such as cisplatin, and labelled with bispecific antibodies. This resulted in endocytosis and ultimately drug release in cancer cells [8].</p><p>Platinum(iv) prodrugs can be used to overcome some of the problems associated with cisplatin and its analogues [9]. The high kinetic inertness of Pt(iv) complexes relative to their Pt(ii) analogues introduces drug stability and the two extra ligands on the octahedral metal centre offer many possibilities for modification of pharmacokinetic parameters. As intracellular reduction of platinum(iv) to platinum(ii) is usually essential for activation and subsequent cytotoxicity, these prodrugs essentially present better ways of delivering cisplatin (or its analogues) to the target tumour cell. Synthetic advances now allow the inclusion of various bioactive ligands in the axial positions, and, hence, targeting to specific types of cancer cells.</p><p>The observation that estrogen receptor-positive, ER(+), breast cancer cells treated with estrogen are sensitized to cisplatin, led Lippard et al. to synthesize the estrogen-tethered platinum(iv) complex 1 (see Figure 1a). Intracellular reduction releases one equivalent of cisplatin and two equivalents of estradiol. The latter induces up-regulation of high-mobility group (HMG) domain protein HMGB1, a protein that shields platinated DNA from nucleotide excision repair [10].</p><p>Along a similar vein, but with the intention of overcoming platinum drug resistance rather than cell sensitization, Dyson et al. used the Pt(iv) complex 2 (Figure 1a) to target cytosolic glutathione-S-transferase (GST), which constitutes the main cellular defence against xenobiotics. Ethacraplatin (2) has the GST inhibitor ethacrynic acid, a diuretic in clinical use, attached. Reduction of ethacraplatin in the cell results in the release of two equivalents of a potent GST inhibitor together with one equivalent of the cytotoxic cisplatin [11].</p><p>Lippard et al. have tackled the problem of drug delivery by attaching a Pt(iv) prodrug to functionalized soluble single-walled carbon nanotubes (SWNT), highly effective carriers that can transport various cargos across the cell membrane through clathrin-dependent endocytosis [••12]. Platinum(iv) complex 3 has two different axial ligands (Figure 1a) and binds non-covalently to the nanotube surface and one SWNT longboat carries on average 65 platinum complexes. The conjugate shows a substantial increase in cytotoxicity compared to the untethered complex and to cisplatin.</p><p>Sadler et al. are using a strategy that relies on the photochemical activation of platinum(IV) drugs to release active antitumour agents, rather than spontaneous intracellular reduction. This can then provide localized treatment of cancers accessible to irradiation. The trans-dihydroxy platinum(iv) prodrugs are non-toxic in the dark and incorporate two azide ligands, either positioned cis (4) [•13] or trans (5) [14] to each other (Figure 1b). Irradiation results in growth inhibition of human bladder cancer cells (5) and cytotoxicity towards human skin cells (HaCaT keratinocytes) (4 and 5). The discovery that the trans-isomer, a potential precursor of the inactive transplatin, is as active as cisplatin is notable [14]. It is also notable that the cis-azide complex is not cross-resistant to cisplatin and different DNA platination pathways seem to be involved [•13]. Incorporation of a pyridine ligand into these complexes can greatly increase their potency (FS Mackay et al., unpublished).</p><p>The activity of the clinically ineffective transplatin itself is markedly enhanced upon irradiation [15]. This, together with the observed increase in cytotoxicity upon irradiation of dirhodium complexes [16], further illustrates the potential of photoactivation.</p><p>A variety of ruthenium complexes have been designed which interact specifically with the classical target, DNA [17,18]. A family of ruthenium(ii)-arene complexes developed by Sadler et al. [•19], for instance, exhibits high in vitro and in vivo anticancer activity [20]. For example, the direct coordinative binding of the monofunctional Ru-arene complex (6) (Figure 1c) to N7 of G bases in DNA is complemented by intercalative binding of the biphenyl ligand and specific hydrogen bonding interactions of the ethylenediamine NH2 groups with C6O of guanine. These additional interactions result in unique binding modes to duplex DNA and induce different structural distortions in DNA compared to cisplatin, which may explain why these complexes are not cross-resistant with cisplatin [18]. Interestingly, this chemistry has recently been extended to include osmium(ii)-arene analogues, such as 7 (Figure 1c), whose hydrolytic properties can be tuned to achieve promising activity against human A549 and A2780 ovarian cancer cells [•21].</p><p>Other ruthenium-based anticancer drugs, including the NAMI-A (9) and KP1019 (10) drugs which are under clinical evaluation, have different modes of action and specifically aim at non-classical targets such as gene products and cellular transduction pathways [22]. This shift in interest, which complements the classical approach, is one of the major trends in the field and is discussed in the later sections of this review.</p><p>Although much less studied than the metallodrug-DNA interactions, the interaction of metallodrugs with protein targets and the proteome deserves more attention, especially since such studies will not only shed light on the mechanisms of action, but also help to identify new targets for drug therapy [17,23]. Metallodrug-protein interactions studied by various advanced analytical techniques have been recently reviewed [23]. Recent characterization of protein adducts of platinum and ruthenium anticancer drugs by X-ray crystallography [24,25] or advanced mass spectrometry [26,27] show the timeliness of this approach.</p><!><p>Single-stranded ends of human telomeric DNA, which consist of guanine-rich TTAGGG repeats known to fold into G-quadruplex structures (Figure 2a), provide interesting targets for drug design. Telomeric DNA shortens after every cell division and after critical shortening of the telomeres, cells stop dividing and commit suicide. Telomerase, however, maintains the length of the telomeric DNA and overexpression of this enzyme endows the (cancer) cell with the ability to replicate indefinitely and thus proliferate. Since telomerase accepts only the single-stranded overhang, stabilization of the G-quadruplexes provides an attractive means of preventing telomerase from maintaining telomere length. The Ni(ii)-salphen complex (11) (Figure 2a) incorporates the main requirements for quadruplex-stabilizing molecules, i.e. a π-delocalized system prone to stacking on a G-quartet, a positive charge that is able to lie in the centre of the quartet, and finally, positively-charged substituents which can interact with the grooves and loops of the quadruplex. Indeed, 11 induces a high degree of quadruplex DNA stabilization and telomerase inhibition with telEC50 values in the range of 0.1 μM [28].</p><p>An important challenge in this field is the design of complexes that bind selectively to quadruplex over duplex DNA. Whereas the Ni(ii)-salphen complex 11 shows selectivity of > 50-fold, the manganese porphyrin 12 (Figure 2a) which follows the same design criteria as mentioned previously, shows an exceptional 10,000-fold selectivity for quadruplex over duplex DNA (IC50 of around 0.6 μM) [•29]. These results illustrate the potential of metal complexes as telomere-targeted chemotherapeutics.</p><p>Besides direct coordinative binding of metallo-agents to DNA bases, other potential DNA binding modes include intercalation and groove binding. The latter modes are generally non-covalent in nature.</p><p>Farrell et al. have described a discrete binding mode based on interactions that utilize exclusively backbone functional groups [•30]. The new binding mode was observed in the crystal structure of a double-stranded B-DNA dodecamer with TriplatinNC (13) (Figure 2b). This trinuclear Pt(ii) complex is related to the trinuclear trans-platinum drug BBR3464, but lacks the reactive chloride ligands. TriplatinNC displays micromolar activity against human ovarian cancer cell lines [31]. The phosphate-selective complex binds through a multitude of specific "phosphate clamps" (see Figure 2b), bifurcated ammine(NH)⋯phosphate(O)⋯amine hydrogen bonds. A series of such phosphate clamps with one strand of DNA results in so-called "backbone tracking", and a combination of two interstrand clamps gives rise to (minor) "groove-spanning". Both interactions may be present in solution [•30].</p><p>A X-ray crystallographic study of the adduct between triple helicate [Fe2L3]4+ (Figure 2c) and palindromic DNA 5′-d(CGTACG) by Hannon, Coll et al. reveals the metallosupramolecular helicate comfortably occupying the central hydrophobic cavity of a three-way (Y-shaped) junction (Figure 2c) [•32]. The positive charge of the helicate, together with the large hydrophobic surface of the aromatic rings (together with its π-stacking potential), are the driving forces behind this specific interaction. This unprecedented mode of non-covalent DNA recognition shows that three-way junctions, naturally occurring both in DNA and RNA, provide a new structural target for design of novel, highly specific drugs [•32]. In principle, the palindromic sequence allows for the formation of any oligomeric formulation through Watson-Crick hydrogen bonding, thus presenting a dynamic combinatorial library. Addition of the triple helices then drives the selective formation of the three-junction member of this library. The DNA junction recognition compliments the previously reported major groove binding of the iron metallohelicates, which led to remarkable intramolecular DNA coiling [33]. The ruthenium metallohelicate induces a similar bending/coiling effect, further illustrating that the cylinder is responsible for this effect [34]. The Ru triple helicate exhibits cytotoxicity towards human breast cancer HBL-100 and T47D cells, albeit with modest activity (2-5 fold less potent than cisplatin). Related unsaturated dinuclear ruthenium double helicates, capable of classic coordinative binding to DNA, show greatly improved cytotoxicity towards the same cell lines (30-fold more active than cisplatin). These complexes illustrate the many possibilities of using metallosupramolecular architectures in anticancer drug design [35].</p><p>These examples not only illustrate the role of non-covalent DNA interactions, but also the newly-emerging trend of using the metal in a scaffold, rather than as the reactive centre. In a highly modular approach, the use of a metal centre as a building block allows for the spatial orientation of other functionalities (as part of the ligands), which in turn interact favourably with the target via, for instance, hydrogen bonds (phosphate clamps) or π-stacking interactions (helicates). This trend will be discussed in the next section.</p><!><p>Metals ions have been traditionally included in anticancer agents to exploit their reactivity and have been particularly attractive because of the exceptionally wide range of reactivities available. On the other hand, metals can also be used as building blocks for well-defined, three-dimensional constructs. In this way, the availability of many different coordination geometries allows for the synthesis of structures with unique stereochemistry and orientation of organic ligands and structures which are not accessible through purely organic, carbon-based compounds. The kinetic inertness of the coordination/organometallic bonds make these compounds in principle behave like organic compounds. This approach immensely expands our ability to chart biologically-relevant chemical space [••36]. The group of Meggers has pioneered this approach in their development of organometallic ruthenium complexes that mimic organic enzyme inhibitors. The natural product staurosporine, for instance, is a highly potent inhibitor for various kinases (Figure 3a). Meggers et al. replaced the carbohydrate unit with ruthenium fragments. Structural variation by simple substitution of the ligands on the metal to optimize the enzyme-inhibitor interactions has resulted in the discovery of nanomolar and even picomolar protein kinase inhibitors (Figure 3b) [37]. The co-crystal structures of Pim-1 with the organometallic complexes nicely illustrate all salient features of these potent kinase inhibitors (Figure 3c) [38]. The relevance of these organometallic inhibitors as anticancer agents has been demonstrated recently. They are highly cytotoxic towards human melanoma cells. The organometallic GSK-3β inhibitor DW1/2 is a potent activator of p53 and thus induces p53-activated apoptosis via the mitochondrial pathway in otherwise highly chemoresistant melanoma cells [39]. The anticancer agent DW1/2 works by specifically targeting a protein, rather than DNA. The development of novel drugs with non-classical protein targets is becoming a major new theme in metal-based drug development and will be discussed in the next section.</p><!><p>Traditional anticancer drugs that target DNA make use of the fact that malignant cells divide rapidly. A drawback of this strategy is that rapidly dividing healthy cells are affected as well, causing severe toxic side-effects. Alternatively, the design of novel agents that target cellular signalling pathways specific to cancer cells would therefore be preferred. As genomics and proteomics have resulted in an explosion of available information concerning the biology of cancer cells, such targeted therapies have come within reach and offer much potential. This current shift of focus will be illustrated with a few typical examples [17].</p><p>Human thioredoxin reductase (hTrxR) is associated with many cellular processes such as antioxidant defence and redox homeostasis. hTrxR is found at elevated levels in human tumour cell lines. A strong connection with the apoptosis regulator protein p53 has been established and it is strongly associated with tumour proliferation, making hTrxR an interesting target for anticancer drugs [40]. Gold(i) complexes are amongst the most potent inhibitors of hTrxR, a feature attributable to the high electrophilicity of Au(i) and its preference for the selenocysteine residue of hTrxR. For example, phosphole-gold(i) complexes (14, Figure 4a) are highly potent, nanomolar inhibitors of hTrxR and the related human glutathione reductase (hGR) [41]. A crystal structure of 14 with hGR shows, surprisingly, the coordinative binding of one phosphole-gold unit to an exposed cysteine and a second gold atom that has lost both its chloride and phosphole ligand to form a linear S-Au-S adduct at the active site (Figure 4a). IC50 values for gliobastoma cells are in the 5-15 μM range.</p><p>This is one example from the active field of gold anticancer drugs [42], many of which target mitochondria, increasingly recognized as a regulator of cell death. Another promising group of gold-containing anticancer agents are the gold(iii) porphyrins (15) studied by Che et al. (Figure 4b). These complexes exhibit potent in vitro and in vivo anticancer activity towards hepatocellular and nasopharyngeal carcinoma [•43]. A functional proteomics study indeed suggests involvement of the mitochondria in the induced apoptosis [•44]. Finally, gold(iii) thiocarbamates are more cytotoxic in vitro than cisplatin, including intrinsically-resistant cell lines. The primary target is thought to be the proteasome, inhibition of which results in induction of apoptosis [45].</p><p>Other metal-containing anticancer agents are known to target specific enzymes. The cobalt-alkyne analogue 16 of the non-steroidal anti-inflammatory drug aspirin (acetylsalicylic acid) (Figure 4c) exhibits high cytotoxicity in breast cancer cell lines. The cytotoxicity correlates with cyclooxygenase (COX) inhibition. COX is involved in eicosanoid metabolism and interference with this pathway is a promising strategy for the development of new cytostatics [46].</p><p>Hambley et al. have explored the use of cobalt-containing compounds for the selective inhibition of enzymes involved in the process of tumour metastasis [47]. Their interesting strategy focuses on the selective delivery of the established maxtrix metalloproteinase (MMP) inhibitor marimastat by complexing it to a 'chaperone' Co(iii)-complex (17, Figure 4c). MMPs are overexpressed in tumour cells and high levels of MMPs in cancer patients correlate with poor prognosis. The Co(iii) carrier provides an inert framework for the transportation of the inhibitor and the prodrug is activated by a bioreduction pathway generating the more labile Co(ii)-complex, which leads to inhibitor release. Back-oxidation to the prodrug would be prevented by the hypoxic nature of tumour tissue, thus achieving selective release in targeted cells only. showed that Prodrug 17 shows significantly more growth inhibition towards 4T1.2 tumours in vivo in mice than marimastat alone, but also, unexpectedly, that both the inhibitor and the prodrug complex potentiates metastasis [47].</p><p>Inspired by the remarkable properties of NAMI-A (9), a compound devoid of in vitro cytotoxicity but capable of in vivo metastasis inhibition, metastasis was identified as a primary new target [•48] and a number of other ruthenium(iii) complexes have been studied for such potential activity [•48]. Remarkably, the lead complexes of the RAPTA family of ruthenium(ii)-arene compounds (18, Figure 4d) developed by Dyson et al. also showed the same low in vitro cytotoxicity but in vivo inhibition of lung metastases in CBA mice bearing MCa mammary carcinoma [49].</p><p>These examples illustrate the need to develop new assays that look beyond traditional in vitro cytotoxicity tests [22].</p><!><p>The fact that many of the compounds mentioned above are organometallic complexes illustrates the emergence of the field of medicinal organometallic chemistry. More generally, bioorganometallic chemistry is relatively new, but has already led to exciting developments [50]. A prototypical example of a promising organometallic (pro)drug is the now well-established ferrocifen (19, Figure 4e) system. Ferrocifen, in contrast to its parent tamoxifen, is active against both ER(+) and ER(−) human breast cancer cell lines. The antiproliferative effect arises from the anti-estrogenic effect of the tamoxifen-like unit combined with cytotoxicity caused by the redox properties of the attached ferrocene group. Electrochemical studies on a variety of ferrocifen-like compounds have revealed a structure-activity relationship and thus established minimal structural requirements for cytotoxic effect [••51]. Organometallic iron complexes with nucleosides appended to either a ferrocenyl [52] (21) or an η4-butadiene-Fe(CO)3 group[53] (20) also show pronounced cytotoxic activity resulting induction of apoptosis.</p><p>Metzler-Nolte et al. have reported the directed nuclear delivery of an organometallic cobalt compound (22). To achieve this, the SV4-40 T antigen nuclear localization signal (NLS) was modified with a cobaltocenium cation and other groups. These cobaltocenium-NLS-bioconjugates present an intriguing opportunity for the targeted delivery of therapeutics to the cell nucleus [•54].</p><p>The large diversity in recently-reported organometallic anticancer complexes, illustrate that the full arsenal of synthetic organometallic chemistry, is now available to the medicinal chemist. Many more exciting developments can therefore be expected.</p><!><p>This survey of recent literature illustrates that many different new creative approaches are being taken towards the design of innovative metal-based anticancer drugs. The clinical success of cisplatin remains a stimulus for the development of new complexes that address the downsides associated with cisplatin, especially the systemic toxicity and (acquired) resistance. Targeted delivery and/or controlled prodrug activation, be it by light, intracellular reduction or other means, hold the promise of more selective and effective drug administration. The field of classical chemotherapy with DNA as the established target continues to produce interesting discoveries. A clearly discernible, emerging trend, however, is the departure from the cisplatin paradigm of activity. The newly discovered, mainly non-covalent DNA interactions offer a glimpse of the rich chemistry that remains to be discovered, most possibly with applications reaching further than medicinal chemistry. The concept of the metal as scaffold for the construction of unique, yet well-defined three-dimensional structures, rather than reactive centre, holds much promise. This highly modular approach, combined with currently available combinatorial techniques and knowledge of supramolecular chemistry, yields a very powerful method for optimizing drug interactions with carefully selected targets.</p><p>Future clinical success will benefit from targets which are highly specific for cancer cells. The rapidly expanding knowledge of their cellular characteristics offers many new opportunities for drugs that show low systemic toxicity and efficiently tackle the problem of drug resistance. The different examples mentioned in this review offer a promising start. Finally, the advent of medicinal bioorganometallic chemistry has further expanded the toolbox of the medicinal inorganic chemist. The nature of the research will rely ever more heavily on interdisciplinary collaboration, but many exciting discoveries and applications almost certainly lie ahead.</p>

PubMed Author Manuscript

Charge Neutralization Drives the Shape Reconfiguration of DNA Nanotubes

Reconfiguration of membrane protein channels for gated transport is highly regulated under physiological conditions. However, a mechanistic understanding of such channels remains challenging owing to the difficulty in probing subtle gating-associated structural changes. Herein, we show that charge neutralization can drive the shape reconfiguration of a biomimetic 6-helix bundle DNA nanotube (6HB). Specifically, 6HB adopts a compact state when its charge is neutralized by Mg2+; whereas Na+ switches it to the expanded state, as revealed by MD simulations, small-angle X-ray scattering (SAXS), and FRET characterization. Furthermore, partial neutralization of the DNA backbone charges by chemical modification renders 6HB compact and insensitive to ions, suggesting an interplay between electrostatic and hydrophobic forces in the channels. This system provides a platform for understanding the structure-function relationship of biological channels and designing rules for the shape control of DNA nanostructures in biomedical applications.

charge_neutralization_drives_the_shape_reconfiguration_of_dna_nanotubes

<p>The shape reconfiguration of membrane protein channels plays an important role in living systems. It is closely related to the gating of ions, water, and other entities that are vital for many cell functions. These shape changes are often stimulated by membrane tension and electric fields.[1] Although the ionic selectivity, rectification, and gating function of membrane protein channels have been well studied, elucidation of associated subtle structural changes remains challenging.</p><p>Artificial nanotubes (for example, carbon nanotubes) have emerged as promising models for biological channels owing to their nanoscale features and tailorable properties.[2] Because of the endogenous nature and high programmability of DNA, self-assembled DNA nanotubes have attracted intense interest.[3] These DNA nanotubes, when appropriately modified, can be readily inserted into membranes[4] to function as biomimetic channels. Molecular dynamics (MD) simulations have also revealed interesting properties,[5] including ion flow and gating-like behaviors in these DNA-based nanochannels.[6]</p><p>In this work, we designed a computational/experimental approach to study the charge neutralization-induced shape reconfiguration of a 6-helix bundle (6HB) DNA nanotube with three typical states, namely, expanded, compact, and partially compact states. Specifically, by combining MD simulation with structural analysis using small-angle X-ray scattering (SAXS) and Förster resonance energy transfer (FRET), we demonstrated that 6HB adopted a more compact and less expanded shape in a solution with Mg2+ as compared to that with Na+. At lower ion concentrations, 6HB underwent considerable shape expansion; whereas 6HB with an ethyl-phosphorothioate substitution of the DNA backbone remained compact within a volume with a smaller radius.</p><p>A typical membrane protein channel undergoes the transition from the closed state to the open state to function in physiological conditions,[7] probably via a partially or transiently closed state.[8] Similarly, our MD simulations of the biomimetic 6HB under three different physiological conditions (Na+ and Mg2+ conditions or with DNA backbone modification) show that the conformations of 6HB can adopt three states, an expanded state, a compact state, and a partially compact state (Figure 1). In the expanded state, the adjacent helixes within 6-HB were repelled by the electrostatic repulsion to form an "O" shape, whereas in the compact state, the adjacent helixes kept a "II" shape with balanced electrostatic repulsion, and in the partially compact state, owing to the elimination of the electrostatic repulsion by the ethyl-phosphorothioate, the adjacent helixes formed an "8" shape. The possibility of shape reconfiguration of the 6HB under different physiological conditions facilitates its potential applications as biomimetic channels.</p><p>The 70-ns MD simulations were carried on the following four systems: 6HB solvated with 250 mM Na+ (NaL) and 500 mM Na+ (NaH) and 6HB solvated with 125 mM Mg2+ (MgL) and 250 mM Mg2+ (MgH). 6HB is composed of 936 bases, with a channel inner diameter of 2.0 nm, an outer diameter of approximately 6.0 nm, and a length of 20.7 nm (Supporting Information, Figures S1-S4 and Tables S4-S8). The simulation results (Figure 2a) show that the nanostructures of 6HB with both Na+ and Mg2+ counterions at low concentrations are in expanded state. With the increase of ion concentration, 6HB shows a more compact conformation. In the presence of Mg2+, 6HB nanostructures have much smaller inner volumes (Figure 2b and the Supporting Information, Figure S10), inter-helix distances, and cross-sectional area compared to the case with Na+ (Figure S5). Moreover, our calculations of the stretch modulus and persistence length of 6HB in Na+ and Mg2+ (Figure 2c and the Supporting Information, Figure S9 and Table S1) showed that 6HB with Mg2+ are more rigid along the length direction. Experimentally, the structural difference of 6HB in different solutions (12.5 mM Mg2+, 50.0 mM Mg2+, 125.0 mM Mg2+, and 100.0 mM Na+) was first examined by gel electrophoresis on native 6% polyacrylamide gel (PAGE). As shown in the Supporting Information, Figure S11 a, there were no obvious differences between the bands of the magnesium buffered samples (12.5, 50.0, 125.0 mM Mg2+), indicating well-folded 6HB-DNA nanostructures, while the 100.0 mM sodium buffered sample exhibited evident dispersion on the band, indicating a possible loosely folded 6HB structure under this condition.</p><p>FRET and SAXS were utilized to track the change in distance between two DNA helixes. The ideal initial distance between the donor Cy3 fluorophore and the acceptor Cy5 is 4.2 nm (Figure S11c). At a higher ionic strength, the fluorescent intensity of Cy5 (acceptor) was higher, suggesting a higher FRET efficiency and closer distance between Cy3–Cy5 FRET pairs (Figure S11 b,c and Table S2). This implies that 6HB becomes more rigid when the ionic environment changes from 100.0 mM Na+, 12.5 mM Mg2+, and 50 mM Mg2+ to 125 mM Mg2+. The global reconfiguration of 6HB was further validated using SAXS[9] (see detailed descriptions in the Supporting Information, Figures S11-S14).</p><p>The molecular model of 6HB at different expanded states are presented in Figure S11 f. In the presence of 125 mM MgCl2, 6HB showed a most compact structure, which is consistent with MD simulations. With the decrease of ion concentration, or when changing from a divalent to a monovalent cation (Mg2+ to Na+), 6HB expanded, which is also reflected in their 2D SAXS profiles. In comparison to the initial models, some regions in the refined 6HB models swelled and the helices at the ends of the bundle slightly bent outwards, which means a more severe electrostatic repulsion force observed in experiments than in the theoretical simulation. These phenomena are consistent with the observation of large DNA origami objects.[10] Taken together, these data suggest an important role for the electrostatic neutralization, which can stabilize the highly negatively charged DNA nanotube.</p><p>The combination of MD simulations with SAXS and FRET characterization shows that 6HB adopts a compact state when its charge is neutralized with divalent ions (Mg2+) at high concentrations; whereas monovalent ions (Na+) switch it to the expanded state. Counterions can form an ionic atmosphere around DNA, hence neutralizing the negatively charged phosphate and stabilizing the DNA system.[11] Mg2+ has stronger interactions than Na+ with water molecules, so Mg2+ can form a tighter and more stable Mg2+–6H2O complex, which can function as hydrogen bond donors and interact with DNA mainly through hydrogen bonds. In contrast, Na+ ions prefer a direct interaction with phosphate. Moreover, Mg2+–6H2O can bridge two hydrogen bond acceptors at a distance of 10 Å. Moreover, the hydrogen bonds between negatively charged atoms and hydrated Mg2+ ions are much stronger than those of hydrated Na+. Our MD trajectory shows that Mg2+ is mainly distributed in following zones: Hydrated Mg2+ bridging the phosphate outside the minor groove, hydrated Mg2+ binding to O6 and N7 of the CG base in the major groove (Figure 2e and the Supporting Information, Figure S6), which is consistent with previous reports.[11b,12] These two kinds of distribution make DNA more rigid in hydrated Mg2+ systems. Moreover, hydrated Mg2+ can also bridges phosphates of two adjacent helixes in 6HB (Figure 2e). The strong interactions between 6HB and hydrated Mg2+ kept 6HB in the compact state, which make the DNA helix more rigid with Mg2+ than with Na+.</p><p>Although counterions (such as Mg2+ at high concentration) can reduce the electrostatic repulsion of adjacent bundles and hence stabilize 6HB, the repulsion between two negative helical bundles will still deform the 6HB. Removing the negative charges on the DNA backbone would be an alternative to stabilize 6HB. To demonstrate this concept, we replaced 12 phosphate groups of 6HB with ethyl-phosphor-othioate. A 70-ns MD simulation with 500 mM Na+ (NaH_12E6HB) (see Table S5 for the position of ethyl-phosphorothioate 6HB) indicated that ethyl-phosphoro- thioate nucleotides can reduce the expansion of 12E6HB with a finger crossed conformation in the middle of O-ring, which can keep two adjacent helixes tight (Figure 2a and Figure 3a). The inner volume of last 20 ns (of MD) of NaH_12E6HB is 259 nm3 and cross-sectional area is 31.3 nm2, while for NaH system, these two values are 275 nm3 and 33.4 nm2, respectively (Figure 2b and Figure S5). Especially, the "O" ring in unmodified 6HB in the presence of Na+ is converted to an "8" shape (Figure 1 and the Supporting Information, Figure S8). Both of the distances between helixes 1 and 2 and between helixes 4 and 5 decrease by about 1 nm. Moreover, the stretch modulus and persistence length analysis of 6HB (Table S1, Figure 2c, and Figure S9) show that with ethyl-phosphorothioate substitution, 12E6HB was more rigid in the length direction and bending angle than the NaH system.</p><p>Experimentally, we synthesized partially ethylated 6HB (Figure 3b) and diluted it in 450.0 mM Na+ solution for SAXS analysis (Figure 3d). The slight partial reconfiguration resulted in two very similar scattering curves of 6HB with and without ethylation. To describe the changes of 6HB-DNA brought about by the "ethyl ring", the changes of the total volume and section area were calculated (Figure 3e). Both the total volume and section area of 6HB-DNA decrease by introducing the "ethyl ring" because the electrostatic repulsion force is reduced by the uncharged two-base-wide ethyl group. The radius of gyration (Rg) for a hollow cylinder[13] can be calculated from Equation (1): (1)Rg=(ri2+ro22+h212)12 where ri andro are the inner and outer shell radii, respectively, and h is the length of cylinder. Judging from the geometrical parameters of the PDB models, since h did not change much, the Rg decrease of C2H5 sample (Table S3) was caused by the decrease of the radius near the ethyl ring area.</p><p>Our computational/experimental method shows that partial ethylation of the 6HB backbone can switch the 6HB from expanded state to partially compact state, mainly owing to the elimination of the negative charge on the DNA backbone and addition of Van der Waals (vdW) interactions between helixes (Figures 2d and 3a). Finally, we simulated a neutral 6HB, the fully ethyl-phosphorothioated 6HB (Full- E6HB). After a 70-ns MD simulation, FullE6HB holds a more compact conformation than unmodified 6HB in the presence of Mg2+ of high concentration (simulation data are shown in Figure S7 in the Supporting Information). We thus expect that the design of DNA nanostructures with reduced electrostatic repulsive forces might be promising drug delivery tools.</p><p>In summary, we have demonstrated that a biomimetic 6HB can be switched from an expanded state to a (partially) compact state in aqueous solution by charge neutralization through changing ion type and concentration or through chemical modification of the DNA backbone. 6HB is a user-defined model channel based on its programmable structural design and controllable modification features. As DNA is a biocompatible material present in the organisms themselves, DNA-based 6HB nanostructures could have potential biomedical and biological applications. For example, drug molecules can be incorporated into the DNA nanotube and released through shape reconfiguration. The charge neutralization-driven shape reconfiguration of 6HB is highly relevant to functional membrane protein channels. The changes in ion type and ionic strength are very common in living organisms, and are closely related to vital biological functions, such as for voltage-gated potassium channels[14] and calcium-activated potassium channels.[15] Thus, our finding has important implications in uncovering the physiological function of natural ion channels. The mechanism presented herein helps explain previous observations on interhelical spacing of DNA nanostructures.[16] Furthermore, the concept of charge neutralization-driven shape reconfiguration of 6HB, either by ion-atmosphere variation or chemical modification, provides an innovative, simple, and convenient route for controlled release in drug delivery.</p>

PubMed Author Manuscript

Solvation of Biomolecules by the Soft Sticky Dipole-Quadrupole-Octupole Water Model

The soft sticky dipole-quadrupole-octupole (SSDQO) potential energy function represents a water molecule by a single site with a van der Waals sphere and point multipoles. Previously, SSDQO was shown to give good properties for liquid water and solvation of simple ions and is faster than three point models. Here, SSDQO is assessed for solvating biologically relevant molecules having a multi-site, partial charge description. Monte Carlo simulations of ethanol, benzene, and N-methylacetamide in SSDQO with SPC/E moments showed the water structure was as good as in SPC/E. Thus, SSDQO is potentially useful for simulations of biological macromolecules in aqueous solution.

solvation_of_biomolecules_by_the_soft_sticky_dipole-quadrupole-octupole_water_model

INTRODUCTION<!>METHODOLOGY<!>RESULTS AND DISCUSSION<!>CONCLUSIONS<!>

<p>The structure and activity of biological macromolecules such as proteins are highly dependent on being solvated by water. Computer simulations of these systems must include solvent effects and the most accurate way of treating these effects is by using explicit water models. Moreover, simulations with explicit water are also a means of studying the underlying molecular basis of aqueous solvation. However, evaluating the water-water interaction is the most time consuming process in such simulations because of the large number of water molecules needed to solvate the solute. Since most water models use partial charges on fixed interaction sites to describe the electrostatics, a greater number of interaction sites exacerbates the problem because of the increased number of internuclear distances that must be calculated. On the other hand, a greater number of interaction sites generally makes it easier to fit more properties of liquid water. For instance, three-site models such the SPC/E [1] and TIP3P [2] models give a reasonable description of water but have problems with dielectric and dynamical properties, respectively [3]. On the other hand, the five-site TIP5P [4] model has excellent properties for pure water but is computationally expensive.</p><p>Recently, we have developed the soft sticky dipole-quadrupole-octupole (SSDQO) model of water [5], which unlike typical models, has a single-site with a van der Waals sphere and point dipole, quadrupole, and octupole moments. Because the electrostatic interaction potential is the exact moment expansion up to order 1/r4 and contains an approximation for the 1/r5 term, it is both efficient and accurate. When the moments, geometry, and van der Waals parameters of SPC/E, TIP3P, and TIP5P are used, SSDQO reproduces the water dimer potential energy and radial distribution function of the corresponding multipoint model [5]. Moreover, SSDQO using SPC/E moments, geometry, and van der Waals parameters has good thermodynamic, dielectric, and dynamic properties [6] as well as good structural properties around simple ions [7]. Although it also reproduces many of the flaws of SPC/E water, it is actually an improvement in being somewhat more structured than SPC/E. Moreover, SSDQO is computationally faster than even the three-site models because the nine distances required for three-site models are slower than the matrix multiplications and the single intermolecular distance needed in SSDQO for the moment expansion, since the approximation for the 1/r5 require only the lower order matrix multiplications and higher order terms are neglected. Previous studies have shown that SSDQO is about three times faster than three point models such as SPC/E and TIP3P in Monte Carlo simulations [5] and about two times faster in molecular dynamics simulations, where the dipole-dipole interaction was treated with Ewald summation and the higher order interactions by truncation [6]. We are also improving the properties of SSDQO by refining the moments against quantum chemical calculations to reproduce the experimental properties of liquid water.</p><p>However, despite the efficiency and accuracy of the SSDQO water model, it can be useful in improving the speed in typical simulations of biological macromolecules only if it describes the interaction with the solute accurately. This is not a trivial question since most force fields used for such simulations as CHARMM [8] and AMBER [9] describe the electrostatics of proteins, nucleic acids, carbohydrates, lipids, etc. by partial charges while the electrostatics of the water molecules in the SSDQO model are described by higher order multipoles. Thus it is crucial to assess the ability of SSDQO to interact with a multi-site, partial charge description of a solute.</p><p>Here, SSDQO is assessed for solvating biologically relevant molecules described by typical partial charge descriptions. The solutes studied are ethanol, benzene, and N-methylacetamide (NMA), which represent different molecular fragments found in proteins. Ethanol is important since it is amphiphilic, with a hydroxyl group and a hydrocarbon tail, and it mimics the side chain of serine. Benzene tests the ability to solvate a hydrophobic molecule and mimics the side chain of phenylalanine. Finally, NMA is the simplest compound with a peptide bond and thus serves as a test for solvating the polypeptide backbone. Since SSDQO is based on a multipole expansion, which is most accurate at long range, the focus here is on the accuracy of the short-ranged structure where the model is most uncertain. Thus, the radial distribution functions are examined, which are sensitive to short-ranged structure even when long-range correlations are poorly described [10]. Monte Carlo simulations of these solutes in SSDQO using moments and van der Waals parameters from SPC/E and in SPC/E water demonstrate that the solvation by SSDQO is as good as by SPC/E. Moreover, the slight differences indicate that SSDQO actually acts a better hydrogen bond donor than SPC/E. The efficiency and accuracy of the SSDQO model of water indicate that it is potentially very useful for computer simulation of macromolecules in aqueous solution.</p><!><p>Detailed descriptions of the SSDQO water-water and water-ion potentials can be found elsewhere [5,7] so only a brief description is given here. The non-bonded interaction potential is given by</p><p> Uij(r)=4εij{(σijr)12−(σijr)6}+1r[qiqj]+1r2[−qi(μj·n)+(n·μi)qj]+1r3[qi(Θj·n(2))+(n(2)·Θi)qj−3(n·μi)(μj·n)+μi·μj]+1r4[−qi(Ωj∴n(3))+(n(3)∴Ωi)qj+5(n·μi)(Θj:n(2))−5(n(2):Θi)(μj·n)−2μi·Θj·n+2n·Θi·μj]+1r5[qiΦj(μj·n)(oj∴n(3))+(n(3)∴oi)(n·μi)Φiqj−cDO(n·μi)(Ωj∴n(3))−cDO(n(3)∴Ωi)(μj·n)+cQQ(n(2):Θi)(Θj:n(2))] where r = rn is the internuclear vector from particle i to j. Also, the dyadic products are denoted by [n(2)]ij = ninj and [n(3)]ijk = ninjnk, and the matrix contractions are denoted by A·B = Σi AiBi, A:B = Σij AijBij, and A∴B = Σijk AijkBijk. The electrostatic potential is an exact multipole expansion up to order 1/r4 and contains an approximate 1/r5 term. For water-water interactions, the water molecules i and j interact through the dipole μ, quadrupole Θ, and octupole Ω moments of water, with the monopole q=0. For the solute-water interactions, the partial charges qi of the solute molecule interact with the multipole moments of SSDQO water molecule j up to the hexadecapole Φ. In the approximate charge-hexadecapole interaction, m is a unit vector along the direction of μ, and o is a unit vector along the direction of Ω. The factor Φ for a tetrahedral molecule is defined as 7ΓbOH/10√3 = −Hzzzz/2, where H is the hexadecapole moment matrix. For SPC/E, Γ = 5/2Ω so Φ = 7ΩbOH/4√3. This potential allows straightforward combining rules for interaction with other molecules.</p><p>The Monte Carlo simulations used standard Metropolis sampling [11] in the NVT ensemble at 298 K for a cubic box (box length, b = 24.835 Å). In each case, one solute was solvated in box of water created at the experimental density of water (0.033 46 molecules/Å3). The simulations consisted of one ethanol in 504 water molecules, one benzene in 502 water molecules, and one NMA in 498 water molecules. Periodic boundary conditions and spherical switching functions between (b/2 – 1) Å and b/2 Å were applied.</p><p>The starting configurations of the ethanol and the NMA were taken from the Cambridge Structural Database while benzene was created from CHARMM parameters (C-C and C-H bond lengths are 1.375 Å and 1.080 Å respectively; C-C-C and H-C-C angles are 120o). The CHARMM22 potential energy function [12] was used for all solutes. The SSDQO potential energy function with SPC/E Lennard-Jones parameters and moments (μ= 2.3503 D, Θ = 2.0355 × 10−26 esu-cm2, Δ=0, Ω=0.7834 × 10−34 esu-cm3, and Γ = 1.9585 × 10−34 esu-cm3), which is referred to as SSDQO:SPC/E, was used for the water molecules [5]. In these calculations, the solute coordinates were fixed.</p><p>The initial configurations were equilibrated for 400,000 MC "passes" (one pass equals N attempted translational and rotational moves) and the radial distribution functions were calculated from the subsequent 400,000 MC passes. The acceptance ratio in all MC runs was approximately 40%. For comparison, the solutes were also simulated in SPC/E water using the same conditions.</p><!><p>The radial distribution functions around each of the solutes were calculated for both SSDQO and SPC/E. For reference, the pure water radial distribution function is reproduced in Fig. 1. In the discussion below, gXO and gXH refer to the radial distribution function around the solute atom X of the water O and H, respectively.</p><p>The radial distribution functions of SSDQO and SPC/E water around ethanol demonstrate that the solvation of an hydrogen bonding amphiphilic solute by both models is very similar but with some minor structural differences (Fig. 1). The first solvation shell around the ethanol oxygen is at approximately the same distance for both models, with the first peak at ~2.8 Å in gOO for both (Fig. 1a). However, the first peak of SSDQO was higher and wider than in SPC/E and integration of the first peak to 3.475 Å (the minimum for SSDQO) gave coordination numbers of 3.35 and 2.90 for SSDQO and SPC/E respectively, while integration of the first peak to 3.3 Å (the minimum for SPC/E) gave coordination numbers of 2.85 and 2.31 for SSDQO and SPC/E respectively (Table 1). This reflects the somewhat stronger hydrogen bonding capability of SSDQO with SPC/E moments. The more strongly pronounced peak in gOO for SSDQO is consistent with an ab initio molecular dynamics (AIMD) simulation study [13] and the larger coordination number is consistent with neutron diffraction value of less than 3 waters within 3.0 Å [14,15] and the AIMD value of 3 waters within 3.3 Å [13]. The peaks in gOH and gHO had minima at similar positions for SSDQO and SPC/E and the slightly more pronounced peaks are again consistent with the AIMD study [13]. The number of hydrogen bond donating waters in the first shell found by integrating the first peak in gOH to 2.55 Å (the minimum for SSDQO) gave coordination numbers of 1.86 and 1.33 for SSDQO and SPC/E, respectively, while the number of hydrogen bond accepting waters found by integrating the first peak in gHO to 2.6 Å (the minimum for SSDQO) gave coordination numbers of 0.90 and 0.77 for SSDQO and SPC/E, respectively. Thus, both models are reasonable in showing that more waters act as hydrogen bond donors than acceptors [16]. Furthermore, this demonstrates that even though SSDQO does not have an explicit hydrogen, it can act as better hydrogen bond donor than SPC/E. Also, the solvation around the methyl carbon shows that the water is shows no hydrogen bonding to it for both models (Fig. 2b), as expected for a nonpolar atom.</p><p>The radial distribution functions of SSDQO and SPC/E water around benzene show that both models were almost identical in their solvation of this hydrophobic solute (Fig. 3). The distribution functions around the carbons, gCO and gCH, showed that water has relatively little structure around the carbon atoms for both models (Fig. 3a), as expected. In addition, the distribution functions around the center of mass (cm) showed that the packing was similar (Fig. 3b), with small peak in gcmO at ~3.5 Å corresponding to water near the center of the ring and a larger peak at ~5 Å corresponding to water along the periphery of the ring. Moreover, coordination numbers found by integrating the small peak in gCO at ~3.5 Å together with the peak in gcmH at ~2.5 Å indicate that on average, one water molecule is found at face of the benzene near the center of the ring with the hydrogen pointing inward (Table 1) even though there are two equivalent positions, one on each face. To our knowledge, no experimental data was available for comparison.</p><p>The radial distribution functions of SSDQO and SPC/E water around NMA also showed similar structure of both with minor variations (Fig. 4). The first solvation shell around the carbonyl oxygen was very similar (Fig. 4a), with the first peak at 2.8 Å in gOO for both. The peak was slightly wider for SSDQO, so the coordination numbers found by integrating to 3.35 Å (the minimum for SSDQO) were 2.55 and 2.35 for SSDQO and SPC/E, respectively, while coordination numbers found by integrating to 3.25 Å (the minimum for SPC/E) were 2.35 and 2.25 for SSDQO and SPC/E, respectively. The gOO for both were consistent with an AIMD simulation [17,18] and the coordination numbers were consistent with a neutron diffraction value of two hydrogen bond donor waters [19] and the AIMD value of two waters [17,18]. The first peak at 1.8 Å in gOH for both showed that water is acting as a hydrogen bond donor to the NMA oxygen in both. The second peak in gOO for both models was at ~5 Å, corresponding to the expected position for water hydrogen bonded to the trans NH group [20], although it was somewhat more defined for SSDQO. On the other hand, while the solvation around the amide nitrogen was relatively less structured than around the carbonyl oxygen for both SSDQO and SPC/E (Fig. 4b), SSDQO had a slight peak in gNO at ~3 Å with a population of 0.77 whereas there was only a shoulder for SPC/E with a population of 0.5 (Table 1). The stronger peak in SSDQO is consistent with the AIMD study [17,18] and the larger coordination number is consistent with the neutron diffraction value of one hydrogen bond acceptor water [19] and the AIMD value of one water [17,18]. The peak in gHO at ~2 Å shows that the water is acting as a hydrogen bond acceptor. Also, the density in gNH does not become significant until ~3.5 Å, which indicates that the water in the first shell is in dipolar orientation with respect to the amide nitrogen. Finally, the solvation around the methyl carbons is very similar to the methyl carbon of ethanol for both models and so are not shown.</p><!><p>Here, the SSDQO model using moments and van der Waals parameters from SPC/E was shown to give reasonable solvation of molecules with varying polarities in which the electrostatics were described by partial charges, which were in good agreement with solvation by SPC/E. There was a tendency for stronger hydrogen bonds in SSDQO than SPC/E, which was also shown for the interactions in pure liquid water [5], and we are currently optimizing the moments and van der Waals parameters for SSDQO to reproduce thermodynamic, dielectric, dynamical, and solvation properties of water under different conditions. However, overall these studies demonstrate that the single point, multipole moment interaction potential of SSDQO can be used with the multiple point, partial charge interaction potential commonly used in force fields used to describe the aqueous solvation of biological macromolecules. The radial distribution functions, which are sensitive to short-range interactions [10], demonstrates that SSDQO can reproduce the local solvation structure around a solute, including solutes that have hydrogen-bonding interactions with water. This is an important test because while the multipole approximation is expected to good for long-range interactions, the ability to reproduce short-range interactions needed to be demonstrated. Furthermore, the computational efficiency of SSDQO makes it potentially valuable for computational simulations of large macromolecules in aqueous solution, where the number of water molecules needed is substantial.</p><!><p>Radial distribution for liquid water: oxygen – oxygen, oxygen – hydrogen (shifted upward by 1), and hydrogen – hydrogen (shifted upward by 2) for SSDQO (black) and SPC/E (gray) water.</p><p>Radial distribution functions for ethanol in SSDQO (black) and SPC/E (gray) water: a) ethanol oxygen - water oxygen, ethanol oxygen - water hydrogen (shifted upward by 1), and ethanol hydrogen – water oxygen (shifted upwards by 2) and b) ethanol methyl carbon – water oxygen and ethanol methyl carbon – water hydrogen (shifted by 2).</p><p>Radial distribution functions for benzene in SSDQO (black) and SPC/E (gray) water: a) benzene carbon - water oxygen and benzene carbon - hydrogen (shifted upward by 2) and b) benzene center of mass - water oxygen and benzene center of mass - hydrogen (shifted upward by 2).</p><p>Radial distribution functions for N-methylacetamide in SSDQO (black) and SPC/E (gray) water: a) NMA oxygen - water oxygen and NMA oxygen - hydrogen (shifted upward by 2) and b) NMA nitrogen - water oxygen, NMA nitrogen - hydrogen (shifted upward by 1), and NMA hydrogen - oxygen (shifted upward by 2).</p><p>Comparison of the number of water molecules surrounding various atoms of the solutes, integrated to the minimum in the SSDQO (SPC/E) radial distribution function.</p>

PubMed Author Manuscript

Novel Carvedilol Analogs that Suppress Store Overload Induced Ca2+ Release

Carvedilol is a uniquely effective drug for the treatment of cardiac arrhythmias in patients with heart failure. This activity is in part due to its ability to inhibit store overload-induced calcium release (SOICR) through the RyR2 channel. We describe the synthesis, characterization and bioassay of ca. 100 compounds based on the carvedilol motif in order to identify features that correlate with and optimize SOICR inhibition. A single cell bioassay was employed based on the RyR2-R4496C mutant HEK-293 cell line, in which calcium release from the endoplasmic reticulum through the defective channel was measured. IC50 values for SOICR inhibition were thus obtained. The compounds investigated contained modifications to the three principal subunits of carvedilol, including the carbazole and catechol moieties, as well as the linker chain containing the \xce\xb2-amino alcohol functionality. The SAR results indicate that significant alterations are tolerated in each of the three subunits.

novel_carvedilol_analogs_that_suppress_store_overload_induced_ca2+_release

Introduction<!>Results<!>Chemistry<!>SOICR Inhibition<!>Discussion<!>Conclusions<!>Experimental Section<!>Preparation of 1-(9H-carbazol-4-yloxy)-3-{[2-(4,5-dichloro-2-methoxyphenoxy)ethyl]amino}-2-propanol (6)43<!>1-(9H-Carbazol-4-yloxy)-3-{[2-(2,4,5-trichlorophenoxy)ethyl]amino}-2-propanol (7)43<!>1-(9H-Carbazol-4-yloxy)-3-[2-(phenoxyethyl)amino]-2-propanol (8)44,45<!>1-(9H-Carbazol-4-yloxy)-3-{[2-(2-methylphenoxy)ethyl]amino}-2-propanol (10)44,45<!>1-(9H-Carbazol-4-yloxy)-3-{[2-(2-methylthiophenoxy)ethyl]amino}-2-propanol 1145<!>1-(9H-Carbazol-4-yloxy)-3-{[2-[(2-methoxyphenylthio)ethyl]amino}-2-propanol (12)45<!>1-(9H-Carbazol-4-yloxy)-3-{[3-(2-methoxyphenyl)propyl]amino}-2-propanol (13)45<!>1-(9H-Carbazol-4-yloxy)-3-[3-(phenylpropyl)amino]-2-propanol (14)45<!>1-(9H-Carbazol-4-yloxy)-3-{2-(2-pyridyloxy)ethyl]amino}-2-propanol (15)<!>1-(9H-Carbazol-4-yloxy)-3-{2-[(cyclohexyloxy)ethyl]amino}-2-propanol (16)43<!>Preparation of 6-[(9H-carbazol-4-yloxy)methyl]-4-[2-(2-methoxyphenoxy)ethyl]-3-morpholinone (17)43<!>6-[(9H-Carbazol-4-yloxy)methyl]-4-[2-(4,5-dichloro-2-methoxyphenoxy)ethyl]-3-morpholinone (18)43<!>6-[(9H-Carbazol-4-yloxy)methyl]-4-[(2,4,5-trichlorophenoxy)ethyl]-3-morpholinone (19)43<!>6-[(9H-Carbazol-4-yloxy)methyl]-4-[2-(cyclohexyloxy)ethyl]-3-morpholinone (20)43<!>Preparation of 4-{4-[2-(2-methoxyphenoxy)ethyl]-2-morpholinyl]methoxy}-9H-carbazole (21)43<!>4-{[4-[2-(4,5-Dichloro-2-methoxyphenoxy)ethyl]-2-morpholinyl]methoxy}-9H-carbazole (22)43<!>4-{[4-[2-(2,4,5-Trichlorophenoxy)ethyl]-2-morpholinyl]methoxy}-9H-carbazole (23)43<!>4-{[4-[2-(Cyclohexyloxy)ethyl]-2-morpholinyl]methoxy}-9H-carbazole (24)43<!>Preparation of 5-[(9H-carbazol-4-yloxy)methyl]-3-[2-(2-methoxyphenoxy)ethyl]-2-oxazolidinone (25)43,46,47<!>5-[(9H-Carbazol-4-yloxy)methyl]-3-[2-(4,5-dichloro-2-methoxyphenoxy)ethyl]-2-oxazolidinone (26)43<!>5-[(9H-Carbazol-4-yloxy)methyl]-3-[2-(cyclohexyloxy)ethyl]-2-oxazolidinone (27)43<!>Preparation of 5-[(9H-carbazol-4-yloxy)methyl]-3-[2-(2-hydroxyphenoxy)ethyl]-2-oxazolidinone (28)<!>Preparation of 1-(9H-carbazol-4-yloxy)-3-{[2-(2-hydroxyphenoxy)ethyl]amino}-2-propanol (9)44,45<!>Preparation of 1-(9H-carbazol-4-yloxy)-3-{[2-(2-methoxyphenoxy)ethyl]methylamino}-2-propanol (29)45<!>Preparation of 1-{[2-(2-methoxyphenoxy)ethyl]amino}-3-[(9-methyl-9H-carbazol-4-yl)oxy]-2-propanol (30)43,48<!>Preparation of 2-methoxy-N-[2-(2-methoxyphenoxy)ethyl]-N-methyl-3-[(9-methyl-9H-carbazol-4-yl)oxy]-1-propanamine (31)43<!>Preparation of 6-[2-(9H-carbazol-4-yloxy)ethyl]-4-[2-(2-methoxyphenoxy)ethyl]-3-morpholinone (32)<!>Preparation of 6-(9H-carbazol-4-yloxy)-1-{[2-(2-methoxyphenoxy)ethyl]amino}-2-hexanol (33)<!>Preparation of 1-(9H-carbazol-4-yloxy)-4-[[2-(2-methoxyphenoxy)ethyl]amino]-2-butanol (34)<!>Preparation of 7-[(9H-carbazol-4-yloxy)methyl]-4-[2-(2-methoxyphenoxy)ethyl]-1,4-oxazepan-3-one (35)<!>Preparation of 1-(9H-carbazol-4-yloxy)-3-{[3-(2-methoxyphenyl)propyl]amino}-2-propanol (36)43<!>Preparation of 1-(9H-carbazol-4-yloxy)-3-{[2-(2-methoxybenzyloxy)ethyl]amino}-2-propanol (37)<!>Preparation of 1-(9H-carbazol-4-yloxy)-3-{[2-(benzyloxy)ethyl]amino}-2-propanol (38)<!>Preparation of 1-{-2-[(9H-carbazol-4-yloxy)ethyl]amino}-3-(2-methoxyphenyloxy)-2-propanol (39)<!>Preparation of 4-[2-(9H-carbazol-4-yloxy)ethyl]-6-(2-methoxyphenoxymethyl)morpholine-3-one (40)<!>Preparation of 6-[(9H-carbazol-3-yloxy)methyl]-4-[2-(2-methoxyphenoxy)ethyl]morpholine-3-one (41)<!>Preparation of 6-[(9H-carbazol-3-yloxy)methyl]-4-[2-(2-methoxyphenoxy)ethyl]-1,3\xce\xbb6,4-oxathiazinane-3,3-dione (42)<!>Preparation of 4-(9H-carbazol-3-yloxy)-1-{[2-(2-methoxyphenoxy)ethyl]amino}-2-butanol (43)<!>Preparation of 5-(9H-carbazol-3-yloxy)-1-{[2-(2-methoxyphenoxy)ethyl]amino}-2-pentanol (44)<!>Preparation of 6-(9H-carbazol-3-yloxy)-1-{[2-(2-methoxyphenoxy)ethyl]amino}-2-hexanol 45<!>Preparation of 1-(9H-carbazol-3-yloxy)-4-{[2-(2-methoxyphenoxy)ethyl]amino}-2-butanol (46)<!>Preparation of 1-(6-fluoro-9H-carbazol-3-yloxy)-3-{[2-(2-methoxyphenoxy)ethyl]amino}-2-propanol (47)<!>Preparation of 1-[(9H-carbazol-3-yl)amino]-3-{[2-(2-methoxyphenoxy)ethyl]amino}-2-propanol (48)<!>6-[(9H-Carbazol-2-yloxy)methyl]-4-[2-(2-methoxyphenoxy)ethyl]morpholine-3-one 49<!>4-(9H-Carbazol-2-yloxy)-1-{[2-(2-methoxyphenoxy)ethyl]amino}-2-butanol (50)<!>6-[2-(9H-Carbazol-2-yloxy)ethyl]-4-[2-(2-methoxyphenoxy)ethyl]-3-morpholinone (51)<!>5-[2-(9H-Carbazol-2-yloxy)ethyl]-3-[2-(2-methoxyphenoxy)ethyl]-2-oxazolidinone (52)<!>1-(6-Fluoro-9H-carbazol-2-yloxy)-3-{[2-(2-methoxyphenoxy)ethyl]amino}-2-propanol (53)<!>1-(6,8-Difluoro-9H-carbazol-2-yloxy)-3-{[2-(2-methoxyphenoxy)ethyl]amino}-2-propanol (54)<!>1-(9H-Carbazol-2-yloxy)-3-{[2-(4-fluoro-2-methoxyphenoxy)ethyl]amino}-2-propanol (55)<!>1-(9H-carbazol-2-yloxy)-3-{[2-(2-trifluoromethylphenoxy)ethyl]amino}-2-propanol (56)<!>1-(9H-carbazol-2-yloxy)-3-{[2-(2-fluorophenoxy)ethyl]amino}-2-propanol (57)<!>1-(9H-carbazol-1-yloxy)-3-{[2-(2-methoxyphenoxy)ethyl]amino}-2-propanol (58)50<!>Preparation of 1-[4-(2-hydroxy-3-{[2-(2-methoxyphenoxy)ethyl]amino}propoxy-9H-carbazol-9-yl]octadecan-1-one (59)<!>4-[2-(2-Methoxyphenoxy)ethyl]-6-{[(9-octadecanoyl-9H-carbazol-4-yl)oxy]methy}morpholin-3-one (60)<!>1-{[2-(2-Methoxyphenoxy)ethyl]amino}-3-(2,3,4,9-tetrahydro-1H-carbazol-6-yloxy)- 2-propanol (61)<!>4-(2-Hydroxy-3-{[2-(2-methoxyphenoxy)ethyl]amino}propoxy)-9H-fluoren-9-one (63)<!>1-{[2-(2-Methoxyphenoxy)ethyl]amino}-1-{8-oxatricyclo[7.4.0.02,7]trideca-1(9),2(7),3,5,10,12-hexaen-4-yloxy}-2-propanol (64)<!>1--{[2-(2-Methoxyphenoxy)ethyl]amino}-1-(10H-phenothiazin-2-yloxy)-2-propanol (66)<!>2-(2-Hydroxy-3-{[2-(2-methoxyphenoxy)ethyl]amino}propoxy)-10H-5\xce\xbb6,10-phenothiazine-5,5-dione (67)<!>1--{[2-(2-Methoxyphenoxy)ethyl]amino}-3-(naphthalen-1-yloxy)-2-propanol (68)<!>1--{[2-(2-Methoxyphenoxy)ethyl]amino}-4-(naphthalen-1-yloxy)-2-butanol (69)<!>1-(Adamantan-1-yloxy)-3-{[2-(2-methoxyphenoxy)ethyl]amino}-2-propanol (70)<!>1-(Adamantan-2-yloxy)-3-{[2-(2-methoxyphenoxy)ethyl]amino}-2-propanol (71)<!>1-(Adamantan-1-ylmethoxy)-3-{[2-(2-methoxyphenoxy)ethyl]amino}-2-propanol (72)<!>6-(2-Hydroxy-3-{[2-(2-methoxyphenoxy)ethyl]amino}propoxy)-1,2-dihydroquinolin-2-one (73)<!>6-(2-Hydroxy-3-{[2-(2-methoxyphenoxy)ethyl]amino}propoxy)-1,2,3,4-tetrahydroquinolin-2-one (74)<!>7-(2-Hydroxy-3-{[2-(2-methoxyphenoxy)ethyl]amino}propoxy)-1,2,3,4-tetrahydroquinolin-2-one (75)<!>1-(1H-Indol-4-yloxy)-3-{[2-(2-methoxyphenoxy)ethyl]amino}-2-propanol (76)51<!>1-(1H-Indol-5-yloxy)-3-{[2-(2-methoxyphenoxy)ethyl]amino}-2-propanol (77)<!>Preparation of 1-(1H-indazol-5-yloxy)-3-{[2-(2-methoxyphenoxy)ethyl]amino}-2-propanol (78)<!>1-(1H-indazol-6-yloxy)-3-{[2-(2-methoxyphenoxy)ethyl]amino}-2-propanol (79)<!>1-(1H-Benzodiazol-5-yloxy)-3-{[2-(2-methoxyphenoxy)ethyl]amino}-2-propanol (80)<!>Preparation of 3-(9H-Fluoren-4-yloxy)-1-{[2-(2-methoxyphenoxy)ethyl]amino}-2-propanol (62)<!>Preparation of 3-(2-Hydroxy-3-{[2-(2-methoxyphenoxy)ethyl]amino}propoxy-N-phenylaniline (65)<!>Preparation of 1-[(benzofuran-2-ylmethyl)amino]-3-(9H-carbazol-2-yloxy)-2-propanol (81)<!>Preparation of 1-[(benzoxazol-2-ylmethyl)amino]-3-(9H-carbazol-2-yloxy)-2-propanol (82)<!>1-(9H-Carbazol-4-yloxy)-3-(3,4-dihydro-2H-1,4-benzoxazin-4-yl)-2-propanol (83)<!>Preparation of 4-[3-(9H-carbazol-4-yloxy)-2-hydroxypropyl]-3,4-dihydro-2H-1,4-benzoxazin-3-one (84)<!>4-(9H-Carbazol-4-yloxy)-1-(3,4-dihydro-2H-1,4-benzoxazin-4-yl)-2-butanol (85)<!>1-(9H-Carbazol-3-yloxy)-3-(3,4-dihydro-2H-1,4-benzoxazin-4-yl)-2-propanol (86)<!>1-(9H-Carbazol-2-yloxy)-3-(3,4-dihydro-2H-1,4-benzoxazin-4-yl)-2-propanol (87)<!>1-(3,4-Dihydro-2H-1,4-benzoxazin-4-yl)-3-(6-fluoro-9H-carbazol-2-yloxy)-2-propanol (88)<!>1-(9H-Carbazol-4-yloxy)-3-[4-(2-methoxyphenyl)piperazin-1-yl]-2-propanol (89)<!>1-(9H-Carbazol-4-yloxy)-3-(4-phenylpiperazin-1-yl]-2-propanol (90)<!>N-[3-(9H-Carbazol-4-yloxy)-2-hydroxypropyl]-2-(2-methoxyphenoxy)aniline (92)<!>Preparation of 2-(9H-carbazol-4-yloxy)-N-[2\xe2\x80\x932-(methoxyphenoxy)ethyl]acetamide (93)<!>Preparation of 2-(9H-carbazol-4-yloxy)-N-[1-hydroxy-3-(2-methoxyphenoxy)propan-2-yl]acetamide (94)<!>Preparation of 4-{[4-(2-methoxyphenoxymethyl)-4,5-dihydro-1,3-oxazol-2-yl]methoxy}-9H-carbazole (91)<!>Preparation of 1-(4,6-dibromo-9H-carbazol-3-yloxy)-3-{[2-(2-methoxyphenoxy)ethyl]amino}-2-propanol (97)<!>Preparation of 6-[(-(4,6-dibromo-9H-carbazol-3-yloxy)methyl]-4-[2-(2-methoxyphenoxy)ethyl]morpholin-3-one (98)<!>Preparation of 7-methoxy-5-methyl-2,3,4,5-tetrahydro-1,5-benzothiazepine (101)<!>Bioassay: Single-cell Ca2+ imaging of HEK293 cells

<p>Ventricular arrhythmias are a leading cause of sudden death, particularly in patients with heart failure. Consequently, a variety of antiarrhythmic drug therapies have been evaluated in clinical trials, which revealed only limited survival benefits.1–3 Antagonists of β-adrenergic receptors (β-blockers) have been of special interest in these studies, as overstimulation of these receptors can trigger fatal ventricular arrhythmias.4–6 The underlying mechanism of this process involves, in part, an overload of Ca2+ in the sarcoplasmic reticulum, which results in spontaneous Ca2+ efflux through the RyR2 Ca2+ release channel.7,8 In turn, this store overload-induced calcium release (SOICR) through a defective RyR27–14 triggers delayed afterdepolarizations (DADs),15–21 which have been implicated in catecholaminergic polymorphic ventricular tachycardias (CPVTs), as well as in ventricular tachyarrhythmias and sudden death.4,5,22,23</p><p>The nonselective β-blocker carvedilol (1) and certain congeners also inhibit the α-adrenergic receptor24 and are reported to display antioxidant activity.25,26 Thus, 1 has proven uniquely effective in suppressing ventricular arrhythmias in patients with failing hearts.27–30 Unfortunately, the benefits of carvedilol therapy are limited by drug intolerance and excessive β-blockade, with attendant complications of bradycardia and hypotension.2,31 More recently, we demonstrated that a variety of other α- and β-blockers, as well as antioxidants, failed in the suppression of SOICR.32 This suggests that the unique efficacy of carvedilol in suppressing SOICR occurs independently of its α- and β-blocking activity and its antioxidant properties, and is instead principally due to its ability to stabilize Ca2+ handling via the RyR2 channel. Indeed, we recently reported three novel carvedilol analogs 2–4 with comparable abilities to inhibit SOICR to that of the parent compound 1 (ca. 10 μmolar), but with strongly attenuated β-blockade (ca. μmolar compared to nanomolar for 1). Compounds 2–4 proved highly effective in preventing stress-induced ventricular arrhythmias in mice (vide infra), without the undesired effects of excessive β-blockade.32</p><p>For the purpose of these studies, a convenient single cell bioassay was developed for measuring SOICR suppression by drug candidates, based on human embryonic kidney (HEK 293) cells expressing a mutant RyR2 channel (R4496C).7,8 This mutation results in spontaneous calcium release from the endoplasmic reticulum of the cells through the defective channel, with calcium efflux detected by the measurement of fluorescence from the Ca2+-sensitive indicator dye fura 2/AM (Invitrogen).</p><p>Moreover, the same R4496C mutation in mice renders them highly susceptible to stress-induced ventricular arrhythmias, which are easily triggered by stimulants such as caffeine and epinephrine. This provides a useful animal model for evaluating potential antiarrhythmic drugs.9–14</p><p>Our initial efforts that led to the discovery of 2–4 were guided by the x-ray crystal structure of the complex of the carvedilol analog carazolol (5) with the β-adrenergic receptor that had been previously reported by Stevens, Kobilka and their coworkers.33 Their results indicated that the carbazole amino group participates in hydrogen-bonding with residue S203 of the receptor, while multiple hydrogen bonds occur between the secondary alcohol and the protonated amino group of the β-amino alcohol moiety with residues D113, Y316 and N312. Furthermore, a series of hydrophobic interactions were observed involving the aromatic rings of the carbazole and the N-isopropyl group of 5 with corresponding hydrophobic pockets in the receptor. Blocking, modifying or relocating the key hydrogen bonding functionalities and hydrophobic regions of carvedilol might therefore be expected to disrupt its binding to the β-adrenergic receptor. Similarly, a catechol moiety is a common structural motif in many α-blockers,24 suggesting that manipulation of this group in 1 could result in diminished α-blockade. However, such structural changes might also interfere with binding to RyR2 and SOICR suppression. We now report the synthesis of a wide range of novel structures, in addition to the carvedilol analogues 2–4, as part of a SAR investigation to determine how such structural modifications would affect the abilities of these compounds to decrease SOICR in the HEK 293 (R4496C) cell line.</p><!><p>The SOICR-suppressing ability of a series of carvedilol analogs 2–94, 97 and 98, as well as of reference compounds 1, 95, 96 and 99–102 that were included for comparison, is shown in Tables 1–8. All compounds were tested for SOICR inhibition in the RyR2-R4496C mutant HEK293 cell line and the IC50 values, along with the number of replicates and the total number of cells employed in the assay, are provided in the Tables.</p><!><p>The synthesis of compounds 1 and 6–8 and 10–16 in Table 1 was achieved by reacting the commercially available epoxide 103a with the corresponding amines 104a-l, as shown in Scheme 1. Cyclization of amino alcohols 1, 6, 7 and 16 with chloroacetyl chloride afforded lactams 17–20, while the use of 1,1′-carbonyldiimidazole afforded products 25–27 in Table 2. The reduction of 17–20 with lithium aluminum hydride provided 21–24. The free phenol 28 was obtained by demethylation of 25 with boron tribromide, while hydrolysis of 28 afforded the amino alcohol 9.</p><p>The methylated product 29 in Table 2 was prepared by methylation of 1 with iodomethane, while 30 was obtained from 1 by cyclization and N-methylation with dimethyl carbonate in one step, followed by hydrolysis (Scheme 2). Compound 31 was produced from the reaction of epoxide 103b with the N-methyl derivative of amine 104a, followed by O-methylation with iodomethane.</p><p>The products 2 and 33 in Table 3 were prepared from the reactions of amine 104a with the homologated epoxides 105 and 106, in turn obtained from 4-hydroxycarbazole and the corresponding homologated haloepoxides. Alternatively, 34 was produced from 4-hydroxycarbazole and amine 104a via the monotosylate 109. Compounds 36–38 were obtained from epoxide 103a and amines 110 and 111a,b, respectively, while the product 39, containing transposed alcohol and amine functionalities, was prepared from the carbazole derivative 112 and epoxide 113. These processes are summarized in Scheme 3. Cyclizations to afford lactams 32, 35 and 40 were effected from the corresponding amino alcohols 2, 34 and 39, respectively, with chloroacetyl chloride, as shown previously for 17–20 in Scheme 1.</p><p>While the products in Tables 1–3 originated from 4-hydroxycarbazole, the syntheses of those in Table 4 were carried out similarly, but employing 3-hydroxycarbazole, its 6-fluoro derivative or 3-aminocarbazole instead, as shown in Scheme 4. Compound 46 was obtained by alkylation of amine 104a with bromide 115, which was obtained as a byproduct in the preparation of epoxide 114b. Similarily, compound 48 was obtained by alkylation of 104a with chloride 116, in turn obtained from 9-t-Boc-protected 3-aminocarbazole and epichlorohydrin. Furthermore, the products in Table 5 were obtained by analogous methods, employing 2-hydroxycarbazole, its 6-fluoro or 6,8-difluoro analogs, or 1-hydroxycarbazole as starting materials (Scheme 5). Lactams 41, 49 and 51 were again obtained by cyclization of the corresponding amino alcohols 3, 4 and 50, respectively, with chloroacetyl chloride, while cyclizations to afford 42 and 52 from 3 and 50, respectively, were effected with bromomethanesulfonyl chloride or 1,1′-carbonyldiimidazole.</p><p>The N-acylated carbazole derivatives 59 and 60 in Table 6 were prepared as shown in Scheme 6. While 60 was prepared by direct acylation of 17, compound 59 was more easily obtained by prior acylation of epoxide 103a, followed by ring-opening with amine 104a in the usual manner. The other compounds in Table 6 were prepared by alkylating the other indicated phenols or alcohols instead of 4-hydroxycarbazole with the corresponding haloalkyl epoxides, followed by treatment with amine 104a, as indicated in Scheme 6. Product 62 was obtained by Wolff-Kishner reduction of ketone 63, while desulfurization of 66 with nickel boride34 afforded 65.</p><p>The products 83–90 and 92 in Table 7 were obtained by treating epoxides 103a (entries 3, 4, 9, 10 and 12), 105 (entry 5), 114a (entry 6), 117a (entry 7) or 117c (entry 8) with the corresponding amines 104a or 118a-f, as shown in Scheme 7. The benzofuran and benzoxazole derivatives 81 and 82 (entries 1 and 2) were produced by reductive amination or alkylation of amine 119 with aldehyde 121 or chloride 122, respectively, followed by debenzylation of the intermediate benzylamines. Amides 93 and 94 (entries 13 and 14) were formed by amidation of carboxylic acid 120 with amines 104a or 118f, respectively, using N-(3-diethylaminopropyl)-N′-ethylcarbodiimide hydrochloride (EDC hydrochloride) and 1-hydroxybenzotriazole hydrate (HOBT·H2O) as coupling reagents.35 Product 91 (entry 11) was prepared by cyclization of 94 with diethylaminosulfur trifluoride (DAST).36</p><p>Metoprolol (95) was obtained from commercial sources, while 4,6-dibromo-3-hydroxycarbazole (96) was prepared by a literature procedure.37 The conversion of 96 to 97 and its further transformation to 98 was achieved via the same procedure that was employed for the preparation of 1 from 4-hydroxycarbazole, and for the cyclization of 1 to 17, as shown in Scheme 1. Products 99 and its hydrochloride salt 100,38 as well as 10239 were obtained by known procedures. Alternatively, the regioisomer 101 of 99 was prepared, along with 101, from the known thiapyrone 123,40 as shown in Scheme 8.</p><!><p>The benchmark compound 1 displayed an IC50 of 15.9 μM (Table 1, entry 1) in this assay. We first prepared and assayed several derivatives (6–16) with modified catechol groups, as shown in Table 1. Aromatic hydroxylation of the catechol moiety of carvedilol at the 4′- and 5′-positions is known to afford phenolic metabolites.41 Modification of these sites to probe their effect on SOICR inhibition was therefore of particular interest (entries 2–4). The introduction of 4′- and 5′-chloro substituents in 6 slightly improved SOICR activity, but replacement of the methoxy group with a third chlorine atom in 7 diminished the potency significantly. Surprisingly, the simple phenyl derivative 8 proved comparable to 1, while replacement of the methoxy substituent of 1 with a hydroxyl group in derivative 9 lowered the IC50 by half. The similar replacement by methyl or methylthio groups (10 and 11, respectively) had no effect on activity. When the other catechol oxygen atom was replaced with sulfur or a methylene group (12 and 13, respectively), or when the methoxy group was also absent, as in 14, slightly higher activity compared to 1 was observed. On the other hand, replacement of the catechol moiety by a 2-pyridyl group (15) or by a cyclohexyl substituent (16) resulted in a three-fold and two-fold loss of activity, respectively.</p><p>We also investigated whether SOICR inhibition would be affected by manipulation of the β-amino alcohol moiety in the linker chain, which plays a critical role in the excessive β-blockade encountered with carvedilol. Entries 1–12 in Table 2 show the IC50 values for analogs where this functionality was incorporated into various rings, with or without the previous modifications to the catechol region of the carvedilol molecule given in Table 1. In the case of δ-lactams 17–20, there was a moderate loss of activity compared to 1 when either the intact catechol moiety was retained (17), or when it was replaced by a simple cyclohexyl group (20). Moreover, the chlorinated derivatives 18 and 19 were essentially devoid of activity when introduced together with lactamization (cf. the relatively active chlorinated compounds 6 and 7 in Table 1). The morpholine analogs 21–24 revealed no consistent pattern of behavior when compared with the corresponding free β-amino alcohols 1, 6, 7 and 16, and with the corresponding lactams 17–20, respectively. While a slight decrease in activity was observed in 17 and 21 compared to 1, the chlorinated analogs 22 and 23 were intermediate between the corresponding amino alcohols 6 and 7 and the inactive lactams 18 and 19. The cyclohexyl derivative 24 was devoid of activity, in contrast to the weakly active amino alcohol 16. The cyclic carbamates 25–27 all showed poor SOICR inhibition compared to the corresponding amino alcohols 1, 6 and 16, while the free phenol analog 28 was comparable to carvedilol, but less active than the phenol 9 in this regard. Furthermore, alkylation of the aliphatic secondary amino group in 29 had little effect, while alkylation of the carbazole nitrogen in 30 resulted in diminished activity. Surprisingly, exhaustive alkylation of both nitrogens and of the secondary alcohol group in 31 produced a more strongly SOICR-inhibiting compound than either 1, 29 or 30.</p><p>Attempts to determine the effects of homologation of the linker chain at various sites upon SOICR inhibition were also made (Table 3, entries 1–8). Homologation was effected by insertion of one or more extra methylene units between the carbazole ether and secondary alcohol of 1 to afford 2, its cyclized derivative 32, and 33, respectively. Similarly, homologation between the alcohol and amino group provided 34 and the corresponding lactam 35, while insertion of an extra methylene unit between the amino group and the catechol ether and between the catechol ether and aromatic ring afforded derivatives 36 and 37, respectively. Products 2, 32, 33, 34, 37 and 38 proved comparable to 1, while 36 was slightly less potent. When homologation was combined with lactamization of the linker chain of 34 to afford 35, activity diminished more than 6-fold. Removal of the methoxy group from homologue 37 to give 38 had essentially no effect on the activity. Transposition of the alcohol and secondary amino groups in the linker chain of 1 and 17 was also investigated (entries 9 and 10). Thus, 39 and its lactam analog 40 were prepared and found to have IC50s ca. 1.7 times those of 1.</p><p>When the point of attachment of the linker was moved from the 4-position of the carbazole moiety (as in carvedilol) to the 3-position, a series of highly active compounds was obtained (Table 4, entries 1–7). Thus, compounds 3, 41, 43 and 44 all proved superior to carvedilol in their inhibition of SOICR. Cyclization of 3 to afford lactam 41 retained the high activity of the parent amino alcohol and only the sultam moiety of derivative 42 strongly impeded SOICR inhibition. Furthermore, homologation of 3 by one, two or three methylene units between the carbazole and hydroxyl functions produced the highly active analogs 43–45, respectively, with the lowest IC50 of 4.66 μmol observed for the doubly homologated derivative 44. Single homologation between the alcohol and amino functions in 46 also produced a more potent product than carvedilol, while the 6-fluoro derivative 47 showed essentially identical activity to that of the parent compound 3 of this series. Replacement of the ether oxygen with an amino group at C-3 of the carbazole moiety in 48 resulted in comparable activity to that of carvedilol.</p><p>The 2-substituted carbazole series (Table 5, entries 1–10) was also investigated and revealed several highly active SOICR inhibitors. The parent compound 4, as well as its lactam and homologated counterparts 49 and 50, respectively, proved similar to carvedilol. In contrast, the homologated lactam 51 and cyclic carbamate 52 were ineffective in SOICR inhibition. Monofluorination of the 6- position or 6,8-difluorination of the carbazole had a beneficial effect on activity, leading to the highly potent analogs 53 and 54. Fluorination of the catechol moiety in compound 55, or fluoro or trifluoromethyl substitution of the methoxy group in 57 and 56, respectively, had little effect compared to 1 or 4. Finally, attachment of the linker chain to the 1-position of the carbazole unit in 58 (entry 11) also had little effect upon activity.</p><p>It has been suggested that the carbazole moiety of carvedilol serves to embed the molecule in the lipid bilayer of cell membranes.42 Furthermore, it provides a hydrophobic region and a hydrogen-bonding functionality that are key for binding to the β-adrenergic receptor.33 In order to determine whether these structural features are also required for SOICR inhibition, we prepared a variety of analogs containing modified carbazole units (Table 6). In entries 1 and 2, amidation of the carbazole nitrogens of 1 and 17 with octadecanoic acid afforded 59 and the lactam derivative 60, respectively. Unlike the N-methylated derivatives 30 and 31 in Table 2, which showed comparable activity to carvedilol (1), amides 59 and 60 unfortunately revealed negligible or significantly diminished activity relative to 1 and 17.</p><p>Other carbazole modifications are shown in entries 3–22 of Table 6. The tetrahydrocarbazole 61, the fluorene and fluorenone analogs 62 and 63, respectively, that lack the carbazole nitrogen atom, as well as the diphenylamine 65, all showed comparable or only slightly lower activity than carvedilol. On the other hand, the dibenzofuran 64 and the phenothiazine 66 were more strongly active than 1, while oxidation of 66 to the corresponding sulfone 67 decreased activity by more than 10-fold. Replacement of the carbazole with naphthyl residues in 68 and its homologated analog 69 afforded potent SOICR inhibitors. The adamantyl residues in 70–72 resulted in less efficacious compounds and the quinolinone 73 and partially reduced quinolinones 74 and 75 were essentially inactive. The installation of indole (compounds 76 and 77), benzodiazole (compounds 78 and 79) and benzimidazole (compound 80) residues in place of the carbazole also produced very weak or essentially inactive products.</p><p>We also introduced additional heterocyclic groups into the linker chain or in place of the catechol moiety (Table 7, entries 1–12), as well as amide instead of amino alcohol functionalities (entries 13–14). The benzofuran and benzoxazoles 81 and 82 were devoid of activity, but the benzomorpholine 83 proved more than twice as potent as carvedilol (1). The lactam derivative 84 and homologue 85 were comparable to 1. Repositioning the linker chain of 85 to the 3- and 2-position of the carbazole moiety resulted in strong and negligible activities in 86 and 87, respectively. Fluorination of the 6-position of 87 to afford 88 failed to improve the bioactivity. Piperazines 89 and 90, as well as dihydrooxazole 91 and diaryl ether 92 displayed weaker activity than 1. Surprisingly, replacement of the amino alcohol moiety of 1 with amide linkages in 93 and 94 afforded the most potent SOICR inhibitors of this investigation, with IC50 values of ca. 3.6 μM compared with 15.9 μM for carvedilol.</p><p>Several other classes of compounds have been reported to provide salutary effects in the treatment of cardiac arrhythmias. We therefore measured their SOICR inhibiting properties in order to compare them with the above carvedilol analogs (Table 8). The clinically useful β-blocker metoprolol (95) was devoid of any SOICR-inhibition in the mutant HEK293 single cell bioassay. This result is consistent with our previous finding32 that carvedilol is unique in the family of β-blocker drugs in effectively suppressing SOICR in the HEK293 cell line and in the knock-in mouse model.</p><p>Furthermore, compound 96 has been reported to inhibit Ca2+-induced Ca2+ release (CICR) in skeletal muscle sarcoplasmic reticulum.37 In the RyR2-R4496C mutant HEK293 cell assay it exhibited negligible activity, which improved substantially when its structure was incorporated into the novel carvedilol derivative 97, while further lactamization of the amino alcohol in 98 resulted in complete loss of activity. The thiazepine compounds 9938 and 10239 have been reported to suppress ventricular arrhythmias and sudden death in mice through enhanced binding of the 12.6 kDa FK506 binding protein (FKBP12.6) to the RyR2 channel. However, neither 99, its hydrochloride salt 100, nor its isomer 101 displayed any measurable activity in the present assay. On the other hand, the amide derivative 102 showed activity about one-half that of 1.</p><!><p>These results demonstrate that considerable variation in the structure of carvedilol is possible while retaining strong SOICR-suppressing activity in the RyR2-R4496C mutant HEK293 single cell assay. Thus, 34 of the above compounds (3, 6, 9, 10, 12–14, 31–34, 37, 38, 41, 43–47, 50, 53–56, 58, 64, 66, 68, 69, 83, 84, 86, 93 and 94) proved equal or superior to carvedilol (1) in this assay. This subset of highly potent analogs reveals that significant changes can be tolerated in the catechol, linker and carbazole moieties without loss of activity relative to the clinically useful drug 1. All of the compounds in Table 1, where modifications to the catechol subunit are listed, show significant activity. Beneficial catechol modifications include chlorination of the 4′- and 5′-positions (compound 6), which is expected to block metabolic oxidation at those sites and possibly retard clearance. The 2′-methoxy group can be replaced by H, OH, Me or MeS substituents (compounds 8, 9, 10 and 11) and the 1′-ether oxygen can be replaced by S (compound 12) or CH2 (compounds 13 and 14) without deleterious effects on SOICR inhibition. Remarkably, the simple phenyl derivative 14 is roughly twice as potent as 1, while even the pyridyl and cyclohexyl analogs 15 and 16 show only a three- or two-fold loss of activity. This suggests that the compounds in Table 1 may prove good candidates as SOICR inhibitors.</p><p>The β-amino alcohol functionality plays a key role in mediating β-adrenergic blockade via multiple hydrogen-bonding interactions with the β-receptor.33 In order to determine whether or not this functionality plays a similar role in SOICR inhibition, we investigated the alkylation of these key groups via incorporation into cyclic structures or by simple methylation. However, Table 2 indicates mixed results with respect to the SOICR inhibition shown by such compounds. While the lactam and morpholine derivatives of carvedilol (17 and 21, respectively), as well as the O- and N-methylated compounds 29–31 showed similar or slightly lower SOICR inhibition compared to 1, further alteration of these structures was generally accompanied by severe or total loss of activity.</p><p>The variously homologated analogs in Table 3 (2, 32–34 and 36–38) and the compounds with transposed amino alcohol linkers (39 and 40) all showed strong SOICR suppression in the mutant HEK293 cells. Similarly, relocation of the point of attachment of the linker chain from the 4-position (as in 1) to the 3-, 2- or 1-positions of the carbazole subunit generally had a salutary effect on the compounds in Tables 4 and 5, except for the sultam derivative 42, the lactam 51 and the cyclic carbamate 52.</p><p>The alterations to the carbazole moiety shown in Table 6 produced mixed results. The relatively strong activity of several analogs lacking the carbazole nitrogen or other hydrogen-bonding functionalities is noteworthy. Thus, the fluorenyl derivative 62, naphthyl derivatives 68 and 69, and the weaker but still significantly active adamantyl analogs 70–72 demonstrate that the hydrogen-bonding carbazole NH functionality that participates in binding to the β-receptor is not required for SOICR inhibition. However, N-acylation of the carbazole nitrogen suppressed activity (compounds 59 and 60), while other heterocyclic groups displayed IC50s ranging from 5.7 μM (compound 66) to complete loss of activity. The more extensive structural modifications to the compounds listed in Table 7 also had varied effects on SOICR suppression. Of special interest are the benzomorpholine derivative 83 and the amides 93 and 94, with IC50s ranging from 3.55 to 5.76 μM, compared to 15.9 μM for 1.</p><p>Finally, we note the striking contrast between the most active SOICR inhibitors described above and the SOICR-suppressing abilities of several conventional β-blockers and antiarrhythmic agents shown in Table 8. The complete absence of such activity in the case of metoprolol (95), the dibromocarbazole derivatives 96 and 98, and the thiazepine derivatives 99–101 strongly suggest that the latter compounds express their antiarrhythmic effects through different mechanisms to the active compounds described in Tables 1–7. Only the carvedilol-resembling analog 97 and the thiazepine 102 exhibited any measurable effect on SOICR suppression.</p><!><p>It was recently demonstrated that the effectiveness of carvedilol as an antiarrhythmic therapy for patients with heart failure was due in part to its ability to suppress SOICR by regulating calcium efflux through the RyR2 channel.32 As a continuation of this investigation, we screened ca. 100 mostly novel compounds for their SOICR-suppressing effects on the RyR2-R4496C mutant HEK293 cell line. The wide variety of modifications that are tolerated suggests that RyR2 is a promiscuous channel that accommodates a broad range of ligands, including ones with structural changes to the carbazole, linker chain and catechol moieties of carvedilol.</p><!><p>All compounds subjected to bioassay for which elemental analyses were not provided were of >95% purity as determined by HPLC analysis employing the following conditions: Novapak C18 reversed-phase column, 3.9 × 150 mm; solvent: acetonitrile–water, 70:30, 0.8 mL/min; UV detector: 254 nm. 1H NMR spectra were obtained at 300 or 400 MHz. 13C NMR spectra were obtained at 75 or 101 MHz. 19F NMR spectra were obtained at 376 MHz with hexafluorobenzene (-164 ppm) as the external standard, relative to trichlorofluoromethane (0.00 ppm). Products 2–4 were prepared as described previously in the Supporting Information of ref. 32.</p><!><p>Epoxide 103a (169 mg, 0.706 mmol) and amine 104b (250 mg, 1.06 mmol) were refluxed for 2 h in isopropanol (4 mL). The solvent was then removed under vacuum, the residue was purified by flash chromatography over silica gel (methanol-dichloromethane) to obtain 208 mg (62%) of product 6 as a white solid; mp 149–151 °C; IR (film): 3294, 1544, 1258, 1092 cm−1; 1H NMR (300 MHz, CDCl3) δ 8.21 (d, J = 7.7 Hz, 1H), 8.06 (s, 1H), 7.46–7.14 (m, 4H), 7.06 (d, J = 7.7 Hz, 1H), 6.87 (s, 1H), 6.75 (s, 1H), 6.64 (d, J = 7.7 Hz, 1H), 4.29–3.91 (m, 5H), 3.57 (s, 3H), 3.30–3.12 (m, 4H); 13C NMR (101 MHz, CDCl3) δ 155.2, 149.0, 147.5, 141.1, 138.8, 126.8, 125.2, 124.2, 123.5, 123.0, 122.6, 119.8, 115.3, 113.4, 112.8, 110.2, 104.0, 101.4, 70.4, 69.2, 68.5, 56.3, 52.1, 48.6; MS (ESI), m/z (relative intensity) 475 [M + 1]+. HRMS (EI) calcd for C24H2435Cl2N2O4 [M+]: 474.1113; found: 474.1123.</p><p>Compounds 7–8 and 10–16 were prepared similarly.</p><!><p>Yield: 66%; white solid; mp 146–148 °C; IR (film): 3285, 1452, 1095 cm−; 1H NMR (300 MHz, CDCl3) δ 8.24 (d, J = 7.8 Hz, 1H), 8.07 (s, 1H), 7.44–7.35 (m, 3H), 7.32 (dd, J = 8.0, 8.1 Hz, 1H), 7.20 (ddd, J = 1.6, 6.7, 8.1 Hz, 1H), 7.06 (dd, J = 0.5, 8.1 Hz, 1H), 6.96 (s, 1H), 6.68 (d, J = 7.6 Hz, 1H), 4.38–4.24 (m, 3H), 4.15–4.07 (m, 2H), 3.20–3.11 (m, 3H), 3.06 (dd, J = 6.9, 12.2 Hz, 1H); 13C NMR (75 MHz, CDCl3) δ 154.9, 153.2, 141.4, 139.1, 131.1, 130.7, 126.3, 124.6, 124.3, 122.5, 122.1, 118.8, 114.9, 112.3, 110.1, 104.1, 100.4, 70.1, 68.6, 68.4, 52.2, 48.0; MS (EI), m/z (relative intensity) 478 (4) [M+], 184 (14), 183 (100). HRMS (EI) calcd for C23H2135Cl3N2O3 [M]+: 478.0618; found: 478.0619.</p><!><p>Yield 76%; white solid; mp 109–111 °C; IR (film) 3407, 3244, 3055, 2924, 2873, 1597, 1506, 1453, 1242, 1095, 750, 725 cm−1; 1H NMR (400 MHz, CDCl3) δ 8.27 (d, J = 7.8 Hz, 1H), 8.07 (br s, 1H), 7.19–7.44 (m, 6H), 7.07 (d, J = 7.7 Hz, 1H), 6.95 (dd, J = 10.5, 4.2 Hz, 1H), 6.90 (dd, J = 8.7, 1.0 Hz, 2H), 6.68 (d, J = 7.9 Hz, 1H), 4.38–4.20 (m, 3H), 4.12 (t, J = 5.2 Hz, 2H), 3.19–3.10 (m, 3H), 3.05 (dd, J = 12.3, 7.3 Hz, 1H); 13C NMR (101 MHz, CDCl3) δ 158.9, 155.3, 141.1, 138.9, 129.6, 126.8, 125.2, 123.1, 122.7, 121.1, 119.9, 114.7, 112.9, 110.2, 104.0, 101.5, 70.5, 68.7, 67.3, 52.1, 49.0; MS (ESI) m/z (relative intensity) 377 (100) [M+H]+; HRMS (ESI) calcd for C23H25N2O3 [M+H]+: 377.1860; found: 377.1856.</p><!><p>Yield: 76%; white solid; mp 118–119 °C; IR (film) 3407, 3306, 3055, 2924, 1606, 1497, 1453, 1242, 1098, 750, 719 cm−1; 1H NMR (400 MHz, CDCl3) δ 8.26 (d, J = 7.7 Hz, 1H,), 8.07 (br s, 1H), 7.36–7.44 (m, 2H), 7.32 (dd, J = 8.0, 7.9 Hz, 1H), 7.21 (ddd, J = 8.1, 6.5, 1.8 Hz, 1H), 7.04–7.17 (m, 3H), 6.91–6.79 (m, 2H), 6.68 (d, J = 7.9 Hz, 1H,), 4.40–4.20 (m, 3H), 4.12 (t, J = 5.1 Hz, 2H,), 3.22–3.12 (m, 3H), 3.06 (dd, J = 12.3, 7.1 Hz, 1H), 2.20 (s, 3H); 13C NMR (101 MHz, CDCl3) δ 156.9, 155.2, 141.1, 138.8, 130.8, 127.0, 126.9, 126.8, 125.1, 123.0, 122.6, 120.7, 119.8, 112.9, 111.3, 110.1, 104.0, 101.4, 70.5, 68.6, 67.4, 52.0, 49.0, 16.3; MS (ESI) m/z (relative intensity) 391 (100) [M+H]+. HRMS (ESI) calcd for C24H27N2O3 [M+H]+: 391.2012; found: 391.2016.</p><!><p>Yield 65%; white solid; mp 103–104 °C; IR (film) 3407, 3297, 2917, 1606, 1456, 1435, 1236, 1095, 756, 722 cm−1; 1H NMR (400 MHz, CDCl3) δ 8.27 (d, J = 7.8 Hz, 1H), 8.08 (br s, 1H), 7.35–7.46 (m, 2H), 7.32 (dd, J = 8.0, 8.0 Hz, 1H), 7.21 (ddd, J = 8.0, 6.7, 1.6 Hz, 1H), 7.08–7.15 (m, 2H), 7.06 (d, J = 8.0 Hz, 1H,), 6.97 (ddd, J = 7.6, 7.4, 1.2 Hz, 1H,), 6.81–6.88 (m, 1H), 6.69 (d, J = 7.9 Hz, 1H), 4.15–4.39 (m, 5H), 3.14–3.22 (m, 3H), 3.06 (dd, J = 12.3, 7.0 Hz, 1H), 2.36 (s, 3H); 13C NMR (101 MHz, CDCl3) δ 155.3 (C), 155.2 (C), 141.1 (C), 138.8 (C), 127.9 (C), 126.7 (CH), 125.8 (2 x CH), 125.1 (CH), 123.0 (CH), 122.6 (C), 121.8 (CH), 119.8 (CH), 112.9 (C), 112.0 (CH), 110.1 (CH), 103.9 (CH), 101.4 (CH), 70.3 (CH2), 68.6 (CH), 68.4 (CH2), 51.9 (CH2), 48.7 (CH2), 14.5 (CH3); MS (EI) m/z (relative intensity) 422 (20) [M+], 196 (66), 182 (100), 153 (38). HRMS (EI) calcd for C24H26N2O3S [M]+: 422.1664; found: 422.1676.</p><!><p>Yield 69%; white solid; mp 48–49 °C; IR (film) 3407, 2930, 2833, 1603, 1581, 1456, 1098, 910, 750, 725 cm−1; 1H NMR (400 MHz, CDCl3) δ 8.25 (d, J = 7.8 Hz, 1H), 8.08 (s, 1H), 7.47 – 7.29 (m, 4H), 7.26 – 7.18 (m, 2H), 7.06 (d, J = 8.0 Hz, 1H), 6.90 (td, J = 7.5, 1.1 Hz, 1H), 6.86 (d, J = 8.2 Hz, 1H), 6.68 (d, J = 7.9 Hz, 1H), 4.40–4.15 (m, 3H), 3.88 (s, 3H), 3.14–3.01 (m, 3H), 2.99–2.85 (m, 3H); 13C NMR (101 MHz, CDCl3) δ 158.1, 155.1, 141.0, 138.8, 131.3, 128.1, 126.7, 125.1, 123.2, 123.0, 122.6, 121.2, 119.8, 112.8, 110.8, 110.1, 104.0, 101.3, 70.3, 68.5, 55.9, 51.7, 48.2, 33.0; MS (EI) m/z (relative intensity) 422 (45) [M+], 269 (100), 195 (50), 182 (68), 153 (52), 120 (50). HRMS (EI) calcd for C24H26N2O3S [M]+: 422.1664; found: 422.1658.,</p><!><p>Yield: 64%; white solid; mp 104–105 °C; IR (film) 3407, 2933, 2836, 1603, 1503, 1456, 1239, 1098, 907, 750, 719 cm−1; 1H NMR (400 MHz, CDCl3) δ 8.26 (d, J = 7.8 Hz, 1H), 8.09 (s, 1H), 7.47–7.35 (m, 2H), 7.32 (t, J = 8.0 Hz, 1H), 7.22 (ddd, J = 8.0, 6.6, 1.6 Hz, 1H), 7.17 (td, J = 8.0, 1.7 Hz, 1H), 7.12 (dd, J = 7.3, 1.5 Hz, 1H), 7.06 (d, J = 8.0 Hz, 1H), 6.88 (td, J = 7.4, 1.0 Hz, 1H), 6.84 (d, J = 8.1 Hz, 1H), 6.67 (d, J = 7.9 Hz, 1H), 4.47–4.09 (m, 3H), 3.80 (s, 3H), 3.06 (dd, J = 12.2, 3.5 Hz, 1H), 2.96 (dd, J = 12.2, 7.6 Hz, 1H), 2.86–2.61 (m, 4H), 1.81–1.91 (m, 2H); 13C NMR (101 MHz, CDCl3) δ 157.6, 155.3, 141.1, 138.9, 130.4, 130.0, 127.2, 126.8, 125.1, 123.0, 122.7, 120.6, 119.8, 112.9, 110.4, 110.1, 104.0, 101.4, 70.6, 68.4, 55.4, 52.1, 49.5, 30.2, 27.8; MS (EI) m/z (rel intensity) 404 (40), 182 (100), 178 (88), 153 (50); HRMS (EI+) m/z calcd for C25H28N2O3 [M]+: 404.2100. Found: 404.2101.</p><!><p>Yield: 64%; white solid; mp 110–111 °C; IR (film) 3407, 3084, 2917, 1603, 1450, 1095, 756, 716 cm−1; 1H NMR (400 MHz, CDCl3) δ 8.26 (d, J = 7.8 Hz, 1H), 8.08 (s, 1H), 7.48–7.36 (m, 2H), 7.32 (dd, J = 8.0, 8.0 Hz, 1H), 7.29–7.13 (m, 6H), 7.06 (d, J = 7.9 Hz, 1H), 6.67 (d, J = 7.9 Hz, 1H), 4.36–4.17 (m, 3H), 3.04 (dd, J = 12.2, 3.5 Hz, 1H), 3.00–2.89 (m, 1H), 2.85–2.61 (m, 4H), 1.87 (p, J = 7.2 Hz, 2H); 13C NMR (101 MHz, CDCl3) δ 155.2, 142.1, 141.1, 138.9, 128.5 (2C), 126.8, 125.9, 125.1, 123.0, 122.6, 119.8, 112.8, 110.2, 104.0, 101.4, 70.5, 68.5, 52.2, 49.5, 33.6, 31.7; MS (EI) m/z (relative intensity) 374 (18) [M+], 183 (100). HRMS (EI) calcd for C24H26N2O2 [M]+: 374.1994; found: 374.1994.</p><!><p>Yield 54%; white solid; mp 122–123 °C; IR (film) 3396, 3291, 2916, 1589, 1432, 1286, 1094, 783, 751, 720 cm−1; 1H NMR (400 MHz, CDCl3) δ 8.26 (d, J = 8.0 Hz, 1H), 8.14 (ddd, J = 5.0, 2.0, 0.8 Hz, 1H), 8.08 (s, 1H), 7.55 (ddd, J = 8.3, 7.1, 2.0 Hz, 1H), 7.34–7.43 (m, 2H), 7.32 (dd, J = 8.0, 8.0 Hz, 1H), 7.21 (ddd, J = 8.1, 6.8, 1.4 Hz, 1H), 7.06 (dd, J = 8.1, 0.5 Hz, 1H), 6.86 (ddd, J = 7.1, 5.1, 0.9 Hz, 1H), 6.72 (dt, J = 8.4, 0.9 Hz, 1H), 6.68 (d, J = 8.0 Hz, 1H), 4.43–4.52 (m, 2H), 4.20–4.38 (m, 3H), 2.98–3.25 (m, 5H); 13C NMR (101 MHz, CDCl3) δ 163.8, 155.2, 146.9, 141.1, 138.8, 138.8, 126.8, 125.2, 123.1, 122.6, 119.9, 117.1, 112.9, 111.2, 110.1, 104.0, 101.4, 70.4, 68.5, 65.3, 52.1, 49.0; MS (ESI) m/z (relative intensity) 378 (100) [M+H]+. HRMS (EI) calcd for C22H23N3O3 [M]+: 377.1739; found: 377.1742.</p><!><p>Yield: 68%; viscous oil; IR (neat) 3401, 3280, 3051, 1450, 1094 cm−1; 1H NMR (300 MHz, CDCl3) δ 8.28 (d, J = 7.7 Hz, 1H), 8.14 (s, 1H), 7.41–7.22 (m, 4H), 7.06 (d, J = 7.7 Hz, 1H), 6.68 (d, J = 7.7 Hz, 1H), 4.32–4.22 (m, 3H), 3.62–3.59 (m, 2H), 3.27–3.24 (m, 1H), 3.12–3.07 (m, 1H), 3.01- 2.78 (m, 4H), 2.00–1.89 (m, 2H), 1.72–1.71 (m, 2H), 1.54–1.49 (m, 1H), 1.39–1.19 (m, 5H); 13C NMR (75 MHz, CDCl3) δ 155.2, 141.1, 138.9, 126.8, 125.1, 123.1, 122.7, 119.8, 112.83, 110.2, 104.0, 101.4, 78.1, 70.4, 68.3, 66.6, 52.1, 49.6, 32.4, 25.9, 24.3; MS (EI), m/z (relative intensity) 382 (4) %) [M+], 269 (18), 183 (100). HRMS (EI) calcd for C23H30N2O3 [M]+: 382.2256; found: 382.2270.</p><!><p>Chloroacetyl chloride (113 mg, 1.00 mmol) was added to an ice-cooled solution of carvedilol (1) (406 mg, 1.00 mmol) and triethylamine (152 mg, 1.50 mmol) in chloroform. The reaction mixture was warmed to room temperature and stirred for 6 h. It was then quenched with water (10 mL) and extracted with chloroform. The combined organic layers were washed with saturated NH4Cl solution (25 mL), dried over Na2SO4 and evaporated under reduced pressure. The residue was purified by flash chromatography over silica gel (methanol-dichloromethane) to afford the corresponding chloroacetyl derivative. A suspension of NaH (1.1 mmol) in anhydrous THF was cooled to 0° C. The above product was dissolved in THF and was added to the NaH mixture. The reaction mixture was stirred for 12 h at room temperature, quenched cautiously with water (10 mL) and extracted with ethyl acetate. The combined organic layers were washed with sat. NH4Cl solution, dried over Na2SO4 and evaporated under reduced pressure. Flash chromatography of the residue over silica gel column afforded 321 mg (72%) of 17 as a viscous oil; IR (neat): 3270, 1652, 1504, 1252 cm−1; 1H NMR (300 MHz, CDCl3) δ 8.28–8.19 (m, 2H), 7.44–7.06 (m, 6H), 7.01–6.81 (m, 3H), 6.65 (d, J = 7.7 Hz, 1H), 4.45–4.21 (m, 7H), 3.96–3.79 (m, 4H), 3.69 (s, 3H); 13C NMR (75 MHz, CDCl3) δ 167.1, 154.9, 149.8, 148.1, 141.2, 138.9, 126.7, 125.3, 123.1, 122.5, 122.0, 121.0, 119.9, 114.1, 113.0, 112.0, 110.2, 104.4, 101.3, 72.2, 68.4, 68.0, 67.9, 55.8, 51.3, 47.1; MS (EI), m/z (relative intensity) 446 (17) [M+], 323 (100). HRMS calcd for C26H26N2O5 [M+]: 446.1842; found 446.1844.</p><p>Compounds 18–20 were prepared similarly from the corresponding amino alcohols.</p><!><p>Yield: 92%; viscous oil; IR (neat) 3275, 1650 cm−1; 1H NMR (300 MHz, CDCl3) δ 8.20 (d, J = 7.7 Hz, 1H), 8.09 (s, 1H), 7.40–7.34 (m, 2H), 7.31 (t, J = 8.0 Hz, 1H), 7.12 (dd, J = 7.9, 4.5 Hz, 1H), 7.05 (d, J = 8.1 Hz, 1H), 6.92 (s, 1H), 6.78 (s, 1H), 6.66 (d, J = 7.7 Hz, 1H), 4.48–4.18 (m, 7H), 4.01–3.78 (m, 4H), 3.57 (s, 3H); 13C NMR (75 MHz, CDCl3) δ 167.2, 154.8, 148.9, 147.3, 141.1, 138.8, 128.5, 126.7, 125.4, 123.5, 123.0, 122.4, 119.9, 115.2, 113.3, 112.9, 110.2, 104.4, 101.3, 72.1, 68.3, 68.2, 67.9, 56.1, 51.3, 46.8; MS (EI), m/z (relative intensity) 514 (6) [M+] 323 (100), 183 (18), 154 (48). HRMS calcd for C26H2435Cl2N2O5 [M+]: 514.1062; found: 514.1067.</p><!><p>Yield: 77%; white solid; mp 106–108 °C; IR (film): 3268, 1652 cm−1; 1H NMR (300 MHz, CDCl3) δ 8.22 (d, J = 7.7 Hz, 1H), 8.08 (s, 1H), 7.44–7.18 (m, 5H), 7.09 (d, J = 7.7 Hz, 1H), 6.96 (s, 1H), 6.67 (d, J = 7.7 Hz, 1H), 4.44–4.19 (m, 7H), 4.02–3.82 (m, 4H); 13C NMR (75 MHz, CDCl3) δ 167.7, 154.5, 152.9, 141.3, 139.0, 131.3, 130.8, 126.3, 124.9, 124.6, 122.7, 122.1, 121.7, 119.2, 114.6, 112.5, 110.2, 104.4, 100.8, 71.9, 67.9, 67.8, 67.5, 51.0, 46.7; MS (EI), m/z (relative intensity) 518 (14) [M+], 323 (33), 183 (100). HRMS calcd for C25H2135Cl3N2O4: 518.0567; found: 518.0599.</p><!><p>Yield: 41%; viscous oil; IR (neat) 3267, 1640; 1H NMR (300 MHz CDCl3) δ 8.24 (d, J = 7.7 Hz, 1H), 8.14 (s, 1H), 7.49–7.23 (m, 4H), 7.10 (d, J = 7.7 Hz, 1H), 6.68 (d, J = 7.7 Hz, 1H), 4.50–4.24 (m, 5H), 3.78 (d, J = 6.9 Hz, 2H), 3.66 (dd, J = 15.8, 4.6 Hz, 4H), 3.28–3.17 (m, 1H), 1.92–1.78 (m, 2H), 1.76–1.42 (m, 3H), 1.32–1.08 (m, 5H); 13C NMR (75 MHz, CDCl3) δ 166.9, 154.8, 141.2, 139.0, 126.7, 125.3, 123.1, 122.5, 119.8, 112.9, 110.3, 104.5, 101.3, 72.1, 68.3, 67.9, 66.2, 51.1, 47.7, 32.3, 29.9, 25.8, 24.1; MS (EI), m/z (relative intensity) 422 (24) [M+], 296 (16), 183 (100). HRMS (EI) calcd for C25H30N2O4 [M+]: 422.2206; found: 422.2225.</p><!><p>A solution of 17 (446 mg, 1.00 mmol) in THF was added to a suspension of LiAlH4 (38 mg, 1.0 mmol) in THF and the mixture was stirred at room temperature for 3h. The reaction was quenched with ethyl acetate followed by saturated Na2SO4 solution. The crude mixture was filtered through Celite and the residue was washed with ethyl acetate. The filtrate was dried over Na2SO4, evaporated under reduced pressure and the residue was purified by flash chromatography over silica gel to afford 259 mg (60%) of 21 as a viscous oil; IR (neat) 3290 cm−1; 1H NMR (300 MHz, CDCl3) δ 8.30 (d, J = 7.7 Hz, 1H), 8.09 (s, 1H), 7.46–7.17 (m, 4H), 7.06 (d, J = 7.7 Hz, 1H), 6.96–6.84 (m, 4H), 6.66 (d, J = 7.7 Hz, 1H), 4.40–4.05 (m, 5H), 4.04–3.93 (m, 1H), 3.91–3.76 (m, 1H), 3.78 (s, 3H), 3.34–3.22 (m, 1H), 3.04–2.84 (m, 3H), 2.58–2.31 (m, 2H); 13C NMR (75 MHz, CDCl3) δ 155.3, 149.9, 148.4, 141.1, 138.9, 126.8, 125.1, 123.2, 122.7, 121.8, 121.1, 119.8, 114.2, 112.9, 112.2, 110.1, 104.0, 101.3, 74.2, 69.3, 67.0, 66.9, 57.7, 56.8, 56.0, 53.8; MS (EI), m/z (relative intensity) 432 (17) [M+], 295 (100). HRMS (EI) calcd for C26H28N2O4 [M+]: 432.2049; found: 432.2017.</p><p>Compounds 22–24 were prepared similarly.</p><!><p>Yield: 86%; viscous oil; IR (neat): 3280, 1600, 1503 cm−1; 1H NMR (300 MHz, CDCl3) δ 8.29 (d, J = 7.7 Hz, 1H), 8.09 (s, 1H), 7.46–7.13 (m, 3H), 7.06 (d, J = 7.7 Hz, 1H), 6.98–6.80 (m, 3H), 6.67 (d, J = 7.7 Hz, 1H), 4.37–3.97 (m, 7H), 3.80 (s, 3H), 3.32–3.22 (m, 1H), 3.02–2.84 (m, 3H), 2.58–2.29 (m, 2H); 13C NMR (75 MHz, CDCl3) δ 155.2, 149.9, 141.1, 138.9, 126.8, 125.1, 123.2, 122.7, 121.9,121.0, 120.5, 119.8, 114.5, 113.0, 112.7, 112.1, 110.1, 104.0, 101.4, 74.2, 69.2, 57.6, 56.1, 56.0, 53.8; MS (EI), m/z (relative intensity) 500 (10) [M+], 296 (20), 295 (100). HRMS (EI) calcd for C26H2735Cl2N2O4 [M+]: 500.1270; found: 500.1263.</p><!><p>Yield: 66%; viscous oil; IR (neat): 3308 cm−1; 1H NMR (300 MHz, CDCl3) δ 8.25 (d, J = 7.7 Hz, 1H), 8.07 (s, 1H), 7.46–7.23, (m, 4H), 7.20–7.12 (m, 1H), 7.07 (d, J = 7.7 Hz, 1H), 7.00 (s, 1H), 6.67 (d, J = 7.7 Hz, 1H), 4.38–4.10 (m, 6H), 4.08–3.86 (m, 2H), 3.46–3.25 (m, 1H), 3.16–2.92 (m, 2H), 2.74–2.48 (m, 2H); 13C NMR (75 MHz, CDCl3) δ 155.2, 153.3, 141.0, 138.8, 131.3, 131.0, 126.8, 125.2, 124.8, 123.1, 122.6, 122.3, 119.8, 115.1, 112.9, 110.1, 104.1, 101.4, 74.0, 69.1, 67.8, 66.8, 57.2, 56.6, 53.8; MS (EI), m/z (relative intensity) 504 (7) [M+], 295 (44), 43 (100). HRMS (EI) calcd for C25H2335Cl3N2O3 [M+]: 504.0774; found: 504.0757.</p><!><p>Yield: 60%; viscous oil; IR (neat) 3293, 1606, 1341, 1100 cm−1; 1H NMR (300 MHz, CDCl3) δ 8.31 (d, J = 7.7 Hz, 1H), 8.11 (s, 1H),7.46–7.20 (m, 4H), 7.06 (d, J = 7.7 Hz, 1H), 6.66 (d, J = 7.7 Hz, 1H), 4.38–4.28 (m, 1H), 4.23–4.12 (m, 2H), 3.98 (d, J = 9.9 Hz, 1H), 3.84 (ddd, J = 11.3, 11.2, 2.3 Hz, 1H), 3.66 (t, J = 6.1 Hz, 2H), 3.28–3.17 (m, 2H), 2.85 (d, J = 11.6 Hz, 1H), 2.66 (t, J = 6.0 Hz, 2H), 2.37 (ddd, J = 11.4, 11.3, 3.3 Hz, 1H), 2.32–2.23 (m, 1H), 1.96–1.83 (m, 2H), 1.78–1.62 (m, 2H), 1.55–1.48 (m, 1H), 1.36–1.14 (m, 5H); 13C NMR (75 MHz, CDCl3) δ 155.3, 141.1, 13.9, 126.8, 125.1, 123.3, 122.7, 119.8, 113.0, 110.1, 103.9, 101.3, 74.2, 69.4, 67.0, 65.5, 58.8, 56.9, 53.9, 32.4, 29.8, 25.9, 24.4; MS (EI), m/z (relative intensity) 408 (6) [M+], 295 (100). HRMS (EI) calcd for C25H32N2O3 [M+]: 408.2413; found: 408.2415.</p><!><p>A solution of carvedilol 1 (99 mg, 0.24 mmol), triethylamine (51 μL, 0.37 mmol) and 1,1′-carbonyldiimidazole (43 mg, 0.27 mmol) in 3 mL of dichloromethane was stirred at room temperature for 5 h. The mixture was diluted with dichloromethane and washed with water and saturated aqueous NH4Cl solution, dried over Na2SO4 and concentrated under vacuum. The residue was purified by flash chromatography over silica gel (methanol-dichloromethane) to afford 96 mg (93%) of 25 as a white solid; mp >350° C; IR (film) 3288, 1730, 1107 cm−1; 1H NMR (300 MHz, CDCl3) δ 8.16 (d, J = 7.7 Hz, 1H), 8.08 (s, 1H), 7.46–7.19 (m, 4 H), 7.09 (d, J = 7.7 Hz, 1H), 6.98–6.79 (m, 4H), 6.64 (d, J = 7.7 Hz, 1H), 5.13–4.99 (m, 1H), 4.40 (dd, J = 10.0, 4.5 Hz, 1H), 4.33 (dd, J = 9.9, 5.7 Hz, 1H), 4.28–4.14 (m, 3H), 3.99 (dd, J = 9.0, 6.0 Hz, 1H), 3.88–3.72 (m, 2H), 3.68 (s, 3H); 13C NMR (75 MHz, CDCl3-CD3OD) δ 158.4, 154.5, 149.6, 147.7, 141.4, 139.1, 126.3, 125.0, 122.7, 122.1, 122.0, 121.0, 119.2, 114.1, 112.2, 112.0, 110.2, 104.6, 100.5, 71.9, 68.0, 67.9, 55.6, 43.9; MS (EI), m/z (relative intensity) 432 (58) [M+], 183 (53), 44 (100). HRMS (EI) calcd for C25H24N2O5 [M+]: 432.1685; found: 432.1672.</p><p>Compounds 26 and 27 were prepared similarly.</p><!><p>Yield: 65%; white solid; mp 104–106 °C; IR (film): 3290, 1749 cm−1; 1H NMR (300 MHz, CDCl3) δ 8.12–8.05 (m, 2H), 7.44–7.20 (m, 4H), 7.13 (d, J = 7.7 Hz, 1H), 6.83 (d, J = 7.7 Hz, 1H), 6.67 (m, 2H), 5.08–5.02 (m, 1H), 4.43–4.39 (m, 2H), 4.14–4.04 (m, 4H), 3.85–3.62 (m, 2H), 3.52 (s, 3H); 13C NMR (75 MHz, CDCl3) δ 157.6, 154.4, 148.5, 146.9, 139.7, 137.5, 128.2, 126.8, 124.9, 124.6, 124.4, 123.5, 123.4, 115.5, 114.8, 113.8, 112.9, 107.9, 105.9, 71.3, 68.5, 68.3, 56.0, 48.6, 43.8; MS (EI), m/z (relative intensity) 500 (6) [M+], 222 (10), 68 (100). HRMS (EI) calcd for C25H2235Cl2N2O5 [M+]: 500.0906; found: 500.0936.</p><!><p>Yield: 74%; white solid; mp 146–148 °C; IR (neat) 3290, 1732 cm−1; 1H NMR (300 MHz, CDCl3) δ 8.22–8.14 (m, 2H), 7.45–7.19 (m, 4H), 7.10 (d, J = 7.7 Hz, 1H), 6.66 (d, J = 7.7 Hz, 1H), 5.08–4.98 (m, 1H), 4.46–4.36 (m, 2H), 4.08–4.00 (m, 1H) 3.94–3.83 (m, 2H), 3.68–3.45 (m, 3H), 3.24–3.16 (m, 1H), 1.85–1.39 (m, 6H), 1.28–1.04 (m, 4H); 13C NMR (75 MHz, CDCl3) δ 157.8, 154.7, 141.1, 138.9, 126.7, 125.3, 123.1, 122.5, 120.0, 112.9, 110.2, 104.5, 101.2, 77.9, 71.2, 68.2, 66.5, 48.8, 44.8, 32.2, 25.8, 24.0; MS (EI), m/z (relative intensity) 408 (100) [M+], 183 (87). HRMS (EI) calcd for C24H28N2O4 [M+]: 408.2049; found 408.2027.</p><!><p>A solution of 25 (160 mg, 0.37 mmol) in dry dichloromethane (7 mL) was cooled in an ice bath and a 1.0 M solution of BBr3 in dichloromethane (1.2 mL) was added dropwise. After 1 h, the reaction was warmed to room temperature and stirred an additional 1 h. The reaction was cooled in an ice bath and slowly quenched with water and diluted further with ethyl acetate. The organic phase was separated and washed with water and brine, dried over MgSO4 and concentrated under vacuum. The residue was purified by flash chromatography over silica gel (ethyl acetate-hexanes) to afford 131 mg (85%) of 28 as a white solid; mp 194–196 °C; IR (film) 3423, 3270, 2919, 2847, 1709, 1505, 1114, 746, 742 cm−1; 1H NMR (400 MHz, DMSO-d6) δ 11.28 (s, 1H), 8.91 (s, 1H), 8.06 (d, J = 7.8 Hz, 1H), 7.45 (d, J = 8.1 Hz, 1H), 7.30–7.37 (m, 1H), 7.30 (dd, J = 8.0, 8.0 Hz, 1H), 7.15–7.07 (m, 2H), 6.91 (dd, J = 7.9, 1.3 Hz, 1H), 6.84–6.75 (m, 2H), 6.74–6.67 (m, 2H), 5.15–5.02 (m, 1H), 4.43 (dd, J = 10.8, 3.1 Hz, 1H), 4.34 (dd, J = 10.8, 4.8 Hz, 1H), 4.16 (t, J = 5.4 Hz, 2H), 4.02 (t, J = 9.2 Hz, 1H), 3.80–3.68 (m, 2H), 3.59 (dt, J = 14.5, 5.4 Hz, 1H); 13C NMR (101 MHz, DMSO-d6) δ 157.2, 154.2, 147.1, 146.3, 141.0, 138.8, 126.3, 124.6, 122.1, 121.7, 121.4, 119.1, 118.6, 115.8, 114.7, 111.4, 110.3, 104.2, 100.4, 68.1, 68.2, 66.8, 46.6, 43.2; MS (EI) m/z (relative intensity) 418 (100) [M+], 222 (57), 154 (23); HRMS (EI) calcd for C24H22N2O5 [M]+: 418.1529; found: 418.1524.</p><!><p>Compound 28 (77 mg, 0.18 mmol) was dissolved in ethanol (1 mL) and a 2 M NaOH solution (1.5 mL), and refluxed for 3.5 h. The mixture was cooled to room temperature, neutralized with a 1 M HCl solution, and extracted with ethyl acetate. The combined extracts were dried over Na2SO4 and concentrated under vacuum. The residue was purified by flash chromatography over silica gel (methanol-dichloromethane) to afford 56 mg (78%) of 9 as an off-white solid; mp 160–162 °C; IR (KBr) 3415, 1605, 1503, 1263, 1100, 755, 722 cm−1; 1H NMR (400 MHz, DMSO-d6) δ 11.24 (s, 1H), 8.21 (d, J = 7.6 Hz, 1H), 7.43 (d, J = 8.1 Hz, 1H), 7.32 (ddd, J = 7.1, 7.0, 1.2 Hz, 1H), 7.27 (d, J = 8.0 Hz, 1H), 7.09 (ddd, J = 8.0, 7.0, 1.0, 1H), 7.06 (d, J = 8.0 Hz, 1H,), 6.91 (dd, J = 8.1, 1.3 Hz, 1H,), 6.82–6.74 (m, 2H), 6.73–6.68 (m, 1H), 6.68 (d, J = 8.1 Hz, 1H), 4.22–4.08 (m, 3H), 4.06–3.97 (m, 2H), 2.96–2.86 (m, 3H,) ), 2.80 (dd, J = 12.3, 6.9 Hz, 1H); 13C NMR (101 MHz, DMSO-d6) δ 154.9, 148.0, 146.6, 141.0, 138.8, 126.4, 124.4, 122.4, 122.0, 121.6, 119.0, 118.5, 116.0, 115.8, 111.5, 110.2, 103.7, 100.3, 70.4, 69.2, 68.3, 52.1, 48.2; MS (EI) m/z (relative intensity) 392 (69) [M+], 183 (100), 166 (30). HRMS (EI) calcd for C23H24N2O4 [M]+: 392.1736; found: 392.1731.</p><!><p>A suspension of NaH (60% in oil) (10 mg, 0.25 mmol)) in anhydrous THF was cooled to 0 °C. A solution of carvedilol (1) (100 mg, 0.246 mmol) in THF was added and stirred for 20 min. at the same temperature, followed by iodomethane (35 mg, 0.25 mmol). The reaction mixture was stirred for 4 h at room temperature, quenched with water and extracted with ethyl acetate. The combined organic layers were washed with saturated NH4Cl solution, dried over Na2SO4, evaporated under reduced pressure and the residue was purified by flash chromatography over silica gel to obtain 57 mg (54%) of 29 as viscous oil; IR (film) 3233, 2923, 1600, 1504, 1452, 1252, 732 cm−1; 1H NMR (300 MHz, CDCl3) δ 8.29–8.20 (m, 2H), 7.41–7.31 (m, 3H), 7.28–7.26 m, 1H), 7.06 (d, J = 8.1 Hz, 1H), 6.93–6.87 (m, 4H), 6.66 (d, J = 8.1 Hz, 1H), 4.36–4.21 (m, 5H), 3.78 (s, 3H), 3.27–3.13 (m, 4H), 2.72 (s, 3H); 13C NMR (75 MHz, CDCl3) δ 154.9, 149.7, 147.6, 141.0, 138.8, 126.7, 125.0, 122.9, 122.5, 122.1, 120.9, 119.7, 114.3, 112.7, 111.9, 110.1, 104.0, 101.3, 69.9, 66.1, 66.0, 61.2, 56.4, 55.7, 43.0; MS (EI) m/z (relative intensity) 420 (5) [M+], 283 (41), 194 (100). HRMS (EI) calcd for C25H28N2O4 [M+]: 420.2049; found 420.2038.</p><!><p>A solution of carvedilol (1) (50 mg, 0.12 mmol) and DABCO (2 mg) in dimethyl carbonate (4 mL) and DMF (1 mL) was heated at 95 °C for 18 h. The reaction mixture was partitioned between ether and water. The ether layer was washed with water and brine, dried over Na2SO4 and evaporated under reduced pressure. The residue was purified by flash chromatography on silica gel to give (41 mg, 77%) of the N-methyl derivative of 25 as an oil. The latter product was dissolved in ethanol, 2 N NaOH (2 mL) was added and the mixture was refluxed for 6 h. The ethanol was removed under vacuum and the residue was taken in ethyl acetate and washed with 1 M HCl. The aqueous layer was extracted with ethyl acetate and the combined organic layer was washed with water, brine, dried over Na2SO4 and evaporated under reduced pressure. The residue was purified by flash chromatography over silica gel to obtain 38 mg (73% overall from 1) of 30 as a white solid; mp 130–132 °C; IR (film) 3300, 1587, 1504, 1252 cm−1; 1H NMR (300 MHz, CDCl3) δ 8.30 (d, J = 8.2 Hz, 1H), 7.48–7.36 (m, 3H), 7.24 (dd, J = 12.8, 5.1 Hz, 1H), 7.06 (d, J = 8.2 Hz, 1H), 6.98–6.88 (m, 4H), 6.70 (d, J = 8.2 Hz, 1H), 4.36–4.22 (m, 3H), 4.17 (t, J = 5.1 Hz, 2H), 3.85 (s, 3H), 3.84 (s, 3H), 3.18–3.06 (m, 3H), 3.00 (dd, J = 12.2, 6.9 Hz, 1H), 2.72 (br s, 1H); 13C NMR (75 MHz, CDCl3) δ 155.3, 149.9, 148.3, 142.7, 140.4, 126.6, 125.0, 123.1, 122.2, 121.8, 121.0, 119.3, 114.4, 112.1, 112.0, 108.1, 101.9, 101.1, 70.5, 69.0, 68.6, 56.0, 52.1, 48.8, 29.4; MS (EI) m/z (relative intensity) 420 (13) [M+], 197 (100). HRMS (EI) calcd for C25H28N2O4 [M+]: 420.2049; found: 420.2085.</p><!><p>A solution of epoxide 103a (307 mg, 1.28 mmol) in DMF (5 mL) was added dropwise to a mixture of 60% NaH (101 mg, 2.50 mmol) in DMF (5 mL), cooled in an ice bath. After 5 min, iodomethane (87 μL, 1.4 mmol) was added and the mixture was stirred at room temperature for 4 h. The reaction was quenched with water and brine, and diluted with ethyl acetate. The organic phase was separated and washed with brine, dried over Na2SO4 and concentrated under reduced pressure. The crude material was subjected to flash chromatography over silica gel (ethyl acetate-hexanes) to afford 300 mg (92%) of the corresponding N-methyl derivative 103b as a white solid; mp 76–77 °C; 1H NMR (400 MHz, DMSO-d6) δ 8.21 (d, J = 7.4 Hz, 1H), 7.57 (d, J = 8.2 Hz, 1H), 7.44 (ddd, J = 8.3, 7.2, 1.2 Hz, 1H), 7.39 (dd, J = 8.1, 8.1 Hz, 1H), 7.27 – 7.17 (m, 2H), 6.77 (d, J = 7.9 Hz, 1H), 4.58 (dd, J = 11.3, 2.6 Hz, 1H), 4.12 (dd, J = 11.3, 6.3 Hz, 1H), 3.86 (s, 3H), 3.58 – 3.51 (m, 1H), 2.94 (dd, J = 5.0, 4.3 Hz, 1H), 2.85 (dd, J = 5.1, 2.7 Hz, 1H); 13C NMR (101 MHz, DMSO-d6) δ 154.4, 142.0, 139.8, 126.6, 124.8, 122.3, 121.1, 118.9, 110.9, 108.67, 102.4, 101.1, 68.8, 49.9, 43.8, 29.1.</p><p>The above product (90 mg, 0.36 mmol) was refluxed with the N-methyl derivative of amine 104a (72 mg, 0.40 mmol) in isopropanol for 2 h. The solution was evaporated and the residue was purified by flash chromatography over silica gel (methanol-dichloromethane) to obtain 117 mg (76%) of the corresponding N,N-methylated product as a colorless oil; IR (film) 3496, 3059, 1589, 1503, 1466, 1252, cm−1; 1H NMR (300 MHz, CDCl3) δ 8.35 (d, J = 7.7 Hz, 1H), 7.49–7.34 (m, 3H), 7.28–7.19 (m, 1H), 7.03 (d, J = 8.1 Hz, 1H), 7.00–6.84 (m, 4H), 6.71 (d, J = 7.9 Hz, 1H), 4.58 (dd, J = 11.3, 2.6 Hz, 1H), 4.41–4.22 (m, 3H), 4.16 (t, J = 5.7 Hz, 2H), 3.85 (s, 3H), 3.83 (s, 3H), 3.18–3.03 (m, 1H), 3.00–2.84 (m, 3H), 2.51 (s, 3H); 13C NMR (101 MHz, CDCl3) δ 155.4, 149.8, 148.4, 142.6, 140.4, 126.6, 124.8, 123.1, 122.2, 121.5, 120.9, 119.2, 113.7, 112.1, 112.0, 108.0, 101.8, 101.0, 70.4, 67.2, 67.0, 61.1, 56.8, 55.9, 43.5, 29.4; MS (EI) m/z (relative intensity) 434 (4) [M+], 297 (10), 194 (100). HRMS (EI) m/z calcd for C26H30N2O4 [M+]: 434.2206; found: 434.2206.</p><p>The above carvedilol derivative (186 mg, 0.428 mmol) in DMF (5 mL) was cooled in an ice bath and 60% NaH (34 mg, 0.85 mmol) was added. After 20 min, iodomethane (30 μL, 0.47 mmol) was added and the mixture stirred at room temperature for 5 h. The reaction was quenched with water and brine, and diluted with ethyl acetate. The organic phase was separated and washed with brine, dried over Na2SO4, filtered, and concentrated under reduced pressure. The crude material was subjected to flash chromatography over silica gel (methanol-dichloromethane) to afford 151 mg (79%) of 31 as a viscous, colorless oil; IR (film) 3062, 1591, 1465, 1249, 1104, 1026 cm−1; 1H NMR (400 MHz, CDCl3) δ 8.37 (d, J = 8.0 Hz, 1H), 7.45 (ddd, J = 8.2, 7.1, 1.2, 1H), 7.42–7.32 (m, 2H), 7.23 (ddd, J = 8.0, 7.1, 1.1, 1H), 7.03 (d, J = 8.0 Hz, 1H), 6.95–6.74 (m, 4H), 6.70 (d, J = 7.8 Hz, 1H), 4.44–4.30 (m, 2H), 4.13 (t, J = 6.3 Hz, 2H), 3.97–3.89 (m, 1H), 3.84 (s, 3H), 3.81 (s, 3H), 3.61 (s, 3H), 2.99 (t, J = 6.3 Hz, 2H), 2.96–2.81 (m, 2H), 2.50 (s, 3H); 13C NMR (101 MHz, CDCl3) δ 155.5, 149.6, 148.4, 142.6, 140.3, 126.5, 124.8, 123.2, 122.3, 121.2, 120.9, 119.3, 113.4, 112.1, 111.9, 107.9, 101.6, 100.9, 78.5, 68.5, 67.2, 59.2, 58.1, 57.1, 55.9, 44.3, 29.4; MS (EI) m/z (relative intensity) 448 (2) [M+], 194 (100). HRMS (EI) calcd for C27H32N2O4 [M+]: 448.2362; found: 448.2345.</p><!><p>Compound 32 was obtained from 2 by the same procedure used to obtain 17 from 1. Yield: 56%; white solid, mp 157–160 °C; IR (KBr) 3256, 1634, 1507, 1452, 1257, 1122, 1098, 722 cm−1; 1H NMR (300 MHz, CDCl3) δ 8.36 (br s, 1H), 8.26 (d, J = 7.8 Hz, 1H), 7.39–7.36 (m, 2H), 7.32 (d, J = 8.1 Hz, 1H), 7.19–7.12 (m, 1H), 7.04 (d, J = 8.1 Hz, 1H), 6.98–6.90 (m, 4H), 6.69 (d, J = 8.0 Hz, 1H), 4.44–4.05 (m, 7H), 3.81 (s, 3H), 3.8 –3.74 (m, 2H), 3.70–3.66 (m, 2H), 2.26–2.12 (m, 2H); 13C NMR (75 MHz, CDCl3) δ 167.1, 155.1, 149.4, 147.9, 141.0, 138.7, 126.6, 124.9, 122.7, 122.5, 121.6, 120.9, 119.5, 113.5, 112.6, 111.8, 110.0, 103.7, 101.0, 70.7, 67.8, 67.6, 63.6, 55.7, 53.6, 46.6, 32.7; MS (EI) m/z (relative intensity) 460 (14) [M+], 338 (20), 337 (100). HRMS (EI) calcd for C27H28N2O5 [M+]: 460.1998; found: 460.2005.</p><!><p>The product was obtained from epoxide 106 and amine 104a by the same procedure as used to obtain 6. Yield: 54%; solid white foam; IR IR (film) 3398, 3227, 3055, 2935, 1608, 1501, 1453, 1250, 1092, 787, 749, 720 cm−1; 1H NMR (400 MHz, CDCl3) δ 8.33 (d, J = 7.7 Hz, 1H), 8.28 (s, 1H), 7.42–7.20 (m, 4H), 7.01 (d, J = 8.0 Hz, 1H), 6.99–6.88 (m, 4H), 6.65 (d, J = 8.0 Hz, 1H), 4.22 (t, J = 6.3 Hz, 2H,), 4.09 (t, J = 5.1 Hz, 2H), 3.85 (s, 3H), 3.69–3.58 (m, 1H), 3.06–2.93 (m, 2H), 2.75 (dd, J = 12.0, 2.8 Hz, 1H,), 2.47 (dd, J = 12.0, 9.5 Hz, 1H,), 2.20–2.80 (br s, 2H), 2.05–1.96 (m, 2H), 1.82–1.48 (m, 4H); 13C NMR (101 MHz, CDCl3) δ 155.8, 149.8, 148.4, 141.2, 138.9, 126.8, 125.0, 123.2, 122.9, 121.8, 121.1, 119.7, 114.3, 112.9, 112.1, 110.1, 103.5, 101.2, 69.4, 68.9, 67.9, 56.0, 55.2, 48.6, 34.8, 29.6, 22.5; MS (CI) m/z (relative intensity) 449 (100) [M + H]+. HRMS (CI) calcd for C27H33N2O4 [M + H]+: 449.2440; found: 449.2460.</p><!><p>4-Hydroxycarbazole (56 mg, 0.31 mmol), 107 (160 mg, 0.31 mmol) and K2CO3 (85 mg, 0.62 mmol) were heated in DMF (10 mL) at100 °C for 12 h. The reaction mixture was diluted with water (50 ml) and extracted with dichloromethane. The combined organic layers were dried over Na2SO4, the residue was purified by flash chromatography over silica gel to afford 120 mg (63%) of 108 as a colorless oil. The latter product (70 mg, 0.11 mmol) was dissolved in THF, TBAF in THF (0.22 mL, 1 M) was added, and the mixture was stirred at room temperature overnight. The solvent was removed under vacuum and the residue was chromatographed similarly to afford 24 mg (80%) of the corresponding diol as colorless oil. The diol (735 mg, 2.71 mmol), triethylamine (1.5 mL, 10.8 mmol), p-toluenesulfonyl chloride (1.03 g, 5.4 mmol) and DMAP (5 mol %) were stirred in dry chloroform at room temperature for 2 d. Concentration and flash chromatography afforded 200 mg (17%) of tosylate 109, along with 500 mg of recovered starting material. The tosylate (200 mg, 0.471 mmol), amine 104a (400 mg, 2.40 mmol) and LiBr (104 mg, 1.20 mmol) were heated briefly at 100 °C in DME (1 mL). The reaction mixture was then stirred for 2 d at room temperature, diluted with water (20 mL) and extracted with chloroform. The combined organic layers were dried over Na2SO4, concentrated and subjected to flash chromatography over silica gel to afford 100 mg (51%) of 34 as solid foam; IR (KBr) 3345, 2919, 1605, 1502, 1454, 1255, 1123, 1100, 749, 718 cm−1; 1H NMR (400 MHz, CDCl3) δ 8.28 (d, J = 7.7 Hz, 1H), 8.15 (br s, 1H), 7.43–7.38 (m, 2H), 7.33 (t, J = 8.0 Hz, 1H), 7.23 (ddd, J = 8.0, 6.7, 1.5 Hz, 1H), 7.06 (d, J = 8.0 Hz, 1H), 7.00–6.90 (m, 4H), 6.70 (d, J = 8.0 Hz, 1H), 4.48–4.42 (m, 1H), 4.29 (dd, J = 9.2, 5.2 Hz, 1H), 4.18–4.12 (m, 2H), 3.87 (s, 3H), 3.22–2.98 (m, 5H), 2.10–2.03 (m, 1H), 1.94–1.84 (m, 1H); 13C NMR (101 MHz, CDCl3) δ 155.3, 149.8, 148.1, 141.0, 138.8, 126.7, 124.9, 123.0, 122.6, 121.9, 121.0, 119.6, 114.5, 112.7, 111.9, 110.0, 103.7, 101.2, 71.6, 71.5, 68.4, 55.8, 48.4, 47.8, 31.8; MS (EI) m/z (relative intensity) 420 (11) [M+], 238 (46), 154 (100). HRMS (EI) calcd for C25H28N2O4 [M+]: for 420.2049; found: 420.2069.</p><!><p>Compound 35 was obtained from 34 by the same procedure used to obtain 17 from 1. Yield 23% (overall); viscous oil; IR (KBr): 3285, 2920, 1647, 1504, 1455, 1252, 1221, 1123, 1032, 752 cm−1; 1H NMR (400 MHz, CDCl3) δ 8.26 (br s, 1H), 8.24 (d, J = 7.4 Hz, 1H), 7.41–7.39 (m, 2H), 7.32 (t, J = 8.0 Hz, 1H), 7.23 (ddd, J = 8.1, 6.7, 4.0 Hz, 1H), 7.06 (d, J = 8.1 Hz, 1H), 6.97–6.87 (m, 4H), 6.65 (d, J = 8.0 Hz, 1H), 4.57 (d, J = 15.5 Hz, 1H), 4.36–4.31 (m, 2H), 4.27–4.10 (m, 4H), 3.98–3.83 (m, 3H), 3.80 (s, 3H), 3.76–3.70 (m, 1H), 2.39–2.25 (m, 1H), 2.15–2.08 (m, 1H); 13C NMR (101 MHz, CDCl3) δ 172.1, 154.9, 149.4, 148.0, 140.9, 138.7, 126.6, 125.0, 122.8, 122.4, 121.5, 120.9, 119.6, 113.2, 112.7, 111.7, 110.03, 103.9, 101.2, 78.6, 72.3, 69.8, 67.7, 55.7, 49.0, 47.8, 31.9; MS (EI) m/z (relative intensity) 460 (16) [M+], 337 (100). HRMS (EI) calcd for C27H28N2O5 [M+]: 460.1998; found: 460.1990.</p><!><p>Epoxide 103a (100 mg, 0.418 mmol) and amine 110 (151 mg, 0.836 mmol) were stirred for 24 h at 50° C in anhydrous DME (6 mL) in the presence of a catalytic amount of LiBr. The solvent was then removed under vacuum, the residue was taken up in ether (10 mL) and washed with water. The aqueous layer was extracted with ether and the combined organic layers were washed with brine, dried over Na2SO4 and evaporated under reduced pressure. The residue was purified by flash chromatography over silica gel (ethyl acetate-hexanes) to obtain 86 mg (49%) of product 36 as a viscous oil; IR (film) 3352, 1504, 1452 cm−1; 1H NMR (300 MHz; CDCl3) δ 8.29–8.24 (m, 2H), 7.42–7.28 (m, 2H), 7.22–7.18 (m, 1H), 7.06 (d, J = 8.2 Hz, 1H), 6.96–6.82 (m, 5H), 6.65 (d, J = 8.2 Hz, 1H), 4.39–4.09 (m, 2H), 4.12 (t, J = 5.1 Hz, 2H), 3.88–3.79 (m, 1H), 3.85 (s 3H), 3.60–3.20 m, 2H), 3.20–2.93 (m, 3H), 2.15–2.05 (m, 2H); 13C NMR (75 MHz; CDCl3) δ 155.6, 149.7, 148.3, 141.1, 138.9, 126.9, 125.0, 123.0, 122.8, 121.8, 121.1, 119.7, 114.2, 112.7, 112.0, 110.2, 103.7, 101.3, 70.3, 68.1, 68.0, 56.1, 52.1, 47.7, 28.6; MS (EI) m/z (relative intensity) 420 (12) [M+], 183 (44), 154 (70), 45 (100). HRMS (EI) calcd for C25H28N2O4: 420.2049; found 420.2088.</p><!><p>The product was obtained from epoxide 103a and amine 111a by the same procedure used to obtain 6. Yield 82%; off-white solid; mp 94–95 °C; IR (film) 3402, 3291, 3050, 1606, 1455, 1243, 1100 cm−1; 1H NMR (400 MHz, CDCl3) δ 8.27 (d, J = 7.8 Hz, 1H), 8.15 (s, 1H), 7.42–7.19 (m, 6H), 7.04 (d, J = 8.0 Hz, 1H), 6.93 (td, J = 7.4, 0.9 Hz, 1H), 6.86 (d, J = 8.2 Hz, 1H), 6.66 (d, J = 7.9 Hz, 1H), 4.59 (s, 2H), 4.33–4.15 (m, 3H), 3.82 (s, 3H), 3.67 (t, J = 5.1 Hz, 2H), 3.05 (dd, J = 12.3, 3.5 Hz, 1H), 3.02–2.86 (m, 5H); 13C NMR (101 MHz, CDCl3) δ 157.4, 155.3, 141.1, 138.9, 129.4, 129.0, 126.8, 126.6, 125.1, 123.1, 122.7, 120.6, 119.8, 112.9, 110.5, 110.1, 104.0, 101.4, 70.4, 69.6, 68.5, 68.2, 55.5, 52.1, 49.4; MS (EI) m/z (relative intensity) 420 (8) [M+], 194 (46), 183 (100), 154 (48), 121 (92), 91 (50); HRMS (EI) calcd for C25H28N2O4 [M]+: 420.2049; found: 420.2056.</p><!><p>The product was obtained from epoxide 103a and amine 111b by the same procedure used to obtain 6. Yield 72%; solid white foam; IR (film) 3405, 3299, 3056, 2919, 2856, 1603, 1455, 1100, 751, 720 cm−1; 1H NMR (400 MHz, CDCl3) δ 8.27 (d, J = 7.8 Hz, 1H), 8.09 (s, 1H), 7.46–7.18 (m, 9H), 7.06 (d, J = 8.0 Hz, 1H), 6.68 (d, J = 8.0 Hz, 1H), 4.52 (s, 2H), 4.34–4.18 (m, 3H), 3.63 (t, J = 5.1 Hz, 2H), 3.06 (dd, J = 12.2, 3.4 Hz, 1H), 3.01–2.86 (m, 3H), 2.67 (br s, 2H); 13C NMR (101 MHz, CDCl3) δ 155.2, 141.1, 138.8, 138.2, 128.5, 127.9, 127.8, 126.7, 125.0, 123.0, 122.6, 119.7, 112.7, 110.1, 104.0, 101.3, 73.3, 70.4, 69.5, 68.5, 52.1, 49.4; MS (EI) m/z (relative intensity) 390 (5) [M+], 154 (38), 149 (86), 91 (100); HRMS (EI) calcd for C24H26N2O3 [M+]: 390.1943; found: 390.1939.</p><!><p>The product was obtained from epoxide 113 and amine 112 by the same procedure used to obtain 36. Yield 55%; solid off-white foam; IR (KBr) 3399, 1606, 1586, 1455, 1254, 1119, 751 cm−1; 1H NMR (400 MHz, CDCl3) δ 8.26 (d, J = 7.8 Hz, 1H), 8.17 (br s, 1H), 7.40–7.35 (m, 2H), 7.31 (t, J = 8.0 Hz, 1H), 7.20 (ddd, J = 8.1, 6.4, 1.9, 1H), 7.04 (d, J = 8.0 Hz, 1H), 6.97–6.85 (m, 4H), 6.66 (d, J = 8.0 Hz, 1H), 4.35 (t, J = 5.3 Hz, 2H), 4.19–4.13 (m, 1H), 4.09–4.01 (m, 2H), 3.81 (s, 3H), 3.26 (dd, J = 5.8, 4.4 Hz, 2H), 3.22 (br s, 1H), 3.06 (dd, J = 12.2, 4.1 Hz 1H), 2.97 (dd, J = 12.2, 7.3–1H); 13C NMR (101 MHz, CDCl3) δ 155.2, 149.9, 148.2, 140.9, 138.7, 126.6, 125.0, 122.9, 122.5, 122.0, 120.9, 119.7, 115.0, 112.3, 111.9, 110.0, 103.8, 101.2, 72.7, 68.2, 67.2, 55.8, 51.7, 48.9; MS (EI) m/z (relative intensity) 406 (5) [M+], 238 (100). HRMS (EI) calcd for C24H26N2O4 [M+]: 406.1893; found: 406.1904.</p><!><p>Compound 40 was obtained from 39 by the same procedure used to obtain 17 from 1. Yield 52% (overall); solid white foam; IR (KBr) 3257, 1646, 1501, 1450, 768, 717 cm−1; 1H NMR (300 MHz, CDCl3) δ 8.50 (br s, 1H), 8.27 (d, J = 7.8 Hz, 1H), 7.36–7.33 (m, 2H), 7.29 (d, J = 8.0 Hz, 1H), 7.15 (ddd, J = 8.0, 5.1, 3.1 Hz, 1H), 7.02 (d, J = 8.0 Hz, 1H), 6.97 (dd, J = 7.0, 1.3–1H), 6.89–6.86 (m, 2H), 6.79 (td, J = 7.8, 1.8 Hz, 1H), 6.63 (d, J = 8.0 Hz, 1H), 4.44–4.38 (m, 2H), 4.36 (d, J = 16.7 Hz, 1H), 4.21 (d, J = 16.6 Hz, 1H), 4.10–3.99 (m, 3H), 3.91–3.82 (m, 2H), 3.77 (s, 3H), 3.80–3.72 (m, 1H), 3.62 (dd, J = 11.9, 2.9 Hz, 1H); 13C NMR (75 MHz, CDCl3) δ 167.0, 154.7, 149.8, 147.6, 141.0, 138.7, 126.6, 125.0, 122.6, 122.3, 120.8, 119.5, 114.8, 112.4, 111.9, 110.1, 104.0, 101.0, 71.8, 69.4, 67.6, 65.9, 55.6, 50.1, 46.6; MS (CI) m/z (relative intensity) 464 (23) [M+ NH4]+, 447 (100) [M + H]+. HRMS (CI) calcd for C26H27N2O5 [M + H]+: 447.1920; found: 447.1942.</p><!><p>Compound 41 was obtained from 3 by the same procedure used to obtain 17 from 1. Yield 49% (overall); white solid; mp 63–65 °C; IR (KBr) 3458, 3378, 2922, 2834, 1507, 1452, 1253, 1222, 1186, 1121, 745 cm−1; 1H NMR (300 MHz, CDCl3) δ 7.98 (d, J = 7.8 Hz, 1H), 7.96 (br s, 1H), 7.54 (d, J = 2.4 Hz, 1H), 7.43–7.36 (m, 2H), 7.32 (d, J = 8.7 Hz, 1H), 7.23–7.15 (m, 1H), 7.05 (dd, J = 8.7, 2.4 Hz, 1H), 6.95–6.82 (m, 4H), 4.40–4.03 (m, 7H), 3.78 (s, 3H), 3.92–3.76 (m, 4H); 13C NMR (101 MHz, CDCl3) δ 167.1, 152.7, 149.7, 148.1, 140.5, 135.0, 126.2, 123.9, 123.3, 121.9, 120.4, 119.3, 115.6, 113.8, 112.1, 111.5, 111.0, 104.9, 72.2, 69.5, 68.0, 67.9, 56.0, 51.0, 47.0); MS (EI) m/z (relative intensity) 445 (20), 322 (100). HRMS (EI) calcd for C26H26N2O5 [M+]: 446.1842; found: 446.1861.</p><!><p>A suspension of 3 (101 mg, 0.249 mmol) and Et3N (60 μL, 0.4 mmol) in dry dichloromethane (3 mL) was cooled in an ice bath. After 5 min, a solution of bromomethanesulfonyl chloride49 (84 mg, 0.43 mmol) in dry dichloromethane (0.5 mL) was added dropwise. The ice bath was removed and the reaction was stirred at room temperature for 24 h. The reaction was concentrated under vacuum and the residue purified by flash chromatography over silica gel (ethyl acetate hexanes) to afford the intermediate sulfonamide as a solid white foam. A solution of the sulfonamide (48 mg, 0.086 mmol) in dry THF (2 mL) was cooled to 10 °C and treated with NaH (6 mg, 60% dispersion in oil, 0.15 mmol). The reaction was stirred for 20 h at room temperature, quenched with water and extracted with ethyl acetate. The combined organic extracts were dried over Na2SO4 and concentrated under vacuum. The residue was purified by flash chromatography over silica gel (ethyl acetate-hexanes) to afford 28 mg (24% overall) of 42 as an off-white solid; mp 71–73 °C; IR (film) 3396, 3007, 2919, 1503, 1452, 1329, 1253, 1160, 1120, 745 cm−1; 1H NMR (300 MHz, DMSO-d6) δ 11.05 (s, 1H), 8.06 (d, J = 7.6 Hz, 1H,), 7.72 (d, J = 1.6 Hz, 1H,), 7.47–7.29 (m, 3H), 7.15–6.83 (m, 6H), 5.02 (d, J = 11.7 Hz, 1H), 4.94 (d, J = 11.7 Hz, 1H), 4.44–4.37 (m, 1H), 4.16–4.10 (m, 4H), 3.72 (s, 3H), 3.92–3.55 (m, 4H,); 13C NMR (101 MHz, CDCl3) δ 152.6, 149.7, 147.8, 140.5, 135.0, 126.2, 124.0, 123.3, 122.2, 121.1, 120.4, 119.4, 115.6, 114.0, 112.1, 111.5, 111.0, 105.0, 81.4, 74.5, 68.9, 68.7, 55.9, 54.2, 47.2; MS (CI) m/z (relative intensity) 500 [M + NH4]+ (100), 418 (30), 354 (20). HRMS (EI) calcd for C25H26N2O6S [M+]: 482.1512; found: 482.1499.</p><!><p>Compound 43 was obtained from epoxide 114b and amine 104a by the same procedure as used to obtain 6. Yield: 62%; off-white solid; mp 126–128 °C; IR (KBr) 3408, 3299, 3060, 2940, 2877, 2834, 1588, 1495, 1459, 1250, 1183, 1123, 1024, 745 cm−1; 1H NMR (300 MHz, DMSO-d6) δ 11.00 (s, 1H), 8.07 (d, J = 7.7 Hz, 1H), 7.67 (d, J = 2.3 Hz, 1H), 7.42 (d, J = 8.0 Hz, 1H), 7.39–7.29 (m, 2H), 7.09 (dd, J = 7.5, 7.4 Hz, 1H), 7.04–6.80 (m, 5H), 4.75 (d, J = 5.0 Hz, 1H), 4.15 (t, J = 6.3 Hz, 2H), 4.02 (t, J = 5.5 Hz, 2H), 3.85–3.77 (m, 1H), 3.74 (s, 3H), 2.92 (t, J = 5.5 Hz, 2H), 2.74–2.53 (m, 2H), 1.96–1.75 (m, 2H); 13C NMR (101 MHz, DMSO-d6) δ 152.2, 149.1, 148.0, 140.3, 134.4, 125.2, 122.7, 122.4, 121.0, 120.7, 120.2, 117.8, 115.1, 113.6, 112.2, 111.4, 110.9, 103.9, 68.2, 66.3, 65.2, 55.5, 55.4, 48.2, 34.8; MS (EI) m/z (relative intensity) 420 (100) [M+], 419 (32), 389 (17), 388 (53). HRMS (EI) calcd for C25H28N2O4 [M+]: 420.2049; found: 420.2044.</p><!><p>Compound 44 was obtained from epoxide 114c and amine 104a by the same procedure as used to obtain 6. Yield: 56%; white solid; mp 116–118 °C; IR (film) 3402, 3236, 2942, 1500, 1449, 1249, 1180, 742 cm−1; 1H NMR (300 MHz, DMSO-d6) δ 11.00 (s, 1H), 8.07 (d, J = 7.6 Hz, 1H), 7.66 (d, J = 2.3 Hz, 1H), 7.43 (d, J = 8.0 Hz, 1H,), 7.36 (d, J = 8.7 Hz, 1H,), 7.33 (dd, J = 7.2, 7.0 Hz, 1H,), 7.09 (dd, J = 7.6, 7.1 Hz, 1H), 7.00–6.81 (m, 5H), 4.67 (d, J = 4.1 Hz, 1H), 4.08–3.99 (m, 4H), 3.74 (s, 3H), 3.70–3.60 (m, 1H), 2.92 (t, J = 5.5 Hz, 2H), 2.71–2.49 (m, 2H), 1.95–1.45 (m, 4H); 13C NMR (101 MHz, DMSO-d6) δ 152.2, 149.1, 148.0, 140.3, 134.4, 125.2, 122.7, 122.4, 121.0, 120.7, 120.2, 117.8, 115.1, 113.6, 112.2, 111.4, 110.8, 103.9, 68.9, 68.3, 68.2, 55.6, 55.4, 48.2, 31.6, 25.4; MS (EI) m/z (relative intensity) 434 (50) [M+], 252 (77), 180 (100), 154 (14). HRMS (EI) calcd for C26H30N2O4 [M+]: 434.2206; found: 434.2203.</p><!><p>Compound 45 was obtained from epoxide 114d and amine 104a by the same procedure as used to obtain 6. Yield: 54%; solid white foam; IR (film) 3406, 3234, 3060, 2938, 1502, 1250, 743 cm−1; 1H NMR (300 MHz, CDCl3) δ 8.00 (d, J = 8.1 Hz, 1H), 7.98 (br s, 1H), 7.53 (d, J = 2.3 Hz, 1H), 7.40–7.33 (m, 2H), 7.30 (d, J = 8.7 Hz, 1H), 7.20–7.14 (m, 1H), 7.03 (dd, J = 8.7, 2.4 Hz, 1H), 6.96–6.86 (m, 4H), 4.10 (t, J = 5.2 Hz, 2H), 4.05 (t, J = 6.4 Hz, 2H), 3.83 (s, 3H), 3.70–3.60 (m, 1H), 3.08–2.98 (m, 2H), 2.81 (dd, J = 12.1, 2.9 Hz, 1H), 2.77–2.53 (br s, 2H), 2.51 (dd, J = 12.0, 9.5 Hz, 1H), 1.90–1.80 (m, 2H), 1.77–1.42 (m, 4H); 13C NMR (101 MHz, CDCl3) δ 153.4, 149.8, 148.3, 140.5, 134.6, 125.9, 123.9, 123.5, 121.9, 121.1, 120.4, 119.1, 115.8, 114.4, 112.0, 111.4, 110.9, 104.5, 69.3, 68.0, 68.8, 56.0, 55.1, 48.5, 34.8, 29.7, 22.5; MS (CI) m/z (relative intensity) 449 (100) [M + H]+. HRMS (CI) calcd for C27H33N2O4 [M + H]+: 449.2440; found: 449.2437.</p><!><p>A solution of bromoalcohol 115 (151 mg, 0.452 mmol), amine 104a (149 mg, 0.892 mmol) and Et3N (0.14 mL, 0.98 mmol) in methanol (2 mL) was heated at 60 °C for 18 h. The reaction mixture was diluted with water and extracted with dichloromethane. The combined organic extracts were washed with brine, dried over Na2SO4 and concentrated under vacuum. The residue was purified by flash chromatography over silica gel (methanol-dichloromethane) to afford 94 mg (49%) of 46 as a white solid; mp 134–136 °C; IR (film) 3402, 3351, 2927, 2839, 1500, 1452, 1252, 1180, 1026, 745 cm−1; 1H NMR (400 MHz, DMSO-d6) δ 11.01 (s, 1H), 8.07 (d, J = 7.8 Hz, 1H), 7.67 (d, J = 2.4 Hz, 1H), 7.43 (d, J = 8.0 Hz, 1H), 7.37 (d, J = 8.8 Hz, 1H), 7.33 (ddd, J = 8.0, 7.9, 1.2 Hz, 1H), 7.09 (ddd, J = 8.0, 7.9, 1.0 Hz, 1H), 7.02 (dd, J = 8.8, 2.5 Hz, 1H), 6.99–6.82 (m, 4H), 4.00 (t, J = 5.7 Hz, 2H), 3.98–3.90 (m, 3H), 3.74 (s, 3H), 2.90 (t, J = 5.7 Hz, 2H), 2.79 (t, J = 6.9 Hz, 2H), 1.80–1.75 (m, 1H), 1.65–1.58 (m, 1H); 13C NMR (101 MHz, DMSO-d6) δ 152.3, 149.2, 148.0, 140.3, 134.5, 125.2, 122.7, 122.4, 121.1, 120.7, 120.2, 117.9, 115.2, 113.8, 112.2, 111.4, 110.8, 104.0, 73.1, 68.1, 67.7, 55.4, 48.1, 46.1, 33.4; MS (EI) m/z (relative intensity) 420 (1) [M+], 238 (100), 180 (30). HRMS (EI) calcd for C25H28N2O4 [M+]: 420.2049; found: 420.2032.</p><!><p>Compound 47 was obtained from epoxide 114e and amine 104a by the same procedure as used to obtain 6. Yield 53%; white solid; mp 62–64 °C; IR (KBr) 3382, 2927, 2836, 1575, 1497, 1456, 1286, 1249, 11955, 1120, 1023, 794, 756 cm−1; 1H NMR (400 MHz, DMSO-d6) δ 11.05 (s, 1H), 7.91 (dd, J = 9.5, 2.6 Hz, 1H), 7.70 (d, J = 2.4 Hz, 1H), 7.42 (dd, J = 8.8, 4.4 Hz, 1H,), 7.37 (d, J = 8.8 Hz, 1H,), 7.18 (ddd, J = 9.4, 8.8, 2.6 Hz, 1H,), 7.05 (dd, J = 8.8, 2.5 Hz, 1H), 6.99–6.93 (m, 2H), 6.92–6.81 (m, 2H), 5.07 (d, J = 2.6 Hz, 1H,), 4.06–3.93 (m, 5H), 3.73 (s, 3H), 2.93 (t, J = 5.5 Hz, 2H), 2.83 (dd, J = 11.4, 3.7 Hz, 1H), 2.71 (dd, J = 11.8, 6.4 Hz, 1H); 13C NMR (101 MHz, DMSO-d6) δ 155.9 (d, 1JC F = 230 Hz), 152.1, 149.1, 148.0, 136.7, 135.6, 122.7 (d, 3JC F = 10 Hz), 122.4 (d, 4JC F = 4 Hz), 121.0, 120.6, 116.1, 113.6, 113.0 (d, 2JC F = 25 Hz), 112.2, 111.8, 111.7 (CH, d, 3JC F = 9 Hz), 105.6 (d, 2JC F = 23 Hz), 104.1, 71.4, 68.2, 55.4, 52.4, 48.4; 19F NMR (376 MHz, DMSO-d6) δ −125.6; MS (EI) m/z (relative intensity) 424 (18) [M+], 201 (97), 180 (100), 172 (28), 44 (39). HRMS (EI) calcd for C24H25FN2O4 [M+]: 424.1798; found: 424.1790.</p><!><p>A mixture of 9-t-butyloxycarbonyl-3-aminocarbazole (576 mg, 2.04 mmol) and epichlorohydrin (0.11 mL, 1.4 mmol) in absolute ethanol (3.5 mL) was refluxed for 6 h. The reaction mixture was cooled to room temperature, concentrated under vacuum and the residue was purified by flash chromatography over silica gel (methanol-dichloromethane) to afford 369 mg (70%) of 116 as a yellow foam. A mixture of 116 (201 mg, 0.536 mmol), 104a (177 mg, 1.06 mmol), K2CO3 (91 mg, 0.66 mmol) and catalytic KI in absolute ethanol (3.5 mL) was refluxed for 3 h. The reaction mixture was cooled to room temperature, filtered and washed with dichloromethane. The filtrate was concentrated under vacuum and the residue purified by flash chromatography over silica gel (methanol-dichloromethane) to afford 177 mg (66%) of t-Boc-protected 48 as a yellow solid foam. A solution of the latter product (100 mg, 0.198 mmol) in dichloromethane (2.5 mL) was stirred in TFA (0.38 mL, 4.9 mmol) at room temperature for 3.5 h. The reaction mixture was treated with a 1 M NaOH and extracted with dichloromethane. The combined organic extracts were dried over Na2SO4 and concentrated under vacuum. The residue was purified by flash chromatography over silica gel (methanol-dichloromethane) to afford 52 mg (65%) of 48 as a white solid; mp 135–137 °C; IR (KBr) 3408, 3299, 3060, 2940, 2877, 2834, 1588, 1495, 1459, 1250, 1183, 1123, 1024, 745 cm−1; 1H NMR (400 MHz, DMSO-d6) δ 10.75 (s, 1H), 7.94 (d, J = 7.6 Hz, 1H), 7.36 (d, J = 8.0 Hz, 1H), 7.32–7.20 (m, 3H,), 7.12–6.80 (m, 6H), 4.92 (br s, 1H,), 4.03 (t, J = 4.8 Hz, 2H,), 3.89–3.80 (m, 1H), 3.74 (s, 3H), 3.20 (dd, J = 12.2, 5.1 Hz, 1H,), 3.05 (dd, J = 12.2, 6.4 Hz, 1H,), 2.93 (t, J = 5.0 Hz, 2H,), 2.79 (dd, J = 11.6, 3.9 Hz, 1H,), 2.68 (dd, J = 11.6, 7.3 Hz, 1H); 13C NMR (101 MHz, DMSO-d6) δ 149.1, 148.0, 142.4, 140.0, 132.6, 124.7, 123.0, 122.3, 121.0, 120.6, 119.8, 117.4, 114.5, 113.7, 112.2, 111.3, 110.6, 101.0, 68.2, 68.1, 55.4, 53.6, 49.1, 48.3; MS (EI) m/z (relative intensity) 405 (100) [M+], 220 (50), 195 (80), 182 (40), 167 (33). HRMS (EI) calcd for C24H27N3O3 [M+]: 405.2052; found: 405.2038.</p><p>Compounds 49–58 were prepared in the usual manner, by conversion of the corresponding hydroxycarbazoles to the epoxides 117a-117e, followed by epoxide-opening with amines 104a or 104m-104o, as shown in Scheme 5. The lactams 49 and 51, and the oxazolidinone 52, were obtained by cyclization of the corresponding amino alcohols with chloroacetyl chloride and 1,1-carbonyldiimidazole, respectively.</p><!><p>Yield: 57% (overall from 4); white solid; mp 69–71 °C; IR (KBr) 3407, 3063, 2929, 2873, 1647, 1607, 1500, 1460, 1250, 1175, 1022, 747 cm−1; 1H NMR (300 MHz, DMSO-d6) δ 11.12 (s, 1H), 7.99 (d, J = 7.6 Hz, 1H), 7.97 (d, J = 8.5 Hz, 1H), 7.42 (d, J = 8.0 Hz, 1H), 7.28 (dd, J = 7.9, 7.2 Hz, 1H), 7.10 (dd, J = 7.5, 7.2 Hz, 1H), 7.04–6.83 (m, 5H), 6.80 (dd, J = 8.5, 2.1 Hz, 1H,), 4.28–4.10 (m, 7H), 3.77 (s, 3H), 3.86–3.58 (m, 4H); 13C NMR (101 MHz, DMSO-d6) δ 165.9, 157.3, 149.2, 147.7, 140.9, 139.7, 124.2, 122.5, 121.4, 120.9, 120.7, 119.2, 118.5, 116.5, 113.7, 112.4, 110.6, 107.8, 95.4, 71.5, 68.1, 66.8, 66.4, 55.6, 49.1, 45.6; MS (EI) m/z (relative intensity) 445 (15), 322 (75), 181 (49), 153 (100), 77 (21). HRMS (EI) m/z calcd for C26H26N2O5 [M+]: 446.1842; found: 446.1824.</p><!><p>Yield: 65% from epoxide 117b and amine 104a; 39% overall from 2-hydroxycarbazole; mp 93–95 °C; IR (KBr) 3415, 3052, 2928, 2871, 1605, 1503, 1263, 1100, 755, 722 cm−1; 1H NMR (400 MHz, DMSO-d6) δ 11.07 (s, 1H), 7.97 (d, J = 7.8 Hz, 1H), 7.94 (d, J = 8.5 Hz, 1H), 7.41 (d, J = 8.0 Hz, 1H), 7.27 (ddd, J = 8.2, 7.1, 1.2 Hz, 1H,), 7.09 (ddd, J = 7.9, 7.8, 1.0 Hz, 1H), 6.99–6.93 (m, 3H), 6.92–6.82 (m, 2H), 6.76 (dd, J = 8.5, 2.2 Hz, 1H), 4.75 (d, J = 4.1 Hz, 1H), 4.14 (t, J = 6.6 Hz, 2H), 4.01 (t, J = 5.6 Hz, 2H), 3.85–3.75 (m, 1H), 3.74 (s, 3H), 2.90 (t, J = 5.6 Hz, 2H), 2.64 (dd, J = 11.8, 4.2 Hz, 1H), 2.63 (dd, J = 11.8, 7.4 Hz, 1H,), 1.99–1.88 (m, 1H), 1.84–1.78 (m, 1H); 13C NMR (101 MHz, DMSO-d6) δ 157.8, 149.1, 148.0, 141.0, 139.6, 124.0, 122.6, 121.0, 120.8, 120.7, 119.1, 118.4, 116.0, 113.6, 112.2, 110.5, 108.0, 95.0, 68.3, 66.3, 64.7, 55.6, 55.4, 48.3, 34.7; MS (EI) m/z (relative intensity) 238 (100), 180 (45), 100 (16). HRMS (EI) calcd for C25H28N2O4 [M+]: 420.2049; found: 420.2029.</p><!><p>Yield: 61% overall from 50; white solid; mp 160–161 °C; IR (film) 3396, 3002, 2932, 2876, 1641, 1503, 1460, 1255, 1219, 768 cm−1; 1H NMR (400 MHz, DMSO-d6) δ 11.07 (s, 1H), 7.97 (d, J = 7.8 Hz, 1H), 7.95 (d, J = 8.6 Hz, 1H), 7.41 (d, J = 8.0 Hz, 1H), 7.27 (dd, J = 8.0, 7.1 Hz, 1H), 7.09 (dd, J = 7.5, 7.4 Hz, 1H), 7.01–6.94 (m, 5H), 6.76 (dd, J = 8.5, 1.9 Hz, 1H), 4.20–4.05 (m, 6H), 4.03–3.95 (m, 1H), 3.77–3.68 (m, 1H), 3.73 (s, 3H), 3.66–3.49 (m, 3H), 2.08–1.94 (m, 2H); 13C NMR (101 MHz, DMSO-d6) δ 166.0, 157.5, 149.1, 147.7, 140.9, 139.7, 124.1, 122.5, 121.3, 120.8, 120.7, 119.2, 118.4, 116.2, 113.6, 112.3, 110.5, 107.9, 95.2, 70.1, 67.0, 66.3, 63.7, 55.5, 52.1, 45.4, 32.0; MS (EI) m/z (relative intensity) 460 (24) [M+], 337 (100). HRMS (EI) calcd for C27H28N2O5 [M+]: 460.1998; found: 460.1988.</p><!><p>Yield: 87% overall from 50; white solid; mp 61–63 °C; IR (film) 3399, 3320, 3002, 2926, 1739, 1500, 1460, 1249, 749 cm−1; 1H NMR (400 MHz, DMSO-d6) δ 11.09 (s, 1H), 7.98 (d, J = 7.8 Hz, 1H), 7.95 (d, J = 8.5 Hz, 1H), 7.41 (d, J = 8.0 Hz, 1H), 7.28 (dd, J = 7.8, 7.3 Hz, 1H), 7.10 (dd, J = 7.6, 7.3 Hz, 1H), 7.01–6.83 (5H, m), 6.76 (dd, J = 8.5, 1.9 Hz, 1H,), 4.80–4.70 (m, 1H), 4.21–4.11 (m, 2H), 4.09 (t, J = 5.3 Hz, 2H), 3.88 (t, J = 8.6 Hz, 1H,), 3.73 (s, 3H), 3.58–3.50 (m, 3H), 2.19–2.11 (2H, m); 13C NMR (101 MHz, DMSO-d6) δ 157.4, 157.1, 149.3, 147.6, 140.9, 139.7, 124.1, 122.5, 121.5, 120.8, 120.7, 119.2, 118.4, 116.3, 114.1, 112.4, 110.5, 107.9, 95.2, 70.9, 66.6, 63.7, 55.5, 50.1, 43.1, 33.9; MS (EI) m/z (relative intensity) 446 (100) [M+], 236 (48). HRMS (EI) calcd for C26H26N2O5 [M+]: 446.1842; found: 446.1835.</p><!><p>Yield: 54% from epoxide 117c and amine 104a; 38% overall from 6-fluoro-2-hydroxycarbazole; white solid; mp 157–159 °C; IR (KBr) 3388, 3059, 2924, 2855, 1631, 1503, 1487, 1456, 1249, 1164, 1104, 1035, 819, 747 cm−1; 1H NMR (400 MHz, DMSO-d6) δ 11.10 (s, 1H,), 7.97 (d, J = 7.8 Hz, 1H), 7.81 (dd, J = 9.5, 2.6 Hz, 1H), 7.40 (dd, J = 8.8, 4.4 Hz, 1H), 7.11 (ddd, J = 9.5, 8.8, 2.6 Hz, 1H), 6.99–6.81 (m, 5H), 6.77 (dd, J = 8.6, 2.2 Hz, 1H), 5.08 (br s, 1H), 4.07–3.94 (m, 5H), 3.74 (s, 3H), 2.93 (t, J = 5.5 Hz, 2H), 2.82 (dd, J = 12.0, 4.2 Hz, 1H), 2.71 (dd, J = 11.7, 6.2 Hz, 1H); 13C NMR (101 MHz, DMSO-d6) δ 158.2, 156.4 (d, 1JC F = 230 Hz), 149.1, 148.0, 142.1, 136.1, 123.2 (d, 3JC F = 10 Hz), 121.3, 121.0, 120.6, 115.9 (d, 4JC F = 4.0 Hz), 113.7, 112.2, 111.3 (d, 2JC F = 33 Hz), 111.2, 108.3, 104.9 (d, 2JC F = 24 Hz), 95.2, 70.9, 68.2, 68.1, 55.4, 52.3, 48.3; 19F NMR (376 MHz, DMSO-d6) δ −125.2; MS (EI) m/z (relative intensity) 424 (5) [M+], 368 (10), 201 (34), 180 (100), 56 (34). HRMS (EI) calcd for C24H25N2O4F [M+]: 424.1798; found: 424.1807.</p><!><p>Yield: 55% from epoxide 117d and amine 104a; 33% overall from 6,8-difluoro-2-hydroxycarbazole; white solid; mp 136–138 °C; IR (KBr) 3309, 3176, 2927, 2860, 1652, 1632, 1592, 1505, 1256, 1117, 984, 838, 815, 735 cm-1; 1H NMR (400 MHz, DMSO-d6) δ 11.55 (s, 1H), 7.99 (d, J = 8.6 Hz, 1H), 7.71 (d, J = 7.7 Hz, 1H), 7.17 (t, J = 9.5 Hz, 1H,), 7.05–6.80 (m, 6H), 5.10 (br s, 1H), 4.09–3.92 (m, 5H), 3.73 (s, 3H), 2.92 (t, J = 5.1 Hz, 2H), 2.87–2.66 (m, 2H); 13C NMR (101 MHz, DMSO-d6) δ 158.9, 155.5 (dd, JC F = 233, 9.9 Hz), 149.2, 148.0, 147.3 (dd, JC F = 242, 13.9 Hz), 142.3, 125.7 (dd, JC F = 11.3, 7.2 Hz), 123.8 (d, JC F = 12.6 Hz), 121.8, 121.1, 120.7, 116.0 (dd, JC F = 4.2, 4.1 Hz), 113.7, 112.2, 109.2, 101.2 (dd, JC F = 23, 3.7 Hz), 98.9 (dd, JC F = 29, 21 Hz), 95.4, 71.0, 68.3, 68.2, 55.5, 52.3, 48.4; 19F NMR (376 MHz, DMSO-d6) δ −122.5, −130.4; MS (EI) m/z (relative intensity) 442 (9) [M+], 219 (13), 180 (100). HRMS (EI+) calcd for C24H24F2N2O4 [M+]: 442.1704; found: 442.1715.</p><!><p>Yield: 60% from epoxide 117a and amine 104m; 37% overall from 2-hydroxycarbazole; white solid; mp 141–143 °C; IR (KBr) 3400, 3303, 3247, 3081, 2927, 2839, 1609, 1503, 1462, 1173, 1032, 948, 750, 728 cm−1; 1H NMR (400 MHz, DMSO-d6) 11.07 (s, 1H), 7.97 (d, J = 7.8 Hz, 1H), 7.94 (d, J = 8.6 Hz, 1H), 7.41 (d, J = 8.0 Hz, 1H), 7.27 (td, J = 8.2, 1.2 Hz, 1H), 7.09 (td, J = 7.9, 1.0 Hz, 1H), 6.99–6.92 (m, 2H), 6.88 (dd, J = 10.7, 3.0 Hz, 1H), 6.77 (dd, J = 8.6, 2.2 Hz, 1H), 6.65 (td, J = 8.6, 3.0 Hz, 1H), 5.08 (d, J = 4.0 Hz, 1H), 4.06–3.92 (m, 5H), 3.75 (s, 3H), 2.90 (t, J = 5.5 Hz, 2H), 2.80 (dd, J = 11.8, 4.0 Hz, 1H), 2.69 (dd, J = 11.8, 6.4 Hz, 1H); 13C NMR (101 MHz, DMSO-d6) δ 157.8, 156.8 (d, 1JC F = 235 Hz), 150.2 (d, 3JC F = 10 Hz), 144.4 (d, 4JC F = 2.0 Hz), 141.0, 139.6, 124.0, 122.6, 120.8, 119.1, 118.4, 116.1, 114.4 (d, 3JC F = 10 Hz), 110.5, 108.0, 105.5 (d, 2JC F = 22 Hz), 100.6 (d, 2JC F = 27 Hz), 95.2, 70.9, 69.1, 68.2, 55.8, 52.3, 48.4; 19F NMR (376 MHz, DMSO-d6) δ −120.4; MS (EI) m/z (relative intensity) 424 (28) [M+], 198 (39), 183 (100), 154 (65), 56 (37). HRMS (EI) calcd for C24H25N2O4F [M+]: 424.1798, found: 424.1792.</p><!><p>Yield: 52% from epoxide 117a and amine 104n; 32% overall from 2-hydroxycarbazole; white solid; mp 138–140 °C; IR (film) 3400, 3292, 1607, 1455, 1319, 1131, 1034, 766, 749 cm−1; 1H NMR (400 MHz, DMSO-d6) δ 11.05 (s, 1H), 7.97 (d, J = 7.7 Hz, 1H), 7.94 (d, J = 8.6 Hz, 1H), 7.57–7.65 (m, 2H), 7.40 (d, J = 8.0 Hz, 1H), 7.32–7.23 (m, 2H), 7.14–7.03 (m, 2H), 6.95 (d, J = 2.2 Hz, 1H), 6.76 (dd, J = 8.5, 2.2 Hz, 1H), 5.03 (d, J = 4.6 Hz, 1H), 4.17 (t, J = 5.2 Hz, 2H), 4.07–3.90 (m, 3H), 2.96 (t, J = 5.5 Hz, 2H), 2.86–2.65 (m, 2H); 13C NMR (101 MHz, DMSO-d6) δ 157.8, 156.3, 140.9, 139.6, 134.1, 126.6 (q, 3JC F = 5 Hz), 124.0 (CH), 123.7 (q, 1JC F = 270 Hz), 122.5, 120.7, 120.1, 119.1, 118.4, 117.1 (q, 2JC F = 30 Hz), 116.1, 113.6, 110.5, 108.0, 95.2, 70.8, 68.5, 68.2, 52.2, 48.0; 19F NMR (376 MHz, DMSO-d6) δ −60.7; MS (EI) m/z (relative intensity) 444 (10) [M+], 198 (25), 183 (100). HRMS (EI) calcd for C24H23N2O3F3 [M+]: 444.1661; found: 444.1663.</p><!><p>Yield: 58% from epoxide 117a and amine 104o; 36% overall from 2-hydroxycarbazole; white solid; mp 160–162 °C; IR (KBr) 3397, 3303, 3068, 2924, 2848, 1609, 1509, 1456, 1311, 1286, 1199, 1173, 1108, 738, 722 cm−1; 1H NMR (400 MHz, DMSO-d6) δ 11.05 (s, 1H), 7.97 (d, J = 7.8 Hz, 1H), 7.94 (d, J = 8.6 Hz, 1H), 7.41 (d, J = 8.0 Hz, 1H), 7.27 (ddd, J = 8.2, 7.1, 1.2 Hz, 1H), 7.23–7.06 (m, 4H), 6.96 (d, J = 2.1 Hz, 1H,), 6.95–6.90 (m, 1H), 6.77 (dd, J = 8.5, 2.2 Hz, 1H,), 5.05 (d, J = 4.5 Hz, 1H), 4.11 (t, J = 5.6 Hz, 2H), 4.07–3.90 (m, 3H), 2.95 (t, J = 5.5 Hz, 2H), 2.81 (dd, J = 11.8, 3.9 Hz, 1H), 2.71 (dd, J = 11.8, 6.5 Hz, 1H); 13C NMR (101 MHz, DMSO-d6) δ 157.8, 151.7 (d, 1JC F = 241 Hz), 146.5 (d, 3JC F = 10 Hz), 141.0, 139.6, 124.7 (d, 4JC F = 3.0 Hz), 124.0, 122.6, 121.0 (d, 3JC F = 7.0 Hz), 120.7, 119.1, 118.4, 116.1, 115.9 (d, 2JC F = 18 Hz), 115.0, 110.5, 108.0, 95.2, 70.9, 68.6, 68.2, 52.3, 48.2; 19F NMR (376 MHz, DMSO-d6) δ −134.9; MS (EI) m/z (relative intensity) 394 (20) [M+], 183 (100), 168 (46), 154 (40), 56 (27). HRMS (EI) calcd for C23H23FN2O3 [M+]: 394.1693; found: 394.1679.</p><!><p>Yield: 66% from epoxide 117e and amine 104a; 44% overall from 1-hydroxycarbazole; white solid; mp 58–60 °C; IR (film) 3215, 3057, 2927, 2833, 1576, 1502, 1452, 1250, 738 cm−1; 1H NMR (300 MHz, DMSO-d6) δ 11.17 (s, 1H), 8.06 (d, J = 7.8 Hz, 1H), 7.68 (d, J = 7.7 Hz, 1H), 7.51 (d, J = 8.0 Hz, 1H), 7.36 (dd, J = 7.2, 7.2 Hz, 1H), 7.13 (dd, J = 7.4, 7.3 Hz, 1H), 7.05 (dd, J = 7.8, 7.8 Hz, 1H), 6.99–6.80 (5H, m), 5.07 (d, J = 4.0 Hz, 1H), 4.20–4.00 (m, 5H), 3.73 (s, 3H), 2.95 (t, J = 5.4 Hz, 2H), 2.99–2.73 (m, 2H); 13C NMR (101 MHz, CDCl3) δ 149.1, 148.0, 144.8, 139.4, 129.8, 125.2, 123.5, 122.6, 121.0, 120.7, 120.1, 119.0, 118.4, 113.7, 112.6, 112.2, 111.3, 107.1, 70.7 (2 signals), 68.2, 55.4, 52.1, 48.3; MS (EI) m/z (relative intensity) 405 (45), 223 (18), 182 (55), 181 (63), 179 (98), 153 (100), 123 (34), 77 (30), 56 (63). HRMS (EI) calcd for C24H26N2O4 [M+]: 406.1893; found: 406.1880.</p><!><p>Epoxide 103a (700 mg, 2.93 mmol) was dissolved in dry THF and cooled to 0 °C. Sodium hydride (220 mg, 60% dispersion in oil, 5.5 mmol) was added, the mixture was stirred for 45 min and then warmed to room temperature. Octadecanoyl chloride (1.26 g, 4.16 mmol) in THF was added and the reaction mixture was stirred for 1 h at room temperature, quenched with water and the aqueous phase was exacted with chloroform and ethyl acetate. The combined organic layers were washed with brine and dried over Na2SO4. After removal of the solvent under reduced pressure, the residue was purified by flash chromatography over silica gel (dichloromethane-hexanes) to afford product 1.26 g, (85%) of epoxide 103c.</p><p>Epoxide 103c (200 mg, 0.396 mmol) and amine 104a (132 mg, 0.792 mmol) were reacted in the usual manner to afford 88 mg (33%) of 59 as a white solid; mp 94–95 ºC; IR (KBr) 3451, 2922, 2836, 1705, 1594, 1507, 1441, 1237, 739, 717 cm−1; 1H NMR (300 MHz, CDCl3) δ 8.31 (d, J = 7.4 Hz, 1H), 8.20 (d, J = 8.3 Hz, 1H), 7.80 (d, J = 8.4 Hz, 1H), 7.45–7.31 (m, 3H), 6.95–6.81 (m, 5H), 4.28–4.13 (m, 5H), 3.82 (s, 3H), 3.14–2.95 (m, 6H), 1.97–1.87 (m, 2H), 1.28 (br s, 28H), 0.90 (t, J = 6.6 Hz, 3H); 13C NMR (75 MHz, CDCl3) δ 173.5, 154.6, 149.6, 148.0, 139.7, 137.8, 127.7, 126.2, 125.5, 123.6, 123.1, 121.6, 120.8, 115.7, 115.3, 114.0, 111.7, 109.1, 105.6, 70.5, 68.6, 68.2, 55.7, 51.9, 48.6, 39.2, 31.9, 29.6, 29.6, 29.5, 29.3, 29.2, 24.7, 22.6, 14.1; MS (EI) m/z (relative intensity) 672 (2) [M]+, 549 (10), 183 (60), 180 (100). HRMS (EI) calcd for C42H60N2O5 [M+]: 672.4502; found: 672.4504.</p><!><p>The product was prepared by the acylation of 17 with octadecanoyl chloride, as in the preceding procedure. Yield: 80%; white solid; mp 98–99 °C; IR (KBr) 1704, 1639, 1500, 1257, 1160, 752, 739, 721 cm−1; 1H NMR (300 MHz, CDCl3) δ 8.29 (d, J = 7.7 Hz, 1H), 8.20 (d, J = 8.4 Hz, 1H), 7.90 (d, J = 8.4 Hz, 1H), 7.45 (d, J = 8.9 Hz, 1H), 7.39 (d, J = 8.5 Hz, 1H), 7.28–7.24 (m, 1H), 6.98–6.91 (m, 3H), 6.84 (t, J = 7.3 Hz, 2H), 4.45–4.23 (m, 7H), 3.99–3.80 (m, 4H), 3.68 (s, 3H), 3.14 (t, J = 7.4 Hz, 2H), 1.98–1.88 (d, J = 7.2 Hz, 2H), 1.27 (br s 28H), 0.89 (t, J = 6.6 Hz, 3H); 13C NMR (75 MHz, CDCl3) δ 173.6, 166.8, 154.3, 149.5, 147.9, 140.00, 137.9, 127.8, 126.5, 125.3, 123.7, 123.2, 121.8, 120.9, 115.7, 115.6, 113.8, 111.8, 109.7, 105.6, 72.0, 68.4, 68.0, 67.8, 55.6, 51.0, 46.9, 39.3, 31.9, 29.7 (2 signals), 29.6, 29.5, 29.4, 29.3, 24.7, 22.7, 14.1; MS (CI) m/z 713 (71) [M + H]+, 447 (100). HRMS (EI) calcd for C44H60N2O6: 712.4451; found: 712.4418 [M+].</p><p>Compounds 61, 63, 64 and 66–80 were prepared by alkylation of the corresponding phenols or alcohols with 3-chloro-1,2-epoxypropane or 4-bromo-1,2-epoxybutane, followed by epoxide–opening with amine 104a, as in the preparation of 6.</p><!><p>Yield: 75% from epoxide 103d; 53% overall from the corresponding phenol; white solid; mp 114–116 °C; IR (film) 3388, 3309, 3060, 2930, 2837, 1592, 1505, 1456, 1253, 1123, 1024, 908, 738 cm−1; 1H NMR (400 MHz, CDCl3) δ 7.55 (br s, 1H), 7.12 (d, J = 8.4 Hz, 1H), 6.96–6.84 (m,5H), 6.75 (dd, J = 8.7, 2.5 Hz, 1H), 4.13 (t, J = 5.3 Hz, 2H), 4.13–4.07 (m, 1H), 4.02 (d, J = 5.2 Hz, 2H), 3.82 (s, 3H), 3.08 (t, J = 5.3 Hz, 2H), 2.98 (dd, J = 12.2, 3.9 Hz, 1H), 2.88 (dd, J = 12.2, 7.8 Hz, 1H), 2.69 (t, J = 5.9 Hz, 2H), 2.63 (t, J = 5.8 Hz, 2H), 1.93–1.80 (m, 4H); 13C NMR (101 MHz, CDCl3) δ 153.1, 150.1, 148.4, 135.4, 131.2, 128.5, 122.0, 121.1, 114.7, 112.2, 111.1, 110.3, 102.0, 71.7, 69.0, 68.7, 56.1, 52.0, 48.9, 23.6, 23.5, 23.4, 21.1; MS (EI) m/z (relative intensity) 410 (30) [M+], 187 (100), 180 (25), 158 (24). HRMS (EI) calcd for C24H30N2O4 [M]+: 410.2266; found: 410.2208.</p><!><p>Yield: 40% from epoxide 103e; 53% overall from the corresponding phenol; yellow solid, mp 97–99 °C; IR (KBr) 3326, 1710, 1598, 1505, 1251, 726 cm−1; 1H NMR (400 MHz, CDCl3) δ 7.81 (d, J = 7.4 Hz, 1H), 7.63 (d, J = 7.3 Hz, 1H), 7.42 (t, J = 7.5 Hz, 1H), 7.30 (d, J = 7.2 Hz, 1H), 7.25 (d, J = 8.0 Hz, 1H), 7.21 (d, J = 7.3 Hz, 1H), 7.06 (d, J = 8.2 Hz, 1H), 6.95–6.87 (m, 4H), 4.21–4.15 (m, 6H), 3.83 (s, superimposed on m, 4H), 3.14–3.11 (m, 2H), 3.07 (dd, J = 12.2, 3.2 Hz, 1H), 2.96–2.89 (m, 1H); 13C NMR (101 MHz, CDCl3) δ 194.1, 154.5, 149.6, 148.0, 143.7, 135.8, 134.9, 133.5, 131.4, 130.4, 128.1, 124.2, 124.0, 121.7, 120.9, 118.8, 116.9, 114.0, 111.8, 70.7, 68.7, 68.1, 55.7, 51.7, 48.6; MS (CI) m/z (relative intensity) 420 (100) [M + H]+. HRMS (CI) calcd for C25H26NO5 [M + H]+: 420.1811; found: 420.1815.</p><!><p>Yield: 56% from epoxide 103f; 42% overall from the corresponding phenol; white solid; mp 110–112 °C; IR (nujol) 3257, 1461, 1371, 1247, 1171, 738 cm−1; 1H NMR (400 MHz, DMSO-d6) δ 8.12 (dd, J = 7.8, 1.9 Hz, 1H), 7.73 (d, J = 2.5 Hz, 1H), 7.65 (dd, J = 8.2, 0.8 Hz, 1H), 7.58 (d, J = 8.9 Hz, 1H), 7.49 (ddd, J = 8.6, 8.4, 1.4 Hz, 1H), 7.36 (ddd, J = 7.7, 7.5, 1.0 Hz, 1H), 7.10 (dd, J = 8.9, 2.7 Hz, 1H), 6.99–6.93 (m, 2H), 6.92–6.82 (m, 2H), 5.09 (d, J = 4.4 Hz, 1H), 4.09–3.93 (m, 5H), 3.73 (s, 3H), 2.93 (t, J = 5.5 Hz, 2H), 2.82 (dd, J = 11.9, 4.3 Hz, 1H), 2.72 (dd, J = 11.9, 6.7 Hz, 1H); 13C NMR (101 MHz, DMSO-d6) δ 156.1, 155.1, 150.0, 149.2, 148.1, 127.4, 124.1, 123.9, 122.7, 121.2, 121.0, 120.7, 115.9, 113.7, 112.2, 112.1, 111.6, 105.1, 71.5, 68.4, 68.3, 55.4, 52.4, 48.5; MS (EI) m/z (relative intensity) 284 (6), 270 (7), 210 (8), 180 (100). HRMS (EI) calcd for C24H25NO5 [M+]: 407.1733; found: 407.1733.</p><!><p>Yield: 63% from epoxide 103g; 33% overall from the corresponding phenol; off-white solid; mp 48–50 °C; IR (KBr) 3388, 3059, 2924, 2855, 1631, 1503, 1487, 1456, 1249, 1164, 1104, 1035, 819, 747 cm−1; 1H NMR (400 MHz, DMSO-d6) δ 8.54 (s, 1H), 7.02–6.82 (m, 5H), 6.79 (d, J = 8.4 Hz, 1H), 6.74 (ddd, J = 7.5, 7.5, 1.0 Hz, 1H), 6.67 (dd, J = 7.9, 0.8 Hz, 1H), 6.38 (dd, J = 8.4, 2.5 Hz, 1H), 6.34 (d, J = 2.5 Hz, 1H), 5.03 (br s, 1H), 4.00 (t, J = 5.5 Hz, 2H), 3.93–3.75 (m,3H), 3.73 (s, 3H), 2.89 (t, J = 5.5 Hz, 2H), 2.73 (dd, J = 11.8, 4.1 Hz, 1H), 2.64 (dd, J = 11.8, 6.2 Hz, 1H); 13C NMR (101 MHz, DMSO-d6) δ 158.6, 149.1, 148.0, 143.2, 141.7, 127.2, 126.7, 126.1, 121.6, 121.0, 120.6, 116.9, 114.3, 113.6, 112.2, 107.8, 107.0, 101.2, 70.6, 68.3, 68.0, 55.4, 52.2, 48.4; MS (EI) m/z (relative intensity) 438 (100) [M+], 224 (22), 215 (82). HRMS (EI) calcd for C24H26N2O4S [M+]: 438.1613; found: 438.1600.</p><!><p>Yield: 54% from epoxide 103h; 28% overall from the corresponding phenol; white solid; mp 67–69 °C; IR (film) 3518, 3309, 3183, 3076, 2930, 2834, 2249, 1622, 1260, 1024, 911, 725 cm−1; 1H NMR (400 MHz, CDCl3) δ 8.70 (s, 1H), 7.94 (dd, J = 8.0, 1.1 Hz, 1H), 7.77 (d, J = 8.9 Hz, 1H), 7.31 (t, J = 8.4 Hz, 1H), 7.07 (t, J = 7.4 Hz, 1H), 6.96 (d, J = 8.3 Hz, 1H), 6.94–6.81 (m, 4H), 6.57 (dd, J = 8.9, 2.1 Hz, 1H), 6.34 (d, J = 2.0 Hz, 1H), 4.09 (t, J = 4.9 Hz, 2H), 4.07–3.98 (m, 1H), 3.90–3.78 (m, 2H), 3.76 (s, 3H), 3.27 (br s, 2H), 3.03 (t, J = 4.9 Hz, 2H), 2.87 (dd, J = 12.3, 4.0 Hz, 1H), 2.76 (dd, J = 12.3, 7.8 Hz, 1H); 13C NMR (101 MHz, CDCl3) δ 162.2, 149.6, 148.1, 139.9, 138.1, 133.0, 124.9, 123.0, 122.0, 121.8, 121.6, 121.3, 116.8, 114.3, 114.2, 112.2, 110.6, 100.3, 70.8, 68.5, 68.1, 56.0, 51.5, 48.8; MS (EI) m/z (relative intensity) 470 (3) [M+], 347 (10), 333 (13), 247 (13), 180 (100), 178 (25), 148 (12), 56 (20). HRMS (EI) calcd for C24H26N2O6S [M+]: 470.1512; found: 470.1490.</p><!><p>Yield: 38% from epoxide 103i; 32% overall from the corresponding phenol; white solid; mp 110–112 °C; IR (KBr) 3435, 1504, 1256, 782, 773, 743 cm−1; 1H NMR (300 MHz, DMSO-d6) δ 8.24 (d, J = 7.5 Hz, 1H), 7.85 (d, J = 7.5 Hz, 1H), 7.35–7.55 (m, 4H), 6.80–7.04 (m, 5H), 5.16 (d, J = 3.9 Hz, 1H), 3.97–4.19 (m, 5H), 3.72 (s, 3H), 2.92 (t, J = 5.5 Hz, 2H), 2.87 (dd, J = 11.7, 3.7 Hz, 1H), 2.77 (dd, J = 12.0, 6.3 Hz, 1H), 1.98 (br s, 1H); 13C NMR (75 MHz, DMSO-d6) δ 154.2, 149.2, 148.1, 134.0, 127.4, 126.4, 126.2, 125.1, 125.0, 121.8, 121.0, 120.7, 119.8, 113.6, 112.2, 105.1, 70.9, 68.4, 68.4, 55.4, 52.5, 48.5; MS (CI) m/z (relative intensity) 368 (100) [M + H]+. HRMS (EI) calcd for C22H25NO4 [M+]: 367.1784; found: 367.1797.</p><!><p>Yield: 51% from 103u; 12% overall from the corresponding phenol; white solid; mp 65–67 ºC IR (KBr) 3310, 2917, 1591, 1578, 1504, 1454, 1391, 1252, 1127, 1098, 770 cm−1; 1H NMR (400 MHz, CDCl3) δ 8.27 (ddd, J = 6.7, 1.7, 0.7 Hz, 1H), 7.82–7.80 (m, 1H), 7.50–7.46 (m, 2H), 7.44 (d, J = 8.3 Hz, 1H), 7.38 (t, J = 8.3 Hz, 1H), 6.97–6.84 (m, 5H), 4.38–4.29 (m, 2H), 4.12 (t, J = 5.2 Hz, 2H), 4.07–4.01 (m, 1H), 3.85 (s, 3H), 3.12–3.02 (m, 2H), 2.94 (dd, J = 12.1, 3.2 Hz, 1H), 2.68 (dd, J = 12.1, 9.2 Hz, 1H), 2.14–1.99 (m, 2H); 13C NMR (101 MHz, CDCl3) δ 154.5, 149.7, 148.2, 134.4, 127.4, 126.3, 125.9, 125.6, 125.1, 121.9, 121.6, 120.9, 120.1, 114.1, 111.8, 104.70, 68.7, 66.9, 65.00, 55.7, 55.1, 48.4, 34.5; MS (EI) m/z (relative intensity) 381 (<1) [M+] 238 (100), 180 (74). HRMS (EI) calcd for C23H27NO4 [M+]: 381.1940; found: 381.1941.</p><!><p>Yield: 50% from epoxide 103j; 13% overall from 1-adamantanol; viscous oil; IR (KBr) 3428, 2904, 1498, 1252, 1119, 742 cm−1; 1H NMR (300 MHz, CDCl3) δ 6.93–6.87 (m, 4H), 4.12 (t, J = 5.3 Hz, 2H), 3.85 (s, 3H), 3.85–3.77 m, 1H), 3.47–3.38 (m, 2H), 3.05 (t, J = 5.3 Hz, 2H), 2.82 (dd, J = 12.0, 4.0 Hz, 1H), 2.72 (dd, J = 12.1, 7.6 Hz, 1H), 2.60 (br s, 1H), 2.09–2.17 (m, 3H), 1.69–1.78 (m, 6H), 1.54–1.68 (m, 6H); 13C NMR (75 MHz, CDCl3) δ 149.6, 148.2, 121.4, 120.8, 114.0, 111.8, 72.2, 69.3, 68.7, 62.5, 55.8, 52.1, 48.7, 41.4, 36.3, 30.4; MS (CI) m/z (relative intensity) 376 (100) [M + H]+. HRMS (CI) calcd for C22H34NO4 [M + H]+: 376.2488; found: 376.2475.</p><!><p>Yield: 50% from epoxide 103k; 32% overall from 2-adamantanol; white solid; mp 53–57 ºC; IR (KBr) 3334, 2896, 1509, 1256, 740 cm−1; 1H NMR (300 MHz, CDCl3) δ 6.97–6.86 (m, 4H), 4.13 (t, J = 5.4 Hz, 2H), 3.86 (s, 3H), 3.92–3.84 (m, 1H), 3.52–3.42 (m, 3H), 3.07 (t, J = 5.4 Hz, 2H), 2.85 (dd, J = 12.1, 4.2 Hz, 1H), 2.77 (dd, J = 12.1, 7.5 Hz, 1H), 1.96–2.07 (m, 4H), 1.73–1.88 (m, 4H), 1.70–1.62 (m, 4H), 1.43–1.52 (m, 2H); 13C NMR (75 MHz, CDCl3) δ 149.7, 148.2, 121.5, 120.8, 114.1, 111.8, 82.0, 70.00, 69.1, 68.8, 55.8, 52.2, 48.7, 37.5, 36.4, 31.63, 31.59, 31.5, 27.3; MS (CI) m/z (relative intensity) 376 (100) [M + H]+. HRMS (CI) calcd for C22H34NO4 [M + H]+: 376.2488; found: 376.2501.</p><!><p>Yield: 87% overall from (1-adamantyl)methanol via epoxide 103l; viscous oil; IR (KBr)) 3211, 1596, 1509, 1452, 1254, 1125, 1029, 744 cm−1; 1H NMR (300 MHz, CDCl3) δ 6.96–6.88 (m, 4H), 4.12 (t, J = 5.3 Hz, 2H), 3.86 (s, 3H), 3.90–3.83 (m, 1H), 3.37–3.47 (m, 2H), 3.05 (t, J = 5.4 Hz, 2H), 2.97–3.07 (m, 2H), 2.82 (dd, J = 12.1, 4.1 Hz, 1H), 2.73 (dd, J = 12.1, 7.7 Hz, 1H), 1.91–2.02 (m, 3H), 1.58–1.78 (m, 6H), 1.49–1.56 (m, 6H); 13C NMR (75 MHz, CDCl3) δ 149.7, 148.1, 121.6, 120.9, 114.3, 111.8, 82.5, 74.0, 68.6, 68.5, 55.8, 52.0, 48.6, 39.6, 37.1, 34.1, 28.2; MS (CI) m/z (relative intensity) 390 (100) [M + H]+. HRMS (CI) calcd for C23H36NO4 [M + H]+: 390.2644; found: 390.2631.</p><!><p>Yield: 40% from epoxide 103m; 9% overall from the corresponding phenol; white solid; mp 154–155 °C; IR (film) 3283, 2909, 1654, 1502, 1249, 1119 cm−1; 1H NMR (400 MHz, DMSO-d6) δ 11.60 (br s, 1H), 7.82 (d, J = 9.6 Hz, 1H), 7.10–7.25 (m, 3H), 6.80–6.99 (m, 4H), 6.47 (d, J = 9.5 Hz, 1H,), 5.05 (br s, 1H), 3.87–4.05 (m, 5H), 3.73 (s, 3H), 2.91 (t, J = 5.4 Hz, 2H), 2.78 (dd, J = 11.8, 3.6 Hz, 1H), 2.68 (dd, J = 11.7, 6.0 Hz, 1H,); 13C NMR (101 MHz, DMSO-d6) δ 161.4, 153.5, 149.2, 148.1, 139.7, 133.3, 122.2, 121.0, 120.7, 119.9, 119.6, 116.3, 113.7, 112.3, 110.2, 71.1, 68.3, 68.1, 55.5, 52.3, 48.4; MS (ESI) m/z (relative intensity), 385 (100) [M + H]+. HRMS (ESI) calcd for C21H25N2O5 [M + H]+: 385.1758; found: 385.1754.</p><!><p>Yield: 29% from epoxide 103n; 10% overall from the corresponding phenol; white solid; mp 126–128 °C; IR (film) 3302, 3167, 2920, 2830, 1625, 1593, 1503, 1450, 1252, 1121, 1022, 743 cm−1; 1H NMR (400 MHz, DMSO-d6) δ 9.87 (br s, 1H), 6.96 (d, J = 7.6, 2.1 Hz, 2H), 6.93–6.83 (m, 2H), 6.81–6.68 (m, 3H), 4.98 (br s, 1H), 4.00 (t, J = 5.6 Hz, 2H), 3.92–3.80 (m, 3H), 3.74 (s, 3H), 2.90 (t, J = 5.5 Hz, 2H), 2.82 (t, J = 7.5 Hz, 2H), 2.75 (dd, J = 12.0, 3.7 Hz, 1H), 2.65 (dd, J = 11.8, 6.1 Hz, 1H), 2.39 (dd, J = 8.3, 6.8 Hz, 2H); 13C NMR (101 MHz, DMSO-d6) δ 169.7, 153.8, 149.2, 148.1, 131.7, 124.8, 121.0, 120.7, 115.7, 114.0, 113.7, 113.0, 112.3, 71.0, 68.3, 68.1, 55.5, 52.3, 48.4, 30.3, 25.1; MS (ESI) m/z (relative intensity) 387 (100) [M + H]+. HRMS (ESI) m/z (relative intensity) calcd for C21H27N2O5 [M + H]+: 387.1914; found: 387.1910.</p><!><p>Yield: 65% from epoxide 103o; 13% overall from the corresponding phenol; white solid; mp 161–163 °C; IR (film) 3584, 3183, 2914, 2831, 1678, 1588, 1223, 1123, 1027, 738 cm−1; 1H NMR (400 MHz, DMSO-d6) δ 9.96 (br s, 1H), 7.14–6.78 (m, 5H), 6.62–6.37 (m, 2H), 5.03 (br s, 1H), 4.01 (t, J = 5.3 Hz, 2H), 3.95–3.76 (m, 3H), 3.74 (s, 3H), 2.91 (t, J = 5.3 Hz, 2H), 2.78 (t, J = 7.3 Hz, 2H), 2.79–2.71 (m, 1H), 2.66 (dd, J = 11.7, 6.2 Hz, 1H), 2.41 (t, J = 7.4 Hz, 2H); 13C NMR (101 MHz, DMSO-d6) δ 170.2, 157.9, 149.2, 148.0, 139.1, 128.3, 121.1, 120.7, 115.5, 113.8, 112.3, 107.5, 101.7, 70.6, 68.3, 68.0, 55.5, 52.2, 48.4, 30.7, 24.0; MS (ESI) m/z (relative intensity) 387 (100) [M + H]+; HRMS (ESI) calcd for C21H27N2O5 [M + H]+: 387.1914; found: 387.1909.</p><!><p>Yield: 63% from epoxide 103p; 43% overall from 4-hydroxyindole; white solid; mp 98–100 °C; IR (film) 3378, 3066, 2924, 2868, 2834, 1586, 1506, 1453, 1249, 1122, 1091, 1051, 1027, 909. 742 cm−1; 1H NMR (400 MHz, CDCl3) δ 8.20 (br s, 1H), 7.11–7.00 (m, 3H), 6.97–6.83 (m, 4H), 6.65–6.59 (m, 1H,), 6.51 (dd, J = 7.6, 0.7 Hz, 1H), 4.24–4.08 (m, 5H), 3.81 (s, 3H), 3.20–3.10 (br s, 1H), 3.10 (t, J = 5.2 Hz, 2H,), 3.02 (dd, J = 12.4, 3.5 Hz, 1H), 2.93 (dd, J = 12.4, 7.5 Hz, 1H), 2.96–2.92 (m, 1H); 13C NMR (101 MHz, CDCl3) δ 152.6, 150.1, 148.4, 137.6, 123.0, 122.9, 122.0, 121.2, 119.0, 114.7, 112.2, 105.0, 101.1, 100.1, 70.6, 69.0, 68.5, 56.1, 52.0, 48.9; MS (EI) m/z (relative intensity) 356 (25) [M+], 180 (100), 133 (35). HRMS (EI) calcd for C20H24N2O4 [M+]: 356.1736; found: 356.1727.</p><!><p>Yield: 58% from epoxide 103q; 44% overall from 5-hydroxyindole; white solid; mp 109–110 °C; IR (film) 3375, 2930, 2867, 1592, 1505, 1456, 1253, 1220, 1157, 1123, 1027, 748, 727 cm−1; 1H NMR (400 MHz, DMSO-d6) δ 10.89 (s, 1H), 7.28–7.24 (m, 2H), 7.03 (d, J = 2.4 Hz, 1H), 6.98–6.93 (m, 2H), 6.92–6.82 (m, 2H), 6.73 (dd, J = 8.7, 2.4 Hz, 1H), 6.32–6.30 (m, 1H), 5.00 (br s, 1H), 4.01 (t, J = 5.6 Hz, 2H), 3.95–3.85 (m, 3H), 3.73 (s, 3H), 3.32 (br s, 1H), 2.92 (t, J = 5.6 Hz, 2H), 2.80 (dd, J = 11.9, 3.9 Hz, 1H), 2.68 (dd, J = 11.9, 6.4 Hz, 1H); 13C NMR (101 MHz, DMSO-d6) δ 152.5, 149.2, 148.0, 131.0, 127.9, 125.7, 121.0, 120.7, 113.7, 112.2, 111.8, 111.6, 102.8, 100.8, 71.3, 68.3, 68.2, 55.4, 52.5, 48.4; MS (ESI) m/z (relative intensity) 357 (100 %) [M + H]+. HRMS (ESI) calcd for C20H24N2O4 [M + H]+: 357.1809; found: 357.1819.</p><!><p>The reaction of the N-t-Boc-protected epoxide derivative 103r with amine 104a was carried out as in the preparation of 6, followed by deprotection with TFA in dichloromethane at room temperature, to afford 78 in 43% overall yield; white solid; mp 114–116 °C; IR (film) 3000–3600 (br), 3062, 2924, 2848, 1591, 1500, 1450, 1249, 1221, 1123, 1026, 948, 744 cm−1; 1H NMR (400 MHz, CDCl3) δ 7.96 (br s, 1H), 7.37 (d, J = 8.8 Hz, 1H), 7.15–7.04 (m, 2H), 6.84–7.02 (m, 4H), 4.17 (t, J = 5.1 Hz, 2H), 4.20–4.11 (m, 1H), 4.04 (d, J = 5.1 Hz, 2H), 3.84 (s, 3H), 3.13 (t, J = 5.0 Hz, 2H), 3.03 (dd, J = 12.3, 3.7 Hz, 1H), 2.91 (dd, J = 12.2, 8.0 Hz, 2H); 13C NMR (101 MHz, CDCl3) δ 153.9, 150.0, 148.3, 135.9, 134.6, 123.5, 122.1, 121.1, 119.5, 114.8, 112.2, 110.8, 101.6, 71.2, 68.9, 68.2, 56.0, 51.8, 48.8; MS (ESI) m/z (relative intensity) 358 (100%) [M + H]+. HRMS (ESI) calcd for C19H24N3O4 [M + H]+: 358.1761; found: 358.1756.</p><p>Products 79 and 80 were prepared similarly.</p><!><p>Overall yield from 103s: 47%; white solid; mp 131–133 °C; IR (film) 3300, 3018, 2933, 2870, 1628, 1503, 1456, 1252, 1123, 1026, 941, 741 cm−1; 1H NMR (400 MHz, CDCl3) δ 7.94 (br s, 1H), 7.56 (d, J = 8.8 Hz, 1H), 7.02–6.77 (m, 6H), 4.30–4.22 (m, 1H), 4.20 (t, J = 5.1 Hz, 2H), 4.10–4.01 (m, 2H), 3.83 (s, 3H), 3.20 (dd, J = 8.1, 4.8 Hz, 2H), 3.14 (dd, J = 12.3, 3.9 Hz, 1H), 3.01 (dd, J = 12.3, 8.2 Hz, 2H); 13C NMR (101 MHz, CDCl3) δ 158.6, 149.6, 147.9, 141.3, 134.2, 122.1, 121.6, 121.2, 118.1, 114.5, 113.4, 112.2, 91.9, 70.6, 68.0, 67.7, 55.9, 51.7, 48.4; MS (ESI) m/z (relative intensity) 358 (100) [M+]. HRMS (ESI) calcd for C19H24N3O4 [M + H]+: 358.1761; found: 358.1752.</p><!><p>Overall yield from 103t: 20%; white solid; mp 125–127 °C; IR (film) 3302, 3167, 2920, 2830, 1625, 1593, 1503, 1450, 1252, 1121, 1022, 743 cm−1; 1H NMR (400 MHz, CD3OD) δ 8.05 (s, 1H), 7.48 (d, J = 8.8 Hz, 1H), 7.13 (s, 1H), 7.03–6.84 (m, 5H), 6.51 (dd, J = 7.6, 0.7 Hz, 1H), 4.18–4.01 (m, 5H), 3.80 (s, 3H), 3.03 (t, J = 5.0 Hz, 2H), 2.96 (dd, J = 12.2, 4.0 Hz, 1H), 2.85 (dd, J = 12.2, 8.0 Hz, 1H), 2.96–2.92 (m, 1H); 13C NMR (101 MHz, CDCl3) δ 154.3, 148.3, 146.7, 139.2, 120.1, 119.2, 119.0, 112.9, 111.2, 110.5, 100.0, 69.6, 67.0, 66.8, 53.5, 50.3, 46.7; MS (EI) m/z (relative intensity) 179 (100), 133 (35). HRMS (ESI) calcd for C19H24N3O4 [M + H]+: 358.1761; found: 358.1760.</p><!><p>The product was prepared by the Wolff-Kishner reduction of 63 in 54% yield; white solid; mp 104.5–106.0 ºC; IR (KBr) 3290, 2922, 1582, 1503, 1457, 1274, 1253, 1221, 738 cm−1; 1H NMR (300 MHz, CDCl3) δ 8.14 (d, J = 7.5 Hz, 1H), 7.53 (d, J = 7.3 Hz, 1H), 7.36 (t, J = 7.1 Hz, 1H), 7.30–7.28 (m, 1H), 7.24 (d, J = 7.8 Hz, 1H), 7.18 (d, J = 7.3 Hz, 1H), 6.98–6.88 (m, 5H), 4.30–4.14 (m, 5H), 3.92 (s, 2H), 3.84 (s, 3H), 3.14–3.07 (m, 3H), 2.97 (dd, J = 7.7, 12.3 Hz, 1H); 13C NMR (75 MHz, CDCl3) δ 154.9, 149.7, 148.2, 145.3, 142.5, 140.8, 129.9, 127.6, 126.7, 125.8, 124.3, 123.6, 121.6, 120.9, 117.6, 114.1, 111.8, 109.6, 70.4, 68.8, 68.4, 55.7, 51.9, 48.6, 37.2; MS (EI) m/z (relative intensity) 405 (3) [M+], 268 (25), 180 (100). HRMS (EI) calcd for C25H27NO4 [M+]: 405.1940; found: 405.1945.</p><!><p>A mixture of 66 (88 mg, 0.20 mmol) and NiCl2·6H2O (337 mg, 1.41 mmol) in methanol-THF-H2O (8 mL; 1:2:1) at room temperature was treated with small portions of NaBH4 (162 mg, 4.28 mmol) over 1 h.34 Stirring was continued for an additional 2 h, the mixture was filtered through a pad of Celite and the filtrate was concentrated under vacuum. The residue was purified by flash chromatography over silica gel (methanol-dichloromethane) to afford 29 mg (36%) of 65 as a yellow oil; IR (film) 3355, 3279, 3061, 2928, 2835, 1588, 1495, 1251, 745 cm−1; 1H NMR (400 MHz, CDCl3) δ 7.28–7.21 (m, 3H), 7.11 (t, J = 8.0 Hz, 1H), 7.08–7.02 (m, 2H), 6.97–6.83 (m, 5H), 6.66–6.58 (m, 2H), 6.44 (dd, J = 8.2, 2.0 Hz, 1H), 4.14 (t, J = 5.1 Hz, 2H), 4.15–4.07 (m, 1H), 4.00–3.92 (m, 2H), 3.81 (s, 3H), 3.10 (t, J = 5.1 Hz, 2H), 3.00 (dd, J = 12.3, 3.7 Hz, 1H), 2.88 (dd, J = 12.3, 8.0 Hz, 1H); 13C NMR (101 MHz, CDCl3) δ 159.9, 150.0, 148.2, 144.9, 142.9, 130.3, 129.5, 122.2, 121.5, 121.2, 118.6, 115.0, 112.2, 110.6, 106.9, 104.1, 70.4, 68.7, 68.0, 56.0, 51.7, 48.7; MS (EI) m/z (relative intensity) 408 (40) [M+], 368 (19), 285 (19), 186 (19), 185 (100), 180 (73). HRMS (EI) calcd for C24H28N2O4 [M+]: 408.2049; found: 408.2030.</p><!><p>A mixture of 117a (304 mg, 1.27 mmol) and benzylamine (0.42 mL, 3.84 mmol) were refluxed for 2 h in isopropanol (3 mL). The solvent was removed under vacuum and the residue was purified by flash chromatography over silica gel (methanol-dichloromethane) to afford 211 mg (48%) of amino alcohol 119 as an off-white solid; mp 160–161 °C; IR (film) 3392, 3043, 2919, 2847, 1607, 1455, 749, 724 cm−1; 1H NMR (400 MHz, DMSO-d6) δ 11.06 (s, 1H), 7.97 (d, J = 8.2 Hz, 1H), 7.94 (d, J = 9.2 Hz, 1H), 7.41 (d, J = 8.0 Hz, 1H), 7.38–7.20 (m, 6H), 7.09 (dd, J = 7.7, 7.2 Hz, 1H), 6.96 (s, 1H), 6.76 (d, J = 8.6 Hz, 1H), 4.09–3.93 (m, 3H), 3.77 (s, 2H), 2.73 (dd, J = 11.7, 3.5 Hz, 1H), 2.65 (dd, J = 11.6, 5.9 Hz, 1H); 13C NMR (101 MHz, d6-DMSO) δ 157.8, 141.0, 140.3, 139.7, 128.1, 128.0, 126.6, 124.1, 122.6, 120.8, 119.2, 118.5, 116.2, 110.5, 108.1, 95.3, 70.9, 68.1, 52.9, 51.6; MS (EI) m/z (relative intensity) 346 (32) [M+], 183 (100), 154 (20), 120 (18), 91 (48). HRMS (EI) calcd for C22H22N2O2 [M+]: 346.1681; found: 346.1686.</p><p>A mixture of 119 (188 mg, 0.542 mmol), aldehyde 121 (85 mg, 0.58 mmol) and NaBH(OAc)3 (168 mg, 0.792 mmol) in 1,2-dichloroethane (5 mL) was stirred at room temperature for 4 h. The mixture was partitioned between water and dichlormethane and the aqueous phase was extracted with dichloromethane. The combined organic extracts were washed with brine, dried over MgSO4 and concentrated under vacuum. The residue was purified by flash chromatography over silica gel (methanol-dichloromethane) to furnish 196 mg (76%) of the N-benzyl derivative of 81 as a white solid foam; 1H NMR (400 MHz, CDCl3) δ 7.94 (d, J = 7.7 Hz, 1H), 7.88 (d, J = 8.4 Hz, 1H), 7.87 (br s, 1H), 7.54–7.50 (m, 1H), 7.48–7.44 (m, 1H), 7.40–7.31 (m, 6H), 7.29–7.16 (m, 4H), 6.84–6.76 (m, 2H), 6.58 (s, 1H), 4.14–4.23 (m, 1H), 4.03 (d, J = 5.1 Hz, 2H), 3.90 (d, J = 15.1 Hz, 1H), 3.89 (d, J = 13.5 Hz, 1H), 3.84 (d, J = 15.1 Hz, 1H), 3.70 (d, J = 13.5 Hz, 1H), 3.31 (s, 1H), 2.77–2.92 (m, 2H); 13C NMR (101 MHz, CDCl3) δ 158.3, 155.3, 155.1, 140.9, 139.7, 138.4, 129.3, 128.7, 128.4, 127.6, 124.8, 124.2, 123.6, 122.9, 121.1, 121.0, 119.8, 119.7, 117.6, 111.4, 110.5, 108.8, 106.0, 95.8, 70.9, 67.0, 58.7, 56.5, 50.7.</p><p>A solution of the above N-benzyl derivative (50 mg, 0.10 mmol) in a 2:1 mixture of methanol-dichloromethane (1.5 mL) was added to 10% palladium on charcoal (27 mg). The mixture was heated at 45 °C for 4 h under positive pressure of hydrogen (balloon). The reaction was cooled to room temperature and filtered through Celite. The filtrate was concentrated under vacuum and the residue was purified by flash chromatography over silica gel (methanol-dichloromethane) to provide 27 mg (68%) of 81 as a white solid; mp 159–160 °C; IR (film) 3355, 3279, 3061, 2928, 2835, 1588, 1495, 1251, 745 cm−1; 1H NMR (400 MHz, DMSO-d6) δ 11.04 (s, 1H), 7.97 (d, J = 7.6 Hz, 1H), 7.94 (d, J = 8.6 Hz, 1H), 7.56 (d, J = 8.4 Hz, 1H), 7.51 (d, J = 8.0 Hz, 1H), 7.41 (d, J = 8.0 Hz, 1H), 7.31–7.17 (m, 3H), 7.09 (dd, J = 7.7, 7.2 Hz, 1H), 6.96 (d, J = 2.2 Hz, 1H), 6.76 (dd, J = 8.5, 2.2 Hz, 1H), 6.73 (s, 1H), 5.03 (d, J = 4.5 Hz, 1H), 3.92–4.09 (m, 2H), 3.91 (s, 3H), 2.79 (dd, J = 11.8, 4.3 Hz, 1H), 2.70 (dd, J = 11.8, 6.0 Hz, 1H); 13C NMR (101 MHz, DMSO-d6) δ 157.8, 157.7, 154.1, 140.9, 139.6, 128.1, 124.0, 123.5, 122.6, 122.5, 120.7, 120.6, 119.1, 118.4, 116.1, 110.7, 110.4, 108.0, 103.1, 95.2, 70.8, 68.2, 51.7, 46.0; MS (EI) m/z (relative intensity) 386 (74) [M+], 183 (100), 131 (82), 43 (54). HRMS (EI) calcd for C24H22N2O3 [M+]: 386.1630; found: 386.1626.</p><!><p>A mixture of amino alcohol 119 (88 mg, 0.25 mmol), chloride 122 (47 mg, 0.28 mmol), N,N-diisopropylethylamine (71 μL, 0.41 mmol) and a catalytic amount of KI in acetonitrile (2 mL) was stirred at 60 °C for 16 h. The reaction was cooled to room temperature and partitioned between water and ethyl acetate. The organic phase was separated and extracted with ethyl acetate. The combined extracts were washed with brine, dried over Na2SO4 and concentrated under vacuum. The residue was purified by flash chromatography over silica gel (methanol-dichloromethane) to provide 101 mg (85%) of the N-benzyl derivative of 82 as an off-white solid; mp 77–79 °C; 1H NMR (400 MHz, DMSO-d6) δ 11.07 (s, 1H), 7.98 (d, J = 7.8 Hz, 1H), 7.92 (d, J = 8.5 Hz, 1H), 7.79–7.73 (m, 1H), 7.72–7.66 (m, 1H), 7.46–7.21 (m, 9H), 7.15–7.06 (m, 1H), 6.91 (d, J = 2.1 Hz, 1H), 6.67 (dd, J = 8.5, 2.2 Hz, 1H), 5.01 (d, J = 4.8 Hz, 1H), 4.13–4.01 (m, 4H), 3.94–3.89 (m, 1H), 3.90 (d, J = 16.0 Hz, 1H), 3.83 (d, J = 16.0 Hz, 1H), 2.85 (dd, J = 13.3, 5.9 Hz, 1H), 2.77 (dd, J = 13.3, 5.8 Hz, 1H); 13C NMR (101 MHz, DMSO-d6) δ 164.0, 157.8, 150.2, 141.0, 140.6, 139.7, 138.7, 128.8, 128.2, 127.0, 125.0, 124.3, 124.0, 122.6, 120.7, 119.6, 119.2, 118.5, 116.1, 110.7, 110.5, 108.0, 95.2, 70.9, 67.5, 58.4, 56.3, 50.5; MS (EI) m/z (relative intensity) 477 (7) [M+], 345 (21), 252 (20), 251 (100), 91 (79). HRMS (EI) calcd for C30H27N3O3 [M+]: 477.2052; found: 477.2068.</p><p>A solution of the N-benzyl derivative of 82 (79 mg, 0.17 mmol) in methanol (1.5 mL) was added to 10% palladium on charcoal (21 mg). The mixture was hydrogenated as in the preceding procedure for 48 h. The reaction was cooled to room temperature and filtered through Celite. The filtrate was concentrated under vacuum and the residue was purified by flash chromatography over silica gel (methanol-dichloromethane) to obtain 22 mg (34%) of 82 as a beige solid; mp 136–138 °C; IR (film) 3355, 3279, 3061, 2928, 2835, 1588, 1495, 1251, 745 cm−1; 1H NMR (400 MHz, DMSO-d6) δ 11.05 (s, 1H), 7.97 (d, J = 7.7 Hz, 1H), 7.93 (d, J = 8.5 Hz, 1H), 7.75–7.65 (m, 2H), 7.43–7.32 (m, 3H), 7.27 (t, J = 7.2 Hz, 1H), 7.10 (t, J = 7.3 Hz, 1H), 6.96 (d, J = 2.0 Hz, 1H), 6.75 (dd, J = 8.5, 2.1 Hz, 1H), 5.07 (d, J = 4.2 Hz, 1H), 4.09–3.93 (m, 3H), 4.07 (s, 2H), 2.83 (dd, J = 12.0, 3.7 Hz, 1H), 2.75 (dd, J = 11.5, 5.6 Hz, 1H); 13C NMR (101 MHz, DMSO-d6) δ 165.8, 157.8, 150.2, 141.0, 140.7, 139.7, 124.9, 124.3, 124.1, 122.6, 120.8, 119.5, 119.2, 118.5, 116.2, 110.7, 110.5, 108.1, 70.9, 68.4, 51.9, 46.2; MS (EI) m/z (relative intensity) 387 (82) [M+], 183 (100), 161 (25), 133 (32), 132 (39). HRMS (EI) calcd for C23H21N3O3 [M+]: 387.1583; found: 387.1587.</p><p>Compounds 83–90 and 92 were prepared by the treatment of the corresponding epoxides 103a, 105, 114a, 117a or 117c with the appropriate amines, by means of the same procedure as for the preparation of 6, except for the variation described below for 84.</p><!><p>From 103a and 118a.Yield: 67%; off-white solid; mp 68–70 °C; IR (film) 3398, 3056, 2930, 2870, 1605, 1502, 1452, 1343, 1097, 722 cm−1; 1H NMR (300 MHz, CDCl3) δ 8.30 (d, J = 7.8 Hz, 1H), 8.08 (s, 1H), 7.47–7.37 (m, 2H), 7.33 (dd, J = 8.0, 7.9 Hz, 1H), 7.30–7.21 (m, 1H), 7.06 (d, J = 8.0 Hz, 1H), 6.90–6.79 (m, 3H), 6.75–6.61 (m, 2H), 4.57–4.45 (m, 1H), 4.40–4.24 (m, 2H), 4.21 (t, J = 4.4 Hz, 2H), 3.69 (dd, J = 14.8, 5.4 Hz, 1H), 3.57 (dd, J = 14.8, 7.1 Hz, 1H), 3.53–3.34 (m, 2H), 2.61 (s, 1H); 13C NMR (101 MHz, CDCl3) δ 154.9, 144.4, 141.1, 138.9, 135.4, 126.9, 125.3, 122.9, 122.5, 121.9, 119.9, 118.5, 116.8, 113.0, 112.8, 110.3, 104.3, 101.4, 69.8, 68.4, 64.4, 55.2, 49.0; MS (ESI) m/z (relative intensity) 375 (100) [M + H]+. HRMS (ESI) calcd for C23H23N2O3 [M + H]+: 375.1709; found: 375.1708.</p><!><p>A mixture of epoxide 103a (179 mg, 0.748 mmol), 118b (101 mg, 0.677 mmol), and Cs2CO3 (309 mg, 0.948 mmol) in DMF (2 mL) was heated at 80 °C for 20 h. The reaction was then partitioned between ethyl acetate and water. The organic phase was separated, washed with brine, dried over Na2SO4 and concentrated under vacuum. The residue was purified by flash chromatography over silica gel (ethyl acetate-hexanes) to afford 94 mg (36%) of 84; off-white solid; mp 133–135 °C; IR (film) 3402, 3342, 3050, 2925, 2870, 1663, 1497, 1094, 748, 720 cm−1; 1H NMR (400 MHz, DMSO-d6) δ 11.25 (s, 1H), 8.30 (d, J = 7.8 Hz, 1H), 7.44 (d, J = 8.1 Hz, 1H), 7.41–7.47 (m, 1H), 7.33 (ddd, J = 8.8, 7.5, 1.5 Hz, 1H), 7.28 (dd, J = 8.0, 7.9 Hz, 1H), 7.10 (ddd, J = 8.0, 7.2, 1.0 Hz, 1H), 7.08 (dd, J = 8.0, 0.5 Hz, 1H), 6.97–7.03 (m, 3H), 6.66 (d, J = 7.6 Hz, 1H), 5.49 (d, J = 5.3 Hz, 1H), 4.67 (d, J = 14.8 Hz, 1H), 4.61 (d, J = 14.8 Hz, 1H), 4.32–4.40 (m, 1H), 4.09–4.31 (m, 4H); 13C NMR (101 MHz, DMSO-d6) δ 164.3, 154.7, 145.0, 141.1, 138.9, 129.3, 126.4, 124.5, 123.4, 122.6, 122.4, 121.6, 118.5, 116.5, 116.1, 111.5, 110.3, 104.0, 100.4, 70.2, 67.1, 66.4, 44.5; MS (EI) m/z (relative intensity) 388 (25) [M+], 206 (100). HRMS (EI) calcd for C23H20N2O4 [M+]: 388.1423; found: 388.1417.</p><!><p>From 105 and 118a.Yield: 20% (35% based on recovered starting material); off-white solid; mp 172–174 °C; IR (film) 3516, 3319, 2945, 2879, 1600, 1500, 1091, 734, 720 cm−1; 1H NMR (400 MHz, DMSO-d6) δ 11.21 (s, 1H), 8.07 (d, J = 7.8 Hz, 1H), 7.42 (d, J = 8.1 Hz, 1H), 7.25–7.35 (m, 2H), 6.97–7.10 (m, 2H), 6.76 (dd, J = 8.1, 1.5 Hz, 1H), 6.62–6.73 (m, 3H), 6.48 (ddd, J = 7.9, 7.2, 1.5 Hz, 1H), 4.99 (d, J = 5.6 Hz, 1H), 4.06–4.40 (m, 5H), 3.52–3.58 (m, 1H), 3.37–3.43 (m, 2H), 3.22 (dd, J = 14.6, 7.5 Hz, 1H), 2.13–2.23 (m, 1H), 1.82–1.94 (m, 1H); 13C NMR (101 MHz, CDCl3) δ 155.0, 143.3, 141.0, 138.8, 135.4, 126.5, 124.4, 122.2, 121.7, 121.3, 118.5, 116.1, 115.7, 111.7, 111.4, 110.3, 103.7, 100.3, 64.6, 64.1, 63.9, 57.5, 48.3, 34.7; MS (EI) m/z (relative intensity) 388 (26%) [M+], 148 (100%). HRMS (EI) calcd for C24H24N2O3 [M+]: 388.1787; found: 388.1786.</p><!><p>From 114a and 118a. Yield: 76%; white solid; mp 61–62 °C; IR (film) 3402, 3056, 2925, 2862, 1500, 1183, 745 cm−1; 1H NMR (300 MHz, CDCl3) δ 8.01 (d, J = 7.8 Hz, 1H), 7.95 (s, 1H), 7.59 (d, J = 2.4 Hz, 1H), 7.38–7.44 (m, 2H), 7.35 (d, J = 8.7 Hz, 1H), 7.17–7.25 (m, 1H), 7.10 (dd, J = 8.7, 2.5 Hz, 1H), 6.76–6.87 (m, 3H), 6.66 (ddd, J = 8.1, 6.2, 2.4 Hz, 1H), 4.31–4.50 (m, 1H), 4.24 (t, J = 4.4 Hz, 2H), 4.07–4.22 (m, 2H), 3.59 (dd, J = 14.8, 5.6 Hz, 1H), 3.40–3.57 (m, 3H), 2.53 (d, J = 4.8 Hz, 1H); 13C NMR (101 MHz, CDCl3) δ 152.8, 144.4, 140.5, 135.5, 134.9, 126.1, 124.0, 123.3, 121.8, 120.4, 119.3, 118.3, 116.8, 115.5, 112.9, 111.5, 110.9, 104.9, 71.0, 68.5, 64.5, 54.8, 49.0; MS (EI) m/z (relative intensity) 374 (42) [M+], 148 (100); HRMS (EI+) m/z calcd for C23H22N2O3 [M+]: 374.1630; found: 374.1621.</p><!><p>From 117a and 118a. Yield: 61%; beige solid; mp 172–174 °C; IR (film) 3402, 3053, 2930, 2870, 1602, 1502, 1097, 908, 725 cm−1; 1H NMR (400 MHz, DMSO-d6) δ 11.08 (s, 1H), 7.98 (d, J = 8.0 Hz, 1H), 7.97 (d, J = 8.0 Hz, 1H), 7.41 (d, J = 8.0 Hz, 1H), 7.28 (ddd, J = 8.2, 7.2, 1.2 Hz, 1H), 7.18–7.05 (ddd, , J = 8.0, 7.0, 1.0 Hz, 1H), 6.99 (d, J = 2.1 Hz, 1H), 6.81 (dd, J = 8.5, 2.2 Hz, 1H), 6.76 (dd, J = 8.1, 1.5 Hz, 1H), 6.71 (ddd, J = 8.1, 7.1, 1.5 Hz, 1H), 6.65 (dd, J = 7.9, 1.5 Hz, 1H), 6.47 (ddd, J = 7.8, 7.2, 1.5 Hz, 1H), 5.26 (s, 1H), 4.23–4.08 (m, 3H), 4.04 (d, J = 5.1 Hz, 2H), 3.56 (dd, J = 14.7, 4.9 Hz, 1H), 3.53–3.40 (m, 2H), 3.28 (dd, J = 14.7, 6.8 Hz, 1H); 13C NMR (101 MHz, DMSO-d6) δ 157.7, 143.4, 141.0, 139.7, 135.4, 124.1, 122.6, 121.3, 120.9, 119.2, 118.5, 116.4, 116.3, 115.8, 111.9, 110.6, 108.1, 95.3, 70.4, 66.8, 63.9, 54.0, 48.2; MS (EI) m/z (relative intensity) 374 (40) [M+], 148 (100). HRMS (EI) calcd for C23H22N2O3 [M+]: 374.1630; found: 374.1613.</p><!><p>From 117c and 118a. Yield: 86%; white solid; mp 180–181 °C; IR (film) 3392, 2917, 2850, 1605, 1502, 1163, 911, 738 cm−1; 1H NMR (400 MHz, DMSO-d6) δ 11.12 (s, 1H), 7.98 (d, J = 8.6 Hz, 1H), 7.82 (dd, J = 9.5, 2.6 Hz, 1H), 7.40 (dd, J = 8.8, 4.5 Hz, 1H), 7.11 (ddd, J = 9.5, 9.4, 2.6 Hz, 1H), 6.98 (d, J = 2.1 Hz, 1H), 6.82 (dd, J = 8.6, 2.2 Hz, 1H), 6.76 (dd, J = 8.1, 1.4 Hz, 1H), 6.70 (ddd, J = 8.1, 7.6, 1.5 Hz, 1H), 6.65 (dd, J = 7.8, 1.5 Hz, 1H), 6.47 (ddd, J = 7.9, 7.8, 1.5 Hz, 1H), 5.24 (d, J = 4.8 Hz, 1H), 4.23–4.07 (m, 3H), 4.04 (d, J = 4.9 Hz, 2H), 3.55 (dd, J = 14.7, 4.8 Hz, 1H), 3.54–3.46 (m, 1H), 3.44–3.35 (m, 1H), 3.35–3.23 (m, 1H); 13C NMR (101 MHz, DMSO-d6) δ 158.1, 156.5 (d, 1JC F = 231 Hz), 143.4, 142.1, 136.1, 135.4, 123.2 (d, 3JC F = 10 Hz), 121.4, 121.3, 116.4, 116.1 (d, 4JC F = 4.0 Hz), 115.8, 111.9, 111.5 (d, 2JC F = 27 Hz), 111.3 (d, 3JC F = 12 Hz), 108.3, 104.9 (d, 2JC F = 24 Hz), 95.3, 70.4, 66.8, 63.8, 54.0, 48.2; 19F NMR (376 MHz, DMSO-d6) δ −127.1; MS (EI) m/z (relative intensity) 392 (24%) [M+], 148 (100%). HRMS (EI) calcd for C23H21N2O3F [M+]: 392.1536; found: 392.1523.</p><!><p>From 103a and 118c. Yield 87%; white solid; mp 88–89 °C; IR (film) 3406, 3273, 3060, 2943, 2828, 1502, 1240, 1098, 909, 751, 724 cm−1; 1H NMR (400 MHz, CDCl3) δ 8.31 (d, J = 7.9 Hz, 1H), 8.12 (s, 1H), 7.42–7.37 (m, 2H), 7.33 (t, J = 8.0 Hz, 1H), 7.29–7.21 (m, 1H), 7.08–6.92 (m, 4H), 6.88 (dd, J = 8.0, 1.1 Hz, 1H), 6.70 (d, J = 7.8 Hz, 1H), 4.43–4.31 (m, 2H), 4.29–4.22 (m, 1H), 3.88 (s, 3H), 3.16 (br s, 4 H), 3.02–2.91 (m, 2H), 2.86–2.79 (m, 2H), 2.79–2.68 (m, 2H); 13C NMR (101 MHz, CDCl3) δ 155.4, 152.4, 141.3, 141.1, 138.9, 126.8, 125.1, 123.2, 123.1, 122.7, 121.2, 119.8, 118.4, 112.9, 111.4, 110.1, 103.9, 101.4, 70.5, 65.9, 61.3, 55.5, 53.8, 50.9; MS (EI) m/z (relative intensity) 431 (25) [M+], 205 (100). HRMS (EI) calcd for C26H29N3O3 [M]+: 431.2209; found: 431.2220.</p><!><p>From 103a and 118d. Yield: 97%; off-white solid; mp 75–77 °C; IR (film) 3405, 3295, 3056, 2940, 2824, 1598, 1505, 1449, 1094, 908, 752, 725 cm−1; 1H NMR (400 MHz, CDCl3) δ 8.31 (d, J = 7.9 Hz, 1H), 8.08 (s, 1H), 7.44–7.37 (m, 2H), 7.37–7.22 (m, 4H), 7.05 (d, J = 7.9 Hz, 1H), 7.01–6.92 (m, 2H), 6.89 (dd, J = 7.3, 7.2 Hz, 1H), 6.70 (d, J = 7.9 Hz, 1H), 4.43–4.30 (m, 2H), 4.31–4.21 (m, 1H), 3.57 (br s, 1H), 3.34–3.18 (m, 4H), 2.97–2.86 (m, 2H), 2.86–2.74 (m, 2H), 2.75–2.64 (m, 2H); 13C NMR (101 MHz, CDCl3) δ 155.3, 151.3, 141.1, 138.9, 129.3, 126.8, 125.2, 123.1, 122.7, 120.0, 119.8, 116.3, 112.9, 110.2, 104.0, 101.4, 70.4, 66.0, 61.2, 53.6, 49.4; MS (EI) m/z (relative intensity) 401 (30) [M+], 175 (100). HRMS (EI) calcd for C25H27N3O2 [M]+: 401.2103; found: 401.2114.</p><!><p>From 103a and 118e. Yield: 88%; white solid; mp 62–64 °C; IR (film) 3407, 3059, 2930, 1603, 1497, 1261, 1208, 725 cm−1; 1H NMR (400 MHz, CDCl3) δ 8.26 (d, J = 7.6 Hz, 1H), 8.08 (s, 1H), 7.46–7.36 (m, 2H), 7.32 (t, J = 8.0 Hz, 1H), 7.24 (ddd, J = 8.0, 6.5, 1.8 Hz, 1H), 7.12–7.05 (m, 2H), 7.04–6.98 (m, 1H), 6.99–6.91 (m, 3H), 6.88 (ddd, J = 8.0, 7.3, 1.5 Hz, 1H), 6.78 (dd, J = 8.0, 1.4 Hz, 1H), 6.72–6.66 (m, 1H), 6.64 (d, J = 8.0 Hz, 1H), 5.31 (br s, 1H), 4.53–4.43 (m, 1H), 4.31 (d, J = 5.2 Hz, 2H), 3.82 (s, 3H), 3.68 (dd, J = 13.4, 4.5 Hz, 1H), 3.53 (dd, J = 13.3, 7.3 Hz, 1H), 2.70 (s, 1H); 13C NMR (101 MHz, CDCl3) δ 155.1, 150.9, 145.7, 144.9, 141.1, 139.4, 138.9, 126.8, 125.2, 124.4, 124.3, 123.0, 122.6, 121.2, 119.9, 119.7, 117.8, 117.7, 112.8, 112.2, 110.2, 104.2, 101.5, 70.1, 69.1, 56.1, 47.2; MS (EI) m/z (relative intensity) 454 (78) [M+], 228 (100), 183 (40), 120 (25). HRMS (EI) calcd for C28H26N2O4 [M+]: 454.1893; found: 454.1885.</p><!><p>A mixture of amine 104a (208 mg, 1.24 mmol) in dry THF (12 mL) was cooled in an ice bath and treated with Et3N (0.34 mL, 2.4 mmol). After 15 min, acid 12052 (292 mg, 1.21 mmol), N-(3-diethylaminopropyl)-N′-ethylcarbodiimide hydrochloride (EDC hydrochloride) (356 mg, 1.86 mmol) and 1-hydroxybenzotriazole hydrate (HOBT·H2O) (280 mg, 1.8 mmol) were added sequentially and the reaction mixture was stirred at room temperature for 3 h. It was partitioned between ethyl acetate and water, the organic phase was separated and washed with saturated aqueous NH4Cl and NaHCO3, H2O, brine, dried over Na2SO4 and concentrated under vacuum. The residue was purified by flash chromatography over silica gel (ethyl acetate-hexanes) to afford 318 mg (67%) of 93 as a white solid; mp 155–156 °C; IR (film) 3425, 3292, 3053, 2937, 2827, 1671, 1502, 1256, 752, 722 cm−1; 1H NMR (400 MHz, DMSO-d6) δ 11.29 (s, 1H), 8.28 (t, J = 5.5 Hz, 1H), 8.22 (d, J = 7.8 Hz, 1H), 7.45 (d, J = 8.1 Hz, 1H), 7.33 (t, J = 8.0 Hz, 1H), 7.26 (t, J = 8.0 Hz, 1H), 7.10 (d, J = 7.9 Hz, 1H), 7.06 (t, J = 8.0 Hz, 1H), 6.82–7.02 (m, 4H), 6.61 (d, J = 7.9 Hz, 1H), 4.76 (s, 2H), 4.05 (t, J = 5.8 Hz, 2H), 3.70 (s, 3H), 3.57 (q, J = 5.7 Hz, 2H); 13C NMR (101 MHz, DMSO-d6) δ 168.0, 154.0, 149.3, 147.8, 141.1, 138.9, 126.3, 124.6, 122.6, 121.4 (2C), 120.7, 118.5, 114.2, 112.4, 111.7, 110.3, 104.5, 100.8, 67.3, 67.2, 55.4, 38.2; MS (EI) m/z (relative intensity) 390 (64) [M+], 267 (100), 196 (42), 154 (57). HRMS (EI) calcd for C23H22N2O4 [M+]: 390.1580; found: 390.1577.</p><!><p>A solution of 118f·HCl (354 mg, 1.51 mmol) in dry THF (20 mL) was cooled in an ice bath and treated with Et3N (0.65 mL, 4.5 mmol). After 15 min, acid 12052 (365 mg, 1.51 mmol), EDC hydrochloride (435 mg, 2.26 mmol) and HOBT·H2O (348 mg, 2.3 mmol) were added sequentially and the reaction mixture was stirred at room temperature for 4 h. It was worked up and purified as in the preceding procedure to afford 481 mg (76%) of 94 as a white solid; mp 169–170 °C; IR (film) 3412, 3285, 3056, 2933, 1661, 1502, 1253, 1216, 1107, 722 cm−1; 1H NMR (400 MHz, DMSO-d6) δ 11.27 (s, 1H), 8.21 (d, J = 7.8 Hz, 1H), 7.97 (d, J = 8.1 Hz, 1H), 7.45 (d, J = 8.1 Hz, 1H), 7.32 (t, J = 7.2 Hz, 1H), 7.27 (t, J = 8.0 Hz, 1H), 7.11 (d, J = 8.0 Hz, 1H), 7.08–7.00 (m, 2H), 6.83–7.00 (m, 3H), 6.64 (d, J = 7.9 Hz, 1H), 4.99 (t, J = 5.4 Hz, 1H), 4.77 (s, 2H), 4.28–4.16 (m, 1H), 4.15–4.00 (m, 2H), 3.70 (s, 3H), 3.69–3.63 (m, 1H), 3.63–3.55 (m, 1H); 13C NMR (101 MHz, DMSO-d6) δ 167.6, 153.89, 149.3, 148.0, 141.1, 138.9, 126.3, 124.6, 122.5, 121.4, 121.3, 120.7, 118.6, 114.1, 112.6, 111.5, 110.4, 104.5, 100.9, 67.3, 67.2, 59.8, 55.6, 50.1; MS (EI) m/z (relative intensity) 420 (40) [M+], 297 (100), 196 (42), 154 (44). HRMS (EI) calcd for C24H24N2O5 [M+]: 420.1685; found: 420.1667.</p><!><p>A solution of 94 (75 mg, 0.18 mmol) in a 2:1 mixture of dichloromethane-THF (4 mL) was cooled to −78 °C and treated with diethylaminosulfur trifluoride41 (DAST) (32 μL, 0.24 mmol). After 1 h, K2CO3 (38 mg, 0.26 mmol) was added and the cold bath was removed. The reaction mixture was stirred at room temperature for 2 h., diluted with saturated aqueous NaHCO3 solution and extracted with dichloromethane. The combined organic extracts were dried over Na2SO4 and concentrated under vacuum. The residue was purified by flash chromatography over silica gel (ethyl acetate-hexanes) to afford 44 mg (62%) of 91 as a white solid; mp 152–154 °C; IR (film) 3402, 3062, 2936, 1669, 1606, 1583, 1500, 1452, 1252, 1117, 751, 722 cm−1;1H NMR (400 MHz, DMSO-d6) δ 11.28 (s, 1H), 8.22 (d, J = 7.8 Hz, 1H), 7.45 (d, J = 8.1 Hz, 1H), 7.34 (ddd, J = 8.2, 7.2, 1.2 Hz, 1H), 7.27 (t, J = 8.0 Hz, 1H), 7.10–7.15 (m, 1H), 7.11 (d, J = 7.7 Hz, 1H), 6.82–7.05 (m, 4H), 6.75 (d, J = 7.7 Hz, 1H), 5.02 (s, 2H), 4.43–4.62 (m, 2H), 4.27–4.35 (m, 1H), 4.09 (dd, J = 9.7, 4.4 Hz, 1H), 3.98 (dd, J = 9.7, 5.4 Hz, 1H), 3.75 (s, 3H); 13C NMR (101 MHz, DMSO-d6) δ 163.8, 153.9, 149.3, 148.0, 141.1, 138.9, 126.3, 124.7, 122.5, 121.5, 121.4, 120.8, 118.6, 114.5, 112.7, 111, 110.4, 104.5, 100.8, 70.6, 69.9, 65.2, 62.3, 55.7; MS (EI) m/z (relative intensity) 402 (100) [M+], 182 (38), 154 (72). HRMS (EI) calcd for C24H22N2O4 [M+]: 402.1580; found: 402.1560.</p><!><p>Compound 97 was obtained from 4,6-dibromo-3-hydroxycarbazole (96) by treatment with epichlorohydrin, followed by amine 104a by the same procedure as employed for the preparation of 3.32 Yield 42%; off-white solid; mp 141–143 °C; IR (KBr) 3458, 3378, 2922, 2834, 1507, 1452, 1253, 1222, 1186, 1121, 745 cm−1; 1H NMR (400 MHz, CDCl3) δ 8.90 (d, J = 1.8 Hz, 1H), 8.22 (br s, 1H), 7.55 (dd, J = 8.6, 2.0 Hz, 1H), 7.31 (d, J = 8.7, 2.6 Hz, 1H), 7.30 (d, J = 8.7 Hz, 1H), 6.98–6.90 (m, 4H), 4.20–4.10 (m, 5H), 3.86 (s, 3H), 3.13 (dd, J = 5.8, 5.7 Hz, 2H), 2.99 (dd, J = 6.4, 4.7 Hz, 2H); 13C NMR (101 MHz, CDCl3) δ 149.9, 149.5, 148.4, 139.2, 136.3, 129.5, 125.5, 125.1, 121.8, 121.1, 115.2, 114.4, 112.1, 112.07, 112.02, 110.1, 110.0, 107.4, 74.1, 69.0, 68.6, 56.0, 51.7, 49.0; MS (CI) m/z (relative intensity) 566 (52) [M+ (81Br2)], 564 (100) [M+ (81Br79Br)], 562 (47) [M+ (79Br2)], 487 (40), 485 (41), 407 (18), 224 (36). HRMS (EI) m/z calcd for C24H2481Br79BrN2O4 [M+]: 564.0082; found: 564.0099.</p><!><p>Compound 98 was prepared from 97 by the same procedure as employed for the preparation of 17. Yield 34%; white solid; mp 208–209 °C; IR (KBr) 3430, 3277, 2926, 2875, 1634, 1504, 1290, 1255, 751 cm−1; 1H NMR (400 MHz, DMSO-d6) δ 11.63 (s, 1H), 8.72 (d, J = 2.0 Hz, 1H), 7.59 (dd, J = 8.6, 2.0 Hz, 1H), 7.51 (d, J = 8.8 Hz, 1H), 7.50 (d, J = 8.8 Hz, 1H), 7.36 (d, J = 8.8 Hz, 1H), 7.02–6.83 (m, 4H), 4.28–4.10 (m, 7H), 3.72 (s, 3H), 3.85–3.61 (m, 4H); 13C NMR (101 MHz, DMSO-d6) δ 165.9, 149.2, 148.3, 147.7, 139.3, 136.5, 128.6, 123.6, 121.4, 120.7, 120.4, 115.6, 113.8, 113.2, 112.3, 111.0, 110.0, 105.5, 104.4, 71.6, 71.0, 66.9, 66.4, 55.5, 49.0, 45.6; MS (EI) m/z (relative intensity) 482 (39), 480 (100), 478 (44). HRMS (EI) calcd for C26H2481Br79BrN2O5 [M+]: 604.0031; found: 604.0007.</p><!><p>Thiapyrone 12340 (1.00 g, 5.15 mmol) was dissolved in TFA (5 mL). Trimethylsilyl azide (0.68 mL, 5.2 mmol) was added at room temperature, the mixture was stirred for 2 d and the reaction was quenched with water and basified with NaOH. The mixture was extracted with dichloromethane, the combined organic phase was dried over Na2SO4, the solvent was evaporated under reduced pressure and the residue was purified by flash chromatography over silica gel (ethyl acetate-hexanes) to afford lactams 12453 (260 mg, 24%) and 12554 (500 mg, 47%). Product 125 had the following properties: 1H NMR (300 MHz, CDCl3) δ 8.25 (br s, 1H), 7.50 (d, J = 8.5 Hz, 1H), 6.73 (dd, J = 8.5, 2.6 Hz, 1H), 6.67 (d, J = 2.6 Hz, 1H), 3.82 (s, 3H), 3.39 (t, J = 7.0 Hz, 2H), 2.62 (t, J = 6.9 Hz, 2H); 13C NMR (75 MHz, CDCl3) δ 173.9, 161.0, 142.7, 136.4, 117.6, 112.1, 109.2, 55.6, 34.4, 33.6; MS (EI) m/z (relative intensity) 209 (100) [M+], 154 (75). HRMS (EI) calcd for C10H11NO2S [M+]: 209.0511; found: 209.0507.</p><p>To lactam 125 (500 mg, 2.39 mmol) in dry THF-ether (15 mL, 1:1), LiAlH4 (181 mg, 4.77 mmol) was added at 0 °C and the mixture was then heated at 40 °C overnight. The reaction was carefully quenched with 0.5 mL of water and filtered through Celite. The solid was washed repeatedly with ether, the filtrate was evaporated and the crude product was purified by flash chromatography over silica gel to give the corresponding amine (260 mg, 56%) as an oil that solidified upon standing: mp 71.5–72.5 °C; 1H NMR (400 MHz, CDCl3) δ 7.33 (d, J = 8.5 Hz, 1H), 6.41 (dd, J = 8.5, 2.6 Hz, 1H), 6.34 (d, J = 2.6 Hz, 1H), 3.76 (s, 3H), 3.24–3.22 (m, 2H), 2.76–2.73 (m, 2H), 2.12–2.06 (m, 2H); 13C NMR (101 MHz, CDCl3) δ 159.8, 153.3, 134.1, 117.0, 106.6, 105.9, 55.3, 47.7, 33.5, 32.1; MS (EI) m/z (relative intensity) 195 (97) [M+], 166 (100). HRMS (EI) calcd for C10H13NOS [M+]: 195.0718; found: 195.0710.</p><p>The above amine (250 mg, 1.28 mmol) was dissolved in 10 mL of methanol and 1.1 mL of 37% formaldehyde. Sodium cyanoborohydride (476 mg, 7.57 mmol) was added at room temperature and the mixture was stirred for 30 min, during which the pH was maintained between 4 and 5 by the addition of a few drops of 1 N HCl solution. The solvent was removed under reduced pressure, the residue was dissolved in dichloromethane, washed with water and dried over Na2SO4. The solvent was evaporated and the residue was purified by flash chromatography over silica gel (ether-pentane) to afford 245 mg (91%) of 101; colorless oil; IR 2933, 2822, 1593, 1555, 1478, 1107, 1074, 1032 cm−1; 1H NMR (300 MHz, CDCl3) δ 7.33 (d, J = 8.4 Hz, 1H), 6.50 (d, J = 2.6 Hz, 1H), 6.40 (dd, J = 8.4, 2.6 Hz, 1H), 3.78 (s, 3H), 3.17–3.13 (m, 2H), 2.91 (s, 3H), 2.78–2.74 (m, 2H), 2.07–2.00 (m, 2H); 13C NMR (75 MHz, CDCl3) δ 159.9, 155.1, 133.7, 117.9, 105.3, 104.4, 55.6, 55.2, 42.6, 30.5, 30.3; MS (EI) m/z (relative intensity) 209 (77) [M+], 180 (100). HRMS (EI) calcd for C11H15NOS [M+]: 209.0874; found: 209.0875.</p><!><p>Stable, inducible HEK293 cells expressing a CPVT-causing RyR2 mutant, R4496C, display robust spontaneous Ca2+ oscillations (SOICR), but parental HEK293 cells do not.55,56 These RyR2-R4496C cells were used to assess the impact of carvedilol, carvedilol analogs, beta-blockers, and other compounds on SOICR. SOICR was measured using single-cell Ca2+ imaging and the fluorescent Ca2+ indicator dye fura-2/AM (Invitrogen) as described previously55,56 Briefly, cells grown on glass coverslips for 18–22 hours after induction by 1 μg/ml tetracycline were loaded with 5 μM fura 2/AM in KRH (Krebs–Ringer–Hepes) buffer (125 mM NaCl, 5 mM KCl, 1.2mM KH2PO4, 6 mM glucose, 1.2 mM MgCl2 and 25 mM Hepes, pH 7.4) plus 0.02% pluronic F-127 and 0.1 mg/ml BSA for 20 min at room temperature (23°C). The coverslips were then mounted in a perfusion chamber (Warner Instruments, Hamden, CT, U.S.A.) on an inverted microscope (Nikon TE2000-S). The Ca2+ concentration was then stepped to 0.5 mM for 5 min before increasing to 1 mM. The cells were continuously perfused with KRH buffer containing 1 mM CaCl2 and different drugs for 8–10 min. Caffeine (10 mM) was applied at the end of each experiment to confirm the expression of active RyR2 channels. Time-lapse images (0.25 frame/s) were captured and analyzed with the Compix Simple PCI 6 software (Compix Inc., Sewickley, PA, USA). Fluorescence intensities were measured from regions of interest centered on individual cells. Only cells that responded to caffeine were used in analyses. All chemicals were obtained from Sigma (St. Louis, MO) unless otherwise specified.</p>

PubMed Author Manuscript

Aroma molecules as dynamic volatile surfactants: functionality beyond the scent

Understanding of non-equilibrium processes at dynamic interfaces is indispensable for advancing design and fabrication of solid state and soft materials. The research presented here unveils specific interfacial behavior of aroma molecules and justifies their usage as multifunctional volatile surfactants. As non-conventional volatile amphiphiles we study commercially available poorly water-soluble compounds from the classes of synthetic and essential flavor oils. Their distinctive feature is high dynamic interfacial activity, so that they decrease the surface tension of aqueous solutions on a time scale of milliseconds. Another potentially useful property of such amphiphiles is their volatility, so that they notably evaporate from interfaces on a time scale of seconds. This behavior allows for control of wetting and spreading processes. A revealed synergetic interfacial behavior of mixtures of conventional and volatile surfactants is attributed to a decrease of the adsorption barrier as a result of high statistical availability of new sites at the surface upon evaporation of the volatile component. Our results offer promising advantages in manufacturing technologies which involve newly creating interfaces, such as spraying, coating technologies, ink-jet printing, microfluidics, laundry, stabilization of emulsions in cosmetic and food industry, as well as in geosciences for controlling aerosols formation.

aroma_molecules_as_dynamic_volatile_surfactants:_functionality_beyond_the_scent

Introduction<!>Experimental Section<!>Results and Discussion<!>Conclusions<!>Table of Contents artwork<!>Aroma molecules as dynamic volatile surfactants: functionality beyond the scent<!>Eq. (S1)

<p>Odor compounds are aliphatic or aromatic amphiphilic molecules with varied carbon backbones and diverse functional groups, including aldehydes, esters, ketones, alcohols, alkenes, carboxylic acids, amines. Essential oils are generated through different biosynthetic routes and extracted from vegetable raw materials. The fragrance and cosmetic industries promoted the development of chemical synthesis of perfume molecules, 1 as well as activated research on the structure-odor relationship. Odortriggering activity is an important part of specific signaling pathways in olfactory system, an evolutionally evolved remarkable molecular sensing mechanism. 2,3 Features of molecules which can act as odorant-triggers are typically considered in terms of molecular weight, geometry, chirality, flexibility (vibrational frequencies and rotation), as well as biochemical (enzymatic) influence. Applications of aroma molecules are currently related to their biological activities (odor, antioxidant, antimicrobial). However, rich and complex composition of these poorly water soluble volatile amphiphiles defines their specific physicochemical properties beyond the scent function.</p><p>In this paper we disclose functional properties of aroma molecules, which are of high interest for material and surface science and which are envisaged to notably expand the application fields of this class of amphiphilic compounds. One of the important parameters which define the formation, properties and functions of aqueous solutions, dispersions and mixtures is the surface/interfacial tension γ which is typically assumed to be an equilibrium value. This thermodynamic parameter defines the energetic costs associated with the existence of a phase boundary, e.g. between liquid and air. A standard approach to facilitate the formation of new surfaces is to apply surfactants -amphiphilic molecules, which concentrate in a thin sub-interfacial region, and thus decrease the surface/interfacial energy (Figure 1a). However, the systems in which a fresh surface is created may not be in an adsorption equilibrium, and therefore exhibit a time-dependent (dynamic) surface tension γ (t). 4,5 For solutions of amphiphilic molecules the γ (t) dependence may span from microseconds to minutes (or even hours) [5][6][7] (Figure 1b). In the simplest case the time effect is controlled by the rate of the surfactant diffusion D to the surface according to the diffusion equation:</p><p>where c is bulk concentration of the surfactant and x is a distance measured away from the surface. Diffusivity D depends on the molecular weight and molecular structure of a molecule (Figure 1b), so that in practical research the choice of surfactants is often considered in terms of their dynamic interfacial activity.</p><p>"Dynamic" interfaces with non-equilibrium adsorption state are relevant to a wide range of processes and technologies in which "fresh" surfaces are created (Figure 1c).</p><p>Examples include cleaning, 8 spraying and coating operations, i.e. liquids emerging from orifices or spray nozzles, 9, 10 or bubbles and droplets formed within liquids, as in microfluidic devices. 11 In ink jet printing surface life-times of as low as fractions of a millisecond are of practical importance. 12 S1, Supporting Information) and of volatile surfactants (filled symbols) (Table 1) measured with bubble pressure tensiometer at 25°C: dashed line -MQ water; red empty squares-sodium dodecyl sulfat (SDS); magenta down triangles-DX4005N fluorosurfactant; black circles -methylestersulfonat (MES); blue triangles -citral; green circles -linalool, orange squares -geraniol. (c) Examples of processes and technologies which involve formation of new surfaces at a timescale of milliseconds and below.</p><p>In this paper we introduce volatile surfactants (from a class of aroma molecules) which possess distinctive features such as volatility and a higher than conventional surfactants dynamic interfacial activity at a time scale of milliseconds (Figure 1b, Table 1). We note these unique and powerful features of aroma (fragrance) molecules remained until now unrevealed, so that intensive fundamental research on the effects of perfumes on the interfacial processes in complex colloidal systems has to be revisited. 13 Here these properties are quantified using dynamic and static (equilibrium) surface tension measurements, and are further demonstrated in dynamic wetting experiments. We present a methodology allowing evaluating interactions of volatile surfactants with other component of complex liquids, and disclose a synergetic interfacial behavior of volatile and conventional surfactants, that offers breakthrough possibilities in surfacesemerging technologies. Preliminary conclusions are made with regards to the understanding the "structurefunction" relationship of volatile surfactants.</p><!><p>Aroma compounds -flavor oils linalool, 1-decanol (both form Sigma-Aldrich), benzyl alcohol, citronellol and conventional surfactants SDS (Sigma-Aldrich), MES (PT Global Eco Chemicals), and DX4005N (Dynax) (Table 1S, Supporting information) have been used as received.</p><p>Solutions for the measurements of isotherms have been prepared by titration method via stepwise addition of weighted amounts of corresponding liquids (or stock solutions) to MQ water under stirring at constant temperature for 10 minutes at 25°C, if not stated otherwise.</p><p>Solutions of conventional surfactants sodium dodecyl sulfat (SDS from Sigma-Aldrich), methylestersulfonat (MES from PT. Global Eco Chemicals Indonesia) anionic surfactants and DX4005N fluorosurfactant (Dynax Corporation) (Table S1, Supporting Information) have been prepared by dilution of stock solutions. The curves presented in Figure 1b are exemplary measurements in course of titration experiments to obtain surface tension isotherm of respective surfactants.</p><p>Dynamic surface tension was measured by maximum bubble pressure method at 25 ± 0.5°C using a SITA pro line t100 apparatus (SITA GmbH, Germany) and capillary from polymethyl ether ketone (MEEK).</p><p>Measuring dynamic surface tension with the bubble pressure method enables high precision and flexibility without adjustment of the immersion depth. This is done by pumping air through a capillary into the liquid to be analyzed. The pressure within the bubble changes continuously as the radius growths. The surface tension is calculated from the deviation between pressure maximum and minimum. The calibration is carried out with MQ water. Thereby, the radius of the capillary (0.8 mm) is taken into account.</p><p>Static surface tension was measured by the pendant drop method at 25 ± 1°C using horizontal microscope. The drop was continuously imaged via DCM-130 digital video camera. The value of the surface tension was determined from the shape analysis of the pendant drops by numerical integration of the Young-Laplace equation and approximation of the resulting droplet shape to experimentally obtained image. To determine the equilibrium surface tension, droplets of the surfactant solution were formed in a closed cuvette with an aliquot of a corresponding solution at the bottom of the cell to maintain the respective partial vapor pressure. To study the effect of the solution evaporation and the volatility of the surfactants on the surface tension, the drops have been formed in the open system (Figure 4a).</p><p>Spreading kinetics was studied in the experimental setup consisting of horizontal microscope equipped with a high-speed video camera (XIMEA xiQ), horizontal adjustable substrate support, and vertical dosing syringe with flat-tip needle mounted above the substrate. In a standard procedure a substrate is placed on substrate support, and a drop with a diameter of ~ 1 mm is formed at the needle tip. The substrate support is gently raised until the drop contact with the substrate is initiated, and the drop is simultaneously transferred and spreads onto the substrate surface. Process of drop transfer and spreading is recorded with 432 x 310 pix resolution at a frame rate of 1000 fps. Geometrical parameters of the spreading drop (contact angle θ and drop base diameter d) have been extracted from the sequence of the drop images using home-made software. For each system at least five drops have been reproduced. Freshly cleaned glass slides (cleaning procedure with chromic mixture followed by UV/ozone treatment) or Melinex® PCS peelable polymer slides have been used as substrates.</p><p>Jetting experiments have been done using pico-liter dosing system (PDDS module from Dataphysics, Germany). 30 pl droplets of studied solutions have been ejected onto fresh polymer substrate (Milinex® PCS). The drop diameter and the contact angle have been monitored and analyzed using SCA Software.</p><!><p>The interfacial behavior of amphiphilic organic substances was studied in connection to their molecular structure and related physico-chemical parameters (Table 1). Figure 2a displays surface tension of solutions of three alcohols measured with maximum bubble pressure tensiometer which assesses the pressure development in a gas bubble (fresh interface) at a tip of a capillary immersed into a liquid (Inset in Figure 2b). 14 This method allows a time window from milliseconds to hundreds of seconds over which dynamic surface tension measurements can be made. 5 2, Table S2 and Figure S1, Supporting Information). Sketch illustrates the measurement cell in bubble pressure method:</p><p>air bubble growing at a tip of a capillary inside the bulk aqueous solution.</p><p>An equation which relates adsorption Γ(t) to the diffusion D of solute molecules from the bulk to the interface as well as to their back diffusion was first obtained by Ward and Tordai. 16 It accounts for the diffusion of monomers from the bulk to interface, and also for the back diffusion into the bulk as the interface becomes more crowded. It is widely used to analyze the adsorption process from the dynamic surface tension measurements. 5 :</p><p>where c is the bulk surfactant concentration, D the monomer diffusion coefficient, c(τ) the concentration in the subsurface, and τ is a dummy variable of integration. The solid lines in Figure 2b are fits according to Equation 2, which has been solved numerically using the code developed by Li at al. 17 . It is assumed that the isotherm of surface tension at equilibrium can be described by semiempirical Szyszkowski Equation which is valid in semidilute regime:</p><p>Parameters K L (a) and Γ max (B) for linalool and benzyl alcohol have been obtained from the surface tension isotherms measured with bubble pressure tensiometer at 20 s surface age time. The approximation of the surface tension to the infinite-time limit γ eq (Table S3, Supporting Information) has been done using method suggested by Christov et al. 14 More details are given in the Supporting Information.</p><p>Dynamic surface tension of linalool solution with a concentration of 6,4•10 -4 mol/L can be well fitted by Ward-Tordai model (Figure 2b). The experimental value of γ at 20 s surface-age time is slightly higher than that evaluated from Equation 3, while the estimated diffusion coefficients for linalool and for 1-decanol are an order of magnitude lower than that evaluated from Einstein relation (Table S2, Supporting Information). This result indicates an existence of the adsorption barrier for the molecules to transform from sub-interfacial layer to the interface.</p><p>The estimated equilibrium surface tension of the 1decanol solution is considerably lower than that measured at 20 s surface age (Table S2). The experimental curve for the solution of benzyl alcohol can not be described in the frame of Ward-Tordai model, since the evaluated diffusion coefficient does not have a physical sense (Figure S1).</p><p>Data presented in Figure 2a,b convincingly demonstrate qualitative and quantitative difference in the behavior of selected alcohols both on a short-term scale and at (quasi)equilibrium conditions. Benzyl alcohol has a relatively good solubility in water (Table 1), so that its solutions exhibit practically a time-invariant surface tension in the whole studied concentration and time-scale ranges. The low interfacial activity of benzyl alcohol is due to small energetic gain in moving solvated molecules from solution to the interface. Poorly soluble 1-decanol reveals a high surface activity at a time scale of seconds, i.e. in a quasi-equilibrium adsorption state. Its retarded diffusivity is clearly seen in the isotherm at 40 ms time (Figure 2a) and in respected kinetic curve in Figure 2b.</p><p>The most interesting behavior from the practical point of view has been found for linalool, which shows a high dynamic activity already at a time scale of tens of milliseconds. Linalool exhibits a higher diffusivity in diluted solutions than that of 1-decanol and a higher dynamic and equilibrium interfacial activity as compared to benzyl alcohol, resulting in 100 times lower amount of linalool to produce the same interfacial effect as benzyl alcohol (Table S2). We note that measurements of the surface tension of solutions of poorly soluble aroma molecules are complicated by so called mesoscale solubility of organic liquids in water. 18 During tensiometric measurements, the mesoscale solubility in a form of "heterogeneous" oil droplets has to be taken into account. The mesoscale droplets, which serve as reservoirs and affect the diffusion-limited regime of adsorption, can be removed from the solution by filtration. According to our observations, the transition from molecular to mesoscale solubility of tested aroma molecules is detected at the concentrations below the solubility limit reported for these compounds in literature, that suggest a revision of the published solubility data.</p><p>Figure 3a compares surface tension isotherms of linalool and of conventional surfactant sodium dodecyl sulfate (SDS) in MQ water. As clearly seen, linalool exhibits a higher interfacial activity, especially in a millisecond range. In contrast to micelle-forming SDS solutions (Figure 2d,A), the cut-off in the concentration dependence of the surface tension of linalool solutions is attributed to the limit of its molecular solubility. The fact that the lowest surface tension is similar for both systems is rather accidental, since this value varies for the saturated solutions of aroma molecules. A practical conclusion for inkjet formulations is that a typically desired value of γ of 55 mN/m at ~50 ms surface age time can be achieved with a three times lower amount of linalool as compared to that of a conventional surfactant (Figure 2a).</p><p>Obviously, important properties and functions of micelleforming surfactants are not innate for amphiphilic organic oils. Therefore, an intriguing question was whether the mixtures of conventional surfactants and of aroma molecules retain the specific features of individual solutions. Displayed in Figure 3b are surface tension isotherms of SDS and linalool individual solutions, as well as of their mixtures with a constant fraction of linalool of 0,5 versus total molar concentration of surfactants in solution. Quantitative analysis of these data is beyond the scope of this publication. Here we emphasize several important observations. In the concentration range below the critical micelle concentration (CMC) of SDS, the dynamic behavior of the mixtures is very similar to that of linalool solutions of the same concentration (Figure 3c i,ii). This observation can be interpreted as an enhancement of the intrinsic diffusivity of SDS molecules by lowering the adsorption barrier in the presence of linalool.</p><p>Around CMC of SDS (8,3 mmol/L) a clear synergetic effect of the surfactants is seen in that the surface tension of the mixture at a time scale of above ~100 ms is significantly lower than that of the individual solutions of each surfactant with the same concentration (Figure 3c, iii). The synergetic adsorption behavior of mixtures of charged and non-ionic surfactants is typically attributed to the formation of a denser adsorption layer as a result of the reduced electrostatic repulsion of the charged headgroups. 19 From a dynamic point of view, with increasing bulk surfactant concentration typically a drop of the evaluated diffusion coefficient in several orders of magnitude is observed. It is explained by thermodynamic factors and by a statistical factor, related to the availability of an empty site for a new molecule to enter a crowded interface. 5 The latter factor presumably explains the enhanced effective diffusivity of SDS in mixed solutions from the bulk towards the interface as a result of the property of aroma molecules to desorb from the interface into the air phase thus reducing the backward diffusion of SDS surfactants (Figure 3d).</p><p>The volatile behavior of surfactants has been studied using pendant drop method, which has been utilized in two combinations: in a closed cuvette allowing for a saturated atmosphere, and in air (Figure 4a). In the latter case the solvent (water) evaporates during the measurements, and at the same time the unsaturated atmosphere affects the adsorption-desorption equilibrium of the volatile surfactant. Shown in Figure 4a,b are kinetic curves of the surface tension, collected over a large time scale by a combination of dynamic (I) and static (II) measurements. For both concentrations of SDS (above CMC) presented in Figure 4b the equilibrium adsorption (constant surface tension value) is achieved on a time scale of tens of seconds. In the open system (III) on a longer time scale the surface tension continues to decrease as a result of the reduction of the drop volume (water evaporation) leading to an increased concentration of the surfactant in the pendant drop.</p><p>Different behavior is observed for linalool solutions (Fig- ure 4c). While in the closed cuvette (system II) the equilibrium surface tensions measured by the pendant drop and by the bubble pressure methods are in a good agreement, in the open system the evaporation of the surfactant into the gas phase overcomes the effect of water evaporation, so that the surface tension of 2.4 mmol/L linalool solution increases with time and almost reaches that of water within 100 seconds. The constant surface tension of 10 mmol/L linalool solution is due to the compensation of the surfactant evaporation by the reduction of the drop volume as a result of water evaporation.</p><p>To clarify the effect of the volatile surfactants on the equilibrium wetting behavior and on the spreading kinetics of aqueous solutions we employed two kinds of measurements. In a first set a ~40 µl-small pendant drop with a life time of minutes, i.e. with a preformed adsorbed layer, was brought into contact with a solid substrate (Figure 5a), and a dynamic wetting behavior (drop diameter and contact angle) has been monitored. Time dependencies of contact angle and of drop base diameter of drops of water and of 7 mmol/L solution of linalool on a freshly cleaned glass surface are presented in Figure 5b in half-logarithm coordinates.</p><p>For both systems spreading process can be divided into two stages: fast spreading at 0-10 ms down to contact angle of ~ 20-25°, and slow spreading up to 5° in 1 s. Figure 5 c shows time dependence of the squared drop base diameter for the fast spreading stage (at t < 10 ms) which can be presented as d 2 = kt, where k is the spreading rate constant. Values of k determined from the experimental data for water and linalool solution are 470±20mm 2 /s and 340±10 mm 2 /s, respectively (Figure 4c). The lower values of k for linalool solution indicate that already on a time scale of <10 ms linalool reduces the surface tension of water. Thus, with these experiments we assess the surface age times, which are even shorter than that currently accessible with bubble pressure tensiometry.</p><p>On the other side, the constant of the kinetic spreading regime is shown to be dependent on the surface tension: 20</p><p>where γ is the surface tension, R -radius of curvature, ρliquid density. The rate constants k evaluated from Equation 4 result in 270 mm 2 /s and 220 mm 2 /s for water and linalool solution, correspondingly. The reduction of the spreading constant is presumably due to the fast decrease of the surface tension of water in the presence of linalool.</p><p>The fact that the calculated from the experimental data and evaluated according to Equation 4 values of the spreading rate constant are of the same order of magnitude can be considered as a confirmation of the applicability of the spreading model for the studied systems. In a time range of 0,01 -1 s a much slower spreading was observed with time exponents of 0,075 and 0,110 for water and for linalool solution, correspondingly (Figure 5d). These values are close to 0,1, which is an indication of viscous spreading regime for the case of perfect spreading, as shown by Tanner 21 and de Gennes. 22 ! " = # (Eq. 5)</p><p>where # = %& ' ( )* , V -drop volume, σ -surface tension, ηliquid viscosity and K -empirical coefficient close to 10. Interestingly, in the viscous regime linalool accelerates spreading of water on glass, i.e. the effect is inversed as compared to the kinetic regime of spreading when the reduction of the spreading driving force resulted from the decreased surface tension of the linalool solution. This apparent contradiction can be attributed to the volatile behavior of linalool on a longer timescale: evaporation from the drop surface should cause depletion of linalool in the drop close to the triple-phase contact line. Consequently, a negative gradient of the surface tension appears from the wetting front to the middle of the droplet interface (Marangoni Effect), forcing the drop to spread faster. Thus, depending on the time scale of the process (as well as on the concentration of the added volatile surfactant) the volatile amphiphiles can either accelerate or retard the spreading behavior of aqueous formulations. These results may be relevant to the role of surfactants in governing interfacial particle motion. 23 Model jetting experiments (Figure 6) demonstrate the appropriateness of volatile surfactants in controlling the contact line receding phenomena of liquid droplets under evaporation. Printed on polymer substrates pl-small droplets, containing conventional (DX4005N) and volatile surfactants, have been analyzed with regards to their diameter (Figure S3b, Supporting Information), evaporation time (Figure 6b) and receding contact angle (Figure 6c). As clearly seen in Figure 6b, the smaller the primary contact angle the shorter is the droplet evaporation time. Droplets of fluoro-surfactant DX 4005N wet the polymer surface well and evaporate in a pinned state keeping liquid/solid area constant during the whole evaporation time. Thus the reduction of droplet volume leads to a continuously declining contact angle (Figure 6c). Droplets of linalool solution transit within 0,5 s from "constant area" to "constant contact angle" mode. The mixture of both surfactants exhibits wetting behaviour which is a combination of both modes. Therefore, these experiments demonstrate that by mixing surfactants with different surface/interfacial activities, it is possible to tune the dynamic surface tension (Figure S4, Supporting Information) and surface modification properties of the formulations, in particular, to trigger the onset of the pinningdepinning transition of the contact line dynamics and thus the quality and resolution of printed structures. 24,25 Furthermore, an improved functionality of the printed structures, e.g. conducting lines, is expected as a result of the reduced final concentration of the volatile amphiphile.</p><!><p>In summary, aroma molecules are shown to possess functional properties beyond the odor trigger, such as high dynamic activity and volatility. These findings are envisaged to launch multiple applications of these volatile amphiphiles, as well as to advance material science, as well as broad range of technologies ranging from surfaceemerging processes to medical diagnostics 26,27 and to geosciences, e.g. to control the cloud droplet formation. 28 Further work on quantification of the interfacial properties and classification of volatile surfactants from a reach class of aroma (fragrance) molecules is in progress.</p><p>Supporting Information. Details to materials section; results of the quantitative evaluation of the dynamic curves shown in Figure 2a; additional data set to the printing experiments in Figure 6. logP ow -partition coefficient of an amphiphile in a mixture on n-octanol and water</p><!><p>Dynamic surface tension is an important parameter which defines processes and technologies dealing with newly creating interfaces, such as spraying, coating, ink-jet printing, microfluidics, cleaning. Aroma molecules are shown to possess functional properties beyond the odor trigger, such as high dynamic activity at ms time scale and volatility at a scale of seconds, Such dynamic volatile surfactants are envisaged to be of high value in material science and manufacturing.</p><!><p>Oxana A. Soboleva, # Pavel V. Protsenko, # Vadim V. Korolev, # Jekaterina Viktorova, †% Alexey Yakuschenko, † Ruth Kudla, ‡ Jochen S. Gutmann, ‡& and Larisa A. Tsarkova* Evaluation of surface tension according to Ward-Tordai model: Equation 2 has been solved numerically using the code developed by Li at al. [1]. It is assumed that the isotherm of surface tension at equilibrium can be described by semi-empirical Szyszkowski Equation 3which is valid in semi-dilute regime. Parameters K L (a) and Γ max (B) for linalool and benzyl alcohol have been obtained from the surface tension isotherms measured with bubble pressure tensiometer at 20 s surface age time. The approximation of the surface tension to the infinite-time limit γ eq (Table S2) has been done using Equation S1: + ≈ -. /0 12/ 3 ⁄ 5 -. 12/ 3 ⁄</p><!><p>For 1-decanolsolutions parameters B and a of the Szyszkowski Equation have been evaluated from the data published in Ref. [2], were the static surface tension has been measured using ring tensiometer.</p><p>Table S2. Results of the evaluation the kinetic curves in Figure 2b. S3.</p><p>Table S3. Details on the solutions and measurement curves presented in Figure S3 Legend in Figure S4 Temperature of the measurement, °C</p>

ChemRxiv

Antibacterial Properties of Charged TiN Surfaces for Dental Implant Application

The formation and characterization of positively surface charged TiN surfaces were investigated for improving dental implant survival. Surface nitrogen atoms of a traditional TiN implant were converted to a positive charge by a quaternization reaction which greatly increased the antibacterial efficiency. Ti, TiN, and quaternized TiN samples were incubated with human patient subgingival bacteria for 4 hours at 37\xc2\xb0C in an anaerobic environment with an approximate 40% reduction in counts on the quaternized surface over traditional Ti and TiN. The samples were challenged with Streptococcus Mutans and fluorescent imaging confirmed significant reduction in the quaternized TiN over the traditional Ti and TiN. Contact angle measurement and X-Ray Photoelectron Spectroscopy (XPS) were utilized to confirm the surface chemistry changes. The XPS results found the charged quaternized nitrogen peak at 399.75 eV that is unique to the quaternized sample.

antibacterial_properties_of_charged_tin_surfaces_for_dental_implant_application

Introduction<!>Results<!>Confirmation of Surface Reaction<!>Impact of Quaternized TiN Surface on Biofilm Growth<!>Conclusion

<p>Approximately 5% of all dental implants anchored by osseointegration will fail within 10 years.[1] A primary culprit for implant failure is peri-implantitis, which is site specific bacterial infection leading to bone loss around the implant and soft tissue inflammation. Controlling the inflammation through improved hygiene by reducing microbial activity is the only current method to combat potential implant failure, as there is no conclusive treatment to stop infection and progression of the disease.[2] Anywhere from 10% to 45% of all dental implants will have detectable peri-implantitis inflammatory reactions. Characterizing peri-implantitis has proven to be difficult due to many factors such as patients' dental history, medical conditions, hygiene and eating habits.[3] For example, patients with previous history of periodontal disease have lower implant survival rates, but diseases such as osteoporosis or diabetes have not demonstrated consistent effects.[2a,b]</p><p>In addition, the bacterial flora has been explored to determine a prevalence of certain bacteria which may lend to formation of peri-implantitis. The sub-gingival flora of sites with peri-implantitis have similar flora to that of periodontitis with mostly gram-negative bacteria.[4] Persson et al. identified 19 bacterial species with higher counts at the site of peri-implantitis infection than at implant sites without infection, with seven bacteria strains accounting for 30% of bacterial flora at the infection site as compared to 14% at non-infected implants.[4a] These results indiciate that peri-implantitis may not directly result from a singular bacteria strain infecting the implant site, rather an imbalance of a healthy flora.</p><p>Use of antibacterial coatings may provide a method for improving implant lifetime. Even use of different metals in an implant has significant effect on bacterial growth. Titanium nitride (TiN) is the most promising metal as this promotes osseointegration while also having lower bacterial growth rate than traditional titanium (Ti) implants.[5] Current implants utilize several microns thick TiN by plasma spraying coatings on implant surfaces. Despite this, peri-implantitis still exists with these types of implants. Many groups have explored addition of metallic particles to the implant such as copper, silver, and metal oxides;[6] however, a galvanic effect was reported where patients who had gold or amalgam in their teeth concurrently with Ti implants displayed yellow nails and were found to have Ti in their nail clippings.[7] Thus, the additional metallic elements in Ti may enhance Ti and other metals to dissolve into the bodily environment. Primarily, the impact of metal dissolution into the surrounding tissue may facilitate peri-implant inflammatory reactions.[8]</p><p>Molecules with quaternary nitrogen atoms have shown the ability to disrupt the cell wall, leading to leakage of the cell contents and eventual apoptosis of the bacteria.[9] To minimize concerns over dissolute metallic species undergoing unfavorable interactions with the surrounding tissues, modifying the nitrogen atoms on TiN surface into quaternary nitrogen atoms may be a better alternative to inhibit peri-implant inflammatory reactions.</p><p>In this study, we aim to test the effectiveness of quaternized TiN on reducing bio-film growth by employing allyl bromide through the Menschutkin reaction to convert nitrogen atoms on the TiN surface into quaternary nitrogen. Sessile drop contact angle and x-ray photoelectron spectroscopy (XPS) were employed to confirm the nitrogen atoms were converted into quaternary nitrogen. The objectives of this study are to test the hypotheses that: (1) quaternized TiN will have greater antibacterial properties compared with Ti and TiN as a function of colony forming units and (2) bacterial film thickness will affect the anti-microbial properties of these coatings.</p><!><p>ANOVA (R2=.288, adjusted R2=.250, F statistic=7.6 on 5 and 94 degrees of freedom) revealed that coating material is significantly associated with bacteria level (p<.0001). Pairwise comparisons among material groups show that across all thicknesses: (1) Quaternized TiN (Q) is significantly lower than Titanium (T) (p<.0001) and for any random thickness or experimental run, Q is estimated to be 45.6 units lower than T (95% CI=[−65.0, −26.2]); (2) Q is significantly lower than Titanium nitride (N) (p=.0003) and for any random thickness or experimental run, Q is estimated to be 37.1 units lower than N (95% CI=[−56.8, −17.4]); (3) T and N are not significantly different (p=.392) (Figure 1).</p><p>Bacteria thickness is significantly associated with bacteria level (p=.002). Pairwise comparisons among thickness groups show that across all materials: (1) 100 μm yields marginally higher values than 75 μm (p=.060), and for any random material and experimental run, 100μm is estimated to be 18.9 units higher than 75 μm (95% CI=[−0.778, 38.6]); (2) 125 μm yields significantly higher values than 75 μm (p=.0004) and for any random material and experimental run, 125 μm is estimated to be 38.2 units higher than 75 (95% CI=[16.6, 55.7]); (3) 125 μm yields marginally higher values than 100 (p=.082) and for any random material and experimental run, 125 μm is estimated to be 17.3 units higher than 100 (95% CI=[−2.24, 38.8]) (Figure 2). The interaction between coating material and bacterial thickness is p=0.977 indicating that the anti-bacterial effectiveness of the coatings is not expected to be affected by the thickness of bacteria. Experimental run was not associated with bacteria level (p=.450).</p><p>The result from the staining assay shows a significant reduction in the number of live bacteria, on the quaternized TiN samples, as shown by the fluorescence images on Figure 2. The calculation of live bacteria coverage indicate the bactericidal effect of the quaternized TiN samples against S. mutans after 4 h of culture, significantly decreased the number of live bacteria (Figure 4).</p><p>Pairwise comparisons among material groups show regard to live bacteria coverage: (1) Quaternized TiN (Q) (10.85%) is significantly lower than Titanium (T) (89.1%) (p=<.0001); (2) Q is significantly lower than Titanium nitride (N) (82.2%) (p<.0001); (3) T and N are not significantly different (p=.312) (Figure 3).</p><p>Sessile contact angle measurement images of part of a single water droplet placed on Ti, TiN or quaternized TiN surface are shown in Figure 4. For both Ti and TiN substrates, a small contact angle was observed indicating that Ti and TiN surface were hydrophilic and had good wettability (Figure 3). On the contrary, a larger contact angle was obtained on the quaternized TiN surface. Table 1 lists the specific average contact angles and standard deviations of Ti, TiN and quaternized TiN substrates. The average contact angle measurements were 12° for Ti; 16° for TiN and 67° for quaternized TiN substrates. zeta potential measurements were used to characterize the change in surface charge density of the samples. Surface charge density measurements were taken in a slightly acidic solution of pH 5.5. As deposited TiN and quaternized TiN exhibited surface charge densities of 1.19 × 10−7 C/cm2 and 1.81 × 10−7 C/cm2, respectively.</p><p>The stacked XPS survey scans of the TiN surface and the quaternized TiN surface are shown in Figure 5a. The spectra were aligned through the presence of adventitious carbon peak at 284.8 eV from the atmospheric exposure during handling and transport between deposition tools. The general surface chemistry between the two samples is similar as only Ti, O, N, and C were identified in both samples.</p><p>The Nitrogen 1 s spectra acquired at different tilting angles for TiN and quaternized TiN sample are shown for 0° (Figure 5b) and 50° (Figure 5c). The main N peak corresponding to Ti─N bonds is located at 396.75 eV and there is an N satellite peak at 398.48 eV. The quaternized nitrogen, N+, peak is 399.75 eV. As compared to the 0° spectra, surface effects are more pronounced for the tilted spectra, and the penetration depth of the x-ray source is greatly reduced.</p><!><p>The integration and longevity of a dental implant requires the interface between bone and implant remain bacteria free. TiN has been shown to greatly improve development of bone around an implant and decrease the bacteria count over Ti implants.[5] Prior to evaluating the effectiveness of quaternized TiN on biofilm reduction, the success of converting surface bound nitrogen from uncharged to positively charged nitrogen needs to be verified. As shown in Figure 4 and Table I, the contact angles of Ti and TiN substrates were very low. The titanium sample after treatment with hydrogen peroxide was expected to be very hydrophilic because of the presence of Ti-O─H and Ti=O on Ti surface. The TiN sample was also very hydrophilic owing to the high polarity of TiN and the ability for hydrogen bonding of the water droplet to the nitrogen rich surface of TiN. For the quaternized TiN surface, surface nitrogen was converted into a positively charged nitrogen through the Menschutkin reaction and the contact angle should be low. However, as shown in Figure 4 and Table I, the contact angle of the quaternized TiN surface dramatically increased from 16 to 67°. This increase in hydrophobicity is due to the carbon chains extending from the surface bound nitrogen, as illustrated in equation (1) and Figure 3.</p><p>Menschutkin reactions are known to progress very rapidly due to the lower enthalpy of formation of higher substituted amines, which is evident by the reaction saturating within an hour. Hence the Menschutkin reaction is well suited for formation of quaternary ammonium salts and is difficult to stop at secondary or tertiary products.[10]</p><p>XPS surface analysis was employed to further confirm and identify changes in surface chemistry of quaternized TiN surface. As shown in Figure 6, the wide range analyses were acquired from the surface of the TiN and quaternized TiN substrates and Ti, O, N and C were the only elements identified on these two samples. This indicates that the general surface chemistry of TiN and quaternized TiN samples are identical. Therefore, a detailed peak analysis was necessary for confirmation of the surface reaction. High resolution XPS spectra of the N 1s regions were acquired at 0 (Figure 6b) and 50° (Figure 6c) angle with respect to the normal for the TiN and quaternized TiN surface. The main N peak corresponding to Ti─N bonds was located at 396.75 eV and there was an N satellite peak at 398.48 eV. The shoulder peak of N for Ti─N at 394.5 eV was resultant of N on N─Ti-O. The XPS peak[11] for quaternized nitrogen bound to the allyl group (−CH3CH5), N+, was 399.75 eV. Surface effects were more pronounced for the spectra taken at a 50° tilt and the penetration depth of the x-ray source was greatly reduced, compared with the 0° spectra, . For the N spectra of TiN substrate, the peak corresponding to the N of N─Ti-O was more pronounced because of the presence of oxide on the TiN surface. For the N 1s spectra of the quaternized nitrogen, this effect was evident in the increased relative intensity of the quaternized nitrogen peak at 399.75 eV compared with the TiN peak at 396.5 eV at 50°; indicating that N+ atoms are situated on the surface of quaternized TiN. In addition, after a gentle sputtering with 500 eV Ar ions for 30 seconds to remove the top atomic layers of the quaternized TiN substrate, the quaternized nitrogen peak had completely disappeared. This also confirmed that quaternized nitrogen only exists on the top surface of the quaternized TiN substrate.</p><!><p>As previously stated, the oral microbiome is incredibly diverse, and no singular bacteria has been identified as the primary causating agent of peri-implantitis. Thus, preventing all types of microbial growth near the implant site may provide the best avenue for implant survival. The variance experienced with efficacy is expected as each patient's flora will vary day-to-day and may respond differently depending on environmental factors such as diet, hygiene, periodontal condition, etc.</p><p>This study shows the potential anti-microbial effect of quaternized TiN compared with Ti and TiN at reducing CFUs (Figure 2) and bacteria coverage (Figure 3). The antibacterial mechanism of quaternized nitrogen has been proposed to interact with the cell wall, destroy the cytoplasmic membrane leading to the leakage of intracellular components and consequent cell death.[12] The quaternized surface outperformed both the TiN and Ti surfaces significantly within the short 4-hour testing period for bacteria film thickness of 75 μm. Even with thick bacteria films that are 125μm thick, the quaternized surfaces still significantly reduced bacterial counts compared with the other two substrates, indicating that the surface has the ability to neutralize bacteria that are not directly in contact with the surface. The lack of statistical significance in interaction between coating type and thickness indicates that the superior anti-bacterial properties of quaternized TiN versus TiN and Ti is evident across different thicknesses of bacteria. Also, the fluorescent images showed the proliferation of live bacteria was lower in quaternized TiN compared to Ti and TiN. Implant surfaces should have the ability to affect bacteria that is not only in direct contact with the surface but also in the immediate vicinity. This is because during the process of osseointegration, the implant will initially be surrounded by saliva and soft tissue. Additionally, failure of an implant will be induced by an infection beginning near the soft tissue which will propagate downward towards the base of the implant. However, with the quaternized surface that propagation can be hindered or completely prevented.</p><!><p>A novel application of the well-known Menschutkin reaction has been applied to convert the surface nitrogen of a TiN coated implant from neutral to positively charged. The surface change was monitored and confirmed by Sessile contact angle and XPS measurements. Biofilm growth noted a 40–50% reduction in bacteria over traditional implant surfaces within 4 hours and noticeable effects many microns away. Considering current technology and other works pursuing high biocidal activity for implant structures, this methodology provides a simple method that would require little manufacturing line changes to accommodate and bring to market.</p>

PubMed Author Manuscript

RNA-Seq reveals placental growth factor regulates the human retinal endothelial cell barrier integrity by transforming growth factor (TGF-\xce\xb2) signaling

Placental growth factor (PlGF or PGF) is a member of the VEGF (vascular endothelial growth factor) family. It plays a pathological role in inflammation, vascular permeability, and pathological angiogenesis. The molecular signaling by which PlGF mediates its effects in nonproliferative diabetic retinopathy (DR) remains elusive. This study aims to characterize the transcriptome changes of human retinal endothelial cells (HRECs) with the presence and the absence of PlGF signaling. Primary HRECs were treated with the PlGF antibody (ab) to block its activity. The total RNA was isolated and subjected to deep sequencing to quantify the transcripts and their changes in both groups. We performed transcriptome-wide analysis, gene ontology, pathway enrichment, and gene-gene network analyses. The results showed that a total of 3760 genes were significantly differentially expressed and were categorized into cell adhesion molecules, cell junction proteins, chaperone, calcium-binding proteins, and membrane traffic proteins. Functional pathway analyses revealed that the TGF-\xce\xb2 pathway, pentose phosphate pathway, and cell adhesion pathway play pivotal roles in the blood-retina barrier (BRB) and antioxidant defense system. Collectively, the data provide new insights into the molecular mechanisms of PlGF\xe2\x80\x99s biological functions in HRECs relevant to DR and diabetic macular edema (DME). The newly identified genes and pathways may act as disease markers and target molecules for therapeutic interventions for the patients with DR and DME refractory to the current anti-VEGF therapy.

rna-seq_reveals_placental_growth_factor_regulates_the_human_retinal_endothelial_cell_barrier_integri

Introduction<!>Primary human retinal endothelial cells (HRECs) culture<!>Cell treatments<!>RNA extraction<!>RNA sequencing<!>RNA-Seq bioinformatics analysis<!>Functional annotation<!>Gene-gene network analysis<!>Statistical analysis<!>RNA sequencing<!>Functional annotation<!>Pathway-focused gene interaction network analysis<!>Gene-interaction network analysis of genes within the pentose phosphate pathway<!>Gene-interaction network analysis of genes within the TGF-\xce\xb2 signaling pathway<!>Discussion<!><!>Conclusion<!>

<p>Diabetic Retinopathy (DR) is the most common complication of diabetes and the leading cause of visual damage in the working-age adult population worldwide [1]. Increased glucose levels can lead to microvascular impairment in the retina. However, the mechanism by which hyperglycemia initiates retinal blood vessel damage in retinopathy remains intangible. It is well established that oxidative stress and inflammation play pivotal roles in the pathological process, consequently resulting in the breakdown of the blood-retinal barrier (BRB) [2]. An impaired BRB can further grow into augmented vascular retinal permeability and unbalanced growth of new blood vessels, thus prompting moderate to severe vision loss. When DR progresses into further advanced stages, vitreous hemorrhages can occur and result in retinal detachment and fibrovascular contraction, further advancing to vision loss, and eventually leading to blindness [3].</p><p>The identification of VEGF (vascular endothelial growth factor) as a crucial stimulus for proliferative diabetic retinopathy (PDR) and diabetic macular edema (DME) leads to the development of anti-VEGF therapies, which have improved the clinical management of these disease conditions. However, repeated anti-VEGF therapy in severe PDR and DME patients can lead to the formation of fibrovascular membranes (FVM) and tractional retinal detachment, which can cause severe vitreoretinal traction and hemorrhage [4]. Another study has also found that long-term intravitreal anti-VEGF injection may have detrimental effects on the neuronal cells, based on the survival and maintenance function of VEGF [5]. Thus, there is a clear need for alternative therapies for DR, especially for the non- or poor-responders to the current standard of care, with the potential to reduce the risk of treatment-related complications. Indeed, other crucial angiogenic growth factors are likely involved in this condition [6], which could serve as additional or even alternative targets for therapy. Cumulative evidence supports the pathogenic role of PlGF (placental growth factor), which is a member of the VEGF family and it was first identified from a human placental cDNA library in 1991 [7] [8]. PlGF is expressed in a number of cell types, including endothelial cells (ECs) and retinal pigment epithelial cells (RPEs) in response to hypoxia [9]. Like VEGF, PlGF can bind to fms-like tyrosine kinase-1 (FLT1), vascular endothelial growth factor receptor-1 (VEGFR-1), or its soluble isoform. The circulating isoform of sVEGFR-1 lacks the transmembrane and intracellular domains. The PlGF activates the FLT1 that affects the VEGF-VEGFR2 signaling, suggesting synergistic interactions of PlGF and VEGF. It can also bind to VEGF and form heterodimers [10] and employs a strong effect on blood vessel maturation and growth. Also, PlGF has direct proangiogenic effects on ECs [11]. Previously, we described that deleting PlGF in mice leads to reduced expression of diabetes-activated hypoxia-inducible factor (HIF)1α in the retina. We also found changes in the VEGF pathway as well as expressions of VEGF, VEGFR1–3, HIF1α, and levels of phospho (p)–endothelial NOS (nitric oxide synthase), (p)-VEGFR1, and p-VEGFR2 in the retinas of diabetic PlGF knockout mice [12].</p><p>Despite the advances made, the biological function of PlGF and associated mechanisms are less well understood than those of its homolog VEGF. To bridge this gap, we sought to investigate the roles of PlGF in the EC barrier function and to elucidate the underlying molecular mechanisms. In one of the most recently completed studies [13], we found that PlGF inhibition promotes human retinal endothelial cell (HREC)'s barrier function, as indicated by increased resistance, upregulated expression levels of tight and adherens junction proteins, and their reinforced cell membrane localization. PlGF inhibition also prevents HREC's barrier dysfunction caused by high glucose, indicative of implication in DR. Additionally, we found that PlGF inhibition improves EC barrier function through activation of glucose-6-phosphate dehydrogenase (G6PD) and pentose phosphate pathway (PPP), as well as upregulation of antioxidant proteins.</p><p>In the present study, we identified transcriptome and pathways that are regulated by PlGF using the comparative transcriptomic analysis between the PlGF antibody (ab) treatment and the phosphate-buffered saline (PBS) control in the primary HREC. We provide a thorough report about the differentially expressed genes (DEGs) in HREC with the presence and absence of PlGF signaling and highlight several signaling pathways regulated by PlGF. Notably, the expression levels of many genes involved in the TGF-β signaling pathway, which acts as a primary regulator of many other identified genes and pathways, are altered in the PlGF ab treatment relative to the PBS control.</p><!><p>Primary HRECs were bought from Cell Systems (Cat#: ACBR1 181, Kirkland, WA). HRECs were seeded on fibronectin-coated (10 μg/ml, overnight, 33016015, Gibco) plastic culture vessels and grown using the EBM2-MV medium (Cat#: cc-4176, Lonza, Walkersville, MD) supplemented with 10% fetal bovine serum (FBS) (Cat#: SH3039603HI, Fisher Scientific, Waltham, MA), 1% of penicillin/streptomycin (P/S), and EGM MV Singlequots growth supplement kit (Cat#: cc-4147, Lonza). Cells were used during passage 5 to 6.</p><!><p>At 80% confluence, the culture media was replaced with fresh media containing mouse anti-PlGF antibody (PL5D11D4), (25 μg/ml), and HRECs were collected 48 hours after the start of the incubation. PBS treated cells were used as a negative control. The primary HERC cultures treated with the validated neutralizing PlGF antibody have bearing on diabetic retinopathy (DR) in terms of two aspects. First, primary HREC cultures form a mono cell layer with barrier function maintained by the tight junctions and adherin junctions, such as ZO-1, Claudin 5, Occludin-1, and VE-Cadherin. These characteristics are highly relevant to DR because they can be disrupted by hyperglycemia and their expression levels represent whether retinas develop diabetic complications to some extent. Second, PlGF antibody treatment promotes HREC barrier function not only in normal glucose conditions but also in high glucose condition (25mM D-glucose); the HREC with this glucose concentration is widely accepted in vitro model to study DR [13].</p><!><p>Forty-eight hours after treatment, the cells were washed with PBS, and total RNA was extracted using the Qiagen mini RNA preparation kit (Qiagen) according to the manufacturer's protocol. RNA concentration was determined using a NanoDrop spectrophotometer (Thermo Scientific). RNA quality was determined using the Agilent bioanalyzer 2100 (Agilent Technologies). The analysis showed clear, defined 28s and 18s rRNA peaks, an indication of high-quality RNA. RIN value ≥ 8 was set as the cut-off for sample inclusion for downstream processing for RNA sequencing analysis.</p><!><p>RNA samples (6 PBS controls vs. 6 PlGF ab treatment) were submitted to Novogene Leading Edge Genomic Services & Solutions, California, USA, for sample preparation and sequencing. The DNase-treated to samples initially and evaluated for total RNA quality using the Agilent 2100 Bioanalyzer, subsequently 2 rounds of polyadenylate positive (poly A+) selection and conversion to cDNA. The Illumina HiSeq 2500 using for RNA sequencing with the latest versions of sequencing reagents and flow cells, providing up to 300 GB of sequence data per flow cell. In addition to that, the TruSeq library generation kits were used, giving to the manufacturer's instructions (Illumina) [14]. Library construction comprised of random fragmentation of the poly A+ mRNA, then cDNA production using random primers. The ends of the cDNA were repaired, adaptors and A-tailed ligated for indexing (up to 12 different barcodes per lane) in the sequencing run. The cDNA libraries were quantitated using qPCR in a Roche LightCycler 480 with the Kapa Biosystems kit for library quantitation (Kapa Biosystems) before cluster generation. Clusters were created to produce approximately 725K–825K clusters/mm2. Cluster quality and density were determined during the run after the first base addition parameters were measured. Paired-end 2 × 50 bp (basepair) sequencing runs were done to align the cDNA sequences to the reference human genome. Almost 15 million paired 50 bp reads were obtained per sample.</p><!><p>The raw data were used for visualization of reads quality before and after pre-processing using FastQC software: (https://www.bioinformatics.babraham.ac.uk/projects/fastqc/) [15]. Then, the reads were processed to remove adapters, and ambiguous quality reads using the Trimmomatic-0.36 tool [16] [17] with trimming of bases from 3′ and 5′ end, maintaining the Phred-score ≤ 30. The human genome was downloaded from the National Centre for Biotechnology Information (NCBI) genome (https://www.ncbi.nlm.nih.gov/genome/?term=human) for reference-based assembly. We mapped all the datasets onto the human genome reference sequence GRCh38 using TopHat2.0.9 (http://ccb.jhu.edu/software/tophat/index.shtml). The expression levels were transformed into fragments per kilobase of exon per million mapped fragments (FPKM). We identified DEGs that satisfy the significance expressed as q-value representing FDR-adjusted P value < 0.05 by using Cufflinks 2.1.1 (http://cufflinks.cbcb.umd.edu). The Bioconductor tool with the CummeRbund package was employed to analyze differential expression analysis in the assembled transcriptome (http://compbio.mit.edu/cummeRbund/manual_2_0.html). Finally, both control and PlGF ab treated comparison transcript counts (matrix file) were used for differential gene expression using the CummeRbund package of Bioconductor with primary parameters such as FDR (false discovery rate), logFC (log fold-change), logCPM (log counts per million), and p-value. Unigenes with adjusted q-values of less than 0.05 (p<0.05) and the fold change of more than 2 (logFC>2) were considered as significantly differentially expressed genes.</p><!><p>Gene ontology (GO) Enrichment Analysis (http://geneontology.org/page/go-enrichment-analysis) and DAVID annotation (https://david.ncifcrf.gov/) were used for functional annotation and pathways analysis. An adjusted EASE (Expression Analysis Systemic Explore Score) score of 0.05 and a threshold count of >2 genes were employed. Benjamin–Hochberg multiple testing correction was applied to the p-values. GO terms with FDR q<0.05 were considered significantly enriched within the gene set [16, 18].</p><!><p>We performed the protein-protein network analysis for all DEGs using the STRING 10.5 database (https://string-db.org/). This is calculated in a variety of classification systems (Gene Ontology, KEGG, Pfam, and InterPro), and used the Fisher's exact test followed by a correction for multiple testing [19].</p><!><p>All numeric values were expressed as the mean ± standard deviation (SD) for the respective groups. Statistical analyses were performed using the cummeRbund R-package (https://www.bioconductor.org/packages/release/bioc/html/cummeRbund.html). Fisher's exact test and Benjamin–Hochberg corrections (FDR) were also used in the analyses of the two group comparisons. A p-value of less than 0.05 was considered significant.</p><!><p>The cuffdiff results file was statistically analyzed and visualized using the cummeRbund package. The quality of the model fitting dispersion plot (Fig. 1A) showed the highest number of differentially expressed genes (DEGs). In addition, we calculated the distributions of FPKM (Fragments Per Kilobase of transcript per Million) scores across samples using the Density (Fig. 1B) plot and Box plot (Fig. 1C) respectively. We are visualized the coding sequence length (bp) (Fig. 1D), transcript length (bp) (Fig. 1E), genome span (bp) (Fig. 1F), 5' UTR length(bp) (Fig. 1G), 3' UTR length(bp) (Fig. 1H), and percentage of the GC content(Fig. 1I) of DEGs were predicted. In addition, the distribution of DEGs was predicted based on the human chromosomes, distribution of gene type, the number of exons (coding genes), and the number of transcript isoforms per coding gene (Fig. 2). We identified 53808 significant transcripts in total from the datasets. Among them, a total of 3760 differentially expressed genes (1750 up-regulated and 2010 down-regulated genes) that satisfy q-value (FDR-corrected p-value) < 0.05 and fold change ≥ ±2.0, were identified in the PlGF ab treated group relative to the control. A hierarchical clustering heatmap (Fig. 3) and a volcano plot (Fig. 3A, Huang et al., 2019 [13]) are generated to represent the up- and down-regulated genes. These DEGs were used for further gene ontology and functional pathway analysis.</p><!><p>DEGs were used for GO enrichment analysis using the DAVID annotation tool [20] with the complete human genome as the background. The following GO terms were enriched: molecular function (MF), biological process (BP), cellular component (CC), and protein classes (PC). Most of the DEGs were found to be involved in several molecular functions, such as binding (GO:0005488), catalytic activity (GO:0003824), and translation regulator activity (GO:0045182), respectively (Fig. 4A). The DEGs were also involved in various cellular components, i.e., cell (GO:0005623), organelle (GO:0043226), and protein-containing complex (GO:0032991) respectively (Fig. 4B). The DEGs were also involved in various biological processes, i.e., biological regulation (GO:0065007), cellular processes (GO:0009987), and metabolic process (GO:0008152), respectively (Fig. 4C). Most of the DEGs are classified as nucleic acid binding (PC00171), hydrolase (PC00121), transcription factor (PC00218), and enzyme modulator (PC00095) (Fig. 4D) respectively.</p><p>The gene ontology results revealed that the 36.9% of the genes were involved in binding (GO:0005488), 37.1% of genes were involved in catalytic activity (GO:0003824), and 6.9% of genes were involved in transcription regulator activity (GO:0140110) of molecular functions. A total of 44.5% of genes participated in the cell (GO:0005623), 34.0% of genes participated in the organelle (GO:0043226), and 11.4% of genes participated in the protein-containing complex (GO:0032991) of cellular components. A total of 32.8% of the genes were involved in cellular processes (GO:0009987), 24.6% of genes were involved in the metabolic process (GO:0008152), 15.9% of genes were involved in biological regulation (GO:0065007) and 11.0% of genes were involved in localization (GO:0051179) of biological processes. Finally, 16.2% of genes belonged to nucleic acid binding (PC00171), 11.9% of genes belonged to hydrolase (PC00121), 11.2% of genes belonged to the transcription factor (PC00218), 10.0% of genes belonged to enzyme modulator (PC00095), 9.4% of genes belonged to transferase (PC00220), 5.5% of genes belonged to cytoskeletal protein (PC00085), and 5.0% of genes belonged to the transporter (PC00227) of protein classes.</p><p>The functional enrichment results revealed genes (Fig.3B, Huang et al., 2019 [13]) mainly involved in the pentose phosphate pathway (Table 1), TGFβ signaling pathway (Table 2), as well as cell adhesion and antioxidant genes that we have identified and tabulated (Table 3); these are all up-regulated in PlGF ab treated conditions compared with the control. These results suggested that most of the genes involved in the TGFβ signaling pathway might have a beneficial role in diabetic retinopathy.</p><!><p>We performed the gene-gene network analysis of genes within each pathway using the STRING tool (https://string-db.org/). Based on functional enrichment analysis, here we focused on gene-gene interaction network analysis on the genes identified to be involved in the pentose phosphate pathway, carbon metabolism pathway, p53 signaling pathway, apoptosis pathway, pyrimidine metabolism pathway, ubiquitin-mediated proteolysis pathway, TGF-β pathway, and glycolysis pathways respectively.</p><p>The pentose phosphate pathway has 11 nodes, 33 edges, 6 average node degree, 0.842 avg. the local clustering coefficient, and 9 expected number of edges with a PPI enrichment p-value < 1.0e-16 (Suppl. Fig. 1A). The carbon metabolism pathway has 30 nodes, 195 edges, a 13 average node degree, 0.749 avg. the local clustering coefficient, and 9 expected number of edges with a PPI enrichment p-value < 1.0e-16 (Suppl. Fig. 1B). The p53 signaling pathway has 22 nodes, 111 edges, 10.1 average node degree, 0.75 avg. local clustering coefficient and 15 expected number of edges with a PPI enrichment p-value < 1.0e-16 (Suppl. Fig.1C). The apoptosis pathway has 31 nodes, 207 edges, a 13.4 average node degree, 0.728 avg. the local clustering coefficient, and 30 expected number of edges with a PPI enrichment p-value < 1.0e-16 (Suppl. Fig. 1D). The pyrimidine metabolism pathway has 27 nodes, 159 edges, 11.8 average node degree, 0.715 avg. the local clustering coefficient, and 12 expected number of edges with a PPI enrichment p-value < 1.0e-16 (Suppl. Fig. 2A). The ubiquitin-mediated proteolysis pathway has 15 nodes, 56 edges, a 7.47 average node degree, 0.717 avg. the local clustering coefficient, and 8 expected number of edges with a PPI enrichment p-value < 1.0e-16 (Suppl. Fig. 2B). The TGFβ signaling pathway has 37 nodes, 162 edges, an 8.76 average node degree, 0.606 avg. the local clustering coefficient, and 47 expected number of edges with a PPI enrichment p-value < 1.0e-16 (Suppl. Fig. 2C). The glycolysis pathway has 6 nodes, 15 edges, a 5.0 average node degree, 1 avg. the local clustering coefficient, and 1 expected number of edges with a PPI enrichment p-value < 1.0e-16 (Suppl. Fig. 2D). The results revealed that all pathway genes interact with each other directly or indirectly, except the EIF2S1 gene in the apoptosis pathway.</p><!><p>Given our previous observation about the effect of PIGF on glucose-6-phosphate dehydrogenase (G6PD) and the antioxidant system [13], here we focused our analysis on the genes identified in the pentose phosphate pathway. Evidenced-based analysis of the genes showed that all the genes were inter-connected, and ALDOC was the query protein and first shell of interactors. Interestingly, G6PD, a gene that we previously found to be modulated under high glucose and by PlGF, was shown to interact with many genes such as PFKM, ALDOA, RPE, PGLS, and ALDOC (Suppl. Fig. 3A). An interaction analysis based on molecular action was also performed to gain insights into how these genes affect each other, and the results reveal that G6PD binds directly with PRPS1, PGLS, GPI, ALDOA, and ALDOC (Suppl. Fig. 3B). Based on confidence analysis, G6PD showed the highest interaction with GPI, RPE, PGLS, ALDOA, ALDOC, and high interaction with PRPS (Suppl. Fig. 3C). Reactome pathway analysis revealed that glycolysis and TP53 regulate metabolic genes and are among the enriched pathways, with FDR of 1.30e-09 and 0.0023, respectively. In addition, G6PD was involved in three of the 6 enriched Reactome pathways. The clustering of the 11 genes involved in the pentose phosphate pathway revealed two distinctive clusters. In one cluster, genes, G6PD, PGLS, RPE, and GP1 were grouped, while PFKL, PFKM, ALDOC, ALDOA, DERA, and GLYCTK clustered together. PRPS1 was the only gene that did not cluster with any of the above (Suppl. Fig. 3D).</p><!><p>TGF-β signaling has been implicated in the pathophysiology of DR, for instance during the thickening of the capillary basal lamina, mediated via pericytes. Here, we performed an in-depth analysis of the genes identified from the pathway analysis involved in TGF-β signaling to understand how these genes interact with each other in modulating cell behavior after antibody-mediated inhibition of PlGF in vitro. The evidence-based analysis identified integrin beta-3 (ITGB4) as the query protein and the first shell of interactors (Fig. 5A), a protein thought to play a role in the hemidesmosome of epithelial cells. Analysis based on molecular function revealed that ITGB4 binds directly to PTK2, MET, ITGB3, CAV1, and SHC1, which in turn inhibit ZEB1 (Fig. 5B), and a confidence-based interaction analysis revealed that ITGB4 binds strongly with the above proteins (Fig. 6A). Cluster analysis of all the genes involved in the TGF-β pathway revealed three distinctive clusters of genes; Cluster 1 (green), cluster 2 (red), and cluster 3 (Dark cyan) (Fig. 6B). The volcano plot showed fold change and p-value of all TGFβ signaling pathway genes. The green color indicates down-regulated genes and the red color indicates up-regulated genes (Fig. 7).</p><p>The evidence-based analysis reveals that the ITGB4 binds directly to PTK2, MET, ITGB3, CAV1, and SHC1 genes, which in turn inhibit E-box-binding homeobox1 (ZEB1) in upstream condition but gene expression analysis results showed the PTK2, MET, ITGB3, CAV1, and SHC1 genes are down-regulated. It means the PTK2, MET, ITGB3, CAV1, and SHC1 genes are negatively up-regulate the ZEB1 gene. Overall, our results indicate that the ZEB1 gene may activate the transforming growth factor-β (TGFβ) signaling pathway.</p><!><p>To reveal functional and molecular events in HREC with the presence and the absence of PlGF signaling, we performed a global transcriptome analysis to identify DEGs and functional enrichment analysis based on Illumina HiSeq2500 sequencing. To our knowledge, this is the first transcriptome-wide analysis to uncover DEGs that are expressed in presence and the absence of PlGF signaling. We anticipate that these data will be a powerful resource for future investigations into the effects of PlGF signaling responses in HREC cells.</p><p>The DEGs identified in the PIGF ab treated group relative to the control were further characterized and annotated by gene ontology and functional enrichment analysis. The downstream genes and pathways regulated by PlGF are potentially involved in the biological functions of PlGF, such as angiogenesis and EC barrier function.</p><p>One interesting class of genes are those involved in the pentose phosphate pathway and TGFβ signaling pathway. Our previous studies [13] [21] already disclosed the many attractive candidates of PPP pathway and their role in the activation of the antioxidant defense system in DR. The focus of the present study is TGFβ signaling pathway genes and their role in the activation of the antioxidant defense system in DR. The functional enrichment results revealed that the AKT1, APP, CAV1, CCND1, CDKN1A, DAB2, ETS1, FN1, HDAC1, ITGB3, ITGB4, KLF11, LIMK2, MAPK9, MEF2C, MET, NEDD9, PML, PRKAR2A, PTK2, RAF1, ROCK1, SHC1, SMAD2, SMAD4, SNW1, SOS1, SPTBN1, STRAP, TGFB1I1, TGFBR3, TRAP1, UCHL5, WWP1, YAP1, ZEB1, and ZEB2 genes are involved in the TGFβ signaling pathway. TGFβ1 is a signaling protein involved in many processes, including immune system modulation, cell proliferation, cell differentiation, and apoptosis [22]. After activation, TGF-β binds to the type 2 TGF-β receptor, and this interaction leads to recruitment and subsequential phosphorylation of type 1 TGF-β receptor. In turn, the intracellular proteins Smad2 and Smad3 are recruited, and after forming a complex with Smad4, TGF-β translocates into the nucleus, where it activates downstream gene transcription [23].</p><p>There is evidence supporting that TGF-β1 can directly affect endothelial cell permeability such as in the blood-brain barrier (BBB) and BRB. Behzadian et al., (2001) [24] reported that TGF-β1 increases the retinal endothelial cell permeability by activation of MMP9 expression. Our results reveal that the MMP14 gene is up-regulated, which is in agreement with the observations by other investigators. A previous study reported that TGF-β1 signaling is complicated with autocrine and paracrine signaling conveyed by and for multiple cell types and with extensive detection of various receptor and ligand expressions [25]. Similarly, recent studies demonstrated that TGF-β1 signaling is required for both axon formation and migration [26]. The experimental results by Brionne et al. (2003) [27] showed that neural degeneration occurs in the brain of the TGF-β1 knockout mice. TGF-β1 is a multifunctional growth factor that is a well-established modulator of vascular cell integrity and function [28]. TGF-β1 is activated when endothelial and mesenchymal cell contact, and it is involved in the inhibition of EC proliferation and migration, vessel maturation, production of basement membranes, and induction of pericyte differentiation [29]. Pericytes and astrocytes may also release TGF-β1, which contributes to BBB/BRB integrity and function [30]. Recent studies have demonstrated that contact between endothelium cells (ECs) and pericytes or astrocytes leads to TGF-β1 activation (up-regulated), a major determinant of TGF-β1 availability and signaling [31]. Moreover, the loss of retinal pericytes has been speculated to be permissive for the progression of diabetic retinopathy [32]. These findings support the notion that the TGF-β1 signaling keeps the retinal microvascular integrity due to the high pericytes coverage. Our findings correlate well with the literature reports. Braunger et al. (2013) [33] described that the TGF-β1 signaling exerts several functions like the promotion of neuronal differentiation and the maintenance of neuronal survival in the retina. Shen et al. (2011) [34] demonstrated that TGF-β1 stimulus increases BBB permeability of both human brain ECs and bovine retinal ECs. The increased BBB permeability can be caused by tyrosine-phosphorylation of both VE-cadherin and claudin-5. TGF-β1signaling plays an essential role in the differentiation of vascular smooth muscle cells/pericytes at mid-gestation, as revealed by gene knockout studies on the signal components, such as Tgfbr2, TGF-β1, endoglin, Alk1, Alk5, Smad5, and Smad4 [35]. Our findings suggest that the SMAD2 and SMAD4 genes are down-regulated in the TGF-β1signaling pathway. This might have a significant role in dysregulation of the ubiquitin-proteasome pathway, angiogenesis pathway, p53 signaling pathway, and apoptosis pathway genes, as well as in up-regulation of the many cell-cell adhesion genes like LGALS3, FBN2, FBN1, LTBP4, GPNMB, CELSR1, ITGB4, TMEM8B, MFGE8, ITGB3, SDC1, MAGED4B, MPZL1, ITGB2, CCBE1, and MAGED2, and increase of the BBB permeability.</p><p>The other genes of interest from the pentose phosphate pathway and antioxidant defense system, the glycolysis, and carbon metabolism genes, play a beneficial role in diabetes-related oxidative damage to retinal cells (diabetic retinopathy). Oxidative stress is triggered by an imbalance between the production of reactive oxygen species (ROS) and the loss of antioxidant defense components [36] [37]. G6PD gene plays a key role in regulating carbon flow through the pentose phosphate pathway. Specifically, the enzyme affects the production of the reduced form of the extramitochondrial nicotine adenosine dinucleotide phosphate (NADPH) coenzyme by controlling the conversion from glucose-6-phosphate to 6-phosphogluconate in the pentose phosphate pathway. In red blood cells, defense against oxidative damage is heavily dependent on G6PD activity, which is the only source of NADPH [38] [21]. Our outcomes propose that the G6PD gene is up-regulated in the PlGF ab treated condition. It stimulates the oxidative branch of PPP to supply cytosolic NADPH to counteract oxidative damage as well as up-regulating antioxidant genes such as Peroxiredoxin (Prdx)1, Prdx3, and Prdx6. Prdxs are highly conserved and small molecular weight (20–30 kDa) thiol peroxidases that scavenge alkyl hydroperoxides, hydrogen peroxide (H2O2), and peroxynitrite inactive cells [39]. Prdxs play a pivotal role in the protection of cells from oxidative stress. Furthermore, cumulative evidence supports that Prdxs can also perform as redox sensors in the condition of oxidative stress and the local accumulation of hydrogen peroxide [40]. In the retina, after expressing of Prdx6, the Müller cells and astrocytes display central roles in the maintenance of the BRB (blood-retinal barrier) function. Moreover, several studies found that the blood-retinal barrier is compromised due to decreased Prdx6 in several disease conditions like diabetic retinopathy (DR), age-related macular degeneration (AMD), arterial and venous occlusions [16] [39] [41]. Our previous proteomics studies also reported that PlGF absence increases neuroprotective and antioxidant proteins in the diabetic mouse retina such as Prdx6 and Map2 [16]. Recent studies on tears from patients with glaucoma have also identified Prdx1 as having possible involvement in inflammation [42]. Furthermore, apart from their role as antioxidants, the peroxiredoxins can affect a diverse range of biological processes that include cellular proliferation, differentiation, and apoptosis by influencing signal transduction pathways that employ hydrogen peroxide as a secondary messenger [43]. The above reports relavant to our results and suggest that Prdx1, Prdx3, and Prdx6 play a role in defending oxidative stress in DR. Based on all together, the relevance and possible regulatory role of TGFβ pathway DEGs increased BBB permeability. However, further work must be perform to verify the function and relationship between TGFβ signaling and antioxidant defence process in BBB permeability.</p><!><p>The present study only presented the transcriptome and pathway differences between HREC cultures treated with PlGF antibody and the cultures with the presence of PlGF signaling. Although some of the differentially expressed genes, such as G6PD, PRDX3, PRD6, and VECadherin, were validated on the mRNA and/or protein levels and reported in our publications [13] [21], additional validation results of RNA seq and bioinformatics results would corroborate our conclusions, particularly for those TGF-β pathway genes.</p><p>HREC cultures were used for experimental models to investigate the molecular pathways modulated by PlGF signaling and relevant to diabetic retinopathy. Although primary HREC culture is a widely accepted in vitro model, some more physiologically relevant model, such as vascular organoids and transgenic mice, would be used for further study on the function of the candidate genes identified through RNA seq and bioinformatics analyses.</p><p>We used the PlGF neutralizing antibody (PL5D11D4), which was developed to block mouse PlGF signaling through the interaction with VEGFR1 [44]. Characterization of this antibody and effects on cancer and eye disease have been performed in both rodent and human models [45]. We validated that this antibody can bind with human PlGF and also observed the clear biological effects on HREC integrity and function1. Despite these characterizations, whether this antibody can interfere with the interaction of human PlGF and VEGFR1 and what concentrations can effectively block its downstream signaling are to be defined.</p><p>The effects of PlGF on HREC barrier function are explicit, however, it is unclear why and how the neutralizing antibody to PIGF would be expected to have effects on the HREC. Given PlGF is expressed by HREC and can be regulated by VEGF [46], it is most likely that PlGF acts on HREC in an autocrine fashion.</p><!><p>Our results demonstrated that neutralizing PlGF regulates a variety of gene expressions that are relevant to its pathophysiological functional roles, such as angiogenesis and EC barrier function. Among the most important ones are those genes involved in TGFβ, PPP, and the antioxidant defense system. These newly identified genes and pathways may act as potential target molecules for therapeutic interventions for those patients with DR refractory to the current anti-VEGF therapy.</p><!><p>Figure S1: Pentose phosphate pathway, Carbon metabolism pathway, p53 signaling pathway apoptosis pathway genes interactions sub-network. Figure S2: Pyrimidine metabolism pathway, Ubiquitin-proteasome pathway, TGF-beta signaling pathway, Glycolysis metabolism pathway genes interactions sub-networks. Figure S3: Interaction network analysis of genes involved in the pentose phosphate pathway.</p>

PubMed Author Manuscript

Tuning cyclometalated gold(III) for cysteine arylation and ligand-directed bioconjugation

Transition metal-based approaches to selectively modify proteins hold promise in addressing challenges in chemical biology. Unique bioorthogonal chemistry can be achieved with preformed metal-based compounds, however, their utility in native protein sites within cells remain underdeveloped. Here, we tune the ancillary ligands of cyclometalated gold(III) as a reactive group and the gold scaffold allow for rapid modification of a desired cysteine residue proximal to the ligand binding site of a target protein. Moreover, evidence for ligand association mechanism toward C-S bond formation by X-crystallography is established. The observed reactivity of cyclometalated gold(III) enables the rational design of a cysteine-targeted covalent inhibitor of mutant KRAS. This work illustrates the potential of structure-activity relationship studies to tune kinetics of cysteine arylation and rational design of metal-mediated ligand affinity chemistry (MLAC) of native proteins.

tuning_cyclometalated_gold(iii)_for_cysteine_arylation_and_ligand-directed_bioconjugation

Introduction<!>Cyclometalated gold scaffold optimization, rationale, and synthesis<!>Kinetics of Au(III)-mediated cysteine arylation<!>Evidence for ligand associative substitution<!>Electrochemical characterization<!>Design and Synthesis of Covalent Inhibitors for KRAS (G12C)<!>Summary and conclusion<!>Materials and methods:<!>Physical measurements:<!>Synthetic procedures :<!>Rate constant determination (Kinetics):<!>X ray crystallography:<!>Electrochemical studies:<!>Peptide and protein labeling studies using electron spray mass spectrometry:<!>In Silico experiment:<!>Stability in PBS and DMEM at 37 \xc2\xb0C using UV - Vis:<!>Cell culture :<!>Cell viability studies :<!>Cellular uptake:

<p>Site-selective protein modification using specifically designed small-molecules is a robust tool in drug discovery and chemical biology1. The design of small molecules often interact with proteins covalently or non-covalently. Compounds that covalently modify proteins maintain strong and stable interactions. Covalent modifiers are attractive for therapeutic applications due to optimal stability of the overall protein-ligand interaction. Strategies for designing targeted covalent modifiers include designing unique ligands with affinity for specific protein sidechains and introducing a chemo or regioselective reactive handle for target binding. 2, 3</p><p>The nucleophilic amino acid side chains in proteins such as cysteine, lysine, histidine, tyrosine, and tryptophan provide the ground to develop reactive handles for covalent modification. Among these amino acids, cysteine is intrinsically nucleophilic. The innate nucleophilicity in physiological pH and less abundance nature of cysteine makes it easier to develop a covalent modifier with electrophiles. 2, 4</p><p>Strategies for developing agents that target cysteine frameworks include the incorporation of electrophiles into chemical probe designs. In the past, significant number of organic chemoselective cysteine modifiers (i.e. maleimides, α-halocarbonyls, disulfides, and acrylamides)2 have been developed by several researchers. However, their limitations in stability, rate of the reaction, and electrophilicity leave room to develop new chemoselective cysteine modifiers with tunable properties. Therefore, expansion of several chemoselective modifiers with vast tunable properties is important.</p><p>Developing transition metal-based warheads for chemoselective modification is burgeoning. The electrophilicity of the metal center can be tuned by changing the ligand environment. This can lead to a chemoselective reagent with favorable reaction rate with amino acids. Transition metal complexes such as Au, Ru, Pd, Ni and Rh have been reported to modify amino acids via direct metalation or arylation.5–89–16Au complexes are known to modify cysteine significantly due to their inherent Lewis acidic character.15, 17 Au or Pd mediated cysteine arylation have been reported by Wong et. al.18, Spokoyny et. al.4,19Casini et. al.20–22 and Buchwald/Pentelute et al.23 However, Au complex reactivity can be tuned to modify other amino acids as well.24 In our previous work, we reported lysine modification in c-MYC protein using Au-based probes.24 Thus, tuning the ancillary ligand in the metal center can control the reactivity towards nucleophilic amino acids. Despite the progress made with transition metal frameworks that have been developed to target proteins, their utility in selective modification of proteins in their native sites remain a challenge.5 We hope to overcome this challenge by expanding the toolbox of metal-based reagents capable of covalently modifying endogenous protein. 4, 12, 23, 25–28</p><p>Inspired by the versatility of transition metal-based electrophilic handles, we report our efforts toward the synthesis of C^N type cyclometalated Au(III) complexes bearing different ancillary ligands for cysteine modification and evaluate the rate of reaction. Further, evidence for ligand associative mechanism toward cysteine arylation via X-ray crystallography is established. Insight into cysteine arylation by the cyclometalated framework using electrochemistry and mass spectrometry is outlined. Finally, we demonstrate the utility of our metal-mediated ligand affinity chemistry (MLAC) approach to selectively modify mutant protein (KRAS(G12C) using a ligand-directed gold reagent.</p><!><p>In an attempt to tune the cyclometalated scaffold for cysteine reactivity, we evaluated structure activity profile across cycloaurated compounds bearing different ancillary ligands. A small library of six different cyclometalated compounds were synthesized and their reactivity with cysteine was assayed by UV absorbance kinetics, mass spectrometry and electrochemistry. The investigations revealed a broad range of reactivities and provides a guiding framework to design cyclometalated gold(III) reagents for unique applications.</p><p>Applying organometallic reagents in biology is a fruitful arena for protein modification due to their inherent reactivity, unique geometry, and ligand tuning potential. This work takes advantage of the biocompatibility of gold-based compounds to investigate the kinetics of cysteine arylation and selective bioconjugation of proteins based on a proximity-guided approach using an affinity ligand. We proceeded by studying the effect of ancillary ligand modification on the cyclometalated Au(III)[C^N] framework, which have been shown to arylate the cysteine amino acid residue.18, 29 Following the synthesis of the cyclometalated dichlorido gold(III) [C^N] compound (1),10 we applied ligand substitution methods to synthesize different compounds bearing bromide (2), iodide (3), azido (4), thiocyanato (5), acetylacetonato (6) ligands (Fig. 1a). Single crystals of 2 and 4 were grown by vapor diffusion of diethyl ether into a dichloromethane solution of concentrated gold complexes and analyzed by X-ray diffraction. Complex 2 has Au – C bond distance of 2.02 Å between benzoyl pyridine and gold while complex 4 has Au – C bond distance of 2.10 Å, which is slightly greater than the Au – C bond length (2.04 Å) of [C^N]-cyclometalated dichlorido gold(III) (complex 1). In complex 2, the Au – Br (2.43 Å) bond trans to the N(aryl) in the (C^N) framework is shorter compared to the Au – Br (2.46 Å) bond trans to the C(aryl). A similar observation occurs for complex 4, as evidenced by the Au – N (2.04Å) trans to N in the C^N-framework and the Au –N (2.05 Å) trans to C. Both azido ligands in complex 4 has N=N (one N bonded to the Au) bond length of 1.23 Å and the other N=N bond length of 1.14 Å. Complexes 2 and 4 still maintain the square planer geometry around the Au(III) center. Both complexes maintain the same crystal packing as monoclinic (Fig. 2b and Table S1).</p><p>José Vicente et al. structurally characterized a gold(III) complex [Au(C6H4N=NPh-2)(acac-C)CI bearing 1,2-bis(4-methylphenyl)hydrazine as C^N framework. The acetylacetonato ligand was bonded via C to the Au central atom in a trans configuration to the N atom of the arylpyridine, in a manner consistent with what we propose for compound 6 in Fig. 1. The reported 1H-NMR spectrum chemical shift for C-bound acac to Au was -CH3CO, 2.38 ppm, S, 6H and –CH, 4.42 ppm, S, 1H, which was consistent with the 1H-NMR spectrum of complex 6 ( for -CH3CO 2.44 ppm, 6H and -CH 4.54 ppm, 1H, Fig. S8). This led to the conclusion that the acetylacetonato ligand was bonded to Au atom via C bond.30, 31 It is also plausible that in solution a compound with a rearranged O,O-chelate to the Au center can form.</p><p>The monomeric gold(III) complexes were fully characterized by NMR spectroscopy and purity established by elemental analysis (Figure S1 – S9).</p><!><p>To enhance our understanding of the rate of cysteine arylation using preformed Au(III) complexes, we conducted reactivity studies of compounds 1 – 6 with N-acetyl-L-cysteine. Given the established phenomenon by [C^N]-cyclometalated Au(III) to undergo fast nucleophile-assisted reductive elimination, the pseudo-first order reaction kinetics using UV-vis spectroscopy was employed to assess the effect on respective ancillary ligands on cysteine arylation. We calculated the kobs (s−1) (Fig. S24 and Table 1) for each compound in triplicate by monitoring the decay of the absorption band corresponding to the AuIII. We found that the calculated rate constant k (M−1s−1) for complexes 1 – 5 was in the range of 0.8 x 101 M−1s−1 - 11 x 101 M−1s−1 (Table 1). Specifically, Au(III) complexes bearing halide ancillary ligands (-Cl, -Br, -I) had rate constants of 11 X 101 M−1s−1, 6.8 X 101 M−1s−1, and 6.6 X 101 M−1s−1, respectively (Table 1). The impact of the degree of polarizability of the halogen ligands on reactivity with cysteine appeared to be similar when complexes with Cl, Br, or I ligands were considered. The compound with the thiocyanato ligands demonstrated a rate of 4.7 X 101 M−1s−1. We observed the least rate for complex 4 bearing the azido ligand. Strikingly, the reaction involving compound 6 with N- acetyl cysteine was ultra fast (< 1 min). Due to the rapid reactivity, we were not able to calculate the rate constant by UV –vis spectroscopy. While electronegativity of the ligands may play a role in Au(III) reactivity towards cysteine, the conjugated acetylacetonato ligand imparts increased polarizability coupled with an enhanced trans labile effect, which facilitates rapid substitution of the acac ligand toward reductive elimination. This small structure activity library set demonstrates the feasibility to control reaction kinetics of Au(III) complexes with cysteine residues for varied applications.</p><!><p>Experimental evidence for the mode of reactivity of cyclometalated Au(III) and cysteine is not well elucidated. Our previous work showed that modifications to the cyclometalated framework can affect nucleophile reactivity and rate of reductive elimination.32 Thus, we synthesized an aryl-pyridine Au(III) complex bearing acac and chlorido ligands, 7 (Fig. S10 –S11). The rationale was to slow the reductive elimination step in order to characterize the intermediate of Au(III) and cysteine reaction. Using crystallography, the structure of N-acetyl-L-cysteine ligated to cyclometalated Au(III), 8 was solved (Fig. 2, Table S1). Single crystals of 8 were grown by slow diffusion of diethyl ether into concentrated dichloromethane solution at room temperature. Aryl-pyridine of the Au(III) complex appeared to be planar without any major twists in the solid state. The crystal structure confirmed the significant trans effect for ligating N-acetyl-L-cysteine in the (C^N) framework while analyzing the bonding motif. The Au-S bond trans to nitrogen in the (C^N) framework is shorter than the Au-Cl bond trans to the carbon (Table 2). This leads to a slightly distorted square planar geometry around the Au(III) center. The bond angles around the Au(III) center was in the range of 81 – 96 degree, consistent with a square planar geometry (Table 2). The stereochemistry of the cysteine used was conserved in the ligated product as supported by the crystal structure (Fig. 2b).</p><!><p>The electrochemical behavior of the Au(III) warheads (1 – 6) were characterized by cyclic voltammetry in anhydrous DMSO with 0.1 M N(Bu)4PF6 as the supporting electrolyte. Representative voltammograms for Table 3 are shown in the supporting information (Fig S25 – S30). We found that the irreversible reduction events attributed to Au(III)/Au(I) reduction potentials were more negative than reported reduction potentials of Au(III)/Au(I), which are typically in the range of −0.95 V to −1.50 V.33 It is possible that the ligand – metal electron transfer contributes to the observed differences in Au(III)/Au(I) reduction potential. Specifically, the halide ( -Cl, -Br, -I) ligands bearing Au(III) warheads were shown to possess Au(III)/Au(I) reduction potentials at −1.92 V, −1.78 V and – 1.87 V respectively (Table 3.). This data is consistent with the kinetic rate constant of the Au(III) warheads 1, 2, and 3, which had shown similar reactivity towards N-acetyl-L-cysteine. Compound 4 had the most negative reduction potential at −2.08 V. It is conceivable that the less positive reduction potential of 4 makes it less likely to be reduced, leading to the relatively slow rate of reactivity. Additionally, quasi-reversible reduction/oxidation waves at −0.88 V are assigned to ligand– metal electron transfer based on the voltammograms obtained for the benzoyl pyridine ligand alone and the Au(III) complexes in our previous work.34 The oxidation event at ~ 0.68 V was observed in all Au(III) war heads may be related to ligand event. The cyclic voltammograms of 1, 2, and 3 exhibited similar redox wave features while 4, 5, and 6 exhibited slightly different patterns (Fig S25 –S30). We posit that reductive potentials can be predictive indicator for cysteine reactivity for the cyclometalated gold complexes used in this study.</p><!><p>Inspired by the differences in reactivity of 1 and 6 towards N-acetyl-L-cysteine, we sought to apply these warheads in covalent protein modification. We chose a clinically relevant native protein, KRAS mutant (G12C), which has a few cysteine residues for regioselective protein modification using our metal-mediated ligand affinity chemistry (MLAC) approach. We functionalized the Au(III) warheads with a KRAS(G12C) affinity ligand, KRASi to demonstrate the ligand-directed bioconjugation approach, herein, MLAC. KRASi was reported by Ostrem et. al.35 In their work, the KRASi was co crystallized with KRAS(G12C) mutant protein and revealed affinity for the switch – II pocket. This stimulated interest to use KRASi ligand for our MLAC strategy (Fig. 3a). Notably, inhibition of the KRAS(G12C) mutation using small molecules has become a very attractive therapeutic strategy in recent years.36–39 In five synthetic steps, we synthesized our MLAC reagents via standard coupling and substitution reactions (Scheme 1).18, 29, 35 The final compounds KRASi -Au-1, KRASi-Au-2 and KRASi-Au-3 (Fig. 3b) were fully characterized using 1H-NMR, 13C-NMR, high-resolution mass spectrometry (HRMS) and purity was established by HPLC (>97%) (Fig. S12 – S23). The compounds were stable in PBS at physiological conditions (Figure S31 –33).</p><p>We sought structural insights of our reagents, as in, the ability of our compounds to fit into the switch-II binding cleft of KRAS(G12C), via molecular modelling. The in silico study was done using a GDP bound KRAS(G12C) protein (PDB:4LYH).35 The MLAC reagent, KRASi -Au - 2 was used as a representative candidate (docking score = −8.34) and the Au(III)center appeared to have a preferred geometrical orientation toward the G12C position with a bond-length of 5.7 Å as shown in Fig. 3c. This illustrates proximity of the gold to the cysteine thiol for binding; and subsequent reductive elimination for the C–S arylated product.</p><p>Using a peptide sequence (GACGVGKIE) found within the KRAS mutant protein, we determined the reaction conditions of Au(III) towards cysteine arylation. We carried out the site-selective modification studies with the short model peptide under the following conditions: peptide (0.3 mM) in 20 mM HEPES buffer (pH =7.4) was incubated with 2.0 mM of the 1 and KRASi–Au–3 respectively for 2 h at 37 °C (peptide : Au(III)complex = 1 : 6.6). Tandem LC-MS/MS characterization as shown in Fig. 4 confirmed that the cysteine (12 C ) was selectively modified. Both 1 and KRASi–Au–3 modified the cysteine via Au(III)-mediated arylation respectively by m/z = 181 Da and m/z = 567 Da.</p><p>Further, we applied this labeling concept to the native protein. We carried out labeling studies with mutant cysteine containing KRAS(G12C) recombinant protein. We incubated KRASi -Au −3 (2.0 μM) with recombinant KRAS(G12C) (BPS bioscience, 0.4 μM), at 37 °C, in HEPES buffer, pH =7.4 (Protein : Au(III)complex = 1 : 5). The labelled protein was analyzed by using tandem LC-MS/MS. Tryptic digested peptides were further analyzed using MS/MS to reveal covalent modification of KRAS at 12C by KRASi–Au-3 in a manner consistent with arylation, as shown by the mass spectrometry result, m/z = 567.14 (Fig. 5 and S34–39). Further analysis of the tryptic digested peptides confirmed that specific arylation occurs only at 12C of the KRAS (G12C) protein. No modifications were found for the other cysteine residues that are located at the non-active sites of the protein. The calculated modification efficiency of the native protein was ~ 32 % (Table S2). To determine the efficacy of this labeling in the presence of thiol containing competitor, the same experimental condition was maintained, and glutathione (GSH) was added to the labeling reaction (Protein : GSH : Au(III) complex = 1: 1: 5). The labelled protein (in the presence of GSH) was analyzed by tandem LC-MS/MS following tryptic digestion (Fig. S40–S42). GSH did reduce KRASi–Au-3-induced protein modification efficiency by ~55% (Table S3). Of note, the native KRAS protein is in a buffer solution containing 0.25 mM of DTT to maintain protein stability. Importantly, KRASi-Au-3 tolerated DTT to label the protein of interest. Taken together, these results confirm the potential for Au(III) reagents for regioselective protein modification using a ligand-directed approach.</p><p>We used a comparative study to evaluate the selective antiproliferative effect of 1, KRASi-Au-1, KRASi-Au-2, or KRASi-Au-3 in a panel of cancer cell lines: MiaPaCa (pancreas, homozygous G12C), H358 (lung, heterozygous G12C), H460 (lung, Q61H mutation). The cells were treated with 0 – 100 μM of Au(III) warhead 1 or KRASi–Au-1, KRASi–Au–2, KRASi–Au-3 reagents, followed by MTT assay to determine the percentage cell viability (Fig. 6a and Table S4). KRASi-Au-1 showed considerable cytotoxicity in MiaPaCa and H358 cell lines (IC50 of 42.65±0.99 μM and 16.59±4.22 μM), harboring homozygous and heterozygous G12C mutations respectively. Of note, inhibitors of KRASG12C currently in the clinic perform poorly in heterozygous mutants, so the ability for KRASi-Au-1 to significantly inhibit H358 cells is promising. In the H460 cell line harboring a Q61H and not the G12C mutation, KRASi-Au-1 did not induce any cytotoxic effect, demonstrating exceptional selectivity for G12C mutants. KRASi-Au-3 was fairly cytotoxic to MiaPaCa cell lines (69.18±2.22 μM) and significantly inhibited H358 growth (Fig. 6a,b). KRASi-Au–2 did not demonstrate cytotoxic activity across all cell lines used potentially due to the relatively low cellular uptake (Fig. 6c). The Au(III) warhead, 1 did not show any selective cytotoxicity for KRAS(G12C) mutant cell lines. Overall, the ligand-directed Au(III) complexes show good selectivity for G12C mutant cancer cells.</p><p>Intracellular accumulation of Au(III) complexes in MiaPaCa cell lines were evaluated by cellular uptake studies. Cells were treated with 10 μM of complexes KRASi-Au-1, KRASi-Au-2, and KRASi-Au–3 and the Au content of the digested cell pellets were determined by GF-AAS (Graphite - Furnace Atomic Absorption Spectroscopy. In general, we found relatively low uptake profiles for the gold complexes, specifically, KRASi–Au-1 - 202 picomol/106cells; KRASi–Au-2 - 141 picomol/106cells; and KRASi–Au-3 - 189 picomol/106cells (Fig. 6c). We posit that the low gold accumulation in cells accounts for the modest cytotoxic activity in the MiaPaCa cells. Similarly, this reasoning could be extended to the poor cytotoxicity of KRASi–Au-2 in MiaPaCa cell lines, as it had the lowest cellular uptake.</p><!><p>Ten complexes have been synthesized and characterized to study cysteine arylation comprising, Au(III) warheads (seven compounds) and warheads conjugated to affinity ligand (three compounds). Rate constant calculation of the gold-mediated arylation reaction confirmed that the Au(III) reactivity towards cysteine can be kinetically controlled by the choice of ancillary ligands around the gold center. Using LC-MS/MS we confirmed that transfer arylation of affinity ligands can be achieved to covalently modify recombinant KRAS(G12C) protein. Cellular evaluation shows cytotoxic effects imposed by MLAC reagents in homozygous or heterozygous KRAS(G12C) with no effect on other mutants. This study can be used as a very good premise to control the kinetics of site-selective protein modification using tuned metal complexes. Further ligand tuning around the gold center may optimize kinetic parameters towards chemoselective Au(III) warheads beyond cysteine residues, which remains an active area of research in our laboratory.</p><!><p>All the reagents and solvents used in this work were bought from commercial vendors (Acros, Millipore-Sigma, TCI -America, AlFa Aesar, and Pharmco Aaper USA) and were used without additional purification unless explicitly stated otherwise. Anhydrous dimethylformamide (DMF) was purchased from Sigma – Aldrich sealed bottle containing 3Å molecular sieves and used as is. Human recombinant K-Ras(G12C) (N-terminal His - tag) protein was purchased from BPS Bioscience, USA. Model peptide (GACGVGKIE) was purchased from GeneScript, USA. Deuterated solvents for NMR were acquired from Cambridge Isotope Laboratories and used as is. Compounds were purified by silica gel (Silicycle, P/N: R10030B SiliaFlash ®F60, Size: 40-63 μm, Canada) for column chromatography. Analytical thin-layer chromatography (TLC) was performed by aluminum backed silica-gel plates (20 X 20 cm2, TLA-R10011B-323) from Silicycle. Reactions were conducted under standard atmospheric conditions with the exception of amide coupling reactions, which were performed under Schlenk conditions. Reactions were performed in Schlenk flasks equipped with Teflon-coated magnetic stir bars for stirring to maintain homogeneity in the reaction solutions. Progress of reactions was evaluated by NMR and TLC. Low-wavelength light (254 nm) or Iodine stain were used to visualize the TLC plates.</p><!><p>1H-NMR, and 13C{1H} NMR spectra were collected on a 500 MHz Bruker spectrometer, 400 MHz Varian spectrometer, or 500 MHz JEOL ECZr at the University of Kentucky (UK) NMR Center. Solvent signals were used as internal reference (1H NMR: DMSO at δ = 2.50 ppm and CD3OD at δ = 3.31, 13C NMR: DMSO at δ = 39.52 ppm and CD3OD at δ = 49.00 ppm) for chemical shifts in 1H NMR spectra collected. X- ray crystal structures were obtained from UK X-Ray crystallography center at Department of Chemistry. A Shimadzu UV – vis spectrometer (Shimadzu, model: UV1280) was used to collected and analyze the UV–vis spectra and kinetics data. Waters Synapt G2 HD mass spectrometer was used to obtain the High-resolution mass spectra (HRMS). Direct injection method was used to introduce samples to the instrument at 50 µL/min and ionized with ElectroSpray Ionization (ESI) module in the positive mode. Parameters used on the instrument were capillary = 2.8 kV, sampling cone = 40, extraction cone: 5.0, source temperature = 80 °C, desolvation temperature = 150 °C, and desolvation gas flow = 500L/h. Mass spectrometry studies and analysis were performed at the Central Analytical Laboratory at the University of Colorado, Boulder. Low resolution LC/MS were obtained using Agilent 1200 LCMS at College of Pharmacy, University of Kentucky. Synergy H1 hybrid plate reader at College of Pharmacy (UK) was used for cell viability experiment. Flash silica-gel chromatography was carried out using Combi-Flash from Teledyne ISCO. An Agilent 1100 series HPLC instrument equipped with a reverse-phase (C18) column was used to obtain the trace HPLC data.</p><!><p>Synthesis of 1: [C^N]Au(III)-(pcp) Cl2] [pcp=2-(2- pyridyl carbonyl)phenyl] was synthesized according to reported literature procedure. 10</p><p>Synthesis of 2 : To a 25 mL round bottom flask, [C^N]Au(III)-(pcp) Cl2] (45.5 mg, 0.1 mmol), AgSbF6 (69.4 mg, 0.2 mmol, 2 equiv.) and 10.0 mL of MeOH/DCM (v/v = 1:1) was added and stirred for 2 h at room temperature. A white precipitate of AgCl formed and removed by centrifuging the reaction solution. The decant was transferred to another 25 mL round bottom flask and NaI (30.0 mg, 0.2 mmol, 2 equiv.) was added and stirred for 8h at room temperature. The progress of the reaction was monitored by using TLC. The solvent was evaporated in vacuo and washed several times with DCM to obtain the desired product (36.5 mg, Yield = 57%). 1H NMR (400 MHz, DMSO-d6) δ 8.65 (s, 1H), 8.17 – 8.08 (m, 2H), 7.92 (d, J = 7.9 Hz, 1H), 7.73 – 7.67 (m, 1H), 7.51 (dd, J = 7.5, 1.0 Hz, 1H), 7.40 (dd, J = 7.6, 1.6 Hz, 1H), 7.27 (td, J = 7.7, 1.7 Hz, 1H).</p><p>Synthesis of 3 : To a 25 mL round bottom flask, [C^N]Au(III)-(pcp) Cl2] (42.5 mg, 0.09 mmol), AgSbF6 (58.0 mg, 0.18 mmol,2 equiv.) and 10.0 mL of MeOH/DCM (v/v = 1:1) was added and stirred for 2 h at room temperature. White precipitate of AgCl was formed and removed by centrifuging the reaction mixture. The decant was transferred to another 25 mL round bottom flask and NaBr (19.4 mg, 0.18 mmol, 2 equiv.) was added and stirred for 8h at room temperature. The progress of the reaction was monitored by TLC. The solvent was evaporated in vacuo and the residue was dissolved in minimal amount of DCM and precipitated with diethyl ether. The desired pale-yellow precipitate was filtered under vacuum (37.3 mg, Yield = 73%) 1H NMR (400 MHz, DMSO-d6) δ 9.29 (s, 1H), 8.38 (t, J = 7.5 Hz, 1H), 8.20 (d, J = 7.6 Hz, 1H), 7.91 (t, J = 6.2 Hz, 1H), 7.72 (d, J = 7.8 Hz, 1H), 7.61 (d, J = 7.3 Hz, 1H), 7.35 (dt, J = 24.1, 7.2 Hz, 2H). 13C NMR (101 MHz, DMSO) δ 191.62, 151.54, 141.31, 138.44, 135.48, 132.73, 131.47, 128.15, 126.74, 125.89, 39.73, 39.52, 39.31, 39.20, 39.10.</p><p>Synthesis of 4 : To a 100.0 mL round bottom flask [Au(C^N)(pcp)-Cl2] [pcp=2-(2-pyridyl carbonyl)phenyl] (103.0 mg, 0.23 mmol) and 40.0 mL of DCM was added. NaSCN (42.6 mg, 0.52 mmol, 2.3 equiv.) was dissolved in 10.0 mL water and added to the round bottom flask. The reaction mixture was stirred for 4-5 h at rt in the dark. The progress of the reaction was monitored by TLC. Upon completion of the reaction, pale yellow organic layer was separated and dried over anhydrous MgSO4, and solvent was reduced to small volume in vacuo. Then diethyl ether was added to precipitate the product and it was cooled in the freezer. The resulted pale orange solid was filtered and washed with diethyl ether (50 ml x 3).40 (54.0 mg, Yield = 48%) 1H NMR (400 MHz, DMSO-d6) δ 9.41 (d, J = 5.5 Hz, 1H), 8.56 (d, J = 7.4 Hz, 1H), 8.38 (d, J = 7.9 Hz, 1H), 8.10 (t, J = 6.7 Hz, 1H), 7.72 (t, J = 7.1 Hz, 2H), 7.62 – 7.48 (m, 3H).</p><p>Synthesis of 5 : To a 25 mL round bottom flask, [C^N]Au(III)-(pcp) Cl2] (43.5 mg, 0.09 mmol), AgSbF6 (66.4 mg, 1.9 mmol, 2 equiv.) and 10.0 mL of MeOH/DCM (v/v = 1:1) was added and stirred for 2 h at rt. White precipitate of AgCl was formed and removed by centrifuging the reaction mixture. The decant was transferred to another 25 mL round bottom flask and NaN3 (12.5 mg, 1.9 mmol, 2 equiv.) was added and stirred for 8h at rt. The progress of the reaction was monitored by using TLC. The solvent was evaporated in vacuo. Pure compound was obtained by doing precipitation with DCM and diethyl ether. The precipitate was filtered under vacuum. (38.5 mg, Yield = 86%) 1H NMR (400 MHz, DMSO-d6) δ 9.07 (d, J = 5.7 Hz, 1H), 8.53 (t, J = 8.3 Hz, 1H), 8.37 (d, J = 7.8 Hz, 1H), 8.08 (t, J = 6.7 Hz, 1H), 7.75 (dd, J = 13.8, 6.7 Hz, 2H), 7.58 (t, J = 8.4 Hz, 1H), 7.48 (t, J = 7.3 Hz, 1H). 13C NMR (101 MHz, DMSO) δ 189.28, 151.00, 146.12, 144.26, 137.51, 134.32, 132.13, 130.15, 129.80, 129.68, 128.67, 127.25, 40.15, 39.94, 39.73, 39.52, 39.31.</p><p>Safety note: NaN3 is dangerous, acute toxic hazardous chemical considered by OSHA hazard communication standard. We followed all the safety measures while handling and disposal.</p><p>Synthesis of 6 : To a 25 mL round bottom flask, [C^N]Au(III)-(pcp) Cl2] (100.0 mg, 0.22 mmol), N(Bu)4acac ( 75.9 mg, 0.22 mol) and 10.0 mL of CHCl3 was added and stirred for 1 h at rt. The progress of the reaction was monitored by using TLC. The desired product was purified by using Silica Flash column chromatography31 ( Eluent - 0 – 5 % MeOH in DCM)( 56.4 mg, Yield = 49%) 1H NMR (400 MHz, Chloroform-d) δ 9.39 (ddd, J = 5.7, 1.6, 0.7 Hz, 1H), 8.32 (ddd, J = 7.8, 1.6, 0.7 Hz, 1H), 8.24 (td, J = 7.7, 1.5 Hz, 1H), 7.93 – 7.88 (m, 1H), 7.82 (ddd, J = 7.5, 5.6, 1.6 Hz, 1H), 7.76 (dd, J = 7.5, 1.8 Hz, 1H), 7.49 – 7.37 (m, 2H), 4.86 (s, 1H), 2.24 (s, 6H). 13C NMR (101 MHz, CDCl3) δ 205.03, 190.23, 150.92, 149.04, 141.90, 139.66, 134.45, 133.94, 132.49, 130.16, 128.48, 128.19, 126.05, 77.16, 76.84, 76.52, 58.56, 31.30. Anal. % Calcd for C17H15AuClNO3: C, 39.75%; H, 2.94%; N, 2.73%. Found: C, 40.04%; H, 3.09%; N, 2.73%.</p><p>Synthesis of 7 : To a 25 mL round bottom flask, cyclometalated Au(C^N) -(pp) Cl2] [pp=2-phenyl pyridine] (99.0 mg, 0.23 mmol), N(Bu)4acac ( 80.1 mg, 0.23 mol) and 10.0 mL of CHCl3 was added and stirred for 1 h at rt. The progress of the reaction was monitored by using TLC. The desired product was purified by using Silica Flash column chromatography ( Eluent - 0 – 5 % MeOH in DCM)( 54.4 mg, Yield = 48%) 1H NMR (400 MHz, Chloroform-d) δ 9.58 (ddd, J = 5.8, 1.7, 0.7 Hz, 1H), 8.09 (ddd, J = 8.1, 7.5, 1.6 Hz, 1H), 7.96 (dt, J = 8.3, 0.9 Hz, 1H), 7.85 – 7.81 (m, 1H), 7.73 – 7.69 (m, 1H), 7.53 (ddd, J = 7.3, 5.8, 1.4 Hz, 1H), 7.44 – 7.36 (m, 2H), 4.54 (s, 1H), 2.44 (s, 6H). 13C NMR (101 MHz, CDCl3) δ 205.41, 163.25, 148.61, 148.46, 143.82, 142.34, 133.18, 131.59, 129.01, 125.93, 124.56, 120.59, 77.80, 77.48, 77.16, 60.75, 32.21. Anal. % Calcd for C16H15AuClNO2: C, 39.57%; H, 3.11%; N, 2.88%. Found: C, 39.7%; H, 3.08%; N, 2.82%.</p><p>Synthesis of (A): To a well purged Schlenk flask was placed dichlorophenoxy acetic acid (1.0 g, 4.5 mmol), N-Boc piperazine (0.85 g, 4.5 mmol), DIPEA (1.58 mL, 9.0 mmol) and HBTU (2.42 g, 6.3 mmol) and 4.0 mL anhydrous DMF was added. The reaction mixture was stirred for 20 h, at room temperature. Upon completion of the reaction, the solution was diluted into 75 mL EtOAc, washed with 3x30 mL H2O, 30 mL saturated NaHCO3, then 30 mL brine. The organic layer was dried over anhydrous MgSO4, filtered, and solvent was evaporated to dryness, leaving an off- white solid, which was used for the next step without further purification. 1H NMR (400 MHz, Chloroform-d) δ 7.37 (d, J = 2.6 Hz, 1H), 7.17 (dd, J = 8.8, 2.5 Hz, 1H), 6.94 (d, J = 8.8 Hz, 1H), 4.76 (s, 2H), 3.62 – 3.56 (m, 4H), 3.42 (ddd, J = 22.0, 6.7, 3.8 Hz, 4H), 1.46 (s, 9H). (Yield = Quantitative)35</p><p>Synthesis of KRASi : To a solution of (A) (~1.0 g) in DCE (10.0 mL), was added TFA (10.0 mL) and the reaction solution was stirred at room temperature for 6 h. Excess TFA was removed in vacuo and the product was washed several times with DCM. The solvent was removed in vacuo. Flash column (DCM: MeOH: 1-5 %) was used to purify the desired product. 1H NMR (400 MHz, Chloroform-d) δ 9.88 (s, 1H), 7.37 (d, J = 2.5 Hz, 1H), 7.19 (dd, J = 8.8, 2.5 Hz, 1H), 6.89 (d, J = 8.9 Hz, 1H), 4.75 (s, 2H), 3.90 (d, J = 13.6 Hz, 4H), 3.19 (d, J = 24.7 Hz, 4H). (51.99 mg, Yield = 70 %)35</p><p>Synthesis of (B): To a 25 mL round-bottom flask was added [C^N]Au(III)-(pcp) Cl2] (1) (200.0 mg, 0.4 mmol), O-(Carboxymethyl)hydroxylamine hemihydrochloride ( 43.75 mg, 0.4 mmol) and 6.0 mL of DCM/MeOH (v/v = 1:1). The solution was stirred for 48 h at room temperature. Upon completion of the reaction, the white precipitate was washed several times with MeOH, and the solvent was evaporated to obtain the desired product. 1H NMR (500 MHz, DMSO-d6) δ 9.34 (d, J = 5.6 Hz, 1H), 8.49 (d, J = 3.6 Hz, 2H), 7.96 (d, J = 4.6 Hz, 1H), 7.55 (d, J = 7.8 Hz, 1H), 7.34 (dd, J = 17.4, 4.1 Hz, 3H), 4.93 (s, 2H). (178. 6 mg, Yield = 77 %)35</p><p>Synthesis of (C): To a 100 mL round – bottom flask 1-(4-aminopieridin-1-yl)2-(2,4-dichlorophenoxy) ethenone KRASi (1.0 g, 3.45 mmol), 2-bromoethylamine (2.12 g, 10 mmol), Na2CO3 (1.1 g, 10 mmol) and ACN (15 mL) was added. The reaction mixture was refluxed at 80 °C for 36 h. Upon completion of the reaction as monitored by TLC, the solvent was evaporated in vacuo and diluted with 1 M NaOH (5.0 mL) and extracted with ethyl acetate (20.0 mL X 3). The combined organic layers were dried over anhydrous MgSO4, and the solvent was evaporated in vacuo. The reaction was purified via flash column silica chromatography with a gradient (DCM: MeOH = 0 – 7 %). 1H NMR (400 MHz, Chloroform-d) δ 7.29 (d, J = 2.5 Hz, 1H), 7.12 – 7.08 (m, 1H), 6.89 – 6.86 (m, 1H), 4.69 (s, 2H), 3.52 (dd, J = 13.7, 4.9 Hz, 4H), 2.76 (dd, J = 17.4, 5.1 Hz, 4H), 2.36 (dt, J = 16.1, 5.4 Hz, 4H). (50.55 mg, Yield = 44 %)</p><p>Synthesis of Complex KRASi–Au–1 : To a well purged Schlenk flask, KRASi (21.44 mg, 0.27 mmol), B (50.0 mg, 0.09 mmol), HOBt (36.64 mg, 0.27 mmol), EDC (51.76 mg, 0.27 mmol) and 1.0 mL of anhydrous DMF was added and stirred under nitrogen for 20 h at room temperature. The reaction mixture was diluted with 50.0 mL of water and extracted with 3 X 20 mL of DCM. Organic layers were washed with 30.0 mL of brine and dried over anhydrous MgSO4. The solvent was evaporated in vacuo and flash column (Eluent - DCM: MeOH = 0 – 5 %) was used to purify the desired product. ( 9.8 mg, Yield = 13 %) 1H NMR (400 MHz, Chloroform-d) δ 9.42 (d, J = 6.0 Hz, 1H), 8.88 – 8.78 (m, 1H), 8.14 (t, J = 7.9 Hz, 1H), 7.71 – 7.60 (m, 2H), 7.44 – 7.37 (m, 2H), 7.24 – 7.17 (m, 2H), 6.97 (d, J = 8.8 Hz, 1H), 5.17 (d, J = 15.2 Hz, 1H), 4.92 (d, J = 15.1 Hz, 1H), 4.81 (s, 2H), 3.70 (d, J = 24.8 Hz, 6H), 3.45 (d, J = 28.1 Hz, 2H). 13C{1H} NMR (101 MHz, CDCl3) δ 166.32, 152.26, 152.09, 146.03, 142.21, 138.94, 133.89, 130.81, 130.70, 130.23, 129.00, 128.42, 127.83, 127.31, 114.73, 77.48, 73.15, 69.21, 45.75, 45.19, 44.53, 42.44. TOF-MS-ES+: m/z (%) 801.0307 (100) [M + Li]+, calculated m/z for [M + Li]+ 801.0273. Using RP-HPLC, purity was established to be >97%. Rf = 7.69 minutes by the following HPLC method: Flow rate: 1 mL/min; λ = 280 nm; Eluent A = H2O with 0.1% TFA; Eluent B = MeOH with 0.1% TFA; Solvent Gradient: 0 – 3 min (50:50 H2O:MeOH), 5 min (40:60 H2O:MeOH), 7 min (30:70 H2O:MeOH), 9 min (0:100 H2O:MeOH), 10 min (20:80 H2O:MeOH) 12 min until end of run (100:0 H2O:MeOH).</p><p>Synthesis of Complex KRASi–Au–2 : To a 25.0 mL round – bottom flask, Complex KRASi–Au–1 (20.0 mg, 0.025mmol), (Bu)4N(acac) (8.6 mg, 0.025 mmol) and CHCl3 (5.0 mL) was added. The reaction mixture was stirred for 10.0 min at room temperature. The desired product was purified by using flash column chromatography (Eluent - DCM:MeOH = 0 – 5 %). ( 11.23 mg, Yield = 52%) 1H NMR (500 MHz, Chloroform-d) δ 9.19 (s, 1H), 8.70 – 8.64 (m, 1H), 8.06 (d, J = 7.4 Hz, 1H), 7.73 (s, 1H), 7.59 (s, 1H), 7.47 – 7.36 (m, 3H), 7.21 (d, J = 7.8 Hz, 1H), 6.97 (s, 1H), 5.12 (d, J = 15.3 Hz, 1H), 4.90 (d, J = 7.4 Hz, 1H), 4.80 (s, 2H), 3.70 (d, J = 28.9 Hz, 6H), 3.43 (d, J = 25.8 Hz, 2H), 2.25 (s, 6H). 13C{1H}c NMR (101 MHz, CDCl3) δ 205.90, 204.84, 167.12, 166.02, 155.20, 151.94, 150.78, 146.71, 140.71, 140.11, 133.43, 131.17, 130.51, 128.91, 128.14, 127.33, 126.40, 123.72, 114.40, 77.48, 72.67, 68.86, 58.03, 53.57, 44.91, 42.13, 32.07, 31.26, 29.83. TOF-MS-ES+: m/z (%) 865.1035 (100) [M + Li] +, calculated m/z for [M + Li] + 865.1034. Using RP-HPLC, purity was established to be >97%. Rf = 8.56 minutes by the following HPLC method. Flow rate: 1 mL/min; λ = 260 nm; Eluent A = H2O with 0.1% TFA; Eluent B = MeOH with 0.1% TFA; Solvent Gradient: 0 – 3 min (50:50 H2O:MeOH), 5 min (40:60 H2O:MeOH), 7 min (30:70 H2O:MeOH), 9 min (0:100 H2O:MeOH), 10 min (20:80 H2O:MeOH) 12 min until end of run (100:0 H2O:MeOH).</p><p>Synthesis of Complex KRASi–Au–3 : To a well purged Schlenk flask, C (95.46 mg, 0.27 mmol), B (50.0 mg, 0.09 mmol), HOBt (36.64 mg, 0.27 mmol), EDC (51.76 mg, 0.27 mmol) and 1.0 mL of anhydrous DMF was added and stirred under nitrogen for 20 h at room temperature. The reaction mixture was diluted with 50.0 mL of water and extracted with 3 X 20 mL of DCM. Organic layers were washed with 30.0 mL of brine and dried over anhydrous MgSO4. The solvent was evaporated in vacuo and flash column (Eluent - DCM:MeOH = 0 – 5 %) was used to purify the desired product. ( 8.0 mg, Yield = 10 %) 1H NMR (500 MHz, Chloroform-d) δ 9.50 (d, J = 5.8 Hz, 1H), 8.43 (d, J = 7.8 Hz, 1H), 8.20 (t, J = 7.8 Hz, 1H), 7.69 (q, J = 6.8, 6.1 Hz, 2H), 7.42 (d, J = 5.5 Hz, 1H), 7.36 (s, 1H), 7.25 – 7.22 (m, 1H), 7.17 (d, J = 6.3 Hz, 1H), 6.93 (d, J = 8.8 Hz, 1H), 4.85 (d, J = 7.9 Hz, 2H), 4.73 (s, 2H), 3.46 (d, J = 25.9 Hz, 8H), 2.51 (d, J = 59.2 Hz, 7H). 13C{1H} (101 MHz, CDCl3) δ 167.79, 165.27, 153.82, 152.05, 144.88, 141.63, 138.16, 133.45, 130.39, 129.95, 128.60, 127.61, 126.97, 126.53, 123.38, 114.40, 76.84, 74.22, 68.28, 56.17, 52.85, 52.21, 45.06, 41.62, 35.56, 29.53. TOF-MS-ES+: m/z (%) 799.0300 (100) [M - Cl] +, calculated m/z for [M - Cl] + 799.0800. Using RP-HPLC, purity was established to be >97%. Rf = 6.34 minutes by the following HPLC method. Flow rate: 1 mL/min; λ = 260 nm; Eluent A = H2O with 0.1% TFA; Eluent B = ACN with 0.05% HCOOH; Solvent Gradient: 0 – 11 min (0:100 H2O: ACN). 11 min until end of run (0:100 H2O: ACN).</p><!><p>Reactions were performed in 2.25 % ACN in pure water at pH = 7.2. measurements were performed using Shimadzu, model: UV1280 spectrophotometer. Au(III) war heads 1,2,3,4 and 5 were dissolved in ACN separately to make the stock solution (It will give the 2.25 % ACN in the final reaction mixture). Each complex stock solution was diluted using water to 5 different concentrations (100 μM, 80 μM, 60 μM,40 μM and 20 μM). 5.0 mM of N- acetyl cysteine stock solution was prepared in water. The complex solution (1.5 mL) was quickly mixed with N-acetyl cysteine (1.5 mL) in the cuvette and absorbance at particular wavelength (Au(III) warhead 1 = 282 nm, 2 = 275 nm, 3 = 280 nm, 4 = 271 nm and 5 = 279 nm) was recorded every 4 sec until the absorbance was constant. The same procedure was followed for all the concentration as triplicates. To maintain the pseudo first order reaction condition, the concentration of the N- acetyl cysteine was kept as 100 times more than the complex. Rate constant was calculated by using the following equations (Table 1 and Fig. S24).2, 41 A+B→Products R=k[A]n[B]m [B]∘>>>[A]∘ So,[B]=[B]∘ R=k[A]n[B]∘m R=kobs[A]n,kobs=k[B]∘m</p><p>Since it is a pseudo first order condition, n = 1 So,In[A]=−kobst+ln[A]o k=kobs/[A]</p><p>kobs can be obtained from experiment.</p><!><p>A Bruker D8 Venture dual source diffractometer was used to collect the Low temperature (90K) X-ray diffraction data of 2, 4, and 8, and the summary of the crystallographic information is reported in Table 2 and S1. Complexes 2,4 and 8 were dissolved in minimal amount of DMF or DCM and diethyl ether was used to grow the crystals at room temperature by vapor diffusion method. All crystals were mounted using polyisobutene oil on the tip of a fine glass fiber, which was fastened in a copper mounting pin with an electrical solder. Mounted crystals were placed directly into the cold gas stream of a liquid-nitrogen based cryostat.42, 43 APEX3 package was used to for Lorentz polarization effects such as raw data integration, scaling, merging and correction. SHELXT and SHELXL44 were used respectively to determine the space group and refinement. Refinement of non -hydrogen atoms were performed using anisotropic displacement parameters. Hydrogen atoms were placed at calculated positions and refined using a riding model with their isotropic displacement parameters (Uiso) set to either 1.2Uiso or 1.5Uiso of the atom to which they were attached. Graphical program SHELXTL-XP was used to draw the ellipsoid plots.45 All the structures were checked for missed symmetry, twinning, and overall quality with PLATON,46 an R-tensor47 before submitting to Cambridge Structural Database (CSD). Finally CheckCIF was used for validation of the structures.46</p><!><p>Complexes 1, 2, 3, 4, 5 and 6 were dissolved in DMSO to prepare 5.0 mM of solutions. As an electrolyte, 0.1 M N(Bu)4PF6 was added to each sample. CH instruments 650E potentiostat, containing custom 3 electrodes cell comprised of a 3 mm diameter glassy-carbon working electrode, platinum-wire counter electrode, and Ag/AgCl wire reference electrode was used to perform the electrochemical measurements at room temperature. During our electrochemical studies ferrocene was used as an internal reference. 100 mV/s was used as a scan rate in all the measurements. And the measurements were referenced to Ag/AgCl based on the position of Fc/Fc+ couple. The results are reported in the main manuscript and Supporting Information (Table 3. and Fig. S25 – S30)</p><!><p>To determine the site selective modification in cysteine, 100 μL of short model peptide GACGVGKIE ( 0.3 mM) was separately incubated with 2.0 mM of complexes 1 and KRASi–Au-3 in 20 mM HEPES, 100 mM NaCl, pH= 7.4 for 2 h at 37 °C (Peptide : Au(III) complex = 1: 6.6) Both complexes were dissolved in minimal amount of ACN and diluted using buffer to bring the 2.0 mM concentration. The labeled peptide was analyzed by ESI – LC- MS/MS mass spectrometry (Fig. 4.).</p><p>To determine the covalent modification of Au(III) MLAC reagents, 0.4 μM (10 μg/mL) His tagged K-Ras(G12C) (BPS Bioscience) protein was incubated with 2. 0 μM of complex KRASi–Au-3 in 20 mM HEPES, 100 mM NaCl, pH= 7.4 for 2 h at 37 °C (Au(III) complexes : Protein = 5:1). The complex KRASi–Au-3 was dissolved in minimal amount of DMSO and diluted to 2.0 μM using buffer. The labeled protein was analyzed by ESI- LC-MS/MS mass spectrometry following tryptic digestion. Tryptic digested peptide (LVVVGACGVGK) containing G12C was analyzed manually to confirm the covalent modification on 12C. The other tryptic digested peptides containing non-active site cysteines as well as other peptides were analyzed to confirm the modification only at the 12C position. (Fig. 5 and S34–S39). The labeling efficiency of KRASi–Au-3 was calculated using area of the peak (Table S2).</p><p>To determine the covalent modification of Au(III) MLAC reagents in the presence of glutathione (GSH), 0.4 μM (10 μg/mL) His tagged K-Ras(G12C) (BPS Bioscience) protein, and 0.4 μM of GSH were incubated with 2. 0 μM of complex KRASi–Au-3 in 20 mM HEPES, 100 mM NaCl, pH= 7.4 for 2 h at 37 °C (Au(III) complexes : Protein : GSH = 5:1:1). The complex KRASi–Au-3 was dissolved in minimal amount of DMSO and diluted to 2.0 μM using buffer. The labeled protein was analyzed by ESI- LC-MS/MS mass spectrometry following tryptic digestion. Tryptic digested peptide (LVVVGACGVGK) containing G12C was analyzed manually to confirm the covalent modification on 12C. Labeling efficiency of was calculated KRASi–Au-3 in the presence of GSH (Fig. S40–S42 and Table S3).</p><!><p>In Silico studies were performed with the MOE 2019.02 program (Chemical Computing Group Inc., Canada). The starting coordinates of the X-ray crystal structure of K-Ras (G12C) with PDB code 4LYH were obtained from the RCSB Protein Data Bank. The Protein (4LYH) structure was prepared by protonation and assignment of partial charges via QuickPrep application in MOE. Standard energy minimization was then prepared. The structure of the ligand (KRASi–Au-2) was drawn in the ChemDraw Prime 16.0.1.4 software (PerkinElmer, Waltham, MA), and then copied as SMILES into MOE via the builder tool. Standard preparation and energy minimization procedures were also performed on KRASi–Au-2 via the QuickPrep application in MOE. The active site of the 4LYH protein was selected and defined as docking site. Molecular docking was performed in two stages. Firstly, induced-fit docking experiments were done using the dummy atoms that defined the binding pocket of interest. Alpha PMI method was used to align the various conformations of the ligand at the active site of interest, then scored by using the London dG method. A second stage refinement-and-rescoring were required to generate the best poses of the complex KRASi–Au-2 at the active site; this was accomplished by using the AffinitydG method.</p><!><p>All UV- vis spectra were collected on a Shimadzu UV-1280 model equipment. Phosphate buffered saline (PBS) was used as received from Corning® (without calcium or magnesium). DMEM was purchased from Corning© and used as is. Each medium was warmed to 37 °C prior to dilution of the complexes. Freshly prepared complexes were used to prepare 1 mM stock in DMSO/media (% v/v=20/80) solution. Stock solutions were diluted to 50 µM with the respective biological medium. The amount of DMSO in each sample was 0.5%.Therefore the instrument was blanked with either a 0.5% DMSO in PBS/DMEM solution prior to each scan. The solutions were kept in an incubator at a controlled temperature of 37 °C in between each run. Spectrum were recorded at each time periods of t = 0 h (after preparation of sample), 1 h, 3 h, 6 h, 9 h, 12 h, and 24 h. , The instrument was blanked with corresponding 0.5% DMSO in PBS/DMEM solution prior to each run. 600 nm to 200 nm of absorbance profile was recorded for each sample.</p><!><p>MiaPaCa, H358, and H460 cell lines were purchased from ATCC. H460 and H358 cells were cultured in RPMI 1640 medium with 10% fetal bovine serum (FBS), 1mM sodium pyruvate, 2 mM L-glutamine, 100 units/mL penicillin, and 100 μg/mL streptomycin (Invitrogen). MiaPaCa cells were cultured in DMEM with 10% FBS, 2 mM L-glutamine, 100 units/mL penicillin, and 100 μg/mL streptomycin. All cells were maintained at 37 °C in a humid atmosphere containing 5% CO2.</p><!><p>The three human cancer cell lines were seeded in a 96-well plate (3x103 cells/well) in normal growth conditions for 24 h at 37 °C. Media was removed, and cells were then treated with 0 −100 μM of compounds 1, KRASi–Au–1, KRASi–Au–2, and KRASi–Au-3 in media for 72 h. The various stock solutions were prepared in DMSO (final DMSO concentration was 0.1%). Cells in each well were treated with 100 µL of resazurin solution (12.5g/50ml phosphate buffered saline) and incubated for 3 h. Measurements of absorbance were subsequently performed using Synergy H1 hybrid plate reader at 560/590nm (excitation/emission). Data were collected using BioTek Gen5 software. All experiments were conducted in triplicates.</p><!><p>MiaPaCa cells (5 × 105) were plated in each 6-well plate. After the cells were incubated overnight,10 μM of complexes KRASi–Au–1, KRASi–Au–2, and KRASi–Au-3 were added and incubated at 37 °C for15 h. The cells were collected by trypsinization and centrifugation in a 15mL centrifuge tube for 5 min. The cells pellets were transferred to 1.5 mL eppendorf tubes. Pellets were washed twice with PBS and then resuspended in 0.5 mL con. HNO3 (70 %) and agitated for 30 s. Solutions were transferred to 15 mL tubes containing 4.5 mL DI water to make the HNO3 concentration 7%. The samples were subjected to Graphite Furnace – Atomic Absorbance Spectroscopy to assess the concentration of gold in each sample. All studies were performed in triplicates.</p>

PubMed Author Manuscript

Development of Organometallic S6K1 Inhibitors

Aberrant activation of S6 kinase 1 (S6K1) is found in many diseases, including diabetes, aging, and cancer. We developed ATP competitive organometallic kinase inhibitors, EM5 and FL772, which are inspired by the structure of the pan-kinase inhibitor staurosporine, to specifically inhibit S6K1 using a strategy previously used to target other kinases. Biochemical data demonstrate that EM5 and FL772 inhibit the kinase with IC50 value in the low nanomolar range at 100 \xce\xbcM ATP and that the more potent FL772 compound has a greater than 100-fold specificity over S6K2. The crystal structures of S6K1 bound to staurosporine, EM5, and FL772 reveal that the EM5 and FL772 inhibitors bind in the ATP binding pocket and make S6K1-specific contacts, resulting in changes to the p-loop, \xce\xb1C helix, and \xce\xb1D helix when compared to the staurosporine-bound structure. Cellular data reveal that FL772 is able to inhibit S6K phosphorylation in yeast cells. Together, these studies demonstrate that potent, selective, and cell permeable S6K1 inhibitors can be prepared and provide a scaffold for future development of S6K inhibitors with possible therapeutic applications.

development_of_organometallic_s6k1_inhibitors

INTRODUCTION<!>Identification of First Generation Organometallic Ruthenium Inhibitors with Potency and Specificity for S6K1<!>EM5 Binds to S6K1 in the ATP Binding Pocket in an Unusual Conformation<!>Development of Second-Generation Organometallic Inhibitors<!>Characterization of the FL772 S6K1 Inhibitor<!>Cellular Properties of FL772<!>DISCUSSION AND CONCLUSIONS<!>Protein Kinase Profiling<!>Synthesis of EM5, EM6, and FL772<!>Cloning, Expression, and Purification of S6K1 Constructs<!>Cloning, Expression, and Purification of S6K2 Construct<!>Radioactive Kinase Assay<!>Crystallization and Structure Determination<!>Cell Culture and Western Blotting<!>Yeast Cell Culture and Lysis

<p>S6 kinases are members of the AGC serine/threonine kinases of the RSK family, exhibit high homology within their catalytic domain, and are activated by the phosphorylation of a critical residue within the activation loop by phosphoinositide dependent kinase 1 (PDK1). Yeast contains one S6 kinase called Sch9, and humans contain two isoforms called S6K1 and S6K2. S6 kinases act downstream of phosphatidylinositol (3,4,5)-triphosphate (PIP3) in the phosphatidylinositide 3-kinase (PI3K) pathway. Phosphorylation of serine and threonine residues in the C-terminal regulatory domain leads to the phosphorylation of a S6K activation loop residue by PDK1 (residue 252 on the longer splice variant of S6K1).1 In addition to PDK1, mTOR is also involved in the activation of S6K12 and phosphorylates S6K1 at residue T412.</p><p>S6 kinases are associated with many cellular processes, including protein synthesis, mRNA processing, cell growth, and cell survival. S6K1 and S6K2 phosphorylate and activate the 40S ribosomal protein S6, which promotes protein synthesis through an increased rate of mRNA transcription.3 S6K1 also regulates cell size and progression through the cell cycle,4–6 in addition to promoting cell survival by inactivating the proapoptotic protein BAD.7</p><p>The aberrant activation of S6 kinases has been shown to play a role in many disease conditions, including diabetes, obesity, aging, and cancer.8–10 Many melanoma cells harbor constitutive activation of the PI3K-AKT pathway, which results in AKT phosphorylation and leads to activation of the downstream targets mTOR and S6K1.11 This increase in phosphorylation by S6K1 mediates increased protein translation and cell growth. Treatment with rapamycin, an allosteric mTOR inhibitor, leads to significant dephosphorylation of S6K1 and decreased cell growth.12 However, treatment with mTOR inhibitors abrogates feedback inhibition of other pathways,13 which in part leads to side effects such as hyperglycemia, hypercholesterolemia, and hyperlipidemia.14 Because of this, inhibition of S6K1 represents an alternative therapeutic strategy that may bypass the limitations of mTOR inhibition.</p><p>We have previously reported on the development of ATP competitive organometallic kinase inhibitors with high potency and specificity. These inhibitors are structurally inspired by the class of indolocarbazole alkaloids, such as staurosporine, but use a transition metal ion that coordinates up to six ligands to replace the carbohydrate moiety of staurosporine.15 The scaffold design includes a bidentate ligand that is able to target the metal complexes to the ATP-binding site. This mimics ATP and conventional indolocarbazole inhibitors, while the increased size of the bulky transition metal complex allows for exploration of additional chemical space at the edges of the ATP binding site specific to each kinase. Despite being conventional ATP-competitive inhibitors, the combination of unusual globular shape and rigid characteristic of these complexes facilitates the design of highly selective protein kinase inhibitors. It is worth noting that the coordinative bonds to the transition metal are considered to be kinetically stable and are expected to remain intact when exposed to the biological environment, thus avoiding metal-related cytotoxicities.16–18 However, druglikeness of such complexes, including metabolic stability, bioavailability, and pharmacokinetic properties, is not established yet and is subject to current studies. Regardless, this strategy has led to the development of specific and potent kinase inhibitors for GSK3,17 PIM1,19 PI3K,20 MST1,21 and BRAFV600E.22</p><p>Here, we present data on the development of potent and specific organometallic S6K1 inhibitors, EM5 and FL772. We show that FL772 binds to S6K1 with an IC50 value in the single digit nanomolar range at 100 μM ATP and that the more potent FL772 compound has a greater than 100-fold specificity over S6K2. Crystal structures of the S6K1 domain bound to the pan-kinase inhibitor staurosporine, EM5, and FL772 reveal that the organometallic inhibitors bind in the ATP binding pocket in a way that is distinct from staurosporine, likely explaining their more favorable potency and selectivity. Cellular data demonstrate that FL772 is able to inhibit S6K phosphorylation in yeast cells. The data provide an important starting point for the development of S6K inhibitors for possible therapeutic applications.</p><!><p>Inhibitors for S6K1 were initially identified through Millipore KinaseProfiler (Supporting Table 9). Ten different staurosporine-inspired organometallic ruthenium complexes were screened against a diverse panel of 283 protein kinases. This screen led to the identification of EM5 as a potential inhibitor of S6K1, with 7% activity at a concentration of 100 nM in the presence of 10 μM ATP. EM6, a similar complex replacing the isothiocyanate functional group with an isocyanate (Figure 1A), inhibited significantly less, exhibiting 54% activity under the same conditions. In the kinase panel, the EM5 inhibitor inhibited only 41 kinases (16%) to less than 10% activity, including S6K1 and the related S6K family members RSK1, RSK2, RSK3, and RSK4.</p><p>A radioactive kinase assay was used to determine the activity of S6K1 protein constructs prepared in baculovirus-infected insect cells in order to identify a construct that would be suitable for inhibitor testing. Initial tests showed that the full-length αI isoform of S6K1 (S6K(1−525)) and the isolated kinase domain (S6K(84−384)) had low kinase activity, although the full-length kinase showed more activity than the kinase domain (Figure 1B). We reasoned that the S6K1 protein constructs had low kinase activity because the full-length kinase contained the C-terminal autoinhibitory domain. To address this issue and express a more active kinase for further inhibitor studies, we prepared a S6K1(1−421) construct including both the T252 and T412 phosphorylation sites, based on previous data from Keshwani et al.23 indicating that the catalytic domain of the S6K1 aII isoform (residues 1−398) analogous to S6K1(1−421) of the αI isoform was highly expressed in insect cells. To further enhance the catalytic activity of S6K1(1−421), we prepared the T412E mutant to mimic phosphorylation at this position and coexpressed the protein with PDK1 to promote phosphorylation of T252. Preparation of the S6K1(1− 421, T412E, PDK1 activated) protein resulted in highly active kinase that was suitable for inhibition studies in vitro (Figure 1B).</p><p>Both EM5 and EM6 were assayed against S6K1(1−421, T412E, PDK1 activated) in a radioactive kinase assay and determined to have IC50 values of 33.9 nM and 23.5 μM, respectively, at 100 μM ATP (Figure 1C). For comparison, we also determined the IC50 of the nonspecific kinase inhibitor staurosporine, which had an IC50 value of 64.1 nM under the same conditions. Given the apparent specificity and potency of EM5, it became the lead structure for the development of second-generation organometallic S6K1 inhibitors.</p><!><p>Our initial attempts to cocrystallize the S6K1 kinase domain (S6K1KD, residues 84−384) bound to EM5 using several factorial screens were unsuccessful. However, we were able to reproduce the crystals reported by Sunami et al.24 of the S6K1 kinase domain in complex with staurosporine. We then soaked these crystals with high concentrations of the EM5 inhibitor in the hope of exchanging EM5 for staurosporine in the crystals. The EM5-soaked crystals diffracted to about 2.5 Å resolution and formed in space group P21 with two molecules per asymmetric unit. The structure was refined to Rwork and Rfree values of 19.15% and 22.21%, respectively, with excellent geometry (Table 1). During the refinement process, the inhibitors were modeled only after the protein was fully refined.</p><p>Similar to the previously published structures of the S6K1 kinase domain,24,25 the kinase domain is bilobal, consisting of an N-lobe composed largely of β-sheet and a C-lobe that is mostly α-helical. The crystal structure revealed that one protein molecule in the asymmetric unit was bound to staurosporine, while the other molecule was bound to EM5 in the same ATP binding site. This was confirmed by a 25σ Fo − Fc difference peak corresponding to the ruthenium atom in the ATP binding sites of one of the molecules before the inhibitor models were built into the electron density map (Figure 2A). This confirms that the EM5 inhibitor is an ATP-competitive inhibitor. The staurosporine and EM5-bound molecules in the asymmetric units are similar to each other with an overall rmsd of 0.68 Å for the shared atoms.</p><p>Although both staurosporine and EM5 bind in the ATP binding pocket, the more elaborate EM5 compound makes more extensive interactions, correlating with its greater S6K1 potency than staurosporine. Staurosporine forms hydrogen bonds to S6K1 through the backbone oxygen of Glu-222 of the kinase with the nitrogen of the methylamine of staurosporine, and the backbone nitrogen of Leu-175 and backbone oxygen of Glu-173 of the kinase hinge region contact the pyrrolidone oxygen and nitrogen of staurosporine, respectively. The ring system of staurosporine also makes van der Waals contacts to Leu-97, Lys-99, Gly-98, Val-105, Ala-121, Tyr-174, Glu-179, and Met-225 (Figure 2B).</p><p>The EM5 compound retains two hydrogen bonds between the backbone atoms of the hinge residues (Glu-173 and Leu-175) and the maleimide ring of EM5 and all of the van der Waals interactions with the EM5 ring system (Figure 2C). However, in addition to these contacts, EM5 makes additional protein interactions between the ruthenium coordination sphere and the protein. In particular, the additional isothiocyanate group of EM5 makes van der Waals interactions with Gly-100 and Val-105 of the kinase p-loop while the trithiacyclononane ligand makes van der Waals contacts to Gly-100 of the p-loop and Glu-179 and Glu-222 across from the p-loop where the protein substrate is likely to bind and Thr-235 and Asp-236 of the activation loop. Notably, each of these residues (Glu-179, Glu-222, Thr-235, and Asp-236) undergoes a dramatic movement toward the EM5 inhibitor relative to their positions in the staurosporine complex (Figure 2D). The binding of EM5 to S6K1 also introduces significant structural changes in the kinase relative to the staurosporine complex, and these structural changes appear to be indirectly caused by the 1,4,7-trithiacyclononane ligand of the EM5 inhibitor. The αD helix of the staurosporine complex is about two turns longer at its N-terminus than the corresponding helix of the EM5 complex where the corresponding segment takes on a β-strand conformation. This structural difference appears to be driven by the interaction of the tridentate ligand of EM5 with Glu-179.</p><p>On the opposite side of the inhibitor, the staurosporine complex contains an activation loop that is folded toward the ATP active site in an inactive conformation and does not have an ordered αC helix, as previously reported.25 Strikingly, the EM5 complex contains a well-defined αC helix of about two turns. The difference in disposition of the αC helix in the two structures appears to be nucleated around the N-terminal region of the activation loop that undergoes about a 6 Å movement toward the EM5 inhibitor relative to staurosporine. The movement of the activation segment toward the EM5 inhibitor appears to be mediated by the van der Waals interactions that are made between Thr-235 and Asp-236 with the trithiacyclononane ligand of EM5 (Figure 2D). This in turn provides enough room for the αC helix to form and to be stabilized by van der Waals interactions between Phe-237 of the activation loop and Leu-147 of the αC helix and a hydrogen bonding between Lys-123 of the small domain and Glu-143 of the αC helix (Figure 2E). Interestingly, these interactions are characteristic of the active conformations of kinases, even though the activation segment is in an inactive conformation. In contrast, the more out conformation of the activation loop of the staurosporine structure places Phe-237 and Asp-236 in positions that sterically occlude formation of the αC helix (Figure 2D). Taken together, while staurosporine binding to S6K1 places it in the inactive conformation, the S6K1/EM5 complex has characteristics of both the inactive and active kinase conformations.</p><!><p>The EM5 inhibitor bound to S6K1 with an IC50 value in the mid-nanomolar range, and a cocrystal structure confirmed that the inhibitor was binding in the ATP pocket of the kinase domain. As a result, EM5 was a promising lead structure for the design of more selective and potent S6K1 inhibitors. Previous data indicate that modifications to the pyridocarbazole moiety or the coordination sphere can have significant effects on binding affinities or kinase selectivity,26 and the structure of the S6K1/EM5 complex indicated several positions where chemical elaboration could enhance specificity for the kinase. A series of 64 derivatives of EM5 were designed with modifications at the pyridcarbazole heterocycle and the remaining ligand sphere and were tested for inhibition of S6K1 activity using both a radioactive kinase assay (Supporting Information Figure 3A) and an ADP-Glo assay with 1 μM compound.27 Twenty-five of these inhibitors were further screened using 250 nM compound (Figure 3A). The eight compounds that inhibited S6K1 to less than 25% activity, the equivalent of EM5, were assayed to determine their IC50 values (at 100 μM ATP). This analysis produced several compounds that inhibited S6K1 similarly or more potently than EM5 including SEK-222 (IC50 = 18.9 nM), SEK-243 (IC50 = 15.9 nM), SEK-214 (IC50= 14.8 nM), SEK-220 (IC50= 7.7 nM), and FL772 (IC50 =7.3 nM) (Supporting Information Figure 3B). Compound FL772 (Figure 3B) showed the most potent inhibition with an IC50 of 7.3 nM, (100 μM ATP and 2 nM of enzyme) (Figure 3C).</p><!><p>Testing the FL772 inhibitor at a range of concentrations from 1 μM ATP to 500 μM ATP resulted in an increase in the IC50 value concurrent with increasing ATP concentrations from 3.91 nM at 1 μM ATP to 25.79 nM at 500 μM ATP, confirming that FL772 is an ATP competitive inhibitor (Supporting Information Figure 4).27</p><p>To assess the specificity of FL772 for the S6K1 isoform, we also assayed the FL772 compound against recombinant S6K2 (Figure 3D), which resulted in an IC50 value of 975.0 nM, more than 100-fold greater than the IC50 value for S6K1. This confirms that the FL772 is indeed specific for the S6K1 isoform over the S6K2 isoform. For comparison, we tested the published S6K1 inhibitor PF-470867128 against S6K1 using the radioactive kinase assay in order to compare the potency of our compound with another specific S6K1 inhibitor (Figure 3E). The IC50 of PF-4708671 against S6K1 was determined to be 142.8 nM, nearly 20-fold higher than the IC50 of FL772 against S6K1.</p><p>To establish the kinase selectivity profile of FL772, we submitted the compound at a concentration of 100 nM to the KinomeScan profiling of Lead Hunter Discovery Services using 456 kinases (Supporting Information Table 10). FL772 demonstrated a high degree of kinase selectivity, with only 10 of 456 (2.2%) kinases showing less than 10% activity and only 26 of 456 (5.7%) kinases showing less than 35% activity (Supporting Information Figure 2). Like EM5, FL772 showed binding to the CAM, DAP, FLT, PIM, and RSK family member kinases. Unexpectedly, S6K1 itself exhibited 71% activity in the KinomeScan set of Lead Hunter Discovery Services with 70 of 456 (15.3%) showing a higher degree of binding than S6K1. The potency of FL772 therefore appears to be greater against S6K1 prepared by us than S6K1 prepared by Lead Hunter Discovery Services. We hypothesize that the different S6K1 kinase preparation and/or phosphorylation state by Lead Hunter Discovery Services leads to the different FL772 potencies for S6K1 measured by us and Lead Hunter Discovery Services. Nonetheless, taking together our analysis of FL772 against S6K1 and the kinase profiling results, we conclude that FL772 exhibits a high degree of kinase selectivity.</p><p>FL772 is based on the EM5 lead structure and differs by a methylated hydroxy group at the pyridocarbazole moiety and the thioether-containing tridentate ligand. The nine-membered ring of the symmetrical 1,4,7-trithiacyclononane ligand is replaced by a prochiral 1,4,7-trithiacyclodecane bearing a basic N-methylamino group in the 10-membered cyclic tridentate ligand. These structural changes in the tridentate ligand significantly increase the structural complexity of the inhibitor which is exemplified by the number of possible stereoisomers.</p><p>To determine the molecular basis for the increased potency of FL772 over EM5, we determined the X-ray crystal structure of FL772 in complex with S6K1 to 2.7 Å resolution (Table 1). The overall structure for the FL772-bound S6K1 is very similar to the EM5-bound structure, with an rmsd of 0.54 Å for all atoms. In particular the p-loop and activation loop and αD and αC helices take on nearly identical conformations, although the αC-helix is about one turn shorter at its N-terminal end (Figure 4A). In addition, the FL772 inhibitor retains all of the interactions made by EM5 but makes some additional interactions including a hydrogen bond between the backbone carbonyl of Lys-99 of the kinase p-loop with the amine ligand of the N-methyl-1,4,7-trithiacyclodecan-9-amine ligand while the methyl group makes a van der Waals interaction with Tyr-174 of the kinase hinge region (Figure 4B). These additional interactions of FL772 likely contribute to the greater potency of FL772 over EM5. The protrusion of the amine ligand into the region where protein substrate binds for phosphorylation probably also contributes to the greater inhibitor potency.</p><!><p>After establishing that FL772 functions as a potent ATP competitive S6K inhibitor in the in vitro radioactive kinase assay, we carried out studies to characterize its activity in cells. We first tested FL772 for overall cell cytotoxicity and downregulation of phosphorylation of S6 in the 451Lu (BRAFV600E mutant) and 451Lu-MR (BRAF/ MEK-inhibitor resistant) melanoma cell lines. Cells were treated with vehicle or a dose of inhibitor ranging from 0.001 to 10 μM for 22 h (Figure 5A). The 451Lu or 451Lu-MR cell lines exhibited neither a significant decrease in S6 phosphor-ylation nor a decrease in cell viability as indicated by the absence of cleaved PARP. There was also no change in total S6 or peEF2K levels, indicating that mTOR was not a target of FL772.</p><p>We also investigated the effect of FL772 in 293T cells at both 3 and 16 h of treatment (Figure 5B). As controls, we included the compounds AZD8055, PF-4708671, and FL1324. AZD8055 is an ATP-competitive dual mTORC1 and mTORC2 inhibitor that inhibits the phosphorylation of mTORC1 substrates S6K1 and 4EBP1 and the mTORC2 substrate AKT.29 PF-4708671 is a reported S6K1 inhibitor that does not affect the phosphorylation of AKT. The compound FL1324 is an FL772 analogue identified in our screen that inhibited S6K1 with an IC50 of 11 nM (Figure 3A). FL1324 replaces the fluorine of EM5 with a hydroxyl group and is thus a less polar molecule. Since previous studies using PF-4708671 demonstrated a significant reduction in S6 phosphorylation in 293T cells in 30 min,28 we tried both a short (3 h) and long (16 h) time point for treatment. As expected, the AZD8055 mTOR inhibitor showed a significant decrease in pS6 levels at both the S235 and S240 sites, along with a decrease in pAKT at T308 and S473. The PF-4708671 compound showed a modest decrease in phosphorylation of S6 at the 3 h time point, but this phosphorylation returned to near basal levels by the 16 h time point. There was no effect on the phosphorylation of AKT. Notably, neither the FL772 nor the FL1324 compound inhibited phosphorylation of S6 or AKT. Together, these results suggest either that the FL772 inhibitor has poor cell membrane permeability or that inhibition of S6K1 in cells does not significantly reduce S6 phosphorylation. The latter possibility is consistent with the fact that the structurally unrelated compound PF-4708671 also shows poor inhibition of S6 phosphorylation in cells and the observation that S6K2 also targets S6 for phosphorylation.30</p><p>To test if FL772 can inhibit S6 phosphorylation in a setting where S6K2 was not present, we investigated the ability of FL772 to inhibit S6 phosphorylation in budding yeast where a single kinase, Sch9p, is orthologous to human S6K1. We observed that treatment of wild-type budding yeast cells (BY4742) with FL772 significantly decreased the level of phosphorylated S6 in a dose-dependent manner (Figure 5C). At the highest dosage, S6 phosphorylation was reduced to a level similar to the sch9p knockout strain. These data suggest that FL772 functions as an inhibitor of S6 kinases in vivo in a yeast cellular system.</p><!><p>In this study, we developed an organometallic ruthenium compound to inhibit S6K1. Using the Millipore KinaseProfiler and radioactive kinase assays, we showed that the EM5 lead compound was a potent and selective S6K1 inhibitor, with 100 nM compound inhibiting 93% of S6K1 activity and only inhibiting 16% of 283 kinases by less than 90%. We found that the EM6 analogue in which an isocyanate group replaces an isothiocyanate is about 1000-fold less potent, implying that potency and specificity could be further optimized. A crystal structure of EM5 bound to S6K1 provided important molecular insights into EM5 inhibition of S6K1 and led to the development of FL772, a compound containing a novel ligand scaffold with an IC50 in the single digit nanomolar range for S6K1. A crystal structure of FL772 bound to S6K1 revealed the molecular basis for the compound's potent and selective inhibition of S6K1.</p><p>In order to investigate the efficacy of the FL772 inhibitor in cells, we evaluated the inhibitor in both human 293T and BRAFV600E mutant melanoma cells and in budding yeast. We found that FL772 was only able to inhibit S6 phosphorylation in yeast cells, suggesting that either the compound is unable to enter human cells, a significant shift in the IC50 of the compound occurs in the presence of physiological levels of ATP, or the uninhibited activity of S6K2 in human cells was sufficient to maintain S6 phosphorylation. Given that ruthenium compounds similar to FL772 have been used to successfully target MST1,21 PAK1,31 and PI3K20 in cells, we do not believe that the ruthenium compounds are unable to penetrate cells. The radioactive kinase assay prohibits measurements at physiological levels of ATP, but we were able to assay the activity of FL772 against S6K1 using an ATP range from 1 to 500 μM. This analysis revealed that the IC50 values increased with increasing ATP concentrations, consistent with FL772 binding competitively with ATP, as also confirmed with the crystal structure of the S6K1/FL772 complex. Interestingly, the IC50 ranged from 3.91 nM at 1 μM ATP to only 25.79 nM (a 6-fold increase) at 500 μM ATP, suggesting that S6K1 binds ATP relatively loosely and that FL772 is likely to displace ATP even at the higher physiological concentration of ATP. On the basis of these accumulated data, we propose that FL772 is unable to inhibit S6 phosphorylation in human cells because S6 is phosphorylated by the uninhibited S6K2.</p><p>S6K1 is closely related to S6K2, sharing 83% sequence identity in the catalytic domain.32 A study involving S6K1/2 knockdown in mice suggests that both S6K1 and S6K2 are required for full phosphorylation of S6 but that S6K2 may be the more important of the two for phosphorylation of S6.30 The MEK inhibitor AZD6244 showed additive effects on decreasing the phosphorylation of S6 in vitro when treated in combination with siRNA inhibition of both S6K1 and S6K2 combined, indicating the importance of S6K2 in the phosphorylation of S6.33 Furthermore, while normal tissues often express low levels of S6K2, overexpression of S6K2 is more common than overexpression of S6K1 in cancer cells.34–37 Taking these data together suggests that targeting S6K2 either alone or in combination with S6K1 inhibition may be a more viable option for direct S6 inhibition in melanoma and potentially other cancers.</p><p>Despite the similarities in the catalytic domain, homology modeling between S6K1 and S6K2 indicates an important difference in residue Tyr-174 that plays an important role in FL772 binding and is a cysteine in S6K2.38 This residue is located in the hinge region of S6K1 and makes an important van der Waals interaction with the methyl group of the secondary amine, which would not be made with a cysteine residue, suggesting that FL772 may not be a potent inhibitor for S6K2. Indeed, we confirmed in our study that FL772 inhibits S6K2 more than 100-fold more poorly than S6K1. The lethality of S6K1−/−/S6K2−/− knockout mice30 implies that S6K2 targeting may need to be selective for therapeutic value. To date, there are no commercially available S6K2-selective inhibitors, indicating a potential target for the next series of organometallic ruthenium inhibitors. Taken together, the studies reported here provide a potent and selective S6K1 inhibitor that should be useful to probe S6K1 function and as a starting point for the development of efficacious S6K inhibitors for therapeutic use.</p><!><p>Protein kinase profiling of EM5 was performed with the Millipore KinaseProfiler in a panel of 263 human protein kinases. Percentage of kinase activities were determined for EM5 and EM6 at 100 nM in the presence of 10 μM ATP. Measurements were performed in duplicate, and the average was taken. Kinase profiling of FL772 was performed with the DiscoveRx Kinome Screen using a panel of 456 kinases. Active-site-directed competition binding was determined in the presence of 100 nM FL772. See the Supporting Information for more details.</p><!><p>EM5 and EM6 were synthesized in analogy to related compounds reported.39 A detailed synthesis and characterization of FL772 is provided in the Supporting Information.</p><!><p>Full length human S6K1 cDNA (1−525) was purchased from Epitope (catalogue number IHS1380-97652397). S6K1 constructs (84−384, 1−421, 1−421 T412E) were subcloned into the pFASTbac HTB vector for protein expression. Sf9 cells were transfected with the recombinant bacmid DNA using Cellfectin (Invitrogen). Cells were harvested after being incubated for 48 h at 28 °C and stored at −80 °C. The 1−421 T412E construct was coexpressed with PDK1 to phosphorylate the T412E residue (cloned from cDNA purchased from OpenBioSystems). Frozen pellets of the S6K1 kinase domain, S6K1(84−384) used for crystallography were resuspended in sonication buffer (50 mM KPi, pH 7.0, 250 mM NaCl, 5% glycerol, 1:1000 PMSF) and sonicated at a power output of 5.5 for 120 s with 20 s intervals (Misonix Sonicator 3000). Lysates were cleared by high-speed centrifugation at 18 000 rpm for 35 min at 4 °C. Equilibrated Talon metal affinity resin (Clontech) was added to cleared lysates and incubated at 4 °C for 1 h with gentle shaking. The resin/lysate mixture was loaded into a gravity flow column, and the resin was extensively washed with wash buffer (50 mM KPi, pH 7.0, 250 mM NaCl, 5% glycerol). Protein was then eluted with elution buffer (50 mM KPi, pH 7.0, 250 mM NaCl, 500 mM imidazole, and 5% glycerol) in a single step. Pooled Talon eluent was diluted 3.5-fold in dilution buffer (50 mM KPi, pH 7.0, 5% glycerol) and loaded onto an SP anion exchange column pre-equilibrated with buffer A (50 mM KPi, pH 7.0, 50 mM NaCl, 5% glycerol). Protein was eluted with buffer B (50 mM KPi, pH 7.0, 500 mM NaCl, 5% glycerol) in a single step. Elution after the Q column was concentrated and loaded to a Superdex s200 column equilibrated with 50 mM Na citrate, pH 6.5, 300 mM NaCl, 1 mM DTT, 5% glycerol. The eluent was collected and concentrated to 3 mg/mL before protein was flash frozen in dry ice and stored at −80 °C. Purification of the 1−421 T412E construct was completed as above, with gel filtration on the Superdex s200 column using a buffer containing 25 mM Tris, pH 7.5, 200 mM NaCl, 1 mM EDTA, and 5% glycerol.</p><!><p>Full-length human S6K2 cDNA was purchased from GE Healthcare Dharmacon (RPS6KB2, clone identification number 2959036). The S6K2 1−423 construct, equivalent to S6K1 1−421, was subcloned into the pFASTbac HTB vector for protein expression. Sf9 cells were transfected and grown as described above. The construct was coexpressed with PDK1, similar to the S6K1 construct. Frozen pellets were purified identically to the S6K1 1−423 T412E pellets.</p><!><p>Each reaction mixture contained 5 μL of 5× reaction buffer (100 mM MOPS, pH 7.0, 150 mM MgCl2), 2 μL of inhibitor in 50% DMSO, 3.6 μL of S6K1 substrate peptide (RRRLSSLRA), 1 μL of BSA (20 mg/mL), 3.2 μL of S6K1 (concentration as described in Results), and 5 μL of ATP/ATP* mix (concentration as described in Results) in a total reaction volume of 25 μL. Reaction mixtures were incubated for 1 h at room temperature before being transferred to Whatman paper and washed with 0.75% phosphoric acid. Data were collected using a scintillation counter. All experiments were performed in triplicate. IC50 values were determined using sigmoidal dose response with a variable curve in Prism, version 5.</p><!><p>The S6K1 kinase domain crystals were obtained using room temperature hanging drop vapor diffusion by mixing equal volumes of protein (15 mg/mL) preincubated with 1 mM staurosporine with 20−25% (w/v) PEG335, 0.1 M Bis-Tris (pH 5.5−5.7), and 0.2 M LiSO4. Following the growth of crystals, crystal soaking was carried out by incubation with a final inhibitor concentration of 1 mM in cryoprotectant containing the well solution and 15% (w/v) glycerol for 4 h to overnight and flash frozen in liquid nitrogen. Diffraction images were collected at APS beamline 23ID with a 5 μm microbeam. The structures were determined by molecular replacement using the reported S6K1/staurosporine complex (PDB accession code 3A60) as a search model with the staurosporine removed from the coordinate file and refined with CNS and Coot. The inhibitors were modeled last into the refined structures. Simulated annealing omit maps were employed to unambiguously confirm the modeled inhibitors. For the EM5-soaked crystals, this revealed that one protein molecule in the asymmetric unit was bound to staurosporine while the other protein molecule was bound to EM5. For the FL772-soaked crystals, the asymmetric unit contained a single, domain-swapped monomer and only the FL772 inhibitor was modeled in the binding site. The structures were refined to convergence with a final Rwork = 19.15% and Rfree = 22.21% for the S6K1/EM5 structure and a final Rwork = 20.63% and Rfree = 23.01% for the S6K1/FL722 structure with excellent geometry (Table 1).</p><!><p>Human cell lines were cultured in RPMI (10-040-CM; Cellgro) supplemented with 5% fetal bovine serum and harvested at 70% confluence. For immunoblotting, cells were treated for the specified times with the indicated drugs, washed with cold phosphate buffered saline (PBS) containing 100 mM Na3VO4, and lysed using TNE buffer (150 mM NaCl, 1% (v/v) NP-40, 2 mM EDTA, 50 mM Tris-HCl, pH 8.0) supplemented with protease inhibitors (11697498001; Roche). Proteins were separated by SDS−PAGE and transferred to nitrocellulose membranes (9004700; BioRad). After blocking for 1 h in 5% (wt/vol) dry milk/Tris-buffered saline (TBS)/0.1% (v/v) Tween-20, membranes were incubated overnight at 4 °C with primary antibodies followed by incubation with Alexa Fluor-labeled secondary antibodies (IRDye 680LT goat-anti-mouse or IRDye 800CW goat-anti-rabbit antibodies (LI-COR Biosciences) for 1 h. β-Actin (A5441) and vinculin (V9131) antibodies were obtained from Sigma. p-AKT (4056, 4060), S6 (2317), p-S6 (4858, 5364), S6K1 (2708), p-S6K1 (9234), p-eEF2k (3691), peIF4B (3591), and cleaved PARP (5625) were obtained from Cell Signaling Technologies. Fluorescent images were acquired and by LI-COR Odyssey Imaging System.</p><!><p>Overnight cultures of wild-type yeast cells (BY4742) were diluted in synthetic complete (SC) medium and regrown at 30 °C to early log phase (OD600 of 0.2). FL772 was added to an aliquot of culture to the final concentration of 1, 10, 100, and 1000 nM. The treated cultures were further grown at 30 °C for 4 h before harvesting. A culture of sch9Δ cells (KS68) was grown and harvested in parallel as a control. Yeast cell pellets were lysed by spinning down cultures at ~3000 rpm for 3 min at 4 °C, washing with ice-cold water, and broken in lysis buffer as previously described.40 Whole cell extracts were separated on a 4−12% Bolt gel with MOPS running buffer (Life Technologies), followed by transfer to a PVDF membrane in a Mini Trans-Blot cell (Bio-Rad) at 20 V overnight. The blot was blocked with 3% BSA at room temperature for 2 h and then at 4 °C for 4 h, followed by incubation with primary antibodies, Phospho-S6 (Cell Signaling, catalog no. 2211, 1:1000 dilution), and GAPDH (Thermo, catalog no. MA5-15738, 1:1000 dilution) at 4 °C overnight. Incubation with secondary antibodies (anti-rabbit-DyLight-680 and anti-mouse-DyLight-800, Pierce, 1:10000 dilution) was carried out at room temperature for 1 h before imaging with Li-Cor Odyssey.</p>

PubMed Author Manuscript

The expression and function of Vascular Endothelial Growth Factor in Retinal Pigment Epithelial (RPE) cells is regulated by4-Hydroxynonenal (HNE) and Glutathione s-transferase4-4

It is well established that 4-Hydroxynonenal (HNE) plays a major role in oxidative stress-induced signaling and the toxicity of oxidants. Surprisingly our recent studies also demonstrate that low levels of HNE generated during oxidative stress promote cell survival mechanisms and proliferation. Since the expression and secretion of VEGF is known to be affected by Oxidative stress, during present studies, we have examined dose dependent effect of HNE on VEGF expression and secretion in a model of Retinal Pigment Epithelial (RPE) cells in culture. Results of these studies showed that while inclusion of 0.1\xce\xbcM HNE in the medium caused increased secretion of VEGF, its secretion and expression was significantly suppressed in the presence of >5 \xce\xbcM HNE in the media. These concentration dependent hormetic effects of HNE on VEGF secretion could be blocked by the over expression of GSTA4-4 indicating that these effects were specifically attributed to HNE and regulated by GSTA4-4. VEGF secreted in to the media showed angiogenic properties as indicated by increased migration and tube formation of HUVEC in matrigel when grown in media from RPE cells treated with 1 \xce\xbcM HNE. The corresponding media from GSTA4-4 over expressing RPE cells had no effect on migration and tube formation of HUVEC in matrigel. These results are consistent with earlier studies showing that at low concentrations, HNE promotes proliferative mechanisms and suggest that HNE induces VEGF secretion from RPE cells that acts in a paracrine fashion to induce angiogenic signaling mechanism in the endothelial cells. These findings may suggest a role of HNE and GSTA4-4 in oxidative stress induced proliferative retinopathies.

the_expression_and_function_of_vascular_endothelial_growth_factor_in_retinal_pigment_epithelial_(rpe

1. Introduction<!>2.1 Material<!>2.2 Cell lines<!>2.3 Cell viability assay<!>2.4 LDH assay<!>2.5 Transient transfection with hGSTA4<!>2.6 Western Blot Analysis<!>2.7 Wound healing assay<!>2.8 Estimation of VEGF<!>2.9 Tube-formation Assay<!>2.10 Statistical Analysis<!>3.1 HNE causes toxicity to RPE cells<!>3.2 HNE increases VEGF secretion in RPE cells<!>3.3 VEGF secreted from RPE cells increases migration and wound healing capability of HUVEC cells in vitro<!>3.4 RPE conditioned medium increases tube formation in HUVEC cells in an in vitro angiogenesis assay<!>4. Discussion

<p>4-hydroxynoninal (HNE) is the predominant end-product of lipid peroxidation (LPO) that contributes to cytotoxicity of oxidants via electrophilic attack on DNA and proteins [1,2]. It contributes to toxicity by inducing pro-apoptotic signaling through multiple pathways and also by necrosis. We and others have established that HNE is involved in regulation of gene expression and cell cycle signaling in a concentration dependent manner, and that its concentration in cells is regulated through a coordinated action of GSTA4-4, that catalyzes its conjugation to GSH and RLIP76, that transports GS-HNE conjugate out of cells. GstA4 knock-out mice having impaired HNE metabolism and increased HNE levels in tissues are more sensitive to the toxicity of oxidant chemicals/oxidative stress suggesting the role of HNE in the mechanisms of toxicity of oxidant xenobiotics and a protective role of GSTA4-4 against oxidative stress. However, recent studies have shown that unless subjected to oxidative stress, GstA4 knock-out mice have a normal phenotype, and surprisingly show a noticeable increase in their life span [3]. Other intriguing aspects of HNE-induced signaling observed in our studies and also reported by other investigators are its concentration dependent effects causing apoptosis at high concentrations, but promoting proliferation at low concentrations. These contrasting concentration dependent effects of HNE are observed in most of the cell types studied so far. The physiological significance of the contrasting hormetic effects of HNE on signaling and the mechanisms responsible for a surprisingly higher life span of GstA4 knock-out mice are not understood and need to be investigated.</p><p>Our recent studies show that besides being toxic HNE also induces defense mechanisms against oxidative stress to prevent its own toxic effects and protect the neighboring cells from a run-away apoptosis . These studies have shown that HNE induces defense mechanisms such as transcriptional activation of heat shock factors (HSFs), induction of HSP70, induction of anti-oxidant enzymes, and the activation of Daxx mediated anti-apoptotic mechanisms. More importantly, these studies suggest that oxidative stress (UV radiation or H2O2)-induced activation of these defense mechanisms requires HNE. Together, these findings suggest a requirement of HNE for the activation of defense mechanisms against oxidative stress for cell survival. VEGF is a homo-dimeric protein of about 34–45 kDa and it is implicated in angiogenesis in cancers and also in retinal microenvironment [4]. In vaso-proliferative disorders, including ARMD and diabetic retinopathy [5] the retinal pigment epithelial layer has been suggested to be the source of VEGF and it has been shown that oxidative stress causing agents and HNE can induce VEGF secretion from RPE cells. Thus, during present studies we have systematically investigated the dose dependent effect of HNE, an inevitable consequence of oxidative stress, on the expression of VEGF and VEGFR to address the hypothesis that at low concentration HNE activates various survival mechanisms. We have also evaluated the possible physiological consequences of HNE induced secretion of VEGF secreted from RPE cells shows angiogenic effects in an in vitro model. Here we show, for the first time that HNE exerts a hormetic (concentration dependent opposite effect) on the secretion of VEGF, i.e. at low levels HNE causes increased secretion of VEGF from RPE cells but at higher concentration, it inhibits VEGF secretion.</p><!><p>HNE was purchased from Cayman Chemical (Ann Arbor, MI). Bradford reagent, bisacrylamide, and SDS for SDS PAGE were obtained from Bio-Rad (Hercules, CA). Western blot stripping buffer was obtained from Pierce Co. (Rockford, IL). EGM-2 bullet kit medium was purchased from Lonza (Walkersville, MD). The cell culture medium DMEM, Lipofectamine 2000 transfection reagent, and fetal bovine serum were purchased from GIBCO (Invitrogen, Carlsbad, CA). All other reagents and chemicals including DMSO, G418 (geneticin), and 3-(4,5-dimethylthiazol-2-yl)-2,5-diphenyltetrazolium bromide (MTT), etc. were purchased from Sigma-Aldrich (St. Louis, MO).</p><!><p>The simian virus SV40-transformed human fetal male RPE 28 cells (Coriell Institute, Camden, NJ) that exhibit epithelioid morphology and retain physiological functions characteristic of the primary human RPE cells were cultured in standard medium containing 10% fetal bovine serum and antibiotics in a humidified incubator at 37°C in 5% CO2 atmosphere as described before [6]. The HUVEC (Lonza, Walkersville, MD) cells were grown in EGM-2 bullet kit. All studies were conducted by using cells of passages 10–20 for RPE and 2–6 for HUVEC. The cells were trypsinized and passaged every 3–4 days.</p><!><p>The cytotoxicity of HNE to RPE cells was measured by the 3-(4,5-Dimethylthiazol-2-yl)-2,5-diphenyl-2H-tetrazolium Bromide (MTT) assay as described before [7] with minor modifications. Briefly, 2 × 104 cells in 190 μl of medium were seeded in 96-well microtiter plates and allowed to attach for 24 h. Next day, HNE in 10 μL PBS was added to achieve the desired concentration. After 12 h incubation, 10 μl of a stock solution of MTT (5 mg/ml in PBS) was added to each well and the plates were incubated for additional 4 h at 37°C, centrifuged, and the medium was decanted. Cells were subsequently dissolved in 100 μl DMSO with gentle shaking for 2 h at room temperature, followed by measuring absorbance at 562 nm in a micro plate reader (El x 808 BioTek Instruments, Inc). A dose-response curve was plotted and the concentration of HNE causing a 50% reduction in formazan crystal formation (IC50) was determined.</p><!><p>To measure lactate dehydrogenase (LDH) released from the RPE cells, a commercially available cytotoxicity detection kit was used as briefly described below: Cells (2 × 104 in 190 μl of medium) were seeded in 96-well microtiter plates and allowed to attach for 24 h. The next day, 10 μl of PBS containing the desired concentration of HNE was added. After 12 h incubation, total culture medium was collected and centrifuged to remove contaminating cells and cellular debris. The volume of media was then measured. For assay, 100 μl of each sample was transferred to a 96 well microtiter plate, 100 μl of LDH reagent mixture was added to each well and incubated for up to 30 min at room temperature. After incubation 50 μl stop solution was added and the absorbance of samples was measured at 490 nm.</p><!><p>Transient transfection of RPE cells was performed with hGSTA4 as described by us before [6,13]. Briefly, RPE cells at a density of 5 × 105 cells per 100 mm Petri dish were plated and the dishes having >70% confluent cells were used for the transfection with 24 μg of either empty pTarget-T vector (VT) or the pTarget vector containing the open reading frame (ORF) of the restored Kozak hGSTA4 sequence (hGSTA4-Tr). Lipofectamine 2000 reagent (Invitrogen, Carlsbad, CA) was used for transfection as per the manufacturer s instructions.</p><!><p>The cells with or without specified treatment(s), were pelleted, washed thrice with PBS, re-suspended in RIPA lysis buffer (50 mM Tris-HCl, pH 7.5; 1% NP-40; 150 mM NaCl; 1 mg/ml aprotinin; 1 mg/ml leupeptin; 0.5 mM phenylmethylsulfonyl fluoride; 1 mM Na3VO4; 1 mM NaF) at 4°C for 30 min, and lysed by sonication. Cell debris was removed by centrifugation at 14,000 g for 30 min at 4°C to obtain clear extracts. Western blot analyses were performed with the extracts containing 25–75 μg protein. Protein was determined by the method of Bradford [8] throughout these studies.</p><!><p>In vitro wound-healing assay was performed using previously described method [9]. Briefly, HUVECs (8×104) were grown to confluence on 12-well tissue culture plates for 48 h, and then starved in serum-free maintenance medium for 4 h. A "wound" was made by scraping the middle of the cell monolayer with a sterile 10 μL micropipette tip. Floating cells were removed by extensive washing with PBS and the conditioned media obtained from RPE cell cultures with or without HNE treatment was added to each of the wells. Cells were stained with crystal violet and photographed using a phase-contrast microscope at 12 h after wounding.</p><!><p>RPE cells (2×104) grown in 96-well plates were incubated with serum-free medium containing varying concentrations (0–20 μM) of HNE. At the end of 12 h incubation, supernatants were collected and secreted VEGF was measured using an enzyme-linked immunosorbent assay (ELISA) according to manufacturer s (R&D Systems, Inc, Minneapolis, MN) recommendations and absorbance was measured using a microtiter plate reader (Fischer Scientific, PA) with a test wavelength of 450 nm and a reference wavelength set at 550 nm.</p><!><p>Growth factor-reduced (GFR) Matrigel (BD Biosciences, New Bedford, MA) was added to wells of a cold 96-well plate (45 μl/well), and then incubated at 37°C for 1 h to allow gelling. Overnight starved HUVEC cells were seeded onto the Matrigel. Medium collected from HNE-treated and untreated RPE conditioned medium was then placed on these cells. Cells were observed directly and number of tubes and length of the tubes were measured using Image J software under dark field illumination on an inverted light microscope at low power. Photographs were also taken with a digital camera under the same conditions.</p><!><p>Statistical analyses were performed by using Student s t test. P value < 0.05 is considered as statistically significant. Representatives of p-value in figures include "*"<0.05. Analysis of Variance (ANOVA) was used for the analysis of VEGF secretion on exposure to HNE.</p><!><p>HNE cytotoxicity to cells is well known [10–14]. To assess the effect of HNE accumulation on RPE cells in the retinal microenvironment as a result of oxidative stress, we examined the cytotoxicity of HNE to RPE cells. These studies were particularly conducted to assess HNE toxicity at physiologically relevant, low concentrations. HNE treatment decreased the viability of RPE cells in a dose-dependent manner. At physiologically relevant concentrations (below 5 μM), the toxicity of HNE was only minimal but it increased remarkably at higher concentrations. As shown in Figure 1A, the IC50 of HNE was found to be 45 μM ± 4.2 that was similar in range to the value previously reported [6,15] by us for these cells. The results of LDH leakage from the cells we consistent to those of MTT assay indicating only minimal HNE toxicity at physiologically relevant concentrations. The protective role of hGSTA4-4, against HNE cytotoxicity to RPE cells was evaluated by comparing the effect of HNE on empty vector-transfected (VT), and hGSTA4-transfected (GSTA4-Tr) cells using the MTT assay. The over expression of hGSTA4-4 in transiently transfected cells was confirmed by Western blot analysis where a robust expression of hGSTA4-4 was seen. Cells over expressing hGSTA4-4 were significantly protected against HNE toxicity (Fig. 1B) as illustrated by a shift in HNE IC50 value (IC50 values for the vector, and hGSTA4 transfected cells; 45 μM ± 4.2 μM, and 59 ± 3.7 μM, respectively).</p><!><p>Previous studies from our laboratory have shown that HNE is involved in oxidative stress-induced signaling in RPE cells [6,15]. Since VEGF secretion has been shown to be affected by oxidative stress and also by HNE, we studied dose dependent effect of HNE on the expression and secretion of VEGF in these cells. The results of these experiments presented in Fig. 2A suggested that the inclusion of HNE in the media at concentrations as low as of 0.1 to 1.0 μM caused a remarkable increase in the secretion of VEGF from RPE cells into the media. Optimal secretion of VEGF was observed at 1.0 μM HNE that showed a plateau thereafter up to about 5 μM of HNE and decreased when concentrations were higher than 5 μM. HNE-induced secretion of VEGF was abrogated in hGSTA4 transfected cells indicating HNE as the causative factor for induction of VEGF secretion. The effect of HNE on the expression of VEGF receptor (VEGFR-2) was also studied. The results of these experiments (Fig. 2B) showed that 1 μM HNE which optimally induced VEGF secretion caused a significant suppression of VEGFR-2 expression in RPE cells and at high HNE concentration VEGFR-2 expression was almost completely abrogated.</p><!><p>VEGF is implicated as a pro-angiogenic factor that provokes neovascularization in ARMD [16] and RPE has been recognized as an important source for VEGF in this context [5]. Since cell migration is necessary for endothelial cells in neovascularization, we performed an in vitro wound healing migration assay [17] to evaluate the physiological significance of HNE-induced secretion of VEGF in the medium. As described previously for this assay [18], a confluent monolayer of HUVEC cells was scraped across with a pipette tip to create a wounded region. The conditioned medium from RPE cells exposed to 1 μM HNE (that showed maximal secretion of VEGF) or that from control cells without HNE exposure was applied to HUVECs and wound closure or cell migration into the wounding area after 12 h was compared. As shown in Fig. 3, HUVEC cells in the conditioned medium from HNE treated RPE cells showed significantly greater progressive reoccupation of the wounded region as compared to those cells treated with the control RPE conditioned medium without HNE treatment. Similar to the control media, the media from hGSTA4 transfected cells (that did not have increased VEGF secretion Fig. 2A) also did not show any accelerated wound healing in HUVEC cells. Together, these results show that increased secretion of VEGF by low levels of HNE promotes proliferation of endothelial cells in an in vitro model that is consistent with our hypothesis that at low concentrations HNE promotes cell survival mechanisms.</p><!><p>Although angiogenesis is an outcome of multifaceted functionality of several kinds of cells, the tube formation of endothelial cells is a key step in this process [19,20]. To investigate the pro-angiogenic effects of VEGF secreted from HNE-treated RPE cells, we performed in vitro matrigel tube formation assay [20,21]. HUVEC cells, plated on matrigel were grown with and without HNE treated RPE conditioned medium. The length and number of tubes formed in both the cells were measured by Image J software and compared. As shown in Fig. 4, HUVEC cells cultured in media of 1 μM HNE treated RPE cells (with maximal VEGF secretion, Fig. 2) showed increased tube formation as compared to the media of the control cells without HNE treatment. Also, no significant difference in the length and number of tubes formed in HUVEC cells was observed in the presence of conditioned medium from hGSTA4 transfected RPE cells with or without HNE treatment. Together, these results suggest that HNE-induced VEGF secretion contributes to angiogenesis in surrounding epithelial cells that could be inhibited by GSTA4-4.</p><!><p>In present communication we demonstrate that exposure of RPE cells to HNE increases secretion of VEGF from these cells in a concentration dependent manner. Similar to the reported hormetic effect of HNE on EGFR [6], HNE at low levels promotes VEGF secretion in cells and inhibits this secretion at high levels. In a series of experiments we also demonstrate the physiological consequences of the secretion of VEGF by low levels of HNE that may correspond to low levels of oxidative stress. These experiments in which the effect of VEGF secreted in the media of HNE-treated RPE cells was examined on migration and angiogenesis of HUVEC cells were conducted to simulate the effect of HNE-induced secretion of VEGF on retinal micro-environment. The results of these experiments clearly showed that HNE-induced secretion of VEGF promoted increased migration and angiogenesis in HUVEC cells. Furthermore, these effects of HNE were suppressed in GSTA4 transfected cells indicating that the secretion of VEGF and the consequent physiological effects could be specifically attributed to HNE. Increased proliferation of endothelial cells is associated with the excess angiogenesis in the retina. The secreted VEGF exerts its paracrine effect on the surrounding endothelial cells resulting in the proliferation of endothelial cells in the retinal micro-environment that may contribute to the pathologic processes for retinal neo-vascularization. HNE-induced signal to suppress VEGFR-2 expression in RPE cells observed in present studies may be to block the autocrine effects of secreted VEGF on RPE cells. Physiologically relevant concentrations of HNE have been shown to induce EGFR pathway involved in the proliferation of RPE cells [6]. Here we show that at these concentrations, HNE is also associated with the increase in VEGF secretion and proliferation of surrounding endothelial cells. Together, these findings suggest that at physiologically relevant low concentrations HNE seems to promote protective mechanisms but its chronically increased levels due to persistent oxidative stress it may contribute to proliferative retinopathy in the long run.</p><p>Our previous studies have shown that oxidative stress-induced signaling in various cell types is mediated via HNE [12,22]. We have previously shown that even a non-toxic mild transient oxidative stress such as UVA exposure to 5 min leads to HNE formation. Moreover, the cells pre-exposed to 5min UVA shows a significant resistance to oxidative stress induced cell death. Also, it is documented that low levels of HNE promote induction of defense mechanisms such as activation of HSF1, Nrf2, Daxx and EGFR. Thus the results of present studies taken together with these previous studies suggest a novel role of HNE in stress induced signaling. In this model, low levels of HNE generated in cells during mild oxidative stress would act as a sensor of oxidative stress to initiate a multitude of signals for promoting mechanisms for defense against oxidative stress. This would be consistent with the observed resistance of mild stress preconditioned cells [23,24], and the activation of HSF1 [25], Nrf2 [26], Daxx [27], EGFR [6,22] and the secretion of VEGF by low levels of HNE observed here. Thus initial low level of HNE formed during oxidative stress seem to minimize the toxicity of initial oxidative insult by increasing the threshold of stress only beyond which cytotoxicity would be manifested. This model would also explain studies showing activation of HSF1 [25], Nrf2 [26], Daxx [27], and p53 [28] in the tissues of GSTA4 null mice [15] and increased life span of these mice [3].</p>

PubMed Author Manuscript

Anchoring Effects of Surface Chemistry on Gold Nanorods: Modulates Autophagy

Gold nanorods (Au NRs) have been receiving extensive attention owing to their extremely attractive properties which make them suitable for various biomedical applications. Au NRs could induce nano-toxicity, but this trouble could turn into therapeutic potential through tuning the autophagy. However, the autophagy-inducing activity and mechanism of Au NRs is still unclear. Here we showed that surface chemical modification can tune the autophagy-inducing activity of Au NRs in human lung adenocarcinoma A549 cells. CTAB-coated Au NRs induce remarkable levels of autophagy activity as evidenced by LC3-II conversion and p62 degradation, while PSS- and PDDAC-coated Au NRs barely induce autophagy. More importantly, we also demonstrated that the AKT-mTOR signaling pathway was responsible for CTAB-coated Au NRs-induced autophagy. We furthermore showed that CTAB-coated Au NRs also induces autophagy in human fetal lung fibroblast MRC-5 cells in a time-dependent manner. This study unveils a previously unknown function for Au NRs in autophagy induction, and provides a new insight for designing surface modifications of Au NRs for biomedical applications.

anchoring_effects_of_surface_chemistry_on_gold_nanorods:_modulates_autophagy

Introduction<!>Synthesis and Characterization of Au NRs<!>Autophagic cell death induced by CTAB-coated Au NRs<!>Au NRs have no effect on the function of lysosomes<!>The AKT\xe2\x80\x93mTOR signaling pathway is involved in CTAB-coated Au NR-induced autophagy<!>Conclusions<!>Materials and antibodies<!>CTAB-coated Au NRs<!>Polyelectrolyte-coated Au NRs<!>Characterization of Au NRs<!>CCK-8 assay<!>Transmission electron microscopy analysis<!>Evaluation of lysosomal acidity<!>Immunofluorescence analysis<!>Western blotting<!>Statistical analysis

<p>Owing to their unique properties, which depend on their shape, size, and aspect ratio, nanomaterials have attracted intense interest from scientists as therapeutic and diagnostic agents.1-3 Gold nanorods (Au NRs) are one of most promising nanomaterials because their size, aspect ratio (ratio of length to diameter) and coating can all be easily controlled. Their applications in the biomedical field include cell and animal imaging, drug and gene delivery, and therapy and diagnosis in many diseases.4-8 The most convenient synthesizing fashion of Au NRs is seed-mediated method using cetyltrimethylammonium bromide (CTAB), and CTAB is a well-known toxic cationic surfactant. Consequently the CTAB-Au NRs should be further coated by negatively charged PSS, positively charged PDDAC and PEG for biomedical applications. Recent studies have reported preliminary research into the intracellular localization, uptake and cytotoxicity of Au NRs in cells and whole animals. Qiu et al. showed that the aspect ratio and surface chemistry mediated cellular uptake and cytotoxicity of Au NRs.9 In addition, Wang et al. demonstrated that Au NRs can selectively target to mitochondria and induce cell death for cancer therapy.10 Our previous studies also indicated that the penetration and thermotherapy efficacy of Au NRs were determined by surface chemistry in multicellular tumor spheroids.11 However, the mechanism of cell death induced by Au NRs is still unclear, and further study of the cell death mechanism is urgently needed before Au NRs are widely used in clinical studies.</p><p>Autophagy is a lysosome-based degradation process by which eukaryotic cells self-digest long-lived proteins and dysfunctional organelles, and thereby maintains intracellular homeostasis. Autophagy also plays an essential role in a variety of human diseases, including cancer, neurodegenerative disorders and infectious diseases.12-14 In general, autophagy is regarded as a prosurvival mechanism. However, increasing evidence demonstrate that autophagy play a key role in cell death.15-17 Nanomaterials have been suggested to play different roles in autophagy and cell death due to their specific properties.18, 19 Recently, several studies have demonstrated that autophagy can be induced by a variety of nanomaterials, including quantum dots (QDs), polyamidoamine (PAMAM), single-walled carbon nanotubes (SWNTs) and lanthanide-based nanocrystals.20-25 Our previous results show that gold nanoparticles can block autophagy in a size-dependent manner by increasing the lysosomal pH. In this work, we investigated the effect of Au NRs with different surface coatings on autophagy activity, and analyzed the underlying mechanisms and signaling pathway involved in Au NRs-induced autophagy.</p><!><p>In order to investigate the surface chemistry-dependent induction of autophagy activity by Au NRs, we synthesized Au NRs with three different polymer coatings: cetyltrimethylammonium bromide (CTAB), polystyrene sulfonate (PSS) and poly (diallyldimethylammonium chloride) (PDDAC) as described in the Methods section. The CTAB, PSS and PDDAC frequently was used a model polymer coating in Au NRs. As shown in Fig. 1a, the morphology and size of the Au NRs were measured and statistically analyzed based on TEM images. The aspect ratio of all Au NRs was 4, and the mean size of the Au NRs was 55 nm × 14 nm (length × diameter). The UV-Vis-NIR absorption spectra showed that the maximum absorption peaks were close to 808 nm, and the visible absorption spectrum was correlated with shape, size, monodispersion and surface stabilization of the Au NRs (Fig. 1b). Zeta potential is usually used to predict the surface charge and stability of nanomaterials in solution. We measured the zeta potentials and found that the CTAB-coated Au NRs and the PDDAC-coated Au NRs were positively charged, whereas the PSS-coated Au NRs were negatively charged (Fig. 1c-e). This result is consistent with previous studies on coated Au NRs under the same conditions. It has been previously reported that the size and zeta potential of Au NRs are critical for the nonspecific adsorption of serum proteins onto the surface, and the presence of these proteins on the surface of the nanoparticles is related to the cellular uptake and cytotoxicity.9 We further examined the zeta potential of the surface-coated Au NRs after incubating them with DMEM containing 10% fetal bovine serum for 2 hours. The surface charges of all the Au NRs immediately became negative, as we expected (Figure S1 in Supporting Information). Thus the results of the following autophagy studies most likely depend on the surface coating of the Au NRs instead of their zeta potential.</p><!><p>In order to understand the mechanism of cell death induced by Au NRs, we first examined the cytotoxicity of CTAB-, PSS- and PDDAC-coated Au NRs using the autophagy inhibitor 3-methyladenine (3-MA). Starvation, the most widely used inducer of autophagy, was employed as a positive control. The results showed that 70 pM Au NRs exhibited coating-dependent toxicity to human lung adenocarcinoma A549 cells (Fig. 2a). The CTAB-coated Au NRs have notable cytotoxicity, but when further coated with PSS and PDDAC, the cytotoxicity of the Au NRs decreased greatly to negligible levels. Interestingly, we observed that the cytotoxicity induced by CTAB-coated Au NRs could be rescued by treatment with 3-MA, suggesting that CTAB-coated Au NRs likely induced cell death via autophagy.</p><p>To further confirm our hypothesis that CTAB-coated Au NRs induced cell death through autophagy, we used TEM to examine the formation of autophagosomes, which are key intermediate vesicles in the autophagy pathway. TEM imaging revealed that treatment with CTAB-coated Au NRs significantly increased autophagosomes formation in A549 cells (Fig. 2b), compared to those treated with PSS- and PDDAC-coated Au NRs. We next examined the conversion of the autophagy-related protein microtubule-associated protein 1 light chain 3 (LC3).26 LC3 has two isoforms, LC3-I, which is cytosolic, and LC3-II, which associates with autophagosomal membranes. Autophagy is characterized by an increase in LC3-II protein and LC3-positive puncta. We analyzed the expression of LC3 by western blotting and found that CTAB-coated Au NRs induced a remarkable increase in LC3-II expression compared to the control, while PSS- and PDDAC-coated Au NRs did not enhance the LC3-II/LC3-I ratio (Fig. 3a and 3b). Consistently, CTAB-coated Au NRs but not PSS- and PDDAC-coated Au NRs increased LC3 puncta formation (Fig. 3c). Moreover, we also observed that the number of PDDAC-coated Au NRs in cells was much greater than that of CTAB- or PSS-coated Au NRs. These results indicated that the level of Au NR uptake was not related to the level of autophagy induction. In addition, CTAB-coated Au NRs also induce accumulation of LC3-II in a HeLa cell line which stably expresses GFP-tagged LC3 (Figure S2). To confirm these results, the human fetal lung fibroblast (MRC-5) cells, a normal cell line, was chose to analyze LC3-II conversion after treated with CTAB-coated Au NRs (Figure S3). CTAB-coated Au NR also induces autophagy in MRC-5 cells in a time-depended manner. Taken together, these data demonstrate that CTAB-coated Au NRs can influence the autophagy process.</p><!><p>Autophagosome accumulation and the enhancement of the LC3-II/LC3-I ratio may be due to inhibition or induction of autophagy.27 It is well known that lysosome function, especially the internal pH, plays vital roles in maintaining cellular processes, including autophagy. LysoSensor Green DND-189 shows an acidification-dependent increase in fluorescence intensity, which allows us to monitor pH changes in lysosomes.28, 29 Our previous study showed that gold nanoparticles decreased the alkalinization of lysosomes and therefore inhibited autophagy in a size-dependent manner.30 In order to examine whether Au NRs can affect lysosome pH, A549 cells were labeled with LysoSensor Green DND-189 dye after treatment with different Au NRs. Confocal microscopy analysis demonstrated that lysosome pH was unaffected in Au NR-treated cells (Fig. 4a). Flow cytometry analysis also confirmed this result (Fig. 4b and S4).These results indicate that CTAB-, PSS- and PDDAC-coated Au NRs have no effect on lysosome pH and do not influence the autophagy process by this mechanism.</p><!><p>To further distinguish whether the main effect of CTAB-coated Au NRs is on autophagy induction or blockage of autophagic flux, the ratio of LC3-II/LC3-I was checked in the presence and absence of a lysosome degradation inhibitor, bafilomycin A1 (BFA1). It was found that BFA1 further increased the CTAB-coated Au NR-induced conversion of LC3-II (Fig. 5a and 5b). This result indicates that the CTAB-coated Au NRs themselves actually induced autophagy. p62 (also known as SQSTM1/sequestome1), a substrate that is preferentially degraded by autophagy, was also monitored by western blotting analysis.31 As shown in Fig. 5c and 5d, starvation induced rapid down-regulation of p62. CTAB-coated Au NRs also caused marked degulation of p62, while PSS- and PDDAC-coated Au NRs had no effect on p62 protein level. These data suggest that CTAB-coated Au NRs promote autophagy to accelerate p62 turnover through blockade of autophagic flux. In mammalian cells, autophagy is regulated by several classical signaling pathways, most of which involve the inhibition of a serine/threonine protein kinase, mammalian target of rapamycin (mTOR).32, 33 The mTOR protein exists in a phosphorylated form and suppresses autophagy under normal conditions, but when the level of phosphorylated mTOR is down-regulated, such as during starvation, autophagy is up-regulated. To test whether CTAB-coated Au NRs induced autophagy occurred via inhibition of mTOR activity, the level of phosphorylated p70 S6 kinase (p-p70S6K), an indicator for mTOR activity, was examined by western bloting analysis. It was found that both CTAB-coated Au NRs and starvation significantly decreased the level of p-p70S6K), while PSS- and PDDAC-coated Au NRs no effect on the expression of p-p70S6K (Fig.5e). This demonstrated that CTAB-coated Au NRs induced autophagy through down-regulation of mTOR activity. Previous studies have shown that the effect of mTOR on autophagy can be regulated by the PI3K-AKTTSC1/2 pathway.34 To further understand the signaling pathway involved in CTAB-coated Au NR-induced autophagy, we examined phosphorylated AKT, a vital marker upstream of the mTOR pathway.35 As shown in Fig. 5f, the phosphorylated AKT level was significantly decreased when cells were incubated with CTAB-coated Au NRs, and starvation caused similar reductions. PSS- and PDDAC-coated Au NRs were analyzed under the same conditions, and had no effect on the levels of phosphorylated AKT, as we expected. Together, these results revealed that CTAB-coated Au NRs induced autophagy through the AKT-mTOR signaling pathway while other surface coatings did not induce autophagy.</p><!><p>Accumulating evidence show that Au NRs cause cell death for cancer therapy, however, the molecular mechanism to trigger cytotoxicity is poorly understood. As illustrated in Fig. 6, we have provided compelling evidence that CTAB-coated Au NRs promote autophagy while other surface-modifying polymers did not cause obvious autophagy process, indicating that autophagy-inducing activity of Au NRs is tuned by surface chemistry. Meanwhile, CTAB-coated Au NR also induces autophagy in human fetal lung fibroblast MRC-5 cells in a time-dependent manner. Furthermore, we demonstrated that the AKT-mTOR signaling pathway is involved in the induction of autophagy by CTAB-coated Au NRs. Herein, this study unveils a previously unknown function for CTAB-coated Au NRs in autophay induction and provides guidance for rational design of surface coatings for nanoparticles in biomedical applications and pharmaceutical therapy.</p><!><p>Dulbecco's Modified Eagle's Medium (DMEM) and fetal bovine serum (FBS) were obtained from Hyclone (Logan, UT, USA). Cetyltrimethylammonium bromide (CTAB), hydrogen tetrachloroaurate (III) trihydrate (HAuCl4·3H2O), silver nitrate (AgNO3), L-ascorbic acid, and sodium borohydride (NaBH4) were purchased from Alfa Aesar. Poly (sodium-p-styrenesulfate) (PSS, molecular weight: 70,000) and poly (diallyldimethyl ammonium chloride) (PDDAC, 20%) were obtained from Aldrich. Lysosensor green DND-189 (L-7535) was purchased from Invitrogen. The polyclonal anti-LC3 antibody (NB100-2220) was obtained from MBL. The anti-p62 antibody was purchased from MBL. The monoclonal anti-p70S6K (#2708) and anti-phospho-p70S6K antibodies (#9206) were obtained from Cell Signaling Technology. Deionized water (18.2 MΩ cm−1) produced by a Milli-Q system (Millipore Co., USA) was used in all the experiments. Unless specified, all of the commercial products were used without further purification.</p><!><p>As described previously, the CTAB-coated Au NRs were synthesized by seed-mediated growth.36, 37 First, the CTAB-capped Au seeds were obtained by chemical reduction of HAuCl4 with NaBH4: 7.5 ml CTAB (0.1 M) was mixed with 100 μl HAuCl4 (24 mM) and diluted with water to 9.4 ml. Then, 0.6 ml ice-cold NaBH4 (0.01 M) were freshly prepared and added while stirring magnetically. After 2 min of vigorous stirring, the seed solution was kept at room temperature (25 °C) and used within 2-5 h. Second, the Au NR growth solution, consisting of 100 ml CTAB (0.1 M), 2 ml HAuCl4 (24 mM), 2 ml H2SO4 (0.5 M), certain amount of AgNO3 (10 mM), and 800 μl ascorbic acid (0.1 M) was prepared. The amount of Ag ions added was used to control the aspect ratio of the Au NRs. Afterwards, 240 μl seed solution was added to the above growth solution to initiate the growth of the Au NRs. The reaction was stopped after 12 h and the outcome was centrifuged at 8000 rpm for 10 min. The precipitates were collected and re-suspended in deionized water. Au NRs with an aspect ratio of 4 were obtained.</p><!><p>The preparation of multilayer polyelectrolyte-coated Au NRs was performed via a layer-by-layer approach according to references.38, 39 The multilayer polyelectrolyte-coated Au NRs were synthesized by sequentially coating negatively charged PSS and positively charged PDDAC onto the as-synthesized CTAB-coated Au NRs. For PSS coating, 12 ml Au NRs were centrifuged at 12000 rpm for 10 min, and the precipitate was dispersed in 12 ml of 2 mg/ml PSS aqueous solution (containing 6 mM NaCl). The solution was stirred magnetically for 3 h. Afterwards, this solution was centrifuged at 12,000 rpm for 10 min, and the precipitate was redispersed in water. For further coating with PDDAC, a similar procedure was applied to the PSS-coated Au NRs.</p><!><p>UV-Vis-NIR absorbance, transmission electron microscopy (TEM) and dynamic light scattering (DLS) were used for characterization of the optical properties, size and zeta potential of the particles. The UV-Vis-NIR absorption spectra were measured with a Lambda 950 UV/vis/NIR spectrophotometer (Perkin-Elmer, USA). The size and morphology of the Au NRs was determined using a Tecnai G220 STWIN transmission electron microscope (FEI Company, Philips, Netherlands) with 200 kV acceleration voltages. The zeta-potential distribution of the Au NRs was measured by a Zetasizer Nano ZS (Malvern, England), at 25 °C.</p><!><p>The human lung adenocarcinoma A549 cell line was purchased from ATCC, and cultured in DMEM supplemented with 10% FBS, 100 U/ml penicillin/streptomycin at 37 °C in a 5% CO2 incubator at 95% humidity. A549 cells were seeded at a density of 5 ×103 cells per well in 96-well plates in culture medium and incubated overnight. The cells were then pre-treated with or without 3-MA for 4 h and the medium was then replaced with 100 μL 70 pM of Au NRs, Au NRs plus 3-MA or 3-MA. After a further incubation period of 24 h, cytotoxicity assays were performed using CCK-8 Kits (Dojindo Molecular Technologies, Tokyo, Japan). Absorbance was detected at 450 nm with a TECAN Infinite M200 microplate reader (Tecan, Durham, USA). All experiments were conducted in triplicate.</p><!><p>3×105 A549 cells were seeded in 33 mm dishes overnight, then incubated with 70 pM Au NRs for 4 h. The cells were fixed in 2.5% glutaraldehyde in 0.01 M PBS (pH 7.4) for 10 min at room temperature. The fixed cells were then embedded, sectioned and double stained with uranyl acetate and lead citrate for observation under the transmission electron microscope (H-7650B).</p><!><p>A549 cells were collected from growth media after treating with 70 pM Au NRs for 4h and washed three times with 0.01 M PBS. The cells were incubated for 30 min under growth conditions with 500 μL of prewarmed medium containing 2 μmol/L LysoSensor Green DND-189 dye (Invitrogen). After washing, the cells were resuspended in 0.01 M PBS and immediately analyzed by an Attune® acoustic focusing cytometer (Applied Biosystems, Life Technologies, Carlsbad, CA). The cytometer was purchased jointly with the Nanotechnology lab for bioapplications, and was installed by Life Technologies Corp. in the National Center for Nanoscience and Technology, China. The green fluorescence was collected within 1 min from a population of 20 000 cells.</p><!><p>Immunofluorescence analysis was performed as described previously.40 Cells were grown in six-well plates on glass coverslips and treated with 70 pM Au NRs for 4 h. Before incubation with antibodies, cells were fixed for 30 min with 4% paraformaldehyde at room temperature, permeabilized for 10 min with 0.2% Triton X-100 (Sigma) and blocked for 2 h in 10% horse serum albumin. Incubation overnight at 4 °C with the primary antibody was followed by the secondary antibody for 3 h at room temperature. Endogenous LC3 was detected with anti-LC3 antibody. Confocal laser scanning was done on a Zeiss LSM 710 Laser Scanning Microscope.</p><!><p>After incubating with 70 pM Au NRs for 4 h, A549 Cells were lysed in lysis buffer, denatured at 100 °C for 10 min and then the proteins were separated by SDS-PAGE before transferring to PVDF membranes. Membranes were blocked for 1 h with 0.01 M PBS/0.05% Tween/5% milk. The PVDF membranes were incubated with primary antibody at 4 °C overnight, washed in TBST, and detected by a horseradish peroxidase-conjugated secondary antibody at room temperature for 1 hour followed by treatment with the ECL detection system. Quantitative analysis was calculated using AlphaEaseFC software.</p><!><p>All data are presented as the mean ± standard deviation (SD). Differences between groups were analyzed by a one-way analysis of variance (ANOVA) and t-test using the SPSS software package. In all statistical analyses, p < 0.05 was regarded as statistically significant.</p>

PubMed Author Manuscript

EGCG sensitizes chemotherapeutic-induced cytotoxicity by targeting the ERK pathway in multiple cancer cell lines

Epigallocatechin-3-gallate (EGCG), a major polyphenol component of green tea, presents anticancer efficacy. However, its exact mechanism of action is not known. In this study, we evaluated the effect of EGCG alone or in combination with current chemotherapeutics [gemcitabine, 5-flourouracil (5-FU), and doxorubicin] on pancreatic, colon, and lung cancer cell growth, as well as the mechanisms involved in the combined action. EGCG reduced pancreatic, colon, and lung cancer cell growth in a concentration and time-dependent manner. EGCG strongly induced apoptosis and blocked cell cycle progression. Moreover, EGCG enhanced the growth inhibitory effect of 5-FU and doxorubicin. Of note, EGCG enhanced 5-FU\xe2\x80\x99s and doxorubicin\xe2\x80\x99s effect on apoptosis, but not on cell cycle. Mechanistically, EGCG reduced ERK phosphorylation concentration-dependently, and sensitized gemcitabine, 5-FU, and doxorubicin to further suppress ERK phosphorylation in multiple cancer cell lines. In conclusion, EGCG presents a strong anticancer effect in pancreatic, colon, and lung cancer cells and is a robust combination partner for multiple chemotherapeutics as evidenced by reducing cancer cell growth, in part, by inhibiting the ERK pathway.

egcg_sensitizes_chemotherapeutic-induced_cytotoxicity_by_targeting_the_erk_pathway_in_multiple_cance

Introduction<!>Chemicals and Reagents<!>Cell Culture<!>Cell Viability<!>Clonogenic Assay<!>Cell Apoptosis<!>Cell Cycle Analysis<!>Western Blot<!>Immunohistochemistry<!>Statistical Analysis<!>EGCG reduces cancer cell growth in multiple cancer cell lines<!>EGCG reduces cancer cell growth through a strong cytokinetic effect<!>EGCG enhances the cytotoxicity of chemotherapeutics in colon and lung cancer cells<!>EGCG enhances drug sensitivity by the inhibition of Raf/MEK/ERK pathway<!>Discussion

<p>Pancreatic, colon, and lung malignancies have the highest cancer morbidity and mortality for both genders, in the United States [1]. Besides surgery and radiation, the use of chemotherapy, either alone or in combination, is one of the most common ways to treat cancer. Unfortunately, conventional drug therapies have obvious limitations due to chemoresistance, as well as undesirable systemic side effects, which are often severe. For example, gastrointestinal tumors treated with the chemotherapy 5-fluorouracil (5-FU), can easily acquire resistance. Furthermore, 5-FU is associated with health risks, ranging from nausea and diarrhea to neurological disorders and myelosuppression [2, 3]. For these reasons, it is imperative to search for safer treatment strategies.</p><p>Over the last two decades, there has been a growing interest in identifying bioactives with anticancer effects. Due to chemotherapy's significant side effects, combining chemotherapeutic drugs with other agents, such as bioactives, is a promising approach to reduce toxicity while maintaining (or enhancing) the desired efficacy. Among several bioactives under investigation, many phytochemicals have been shown to possess anticancer effects, suppressing cancer growth at various steps. Epigallocatechin-3-gallate (EGCG), a major bioactive component in green tea, is one of these phytochemicals with anticancer activity [4]. Indeed, we have recently shown that EGCG synergized with gemcitabine to suppress pancreatic cancer cell growth [5, 6]. However, the ability of EGCG to enhance the effect of chemotherapeutic drugs in other cancer types is not completely understood.</p><p>Raf/MEK/ERK pathway is frequently activated in various malignancies, correlating to cell growth, cell cycle, and even apoptosis prevention [7]. Notably, activation of Raf/MEK/ERK pathway is also correlated to drug resistance [8]. Thus, inhibitors of Raf, MEK, ERK or some downstream effectors could be the target for therapeutic intervention. However, though the Raf/MEK/ERK pathway plays a vital role in controlling tumor growth and drug resistance, the regulation effect of EGCG remains unclear.</p><p>In this study, we evaluated the efficacy and mechanisms of EGCG in combination with chemotherapeutics (gemcitabine, 5-FU, and doxorubicin) active against pancreatic, colon, and lung cancers to elucidate whether EGCG is a potential adjuvant agent for cancer treatment. We observed that EGCG enhanced gemcitabine, 5-FU, and doxorubicin cell growth inhibition and induced apoptotic cell death in pancreatic, colon, and lung cancer cells, and this effect was associated, in part, with the suppression of the Raf/MEK/ERK pathway.</p><!><p>EGCG (≥98%) was purchased from Tocris (Minneapolis, MN) and a stock solution (100 mM) was prepared in sterile DMSO. Doxorubicin, 3-(4,5-dimethylthiazol-2-yl)-2,5-diphenyltetrazolium bromide (MTT) (≥97.5%), RIPA lysis buffer, Halt Protease Inhibitor Cocktail, and Phosphatase Inhibitor Cocktail were purchased from MilliporeSigma (St. Louis, MO). SuperSignal™ West Dura Extended Duration Substrate were purchased from ThermoFisher Scientific (Waltham, MA). Gemcitabine was purchased from BIOTANG (Waltham, MA, USA). 5-FU (≥99%) was purchased from Alfa Aesar (Haverhill, MA, USA). Bradford protein assay reagent, 30% (w/v) Acrylamide/Bis Solution, 4×Laemmli sample buffer, Immun-Blot Polyvinylidene difluoride (PVDF) Membranes and were purchased from Bio-Rad (Hercules, CA).</p><!><p>Human pancreatic cancer cell lines (Panc-1, MIA PaCa-2, and BxPC-3), human colon cancer cell lines (SW480, HCT15, and HT29), and human lung cancer cell lines (HT1975, H358, and A549) were purchased from the American Type Culture Collection (Manassas, VA). All cell lines were grown as monolayers in the specific medium suggested by the vendor. Although these cells lines were not authenticated in our lab, they were characterized by cell morphology and growth rate, and cultured in our laboratory less than six months after being received.</p><!><p>After treating cells with EGCG alone or together with specific chemotherapeutic drugs for 24, 48 and 72 h, the reduction of MTT dye was determined according to the manufacture's protocol (MilliporeSigma, St. Louis, MO).</p><!><p>This was performed as previously described [9]. Briefly, HCT15 colon cancer cells were plated in 6-well plates (1,000 cells per well), and treated with 5-FU alone or in combination with EGCG for 24 h. Following treatment, cells were then incubated with fresh media for 20 days. Media was replaced once weekly during the incubation. On the last day, colonies were fixed with methanol and stained with 0.1% (w/v) crystal violet in phosphate buffered saline (PBS) (pH 7.4). Cells were then rinsed with distilled water, air-dried, and colonies were counted and analyzed using ImageJ software (V1.46, NIH, Bethesda, MD, USA).</p><!><p>Cells were seeded in 100 mm plates at a density of 1.5 million cells per plate. The following day, cells were treated with EGCG, chemotherapy drugs, or a combination. After 48 h treatment, cells were trypsinized and stained with Annexin V-fluorescein isothiocyanate (FITC) (100× dilution) and propidium iodide (PI) (0.5 μg/mL) for 15 min. Annexin V-FITC and PI fluorescence intensities were analyzed by FACScan (Becton Dickinson, San Jose, CA, USA). Annexin V (+)/PI (−) cells are apoptotic cells, Annexin V (+)/ PI (+) cells have undergone secondary necrosis, and Annexin V (−)/ PI (+) cells are necrotic cells. Results were analyzed by using FlowJo software.</p><!><p>Cells were seeded in 6-well plates and treated the following day with EGCG, chemotherapy drugs, or a combination for 24 h. After each treatment, cells were trypsinized and fixed in 70% ethanol overnight at −20°C, stained with PI (50 μg/ml) and RNase A (10 mg/ml) for 15 min and subjected to flow cytometric analysis by FACScan (Becton Dickinson; San Jose, CA).</p><!><p>Following treatment with EGCG, chemotherapy drugs, or a combination, cells were lysed, and total cell fractions were obtained as previously described [10]. Aliquots of total fractions containing 10–30 μg protein were separated by using 10–12% (w/v) polyacrylamide gel electrophoresis and electroblotted to PVDF membranes. After blocking with 5% (w/v) non-fat milk for 1 h, membranes were probed overnight with the following primary antibodies (1:1000 dilution) from Cell Signaling Technology (Danvers, MA): Caspase-3 (Cat #14220), Caspase-7 (Cat #12827), Caspase-9 (Cat #9508), PARP (Cat #9542), phospho-Chk1 (Ser345) (Cat #2348), phospho-p53 (Ser15) (Cat #9286), p53 (Cat #2527), p21 Waf1/Cip1 (Cat #2947), cdc2 (Cat #28439), Cyclin B1 (Cat #12231), Bcl-xL (Cat #2802), Bad (Cat #9239), XIAP (Cat # 14334), survivin (Cat # 2808), p-ERK1/2 (Cat #4370), and ERK1/2 (Cat #9102). β-Actin (Cat #8457) was used at the same time as a loading control. After incubation for 60 min at room temperature in the presence of the secondary antibody (HRP-conjugated; 1:5,000 dilution), the conjugates were developed and visualized using a Molecular Imager FX™ System (BioRad; Hercules, CA) and analyzed using ImageJ software(V1.46, NIH, Bethesda, MD, USA).</p><!><p>Immunohistochemistry was performed using tumor samples from a previous efficacy study that evaluated the effect of EGCG and gemcitabine on murine pancreatic cancer xenografts [5]. Briefly, immunohistochemical staining for p-ERK1/2 (Cat #4370; Cell Signaling Technology, Danvers, MA, USA) was performed as previously described [33]. Briefly, paraffin-embedded sections (5 μm thick) were deparaffinized and rehydrated, followed by antigen retrieval performed by microwave-heating in 0.01 M citrate buffer (pH 6.0). H2O2 3% was used to block endogenous peroxidase activity for 10 min at room temperature. Slides were blocked for 60 min with serum, and incubated with primary antibody overnight at 4 °C. The following morning, slides were washed thrice with PBS, and then incubated with the biotinylated secondary antibody and the streptavidin-biotin complex (Invitrogen, Carlsbad, CA, USA) for 1 h each at room temperature. After washing with PBS three times, slides were stained with 3,3′-Diaminobenzidine tetrahydrochloride hydrate (DAB) solution, and then counterstained with hematoxylin. Images were captured at 100× magnification. At least five fields per sample were scored and analyzed using Image J software (V1.46, NIH, Bethesda, MD, USA).</p><!><p>The data, obtained from at least three independent experiments, were expressed as mean ± standard deviation (SD). Statistical evaluation was performed using one-factor analysis of variance (ANOVA) followed by the Duncan test for multiple comparisons. T-tests were used to analyze the difference between two groups. A P value<0.05 was regarded as statistically significant.</p><!><p>To test the anticancer effect of EGCG on cancer cell growth, we included nine human cancer cell lines from pancreatic (Panc-1, MIA PaCa-2, and BxPC-3), colon (HCT15, SW480, and HT29) and lung cancer (A549, H358, and HT1975) and treated them with increasing concentrations of EGCG (20–100 μM) for 24, 48, and 72 h. In all nine cell lines, EGCG reduced cancer cell growth in a time- and concentration-dependent manner. However, different cell lines displayed varying sensitivity to EGCG, as BxPC-3, HT1975, and HCT15 were relatively more sensitive to EGCG, while Panc-1, H358, and HT29 showed more resistance (Fig 1). The Inhibitory Concentration at 48 h (48 h-IC50) for EGCG in each cell line is summarized in Figure 1D. Given the high prevalence of Kras mutations in pancreatic, colon, and lung cancer, we chose Panc-1, MIA PaCa-2, HCT15 and A549 cell lines, which are Kras mutant cell lines, for the subsequent studies.</p><!><p>EGCG inhibited tumor growth through a potent cytokinetic effect. Treatment of Panc-1 and MIA PaCa-2 cells with EGCG for 48 h led to a concentration-dependent induction of apoptosis (Fig. 2A). EGCG at 1xIC50 for 48 h induced apoptosis by 3.5 and 2.1-fold over control in Panc-1 and MIA PaCa-2 cells, respectively (p<0.01). Notably, EGCG predominantly induced apoptotic cell death, with no significant induction of cell necrosis [Annexin V(−) but PI (+)].</p><p>These findings were validated by determining the activation and levels of apoptotic-related Caspases by microscopy and Western blot (Fig. 2B–C). In Panc-1 and MIA PaCa-2 cells, EGCG treatment induced the activation of Caspase 9, 7, and 3 in a concentration-dependent manner. For example, EGCG at 1xIC50 activated Caspase 3 levels by 2.9 and 3.0-fold in Panc-1 and MIA PaCa-2 cells, respectively, compared to the control group (p<0.01 for both). As a consequence of Caspase 3 activation, levels of cleaved poly(ADP-ribose) polymerase (PARP) increased in all EGCG treatments (Fig. 3C).</p><p>Next, we evaluated whether EGCG can also induce apoptosis in colon and lung cancer cells. For this purpose, we treated HCT15 colon cancer and A549 lung cancer cell lines with increasing concentration of EGCG for 48 h, and determined apoptotic-related caspases by Western blot. EGCG treatment induced the activation of Caspase 9, and PARP in a concentration-dependent manner in both HCT15 and A549 cells (Fig. 2D).</p><p>To explore the apoptosis mechanism induced by EGCG, we determined, in Panc-1 and MIA PaCa-2 cells, the expression levels of multiple proteins that regulate apoptosis, including proteins in the inhibitor of apoptosis protein (XIAP) and Bcl-2 family. As shown in Figure 3, while EGCG reduced Bcl-xl, XIAP, and survivin levels, it increased the levels of the proapoptotic protein Bad concentration-dependently.</p><p>Because ERK1/2 has been shown to modulate cell survival through the regulation of Bcl-2 protein family [11], we next evaluated the effect of EGCG on ERK phosphorylation. In Panc-1 and MIA PaCa-2 cells, EGCG treatment for 24 h reduced ERK1/2 phosphorylation in a concentration-dependent manner (Fig. 3B).</p><p>Next, we evaluated whether EGCG can also modulate the ERK pathway in colon and lung cancer cells. For this purpose, we treated HCT15 colon cancer and A549 lung cancer cell lines with increasing concentration of EGCG for 24 h, and determined ERK phosphorylation by Western blot. Consistent with our findings in pancreatic cancer cells, EGCG treatment strongly reduced ERK phosphorylation in both HCT15 and A549 cells (Fig. 3C).</p><p>To examine whether EGCG can affect cell cycle progression, we performed flow cytometry to test cell cycle distribution and determined the levels of cell cycle regulators by Western blot. EGCG induced an S/G2 arrest in Panc-1 cells. However, under the same experimental conditions, a lesser effect was observed in MIA PaCa-2 cells (Fig. 4A). Given the effect of EGCG on the S/G2 transition, we examined the expression levels of G2 phase checkpoint proteins. In Panc-1 and MIA PaCa-2 cells, EGCG treatment increased the expression of p-Chk1, p-p53, and p21 Waf1/Cip1, whereas it reduced the levels of cdc2 and Cyclin B1 (Fig. 4B).</p><!><p>The use of drugs in combination to treat cancer patients is a common practice. We have recently documented that EGCG enhances the chemotherapeutic efficacy of gemcitabine in pancreatic cancer cells and xenografts [5, 6]. Here, we evaluated whether EGCG can enhance the efficacy of chemotherapeutic drugs in colon and lung cancer cells. For this purpose, we treated cells with EGCG together with 5-FU or doxorubicin, two chemotherapeutics commonly used clinically and experimentally in colon and lung cancer. As shown in Figure 5A, EGCG increased the cytotoxicity of 5-FU in HCT15 cells. Compared with the control group, 20 μM 5-FU decreased cell growth to 62.7%, while the cell growth was further reduced to 29.4% after treatment together with EGCG at 1xIC50 (p<0.01). In A549, after co-treating cells with EGCG and doxorubicin, cell growth decreased to about 40%, lower than the doxorubicin alone treated groups, while kept a similar level as EGCG alone group (Fig. 5A).</p><p>In agreement with the growth inhibitory results, EGCG and 5-FU together also effectively inhibited the colony formation in HCT15 cells (Fig 5B). For example, 5-FU alone reduced the formation rate by 66.9%, and the inhibitory effect was enhanced to 90.1% when combined with EGCG (p<0.01).</p><p>EGCG and chemotherapy drugs together also induced more apoptosis compared with each treatment alone. In HCT15 and A549 cells, compared to control, treatment with 5-FU or doxorubicin for 48 h resulted in a 1.7 and 2.8 fold increase in apoptosis, respectively. The effect was further enhanced to 2.1 and 3.4 fold after treatment together with EGCG (p<0.01, Fig. 5C). In contrast, EGCG treatment did not enhance the cell cycle arrest inhibitory effect of 5-FU or doxorubicin on cell cycle progression (Fig. 5D).</p><!><p>We next investigated the potential mechanisms by which EGCG plus chemotherapeutics reduced cell growth and induced cell death by apoptosis. Because the ERK pathway plays a critical role in controlling tumor growth and drug resistance [12], we evaluated the effect of EGCG in combination with gemcitabine, 5-FU, and doxorubicin on the ERK pathway.</p><p>We first explored the effect of EGCG in combination with gemcitabine on ERK activation in pancreatic cancer cells. As shown in Figure 6A, EGCG 1xIC50 reduced the levels of ERK phosphorylation in Panc-1 and MIA PaCa-2 cells by 53% and 37.5% (p<0.05 for both), respectively, and this effect was enhanced in both cell lines, when combined with gemcitabine (70% and 49% in Panc-1 and MIA PaCa-2 cells, respectively). Of note, gemcitabine alone did not affect ERK phosphorylation in Panc-1 and MIA PaCa-2 cells (Fig. 6A). Consistent with the in vitro results, EGCG plus gemcitabine had an additive effect reducing the levels of p-ERK (p < 0.05) in pancreatic tumor xenografts [5] by 91%, compared to control, and the effect was stronger than either treatment alone (Fig. 6B).</p><p>Because ERK is known to phosphorylate STAT3 at the serine 727 residue [13], we also tested the effect of EGCG alone, gemcitabine alone, and EGCG and gemcitabine in combination on STAT3 phosphorylation. EGCG alone reduced STAT3 phosphorylation in Panc-1 and MIA PaCa-2 cell lines. While gemcitabine alone only reduced p-STAT3 levels in Panc-1 cells lines, the combination of EGCG plus gemcitabine had an additive effect and reduced STAT3 phosphorylation. In MIA PaCa-2 cells, p-STAT3 expression in the EGCG plus gemcitabine group was similar to that of EGCG alone (Fig. 6A).</p><p>We then evaluated whether EGCG enhanced the effect of 5-FU and doxorubicin on ERK phosphorylation in colon and lung cancer cells. In HCT15 cells, EGCG reduced ERK phosphorylation by 80% and no additional effect was observed when combined with 5-FU (Fig. 7A). In contrast, EGCG treatment reduced ERK phosphorylation in A549 by 27% and this was significantly enhanced when combined with doxorubicin (75% reduction vs. control; Fig. 7B).</p><p>Finally, in both HCT15 and A549 cells, the expression of Bcl-xl and XIAP decreased in the EGCG plus 5-FU as well as in the EGCG plus doxorubicin groups compared to 5-FU alone or doxorubicin alone (Figure 7A–B).</p><!><p>Patients receiving chemotherapy usually experience side effects, many of which are often severe. Therefore, there is an active search for safer and more effective therapeutic approaches. In this study, we show that the polyphenol EGCG, is a successful combination partner of various chemotherapy drugs in pancreatic, colon, and lung cancer cells, and that its anticancer effect is due, in part, through the modulation of the ERK pathway.</p><p>EGCG is a major bioactive component in green tea, with strong anticancer activity in multiple types of cancers [6, 14–19]. Indeed, EGCG strongly reduced the growth of pancreatic, colon, and lung cancer cell in a time- and concentration-dependent manner. The anticancer effect of EGCG results from its strong cytokinetic effect: inhibition of proliferation, induction of apoptosis, and block at the S/G2 cell cycle transition [18]. The apoptotic effect of EGCG seems to be the dominant one. For example, EGCG at 1xIC50 induced apoptosis in pancreatic cancer cells by up to 3.5-fold compared to controls, with arrest of the cell cycle showing only a moderate effect. The apoptotic cascade of the pancreatic cancer cells was manifested by the activation of execution caspases [20], and the modulation of Bcl-2 and the inhibitor of apoptosis protein families by EGCG. The apoptotic effect was not restrained to only pancreatic cancer cells, since EGCG also strongly induced apoptosis in colon and lung cancer cells, showing that this effect is observed in multiple cancer types. Our findings are consistent with others, showing that EGCG strongly induces apoptosis in various types of tumors [21–24].</p><p>Conventional chemotherapy is commonly associated with both acute and chronic toxicity [25]. Based on the World Health Organization classification, signs of toxicity can be classified in grades 1–4 (ranging from mild (grade 1) to life-threatening (grade 4). Common side effects of gemcitabine, 5-FU, and doxorubicin include anorexia, nausea, vomiting, and fatigue. Other, less common but often severe, unwanted effects of these drugs include hair loss and low white blood cell count. For this reason, combining chemotherapy with safer agents, such as bioactives, is a viable option to potentially reduce side effects while maintaining or enhancing anticancer efficacy.</p><p>Over the past decades, there has been increasing interest in exploring the use of phytochemicals that be used as combination partners with chemotherapeutics [26]. For example, curcumin has been shown to enhance the efficacy of multiple chemotherapeutic drugs [27–29]. For example, it potentiates the effect of 5-FU in colon cancer cells by inhibition of NF-κB and Src protein kinase [29], and enhances the anticancer effect of gemcitabine in preclinical models of pancreatic cancer [28].</p><p>Besides curcumin, resveratrol, a polyphenol present in grapes and red wine, has strong antitumor effects [30–35]. Resveratrol has also been shown to be an effective combination partner with chemotherapy drugs [36]. For example, resveratrol has also been shown to sensitize pancreatic cancer and colon cancer to gemcitabine and 5-FU [37, 38]. Moreover, resveratrol has been shown to protect against the myotoxicity of doxorubicin in aged mice [39].</p><p>Specifically for EGCG, we have recently shown that EGCG is a strong combination partner with gemcitabine in pancreatic cancer cells and xenografts [5]. In addition, EGCG is a strong combination partner for 5-FU and doxorubicin in colon and lung cancer cells, respectively. Consistent with our findings, EGCG is also a great partner for many other drugs, including cisplatin [40], paclitaxel [41] and metformin [42], highlighting its translational potential. Of note, the differential sensitivity that the multiple cancer cell lines had to EGCG and the combination treatment could be due to different genetic mutations present in these cancer cell lines, making them, either, more resistant or more sensitive to treatment. Therefore, implementing high-throughput drug screening and single-cell profiling techniques [43, 44] that can rapidly find effective drug combination for cancers with specific mutations, will likely be instrumental in improving cancer care and facilitating personalized treatment.</p><p>Mechanistically, ERK signaling pathway appears to an important pathway modulated by EGCG alone or in combination. The RAS-regulated RAF-MEK-ERK signaling pathway is frequently activated in various malignancies, correlating to cell growth, cell cycle, and even apoptosis prevention [7]. Moreover, activation of Raf/MEK/ERK pathway is also correlated to drug resistance [8]. The importance of this signaling pathway has driven the development of a variety of pharmaceutical agents to inhibit RAF/MEK/ERK axis in cancer and some RAF and MEK inhibitors are already approved and used in the clinic [45].</p><p>However, there is now much interest in targeting ERK directly for multiple reasons. A critical one is the development of acquired resistance to RAF or MEK inhibitors (i.e. KRAS or BRAF amplification, MEK mutation, etc.), which involves relief of negative feedback and pathway re-activation with all signaling going through ERK. This validates the search for ERK inhibitors with RAF or MEK inhibitors as an up-front combination. EGCG strongly reduces ERK phosphorylation and its downstream STAT3 activation. Furthermore, its effect on ERK phosphorylation is enhanced when combined with chemotherapy drugs, suggesting a key pathway is affected.</p><p>Although EGCG has shown promise in preclinical models of cancer, its use in the clinic has been limited, due, in part, to its poor bioavailability and stability. A few studies have shown a benefit of EGCG clinically in cancer therapy and prevention [46–48], as well as in ameliorating side effect from drugs and radiation [49, 50]. These above mentioned limitations have led to the exploration of multiple approaches, including the formulation of EGCG in nanoparticles, delivering it as a pro-drug, or using it in combination [44, 51–54]. All of these strategies are aimed at improving EGCG's bioavailability and stability, with the ultimate goal of improving the clinical use of EGCG.</p><p>In summary, our study provides new insights into the cellular mechanisms responsible for the antitumor effect of EGCG when combined with chemotherapeutics in multiple cancer types. EGCG has a beneficial effect when combined with chemotherapeutics, in part, through the inhibition of the ERK pathway. Further studies are warranted to precisely assess the in vivo effects of EGCG in combination with chemotherapeutic drugs.</p>

PubMed Author Manuscript

Reaction of O2 with [(\xe2\x88\x92)-Sparteine]Pd(H)Cl: Evidence for an Intramolecular [H\xe2\x80\x93L]+ \xe2\x80\x9cReductive Elimination\xe2\x80\x9d Pathway

(Sp)PdCl2 [Sp = (\xe2\x88\x92)-sparteine] catalyzes a number of different aerobic oxidation reactions, and reaction of O2 with a PdII\xe2\x80\x93hydride intermediate, (Sp)Pd(H)Cl (1), is a key step in the proposed catalytic mechanism. Previous computational studies suggest that O2 inserts into the PdII\xe2\x80\x93H bond, initiated by abstraction of the hydrogen atom by O2. Experimental and computational results obtained in the present study challenge this conclusion. Oxygenation of in-situ-generated (Sp)Pd(H)Cl exhibits a zero-order dependence on [O2]. This result is inconsistent with a bimolecular H-atom-abstraction pathway, and DFT computational studies identify a novel \xe2\x80\x9creductive elimination\xe2\x80\x9d mechanism, in which the chelating nitrogen ligand undergoes intramolecular deprotonation of the PdII\xe2\x80\x93hydride. The relevance of this mechanism to other PdII oxidation catalysts with chelating nitrogen ligands is evaluated.

reaction_of_o2_with_[(\xe2\x88\x92)-sparteine]pd(h)cl:_evidence_for_an_intramolecular_[h\xe2\x80\x93

<p>Pd-catalyzed aerobic oxidation reactions have expanded significantly over the past decade, and these advances are closely linked to the identification of oxidatively stable ancillary ligands that promote these reactions.1 These ligands play a key role in stabilizing the reduced Pd catalyst formed upon substrate oxidation and promoting its reaction with molecular oxygen. The mechanism of catalyst oxidation by O2 has been the focus of considerable recent study by us2 and others.3 For catalytic reactions that proceed via PdII–hydride intermediates, two mechanistic pathways appear to be viable (Scheme 1A): (1) direct reaction of O2 with the Pd–H bond via a hydrogen-atom-abstraction pathway (HAA, Path A) and a pathway initiated by H–X reductive elimination from PdII, followed by a reaction of O2 with Pd0 (HXRE, Path B). Insights from recent experimental and computational studies suggest that all catalyst systems with labile monodentate ligands (e.g., pyridine, Et3N, DMSO) strongly favor the HXRE pathway.4 Many catalysts feature chelating nitrogen ligands, however (Scheme 1B), and previous computational studies of the reaction of O2 with (Sp)Pd(H)Cl (1) [Sp = (−)-sparteine] support an HAA mechanism.3a,d,5 Here, we present experimental data that are incompatible with an HAA mechanism, together with DFT computational studies that reveal a novel "reductive elimination" pathway for aerobic oxidation of 1. These and additional results presented herein suggest that no existing Pd catalysts undergo aerobic oxidation via the HAA pathway.</p><p>The reaction (Sp)Pd(H)Cl with O2 has been the focus of two previous computational studies by Goddard and coworkers.3a,d The first study identified the unprecedented HAA pathway as a possible mechanism for O2 "insertion" into a Pd–H bond (Scheme 2A).3a The second study showed that an alternative mechanism, initiated by deprotonation of the Pd–H, has a significantly higher activation energy relative to the HAA pathway (Scheme 2B; ΔΔG‡ = 7.5 kcal/mol).3d,6 These results have not been confirmed experimentally. In principle, these two mechanisms could be distinguished on the basis of their kinetic dependence on [O2]: the HAA pathway predicts a first-order dependence, whereas the deprotonation mechanism predicts a zero-order dependence. This possibility prompted us to pursue the synthesis of (Sp)Pd(H)Cl (1).</p><p>We examined a number of methods for the preparation of 1, including the reaction of (Sp)PdCl2 with various Si-, B- and Sn-hydride reagents,7 β-hydride elimination from in-situ-generated Pd–formate and alkoxide derivatives,8 and addition of HCl to Pd0 sources in the presence of (−)-sparteine.9 With many of these methods, a common major Pd–H species was evident by 1H NMR spectroscopy at reduced temperatures (≤ −35 °C); however, the complex was thermally unstable, and efforts to isolate the complex were unsuccessful. The instability of this Pd–H species is not entirely surprising. No mononuclear, nitrogen-chelated Pd–H complexes have been reported in the literature, and 1 has never been observed under catalytic conditions.</p><p>Among the synthetic routes evaluated, the most promising result was obtained from the reaction of excess (EtO)3SiH (10 equiv) with (Sp)PdCl2 in CD2Cl2 at −10 °C (Scheme 2). Over 25 min, the reaction turned from yellow to dark brown and a single Pd–hydride complex was formed in 40% yield by 1H NMR spectroscopy (δPd–H = −25.97 ppm).10 The Pd–H complex decomposes if it is maintained at the reaction temperature, and it appears to have at least two accessible decomposition pathways (Scheme 3): (1) formation of H2, evident by the emergence of a peak in the 1H NMR spectrum at 4.59 ppm,11 and (2) formation of the protonated sparteine ligand (Sp-H+) and Pd black, formally arising from NSp–H "reductive elimination" and aggregation of the Pd0 byproduct. The former decomposition pathway is especially evident when hydride reagents are used, and it may proceed via a PdII-dihydride intermediate (top pathway, Scheme 3). As discussed below, the latter pathway has important implications for the reaction of 1 with O2.</p><p>The difficulties in isolating 1 prompted us to investigate the O2 reactivity of the PdII–H complex in situ. When the reaction mixture, containing ~40% yield of PdII–H, was exposed to an atmosphere of O2, the solution changed color from dark brown to yellow, and a 1H NMR spectrum of the product solution revealed (Sp)PdCl2 as the predominant product (Figure S1). A plausible explanation for this observation is that the PdII–H reacts with O2 to afford a Pdhydroperoxide that undergoes rapid reaction with (EtO)3SiCl present in the reaction mixture to afford (Sp)PdCl2 and (EtO)3SiOOH (Figure 1A). It was possible to monitor kinetics of the O2 reactivity by 1H NMR spectroscopy at −30 °C (t1/2 ~ 20 min),12 and increasing the O2 pressure from 1–5 atm exhibited no effect on the reaction rate (Figure 1B).</p><p>Interpretation of this result is subject to a number of caveats associated with the complexity of the reaction mixture and lack of full characterization of 1 or the Pd–hydroperoxide intermediate. Conservatively, however, the data indicate that a sparteine-coordinated Pd–hydride species undergoes a reaction that is induced by the presence of O2, but is independent of the [O2]. While this conclusion lacks mechanistic certainty, the data are not consistent with an HAA mechanism, which should exhibit a first-order dependence on [O2].</p><p>Previous studies of the reactions between O2 and well-defined (albeit catalytically unreactive) Pd–hydride complexes demonstrate close agreement between experimental and computational results.2b,c,e,3b,c Consequently, the discrepancy between the experimental and computational results noted above suggest a mechanism different from those shown in Scheme 2 might be involved. In order to examine this possibility, we decided to reinvestigate the reaction of 1 with O2 using DFT computation methods.13</p><p>Analysis of the ground-state structure of (Sp)Pd(H)Cl 1 reveals that the two Pd–N bonds differ in length by 0.15 Å, with the longer one trans to the hydride ligand (Figure 2). This observation, together with the formation of Sp-H+ observed experimentally (cf. bottom pathway, Scheme 3), raised the possibility that sparteine could serve as an internal base upon dissociation of the amine ligand trans to the hydride ligand. This mechanism would be analogous to the deprotonation pathway noted in Scheme 2B, but it would avoid the entropic cost associated with a bimolecular reaction.3d</p><p>A transition state for dissociation of the NB amine of sparteine was identified computationally (2TS, Figure 3). The imaginary frequency associated with this step includes both Pd–NB bond lengthening and rotation of the sparteine ligand about the Pd–NA bond. The rotation enables the basic NB atom to approach the hydride ligand, and deprotonation of the hydride ligand by the NB amine occurs, yielding the zwitterionic-Pd0 product 3 (Figure 3). The calculated free-energy barrier for this concerted process is 18.6 kcal/mol (Figure 3A). The Pd–H bond distance in 3 is 1.72 Å, and the Pd--H–N interaction is best described as a hydrogen bond between a Lewis-basic Pd0 center and an acidic H–NSp fragment.14 Natural charge analysis suggests the deprotonation can be described formally as an N–H "reductive elimination" reaction in which the Pd center starts with a +0.36 charge and ends with a −0.15 charge. The H atom simultaneously shifts from having slight hydridic character (−0.054 charge) in 1 to protic character, (+0.32 charge) in 3. To summarize, the amine-dissociation and deprotonation steps occur as a concerted process. By analogy to the HXRE nomenclature employed when the reductive elimination step involves an anionic X-type ligand, we assign the name HLRE to the present reaction, reflecting the involvement of a neutral L-type ligand in the "reductive elimination" reaction.</p><p>Aerobic oxidation of the zwitterionic intermediate 3 proceeds via an oxygenation/protonolysis sequence in which all of the calculated intermediates and transition states are lower in energy than 2TS (Figures 3A and S3). These results indicate that the HLRE step is rate-limiting, and oxygenation of 1 by this mechanism should exhibit a zero-order dependence on [O2]. Reinvestigation of the HAA mechanism, using the same computation package and methods employed for the HLRE study, revealed that the HLRE mechanism is more favorable than the HAA pathway: ΔG‡HLRE − ΔG‡HAA = −3.5 kcal/mol).15 The free energy of activation derived from the experiments in Figure 1 (ΔG‡expt ~18 kcal/mol) and the calculated barrier for the HLRE pathway are in good agreement,16 and we propose that aerobic oxidation of (Sp)Pd(H)Cl proceeds via this novel HLRE mechanism.</p><p>In order to probe the generality of the HLRE pathway, we performed analogous calculations with three other commonly used bidentate nitrogen ligands: phenanthroline (phen), bipyridine (bpy), and pyridine-oxazoline (py-ox). A slight difference observed in HLRE mechanism with these three ligands relative to (−)-sparteine was the presence of a stepwise, rather than concerted, ligand-dissociation/deprotonation pathway. The highest barrier for the pathway was associated with ligand dissociation.17 A comparison between the rate-limiting transition-state energies for the HLRE and HAA mechanisms with each of these ligands reveals that the HLRE mechanism is always lower in energy (Figure 4), although the difference is relatively small with the phenanthroline and bpy ligands (ΔΔG‡ ~1–2 kcal/mol). Despite the relative small energy differences, however, these observations reveal the potential for these "ancillary' ligands to participate directly in reactions at the Pd center.</p><p>The mechanisms depicted in Figure 4 feature Pd–hydride complexes with chloride as the second anionic ligand; however, the (Sp)PdCl2 catalyst system is unusual in its use of chloride. The vast majority of nitrogen-ligated Pd catalysts for aerobic oxidation reactions feature carboxylates as the anionic ligands. We have previously highlighted the ability of acetate and other carboxylate ligands to undergo intramolecular deprotonation of Pd–hydrides, formally H–O2CR "reductive elimination" from the PdII center.2b–e Calculation of the rate-limiting transition states for this HXRE pathway for (N-N)Pd(H)(OAc) complexes (10N-N) reveals that this mechanism is strongly favored18 relative to the HAA mechanism (ΔΔG‡ = 8.8–11.9 kcal/mol) (Figure 5). This mechanism is also significantly favored relative to the HLRE mechanism with Pd–chloride complexes by (ΔΔG‡ = 6.4–10.1 kcal/mol). The carboxylate HXRE pathway benefits energetically from the ability of the carboxylate ligand to retain partial bonding to the Pd center as it participates in the deprotonation step. In contrast, the bidentate nitrogen ligands must fully dissociate one of the nitrogen atoms before it can interact with the hydride ligand.</p><p>These observations have clear implications for Pd-catalyzed aerobic oxidation reactions. PdII–hydride intermediates in Pd-catalyzed aerobic oxidation reactions are elusive, and insights into their reactivity are limited by the lack of direct investigations. Nevertheless, results from experimental and computational studies of well-defined model systems suggested that nearly all catalyst systems undergo oxidation by O2 via a Pd0 intermediate. The (Sp)PdCl2 catalyst system appeared to be the sole exception. The present work, however, provides evidence for a previously unrecognized mechanism in which the chelating nitrogen ligand participates in intramolecular deprotonation of the hydride ligand, resulting in the formation of a Pd0 species that reacts with O2. This mechanism exhibits many features in common with the previously characterized HX-reductive elimination pathway. In both pathways, one portion of the ligand remains coordinated to the Pd center while a remote site deprotonates the Pd–H. The intramolecular nature of this process avoids the unfavorable entropy associated with a bimolecular reaction. The hydrogen-atom-abstraction pathway remains an intriguing pathway for O2 insertion into a Pd–H bond, but we are not aware of any aerobic oxidation reactions in which it represents a viable pathway for catalyst oxidation.</p>

PubMed Author Manuscript

Effects of Isolated Tobacco Alkaloids and Tobacco Products on Deprivation-Induced Food Intake and Meal Patterns in Rats

The ability of smoking to reduce body weight serves as motivation for continued smoking. It is unclear to what extent non-nicotine constituents in cigarettes are contributing to the weight-reducing effect of smoking. The purpose of the current study was to examine the effects of nicotine and four minor tobacco alkaloids (nornicotine, cotinine, anatabine, and anabasine) on food intake, one of the key regulators of body weight. In addition, a smokeless tobacco extract (STE) and e-cigarette (EC) refill liquid were used to model the effects of actual tobacco product exposure on food intake. Male Holztman rats were trained to lever press for food pellets during daily 2h sessions in operant chambers. In Experiment 1, the effects of subcutaneous injections of saline, nicotine (0.25 \xe2\x80\x93 1.00 mg/kg), nornicotine (0.50 \xe2\x80\x93 6.00 mg/kg), cotinine (1.00 \xe2\x80\x93 100.00 mg/kg), anatabine (0.25 \xe2\x80\x93 3.00 mg/kg), and anabasine (0.50 \xe2\x80\x93 4.00 mg/kg) were assessed. In Experiment 2, rats from Experiment 1 were used to examine the effects of nicotine, STE, and EC liquid. All alkaloids, except cotinine, produced a dose-dependent reduction in overall food intake. The highest doses of all drugs significantly reduced latency and response rate to obtain the first pellet. At some doses, nicotine, anatabine, and nornicotine reduced food intake within the first 45 minutes without compensatory increases in intake later in the session. STE and EC liquid produced dose dependent decreases in food intake similar to nicotine alone. These data suggest that minor tobacco alkaloids have appetite suppressant effects and warrant further investigation into their effects on body weight, energy intake, and energy expenditure under free-feeding conditions. However, findings with STE and EC liquid suggest that nicotine is the primary constituent in these products to effect food intake, whereas levels of minor alkaloids in these products may be too low to influence food intake.

effects_of_isolated_tobacco_alkaloids_and_tobacco_products_on_deprivation-induced_food_intake_and_me

1. Introduction<!>2.1. Animals<!>2.2. Apparatus<!>2.3. Drugs<!>2.4.1. Training<!>2.4.2. Experiment 1: Effects of nicotine and minor alkaloids<!>2.4.3 Experiment 2: Effects of tobacco product exposure<!>2.5.1. Experiment 1<!>2.5.1. Experiment 2<!>3.1.1. Total food intake<!>3.1.2. Latency to first pellet<!>3.1.3. Changes in running rate to first pellet<!>3.1.4. Changes in within-session rates of food intake<!>3.2.1. Nicotine versus Kodiak extract<!>3.2.2. Nicotine versus EC liquid<!>4. Discussion<!>4.1. Effects of Isolated Alkaloids<!>4.2 Effects of Tobacco Product Extracts<!>4.3. Limitations<!>5. Conclusions

<p>Cigarette smoking is associated with reduced weight gain and quitting smoking results in increased weight gain (1–5). The relationship between smoking status and body weight serves as a primary motivation for continued smoking (6). Weight loss is often given as a rationale for starting to smoke, while weight gain is often cited as a deterrent for quitting (7, 8). The average weight gain among smokers is estimated between 4 – 5 kg, with a significant proportion of the population gaining well above this average (>15 kg in 13 – 33% of ex-smokers), making weight gain a major obstacle to smoking cessation (1, 2, 5).</p><p>Nicotine has been the main focus of previous research investigating smoking's effects on body weight (9–11). Nicotine has been shown to increase metabolic rate and physical activity while decreasing food intake in smokers to produce weight loss. On the other hand, withdrawal from nicotine leads to significant weight gain, with some studies showing up to 70% of which is due to increases in energy intake (9, 12). Animal models of nicotine's effects on body weight have shown similar changes in energy expenditure (13–16) or food intake (17–22) following nicotine administration and withdrawal from nicotine. However, nicotine replacement therapy (NRT) has produced variable effects on post-cessation weight gain, ranging from no effect (23–27) to relatively modest reductions (28–32). These findings suggest that other tobacco constituents may also affect body weight.</p><p>It is unclear to what extent other constituents in cigarettes are contributing to the reduced weigh gain in smokers. In light of their behavioral and neuropharmacological similarities to nicotine, minor tobacco alkaloids (i.e., nornicotine, cotinine, anabasine, anatabine, and myosmine) are prime candidates for contributing to these effects. Because minor tobacco alkaloids act at nicotinic acetylcholine receptors, which may be necessary for facilitating nicotine's effects on weight loss (33), these compounds could also produce effects on body weight. Some of these compounds may also be self-administered (34, 35) and can substitute for nicotine in drug discrimination tasks (35–37). Although cotinine, nornicotine and anatabine (as anatabloc®) have all been proposed in patent applications as treatments for weight-loss (38, 39), no empirical studies of their effects on body weight per se have been published. This research gap represents a significant limitation in our knowledge of the underlying factors involved in smoking's effects on body weight. To the extent that minor tobacco alkaloids can reduce food intake, they might have potential as pharmacotherapies for the treatment of post cessation weight gain, as well as obesity. Because obesity is a leading cause of preventable death in the United States (40), and FDA approved drugs for the treatment of obesity have had limited efficacy and undesirable side effects (41), the development of new drugs for weight loss is a continued research priority.</p><p>As the nature of tobacco use continues to change with the evolution and emergence of new tobacco products, it is important to examine how these products impact food intake and body weight. Relative to research on cigarette smokers, little research has been conducted examining the effects of smokeless tobacco products on body weight and food intake in human and animal models. In addition, e-cigarettes, which have become increasingly popular over the last several years, vary considerably between products with respect to their levels of nicotine and other constituents (e.g., minor tobacco alkaloids; (42)). Given that e-cigarettes are now regulated as tobacco products, it is important to assess the extent to which nicotine and minor alkaloids in these products might facilitate use by influencing body weight and food intake. Although the current levels of minor tobacco alkaloids in tobacco products is generally low compared to nicotine, their interaction with nicotine and thousands of other constituents could nonetheless contribute to the effects of tobacco product use on body weight and food intake. Recent human and animal studies have produced conflicting results regarding the effects of e-cigarettes on weight gain. While some studies have shown that e-cigarette vapor produces increases in physical activity and delayed weight gain (43, 44), others have shown that e-cigarette vapor has no effect on weight gain or food intake (45, 46).</p><p>The purpose of the current study was to examine the effects of four minor tobacco alkaloids (nornicotine, cotinine, anatabine, and anabasine) and nicotine alone on food intake. In this initial study, animals were allowed to lever-press for food pellets (2hr/day) in operant conditioning chambers to investigate the acute effects of a range of doses of nicotine and these alkaloids on deprivation-induced food intake. Following the assessment of isolated alkaloids, the effects of nicotine alone were compared to tobacco product formulations containing equivalent nicotine doses. For this purpose, an aqueous extract of Kodiak Wintergreen smokeless tobacco and an EC liquid were used. Because these products contain non-nicotine constituents that are known to enhance the behavioral effects of nicotine, we hypothesized that they would have a greater effect on food intake than nicotine alone.</p><!><p>Sixteen male Holtzman rats (Harlan, Indianapolis, IN) weighing 300–350 g at arrival were used in this study. This strain was chosen to extend our previous study examining changes in food intake during nicotine self-administration (17). Upon arrival, all rats were individually housed in a temperature- and humidity- controlled colony room under a reversed 12 h light/dark cycle (lights off at 09:00 h) for approximately one week. Rats were maintained under a restricted feeding regimen (18 g/day), with ad libitum access to water. After one week, rats began training during 2 h operant sessions (see section 2.4.). Rats were fed approximately 30 – 45 minutes after the session termination, in their home cages. Protocols were approved by the Institutional Animal Care and Use Committee of the Minneapolis Medical Research Foundation in accordance with the 2013 NIH guide for the Care and Use of Mammals in Neuroscience and Behavioral Research.</p><!><p>Each operant conditioning chamber (29 cm × 26 cm × 33 cm; Coulbourn Instruments, Allentown, PA) was made of aluminum and Plexiglas walls, a Plexiglas ceiling, and a stainless steel grid floor. Two response levers (ENV-110RM, Med Associates) were located on the front wall 7 cm above the chamber floor. Standard grain pellets (45 mg; total Kcal/g: 3.303; breakdown: 0.796 protein, 0.345 fat, and 2.162 carbohydrate) were dispensed via a feeder (ENV-203M-45, Med Associates) into a food receptacle on the front wall located between the two levers. Each response on the right lever produced a single food pellet. Responses on the left lever were recorded, but had no programmed consequences. Water was continuously available via a spout mounted on the back wall of the chamber. Each chamber was placed inside a sound-attenuating cubicle equipped with an exhaust fan that provided masking noise. Med-PC IV (Med Associated, St Albans, VT) software was used for operating the apparatus and recording data.</p><!><p>(−)-Nicotine bitartrate, (+/−)nornicotine, (+/−)cotinine, (+/−) and (+/−)anabasine, were obtained from Sigma Chemical Co. (St. Louis, MO). (+/−)Anatabine was obtained from Toronto Research Chemicals, Inc. (Ontario, Canada). All drugs were dissolved into sterile saline. The pH of all solutions was adjusted to 7.4 using dilute NaOH. All drugs were administered subcutaneously (s.c.) in a volume of 1ml/kg. All drug doses are expressed as the base. The doses selected were based on those used for these alkaloids in a prior study of intracranial self-stimulation (ICSS) behavior (47), in order to make direct comparisons between the ICSS and food intake assays.</p><p>Smokeless tobacco extract was selected for use because these products more accurately reflect non-nicotine constituent exposure in humans, while also allowing for the control of nicotine dose (48). Aqueous tobacco extract was prepared from Kodiak Wintergreen smokeless tobacco product (purchased in the Minneapolis area between January 2013 and January 2014) using general procedures described elsewhere (49). Briefly, Kodiak Wintergreen was mixed with saline at concentrations of 400 mg/ml for 18 h using a tube tipper. Saline extract was chosen because it produces a similar alkaloid extraction profile as artificial saliva and simplifies extract preparation while avoiding toxicity (49). The resulting solution was filtered through gauze, centrifuged, and the supernate was filtered. The nicotine concentration was determined by gas chromatography with nitrogen phosphorus detection, according to standard protocol in our laboratory (50), and extract was diluted to the nicotine concentrations required for the current studies.</p><p>Whole Tobacco Alkaloid (WTA) EC refill liquid (Dark Honey Tobacco flavor in 10 ml vials) was obtained from Aroma E-Juice (http://www.aromaejuice.com, Scottsdale, AZ). The label indicated the liquid contained 80% vegetable glycerine (VG) and 20% propylene glycol (PG), and had a nicotine concentration of 24 mg/ml. The nicotine concentration was determined in each 10 ml vial of EC liquid used, allowing dilution in saline to the nicotine concentrations required for the current studies. The pH of all solutions was adjusted to 7.4 using dilute NaOH, and administered s.c. in a volume of 1ml/kg.</p><!><p>Rats were placed in operant chambers at 09:00 h, at the onset of the dark cycle. The house light remained off during the operant sessions. Stimulus lights were located 2 cm above the levels; lever presses on the active lever resulted in the delivery of food pellets and the illumination of the cue light above this lever for 2 s. This was done in order to facilitate training and discrimination between the active and inactive levers. The sessions lasted for 2 h. The FR for the first pellet was gradually increased from FR 1 to FR10 over the first week, while the FR for all other pellets remained at FR1. This mixed FR 10:FR 1 schedule remained in effect for the remainder of the study. Use of a mixed FR schedule allowed for separate measurement of motivation to initiate (FR 10) versus maintain food intake (FR 1, (51, 52)). Differential effects on these measures can help ascertain whether drugs have specific effects on satiety (reduce maintenance without affecting initiation) or more general motivational effects (reduce both measures).</p><p>After one week of training under the mixed FR schedule, s.c. injections of saline were administered twice a week (Mondays and Thursdays) for at least two weeks, 15 minutes before the session, until behavior stabilized. Behavior was considered stable when the coefficient of variance for total food intake was less than 15% across 4 consecutive saline sessions. Mean intake during these saline sessions then served as the baseline from which each drug was compared. To account for potential changes in baseline food intake over time, animals were re-baselined before a new drug was administered. No limit was placed on total food intake during the operant sessions. If intake within the session was less than 18g (i.e., 400 pellets), animals were supplemented with food in their home cages 30 minutes after the end of their session. Therefore, each rat received at least 18g of food per day, with the possibility to earn more. Any pellets left uneaten in the food hopper or that dropped into the bedding were subtracted from the total pellets.</p><!><p>Once food intake was stable following saline injections, each animal received s.c. injections of nicotine or one of the four minor tobacco alkaloids, 15 minutes before the session, twice a week (Tuesdays and Fridays). Injections of saline continued to be administered on Mondays and Thursdays. Animals continued to run on the weekends, however no drugs were administered during this time. The doses of each drug were administered in ascending order for nicotine (0, 0.25, 0.50, 0.75, or 1.00 mg/kg), nornicotine (0, 0.50, 1.00, 3.00, or 6.00 mg/kg), cotinine (0, 1.00, 6.00, 10.00, or 100.00 mg/kg), anatabine (0, 0.25, 0.50, 1.00, or 3.00 mg/kg), and anabasine (0, 0.50, 1.00, 3.00, or 4.00 mg/kg). These doses were selected based on their effects in other behavioral models (35, 37, 47, 53). Myosmine was not investigated in the present experiments, because previous research has shown either no behavioral effects (54) or only aversive effects of this alkaloid (47). If food intake had not returned to baseline levels the day prior to a drug test session, the treatment drug was not administered and another saline session was run. This only occurred after the highest doses of each drug were tested, and the number of sessions required to recover baseline did not differ between drugs. Once all four doses of a drug were administered, animals were re-baselined. Following at least two weeks of continued saline injections twice a week and stability over four consecutive saline sessions, the next drug was administered. Stability criteria remained the same. The order of drug administration was counterbalanced between animals using a Latin-Square design. This procedure continued until all animals had received nicotine (n=15; one animal was removed from the study prior to receiving nicotine due to sores from injections) and three out of four minor tobacco alkaloids (n = 11–12 per alkaloid)</p><!><p>Fourteen of the sixteen rats from Experiment 1 were used in this experiment (another rat was removed from the study due to sores from injections). Following another re-baseline period of at least two weeks of saline injections, nicotine, Kodiak, and EC liquid were administered. Based on the results from Experiment 1, three doses of nicotine (0.25, 0.50, and 1.00 mg/kg) were tested. Each product was separately compared to nicotine alone using equivalent doses. The order of product testing (nicotine v. Kodiak extract or nicotine v. EC liquid) and the order of nicotine doses was counterbalanced using a Latin-Square design. Food intake and weight gain was re-baselined between products.</p><!><p>Mean total food intake (pellets/session), food intake during consecutive 10-min intervals (12 per session), latency (sec) to obtain the first pellet, and running rate to obtain the first pellet (10/(latency to first pellet – latency to first response, i.e. rate of responding from emission of the first response to delivery of the first pellet) were the primary dependent measures. The latter two measures served as indices of motivation to initiate feeding. For each drug, a one-way repeated measures ANOVA was used to analyze changes in total food intake, latency to first pellet, and running rate to first pellet, followed by Dunnet post-hoc tests. Within-session analysis of food intake was analyzed using two-way ANOVAs with dose and interval as factors, followed by Dunnet post-hoc tests to evaluate changes in intake in each of the 12 ten-minute intervals across doses. This interval size provided a good description of the satiation process, showing a typical pattern involving an initial high rate of food intake in the first hour, followed by a lower rate of intake for the remainder of the session. These two rates of intake were demarcated by a distinct pause in feeding. This pattern was further analyzed by dividing total pellets into two bouts of feeding, defined as the number of pellets obtained prior to and following a pause in food intake of at least 10-minutes. This pause duration has been previously used to define a meal for rats (18, 19). In the event there were no pellets obtained within the first 10 minutes of the session, the first bout of food intake was recorded as zero, and all subsequent food intake categorized as falling under the second bout of food intake. This criterion was selected in order to examine changes in food intake within the first 30 minutes following drug injection. Food intake during the first and second bouts was analyzed for each drug using repeated measures two-way ANOVA followed by Sidak post-hoc tests to compare bouts at each dose and Dunnet post-hoc tests to compare each dose to saline within each bout. All statistical analyses were preformed using GraphPad Prism 7 (GraphPad Software, Inc.)</p><!><p>The same measures and analyses were used as in Experiment 1, except the bout analysis. Also, only the two highest doses administered for nicotine alone, Kodiak extract, and EC liquid were analyzed for the bin data analyses because their time course of effects on food intake was the most consistent, whereas all three doses were evaluated for producing changes in the other three parameters. All statistical analyses were preformed using GraphPad Prism 7 (GraphPad Software, Inc.)</p><!><p>Figure 1 shows the mean number of pellets consumed for each dose of nicotine (panel A), nornicotine (panel B), anatabine (panel C) and anabasine (panel D) during the entire 2 h session. There was a significant effect of nicotine on mean total food intake (F(3.228, 45.2) = 17.32; p < 0.001), with a significant reduction in food intake for all four doses of nicotine compared with saline. There was a significant effect of nornicotine (F(2.77, 27.7) = 8.57; p < 0.001), with a decrease in food intake following injections of 3.00 mg/kg and 6.00 mg/kg nornicotine compared to saline. There was a significant effect of anatabine (F(2.33, 23.3) = 28.72; p < 0.001), with a significant decrease in food intake at the lowest (0.25 mg/kg) and highest dose (3.00 mg/kg) compared to saline. The reduction in food intake at the other doses approached significance (p < 0.1). There was a significant effect of anabasine (F(2.6, 26) = 27.62; p < 0.001), with a decrease in total food intake at the two highest doses (3.00 and 4.00 mg/kg) compared with saline. There was no effect of cotinine on total food intake (see Supplementary Figure 1, panel A). It should be noted that animals consumed less than 18g of food following injections of nicotine, nornicotine, cotinine, anabasine, and anatabine 89%, 65%, 19%, 75% and 54% of the time, respectively, resulting in supplemental food intake (1g to 12g) 15 minutes following the session end. All supplementary food was consumed 100% of the time.</p><!><p>Figure 1 shows the mean latency to first pellet for each dose of nicotine (panel A), nornicotine (panel B), anatabine (panel C) and anabasine (panel D), during 2 h operant sessions. There was a significant effect of nicotine on latency to first pellet (F(1.667, 23.34) = 25.45; p < 0.001), with significant increases in latency at 0.50, 0.75, and 1.00 mg/kg nicotine compared with saline. There was a significant effect of nornicotine (F(1.473, 14.73) = 9.54; p < 0.01), with an increase in latency at 6.00 mg/kg nornicotine compared to saline. There was a significant effect of anatabine (F(1.01, 10.1) = 13.58; p < 0.001), with a significant decrease in response latency for the highest dose (3.00 mg/kg) compared to saline. There was a significant effect of anabasine (F(1.804, 18.04) = 13.68; p < 0.001), with an increase in response latency for the two highest doses (3.00 and 4.00 mg/kg) compared with saline. There was no effect of cotinine on response latency (see Supplementary Figure 1, panel A).</p><!><p>Figure 2 shows the mean running rates for the first pellet following injections of each drug. There was a significant effect of nicotine on running rate (F(4,48) = 7.6; p < 0.001), with significant differences from saline at all doses tested. There was also a significant effect of nornicotine (F(4,36) = 5.9; p < 0.001), with decreases in running rate at 6.00 mg/kg nornicotine compared to saline. There was a significant effect of anatabine (F(4,32) = 2.8; p < 0.05), with decreases in running rate at 3.00 mg/kg anatabine. There was a significant effect of anabasine (F(4,40) = 6.0; p < 0.001), with decreases in running rate following injections of 3.00 and 4.00 mg/kg anabasine. There was no main effect of cotinine dose on running rate (see Supplementary Figure 1, panel B)</p><!><p>Figure 3 shows mean total food intake during 10 minute bins following injections of nicotine (panel A), nornicotine (panel B), anatabine (panel C), and anabasine (panel D), during two hour operant sessions. There was a main effect of bin number (F(11, 154) = 17.81; p < 0.001) and dose (F(4,56) = 17.22; p < 0.001) on mean total food intake following injections of nicotine, and a significant bin × dose interaction (F(44,616) = 12.00; p < 0.001). A significant decrease in mean total food intake was observed at all four doses of nicotine compared to saline during the first 20 min of the session, and doses 0.50, 0.75, and 1.00 mg/kg suppressed food intake up to 30 min. The 0.25 mg/kg dose also decreased intake at the end of the session (bin 12). There was a significant main effect of bin number (F(11,110) = 47.14; p < 0.001) and dose (F(4,40) = 8.92; p < 0.001) on mean total food intake following injections of nornicotine, and a significant bin × dose interaction (F(44,440) = 8.27; p < 0.001). Significant decrease in mean total food intake following injections of 3.00 and 6.00 mg/kg nornicotine in the first 30 min of the session. There was also a decrease in total food intake for 40 min following injections of 6.00 mg/kg. There was a significant main effect of bin number (F(11,110) = 34.22; p < 0.001) and dose (F(4,40) = 28.87; p < 0.001) on mean total food intake following injections of anatabine, with a significant bin × dose interaction (F(44,440) = 8.50; p < 0.001). Significant decreases were observed during the first 40 min following injections of 3.00 mg/kg anatabine. Total food intake was also decreased after injections of 0.25 mg/kg in bin 2, and after 1.00 mg/kg in bins 2 and 3 of the session. There was a significant main effect of bin number (F(11,110) = 41.11; p < 0.001) and dose (F(4,40) = 26.75; p < 0.001) on mean total food intake following injections of anabasine, with a significant bin × dose interaction (F(44,440) = 14.37; p < 0.001). Significant decreases in intake were observed for the first 40 min of the session following injections of 3.00 and 4.00 mg/kg anabasine compared to saline. Total food intake was also decreased by 1.00 mg/kg anabasine at the end of the session (bin 12). There was a significant main effect of bin number (F(11,550) = 104.8; p < 0.001), without a significant main effect of cotinine dose or a significant interaction between cotinine dose and bin number, on food intake (see Supplementary Figure 1, panel C).</p><p>Analysis of event records of pellet deliveries revealed that animals consumed either a single big bout or two bouts of food under baseline conditions, with the majority of intake occurring early during the session in bout one (68%). Nicotine and the minor tobacco alkaloids altered this pattern of intake. To convey drug effects at this more molar level of analysis, the session was separated into two bouts of food intake. The first bout began at the start of the session and ended when there was a 10 minute break without food intake. Some animals didn't consume anything for the first 10 minutes or more, resulting in zero intake for bout one. Other rats began eating immediately, and consumed all of the pellets for the session in bout 1. In this case bout two intake was considered zero (occurred ~30% of the time). Three or more bouts were rare (~10% of the time). In this case, the final two bouts were combined and analyzed as the second bout.</p><p>Figure 4 shows mean food intake following saline injections and all four doses of nicotine, nornicotine, anatabine, and anabasine, during each bout of feeding before or after a pause in feeding of at least 10-minutes. There was a significant main effect of nicotine dose (F(4,56) = 18.83; p < 0.001), but not bout, and a significant bout × dose interaction (F(4,56) = 23.42; p < 0.001) on mean total food intake. A significant decrease in intake was observed at 0.50, 0.75, and 1.00 mg/kg nicotine during the first bout, and an increase in intake at 0.75 and 1.00 mg/kg nicotine during the second bout, compared to saline. Mean total intake differed at all doses between the first and second bouts of intake. There was a significant main effect of nornicotine dose (F(4,40) = 8.03; p < 0.001) and bout (F(1,10) = 56.73; p < 0.001), but no bout × dose interaction, on mean total food intake. A significant decrease in intake was observed at 6.00 mg/kg nornicotine during the first bout of intake. Mean total food intake differed at all doses, except 6.00 mg/kg nornicotine, between the first and second bouts of intake. There was a significant main effect of anatabine dose (F(4,40) = 30.66; p < 0.001) and bout (F(1,10) = 78.35; p < 0.001), and a significant bout × dose interaction (F(4,40) = 9.78; p < 0.001), on mean total food intake. A significant decrease in intake was observed during the first bout at 3.00 mg/kg anatabine. Mean total food intake differed between bouts at all doses, except for 3.00 mg/kg anatabine. There was a significant main effect of anabasine dose (F(4,40) = 25.97; p < 0.001) and bout (F(1,10) = 62.77; p < 0.001), and a significant bout × dose interaction (F(4,40) = 31.22; p < 0.001), on mean total food intake. Significant decreases in intake were observed at 3.00 and 4.00 mg/kg anabasine during the first and second bouts of intake. Mean total food intake differed at all doses between the first and second bouts of intake.</p><!><p>Figure 5 (panel A) shows the mean total food intake for each dose of nicotine alone or Kodiak. There was a significant main effect of dose (F(3,18) = 13.93; p < 0.001), but no effect of drug or a drug × dose interaction. Significant reductions in food intake occurred at 0.50 and 1.00 mg/kg nicotine and all doses of Kodiak extract. Figure 5 (panel B) shows the mean latency to first pellet for each dose of nicotine and Kodiak extract. There was a significant main effect of dose (F(3,18) = 19.84; p < 0.001), but no significant main effect of drug or a drug × dose interaction. Significant increases in latency occurred at 1.00 mg/kg nicotine only. Figure 5 (panel C) shows the mean running rates following injections of nicotine or Kodiak extract. There was no effect of drug or dose, or a drug × dose interaction, although there was a trend (p = 0.055) for a significant main effect for dose. There was a main effect of dose (F(3,15) = 23.66; p<0.001) on running rates, such that all doses of nicotine and Kodiak extract were decreased relative to saline. Figure 5 (panel D) shows mean total food intake during 10-minute bins following injections of nicotine and Kodiak extract. There was a main effect of bin number (F(11, 66) = 18.89; p < 0.001) and drug (F(4,24) = 6.54; p = 0.001), and a significant bin × drug interaction (F(44,264) = 4.05; p < 0.001). Significant decrease in intake occurred at both doses of nicotine and Kodiak extract compared to saline during the first 20 mins of the session. There was also a significant decrease in intake following 0.50 mg/kg nicotine during the final 30 minutes of the session.</p><!><p>Figure 6 (panel A) shows the mean total food intake for each dose of nicotine or EC liquid. There was a significant main effect of dose (F(3,18) = 12.2; p < 0.001), but no effect of drug or a drug × dose interaction. Significant reductions in food intake occurred at 0.50 and 1.00 mg/kg nicotine and EC liquid. Figure 6 (panel B) shows the mean latency to first pellet for each dose of nicotine and E-Juice liquid. There was a significant main effect of dose (F(3,18) = 18.6; p < 0.001), but no main effect of drug or a drug × dose interaction. Significant increases in latency occurred at 1.00 mg/kg nicotine and EC Liquid. Figure 6 (panel C) shows the mean running rates following injections of nicotine or EC liquid. There was a main effect of dose (F(2,10) = 15.77; p < 0.001), but no effect of drug or drug × dose interaction. There was a main effect of dose (F(3,15) = 27.46; p<0.001) on running rates, such that all doses of nicotine and Kodiak extract (with the exception of 0.25mg/kg Kodiak extract) were decreased relative to saline. Figure 6 (panel D) shows mean food intake during 10-minute bins for nicotine and EC liquid. There was a main effect of bin number (F(11, 66) = 8.08; p < 0.001) and drug (F(4,24) = 6.74; p = 0.001), and a significant bin × drug interaction (F(44,264) = 5.72; p < 0.001). Significant decreases in food intake occurred at both doses of nicotine and EC liquid compared to saline during the first 20 min. The 1.00 mg/kg nicotine dose decreased intake during the first 30 (bin 12).</p><!><p>The primary finding of the present study was that nicotine and all minor alkaloids, except for cotinine, produced a dose-dependent decrease in food intake. However, whereas nicotine and anatabine significantly reduced food intake across a wide range of doses, the other minor alkaloids only reduced food intake at doses more than tenfold higher than the lowest effective nicotine dose. In addition, latency to initiate feeding was increased at doses that produced decreases in food intake, with the exception of 0.25 and 0.50 mg/kg nicotine, 0.25 mg/kg anatabine, and 3.00 mg/kg nornicotine. These results indicate that, similar to nicotine, the minor tobacco alkaloids nornicotine, anabasine, and anatabine can all reduce food intake and motivation to initiate food intake, but with significantly lower potency than nicotine in the case of nornicotine and anabasine. Consistent with these findings, a smokeless tobacco extract and EC liquid produced decreases in food intake that were similar to nicotine alone at equivalent nicotine doses. These findings have important implications for understanding the appetite-suppressant effects of tobacco product use in humans.</p><!><p>In the present study, nicotine reduced food intake at all doses tested. These results are similar to what has been seen previously in several other noncontingent administration studies (16, 21, 55). The differences in potency between nicotine and nornicotine and anabasine observed in the present study are consistent with their different potencies for other behavioral effects. The highest dose of nornicotine reduced food intake by 18%. Considering the difference in dose needed to achieve a similar reduction in food intake (e.g. 0.75 vs 6.0 mg/kg), nicotine appears to be around 8-fold more potent than nornicotine. Although complete dose-response curves were not obtained in the present study, this is consistent with the relative potencies reported in other behavioral and neuropharmacological studies (35, 53, 56, 57). Similar decreases in food intake were also seen with anabasine at 3.00 mg/kg, indicating an approximate order-of-magnitude lower potency compared to nicotine. These findings are consistent with a study of minor alkaloid effects on food-maintained responding in mice by Caine et al. (2014) (33). However, the similar potency of anatabine and nicotine in the present study contrasts with the Caine et al. study, which reported a 13-fold lower potency for anatabine. This discrepancy is possibly due to the use of different species between studies. Cotinine on the other hand, had no effect on any measures of food intake at the doses tested, which is consistent with other studies showing little or no effect of cotinine on operant behavior in this dose range (47). The differences in potency between nicotine and the minor alkaloids may be due to differences in their neural mechanisms of action (56) or pharmacokinetics (58), or both. Some studies suggest the reduced penetrability of minor alkaloids through the blood brain barrier compared to nicotine may be involved (53). As such, the relatively low levels of these alkaloids in smokeless tobacco or e-cigarettes do not likely contribute to any appetite suppressant effects of these products. However, despite their lower potency, the minor tobacco alkaloids could serve as parent compounds to develop new, more potent forms.</p><p>The other minor alkaloids and higher doses of nicotine only reduced overall food intake at doses that also decreased running rate and latency to first pellet. This may reflect disruptive motoric or other side effects. Alternatively, it may simply indicate a general effect on food motivation for several reasons. First, reducing food deprivation produces a similar reduction in both food intake and latency to feeding under a similar mixed schedule of food reinforcement (51). Second, none of the doses of minor alkaloids in the present study had a significant effect on response rates in an operant food-maintained nicotine discrimination assay (35). Finally, these doses of nicotine and minor tobacco alkaloids have no effect on response latencies in an intracranial self-stimulation (ICSS) model (47). It is more likely that the reduction in food intake at high doses of these alkaloids was due to general motivational effects. It should be noted that the highest dose of nicotine produced seizures in some of the rats. These seizures resolved and rats were able to move freely around the cage without any apparent motor disruption by the time they were placed in the chambers. Any other motoric effects (e.g., unsteady gate, freezing) also appeared to resolve prior to starting the session.</p><p>Nicotine, nornicotine, and anabasine have all been shown to elevate ICSS thresholds at the highest doses used in the present study. To the extent that an elevation in ICSS threshold provides a measure of the function of fundamental brain pathways mediating motivation and reinforcement, this suggests that a non-specific motivational deficit (i.e. reduced efficacy of any type of reinforcer) may have contributed to their anorectic effects at these doses. Lower doses of these drugs, however, do not produce elevations in ICSS thresholds, suggesting that the reduction in running rate by nicotine, nornicotine, and anabasine reflects an attenuation of food motivation (i.e. both reduced hunger and enhanced satiation) rather than a non-specific motivational deficit. Similarly, none of the doses of anatabine that reduced food intake in the present study produce motivational or performance deficits in an ICSS model (47), suggesting their effect was due to an attenuation of food motivation per se.</p><p>The analysis of food intake across 10-min segments of the session allowed a relatively fine-grained analysis of the time course of drug effects and revealed some differences between the alkaloids. Nicotine, nornicotine, and anabasine all reduced intake early in the session at the same doses that reduced mean total intake. Other studies have shown a similar time course for the effects of nicotine and nornicotine on food-maintained responding, with levels returning to baseline within 60 minutes (57, 59). In contrast, anatabine reduced early session intake at doses that did not reduce total intake across the entire session. Differences in half-lives between these compounds don't directly correspond to the differences in time course of effects seen between nicotine and the minor tobacco alkaloids, as their elimination half-lives range from 9–20 hours (58). A better correlation between the relative half-lives of the alkaloids and their time course for suppression of food intake may have been observed in the present study had longer sessions and different levels of food deprivation been used.</p><p>Differences between alkaloids in their effect on the within-session pattern of food intake were also apparent at a more macro level of analysis. Baseline sessions typically involved an initial bout of food intake at a high rate, then a pause in feeding, followed by a second bout of intake at a lower rate for the remainder of the session. Nicotine, nornicotine, and anatabine, but not anabasine, typically reduced food intake in the beginning of the session without compensatory increases in food intake in the second bout. At high doses of nicotine, significant decreases in food intake were observed early in the session, followed by a rebound in food intake above saline control levels later in the session. Such compensatory effects were not observed with lower nicotine doses. Significant compensation was also not seen with nornicotine or anatabine at any dose. As such, these compounds may have a therapeutic advantage for the treatment of obesity and/or post-cessation weight gain, in that they are not associated with potentially problematic rebound effects (60, 61). However, given the longer half-lives of these compounds noted above, it is necessary to examine them in extended access sessions to better evaluate this. It is also important to mention that we were unable to verify that the pellets were consumed immediately following lever pressing. However, because of the limited session duration and high number of pellets consumed, it seems unlikely that there would have been long intervals between lever pressing and pellet consumption.</p><!><p>The Kodiak extract and EC liquid used in the present study produced changes in food intake, latency to first pellet, and running rates that were comparable to nicotine alone. The failure to find any differences in food intake between nicotine, Kodiak extract, and EC liquid suggests that the primary driving force behind the reductions in food intake following use of these tobacco products is nicotine. These findings are consistent with a prior animal study reporting that smoking-relevant doses of minor alkaloids and other compounds present in cigarette smoke (e.g. harman, norharmane, acetaldehyde) do not moderate weight gain or energy expenditure in rats when they are self-administered as a cocktail with nicotine (14). Thus, at their present relative concentrations, minor alkaloids may not contribute to any appetite suppressant effects of tobacco products or ECs.</p><p>The results from the present study are consistent with research suggesting that smokeless tobacco use can reduce weight gain in its own right and attenuate increased weight gain during smoking cessation in a manner similar to nicotine replacement therapies (NRT). For example, one study reported that smokeless tobacco use reduced weight gain (62), and another study found an increase in weight gain in individuals who quit using smokeless tobacco relative to nonusers and those continuing to use smokeless tobacco (63). In addition, less weight gain has been reported in individuals who switched from smoking cigarettes to using smokeless tobacco (63) compared to those who quit smoking without switching. Smokeless tobacco may therefore be similar to NRT, which is also able to have some beneficial effects on body weight during smoking cessation (64). However, some studies have failed to show any weight loss with smokeless tobacco use (65). These findings suggest that while smokeless tobacco may reduce weight gain, its effects may be less robust than cigarette smoking. Whether this is due to differences in relative concentrations of nicotine and non-nicotine constituents between cigarettes and smokeless tobacco is unclear. A direct comparison of cigarette smoke extracts and smokeless tobacco extracts using methods similar to the present study would help address this issue.</p><p>Few studies have examined the effects of ECs on food intake. A recent investigation evaluating post-cessation weight gain in individuals who switched from conventional cigarettes to e-cigarettes showed an ability of EC use to mitigate post-cessation weight gain (43). In addition, there was a lower incidence of increases in hunger in individuals who had switched to ECs in that study. These findings are consistent with results from the present experiment and suggest that ECs may be an effective tool for helping individuals quit smoking without significant weight gain. In contrast to the present findings, one study in mice showed that EC aerosol had no significant effect on food intake or body weight during exposure or withdrawal, despite brain nicotine levels being comparable to those from exposure to cigarette smoke, which did reduce food intake (46, 60). This inconsistency with the present study may be due to the EC used, route of administration (vapor v. subcutaneous injection), species used (mouse v. rat), or the methods for calculating food consumption (total consumption divided by number of mice in each box v. individual measurements).</p><p>It is worth mentioning that under the Family Smoking Prevention and Tobacco Control Act, the FDA has the authority to reduce the nicotine content in tobacco products as a population-wide approach to facilitate smoking cessation. One concern is that continued presence of non-nicotine constituents like minor tobacco alkaloids might contribute to the effects of smoking on body weight and facilitate continued smoking of reduced nicotine content cigarettes among smokers concerned about weight gain. The significant effects of the minor tobacco alkaloids on food intake in Experiment #1 of the present study were found with doses considerably higher than those delivered by tobacco products. Accordingly, there was no difference between nicotine and E-juice/Kodiak extracts in Experiment 2. It is therefore unlikely that reduced-nicotine tobacco products would maintain weight loss and thereby facilitate continued use.</p><!><p>There are a number of methodological limitations of the present study that need to be addressed in future studies. First, because the session only lasted 2 h, we were unable to determine the longer-term effects of acute exposure to the alkaloids. Second, drug effects were assessed in a state of high motivation to consume food. As such, the dose-response curves (i.e. potencies) for these alkaloids may differ at lower levels of food deprivation. Third, food restriction made it unfeasible to examine weight gain as a dependent measure. Fourth, the effects of chronic exposure to nicotine and minor tobacco alkaloids, as occurs in humans, were not examined. In addition, there are thousands of other constituents in tobacco products and smoke and numerous other tobacco products on the market that were not examined in the present study. It will be important to examine whether and to what extent these may also affect weight gain and food intake. Despite these limitations, the present study addresses and important knowledge gap regarding the potential role of non-nicotine constituents in moderating the effects of tobacco product use and cessation on food intake and body weight gain. Addressing its limitations will be an important next step to examine the effects of tobacco alkaloids under more clinically relevant dosing and feeding conditions.</p><!><p>In the present study, the minor tobacco alkaloids nornicotine, anatabine, and anabasine, all produced reductions in food intake within the first 30 minutes of injections without significant compensatory increases in food intake later in the session. These reductions in food intake produced changes in meal patterns, specific to meal size, in a manner that was similar to the effects of nicotine alone. This study provides novel preliminary evidence that individual minor tobacco alkaloids have effects on food intake and suggests that further exploration of the potential effects of these compounds on body weight and metabolism are warranted. The present study also suggests that nicotine is the primary component in EC liquid and smokeless tobacco products that affects food intake. The current levels of minor tobacco alkaloids and other non-nicotine constituents do not appear to be sufficient to contribute to any appetite suppressant effects of these products.</p>

PubMed Author Manuscript

Total Synthesis of the Diterpenoid Alkaloid Arcutinidine Using a Strategy Inspired by Chemical Network Analysis

Arcutinidine and other arcutinidine-type diterpenoid alkaloids feature an intricate polycyclic, bridged framework with unusual connectivity. A chemical network analysis approach to the arcutane skeleton enabled the identification of highly simplifying retrosynthetic disconnections, which indicated that the caged structure could arise from a simpler fused ring system. On this basis, a total synthesis of arcutinidine is reported herein, featuring an unprecedented oxopyrrolium Diels-Alder cycloaddition which furnishes a key tetracyclic intermediate. In addition, the synthesis utilizes a diastereoselective oxidative dearomatization/cycloaddition sequence and a SmI2-mediated C-C coupling to forge the bridged framework of the natural products. This synthetic plan may also enable future investigations into the biosynthetic relationships between the arcutanes, the related diterpenoid atropurpuran, and other diterpenoid alkaloids.Scheme 3. Completion of arcutinidine (4). Comple4Celik and Dr. Jeffrey Pelton (UC Berkeley) for assistance with NMR experiments, Dr. Nicholas Settineri (UC Berkeley) for single-crystal X-ray diffraction studies, and Dr. Ulla Anderson for assistance with gathering HRMS data. We also thank Dr. Dave Small and the Molecular Graphics and Computational Facility (NIH S10OD023532) for assistance with computations. We are grateful to Prof. Reinhard Hoffman for insightful discussions.

total_synthesis_of_the_diterpenoid_alkaloid_arcutinidine_using_a_strategy_inspired_by_chemical_netwo

<!>Scheme 2. Synthesis of key tetracycle 12.

<p>Natural products that possess a high degree of three-dimensional structural complexity pose significant challenges to identifying strategies for their chemical synthesis. By adopting a retrosynthetic plan that reduces the number of bridged rings in these architecturally intricate structures, the resulting fused ring systems can prompt retrons that guide subsequent disconnections. In 1975, Corey introduced a formalized 'logic' for the retrosynthesis of bridged, polycyclic frameworks that expounded the virtues of identifying a 'maximally bridged ring', which upon disconnection, provides maximal structural simplification. 1 Furthermore, by applying two-bond disconnections (i.e., 'bicyclization transforms'), a rapid decrease in target complexity in the retrosynthetic direction can be realized. In this regard, cycloaddition transforms have proven indispensable.</p><p>The diterpenoid alkaloids are a large group of secondary metabolites with highly complex, three-dimensional frameworks that possess wide-ranging activity as modulators of voltage-gated ion channels. [2][3][4] These structures lend themselves very well to chemical network analysis, which can provide powerfully simplifying bond disconnections as starting points. In 2015, our laboratory reported the application of Corey's chemical network analysis toward syntheses of aconitine-type diterpenoid alkaloids, including liljestrandinine (1, Figure 1A). 5 Similar analyses have also guided our preparation of denudatine-type alkaloids such as paniculamine (2), and the hetisine-type alkaloid cossonidine (3). 6 Building on Corey's graphical analysis, we also introduced a web-based graphing algorithm to facilitate the identification of the maximally bridged ring for any given structure. 7 In this Communication, we describe the application of chemical network analysis to identify a strategy for the synthesis of arcutinidine (4), an arcutane-type diterpenoid alkaloid. Natural products in this class include arcutine (5), arcutinine (6), and aconicarmicharcutinium (7), [8][9][10] the framework of which is presented from three perspectives in Figure 1B (gray box).</p><p>Perhaps even more captivating than the ornate framework of these natural products is their postulated biosynthesis. Arcutinidine (4) is believed to arise from oxidation and rearrangement of the hetidine skeleton (see inset in Figure 1B). This presumed cation-initiated rearrangement is supported by computations that we reported in 2015. [11][12] These calculations also indicate that the reverse of the proposed biosynthesis (i.e., 9®8) may also be possible under laboratory conditions; a transformation which could provide synthetic access to the hetidine alkaloids from the arcutinidine skeleton. Furthermore, oxidation and hydrolysis of the arcutinidine 1-pyrroline moiety could yield the related diterpenoid atropurpuran (10). [13][14][15] Therefore, a synthesis of arcutinidine could set the stage for investigation of the possible biogenesis of atropurpuran and provide access to other diterpenoid alkaloids. During the preparation of this manuscript, Qin and coworkers published their synthesis of 4 and 6. 16 In contemplating a synthetic approach to 4, we sought disconnections that would rapidly reduce structural complexity by applying multiple cycloaddition transforms. We identified two comparable maximally-bridged rings, one of which is a 'primary' ring whereas the other is an 'envelope' ring 17 (see Scheme 1, gray box). A strategic bond disconnection of a key ex-endo bond, 18 as shown for 4, led back to 11 as a precursor. Application of a [4+2] bicyclization transform to 11 unveiled tetracycle 12 as a precursor. In turn, 12 could arise in the forward sense through another [4+2] cycloaddition of diene 14 19 and oxopyrrolium dienophile 13, a reaction that would forge two contiguous, all-carbon quaternary stereocenters. To our knowledge, no precedent exists for cycloadditions involving oxopyrrolium dienophiles, thus our work would serve to probe the unprecedented reactivity of these heterocyclic cations. In the forward sense, our desired oxopyrrolium 13 was envisioned to arise through a Wittig reaction between known citraconimide 15 20 and methyl enol ether 16 21 and subsequent Friedel-Crafts cyclization.</p><p>We commenced our studies by investigating a more conventional cycloaddition between diene 17 and maleimide 18 (Figure 2). On the basis of HOMO-LUMO considerations (compare HOMO of -5.6 eV for 17 versus LUMO of -2.4 eV for 18), 22 it was anticipated that poor reactivity and selectivity would be observed due to the lack of electronic differentiation at the reacting carbon atoms. Conversely, the LUMO of oxopyrrolium 13 (-4.3 eV) is energetically matched and possesses complementary polarization relative to the HOMO of diene 14 (-6.5 eV). 23 These factors were expected to contribute to a more facile and selective cycloaddition.</p><p>Maleimide derivative 18 and the oxopyrrolium precursor 19</p><p>were prepared as detailed in Scheme 2. The synthesis began with the treatment of citraconimide 15 with PPh3 and methyl enol ether 16 in a modified Wittig reaction 24 to give the non-conjugated alkene adduct, which was isomerized to the desired maleimide (18) upon treatment with DBU. As expected, our initial attempts to employ 18 in a cycloaddition with dienes such as 17 25 resulted in poor regioselectivity. The resulting cycloadducts were separable upon hydrogenation of the double bond and the desired constitutional isomer 20 was advanced through several steps. Ultimately, this synthetic plan proved untenable, in part, because of the inefficiency of the cycloaddition step. Our attention was therefore turned to investigating oxopyrrolium dienophile 13. Upon exposing imide 18 to TfOH, a Friedel-Crafts cyclization ensued to afford key tricyclic intermediate 19. This tricycle serves as a masked oxopyrrolium dienophile that can be unveiled under acidic conditions. When 19 was treated with AlCl3 and diene 14, a Diels-Alder cycloaddition, presumably through the intermediacy of oxopyrrolium 13, occurred to furnish 22 bearing two new vicinal quaternary carbon centers. Tetracycle 22 possesses nearly all of the carbon atoms found in arcutinidine and can be readily synthesized on multigram scale. Reduction of the cyclohexenyl double bond in 22, using Rh/Al2O3 and hydrogen, followed by ionic reduction of the hemiaminal functional group using BF3•OEt2 and Et3SiH, gave amide 23 as a single diastereomer. The amide group was then reduced to the corresponding amine using LiAlH4. Concomitant reduction of the acetate unveiled a secondary hydroxy group, which was subsequently re-acetylated to give amine 12.</p><p>A highly selective mono-demethylation of the veratrole unit using TMSI afforded guiacol 24, which was primed to undergo oxidative dearomatization (Scheme 3). Following the addition of in situ generated lead (IV) acrylate, 24 underwent a diastereoselective Wessely-type oxidation 26 to generate a masked diketone intermediate, appended with an acrylate dienophile (25). Upon heating, an intramolecular [4+2] cycloaddition proceeded to afford the [2.2.2] bicycle that is present in 11 and is characteristic of the arcutanes, as well as a large subset of other diterpenoid alkaloids. With all the carbon atoms found in arcutinidine now in place, we pursued the installation of the last C-C bond to complete the arcutinidine skeleton. We sought to utilize a reductive C-C bond formation by employing a pinacol coupling inspired by the Qin synthesis of atropurpuran. 14 Upon dissolution in methanol, lactone 11 underwent solvolysis to afford the ring opened Scheme 1. Network-analysis-guided retrosynthesis of arcutinidine.</p><!><p>Figure 2. HOMO-LUMO considerations calculated using DFT basis set 6-31G**++ with a B3LYP functional and empirical dispersion. product featuring a diketone on the [2.2.2] bicycle and a methyl ester on the bridge of the bicycle. In the same pot, introduction of SmI2 effected reductive removal of one of the carbonyl groups to give a hydroxy group at C9 as a single, inconsequential, diastereomer. At this stage, solvolysis of the acetate under microwave irradiation and subsequent global oxidation with DMP yielded diketone 26. Pinacol coupling using SmI2 gave the desired hexacyclic diol (27), which possesses the full arcutinidine skeleton. The structure of 27 was unambiguously confirmed through X-ray crystallographic analysis.</p><p>Reduction of the highly strained double bond in the [2.2.2] bicycle of 27 using diimide gave the fully saturated arcutinidine core (28). Upon treatment of vicinal diol 28 with InCl3 and Ph2SiHCl, 27 elimination of the C10 hydroxy group and a subsequent deoxygenation of the other-now allylic-C9 hydroxy group ensued to give alkene 29. At this stage, LiAlH4 reduction of the methyl ester, followed by Mukaiyama hydration to reinstall the tertiary hydroxy group at C10, gave diol 30. Hydrogenolysis of the benzyl group in diol 30 with Pearlman's catalyst in the presence of acetic acid, followed by N-chlorination of the resulting secondary amine gave 31. The N-Cl moiety acted as a protecting group for the amine and allowed for mesylation of the primary hydroxy group to give mesylate 32. With leaving groups installed on the nitrogen and at C17 of the [2.2.2] bicycle, a double elimination was effected by heating mesylate 32 in the presence of DBU and NaI to afford imine 33 bearing an exocyclic alkene. This penultimate imine was advanced to arcutinidine (4) and its C15 epimer (34) through an allylic oxidation with SeO2.</p><p>In summary, we report the synthesis of the arcutane-type diterpenoid alkaloid arcutinidine. 28 Our synthetic approach was inspired by chemical network analysis, which enabled rapid simplification of the three-dimensional architecture of the target compound through [4+2] cycloaddition transforms. Ultimately, these disconnections led us to identify an oxopyrrolium intermediate as a viable dienophile in an unprecedented [4+2] cycloaddition reaction. The synthesis reported herein sets the stage for the preparation of the related congeners arcutine (5) and arcutinidine (6) and may provide a starting point for the preparation of atropurpuran (10). These studies, as well as the development of an enantioselective cycloaddition of the oxopyrrolium dienophile and the potential conversion of the arcutane skeleton to the hetidine skeleton, are the subject of ongoing investigations in our laboratory.</p>

ChemRxiv

The micromechanics of lung alveoli: structure and function of surfactant and tissue components

The mammalian lung´s structural design is optimized to serve its main function: gas exchange. It takes place in the alveolar region (parenchyma) where air and blood are brought in close proximity over a large surface. Air reaches the alveolar lumen via a conducting airway tree. Blood flows in a capillary network embedded in inter-alveolar septa. The barrier between air and blood consists of a continuous alveolar epithelium (a mosaic of type I and type II alveolar epithelial cells), a continuous capillary endothelium and the connective tissue layer in-between. By virtue of its respiratory movements, the lung has to withstand mechanical challenges throughout life. Alveoli must be protected from over-distension as well as from collapse by inherent stabilizing factors. The mechanical stability of the parenchyma is ensured by two components: a connective tissue fiber network and the surfactant system. The connective tissue fibers form a continuous tensegrity (tension + integrity) backbone consisting of axial, peripheral and septal fibers. Surfactant (surface active agent) is the secretory product of type II alveolar epithelial cells and covers the alveolar epithelium as a biophysically active thin and continuous film. Here, we briefly review the structural components relevant for gas exchange. Then we describe our current understanding of how these components function under normal conditions and how lung injury results in dysfunction of alveolar micromechanics finally leading to lung fibrosis.

the_micromechanics_of_lung_alveoli:_structure_and_function_of_surfactant_and_tissue_components

The structural components for gas exchange<!><!>The structural components for gas exchange<!><!>The structural components for gas exchange<!><!>The structural components for gas exchange<!>Function under normal conditions<!>Extracellular matrix, cells and micromechanics of ductal and alveolar airspaces<!><!>Extracellular matrix, cells and micromechanics of ductal and alveolar airspaces<!>Surface tension and alveolar micromechanics<!>Mechanisms of alveolar micromechanics<!><!>Mechanisms of alveolar micromechanics<!>Dysfunction: lung injury and fibrosis<!>Outlook

<p>In biology, function follows form and form follows function. The mammalian lung is a paradigmatic example for this principle. It is a complex organ whose functional capacity as a gas exchanger is directly determined by its microstructure (for review, see Weibel 1984, 2017; Ochs and Weibel 2015; Ochs et al. 2016; Hsia et al. 2016). In humans, at the end of a deep inspiration, over 80% of the total volume of the lung is air, and about 10% is blood. Thus, less than 10% (only a few 100 g) is made of "real" tissue. This tissue consists of more than 40 cell types, originating from all three germ layers, and a sophisticated connective tissue network. Together they form an organ with a complex architecture optimized to serve its main function.</p><p>Gas exchange takes place in lung alveoli. The alveolar region (parenchyma) of the lung comprises about 90% of its total volume. The remaining non-parenchyma consists of conducting airways (which are part of the anatomic dead space) and larger vessels. The airways branch in irregular dichotomy into the lung together with the arteries, thus defining broncho-arterial units "inside-out" (from the hilum to the periphery). In the most distal branching generations, alveoli are connected to the airways. Clusters of alveoli are arranged in functional units termed acini. An acinus is defined as the blind-ending parenchymal unit beginning with a transitional (i.e. the first generation of an alveolated) bronchiole. Within an acinus, all airways (alveolar ducts and, depending on the species, respiratory bronchioles) have alveoli attached to their walls and thus participate in gas exchange. Actually, the "wall" of alveolar ducts consists of nothing more than a network of alveolar openings. In the human lung there are around 30,000 acini (reviewed in Weibel et al. 2005).</p><p>The inter-alveolar septum provides the structural basis for gas exchange in the lung (see Weibel 1973; Weibel and Gil 1977; Maina and West 2006). It separates the air compartment (alveolar airspace) from the blood compartment (capillary lumen). The design of the inter-alveolar septum must meet certain structural requirements for efficient diffusion of oxygen and carbon dioxide between air and blood, in particular providing a large surface area (in the human lung about 140 m2) and a thin diffusion barrier (in the human lung about 2 µm) (Gehr et al. 1978). Moreover, it must also meet the functional requirements of stability (thereby preventing over-distension or collapse of alveoli) and flexibility (thereby being able to follow the movements of the thorax during the breathing cycle). Lung alveoli are mechanically stabilized by two factors: the pulmonary surfactant system and the lung´s connective tissue backbone (see below).</p><p>Within the inter-alveolar septum, the tissue barrier separating air and blood consists of two continuous cell layers: an epithelium facing the alveolar lumen and an endothelium facing the capillary lumen. Between them is an interstitial space of varying thickness and composition. The alveolar epithelium is a mosaic of type I alveolar epithelial cells which cover around 95% of the total alveolar surface with their thin squamous cell extensions interspersed with single cuboidal type II alveolar epithelial cells which are easily recognized by their characteristic secretory organelles, the surfactant-storing lamellar bodies (Fig. 1). Thus, there are two alveolar epithelial cell types with distinct functional specialization and thus distinct structural differentiation. Renewal and repair are provided by the type II cells which, in addition to their secretory function, constitute the progenitor cell population of the alveolar epithelium. The numerical ratio of type I to type II cells is about 1:2 (Crapo et al. 1982). Both type I and type II cells may face more than one alveolar lumen, i.e. they are able to cross the inter-alveolar septum thereby contributing their lining (type I cells) or secretory (type II cells) functions to both sides of the inter-alveolar septum. This level of structural (and thereby functional) complexity has so far not been taken into account in any of the in vitro models of the air-blood barrier. The continuity and tightness of the alveolar epithelium and the capillary endothelium are essential for fluid balance in the alveolar region as these tissue layers separate (and actively regulate) extracellular fluid compartments: blood plasma, interstitial fluid and alveolar lining fluid.</p><!><p>Transmission electron microscopy. Human lung. Inter-alveolar septum with type I (AEC1) and type II (AEC2) alveolar epithelial cell. Note surfactant-storing lamellar bodies (LB) in type II cell. Arrowheads mark tight junctions between type II cell and neighbouring type I cell extensions. Collagen fibrils (col) are present in the interstitium. Alv alveolar lumen, Cap capillary lumen, Endo capillary endothelial cell. Scale bar 2 µm</p><!><p>Careful electron microscopic studies revealed the existence of a thin and continuous alveolar lining layer consisting of a surface film and an aqueous hypophase (Weibel and Gil 1968; Gil and Weibel 1969/70). This means that the alveolar epithelium is not directly exposed to air but covered by a liquid lining layer with an estimated mean thickness of about 200 nm in the rat lung (Bastacky et al. 1995). Surfactant is present in the hypophase and constitutes the surface film at the air–liquid interface (for review, see Perez-Gil 2008; Ochs 2010; Ochs and Weibel 2015; Olmeda et al. 2017). All surfactant components (about 90% lipids, mainly saturated phospholipids, and about 10% proteins, including the surfactant proteins SP-A, SP-B, SP-C and SP-D) are synthesized, stored, secreted and to a large extent recycled by type II alveolar epithelial cells (Fig. 1). Most of the intracellular surfactant (at least lipids and the hydrophobic SP-B and SP-C) is assembled in specific organelles, the lamellar bodies, prior to secretion. Intra-alveolar surfactant includes the surface film and different subtypes in the hypophase that can be distinguished morphologically (Fig. 2). Interestingly, these morphologically distinct subtypes largely correspond to different stages in surfactant metabolism and activity. Freshly secreted lamellar bodies transform into tubular myelin, which may act as precursor of the surface film at the air–liquid interface although additional multilayered surface-associated reservoirs have been suggested. "Spent" surfactant is usually present as small unilamellar vesicles which can be taken up by type II cells for recycling or degradation or by alveolar macrophages (the "vacuum cleaners" within the hypophase) for degradation. Overall, surfactant has biophysical as well as immunomodulatory functions. In particular, surfactant stabilizes alveolar dimensions and thus prevents alveolar collapse by a surface-area dependent reduction of alveolar surface tension and is, therefore, essential for normal alveolar micromechanics and lung function (see below).</p><!><p>Transmission electron microscopy. Human lung. Inter-alveolar septum with collagen fibrils (col) and elastic fibers (el). The alveolar epithelium (thin type I cell extension marked by arrowheads) is covered with a lining layer containing intra-alveolar surfactant (Surf). Alv alveolar lumen. Scale bar 1 µm. Inset shows tubular myelin, a surface-active intra-alveolar surfactant subtype, at higher magnification. Scale bar 0.5 µm</p><!><p>The interstitium, i.e. the bounded space between the alveolar epithelial and capillary endothelial basal laminae, contains cells and an extracellular network of elastic fibers and bundles of banded collagen fibrils forming fibers (Weibel and Crystal 1997). The most abundant cells in the interstitium are the fibroblasts. They are a heterogeneous cell population. While the "classical" fibroblasts produce and maintain the extracellular matrix, many of them have mainly contractile properties. These myofibroblasts contain filaments oriented across the inter-alveolar septum, thus connecting the two epithelial sides of the septum and bracing the interstitial space (Kapanci et al. 1974). Through pores in the basal lamina, myofibroblasts are able to directly link alveolar epithelium and capillary endothelium (Sirianni et al. 2003). The extracellular connective tissue fiber network is interwoven with the alveolar capillary network (Weibel and Crystal 1997; Weibel and Bachofen 1997). Thereby, the air-blood barrier has thick parts where cell nuclei and the fiber network are concentrated (thus providing regenerative capacity and mechanical stability) and thin parts where alveolar epithelium and capillary endothelium share one common basal lamina (thus preventing fluid accumulation and minimizing the thickness of the diffusion barrier to considerably less than 1 µm). In the human lung, about half of the total barrier surface is thin (Weibel 1973; Weibel and Gil 1977). According to its location, the connective tissue network can be subdivided into axial, peripheral and septal fibers (Weibel and Gil 1977; Weibel 2009). Axial fibers enwrap airways from the hilum where the main bronchus enters the lung down to the alveolar ducts where they form a network of opening rings into alveoli. Peripheral fibers extend from the visceral pleura into interlobular septa, thus demarcating broncho-arterial units "outside-in" (from the periphery to the hilum). The septal fibers are anchored to both axial (at the alveolar entrance rings) and peripheral (at the boundary of acini) fibers, thereby constituting a fiber continuum throughout the lung. This continuous network of connective tissue forms a self-stabilizing tensegrity (tension + integrity) structure in the lung (Weibel 2009; Ingber 2003). Of particular interest for alveolar micromechanics are the free edges of the inter-alveolar septa where the axial fiber system of alveolar ducts is connected to the septal fibers of alveoli (Wilson and Bachofen 1982). At these alveolar entrance rings (which also represent the "wall" of alveolar ducts), connective tissue (mostly elastic) fibers and smooth muscle cells can be found (Fig. 3). Air within distal acinar airspaces is either present inside alveolar entrance rings (alveolar airspace) or outside (ductal airspace). Gas exchange is only possible within the alveolar airspace with its capillarized septa. Thus, the ratio of alveolar airspace vs. ductal airspace is highly relevant for proper lung function (see below).</p><!><p>Transmission electron microscopy. Human lung. Inter-alveolar septum with free edge (right) indicating the opening into an alveolar lumen (Alv). Note reinforced entrance ring with elastic fibers (el) at the alveolar opening where the axial fibers are connected to the septal fibers. col collagen fibrils, Fb fibroblast extensions, bl alveolar epithelial basal lamina, arrowhead tight junction between type I alveolar epithelial cells. Scale bar 1 µm</p><!><p>The total surface area of the alveolar capillary endothelium matches that of the epithelium. The individual endothelial cell, however, covers a much smaller surface as compared to its epithelial counterpart, the type I cell (Crapo et al. 1982). This is because endothelial cells have a less complex branching architecture than type I cells which possess several cytoplasmic plates lining the inter-alveolar septum (Weibel 2017). The structure of the alveolar capillary network is considerably different from those of the systemic circulation (Weibel 1963; Mühlfeld et al. 2010; Townsley 2012). Its individual segments form small loops (where the length of the segments is in the range of the capillary diameter), thus resulting in a dense meshwork with vertical tissue pillars in-between. This provides the structural basis for the sheet-flow concept in alveolar capillaries (Fung and Sobin 1969). Because the arteries in the lung follow the conducting airways and their branching in the center of broncho-arterial units whereas the veins run in the connective tissue septa separating them, blood flow is directed from the center to the periphery in these units. In the early postnatal phase a double capillary network layer (one for each alveolar lumen) exists in inter-alveolar septa. After microvascular remodeling, only a single layer is present in the adult lung. As a consequence, this single layer exchanges oxygen and carbon dioxide with two adjacent alveoli. When the septal fibers that are interwoven with this single capillary layer are stretched, the capillaries are spread over both alveolar surfaces of the inter-alveolar septum in a zig-zag pattern, thus maximizing the contact area between air and blood with as little interstitial connective tissue as possible (Weibel and Gil 1977; Ochs and Weibel 2015).</p><!><p>During the respiratory cycle, the distal airspaces of lung parenchyma are continuously subjected to volume changes. These volume changes impose deformation on ductal and alveolar airspaces and most importantly on inter-alveolar septa. Such deformations can best be described with the term strain which is the size (e.g. length, surface or volume) of a structure after deformation in relation to the baseline situation (Vlahakis and Hubmayr 2005). At the organ scale, the strain imposed on the lung is accordingly calculated using the tidal volume, corresponding to the lung deformation, and the functional residual volume, corresponding to the baseline situation of the lung. During mechanical ventilation the tidal volume is given by the ventilator while the functional residual volume equals the volume of the lung at a given positive end-expiratory pressure (PEEP). Stress, on the other hand, is defined as the force per area so that stress and pressure have the same unit (Vlahakis and Hubmayr 2005). At the microscopic level, pressure-change related alterations of the microarchitecture and, therefore, deformation have been described in ductal and alveolar airspaces as well as inter-alveolar septa (Gil et al. 1979; Bachofen et al. 1987; Tschumperlin and Margulies 1999; Roan and Waters 2011). Based on these observations, it makes physiological sense to distinguish ductal from alveolar airspaces since from an anatomical point of view they are differently bordered with partly different stabilizing elements which result in different mechanical properties (Wilson and Bachofen 1982; Haefeli-Bleuer and Weibel 1988).</p><!><p>An economically designed fiber network serves the stabilization of distal airspaces bearing and transmitting the stresses related to the elastic recoil pressure of the lung (= trans-alveolar pressure). The latter is defined as the difference between the pressure in the acinar airspaces and the pleural surface (Loring et al. 2016). In vivo, the pressure at the pleural surface compared to atmospheric pressure, is usually negative and is mainly based on the elastic recoil of the lung which has its foundation in the elastic fiber network and the surface tension (Fredberg and Kamm 2006; Wilson and Bachofen 1982). The axial system of elastic and collagen fibers originates from the walls of the conducting airways, enters the centers of the acini and contributes to the formation of alveolar entrance rings thereby surrounding the alveolar ducts. Hence, the alveolar duct as such does not have a wall of its own but is instead bounded by the entrance rings of the alveoli which include the elements of the axial network of connective tissue fibers. In other words, the axial fiber system coils the ductal airspaces (Fig. 4). Therefore, volume changes of the alveolar ducts result primarily in a deformation of the alveolar entrance rings and stretch of the axial fiber system. In this context, it has been observed that with decreasing lung volume, the diameter of alveolar entrance rings becomes smaller (Mercer et al. 1987). The alveolar walls, however, are formed by inter-alveolar septa which include above all an alveolar capillary network, diverse cell types and a minimum of stabilizing connective tissue elements. The axial system of connective tissue fibers, concentrated at the alveolar entrance rings, is connected with the peripheral system originating from the pleura by alveolar septal wall fibers which are located between the basal laminae of the alveolar epithelium and endothelium, corresponding to the thick side of the air-blood barrier. These fibers in the thick side represent the backbone of the inter-alveolar septa and transmit the distending forces which in case of homogenous stress distribution within the lung are generated by the pressure gradients between acinar airspace and pleural cavity (Mead et al. 1970) (Fig. 4). With this regard it has been shown that the volume densities of collagen and elastic fibers within the septa increase towards the free edge of the septa forming the "wall" of the alveolar duct (Mercer and Crapo 1990; Toshima et al. 2004). Elastic fibers have a linear stress–strain relationship over a wide range of deformation allowing a doubling of its baseline length (= 200% strain) so that these fibers contribute to elastic recoil and stabilization of lung parenchyma at lower lung volumes including also the range of normal breathing, usually defined as the volume spectrum between 40 and 80% of total lung capacity (TLC) (Suki et al. 2011; Yuan et al. 2000). Collagen fibers, on the other hand, have a more or less curly run at low lung volumes. As a consequence, collagen fibers become straight at larger lung volumes and are then characterized by a highly non-linear stress–strain relationship and high rigidity (Suki et al. 2005).</p><!><p>The stress-bearing elements of acinar airspaces. In a previous study (Knudsen et al. 2018), healthy rat lungs were fixed in vivo at airway opening pressure (Pao) of 1 (a) and 10 cm H2O (b). At low pressure, the alveolar ducts are narrow and the inter-alveolar septal walls are characterized by foldings and pleats. The septal walls protrude into the alveolar duct and are connected to the duct via the alveolar entrance. By drawing a straight line between the edges of the septal walls, alveolar and ductal airspaces were separated from each other (fine dashed lines). The axial network of elastic and collagen fibers is concentrated at the edges of alveolar septa and coils the alveolar duct. Here, this system is illustrated as springs spanning the alveolar duct (red springs). At low Pao (or lung volume), the elastic fibers are only slightly stretched (a, b). The fibers exert pulling forces on the alveolar edges/entrance rings in the direction of the ductal lumen (red arrows in a, c) and counteract the surface tension forces (green arrows in a, c) which would pull the septal wall away from the duct and result in a piling up and finally collapse of airspaces. At Pao = 10 cm H2O the alveolar duct is widened, the axial fiber system stretched (red springs in c and d). The forces which are responsible for inflation of the lung are related to the pressure gradient between the pleural space (PPl) and the alveolar space (Palv). The outward forces (FO) are transmitted to the fiber system in the septal walls and correspond here to the inward forces (Fi). Depiction is based on models of Wilson and Bachofen (1982) and Mead et al. (1970). Scale bar 100 µm</p><!><p>The stabilizing components of the inter-alveolar septa and, therefore, of the alveolar airspaces also include the alveolar epithelial basal lamina which is supposed to become also stress bearing at larger lung volumes (Maina and West 2006). An increase in volume of the alveoli can result in a stretch of the alveolar epithelial cells which are fixed to the basal lamina via cell-matrix adhesions (Tschumperlin and Margulies 1999). This represents an important difference compared to the alveolar duct which lacks a boundary with a cohesive epithelial lining. However, electron microscopic evaluations of the basal lamina at the thin side of the air-blood barrier of inter-alveolar septa demonstrated folding even at larger lung volumes (e.g. above 80% of TLC) indicating that at least in some areas the basal lamina (and the covering cells) are not entirely stretched (Bachofen et al. 1987). In the healthy lung, these very economically organized stabilizing systems of connective tissue elements described above, allow volumes to change during respiration with minimal effort and without interfering with the parenchyma's crucial gas exchanging function (Weibel et al. 1991; Weibel et al. 1992). In addition, it is likely that deformation of tissue components occurs without much strain of the alveolar epithelium in an intact lung during normal tidal breathing since the scaffold is stress bearing while surface tension in presence of an intact surfactant system is reduced at low lung volumes.</p><p>The basal lamina and the other components of the extracellular matrix form the scaffold to which the cellular components including alveolar epithelial cells, interstitial cells and endothelial cells are fixed via cell-matrix contacts such as focal adhesions. Although cells can execute deforming forces on surrounding extracellular matrix, it has been shown that lung mechanical properties hold only minor changes during the course of decellularization of the lung scaffold (Nonaka et al. 2014). In this context, it has been estimated that the cellular components of the inter-alveolar septa contribute little to overall lung mechanical properties such as elastic recoil and stiffness (Oeckler and Hubmayr 2008). Moreover, strain and stress acting on the extracellular matrix can be transmitted to the stress-bearing elements of the cells via cell-matrix and cell–cell contacts, such as the plasma membrane and the cytoskeleton, a mechanism which seems to be most relevant at larger lung volumes. These forces result in cellular deformation and have been estimated to amount to 5000 Pa at a focal adhesion. In this regard, Tschumperlin and Margulies reported an increase in the surface area of the epithelial basal lamina by 35% comparing lung volumes which corresponded to 42% and 100% of TLC (Tschumperlin and Margulies 1999). Therefore, at larger lung volumes the cytoskeleton and the plasma membrane of alveolar epithelial and endothelial cells might also become stress bearing due to their linkage with the basal lamina (Cong et al. 2017). Hence, these cell layers become susceptible for stress failure, e.g. upon injurious ventilation with increased tidal volumes which has been shown to result in ultrastructural signs of injury of endothelial and epithelial cells such as disruption of the cellular plasma membrane, blebbing and denudation of the basal lamina (Costello et al. 1992; Fu et al. 1992; Dreyfuss and Saumon 1998). Of note, according to these studies the type II alveolar epithelial cells appeared to be less damageable upon increased strain (Dreyfuss and Saumon 1998). Nevertheless, different mechanisms have been observed allowing cells to resist increased strain without failure. Increased strain at the cellular level might also occur under physiological conditions, e.g. upon deep inspirations, during physical exercise or sighs. Pleats of the plasma membrane unfold to accommodate to lateral tension forces. But even if breaks emerge in the plasma membrane the cells can repair these defects without undergoing cell death. Plasma membrane breaks with a diameter of less than 1 µm can be repaired by thermodynamic lateral flow of plasma membrane forming lipid bilayers in a Ca2+ independent way (Vlahakis et al. 2002; Cong et al. 2017). Moreover, stretching of cells has been demonstrated to activate the so-called deformation-induced lipid trafficking which includes the transfer of endogenous lipid vesicles to the corresponding break but also endocytosis of a disrupted plasma membrane and the formation of a membrane patch (Cong et al. 2017).</p><!><p>The pulmonary surfactant system also contributes to pulmonary mechanics and stabilizes alveoli, in particular at lower lung volumes (Bachofen and Schürch 2001). High surface tension at the air–liquid interface has important effects on the alveolar micro-architecture and leads to a reduction in alveolar surface area by causing collapsibility of airspaces. This is counteracted by the intra-alveolar surfactant through reduction of surface tension at end-expiration. The surface active surfactant layer at the air–liquid interface not only prevents end-expiratory alveolar collapse and edema formation (Possmayer et al. 2001) but also interfacial stress. Interfacial stress in the context of high surface tension is based on fluids which oscillate on the surface of the epithelium during breathing and deform e.g. type II alveolar epithelial cells via shear forces. Hence, interfacial stress alone can result in major dysfunction of type II alveolar epithelial cells as illustrated in in vitro test systems (Hobi et al. 2012; Ravasio et al. 2011).</p><p>In a healthy lung macro- and micromechanical properties are dominated by surface tension at the alveolar air–liquid interface at low lung volumes while at larger lung volumes extracellular matrix components become stress bearing and define lung structure and mechanics (Bachofen et al. 1987; Wilson and Bachofen 1982; Bachofen and Schürch 2001). The axial system of fibers surrounding the alveolar entrance rings forms a circular lattice which surrounds the alveolar duct and balances the surface tension at the alveolar air–liquid interface (Fig. 4). The stresses resulting from surface tension in the alveoli act in a way that the inter-alveolar septa would pile up in the corner of the alveolus so that alveolar surface area decreases. Thereby the elastic and collagen fibers which surround the alveolar ducts and concentrate at the edges of septal walls are stretched. Hence, these fibers balance the forces generated by surface tension at the alveolar air–liquid interface. The intra-alveolar surfactant system, in concert with the fiber network stabilizes the alveolar surface area available for gas exchange, which would otherwise decrease with increasing surface tension (Wilson and Bachofen 1982). Filling the lung with saline abrogates the air–liquid interface and, therefore, the surface tension. This has several effects on the microarchitecture such as a relatively narrow diameter of the alveolar ducts, irregular alveolar texture with bulging capillaries and missing pleats of septa in the alveolar corners (Gil et al. 1979). In the air-filled lung with preserved surface tension the alveolar walls are flat and the capillaries do not bulge into the airspace while alveolar ducts appear to be wider. Hence, compared to saline-filled lungs, air-filled lungs have reduced alveolar surface areas as measured by design-based stereology due to the mouldering effect of surface tension (Gil et al. 1979). Increased surface tension resulting e.g. from injury of surfactant-producing type II alveolar epithelial cells is linked with a further increase of alveolar duct diameter and loss of alveolar surface area, structural alterations which might easily be misinterpreted as pulmonary emphysema (Mouded et al. 2009). These observations indicate that the alveolar surface area is a direct function of surface tension: the higher the surface tension the lower the alveolar surface area at low to intermediate lung volumes up to 80% of TLC (Bachofen et al. 1979; Bachofen and Schürch 2001). Surface tension at the air–liquid interface is reduced by surfactant to nearly zero mN/m at the end of expiration so that alveoli are stabilized and the axial system of elastic fibers is only slightly stretched at lower lung volumes (Wilson and Bachofen 1982; Bachofen et al. 1987). This interplay of surface forces and the fiber system is of high importance since it implicates that during normal breathing, e.g. ranging from 40 to 80% of TLC, the cellular components of the inter-alveolar septa like the alveolar epithelium are protected from potentially harmful mechanical stresses. At larger lung volumes, e.g. above 80% of the TLC, surface tension increases but elastic as well as collagen fibers and the epithelial basal lamina are stretched so that they now stabilize and shape the airspaces and can potentially transmit stress and strain to cells. Since these tissue components become stress bearing and their properties define mechanical characteristics at larger lung volumes the surface tension can be neglected (Wilson and Bachofen 1982; Maina and West 2006). In this context, aging has been shown to result in airspace enlargement and reduction in elastic recoil of the lung. These observations have recently been suggested to be explainable by a pure age-related redistribution of elastic and collagen fibers within the inter-alveolar septa "away from the alveolar duct" emphasizing the relevance of the spatial distribution and orientation of the fiber networks (Subramaniam et al. 2017).</p><!><p>As outlined above, the structural design of the lung suggests that inter-alveolar septa are protected from overwhelming mechanical stress and strain, at least during tidal ventilation under healthy conditions (Bachofen and Schürch 2001). This aspect is of high importance since in vitro cell culture studies have illustrated that alveolar epithelial cells are susceptible to strain induced cell damage (Tschumperlin et al. 2000; Dolinay et al. 2017), an issue which is undoubtedly of relevance in vivo in the context of ventilation-induced lung injury (VILI) (Cong et al. 2017). However, our current knowledge regarding alveolar micromechanics and the mechanics of deformation of alveoli including their walls during ventilation is still very limited (Roan and Waters 2011). This is due to limitations in spatial and temporal resolution of available imaging techniques which do not allow direct visualization of acinar micromechanics, defined as architectural and functional alterations during the respiratory cycle. Indeed, a meaningful investigation taking cellular strain in the inter-alveolar septa also into account would require electron microscopic resolution. Based on quantitative assessments of the alveolar surface area or alveolar size at different stages of the pressure–volume curve it has been estimated that the deformation and thus the strain of the alveolus in linear dimensions during tidal breathing (usually defined between 40 to 80% of TLC), is in the range of 4% (Tschumperlin and Margulies 1999; Mercer et al. 1987) to 10% (Gil et al. 1979) and can in principle even increase to 20% and more (Gil et al. 1979; Mercer et al. 1987) in case the inspiratory reserve is exhausted, e.g. during deep sighs or physical exercise (Fredberg and Kamm 2006). Fredberg and Kamm emphasized the relevance of these estimations of alveolar strain. They calculated that during lifetime the alveolar structures must cope with breathing related alveolar strain for up to 109 strain cycles and further concluded: "By the standards of common engineering materials, these strains are extreme and would appear to call for tissue structures that are rather substantial" (Fredberg and Kamm 2006).</p><p>From this reasoning, the question arises how inter-alveolar septa cope with volume change related strain. In other words, what are the mechanisms by which alveoli, and above all the inter-alveolar septa adapt to respiratory cycle-related alveolar and septal strain without subjecting the alveolar epithelial cells to undue mechanical stress. Elegant studies over 30 years ago, using quasi-static conditions and vascular perfusion fixation of lung tissue at defined inspiratory and expiratory pressures during pressure–volume (PV) loops revealed that different mechanisms of septal wall deformation are involved and can result in individual alveolar volume changes during inspiration and expiration. Gil and co-workers discussed 4 mechanisms: (1) Recruitment/derecruitment of alveolar units; (2) Isotropic stretching and destretching with balloon-like changes of alveolar size; (3) Changes in alveolar shape (e.g. from dodecahedral to spherical and vice versa) which due to different geometry is linked with changes in alveolar size; and (4) Folding and unfolding of alveolar walls and accordion-like deformation which might resemble the folding and unfolding of a paper bag (Gil et al. 1979). These potential mechanisms were derived from ex vivo approaches of isolated and perfused lungs and evaluated the behaviour of the lung in a range of pressures at the airway opening from nearly zero (0.1 cm H2O) to up to 30 cm H2O (= 100% TLC) (Bachofen et al. 1987). These observations are, therefore, difficult to compare to the in vivo situation in which lung volume usually does not drop below residual volume. In the first respiratory cycles of a degassed lung, recruitment of complete alveoli during inspiration plays an important role and explains the increased hysteresis of the initial quasi-static PV loops (Bachofen et al. 1987; Bates and Irvin 2002; Carney et al. 1999). Model-based approaches using magnetic resonance imaging have provided evidence that alveolar recruitment during inspiration might be involved under physiological conditions in healthy humans (Hajari et al. 2012). However, these model predictions are in conflict with a number of other studies using direct visualization of alveoli or quantitative assessments of lung structure by means of design-based stereology to investigate alveolar micromechanics in healthy lungs. These studies did not find any evidence for intra-tidal recruitment and derecruitment of complete alveoli in a healthy lung under in vivo conditions above functional residual volume (Oldmixon and Hoppin 1991; Schiller et al. 2001; Pavone et al. 2007; Perlman et al. 2011; Sera et al. 2013; Knudsen et al. 2018; Lovric et al. 2017). In an interconnected network of alveolar walls the alveoli have been predicted to be very stable with balanced stresses acting on inter-alveolar septa as long as surface tension is reduced and harmonized between alveoli of different sizes (Fung 1975; Mead et al. 1970).</p><p>Figure 5 summarizes probable concepts of alveolar micromechanics during deflation in healthy lungs based on morphometric studies using fixed lung tissue at different stages of quasi-static PV loops (Gil et al. 1979; Bachofen et al. 1987; Oldmixon and Hoppin 1991; Tschumperlin and Margulies 1999; Knudsen et al. 2018). However, it has to be emphasised that the volume history of the lung as well as the conditions under which the lungs were studied, e.g. in vivo vs. ex vivo, are of high relevance so that conflicting results have been published. At rather low lung volumes (= decreasing airway opening pressures) the formation of pleats of inter-alveolar septal walls has been observed to occur quite frequently so that unfolding and folding of alveolar septal walls (but not complete alveolar units) appears to be of relevance for alveolar micromechanics, above all at low lung volumes. This has been shown by structural evaluations performed both under ex vivo (Tschumperlin and Margulies 1999) and in vivo conditions (Knudsen et al. 2018). Indeed, it has been demonstrated that tissue elastance during mechanical ventilation with PEEP of 1 cm H2O was significantly increased compared to mechanical ventilation with a PEEP of 5 cm H2O in healthy rat lungs. In other words, the lung gets stiffer if the pressure at the airway opening is reduced from 5 to 1 cm H2O. Based on the establishment of structure–function relationship, this increase in lung stiffness demonstrated a high correlation with the decrease in mean alveolar size which could be most likely attributed to the occurrence of pleats of inter-alveolar septa during deflation from 5 to 1 cm H2O as observed at the electron microscopic level (Knudsen et al. 2018; Tschumperlin and Margulies 1999). Of note, the number of open alveoli did not differ between lungs fixed in vivo (closed chest) at airway opening pressures of 1 and 5 cm H2O on expiration so that there was no evidence for derecruitment of alveolar units. Instead, a decline in surface area and alveolar volume could be linked with the formation of pleats/foldings of inter-alveolar septa. However, other investigators using a similar experimental setup (in vivo, closed chest) have questioned the relevance of this mechanism in a healthy lung since folding or crumpling of inter-alveolar septa were hardly seen between ranges of airway opening pressure of 3–16 cm H2O. However, the formation of pleats below this pressure range could not be ruled out (Oldmixon and Hoppin 1991). At intermediate to higher lung volumes reaching up to 100% of TLC (usually defined as lung volume at 30 cm H2O transpulmonary pressure), shape changes of alveolar airspaces and stretching/destretching of alveolar walls have been pointed out to occur by several investigators (Roan and Waters 2011). Gil and co-workers described changes in shape from a polyhedral to a more spherical configuration under ex vivo conditions (Gil et al. 1979). Tschumperlin and Margulies, in an ex vivo experimental setup, measured the surface area of the epithelial basal lamina during the deflation limb of a PV loop coming from 25 cm H2O and observed quite stable values in the range of approximately 80–40% of TLC while there was a considerable decrease in the range between 100 to 80% TLC. From these data, the authors suggested that stretching of alveolar epithelial cells is of high relevance above 80% of TLC while below 80% TLC, deformations without much change in the surface area of the basal lamina (and, therefore, stretch of alveolar epithelial cells) dominate micromechanics, e.g. unfolding/folding of septal walls or changes in shape (Tschumperlin and Margulies 1999). The overall strain in two dimensions of the basal lamina and, therefore, of the adhering epithelial and endothelial cells has been estimated to amount to 35% between 40 and 100% TLC while the range between 80 and 100% TLC contributed 80% of this overall strain (Tschumperlin and Margulies 1999). In this context, 25% strain in two dimensions of a monolayer of primary alveolar epithelial cells which might correspond to a permanent ventilation of the lung to lung volumes which are between 80% and 100% TLC, resulted in cellular injury such as endoplasmic reticulum stress and apoptosis (Dolinay et al. 2017). These findings illustrate that lung volume during ventilation does not need to exceed TLC to impose undue, potentially harmful strain on the alveolar epithelium.</p><!><p>Mechanisms of alveolar micromechanics during the deflation limb of a pressure–volume curve. Four mechanisms have been suggested (Gil et al. 1979): (1) Alveolar derecruitment, (2) Isotropic (balloon-like) destretching, (3) Shape changes and (4) Folding of alveolar walls. In vivo, the lung volume usually does not drop below the functional residual volume which is above the inferior infliction point. The occurrence of alveolar derecruitment is unlikely in this range of pressures but can be observed at very low lung volumes, e.g. with negative airway opening pressures. The other 3 mechanisms are likely to occur throughout the partial PV relationship above FRC although there is good evidence that folding dominates at lower volumes while destretching is most prominent at larger volumes. Shape changes have been described to be very dominant at intermediate volumes. This depiction is based on the observations of Gil et al. (1979), Bachofen et al. (1987), Tschumperlin and Margulies (1999) and Knudsen et al. (2018). The light microscopic images were taken from histological sections of a previous study (Knudsen et al. 2018). The description provided in this image is based on evaluations of lungs fixed at different pressures during the PV loop. Isolated phenomena occurring in the septal walls such as folding, shape change or stretching have never been observed in exactly the same alveolus directly. Scale bar 50 µm</p><!><p>Using ex vivo approaches, other investigators found evidence for recruitment and derecruitment in a range of airway opening pressures between 0 and 30 cm H2O (Gil and Weibel 1972; Bachofen et al. 1987) corresponding to the pressure–volume relationship over the complete range of TLC of an isolated lung. Hence, the volume history differed between those ex vivo studies which observed alveolar derecruitment and those which did not. During normal breathing and under in vivo conditions, the lung volume does usually not fall below the functional residual capacity and, therefore, the inferior infliction point of the PV loop (Salazar and Knowles 1964; Venegas et al. 1998), an aspect which is much different from ex vivo experiments mentioned above so that the occurrence of alveolar derecruitment at end-expiration is more than questionable in the range of physiological breathing. Using in vivo microscopy and the un-physiological situation of degassed but healthy dog lungs, Carney and co-workers observed alveolar recruitment till a lung volume corresponding to approximately 80% of TLC while those alveoli which were open were characterized by more or less stable individual volumes (Carney et al. 1999). Synchrotron refraction-enhanced computed tomography for direct visualization of lung acini in situ did not provide any evidence for alveolar recruitment in mice during inflation under quasi-static conditions and supported the concept of shape changes and accordion-like unfolding during inspiration (Sera et al. 2013).</p><p>Several ex vivo and in vivo studies also demonstrated that in the range of physiological breathing (e.g. transpulmonary pressure gradients below 10 cm H2O) volume changes predominantly take place in alveolar ducts while a smaller proportion of volume change takes place in the alveoli (Sera et al. 2013; Mercer et al. 1987; Knudsen et al. 2010, 2018). During spontaneous breathing it has been estimated from synchrotron X-ray imaging that 34% of tidal volume results in an increase of alveolar volume while the remaining 66% end up in the ductal or conducting airspaces during inspiration (Chang et al. 2015). These findings are roughly confirmed by in vivo microscopy showing that alveolar size increase only little during tidal ventilation (Schiller et al. 2001). Hence, tidal ventilation related volume changes in the alveolar compartment seem to be comparably low so that variations in linear dimensions of alveoli have been estimated to amount to 3–4%. At larger volumes, however, e.g. transpulmonary pressure gradients > 10 cm H2O, morphometric data suggest that 50% of volume changes during a PV loop occurs in alveolar ducts and alveoli each (Mercer et al. 1987).</p><!><p>The properties of the elastic and collagen fiber network as well as the surface tension at the air–liquid interface in the alveolar space determine alveolar micromechanics and ultimately the lung´s mechanical properties at the organ scale. The balance of stresses within the lung parenchyma is essential for more or less homogenous ventilation of distal airspaces during the respiratory cycle under healthy conditions. Hence, in their two-dimensional model of an interconnected network of elastic elements in the lung, Mead and co-workers emphasized the fundamental role of homogeneity in ventilation to avoid mechanical stress and strain of septal walls during breathing (Mead et al. 1970). Lung pathologies resulting from acute and chronic lung injury, however, severely interfere with homogenous ventilation. Heterogeneous ventilation occurs in the context of surfactant dysfunction, alveolar collapse, intra-alveolar edema formation, lung inflammation or focal fibrotic remodeling. These pathologies have been assigned the roles of stress concentrators which might impose potential harmful stresses and strains on surrounding tissue during respiration (Mead et al. 1970; Makiyama et al. 2014). The spring model proposed by Mead et al. (1970) was based on the interdependence of hexagonally shaped alveoli which share inter-alveolar septal walls. As long as ventilation is homogenous the intra-pulmonary stresses are balanced, protecting lung units (= alveoli) from end-expiratory collapse and inspiratory overdistension. Inhomogeneity in ventilation of alveoli is linked to severe abnormalities in alveolar micromechanics due to end-expiratory collapse combined with enormous stress and intra-tidal deformation (= strain) of neighbouring alveoli including inter-alveolar septa (Mead et al. 1970; Wilson and Bachofen 1982; Makiyama et al. 2014). These predictions implicate that within an injured lung alveolar micromechanics can be locally severely altered and might become an independent trigger of lung injury progression.</p><p>Surfactant dysfunction results for example from injury of type II alveolar epithelial cells and predates the development of further signs of acute lung injury (ALI) such as formation of alveolar edema, increase in inflammatory markers or even remodelling of lung tissue (Steffen et al. 2017; Lutz et al. 2015; Lopez-Rodriguez et al. 2016; Ikegami et al. 2005; Ochs et al. 1999). Surfactant dysfunction is detrimental for the lung since it counteracts the goal of homogenous ventilation and even distribution of stresses. In addition to increases in surface tension and alveolar instability occuring at low lung volumes, differences in surface tension between different alveoli can occur resulting in intra-acinar-pressure gradients which is the basis of the phenomenon of alveolar pendelluft (see below) (Tabuchi et al. 2016). Conversely, surfactant dysfunction may also be a direct consequence of abnormal alveolar micromechanics. Mechanical but also spontaneous ventilation can be associated with undue alveolar strain and large changes of the alveolar surface area which has been shown to be responsible for alterations of the intracellular and intra-alveolar surfactant system and finally acute lung injury (Milos et al. 2017; Veldhuizen et al. 2002; Mascheroni et al. 1988). Mechanical ventilation with high tidal volumes to induce abnormally high strain of the lung results in a conversion of biophysically active large aggregates of surfactant to inactive small aggregates of surfactant within the airspaces (Veldhuizen et al. 2002). In light of these observations ALI or acute respiratory distress syndrome (ARDS) has been suggested to be primarily a disease of ventilation and altered micromechanics and atelectasis was discussed to play a central part in this pathophysiology (Albert 2012).</p><p>The relevance of abnormal alveolar micromechanics as an independent trigger of lung injury during mechanical ventilation has been well described within the last decades. In this context, the dynamic stresses and strains of lung parenchyma induced by cyclic closure and reopening of alveoli during the respiratory cycle (atelectrauma) (Muscedere et al. 1994; Steinberg et al. 2004; Chu et al. 2004) as well as overdistension of patent alveoli (volutrauma), e.g. due to heterogeneous ventilation (Retamal et al. 2014) have been shown to represent important aggravating mechanisms in ARDS. In a healthy lung the combination of volutrauma with increased pulmonary strain and atelectrauma are commonly both necessary to induce acute lung injury while in pre-injured lungs one of these factors usually suffices to aggravate lung injury (Seah et al. 2011; Nieman et al. 2015). Atelectrauma, sometimes also referred to as low volume trauma results from cyclic opening and closing of distal airspaces or even alveoli and has been shown to augment acute lung injury (Muscedere et al. 1994; Steinberg et al. 2004). During the time course of progressive VILI, increasing alveolar instability with cyclic intratidal alveolar recruitment and derecruitment (R/D) as well as edema formation has been linked to degradation of lung mechanical properties (Smith et al. 2013, 2017). In the context of atelectrauma due to surfactant dysfunction and accumulation of intra-alveolar edema, derecruited alveoli are forced to be reopened during inspiration by an entering air bubble. Based on modelling approaches including computational simulations and in vitro validation, the event of an airway reopening exerts harmful deforming stresses on epithelial cells which result during the time course of airway reopening in pulling, pushing and shearing-off of affected cells (Bilek et al. 2003; Kay et al. 2004). Volutrauma on the other hand results in overdistension of inter-alveolar septa at larger alveolar volumes in situations where stresses of the scaffold are transmitted to cells via cell-matrix contacts which then undergo stress failure. These mechanisms were recently discussed in detail in the context of both acute lung injury and pulmonary fibrosis (Cong et al. 2017). The occurrence of cellular stress failure has been documented in particular at an ultrastructural level for alveolar epithelial and endothelial cells (Dreyfuss and Saumon 1998).</p><p>The effects of ALI on alveolar micromechanics, however, go beyond the mechanisms of alveolar R/D (Gatto and Fluck 2004; Schiller et al. 2001) and alveolar overdistension (Cong et al. 2017). Using in vivo real time imaging techniques such as optical coherence tomography and intravital microscopy investigators described several phenomena which are summarized as asynchronous alveolar dynamics which occur independent from alveolar R/D (Mertens et al. 2009; Tabuchi et al. 2016). Under healthy conditions alveoli and alveolar clusters expand in synchrony with the ventilator. This means that during end-expiratory and end-inspiratory plateau phases, characterized by zero flow at the airway opening, no volume changes can be observed. After induction of acute lung injury, Mertens and co-workers (2009) observed a reduced ventilation-associated expansion in a cohort of alveoli while other alveoli were enlarged. This observation was in line with the concept of re-distribution of air within acinar airspaces and heterogeneous ventilation (Mertens et al. 2009; Schirrmann et al. 2010). Using different mouse models of ALI, Tabuchi et al. (2016) succeeded in tracking alveolar micromechanical changes during the development of ALI. Asynchronies between ventilator and alveolar dynamics were observed as early as 10 min after induction of lung injury and corresponded to the phenomenon of alveolar pendelluft. Alveolar pendelluft has been defined as an asynchronous alveolar micromechanical pattern: A cohort of alveoli increased individual volumes during the end-expiratory plateau phase while during end-inspiratory plateau pressures these alveoli decreased in alveolar volumes (Tabuchi et al. 2016). The fact that alveoli changed their size during plateau phases suggested the existence of intra-parenchymatous pressure gradients which resulted in redistribution of air. With disease progression further phenomena of alveolar asynchronies emerged which included alveolar stunning (lack of volume changes during ventilation) and inverse alveolar ventilation (volume decrease during inspiration, volume increase during expiration) which were both linked to impaired oxygenation of blood in adjoining septal wall capillaries (Tabuchi et al. 2016). Using confocal microscopy to visualize alveolar dynamics in an ex vivo rat lung perfusion model Perlman et al. investigated the alveolar interdependence between edema filled alveoli and air-filled alveoli (Perlman et al. 2011). Based on these observations, edema filled alveoli exert tethering forces on air-filled alveoli since the forces acting on the inter-alveolar septa were no longer balanced, and high surface tension dominates in the edema filled alveolus. Hence, due to the occurrence of edema filled alveoli the air-filled alveoli become prone to overdistension and, therefore, volutrauma as a result of the interdependence of alveoli which share inter-alveolar septa (Perlman et al. 2011; Wu et al. 2014). These observations provided convincing experimental evidence with respect to the model-based predictions of Mead and coworkers (1970).</p><!><p>It is evident that both the connective tissue fiber system and the surfactant system are essential and interdependent components of alveolar micromechanics (Weibel and Bachofen 1997). For alveolar physiology and pathophysiology, it is not only relevant what is going on in the alveolar wall—but also what is going on on its surface. In that sense, surfactant is certainly more than "mere paint on the alveolar wall" (Nicholas 1996). Nevertheless, despite considerable efforts over the last decades, we are still far from a comprehensive understanding of alveolar micromechanics.</p><p>The development of real-time in vivo microscopy techniques have undoubtedly advanced the understanding of alveolar micromechanics during mechanical ventilation under physiological and pathophysiological conditions as outlined above. Due to technical constraints, the access to the alveoli is, however, limited so that most studies using in vivo microscopy analyzed subpleurally located alveoli which might not be representative of the whole population of alveoli within the lung. In addition, while in vivo microscopy allows the analyses of alveolar volume changes, the mechanisms of deformation occurring within the inter-alveolar septa cannot be evaluated since the resolution is not sufficient for this purpose. Hence, if the volume change within an alveolus results in a real stretching of cells in the inter-alveolar septa or is linked with an unfolding of pleats and shape changes remains unanswered by these investigations. In-depth analyses to answer these questions would require electron microscopic resolution (for review, see Ochs et al. 2016) to visualize the basal lamina (and the epithelial cells) which might be stretched so that its surface area increases or simply unfold as ultrastructural investigations from fixed lungs suggest. Until now, however, there is no direct visual evidence whether these mechanisms are really involved during spontaneous or mechanical ventilation in a living subject.</p><p>Some of these limitations might be addressed by computational simulations (Burrowes et al. 2018). Computational modelling has the potential to advance our understanding of the acinar micromechanics and alveolar interdependence including the effects down to the cellular level. Measurements of lung mechanical properties such as elastance and compliance at the organ level, so-called macromechanics, reflect mechanisms which occur at the alveolar level such as alveolar R/D or overdistension (Knudsen et al. 2018). With this regard, computational modelling of alveolar micromechanics has been employed to investigate pathologic alterations in R/D in mice with VILI during injury progression (Smith and Bates 2013) to understand the injurious mechanical forces (Hamlington et al. 2016), and to predict the open fraction of respiratory units and alveolar distension as a function of airway pressure and disease severity (Smith et al. 2013, 2015; Knudsen et al. 2018). These empirical models of alveolar R/D are based on the assumption that at a microscopic level the lung is composed of respiratory units (e.g. alveoli) ventilated in parallel. Each alveolus has a certain elastance and viscoelasticity and the collectivity of all alveoli defines the mechanical properties at the organ scale. During the respiratory cycle, the transpulmonary pressure increases during inspiration and decreases during expiration. Depending on the surface tension and "stickiness" of the fluid alveolar lining, alveoli can derecruit (i.e. collapse) if the pressure falls below a certain limit, the alveolar closing pressure. During inspiration, however, alveoli can again be recruited in case the transpulmonary pressure exceeds a certain alveolar opening pressure (Bates and Irvin 2002). These model-assumptions were advanced in recent years and have been demonstrated to be able to reproduce empirically measured lung mechanical properties as well as structural data in several studies (Smith et al. 2013; Knudsen et al. 2018).</p><p>Spring network models were developed for simulations of the mechanics of lung parenchyma. In its original description, the spring model was composed of a two-dimensional network of springs (i.e. inter-alveolar septa) forming hexagonal (i.e. alveolar) spaces (Mead et al. 1970). This model was further developed and applied to simulate the time course and lung mechanical impairment of pulmonary fibrosis as well as pulmonary emphysema including response to lung volume reduction surgery (Bates et al. 2007; Mishima et al. 1999; Mondoñedo and Suki 2017). In addition, spring models were used to understand aspects of alveolar and alveolar-airway interdependence (Ma and Bates 2012, 2014; Ma et al. 2013a, b, 2015; Mead et al. 1970; Makiyama et al. 2014; Bates et al. 2007). It has long been understood that alveolar interdependence plays an important role in the determining strain at the level of individual septa (Mead et al. 1970; Perlman et al. 2011). However, the influence of surfactant dysfunction and derecruitment on the septal strain distribution is not well described, in particular during disease progression. In this regard, it has been proposed that surfactant dysfunction, R/D dynamics, and alveolar interdependence with increased strain play critical roles in the pathogenesis of fibrotic lung disease (Todd et al. 2015; Lopez-Rodriguez et al. 2017; Knudsen et al. 2017; Cong et al. 2017). Recent imaging studies in lungs suffering from idiopathic pulmonary fibrosis (IPF) provided evidence for instability of distal airspaces in regions of the lung which were not yet subject to fibrotic remodeling (Mai et al. 2017; Petroulia et al. 2018). Mai and co-workers used micro-computed tomography of IPF lung explants and found microatelectases in areas close to fibrotic tissue but not yet affected by fibrosis. Based on these observations microatelectases might be attributed the roles of stress concentrators which trigger disease progression including lung injury, alveolar collapse, fibrotic remodeling and collapse induration (Burkhardt 1989; Knudsen et al. 2017).</p>

PubMed Open Access

The FGF2‐induced tanycyte proliferation involves a connexin 43 hemichannel/purinergic‐dependent pathway

AbstractIn the adult hypothalamus, the neuronal precursor role is attributed to the radial glia‐like cells that line the third‐ventricle (3V) wall called tanycytes. Under nutritional cues, including hypercaloric diets, tanycytes proliferate and differentiate into mature neurons that moderate body weight, suggesting that hypothalamic neurogenesis is an adaptive mechanism in response to metabolic changes. Previous studies have shown that the tanycyte glucosensing mechanism depends on connexin‐43 hemichannels (Cx43 HCs), purine release, and increased intracellular free calcium ion concentration [(Ca2+)i] mediated by purinergic P2Y receptors. Since, Fibroblast Growth Factor 2 (FGF2) causes similar purinergic events in other cell types, we hypothesize that this pathway can be also activated by FGF2 in tanycytes to promote their proliferation. Here, we used bromodeoxyuridine (BrdU) incorporation to evaluate if FGF2‐induced tanycyte cell division is sensitive to Cx43 HC inhibition in vitro and in vivo. Immunocytochemical analyses showed that cultured tanycytes maintain the expression of in situ markers. After FGF2 exposure, tanycytic Cx43 HCs opened, enabling release of ATP to the extracellular milieu. Moreover, application of external ATP was enough to induce their cell division, which could be suppressed by Cx43 HC or P2Y1‐receptor inhibitors. Similarly, in vivo experiments performed on rats by continuous infusion of FGF2 and a Cx43 HC inhibitor into the 3V, demonstrated that FGF2‐induced β‐tanycyte proliferation is sensitive to Cx43 HC blockade. Thus, FGF2 induced Cx43 HC opening, triggered purinergic signaling, and increased β‐tanycytes proliferation, highlighting some of the molecular mechanisms involved in the cell division response of tanycyte. This article has an Editorial Highlight see https://doi.org/10.1111/jnc.15218.

the_fgf2‐induced_tanycyte_proliferation_involves_a_connexin_43_hemichannel/purinergic‐dependent_path

<!>INTRODUCTION<!>Ethics statement<!>Primary cultures of PN1 tanycytes<!>Total RNA extraction<!>Reverse transcription (RT) of total RNA<!>Amplification of cDNA by PCR<!><!>Agarose gel electrophoresis<!>Ethidium uptake and fluorescence imaging<!>Extracellular ATP measurements<!>Immunofluorescence<!>Protein immunodetection<!>Incorporation of BrdU in cell cultures<!>Osmotic pumps preparation<!>Stereotaxic cannula implantation<!>Statistics and image processing<!>In vitro tanycytes retain the expression of undifferentiation markers, Cx43, purinergic receptors, FGFs and their receptor<!><!>In vitro tanycytes retain the expression of undifferentiation markers, Cx43, purinergic receptors, FGFs and their receptor<!><!>Cx43 and P2Y1R are necessary for in vitro proliferation of tanycytes induced by FGF2<!><!>FGF2 increases hemichannel activity and ATP release in cultured tanycytes<!><!>FGF2 increases hemichannel activity and ATP release in cultured tanycytes<!>Extracellular ATP exerts a mitogenic effect on cultured tanycytes<!><!>Extracellular ATP exerts a mitogenic effect on cultured tanycytes<!>FGF2 positively modulates connexin43 expression in tanycytes at 7 hr<!><!>FGF2 positively modulates connexin43 expression in tanycytes at 7 hr<!>Gap27 inhibits FGF2‐induced proliferation of β‐tanycytes in vivo<!><!>Gap27 inhibits FGF2‐induced proliferation of β‐tanycytes in vivo<!>DISCUSSION<!><!>DISCUSSION<!>CONFLICT OF INTEREST<!>AUTHORS’ CONTRIBUTIONS<!>

<p>Recabal A , Fernández P , López S , et al. The FGF2‐induced tanycyte proliferation involves a connexin 43 hemichannel/purinergic‐dependent pathway. J. Neurochem. 2021;156:182–199. 10.1111/jnc.15188 32936929</p><p>This article has an Editorial Highlight see https://doi.org/10.1111/jnc.15218.</p><p>third ventricle</p><p>intracellular free calcium ion concentration</p><p>arcuate nucleus</p><p>adenosine triphosphate</p><p>bromodeoxyuridine</p><p>cerebrospinal fluid</p><p>connexin 43 hemichannels</p><p>connexin 43</p><p>ecto‐nucleoside triphosphate diphosphohydrolase 2</p><p>Extracellular signal‐regulated protein kinases 1 and 2</p><p>ethidium bromide</p><p>fibroblast Growth Factor 1</p><p>fibroblast Growth Factor 2</p><p>fibroblast growth factor receptor 1</p><p>gamma‐aminobutyric acid</p><p>lanthanum ion</p><p>median eminence</p><p>N6‐methyl‐2'‐deoxyadenosine‐3',5'‐bisphosphate</p><p>neuronal precursors</p><p>neuropeptide Y</p><p>Pannexin1</p><p>pro‐opiomelanocortin</p><p>Research Resource Identifier</p><p>sex determining region Y‐box 2</p><p>subventricular zone</p><p>wheat germ agglutinin</p><!><p>Tanycytes are specialized hypothalamic ependymal cells lining the lateral walls and floor of the third ventricle (3V) and are classified as reminiscent radial glia as well owing to their highly polarized morphology (Rodríguez et al., 2005). Their apical poles contact the cerebrospinal fluid (CSF), while some of their basal extensions project to the circumventricular organ median eminence (ME) or different hypothalamic nuclei, such as the arcuate nucleus (ARC) (Flament‐Durand & Brion, 1985), where the neurons responsible for energy balance and feeding behavior are located. Moreover, tanycytes represent a pool of neuronal precursor cells that proliferate and differentiate into functional orexigenic and anorexigenic neurons (Haan et al., 2013; Hajihosseini et al., 2008; Robins et al., 2013; Xu et al., 2005) after dietary exposure to high fat (Bless et al., 2016; Lee & Blackshaw, 2012). This early response adds new players to the neuronal network regulating feeding behavior and restoring energy balance (Gouaze et al., 2013; Kokoeva et al., 2005), prior to the pre‐obesity and pre‐diabetes activated inflammatory microenvironment (Li et al, 2012; Moraes et al., 2009). Furthermore, it has been suggested that tanycytes act as neuro‐modulating cells, regulating the availability and access of satiety‐ and hunger‐inducing hormones from peripheral tissue to ARC neurons (Balland et al., 2014; Collden et al., 2015; Langlet et al., 2013; Prevot, 2002). In addition, they express the molecular machinery for detecting nutrients, such as glucose, and for signaling to the adjacent neurons (Barahona et al., 2018; Cortés‐Campos et al., 2011; Elizondo‐Vega et al., 2016; García et al., 2003; Millán et al., 2010), which include connexin 43 hemichannels (Cx43 HCs) (Orellana et al., 2012). The role of Cx43 HCs and purinergic signaling on the glucosensing potential of tanycytes has been demonstrated in living hypothalamic slices (Frayling et al., 2011), while the specific mechanism through which a rapid (within min) increase in (Ca2+)i occurs has been examined in primary cultures of tanycytes using pharmacological approaches. It sequentially consists of glucose transport and glycolytic metabolism, the controlled ATP release to the extracellular milieu mediated mainly by Cx43 HCs, and subsequent activation of P2Y receptors (Orellana et al., 2012). In HeLa cells, spinal astrocytes and glioma cells, the activity of Cx43 HCs have been shown to increase slowly (within hours) after a mitogen stimulus, such as Fibroblast Grow Factor 1 (Garré et al., 2010; Schalper et al., 2008) and 2 (De Vuyst et al., 2007) (FGF1 and 2, respectively), also triggering ATP release and purinergic signaling activation. FGF2 has been implicated in adult neurogenesis in the classic neurogenic niches such as the subventricular zone (SVZ) and the subgranular zone (SGZ) of the hippocampal dentate gyrus (Woodbury & Ikezu, 2014). Indeed, a combination of FGF2 and EGF, has been widely used for the maintenance and proliferation of neurospheres from different neurogenic niches (Kano et al., 2019; Furube et al., 2020). Extensive in situ evidence informed the expression of FGF2 receptor, FGFR1, in the ventral tanycyte domain (Kaminskas et al., 2019; Samms et al., 2015) and highlighted its importance in the control of body weight and food intake (Samms et al., 2015). Moreover, tanycyte cells divide under FGF2 stimulus (Robins et al., 2013; Xu et al., 2005). Thus, it is feasible to hypothesize that FGF2 activates a purineric signaling in which proliferation of tanycytes could result from a combined FGF2/Cx43 HCs/purinergic interaction.</p><p>Caloric restriction and a high‐fat diet have been identified as metabolic stimuli that influence the proliferation of adult hypothalamic neuronal precursors (NPs), but knowledge of the underlaying mechanisms are currently scarce (Bless et al., 2016; Chaker et al., 2016; Nascimento et al., 2016). It is likely that, under hypercaloric stimuli, tanycytes signal through multiple molecular pathways that may be redundant with those involved in metabolites detection. Here we show that in vitro tanycytes trigger Cx43 HCs opening and ATP release after long‐term exposure to FGF2 mitogen. In addition, a moderate increase in ATP concentration in the extracellular media was enough to induce cell division that can be suppressed by Cx43 HC and P2Y1 receptor inhibition. Moreover, in vivo experiments performed by directly infusion of a Cx43 HC blocker to the 3V showed a decline in the FGF2‐induced ventral tanycyte proliferation. These results suggest an essential role of Cx43 HCs and purinergic signaling in tanycyte self‐renewal.</p><!><p>All the studies performed on rats were approved and reviewed by the Animal Ethics Committee of the National Commission of Chile for Scientific and Technological studies (CONICYT, Fondecyt N°1180871), by the ethics committee of the Faculty of Biological Sciences and by the committee on Ethics, Care and Use of Animals of the University of Concepción, Chile. The animals were handled according to the guidelines of the National Institute of Health for the care and use of animals, USA. Male Sprague–Dawley rats (120–280 g) obtained from Charles River original source (RRID _10395233) were housed on a 12 hr light/dark cycle with food and water ad libitum. Each cage housed at least two adult animals and did not exceed the five animals. The hypothalamus of only one female adult rat was used for immunohistochemistry. Pregnant rats were housed in pairs. Each T25 flask of primary cell culture was performed of at least three and up to six hypothalami from post‐natal (PN) 1 rat Sprague–Dawley pups. In vivo studies were performed using a total of 15 Sprague Dawley adult male rats, weighing 120–180 g. Experimental procedure was incorporated in Figure 7. Animals were arbitrarily assigned to experimental groups; no randomization was performed. The study was not pre‐registered; all relevant information is provided in the manuscript and custom‐made materials will be provided upon request.</p><!><p>Primary rat tanycyte cultures were performed according to the method described previously (García et al., 2003; Orellana et al., 2012). PN 1 rats were quickly decapitated, the brain was removed, and the 3V boundaries were dissected on ice. The samples were incubated with 0.25% trypsin–0.2% EDTA (w/v) (Thermo Fisher Scientific Inc.; Cat# 25200114) for 20 min at 37°C, before being transferred to MEM culture medium (Thermo Fisher Scientific Inc.; Cat# 61100087), supplemented with 10% fetal bovine serum (FBS) (Thermo Fisher Scientific Inc.; Cat# 12484028), 2 mM L‐glutamine, 100 U/mL penicillin, and 100 mg/ml streptomycin (Thermo Fisher Scientific Inc.; Cat# 15140163). The samples were disaggregated, and the cells were seeded in T25 culture flasks covered with 0.2 mg/ml poly‐l‐lysine (Sigma‐Aldrich; Cat# P6407) at a density of 3 million cells per flask. The cells were kept in the same bottle for 2 weeks, and the medium was renewed every other day. For subsequent experiments, monolayer‐grown tanycytes were rinsed twice with 0.1 M phosphate buffer (PBS; in mM: 137 NaCl, 2.7 KCl, 10 Na2HPO4, and 2 KH2PO4) at pH 7.4 and treated with 0.25% trypsin–0.2% EDTA (w/v) for 3 min at 37°C. The cells were disaggregated and reseeded in 6, 12, and 24‐well plates, previously covered with 0.01% poly‐l‐lysine (w/v), at a cell density of 500,000, 250,000, and 800 cells per well, respectively. Only cells in passage one were used. Before all the experiments, tanycytes were cultivated for 24  hr in medium without FBS to prevent inhibition of ATP release and activation of purinergic signaling (Lin et al., 2007).</p><!><p>Total RNA was obtained from samples of total brain, hypothalamus, striated muscle, and primary cultures of tanycytes. The RNA was extracted according to the guanidinum thiocyanate–phenol–chloroform method, homogenizing the samples in 500 ml Trizol® (Life Technology; Cat# 15596018) for 10 min and incubating them 5 min at room temperature (20ºC). Then, the samples were treated with 200 ml chloroform, vigorously shaken for 15 s, and incubated at room temperature (20ºC) for 3 min. The samples were centrifuged at 12,000 g for 15 min at 4°C to separate the phases. The aqueous phase was recovered, and 300 µl of isopropanol was added to each sample, incubated for 10 min at room temperature (20ºC), and centrifuged at 12,000 g for 5 min at 4°C. The supernatant was discarded, and the pellet was washed twice with 70% ethanol, spinning at 12,000 g for 10 min each time. Finally, the pellet was resuspended in 10 µl of RNase‐free water and quantified by measuring its absorbance at 260 nm and its purity according to the 260/280 ratio.</p><!><p>DNA synthesis was performed using RevertAid® H Minus M‐MuLV Reverse Transcriptase Enzyme (Thermo Fisher Scientific Inc.; Cat# EP0451). Prior to synthesis, 2 µg of the total RNA samples were treated with DNase (Fisher Scientific Inc.; Cat# AM2238). For a final volume of 20 µl, the above mixture was incubated with 0.5 µg of oligo‐dT, denatured at 70°C for 5 min, and placed on ice for 2 min. Subsequently, the transcription buffer consisting of 50 mM Tris‐HCl (in mM: 50 KCl, 4 MgCl2, 10 DTT) at pH 8.3, the mixture of dNTPs (1 mM each), and 20 U of the ribonuclease inhibitor (Thermo Fisher Scientific Inc.; Cat# 10,777,019) were added, incubating for 5 min at 37°C. Next, 200 U of RevertAid® H Minus M‐MuLV (Thermo Fisher Scientific Inc.; Cat# EP0451) reverse transcriptase enzyme was added and incubated for 1 hr at 42°C followed by incubation at 70°C for 10 min. Negative controls for sample amplification were treated with the same protocol, but without adding oligo‐dT or reverse transcriptase enzyme to the mixture.</p><!><p>Amplification of cDNA was performed in an Eppendorf® Mastercycler® Nexus Thermal Cycler (Merck KGaA, Darmstadt, Germany) in a mixture of 10 mM Tris‐HCl at pH 8.8 containing 50 mM KCl, 1.5 mM MgCl2, 0.2 mM of each dNTP, 0.2 μM of each set of specific primers (Table 1), 0.31 U Taq DNA polymerase (Thermo Fisher Scientific Inc.; Cat# 10342053), and 1 μL of the reverse transcription product, in a final volume of 11.5 μL. The incubation program consisted of 95°C for 5 min, followed by 35 cycles of: denaturation 95°C for 30 s, annealing 55°C for 30 s, and 72°C for 30–40 s, and a final extension of 72°C for 7 min. The cDNAs synthesis was tested using specific β‐actin primers.</p><!><p>Set of primers used for RT‐PCR</p><!><p>The separation of DNA fragments was performed using 1% agarose gels. The buffer used for the electrophoresis was TAE (Tris‐acetic acid EDTA; 40 mM Tris‐HCl, 30 mM acetic acid and 1 mM EDTA; pH 7.6). Agarose gels were prepared with TAE containing 0.5 µg/ml ethidium bromide (Genesse Scientific, Inc., San Diego, California, USA; Cat# 20‐276). A 100‐bp DNA ladder (GeneRuler, Thermo Scientific; Cat# SM0242) was used as the molecular weight marker. Transilluminator and photodocumentation kit (VilberLourmat PMM9A, France) were used to visualize the bands.</p><!><p>Tanycytes were transferred to poly‐l‐lysine ‐covered coverslips, grown to at least 80% confluency in six‐well plates and treated for 7 hr at 37°C with a final concentration of 20 ng/ml FGF2 (Sigma Aldrich; Cat# SRP4037) conjugated with heparin 10 IU/mL to maintain a ratio of 5 IU heparin for every 10 ng of FGF2, as it has been described by Schalper et al., (2008). Gap27 (Genscript, Piscataway; (SRPTEKTIFFI, second extracellular loop domain of Cx43), a selective Cx43 HC inhibitor, was added at a final concentration of 200 µM at the same time as FGF2. All compounds were added in the MEM culture medium supplemented with L‐glutamine, penicillin, and streptomycin, but without FBS. Each supplementation had a half‐hour lag (time that each registration lasted), starting at 9 a.m. and recording at 5 p.m. For fluorescence detection, cells were washed twice with 0.1 M PBS, pH 7.4, before applying the registration solution (154 mM NaCl, 5.4 mM KCl, 2.3 mM CaCl2, and 5 mM HEPES, pH 7.4), containing 5 µM ethidium bromide (Etd+, Sigma Aldrich; Cat# E7637). Cells on the coverslips were mounted under a live cell microscope (Nikon, Eclipse Ti‐FL model, Japan) and were recorded every 30 s with a 40x objective. At least 10 cell nuclei per coverslip were defined as regions of interest (ROIs), and the average change in their fluorescence intensity was measured over time. After approximately 30 min, 200 µM lanthanum ion (La3+, LaCl3x7H2O, Sigma Aldrich; Cat# L4131), which is a non‐selective inhibitor of Cx HCs as a control of cell vitality, was added.</p><!><p>Tanycytes were seeded on 12‐well plates, previously covered with poly‐l‐lysine and cultured without FBS for 24 hr at 37°C, before being treated with a final concentration of 20 ng/ml FGF2, 10 IU/mL heparin, and/or 200 µM Gap27. The final volume in the culture medium was 300 µl for each well. All the experiments started at 10 a.m., and after 7 hr (5 p.m.), the concentration of ATP contained in 100 µM was measured through the luciferin/luciferase bioluminescence assay (ATP Bioluminescence Assay Kit CLS II, Sigma‐Aldrich; Cat# 11699695001). The amount of ATP in each sample was calculated from a standard curve, whose concentration range includes 10, 20, 40, 60, 80, and 100 pM. CPS (counts per second) values that exceeded the curve values were not considered. ATP concentration was normalized to the total protein concentration of its respective sample, using the Bradford reagent (Bio‐Rad Laboratories; Cat# 5000006EDU) and Optizen Pop spectrophotometer (Comercial Rafer SL, Zaragoza. Spain).</p><!><p>For immunohistochemistry, the rats were anesthetized with 200 µl of intraperitoneal ketamine:xylazine (90 mg/kg‐10 mg/kg) until they exhibited no reflections. They were intravascularly perfused with 200 ml ice cold PBS and then 200 ml of 4% PFA. The cervical were dislocated, the brains were removed, the hypothalami were dissected and slices of 200 µm thickness were performed using vibratome (Leica VT1200, Wetzlar, Alemania). For immunocytochemistry, once treated, cells were washed once with 0.1 M PBS and fixed with 4% (w/v) paraformaldehyde dissolved in PBS for 30 min at room temperature (20ºC). The wheat germ agglutinin procedure was considered a pre‐fixation step that consisted of bathing the cells for 30 s with 1 µg/ml WGA (Sigma‐Aldrich; Cat# L9640) that was dissolved in cold PBS and then washed twice with the saline solution. The BrdU detection was performed with an additional step that allowed DNA denaturation through 1 M HCl at 45°C for 30 min, before blocking. The acid was neutralized by rinsing the slices or cells three times for 10 min with Tris phosphate buffer (84 mM Na2HPO4, 35 mM KH2PO4, 120 mM NaCl, 10 mM Tris, pH 7.8). The following primary antibodies dissolved in Tris phosphate with 0.1% Triton X‐100 were incubated overnight at room temperature (20ºC): sheep anti‐BrdU (1:2,000, Abcam; RRID AB_302659), rabbit anti‐Cx43 (1:200, BD Biosciences Franklin Lakes; RRID:AB_397473) previously described (Schalper et al., 2008), mouse anti‐Nestin (1:1,500, Abcam; RIDD AB_305313), mouse anti‐vimentin (1:400, DAKO; RIDD AB_2827759), goat anti‐SOX2 (1:400, Santa Cruz; RIDD AB_2286684), and rabbit anti‐WGA (1:1,000, Sigma Aldrich; RIDD AB_261669). The samples were washed three times with the Tris phosphate buffer and incubated for 2 hrwith the respective secondary antibodies bound to fluorophores and TOPRO (Thermo Fisher; Cat# T3605) in a final dilution of 1:200 and 1:1,000, respectively. Finally, the coverslips were mounted with FluoromontTM Aqueous Mounting Medium (Sigma Aldrich; Cat# F4680). The samples were analyzed by confocal microscopy (Confocal Spectral Microscope, model LSM780 NLO Zeiss, Centro de Microscopía Avanzada, CMA‐Bio Bio).</p><!><p>Total protein extracts were obtained from samples of heart, liver, and primary cultures of tanycytes. The samples were homogenized in protease inhibitor (Thomas Scientific, ROCHE cOmplete™; Cat# 12352200) and sonicated three times on ice at 300 W in the case of tissues or resuspended in 40 µl of lysis buffer (0.5% Igepal CA030, 10 mM Hepes pH 7.9, 1 mM DTT, 100 mM NaCl, 0.5 mM PMSF) in the presence of protease and phosphatase inhibitors in the case of cells within the culture. Concentrations of the lysed proteins were measured by the Bradford technique, equaled to 25–50 µg and incubated for 1 min at 80°C with a protein loading buffer (62.5 mM Tris‐HCl pH 6.8, 2% SDS, 10% glycerol, 0.01% bromophenol blue) in the presence of 0.1 M DTT. Proteins were separated on a 1.5‐mm width 12% acrylamide denaturing gel (SDS‐PAGE). The samples were seeded next to the pre‐stained standard (Spectra Multicolor, Broad Range Protein Ladder, Thermo Scientific, Cat# 26623) in the gel, then separated at 80 V in a solution containing 25 mM Tris, 250 mM glycine, and 0.1% SDS. Proteins were subsequently transferred to an Immobilon‐P membrane (0.45 μm pore, Merck Millipore; Cat# HVLP04700) in transfer solution (25 mM Tris, 192 mM glycine, 20% methanol) for 2 hr at 250 mA. To verify correct transfer, the nitrocellulose membrane was stained with "Ponceau" red (Sigma Aldrich; Cat# P3504‐10G). Multiple washes were performed with TBS‐Tween (150 mM NaCl, 10 mM Tris, 0.05% Tween20), followed by blocking the membrane with 5% skim milk in TBS‐Tween for 1 hr. The membranes were next incubated with the primary antibody and subsequently with the secondary, both dissolved in 5% skim milk TBS‐Tween at 4°C overnight and at room temperature (20ºC) for 1 hr, respectively. The primary antibodies used were mouse anti‐Cx43 (dilution 1:1,000, BD Biosciences), rabbit anti‐lamina B1 (dilution 1:1,000, Abcam), and anti‐phospho ERK1/2 Thr202/Tyr204 (dilution 1:2,000, BioLegend; RRID:AB_2721734). The secondary antibodies used were peroxidase‐conjugated anti‐mouse and anti‐rabbit (1:1,000; Jackson ImmunoResearch Laboratories; Cat# 715‐035‐150; Cat#711‐035‐152). After secondary antibody incubation, three washes with TBS‐Tween were performed for 10 min. Finally, the membrane was exposed using a Western Lightening Plus‐ECL kit (Perkin Elmer; Cat# NEL103001EA) on the luminescent chemo and fluorescence imaging equipment (PXi, Syngene).</p><!><p>Tanycytes were seeded on 8‐mm coverslips covered with poly‐l‐lysine in 24‐well plates. The cultures were incubated at 37°C for 26 hr with a final concentration of each compound (or the mixture of them): 20 ng/ml FGF2, 10 IU/mL heparin, 200 µM Gap27 and 10, 50, 100, or 200 µM ATP (Sigma Aldrich; Cat# A1852), 10 µM ATPɣS (Sigma Aldrich; Cat# 11162306001), and/or 10 µM MRS2179 (Sigma Aldrich; Cat# M3808). After 20 hr, the medium containing the same compounds was renewed, and BrdU was added in a final concentration of 10 µM for another 6 hr. The 6‐hr timepoint was selected since at 2 hr, few cells incorporate BrdU and at 14 hr, almost all were labeled (unpublished). After incubation, the cells were washed twice in 0.1 M PBS, pH 7.4 and fixed for 30 min with 4% paraformaldehyde for later immunocytochemical analysis.</p><!><p>The osmotic pumps (Alzet Model 1007D), with capacity of 100 µl and a continuous release rate of 0.5 µg/h for 7 days, were filled the day before implantation according to the manufacturer's manual. BrdU, FGF2, and Gap27 were dissolved in filtered CSF (1.25 mM NaH2PO4, 126 mM NaCl, 3 mM KCl, 2 mM MgCl2 7H2O, 2 mM CaCl2 6H2O, 26 mM NaHCO3, and 2 mM glucose, pH 7.4) at a final concentration of 1.5 µg/µl, 25 ng/µl, and 0.26 µg/µl, respectively. The stainless‐steel tube was fitted to a vinyl catheter with an internal diameter of 0.58 mm and a length of 3 cm and both were filled with the same solution. The osmotic pumps were equilibrated in sterile saline solution (0.9% w/v) at 37°C for 16 hr before being implanted. Following implantation, compounds were infused at a rate of 0.75 µg/h (BrdU), 0.0125 µg/h (FGF2), and 0.13 µg/h (Gap27).</p><!><p>The cannulas were implanted by 3V stereotaxis according to the protocol shown in Figure 7. Rats were anesthetized with an intraperitoneal injection of the ketamine/xylazine mixture (90 mg/kg–10 mg/kg) and the fur was shaved over the head to expose the area where the incision was made. Rats were attached to the stereotaxic kit with ear bars that do not rupture the eardrum. The shaved skin was cleaned with clorhexydine, an incision was made with a scalpel, and a hole was drilled with a trephine to implant a guide cannula (28 gauge stainless steel; Plastics One), according to the following coordinates: ‐ 3.14 mm in the anterior–posterior axis of the bregma (confluence point of the sutures of the frontal and parietal bones), 0.0 medial–lateral of the medial sagittal sinus, 9.2 mm dorsal–ventral from the upper part of the skull. The cannula guide was previously connected to the 2 cm catheter attached to the perfusion pump and secured to the skull using 3/32 mm mounting screws and dental acrylic. Once the dental acrylic had dried, the osmotic pump was incorporated through the same incision with the help of forceps and faced with absorbable 5/0 HR15 synthetic sutures (Tagum). Animals received a subcutaneous injection of tramadol (2 mg/kg) and 1% ketoprofen (10 mg/ml). The rats were individually housed after surgery, although the infusion of the osmotic pump content was immediate. Rats were sacrificed 8 days after the implantation of the osmotic pump using an intraperitoneal injection of the ketamine/xylazine mixture (90 mg/kg–10 mg/kg), then they were vascularly perfused and cervically dislocated. The protocol was supervised by the ethics committee of the university.</p><!><p>All values were calculated as the average over each culture or animal. Significant differences were determined using the Student's t‐test or one‐way ANOVA with Bonferroni's multiple comparison post hoc test, as indicated otherwise. The normality of data was carried out using Shapiro–Wilk with a confidence value of 95%. For extracellular ATP measurements, CP values that were not within the standardization curve were excluded. For in vitro BrdU incorporation analysis, the visual field of each sample was blindly defined only by nestin fluorescence. Then, the number of BrdU‐positive cells was quantified using Fiji (Image J) software by first determining a subjective threshold, in which all cell nuclei were detected and detached from each other, and then applying the "Analyze Particles" tool with the following settings: "size (micron2)" in the range of 0.25‐infinity and "circularity" with range 0.00–1.00. Statistical significance was defined by p < .05 using the Graphpad Prism 7.0 program. No randomization was performed to allocate cultures in the study. The quantification of in vivo BrdU‐positive hypothalamic cells was computationally and objectively performed using Imaris 9.1.0 software (Centro de Microscopía Avanzada, CMA‐Bio Bio). The following parameters were established: (1) tanycytes and ependymal cells were considered as those cells with a proximity of 20 µm from the ventricular border, (2) the program determined the number of positive BrdU nuclei as those with a diameter greater than 7 µm and less than 12 µm (in images with 20X magnification), and (3) the total volume of the tissue was considered, excluding the ventricular area. Values with a standard deviation higher to 50% were discarded, no blinding was performed. The minimum sample size (experimental n) was determined according to the recursive equation of experiments that do not use blocks (i.e. sex or age), considering only the number of treatments and a confidence value of 95%.</p><!><p>In mouse and rat hypothalamus, the overlapped expression of vimentin and nestin neuronal precursor markers is primarily restricted to tanycytes (Pellegrino et al., 2018). To evaluate the conservation of undifferentiation markers in primary cultures of tanycytes, the expression of Sox2, Cx43, nestin and/or vimentin (both showing the same reactivity in tanycytes, data not shown) was compared to their expression in vivo. Immunohistochemical assays of a hypothalamic frontal slice using antibodies against Sox2 (Figure 1a‐a''), Cx43 (Figure 1b‐b'') and vimentin (Figure 1c‐c'') showed their joint expression in the lateral walls of the 3V (Figure 1d‐d''), represented by tanycytic and ependymal cells at the ventral VZ and most dorsal portion, respectively as shown in Figure 1a–d. Sox2 was localized in the nuclei of ependymocytes (Figure 1a'), α1 close to the ependymocytes (Figure 1a') and α2 close to the β1‐tanycytes (Figure 1a''), whereas Cx43 was mainly detected in the apical portion of both α and β1‐tanycytes (arrowheads in Figure 1b' and b'', respectively). Cx43 was also present in β2‐tanycytes, but with minor intensity (Figure 1b). As previously observed, Cx43 was found surrounding the blood vessels of rat hypothalamic parenchyma (arrows in Figure 1b'‐b''). Tanycyte primary cultures are usually performed under adherent conditions and consequently, they lose their polarized morphology (Figure 1e–h). Nevertheless, they retain the expression of nuclear Sox2 (Figure 1e‐e''), vimentin (Figure 1f‐f'') and Cx43 (Figure 1g‐g').</p><!><p>Connexin43 and undifferentiated markers are expressed by cultured tanycytes. (a–d) Representative confocal images of rat hypothalamus that show Sox2 (a‐a''), Cx43 (b‐b'') and vimentin (c‐c'') reactivity. (d‐d'') shows the three channels merged. The mentioned markers are present in α1 (a'‐d'), α2, and β1 (a''‐d'') tanycytes, which are specified in the magnification of the box in D. Arrowheads in (b‐b'') point to Cx43 localization in the apical portion of tanycytes, while arrows in (b‐b'') point its localization surrounding the blood vessels. (e–h) Immunocytochemistry of adherent primary cultures of tanycytes with antibodies against Sox2 (e‐e'), vimentin (f‐f'), Cx43 (g‐g'), and the merge of all channels (H‐H'). Hoechst was used for nuclei staining (e', g‐g'). (e', e''‐h') are augmented images of H box. 3V, third ventricle; Ep, ependymocytes; Tan, tanycytes. Scale bar: 50 µm. N: 1 animal and 3 primary culture</p><!><p>Considering that Cx43 forms hemichannels that are crucial to trigger purinergic signaling in response to glucose in tanycytes, we evaluated whether HCs Cx43 could also be activated and enhance purinergic signaling in response to mitogens, such as FGF2. At a first step, the expression of Cx43 and of the purinergic signaling receptors present in other neural precursors (NPs), as well as FGF receptor, which have been shown to increase the activity of Cx43 HCs opening, was evaluated transcriptionally. Conventional RT‐PCR analysis showed that Cx43, the metabotropic receptors p2y1, p2y2, and p2y4 (Lin et al., 2007), fgf1 (Garré et al., 2010; Schalper et al., 2008), fgf2 (De Vuyst et al., 2007) and its receptor, fgfr1, were expressed in both hypothalamic extracts (Hyp) and primary tanycytic cultures (Tan; Figure 2). Although the fgfr1 set of primers were designed to detect the long receptor isoform (containing the I Immunoglobulin‐like domain), the expression of the canonical FGF2 receptor isoform (IIIc) was already described by others (Kaminskas et al., 2019; Samms et al., 2015), specifically restricted to the ventral tanycyte subpopulation (Kaminskas et al., 2019).</p><!><p>The relative mRNA amount of Cx43, purinergic signaling components, and FGF2 pathway in the primary culture of tanycytes. Specific primers were used in order to amplify a fragment indicative of the presence of the following mRNAs: cx43, p2y1, p2y2, and p2y4 purinergic receptors, fgf1, fgf2, and its receptor, fgfr1. Actin was used as a loading control. Total mRNAs were extracted from whole cerebral tissue (Cer), hypothalamus (Hyp), striated muscle (Mus), and primary culture of tanycytes (Tan). Retrotranscription was performed in the presence of MulV retrotranscriptase enzyme (positive signs), whereas the absence of the enzyme was used as a negative control (negative signs). The size of the amplicon is shown at the left side of each electrophoresis. The most intense band in the scale bar (first line) represents 500 bp. N = 2 primary culture</p><!><p>The importance of Cx43 and P2Y1 receptor in the proliferation response of tanycytes in culture was examined using Gap27 and MRS2179, respectively. While MRS2179 is a competitive P2Y1 receptor antagonist, Gap27 is a Cx43 mimetic peptide in its 204‐SRPTEKTIFFI‐214 amino acid sequence, which interacts with the second extracellular loop and blocks the activity of Cx HCs in min, but later on also prevents the pairing of two hemichannels, affecting the formation of gap junction channels (Abudara et al., 2014). The DNA replication that proceeds cellular proliferation of tanycytes was evaluated by BrdU incorporation assay. BrdU treatment was evaluated after 24 hr in culture without serum plus another 26 hr of different treatments. The compounds were dissolved in the culture medium and 6 hr before finishing, the medium was renewed and BrdU was added, as shown in Figure 3a. Quantification of the proportion of tanycytes that incorporated BrdU in the analyzed time frame was performed through immunocytochemistry with antibodies specific for BrdU and nestin. Nuclei were stained with TOPRO (representative images in Figure 3b), which showed that under normal conditions, only 1.7 ± 0.1% (mean ± SEM) of the cells underwent division (Figure 3c). Since FGFs require the presence of heparan sulfate proteoglycans to interact with their receptor (Itoh & Ornitz, 2004), a solution containing heparin was used to dilute and deliver FGF2 to the tanycytes (Schalper et al., 2008). The presence of 20 ng/ml of FGF2 together with its heparin cofactor, increased the proportion of proliferative cells in one order of magnitude, reaching a significant value of 17.7 ± 5.1% (Figure 3c). The heparin cofactor promoted a slight effect on the cell division rate, although not statistically significant, increasing proliferation to 5.0 ± 1.7% (Figure 3c). Interestingly, Gap27 (200 µM) and MRS2170 (10 µM) prevented the FGF2‐induced proliferation, reaching 4.8 ± 1.2% and 3.4 ± 0.9% respectively, which were significantly lower than that observed with FGF2, suggesting the importance of the Cx43/P2Y1R axis in the self‐renewal capacity of this cell type. It is important to note that the exposure to each inhibitor did not affect tanycyte cell division (2.1 ± 0.5% for Gap27 and 1.1 ± 0.5% for MRS2179; Figure 3c). Assuming that nestin is expressed only by tanycytes in primary cultures, the number of cells positive for both nestin and BrdU over the total number of BrdU‐positive cells was quantified to corroborate that the proliferation events observed concerned only tanycytes (Figure 3d). The specificity of the BrdU on tanycytes ranged between 95.5 ± 2.7 (for Gap27) and 100 ± 0% (for heparin and MRS2179; Figure 3d). These results demonstrated that the proliferation observed is restricted to tanycytes. The purity of the cultured tanycytes described here agreed with our previous reports using the same methodology, which have shown to contain more than 90% tanycytes (Orellana et al., 2012). Moreover, their intense reactivity to vimentin, Kir6.1, GLUT2, GK, MC1 and 4, but not to GFAP, MAP2, and βIII‐tubulin rule out the contamination with other hypothalamic cell types such as astrocytes and neurons and strongly suggest a highly prevalence of β‐type tanycytes (Cortés‐Campos et al., 2011; García et al., 2003; Millan et al., 2010; Orellana et al., 2012). Negative immunoreaction for GFAP in β tanycytes has been recently corroborated by Kano et al. (2019). Since the tanycyte cultures are mainly composed of β‐tanycytes, it is therefore possible to argue that the Cx43‐dependent FGF2‐proliferative response seen here applies to this cell subpopulation, without excluding a possible effect on α‐tanycytes.</p><!><p>FGF2‐induced proliferation of cultured tanycytes is inhibited by Gap27 and MRS2179. (a) Timeline detailing the procedures performed to evaluate BrdU incorporation. (b) Immunofluorescence detection of BrdU (green) and nestin (magenta) immunoreactivity. TOPRO was used for nuclei staining (red). Scale bar: 100 µm. (c) Quantification of the BrdU positive cells (percentage) after the exposure to the FGF2 cofactor heparin, the mixture of FGF2/heparin, the mixture FGF2/heparin and blocker of either Cx43(Gap27) or P2Y1 receptor (MRS2179). (d) Quantification of the proliferative nestin‐positive cells over the total proliferative cells to assess the cell type‐specificity response to the stimuli. N = ≥12 replicates and three independent cultures per condition. One‐way ANOVA with Bonferroni post hoc. (**) p < .01, (ns) non‐significant. Data were represented as the average ± SEM</p><!><p>Inhibition of Cx43 or P2Y1 receptor blocked FGF2‐induced proliferation in tanycytic cells in vitro. This suggests that FGF2 signaling might be upstream via the purinergic pathway and activated by auto‐ or paracrine molecules released through Cx43 HCs. Previous studies have shown that FGF1 and FGF2 affect the release of ATP to the extracellular medium, and that this occurs after 7 hr of exposure to the mitogen (Schalper et al., 2008). In the present work, the effect of FGF2 on the functional state of Cx43 HCs was studied. The sensitivity of the ethidium uptake (Etd+) to Gap27 and lanthanum (La3+), a specific Cx HC inhibitor, was evaluated in the presence of physiological concentrations of divalent cations as described previously (Schalper et al., 2008). Confluent tanycyte cultures were treated with 20 ng/ml FGF2, 10 IU/mL heparin, and/or 200 µM Gap27 for 7 hr. Etd+ uptake was evaluated by nuclear fluorescence intensity according to the protocol described in Figure 4a. The FGF2/heparin combination significantly increased the Etd+ nuclear fluorescence over time (around 700 AU on average at 25 min of recording; Figure 4b) compared to the control condition and with heparin (608.7 AU and 614 AU at the same time, respectively). Moreover, the FGF2/heparin‐induced Etd+ uptake was inhibited by La3+ (gray stripe in Figure 4b). Treatment with Gap27, even in the presence of FGF2/heparin, prevented this effect to values similar to the control conditions (607.6 AU at 25 min). The presence of the inhibitor itself had no effect (conjugated to heparin; 632.2 AU at 25 min), suggesting that Cx HCs do not play a relevant role under control conditions. For a more comprehensive visualization, only some treatments are shown in Figure 4b.</p><!><p>FGF2 induces ATP release via connexin hemichannels in cultured tanycytes. (a) Timeline detailing the procedures to measure ethidium (Etd+) uptake. (b and c) Etd+ uptake over time (b) and Etd+ uptake rate (c) after exposure of tanycytes to different treatments and quantified as arbitrary units of emitted nuclear fluorescence (AU) over time (min). For a clearer view, data in (b) show the Etd+ uptake average and SEM curves of only some conditions (control, FGF2/heparin, and FGF2/heparin/Gap27). The gray bar in (b) represents the moment when the Cx HCs inhibitor, lantanum ion (La3+), was added. The graph in (c) was originated from the slopes in (b). N = ≥10 nuclei per culture and three independent cultures per condition were analyzed. (d) Luciferin/Luciferase signal obtained in measurements of the amount of extracellular ATP (pM), normalized to the total protein content (µg/µl). N = ≥3 independent cultures per condition. One‐way ANOVA with Bonferroni post hoc. (*) p < .05, (**) p < .01, (ns) non‐significant. Data were represented as the average ± SEM. Blue and black statistical characters indicate comparison of the responses to control and FGF2 conditions, respectively</p><!><p>The Etd+ uptake rate was calculated as the slope of the curves shown in Figure 4b and allowed us to compare the sensitivity of the Etd+ uptake rate to different treatments (Figure 4c). Under normal conditions, the rate of Etd+ uptake by tanycytes presented a baseline value of 1.65 ± 0.10 AU/min (mean ± standard error), which decreased to 0.94 ± 0.08 AU/min after the addition of La3+‐ , which was not significant. After the addition of the FGF2 cofactor, heparin, the Etd+ uptake increased to 2.35 ± 0.50 AU/min although it was not significant. La3+ applied at 25 min of recording reduced the Etd+ uptake rate to a value similar to that of the control condition, La3+ (0.86 ± 0.14 AU/min). To determine the pathway through which FGF2 increases tanycyte membrane permeability, a pharmacological criterion was applied; the increase in the levels of uptake observed in the presence of the FGF2/heparin combination (4.70 ± 0.35 AU/min) was significantly reduced after the application of La3+ at 25 min of recording (0.71 ± 0.27 AU/min). In addition, values were restored to those similar to the control condition when incubated 7 hr with Gap27 (2.00 ± 0.64 AU/min and 0.58 ± 0.64 AU/min in the presence of La3+). The presence of Gap27 per se had no significant effect on the parameters evaluated with respect to the control (1.57 ± 0.52 AU/min and 0.40 ± 0.22 AU/min with La3+). The data confirm that tanycyte membrane permeability induced by treatment with FGF2 and measured at 7 hr, depends mainly on the activity of Cx43 HCs.</p><p>In primary cultures of cortical astrocytes, Cx43 HCs provide a pathway for the uptake and release of small molecules, including those involved in auto‐ and paracrine signaling, such as ATP (Garré et al., 2010). In order to explore whether the opening of Cx HCs induced by FGF2 contributes to the release of ATP, the concentration of ATP in the culture medium of tanycytes treated for 7 hr with FGF2 was evaluated using the luciferin luciferase assay (Figure 4d). In the presence of the FGF2/heparin complex, the concentration of ATP released to the medium and normalized to the protein concentration was 6.3 times greater than that obtained under control conditions (24.0 ± 2.0 pM/µg × µl−1; mean ± SEM) and 9.1 times more than that induced by heparin alone (16.6 ± 3.6 pM/µg × µl−1), reaching values of 152.6 ± 30.5 pM/µg × µl−1. However, incubation of FGF2/heparin with Gap27 significantly attenuated the ATP release to the extracellular medium, reaching average levels of 48.8 ± 21.1 pM/µg × µl−1. Again, the presence of the Gap27 inhibitor had no per se effect, and the residual ATP values remained close to the control (32.3 ± 6.1 pM/µg × µl−1).</p><!><p>Previous studies have shown that extracellular ATP can exert a long‐term trophic effect in cultured astrocytes that includes promotion of DNA synthesis and cell division (Neary et al., 1998). It has been proposed that these events are mediated mainly by the activation of P2Y receptors (Neary et al., 1998). To investigate the possible mitogenic action of purinergic signaling activated by extracellular ATP, which can be released into the medium through Cx43 HCs once induced by FGF2, the incorporation of BrdU by tanycytes treated with the following conditions was assessed: (1) increasing ATP concentrations, (2) a non‐hydrolyzable analog of ATP, ATPƔS, and (3) ATPƔS in addition to MRS2179, a competitive P2Y1 receptor inhibitor. The different treatments were applied according to the protocol described in Figure 5a. Thus, we conducted concentration‐response type experiments, in which tanycytes were treated with increasing concentrations of ATP ranging from 10 to 200 µM. Immunocytochemistry assays were performed with anti‐BrdU (green) and‐nestin (magenta) specific antibodies, using TOPRO as the nuclear stain (red) (representative images in Figure 5b). The number of BrdU+cells/total cells (Figure 5c) and the number of double positive cells (BrdU+/nestin+) over the total number of BrdU+cells (Figure 5d) was quantified, respectively. The latter serves as a control to assess the tanycyte‐specific response to the stimuli. At 10 µM and 50 µM ATP, DNA synthesis in tanycytic cells was significantly higher than in cells under control conditions (Figure 5c), increasing the percentage of proliferative cells from 1.7 ± 0.1% (mean ± SEM) to 7.5 ± 1.2% (for 10 µM ATP) and 6.9 ± 1.3% (for 50 µM ATP). However, increasing concentrations of ATP did not lead to a concomitant increase in DNA synthesis, since at 100 µM and 200 µM, the percentage of transiting mitotic cells was 3.6 ± 0.8% and 5.5 ± 1.4%, respectively.</p><!><p>Proliferation of cultured tanycytes is promoted by ATP and ATPƔS and repressed by inhibition of P2Y1 receptors. (a) Timeline indicating when the procedures to evaluate BrdU incorporation were performed. (b) Immunofluorescence signal detected with antibodies reactive to BrdU (green) and nestin (magenta). TOPRO was used for nuclei staining (red). Scale bar: 100 µm. (c) Percentage of BrdU‐positive cells after exposure to 10 µM, 50 µM, 100 µM, 200 µM ATP, 10 µM ATPƔS, or 10 µM ATPƔS plus 10 µM MRS2179. (d) BrdU‐ and Nestin‐positive cells/BrdU total cells obtained in measurements of the amount of extracellular ATP (μM). N= ≥9 replicates and three independent cultures per condition. One‐way ANOVA with Bonferroni post hoc for control, 10 µM, 50 µM, 100 µM, and 200 µM ATP (first group analyzed) and control, 10 µM ATPƔS, and 10 µM ATPƔS plus 10 µM MRS2179 (second group independently analyzed). (*) p < .05, (**) p < .01, (ns) non‐significant. Data were represented as the average ± SEM</p><!><p>Two main classes of cell surface purinergic receptors have been described (Burnstock & Kennedy, 1985), which are the ATP P2 receptors and adenosine P1 receptors. The latter can be activated directly by adenosine or indirectly by products that result from the breakdown of ATP to adenosine by ectonucleotidases. To determine if the stimulation of cell division was because of the activation of P2 and/or P1 receptors in tanycytes, the hydrolysis‐resistant ATP analog, ATPƔS (Neary et al., 1998), and MRS2179, a P2Y1 receptor inhibitor, were used (Figure 5c). ATPƔS (10 µM) was sufficient to trigger a significant increase in tanycyte proliferation (4.7 ± 1.4%) compared to the control conditions, while 100 µM MRS2179 prevented it, reducing proliferation values closer to those of the control conditions (2.0 ± 0.8%). As described above, quantification of the double labeled nestin+/BrdU+ cells over the total of proliferating cells ranged from 83.8 ± 4.9 (for 10 mM ATP) to 97.1 ± 1.7% (for control), indicating that the observed response is cell type‐specific (Figure 5d). Along the experiments and conditions, cells expressed nestin and exhibited the typical expanded morphology of cultured tanycytes.</p><!><p>What mechanisms could explain the late involvement of Cx43HCs in the FGF2‐induced permeabilization? The results by far showed that treatment with FGF2 for 7 hr increased Cx43 HC activity and release of ATP to the extracellular medium. The half‐life of Cx43 is about 1.3 hr in cardiac tissue (Pogoda et al., 2016), suggesting that the detected changes in cell permeability mediated by Cx HCs could be a consequence of changes in Cx43 expression. To address this premise, the total amount of Cx43 after the FGF2 induction time was analyzed by immunoblots (Figure 6a–c and f) and immunofluorescence assays (Figure 6d‐e'). The specificity for the antibody used for Cx43 immunodetection was demonstrated using heart protein extract as a positive control and liver protein extract as a known tissue with very low expression (Figure 6a). Immunodetection assays were performed in protein extracts derived from three independent cultures (Figure 6b), which had approximately twofold increase in total Cx43 after 7 hr of heparin/FGF2 treatment (Figure 6c). The values were normalized to the G protein β subunit, which serves as a loading control for cell membrane proteins. For immunofluorescence assays to identify the cell membrane, the cells were fixed and briefly dipped with a solution containing WGA dissolved in PBS. WGA is a lectin that binds to the N‐acetyl‐D‐glucosamine and N‐acetylneuraminic acid monosaccharides and its derivatives (Acarin et al., 1994) and can be detected with anti‐WGA antibodies (in red). Co‐localization of anti‐Cx43 antibody binding sites with anti‐WGA was observed in the outer limits of the cell (Figure 6d‐d', arrows), indicating the presence of Cx43 on the cell membrane surface. Treatment with FGF2 and its cofactor induced an apparent increase in the amount of Cx43 content at the cell membrane and in the intracellular compartment with respect to the control conditions (Figure 6e‐e', arrows and arrowheads, respectively). A more detailed timeframe is shown in Figure 6f, where the amount of Cx43 decreased at 1 and 4 hr of FGF2/heparin treatment and then increased at 7 hr. To study whether these changes in Cx43 abundance were related to the FGF2 signaling pathway, the amount of phosphorylated ERK1/2 was analyzed at the same points (Figure 6f). The analysis showed an inverse relationship between the amount of Cx43 and the state of phosphorylated (p)ERK1/2 once treated with FGF2/heparin, the maximum of which occurred between hours 1 and 4 of activation by the ligand. These findings suggest that components of the FGF2 pathway regulate Cx43 through molecular events that were not discussed here, for example, by its the C‐terminal phosphorylation or the down‐regulation of its degradation (Axelsen et al., 2013).</p><!><p>FGF2 increases the total amount of Cx43 in cultured tanycytes. (a) Specific Cx43 immunodetection in whole cardiac (positive control) and hepatic tissue (negative control), as well as of the total protein extraction of tanycyte primary culture. Lamin B1 was used as a loading control. (b) Cx43 detection by western blot analyses of three independent tanycytes cultures with and without exposure to 7 hr heparin/FGF2 treatment. The G protein β subunit was used as a loading control (Gβ). (c) Densitometric analysis of Cx43 after treatment with heparin/FGF2 with respect to Gβ and normalized to the control situation. N = 3 independent cultures for each condition. (*) p < .01, T‐test. Data were represented as the average ± SEM. Primary culture of tanycytes without treatment (d‐d') or after 7 hr heparin/FGF2 induction (e‐e') were fixed, and immunofluorescence analysis was performed using antibodies against Cx43 (green) and WGA (red). The yellow arrows point to Cx43 at the cell boundaries, co‐localizing with WGA. Arrowheads show the intracellular Cx43. Scale bar: 50 µm. The boxes show an amplification of the respective images in the region marked with the yellow arrows. Scale bar:</p><!><p>In tanycytes, Cx43 HCs seem to have a fundamental role in glucose detection (Orellana et al., 2012) and in the activation of purinergic signaling induced by FGF2. However, in radial glia and adult NPs, gap junctions also participate in the cell cycle synchronization through the existence of a coupling network that permits the propagation of calcium waves (Weissman et al., 2004). Since tanycytes are robustly coupled to each other and to astrocytes and oligodendrocytes (Recabal et al., 2018), we wondered if cultured tanycytes retain the capacity to form coupling networks and if these are affected by FGF2. To address this question, tanycytes were grown on coverslips previously covered with poly‐l‐lysine up to 90% confluence and supplemented with serum‐free culture medium. Using a glass microelectrode (Figure S1a and b, asterisk), a single cell was filled with Lucifer yellow (Stewart & Wiley, 1981) for 5 min, then the spread of the molecule to other cells was observed under by fluorescence microscopy under control conditions (Figure S1a‐a') and after 7 hr of heparin/FGF2 treatment (Figure S1b‐b'). Under normal conditions, tanycytes established gap junctional communication in vitro forming clusters of at least two cells, although there were also more extensive configurations of 10 or more cells, reaching an average of nine coupled cells (Figure S1c). Heparin/FGF2 treatment induced the uncoupling of tanycytes to groups of ~3.2 cells on average (Figure S1d). These results suggest that FGF2 increases HC activity and reduces gap junctional communication between cells. Although these data are not statistically comparable, previous studies have shown the internalization and redistribution of Cx43 gap junction channels during cellular proliferation (Vanderpuye et al., 2016).</p><!><p>In order to detect changes in hypothalamic cell turnover upon FGF2 and Gap27 treatment in vivo, the cell proliferation assay was performed by continuous release of BrdU into the 3V. Osmotic pumps (Alzet, model 1007D) connected to the ventricular brain system by a cannula to continuously deliver FGF2 and Gap27 for 7 days at a rate of 0.0125 µg/h and 0.13 µg/h, respectively. BrdU (0.75 µg/h) was co‐administered to verify stimulation or interruption of cell proliferation, as shown in the schematic representation of Figure 7. The rats were sacrificed 8 days after the implantation of the osmotic pump and the brains were removed for subsequent immunohistochemical analysis through the co‐localization of BrdU (green) with vimentin, a tanycyte marker (red) (Figure 7). The cell division representative of the control situation (BrdU infusion only, Figure 7a and e‐e'), after exposure to FGF2 (Figure 7b and c and f‐g') and FGF2/Gap27 (D‐H') were detailed in the dorsal hypothalamic portion (Figure 7a–d) and in the ventral portion (Figure 7e‐h'). The BrdU positive ventricular cells, conceivably tanycytes, from both the lateral ventricular wall and the floor of the 3V are denoted with yellow arrows in Figure 7b, d and e‐f'. The infusion with FGF2 induced an increase in the number of highly variable proliferative ventricular cells, with sections with great incorporation of BrdU in the lateral VZ (Figure 7b) and ventral (Figure 7f‐f') and others with scarce BrdU labeling in the same areas (Figure 7c and g‐g', respectively). The dispersion of these data is evident in Figure 7l.</p><!><p>Proliferation of hypothalamic cells after ICV administration of FGF2 and Gap27. Immunohistochemistry of rat hypothalamic frontal sections when exposed for 7 consecutive days to vehicle and BrdU ICV infusion (a, e‐e'), FGF2 (b and c, f‐f', g‐g'), and FGF2/Gap27 (d, h‐h'). Using specific antibodies, the incorporation of BrdU (green) by hypothalamic cells, including vimentin‐labeled tanycytes (red), are shown. (a–d) Dorsal hypothalamic portion, which includes the subpopulation of α1‐tanycytes and ependymocytes. (e‐h') Ventral hypothalamic portion, considering α2, β1, and β2‐tanycytes, in addition to ME cells. The yellow arrows indicate the incorporation of BrdU by tanycytes. Scale bar: 100 µm. 3V, third ventricle. ME, Median Eminence. (i–n) Number of BrdU‐positive cells normalized to tissue volume in the parenchyma of the dorsal (i), ventral (j), and ME (j) hypothalamus. (l–n) Proliferation of tanycyte subpopulations normalized to 30 µm slice thickness; (l) α1‐tanycytes and ependymocytes, (m) α2‐tanycytes, and (n) β2‐tanycytes. N = at least 3 animals and 31 slices for each condition. One way ANOVA. (*) p < .05, (ns) not significant. Data are represented as mean ± SEM</p><!><p>The quantitative analysis of BrdU‐positive (BrdU+) cells was computationally performed, considering the number of parenchymal and ventricular cells labeled, both for the dorsal and ventral sections and for the Median Eminence (ME) (Figure 7i–n). All the data are represented as mean ± SEM. The number of proliferative parenchymal cells in the dorsal hypothalamus of control animals was 4.7 × 10−5 ± 1.1 × 10−5 BrdU+cells/per µm3 of tissue (Figure 7i), although FGF2 and FGF2/Gap27 treatment groups did not vary significantly compared to the control (5.3 × 10−5 ± 1.2 × 10−5 and 5.7 × 10−5 ± 4.8 × 10−6 BrdU+cells/µm3, respectively). The number of cells that incorporated BrdU in the ventral hypothalami did not vary significantly between the control, FGF2, and FGF2/Gap27 conditions, being 4.5 × 10−5 ± 4.6 × 10−6, 4.0 × 10−5 ± 2.8 × 10−6, and 5.7 × 10−5 ± 1.7 × 10−5 BrdU+cells/µm3 for each group, respectively (Figure 7j). Neither in the ME were significant changes in BrdU incorporation, being the counts 6.6 × 10−5 ± 9.1 × 10−6, 5.5 × 10−5 ± 7.4 × 10−6, and 3.2 × 10−5 ± 8.5 × 10−6 BrdU + cells/µm3 for control, FGF2, and FGF2/Gap27, respectively (Figure 7k). Next, the number of ventricular cells that underwent proliferation was quantified, defining them as tanycytes and/or ependymocytes if their nuclei were located up to 20 µm apart from the ventricular wall. For the count, the following population were considered: α1‐tanycytes and ependymocytes (Figure 7l), α2‐tanycytes (Figure 7m), and β2‐tanycytes (Figure 7n). The number of all the ventricular cells mentioned was normalized to 30 µm tissue thickness. Robins et al. (2013) showed that FGF2 stimulates the proliferation of α2‐tanycytes, which explains why the infusion of this mitogen was used as a positive control in our procedure. The α1‐tanycytes and ependymocytes showed a basal proliferation of 70 ± 3 BrdU+cells/ 30 µm thickness (Figure 7l), which increased, although not significantly, in the presence of FGF2 (127.3 ± 31.4 BrdU+cells/30 µm tissue thickness). Curiously, the infusion of both compounds (FGF2/ Gap27) did not significantly decrease the proliferation of α1‐tanycytes and ependymocytes with respect to that induced by FGF2 nor to the control (Figure 7d, 131.4 ± 31.9 BrdU+cells/30 µm tissue thickness, respectively). Although with a smaller number of cells, a similar effect was observed for α2‐tanycytes (Figure 7m), whose basal proliferation (8.7 ± 5.6 BrdU+cells/30 µm tissue thickness) increased, but not significantly, in the presence of FGF2 (31.7±8.5 BrdU+cells/ 30 µm tissue tickness) and this could not be reversed in the presence of Gap27 (48.6 ± 27.8 BrdU + cells/ 30 µm tissue thickness). The quantitative analysis of β2‐tanycytes proliferation revealed an increase, although not significant (Bonferroni post hoc analysis), in their cell division after induction with FGF2 (23 ± 4.4 BrdU+cells/ 30 µm tissue thickness) compared to the control (9.7 ± 2.2 BrdU+cells/ 30 µm tissue thickness), which could be significantly blocked by Gap27 (7.0 ± 0.1 BrdU+cells/30 µm tissue thickness; Figure 7n). Thus, β2‐tanycytes were the only cell type affected by Cx43 inhibition, which was required for the FGF2‐induced cell division. Finally, our in vitro results agreed with those in vivo, indirectly evincing the preferentially β subpopulation content in the tanycyte cultures.</p><!><p>Previous studies on Cx43 gap junctions in ex vivo tanycytes showed that they are robustly coupled to each other (Szilvasy‐Szabo et al., 2017) and to astrocytes and oligodendrocytes (Recabal et al., 2018). Moreover, Cx43 was the only connexin responsible for tanycyte coupling network. In the present work, the role of Cx43 as HCs on tanycytes was explored. However, we cannot exclude that some of the effects seen, that is, in cell division, were in part due gap junction blockade. Inhibition of Cx43 HCs in tanycytes prevented FGF2‐induced proliferation in vitro and in vivo, suggesting the involvement of both proteins in a common pathway. In C6 glioma cells (De Vuyst et al., 2007) and HeLa cells (Schalper et al., 2008) transfected with Cx43, as well as in spinal astrocytes (Garré et al., 2010), FGF1 or FGF2 induces a transient opening of the formed Cx43 HCs, through which ATP is released. Etd+ uptake analysis revealed that FGF2 promotes the activity of tanycytic Cx43 HCs, which were partially but drastically inhibited by Gap27, suggesting that this connexin is primarily responsible for the dye uptake. However, the contribution of other connexins (such as Cx45, highly expressed in tanycyte cultures) and pannexins (particularly Panx1 and Panx2) (Recabal et al., 2018), which allow the diffusion of molecules across the cell membrane, cannot be ruled out. Opening of Cx43 HCs induced by FGF2 led to the release of ~6.3‐fold more ATP compared to the control conditions. Although these values are low compared to the ~45‐fold increase of nucleotide release after treatment of tanycytes with 10 mM glucose (Orellana et al., 2012), it is important to consider the timing at which maximal response was elicited by cells treated with FGF2 (7 hr) and glucose (< 1 min). ATP output was significant, but partially inhibited by Gap27, suggesting that in addition to Cx43 HCs, other routes might be involved. It is known that the ATP release induced upon activation of FGFRs may be because of (1) vesicular release; (2) activation of P2X7 receptors, which in turn stimulate the opening of Panx1 channels, and (3) Cx43 HCs (Garré et al., 2010). However, Panx1 channels are not responsible for the ATP release induced by 10 mM glucose in tanycytes (Orellana et al., 2012).</p><p>In cultured tanycytes, we found that FGF2 increased the activity of Cx43 HCs as evaluated by Etd+ uptake in addition to ATP release assays. Previous studies performed in Cx43‐transfected HeLa cells showed that FGF1 increases the activity Cx43 HCs, which is associated with an increase in cell surface distribution rather than changes in the total amount of Cx43 (Schalper et al., 2008; De Vuyst et al., 2007). However, 10 mM glucose increases the ion flux through Cx43 HCs but does not affect the amount of Cx43 on the cell surface (Orellana et al., 2012). The modifications indicated are attributable to the phosphorylation of the Cx43 carboxy‐terminal domain by various kinases, including MAPK, which underlies the regulation of Cx43 HC opening and Cx43 distribution and degradation as well (Pogoda et al., 2016). The present work showed that 7 hr of FGF2 treatment doubled the total amount of Cx43 in tanycytes, an event that might be mediated by an ERK1/2‐dependent mechanism. In support to this statement, ERK1/2 phosphorylation was apparently inversely related to the amount of Cx43 throughout the FGF2 treatment. Nevertheless, the amount of Cx43 correlates positively with ERK1/2‐dependent phosphorylation in endothelial cells (Arshad et al., 2018). It is also possible that FGF2 signaling reduces the degradation instead of increasing expression and synthesis of Cx43 (Axelsen et al., 2013). Hence, the molecular mechanism by which FGF2 regulates the amount of Cx43 and opening of Cx43 HCs in tanycytes requires further studies.</p><p>ATP release can exert long‐term effects, such as proliferation, differentiation, migration, and apoptosis in various cell types, especially in astrocytes (Neary et al., 1998) and embryonic NPs (Weissman et al., 2004). The current work showed that tanycytes increased their cell division when they were exposed to 10 or 50 µM ATP, with concentrations greater than 100 µM having no significant effects. The concentration‐dependent effect agrees with studies in astrocytes (Neary et al., 2008) and NPs of the adult subventricular zone (ZSV) (Mishra et al., 2006), where ~30–50 µM of this nucleotide potentiates FGF2‐induced proliferation and 300 and 1,000 µM inhibit it (Neary et al., 2008). Low and high concentrations of ATP activate P2Y and P2X receptors, respectively, triggering opposite effects in the cells, while activation of P2Y receptors promotes DNA synthesis and activation of P2X7 receptors induces arrest of the astrocyte cell cycle in a resting state (Neary et al., 2008). On the other hand, the expression of hypothalamic P2X4 receptors (ATP ionotropic receptors) is limited to NPY orexygenic neurons and tanycytes (Xu et al., 2016). The functional role of this receptor type in NPY neurons, once activated by ATP, is to facilitate the release of GABA from the pre‐synaptic terminal on the two post‐synaptic targets—ARC POMC anorexigenic neurons and paraventricular nucleus neurons (Xu et al., 2016). Although it is known that under sugary (Frayling et al., 2011) and different sweetener stimuli (Benford et al., 2017), tanycytes undergo activation of a purinergic P2 receptor signaling, and the functional role of P2X4 receptors in tanycytes remains to be elucidated. It is reasonable to propose that FGF2‐induced release of ATP by tanycytes has an impact not only autocrine proliferative through P2Y1 receptors, but it might also be paracrine, through the activation of P2X4 receptors in NPY neurons, as is depicted in the final scheme (Figure 8).</p><!><p>Scheme representing the effect of FGF2 on the activity of connexin hemichannels and consequences on ATP release and proliferation of cultured tanycytes. 1) FGF2 induces proliferation of cultured tanycytes via activation of FGFR1, leading to 2) increase connexin43 hemichannel (Cx43 HC) opening, and 3) ATP release via Cx43 HCs. ATP through autocrine signaling could activate 4a) P2Y1 located in the tanycytes for inducing [Ca2+]i increases, which could mediate 5a) tanycyte proliferation. In addition, P2Y 2,4,11,12, and 13 receptors located in tanycytes and the adenosine receptor P1A could potentiate these effects. Another action that could be attributed to ATP released by the tanycytes, is the 4b) paracrine activation of P2X4R in NPY neurons that release GABA</p><!><p>The eight P2Y receptor subtypes can be activated by ATP, ADP, UTP, UDP, and nucleosides (Figure 8), and are linked to different signaling cascades (Zimmermann, 2006). The presence of ectonucleotidase enzymes, whose catalytic site is facing the cell exterior, controls the functionality of extracellular nucleotides. Specifically, the nucleoside triphosphate diphosphohydrolase 2 (ENTPDase2) enzyme is highly expressed by PNs of the adult neurogenic niches, the subventricular and subgranular zones, and catalyzes the hydrolysis of nucleoside triphosphates, transforming them into di and subsequently, monophosphate (Gampe et al., 2015; Mishra et al., 2006). RNAseq studies in cultured tanycytes (Recabal et al., 2018) indicate that the ENTPDase2 transcripts are highly represented, suggesting that the proliferation observed after their exposure to ATP could be a consequence of the molecular interactions that result from ATP degradation (ADP, AMP, and adenosine) with the P2Y1,2,4,11,12,13 and/or P1 receptors (Zimmermann, 2006). To circumvent the lack of specificity, the non‐hydrolyzable analog of ATP, ATPƔS, and the specific inhibitor of P2Y1, and MRS2179, were used. This approach demonstrated that ATPƔS is sufficient to induce an increase in BrdU incorporation by tanycytes, a response that was blocked by MRS2179, suggesting that the hydrolysis of ATP by ectonucleotidases was not essential to achieve the proliferative effect. However, because the quantification of BrdU incorporated upon treatment with 10 mM ATP was slightly higher than that induced with the same concentration of ATPƔS, it is not possible to completely rule out the participation of ADP and adenosine activating other purinergic receptors. Our data coincide with those provided by the literature for the SVZ NPs, which present high hemichannel activity (Talaverón et al., 2015), through which ATP could be mobilized to the outside of the cell to promote cell proliferation once the purinergic receptors are activated (Suyama et al., 2012). Inhibition of either ATP release or activation of purinergic receptor affected cell proliferation. Figure 8 shows another action that could be attributed to ATP released by the tanycytes. Xu et al. (2016) have shown that ATP could mediate activation of P2X4R in NPY neurons that release GABA for inhibiting POMC neurons and, therefore, contribute to regulating feeding behavior. The effects entailed by purine release from tanycytes are physiologically broad, both in a long and short term, ranging from promoting cellular division to triggering activation of neuronal orexigenic responses.</p><p>In the present work, it was demonstrated that Gap27, a Cx43HC inhibitor, reduced the FGF2‐induced proliferation of β2‐tanycytes, which have been controversially proposed as PNs of the adult hypothalamus (Kano et al., 2019; Lee & Blackshaw, 2012). This proliferative blockade seems to be unique to this cell subtype, since parenchymal cells of hypothalamic neuronal nuclei, ME, and α‐tanycyte cells did not show such dramatic decrease in BrdU incorporation upon FGF2 treatment. Notably, similar unalterable proliferative state of parenchymal cells in the ARC has been previously observed after HFD treatment (Safahani et al., 2019). Studies exploring changes in the metabolic state of the individual support the conception of ME‐residing cells plasticity. Clear examples of this are the morphological changes of the external ME area to facilitate the release of the gonadotropin‐releasing hormone into the portal circulation during the estrous cycle, which consists of a close up of the parenchymal basal lamina to the neuronal terminals (Prevot et al., 1999). Consistent with this, female rats treated with a HFD show increased proliferation and neurogenesis specifically in ME (Lee et al., 2014). Previously, the same researchers elucidated the origin of adult nascent neurons, attributing it to β2‐tanycytes (Lee et al., 2012). Hence, Cx43 HCs plays a crucial role not only in the detection of glucose by tanycytes (Frayling et al., 2011; Orellana et al., 2012), but also in the self‐renewal of β2‐tanycytes, evidencing the versatility of membrane channel. The tanycyte multiplication likely depends on the concentration of FGF2 and Gap27 reached in the CSF, which may have not been optimal to achieve noticeable responses, for example, on the incorporation of BrdU by α‐tanycytes. This possibility remains to be further studied using higher FGF2 concentrations up to a detectable cell proliferation response that could be evaluated using a direct approach, for example through Ki67 measurements. In parallel the use of Gap27 or other selective Cx43 HC blocker would unveil the involvement of Cx43 HCs in FGF2‐induced cell proliferation.</p><p>Neurogenesis in adult life remains a controversial topic within the scientific community (Sorrells et al., 2018). In the last decade, hypothalamic neurogenesis has been proposed as an adaptive response mechanism to nutritional imbalance (Sousa‐Ferreira et al., 2014). Therefore, our results provide a novel mechanism involved in this process. Purinergic signaling is mediated in part by Cx43 HCs (and likely scattered by Cx43 gap junctions) that participate in the proliferation of hypothalamic tanycytes, a mechanism that could underlie the development of pharmacological approaches to regulate body weight and decrease the incidence of obesity.</p><!><p>The authors declare that they have no competing interests.</p><!><p>The experiments were performed at the Department of Cell Biology at the University of Concepcion and Departamento de Fisiología, Facultad de Ciencias Biológicas, Pontificia Universidad Católica de Chile, Santiago. MA.G‐R, A.R, and JC.S conceived the experiments; MA.G‐R., A.R., T.C, and JC.S designed the experiments; A.R, P.F., S.L., P.O, A.P, MJ.B, performed the experiments; A.R., P.O., C.F., R.E‐V. analyzed the data; MA.G‐R, A.R., C.F., T.C., E.U., and JC.S. contributed reagents/materials/analysis tools; MA.G‐R., A.R., and JC.S. wrote the paper, and critically revised the manuscript. All authors have approved the final version of the manuscript and agree to be accountable for all aspects of the work in ensuring that questions related to the accuracy or integrity of any part of the work are appropriately investigated and resolved. All persons designated as authors qualify for authorship, and all those who qualify for authorship are listed.</p><!><p>Supplementary Material</p><p>Click here for additional data file.</p>

PubMed Open Access

NF-κB in the crosshairs: Rethinking an old riddle

Highlights•NF-κB transcription factors are central coordinating regulators of immunity, inflammation and cell survival.•NF-κB pathway is aberrantly and stably activated in cancer.•The ubiquitous presence and pleiotropic physiological role of NF-κB dimers have thus far prevented the development of any clinically useful NF-κB inhibitor.•Emerging therapeutic approaches aim to achieve the cancer-selective inhibition of the NF-κB pathway as a way to overcome the preclusive toxicities of conventional IKKβ/NF-κB-targeting drugs.

nf-κb_in_the_crosshairs:_rethinking_an_old_riddle

Introduction<!><!>The futile pursuit of a specific NF-κB inhibitor: an historical perspective on an obstinate conundrum<!>Embracing complexity as a path to achieve the safe therapeutic inhibition of the NF-κB pathway<!>Conclusions<!>Conflict of interest

<p>The anticancer arsenal has traditionally consisted of a limited number of broadly active cytotoxic chemotherapeutics characterised by a small therapeutic index and a minimal capacity to discriminate between malignant and normal cells. Over the past 25 years, fundamental advances in the field of molecular oncology and the understanding of many of the core mechanisms driving oncogenesis have enabled the generation of rationally designed, targeted therapies which selectively interfere with discrete oncogenic effectors, thereby opening the door to an era of stratified oncology and, consequently, revolutionising the clinical management of cancer patients. Indeed, the oncology field is currently undergoing a new revolution with the boom of anticancer immunotherapies capable of producing long-term remissions and even curative outcomes, breaking away from traditional paradigms by targeting the non-malignant, rather than malignant, cell components within tumours (Hodi et al., 2010; Brahmer et al., 2012).</p><p>The scientific breakthroughs of the past few decades have enabled the creation of a new generation of anticancer medicines, which couple greater specificity with reduced adverse effects, thus equipping the current anticancer armoury with multiple classes of new agents which selectively interfere with a wide spectrum of discrete drivers of oncogenesis. However, while an ever-growing number of cancer-driving mechanisms and signalling pathways have thus far been successfully pharmacologically targeted, leading to improved clinical outcomes in oncology, a select group of other pathways have proven defiant to therapeutic intervention. Among these, the NF-κB pathway stands out as perhaps the most illustrious example and arguably the one that has coalesced the greatest frustration and disappointment.</p><p>Ubiquitous NF-κB transcription factors are central coordinating regulators of the host defence responses to stress, injury and infection (Hayden and Ghosh, 2012; Zhang et al., 2017). In addition to fulfilling these elemental physiological roles, NF-κB contributes to the pathogenesis of most of the chief threats to global human health, including cancer, atherosclerosis, diabetes and chronic inflammatory diseases (Xia et al., 2014; DiDonato et al., 2012). Aberrant NF-κB signalling is a hallmark of the large majority of human cancers, where it drives oncogenesis, disease recurrence and therapy resistance, largely by regulating genes that suppress malignant cell apoptosis and govern inflammation in the tumour microenvironment (TME) (Xia et al., 2014; DiDonato et al., 2012). Unsurprisingly, owing to these pivotal pathogenic roles of NF-κB, the targeting of the NF-κB pathway has been a paramount objective of the pharmaceutical industry and the focus of worldwide research efforts for the past 25 years, as a means to improve the clinical management of both oncological and non-oncological patients, especially within particularly refractory disease indications (Gilmore and Herscovitch, 2006; Begalli et al., 2017). However, as best illustrated by the ill-fated pursuit of a clinically useful inhibitor of IκBα kinase (IKK)β, the kinase responsible for phosphorylating IκB proteins and enabling nuclear NF-κB translocation (Hayden and Ghosh, 2012), achieving this goal has to this day proven an insurmountable problem, owing to the failure of traditional IKK/NF-κB-targeting strategies to preserve the pleiotropic and ubiquitous physiologic functions of NF-κB (Greten et al., 2007; Hsu et al., 2011). This minireview offers a glimpse into some of the more promising emerging approaches currently being considered to circumvent these inherent limitations of conventional NF-κB inhibitors, with a focus on oncology.</p><!><p>The cancer-selective strategy to target the NF-κB signalling pathway. Schematic representation of the canonical pathway of NF-κB activation. Depicted in black are the main conventional therapeutic strategies, which have thus far been used to generate pharmacological NF-κB inhibitors. Also depicted in red is one of the emerging approaches aimed at developing a therapeutic inhibitor of a functionally critical and cancer cell-restricted downstream effector of the pathogenic survival axis of the NF-κB pathway. Also shown is the D-tripeptide inhibitor of the GADD45β/MKK7 complex, DTP3, which selectively targets this GADD45β-dependent survival axis of the NF-κB pathway, yielding cancer cell-selective therapeutic activity, thereby circumventing the preclusive limitations of global IKKβ/NF-κB inhibitors.</p><!><p>Given the multitude of stimuli that can activate NF-κB and the broad spectrum of functions that NF-κB plays in different tissues, it is unsurprising that several feedback mechanisms have evolved to ensure the tight control and timely termination of physiological NF-κB signalling as a way to enable the prompt return to homeostasis and prevent excessive inflammation, tissue damage and the development of malignancy (Hayden and Ghosh, 2012; Zhang et al., 2017; Begalli et al., 2017). Indeed, excessive and stable IKK/NF-κB activation is a typifying feature of a wide range of pathological states, including cancer. Whereas in certain malignancies, such as multiple myeloma, diffuse large B-cell lymphoma (DLBCL), mucosa associated lymphoid tissue (MALT) lymphoma and glioblastoma multiforme (GBM), NF-κB is often constitutively activated by recurrent genetic alterations targeting upstream components of the NF-κB pathway, in the large majority of solid tumours and certain haematological malignancies, such alterations of the NF-κB pathway are relatively infrequent (DiDonato et al., 2012; Annunziata et al., 2007; Keats et al., 2007; Pasqualucci et al., 2001; Bredel et al., 2011). Accordingly, in these cancers, aberrant NF-κB activation generally stems from genetic abnormalities targeting conventional tumour-suppressor and oncogenic mechanisms, such as RAS and PTEN mutations, and/or the steady exposure of tumour cells to inflammatory stimuli and other cues emanating from the TME (DiDonato et al., 2012). These findings, and an accompanying extensive body of other genetic, biochemical and clinical evidence, provide a compelling rationale for therapeutically blocking constitutive NF-κB signalling in a wide range of human cancers in areas of current unmet need. Moreover, there is a strong rationale for developing NF-κB-targeting therapeutics to treat numerous non-malignant human pathologies, such as diabetes, autoimmune disorders, and chronic inflammatory diseases, owing to the central role of NF-κB signalling in governing inflammation, and the underlying low-grade inflammatory reaction that propagates the pathogenesis of these and virtually all other human illnesses (Xia et al., 2014; DiDonato et al., 2012). Notwithstanding, to this day – more than 30 years since the discovery of NF-κB and despite an aggressive and persevering effort by the pharmaceutical industry over the past 25 years – no specific NF-κB inhibitor has been clinically approved, due to the preclusive on-target toxicities associated with the systemic inhibition of NF-κB (Gilmore and Herscovitch, 2006; Begalli et al., 2017).</p><p>Owing to its central role as the downstream signal-integration hub for the pathways of NF-κB activation, IKKβ bore the brunt of the drug discovery effort to inhibit pathological NF-κB signalling since its discovery in 1996 (Fig. 1). Nonetheless, while the initial impetus did succeed in generating a large array of specific molecules and multiple candidate therapeutics, this effort eventually came to an inevitable abrupt end, as soon as IKKβ inhibitors were evaluated in animal models and early-phase clinical trials (DiDonato et al., 2012; Greten et al., 2007; Hsu et al., 2011). In a seminal paper published in 2007, Karin and colleagues demonstrated that the pharmacological inhibition of IKKβ increases IL-1β secretion by myeloid cells, owing to an enhanced processing of pro-IL-1β by caspase 1, leading to overt systemic inflammation and increased animal lethality (Greten et al., 2007). In addition to this unanticipated, dose-limiting adverse effect, subsequently confirmed in human studies, global IKKβ/NF-κB inhibition produced a series of other adverse effects, including immunodeficiencies, hepatotoxicity and a potentially increased risk of malignancies arising from tissues such as the liver and the skin, reflecting the essential roles of NF-κB in innate and adaptive immune responses and tissue homeostasis (DiDonato et al., 2012; Greten et al., 2007; Hsu et al., 2011). Eventually, after the initial, short-lived enthusiasm, these findings irrevocably halted any further significant clinical development of IKKβ/NF-κB inhibitors, as demonstrated by the recent dramatic decline in new patent applications relating to these agents (Begalli et al., 2017).</p><p>Another class of drugs originally developed to therapeutically target pathological IKKβ/NF-κB signalling are proteasome inhibitors, which stabilise IκB proteins, thereby preventing nuclear NF-κB translocation by interfering with the proteolytic activity of the proteasome (Fig. 1) (Zhang et al., 2017; Manasanch and Orlowski, 2017). These molecules, as well as immunomodulatory drugs (IMiDs), are known to impact upon NF-κB signalling and have found broad clinical indication in multiple myeloma and a handful of other malignant pathologies. However, both classes of drugs display broad biological activities, lack any specificity for NF-κB, and, importantly, afford clinical benefit in these indications via a mechanism unrelated to the NF-κB pathway (Manasanch and Orlowski, 2017; Richardson, 2010). Consequently, there remains an urgent need for a fresh and entirely different approach to safely targeting the NF-κB pathway in human diseases.</p><!><p>Historically, the insurmountable problem with conventional NF-κB-targeting strategies has been to achieve the contextual, tissue-specific inhibition of the NF-κB pathogenic activity, while preserving the pleiotropic and ubiquitous physiological functions of NF-κB, including its functions in immunity and inflammation (DiDonato et al., 2012). Since the best documented activity of NF-κB in oncogenesis is to upregulate genes that suppress cancer-cell apoptosis, and despite its ubiquitous nature, NF-κB signalling elicits transcriptional programmes that vary considerably depending upon the type of tissue and activating stimulus, we sought to target a non-redundant, cancer cell-specific downstream effector of this oncogenic NF-κB-mediated survival function, rather than NF-κB itself (Fig. 1) (Begalli et al., 2017; Annunziata et al., 2007; Keats et al., 2007; Bennett et al., 2013). We postulated that this strategy could provide a comparably effective, yet considerably safer alternative to conventional IKKβ/NF-κB-targeting drugs, thus circumventing the dose-limiting toxicities of systemic IKKβ/NF-κB inhibition.</p><p>Our group recently tested this hypothesis in the context of multiple myeloma, a malignancy of plasma cells (PC) responsible for almost 2% of all cancer deaths and representing the paradigm of NF-κB-driven cancers (Annunziata et al., 2007; Keats et al., 2007). Despite the recent introduction of new treatments, almost all multiple myeloma patients eventually relapse and/or develop drug resistance. Consequently, the management of these patients remains a significant medical problem. Given its paramount importance in disease pathogenesis, the NF-κB pathway provides an attractive therapeutic target in multiple myeloma. Indeed, virtually all clinical cases of this neoplasia display constitutive NF-κB signalling and elevated NF-κB target-gene signature, leading to malignant cell addiction to nuclear NF-κB activity for survival and sensitivity to apoptosis upon IKKβ/NF-κB inhibition (Annunziata et al., 2007; Keats et al., 2007).</p><p>Our group, as well as others, previously reported that NF-κB inhibits apoptosis, at least in part, by suppressing the exaggerated activation of the JNK MAPK pathway through a mechanism that involves the transcriptional upregulation of Growth Arrest and DNA Damage 45B (GADD45B), a member of the GADD45 family of inducible genes, and other downstream effectors, such as X chromosome-linked inhibitor of apoptosis protein (XIAP) (Jin et al., 2002; De Smaele et al., 2001; Lin and Karin, 2003). Subsequent studies demonstrated that prolonged JNK activation leads to apoptosis, in part, by causing the phosphorylation-dependent activation of the E3 ubiquitin ligase, Itch, which in turn promotes the polyubiquitination and subsequent proteasome-mediated degradation of the caspase 8/10 inhibitor, cellular FLICE (FADD-like IL-1β-converting enzyme)-inhibitory protein (c-FLIP), leading to caspase 8/10 activation and ultimately cell death (Bennett et al., 2013). Prolonged JNK activation has been shown to also enhance the activity of several proapoptotic members of the B-cell lymphoma (BCL)-2 family of proteins, such as Bim and Bmf, by promoting their release from sequestered cytoplasmic pools normally bound to dynein and myosin V motor complexes (Kuwana and Newmeyer, 2003).</p><p>Recently, we identified the complex formed by GADD45β and the JNK kinase, MKK7, as an essential survival module dependent on constitutive NF-κB signalling and a novel therapeutic target in multiple myeloma (Fig. 1) (De Smaele et al., 2001; Papa et al., 2004; Tornatore et al., 2014; Tornatore et al., 2015; Papa et al., 2007). We demonstrated that GADD45B is upregulated in multiple myeloma cells by constitutive NF-κB activation, promotes malignant cell survival by suppressing proapoptotic MKK7/JNK signalling through its direct binding to and inhibition of MKK7, and is associated with poor clinical outcome in multiple myeloma patients (Tornatore et al., 2014; Tornatore et al., 2015). Importantly, most healthy cells do not constitutively express GADD45β, nor rely on GADD45β for their survival, and, unlike mice lacking IKKβ, any other IKK component, or the NF-κB subunit, RelA – which all die during late embryogenesis – Gadd45β-deficient mice are viable, fertile, and die of old age (Lu et al., 2004). Accordingly, we hypothesised that, in contrast to systemic NF-κB blockade, pharmacological GADD45β inhibition would be well tolerated, in vivo. Therefore, we sought to selectively target the NF-κB oncogenic function in multiple myeloma cells by inhibiting the GADD45β/MKK7 survival module downstream in the NF-κB pathway.</p><p>By screening a simplified combinatorial tetrapeptide library, followed by chemical optimisation, we developed the pharmacological D-tripeptide inhibitor, DTP3, which specifically binds to MKK7 with high affinity, disrupting the GADD45β/MKK7 interaction, and, as a result, selectively kills multiple myeloma cells by inducing MKK7/JNK-dependent apoptosis (Fig. 1) (Tornatore et al., 2014; Tornatore et al., 2015). We showed that, due to its target cell-specific mode of action, DTP3 displays potent and cancer-selective therapeutic activity against multiple myeloma cell lines and malignant PCs from multiple myeloma patients and, importantly, is not toxic to normal cells. Owing to these properties, DTP3 exhibited a more than 100-fold higher cancer-cell specificity than either proteasome or IKKβ inhibitors in primary human cells, ex vivo. Notably, as a result of this cancer cell-selective specificity, DTP3 caused a complete regression of established tumour xenografts, extending host survival in mouse models of multiple myeloma, upon intravenous administration, with excellent tolerability and no adverse effects at the therapeutic dose levels (Tornatore et al., 2014; Tornatore et al., 2015).</p><p>Further toxicology studies demonstrated that DTP3 was well tolerated in both rodent and non-rodent species, upon daily repeated-dose administration at high doses for 28 days, exhibiting no target organs of toxicity and no significant adverse effects, resulting in a wide therapeutic index and exposing no risk for its clinical progression. Accordingly, we are currently conducting the first-in-human phase-I/IIa study of DTP3 in patients with refractory or relapsed multiple myeloma. Upon an initial evaluation, DTP3 demonstrated clinical safety and tolerability at all dose levels investigated thus far, alongside a cancer-selective pharmacodynamic response, in highly refractory oncological patients and as a single agent. Future, larger clinical studies will determine the long-term clinical safety and therapeutic efficacy of DTP3 in patients with multiple myeloma and potentially other types of cancer in which DTP3 is indicated. While DTP3 has thus far produced no significant adverse effects in preclinical models or multiple myeloma patients, and unlike sensitive tumour cells, most normal cells do not constitutively express GADD45β, nor display spontaneous MKK7/JNK activation upon GADD45β inhibition, there remains a possibility that DTP3 administration will result in an exacerbation of MKK7/JNK signalling at sites of pre-existing inflammation, thus aggravating chronic inflammatory comorbidities and/or increasing the risk of autoimmune diseases. Further clinical studies will also consolidate the companion stratification strategy to select those patient subsets who will optimally respond to DTP3 and determine whether and, eventually, how rapidly responding tumours develop resistance to DTP3, for instance by acquiring MKK7 gene mutations or functionally redundant, GADD45β-independent antiapoptotic mechanisms. Notwithstanding, together with the compelling preclinical package, these highly encouraging initial clinical results introduce an unprecedented therapeutic mode of action – possessing none of the preclusive safety constraints of conventional IKKβ/NF-κB inhibitors – into clinical oncology and bode well for the ultimate clinical success of DTP3 as a safe and highly effective NF-κB-targeting therapeutic. These results also provide initial proof-of-concept for a safe and cancer-selective NF-κB-targeting strategy as a novel anticancer therapy which promises to be of profound benefit for patients with multiple myeloma and, potentially, other cancers where NF-κB drives oncogenesis via GADD45β (Fig. 1) (Karin, 2014).</p><p>Importantly, the same principle we developed of therapeutically inhibiting a cancer-restricted axis of the NF-κB pathway, rather than NF-κB globally, could be also applied to selectively targeting the NF-κB oncogenic function in GADD45β-independent malignancies and, plausibly, in the context of non-malignant NF-κB-driven diseases (Tornatore et al., 2014). Indeed, the NF-κB survival function is mediated by the upregulation of a diverse group of antiapoptotic target genes, which are independently transcriptionally regulated in a tissue- and stimulus-specific manner. Therefore, this NF-κB function is neither exclusively dependent upon GADD45β induction, nor is it necessarily dependent upon the suppression of JNK signalling (Xia et al., 2014; Bennett et al., 2013). For example, NF-κB has been shown to transcriptionally regulate the coding genes for several antiapoptotic members of the BCL-2 family, including B-cell lymphoma-extra large (Bcl-XL), myeloid cell leukaemia sequence 1 (MCL1), B-cell lymphoma 2-related protein A1 (BCL2-A1)/Bcl-2-related gene expressed in foetal liver (BFL-1), and, in certain biological contexts, BCL-2 itself. These proteins are involved in maintaining the outer mitochondrial membrane integrity, thereby preventing the release of cytochrome c and other proapoptotic mitochondrial factors, such as second mitochondria-derived activator of caspases (SMAC)/direct inhibitor of apoptosis protein (IAP)-binding protein with Low pI (DIABLO), into the cytosol, ultimately inhibiting the onset of cell death (Vogler, 2014). These antiapoptotic members of the BCL-2 family are also known to promote cancer-cell survival in various types of haematological and solid malignancy (Catz and Johnson, 2001; Cang et al., 2015). Notably, BCL-2-targeting drugs have been successfully developed outside the scope of blocking oncogenic NF-κB signalling, and drugs in this class, including the first-in-class BCL-2-family inhibitor, ABT-199 (venetoclax), have been granted breakthrough status designation by the FDA for treating subsets of patients with relapsed or refractory chronic lymphoid leukaemia (CLL) (Cang et al., 2015; Levy and Claxton, 2017). Therefore, these agents could be additionally developed to therapeutically target pathogenic NF-κB signalling in oncological situations in which NF-κB promotes malignant cell survival through the upregulation of BCL-2-like factors.</p><p>A potential alternative strategy to selectively target the NF-κB pathway in human cancer involves the inhibition of upstream signalling mechanisms that drive oncogenesis by virtue of their role in regulating NF-κB activation. For instance, the members of the IAP-family of E3 ubiquitin ligases, c-IAP1 and c-IAP2, have been shown to contribute to NF-κB activation by tumour necrosis factor (TNF)α and other stimuli through its binding to TNF receptor-associated factor (TRAF)-family proteins through their baculovirus IAP repeat (BIR) domain, leading to the polyubiquitination of their signalling substrates, including TRAF proteins themselves, to enable the ubiquitin-mediated assembly of multimeric, receptor-specific protein scaffolds, in which the IKK complex is brought into physical proximity of the transforming growth factor β-activated kinase (TAK)1 kinase complex, thereby resulting in IKKβ activation by TAK1-mediated phosphorylation. As well as XIAP, which is also involved in the recruitment of the TAK1 complex to the NF-κB signalosome, c-IAP1/2 proteins have been found to be highly expressed in a subset of human cancers, where they can promote NF-κB activation and cancer therapy resistance. Therefore, since small-molecule dual antagonists of c-IAP1 and XIAP, such as ASTX660, have been recently progressed into phase-II clinical studies in patients with various types of advanced haematological or solid cancer, including DLBCL, T-cell lymphoma, head and neck squamous cell carcinoma (HNSCC) and cervical carcinoma, these therapeutic molecules could be further developed to selectively target oncogenic NF-κB signalling in those clinical cases in which NF-κB is activated by the overexpression of c-IAP1 and/or XIAP proteins. This handful of examples underscore how recent advances in the understanding of the biological functions and regulation of the NF-κB pathway, and the contextual make-up of the genetic programmes NF-κB selectively elicits in cancer cells, are currently providing tangible new opportunities for targeted therapeutic interventions in different areas of unmet need across the oncological landscape.</p><!><p>Owing to its central role in disease pathogenesis, the NF-κB pathway has been pursued for decades as an attractive target for therapeutic intervention. Yet, despite the clear need for a specific NF-κB inhibitor to treat a broad range of human diseases, developing such a molecule has so far presented an impenetrable riddle, due to the need to confront a ubiquitous signalling pathway that has many elemental physiological functions. This has resulted in the dismaying absence of an NF-κB-targeting drug from the current pharmacopoeia. Despite this bleak reality, the past three decades have seen a succession of fundamental advances in the understanding of the intertwined signalling networks governing NF-κB activation, the myriad of cellular functions the NF-κB pathway embodies, and the diverse transcriptional programmes it contextually governs in any given cell, whether in normal or unhealthy tissues. Indeed, these advances are now providing important clues to finally untangle the NF-κB conundrum, and these clues, in turn, are beginning to translate into targeted and much safer alternatives to global NF-κB blockade as a way to effectively treat patients, both within and outside oncology. While the initial successes in the experimental settings have yet to transform into a clear healthcare benefit, the conceptual revolution conveyed in the approach to therapeutically targeting the NF-κB pathway is already providing tangible opportunities for developing effective new treatments in refractory disease indications. Indeed, if there is a lesson to be learnt from these initial successes, it is that the deep-rooted complexity in the NF-κB pathway may well hold the key to unlocking the gateway to generating clinically useful NF-κB-targeting medicines. Therefore, embracing, rather than evading, this complexity appears to the path to follow in order to finally seize the therapeutic potential still captured in the NF-κB pathway.</p><p>Recent reports suggest that one way of achieving this goal would be to exploit the contextual diversity of the transcriptional programmes NF-κB elicits in different cell types and, accordingly, inhibit the non-redundant, tissue-restricted downstream effectors of the NF-κB pathogenic functions. Although further clinical evaluation will ultimately determine its safety and clinical benefit, this approach has already added a firm string to the bow of the promising new therapies being developed to selectively inhibit NF-κB signalling in cancer. Additional attractive strategies are also appearing on the horizon of realising the contextual, cancer-cell selective inhibition of the NF-κB pathway by exploiting, for instance, the tissue-specific signalling mechanisms governing the contextual NF-κB activation in cells. Future research will tell whether NF-κB inhibitors will ever become part of the available anticancer arsenal. However, the significant advances recently made in this direction bode well for enabling this new reality in the near future.</p><!><p>The authors declare no conflict of interest.</p>

PubMed Open Access

Cyclic acetals as cleavable linkers for affinity capture

Labeling proteins with biotin is a widely used method to identify target proteins due to biotin\xe2\x80\x99s strong binding affinity for streptavidin. Combined with alkyne-azide cycloaddition, which enables the coupling of probes to targeted proteins, biotin tags linked to an alkyne or azide have become a powerful tool for purification and analysis of proteins in proteomics. However, biotin requires harsh elution conditions to release the captured protein from the bead matrix. Use of these conditions reduces signal to noise and complicates the analysis. To improve affinity capture, cleavable linkers have been introduced. Here, we demonstrate the use of a cyclic acetal biotin probe that is prepared easily from commercially available starting materials, is stable to cell lysates, yet is cleaved under mildly acidic conditions, and which provides an aldehyde for further elaboration of the protein, if desired.

cyclic_acetals_as_cleavable_linkers_for_affinity_capture

Introduction<!>Synthesis of cleavable biotin probe<!>Preparation of biotinylated BSA and evaluation of the cleavable acetal probes<!>Cleavage and capture in cell lysates<!>Suppression of streptavidin monomer release<!>Further labeling of the aldehyde tag on protein after cleavage<!>Conclusions<!>General<!>Acetal 1.30<!>Acetal 2<!>Trifluoroacetamide 3<!>Trifluoroacetamide 4<!>Trifluoroacetamide 5<!>Amine 6<!>Amide 7<!>Biotin-PEG10-N3, 8<!>Alkyne functionalized bovine serum albumin<!>Preparation of bovine serum albumin labeled with biotin probe<!>Cleavage test<!>Analysis of protein capture and release

<p>The azide moiety has been used as a functional group for bioorthogonal reaction because an azide and its reacting partner are not present in biological systems.1, 2 In a reaction termed a "click" reaction, an azide undergoes cycloaddition with an alkyne in the presence of Cu(I)3, 4 or with a highly ring strained alkyne, such as difluorinated cyclooctyne.4, 5 Once an azide and an alkyne form a triazole, the resulting triazole is stable to further reaction conditions such as reduction, oxidation, and hydrolysis.6, 7 Due to its fast reaction kinetics and exceptional functional group tolerance, the azide-alkyne click reaction has been chosen for many biological studies, e.g. selective protein modification in vitro and in vivo,8, 9 and activity-based protein profiling.10</p><p>Biotin is often used to capture the targeted protein due to its strong binding affinity for the egg-white glycoprotein avidin or to the bacterial protein streptavidin. Enrichment of biotinylated proteins from a complex mixture is efficiently achieved using a streptavidin-coated solid support such as an agarose resin. However, conventional methods to release biotinylated proteins from streptavidin bead matrices are harsh because of the strong binding interaction. For example, 2% SDS/6M urea,11 boiling in 2% SDS, or on-bead tryptic digestion are required to release the targeted protein. These non-selective conditions release streptavidin monomer and the proteins that are non-specifically bound to streptavidin bead matrices. Similarly, digested peptides of strepavidin from on-bead trypsin treatment contaminate the protein to be analysed. Even though the signal from streptavidin can be easily subtracted from the LC-MS/MS experiment for bioinformatics searches, the signals from non-specifically bound proteins will hamper the target protein search.</p><p>Recently, several biotin probes containing cleavable linkers have been developed to avoid such harsh elution conditions. Disulfide linkers have been widely used due to their rapid cleavage under mild reducing conditions. However, a disulfide linker is unstable to electrophilic and nucleophilic polar reagents, and thiol exchange with thiols in biological fluids can occur. Long-wave UV light can be used to release photocleavable linkers, but in some conditions, illumination of the sample is limited.11–13 There is also an acid labile linker from Pierce (proprietary structure) that is cleaved in 95% TFA. Another alternative is a dialkoxydiphenylsilane linker invented by Szychoswski et al. that is reported to be efficiently cleaved upon treatment with 10% formic acid for 0.5 h.14 Other types of cleavable linkers15, 16 have been developed such as an enzymatically (TEV) cleaved linker,17 diazobenzene-derived linkers that are cleaved with Na2S2O4,18–20 vicinal diol linkers cleaved with NaIO421 and linkers that are released upon reactions with nucleophiles including levulinoyl ester22 and nitrobenzenesulfonamide.23</p><p>In this report, we exploited the cyclic acetal moiety as an acid-sensitive linker. Orthoesters, ketals, and acetals are accepted cleavable linkers for drug delivery in vivo.24 In consideration of long term storage needs in combination with the requirements for stability in physiological conditions and fast cleavage for target identification, we designed cyclic acetals as acid-cleavable linkers (Figure 1). They are readily prepared by simple chemistry from commercially available starting materials. In addition, the resulting aldehyde after hydrolysis of the acetal can serve as a chemical reporter via further modification of the purified protein. Here we introduce the synthesis and capture utility of cyclic acetal linkers in two model systems.</p><!><p>To prepare the acetal-based cleavable biotin probe, azide-PEG8-amine was coupled to dimethoxy acetal 1 using carbonyl diimidazole as a coupling reagent to generate acetal 2 (Scheme 1). A dimethoxy acetal was used to generate the cyclic acetals rather than forming the cyclic acetal directly from the aldehyde because the kinetics of formation were more favorable. Basic alumina was employed for purification since the dimethoxy acetal is very sensitive to acid. In order to generate penta or hexacyclic acetal, protected 3-amine-1,2-diol or serinol, respectively, was coupled to acetal 2. The free amines in the diols were protected as their trifluoracetamides, 3 or 4. The cyclic acetal was formed using p-toluene sulfonic acid as a catalyst in THF/toluene to generate trifluoroacetamide 5. THF was used to dissolve diol 3 or 4 and dry toluene was used to remove water by azeotrope formation to drive the equilibrium toward cyclic acetal formation. The reaction was monitored by thin layer chromatography. Trifluoroacetamide 5 was obtained in 64% – 72% yield. The trifluoroacetamide moiety was removed to generate free amine for the following coupling reaction to form 6. Cyclic acetal 6 was purified on basic alumina in 82% yield. Finally, the amine was coupled to biotin activated with CDI to produce the final product 7a, 7b or 7c in 50% – 70% yield after gravity column chromatography (neutral alumina).</p><!><p>To examine the efficiency of capture with the acetal biotin probes, BSA was used as a model protein. BSA has one cysteine on the surface and an alkyne was installed on the thiol through N-propynyl-maleimide coupling. To facilitate complete coupling, 100 eq of N-propynyl-maleimide was used. After 12 hours, the remaining N- propynyl -maleimide was removed by precipitation of BSA using cold acetone. The alkyne-functionalized BSA was subjected to azide-alkyne cycloaddition with each of the biotin probes (Scheme 2). After 1 hour, excess click reagents were removed using a microconcentrator with a molecular weight cut off of 3 kDa.</p><p>In order to find effective cleavage conditions, each of the biotinylated BSA conjugates was incubated in 1% TFA at 37 °C with gentle agitation and aliquots were removed at 30 minutes, 1 hour, and 2 hours. The quantity of biotin remaining on the BSA was detected by streptavidin blot. The acetal BSA-7b was successfully cleaved in 30 minutes. On the other hand, non-acetal BSA-8 and acetal BSA-7a were stable to the cleavage conditions (Figure 2A). Upon treatment with higher concentrations of TFA, non-acetal probe BSA-8 was released from the resin (data not shown). Therefore, a higher concentration of TFA could not be used to cleave the more stable six-membered ring acetal BSA-7a selectively.</p><p>Acetal BSA-7b was further tested to evaluate the efficiency of cleavage when bound to streptavidin beads. BSA-8 was used as a negative control to confirm that the elution of the streptavidin protein is not responsible for cleavage. The BSA conjugated to probe 7b or 8 was captured on streptavidin ultralink resin for 1 hour at room temperature, and the beads were washed sequentially with 1% SDS in PBS, 6M urea in 250 mM ammonium bicarbonate, 1 M NaCl in PBS to remove as much non-specifically bound protein as possible, and finally, washed two times with water. The loaded beads were incubated in 1% TFA at 37 °C with gentle agitation. After 1 hour, the supernatant was collected, the resin was washed with 0.1% SDS in PBS and PBS, and all eluate and wash fractions were combined. After washing, the beads were boiled to release BSA that was not eluted during the cleavage procedure. As shown in Figure 2B, BSA-7b was successfully released from the streptavidin resin under the cleavage conditions. However, the cleavage of BSA-7b did not go to completion. In addition, a small amount of BSA-8 was released under the cleavage conditions, although the probe linker remained intact as evidenced by the biotin signal in the streptavidin blot. We reasoned that inefficient cleavage of 7b was due to limited solvent access to the acetal because the short linker between acetal and biotin places the acetal in close proximity to the biotin-binding pocket on the streptavidin. We investigated addition of various additives, e.g. SDS and guanidinium hydrochloride, to increase cleavage efficiency with limited success. Therefore, an extended linker with an additional seven atoms was introduced between the acetal and biotin by coupling NHS-LC-biotin with compound 6 (Scheme 1) to provide probe 7c.</p><p>We repeated the capture/cleavage procedures with streptavidin-ultralink resin to compare cleavage of BSA-7c to BSA-7b. However, the cleavage efficiencies of BSA-7b and BSA-7c were similar (data not shown). Because the pore size of the bead can also affect solvent access to acetal and dissociation of the product aldehyde, we tried a different capture medium, streptavidin-agarose beads. As shown in Figure 3A, the probe with the extended linker, 7c, was released more efficiently from the streptavidin-agarose bead complex than probe 7b which has a shorter linker. This result suggests that the combination of the extended linker and larger pore size are required to favor cleavage and dissociation of the acetal moiety.</p><!><p>Since the linker should be stable to physiological conditions and allow efficient capture from a complex mixture, the cleavage test was performed in the presence of bacterial cell lysates. Whole bacterial cell lysates (1 mg) mixed with BSA-7c (100 µg) was incubated with streptavidin beads. After washing as described above, the loaded beads were treated with 1% TFA at 37 °C with gentle agitation. After 1 hour, the supernatant was collected, the resin washed and protein eluted as described above for BSA-7b. The cyclic acetal linker 7c remained intact in the cell lysate, and allowed successful capture of the BSA conjugate on the bead matrix. Subsequent release with mild acid treatment efficiently yielded the cleaved BSA with little protein remaining on the bead (Figure 3B).</p><p>To further establish the optimal cleavage conditions, we tested capture of another protein, RNase A that has a low molecular weight, 13.7 kDa. RNase A has two free cysteines that were used to conjugate an alkyne handle via maleimide chemistry as described above for BSA. The alkyne was further conjugated with cleavable biotin probe 7c through azide-alkyne cycloaddition in the presence of Cu(I). RNase A-7c was captured from bacterial whole cell lysates and released as desired (Figure 4A). However, the eluted protein was contaminated with streptavidin monomer that was released from the bead matrix during the cleavage step (Figure 4A).</p><!><p>When the amount of a targeted protein to be eluted is low, co-elution of the streptavidin monomer should be minimized in order to avoid signal suppression and elevated noise in the subsequent mass spectral analysis. We observed during the course of our investigations to improve cleavage efficiency that 1M guanidinium hydrochloride improved the efficiency of cleaved protein release and suppressed the release of streptavidin. Therefore, we analyzed the use of guanidinium hydrochloride during cleavage and compared its use to cleavage conditions for other cleavable biotin probes in order to determine whether the problem of streptavidin monomer release is widespread.</p><p>Streptavidin agarose beads were incubated separately under the following cleavage conditions: 5% Na2S2O4 for 1 hour at 25°C, 2% of 2-mercaptoethanol for 1 hour at 25 °C, 5% formic acid for 2 hours at 25 °C, and 1M guanidine in 1% TFA at 37 °C for 1 hour. As shown in Figure 4B, formic acid treatment also resulted in the release of streptavidin from agarose beads, whereas, reducing conditions did not. Gratifyingly, inclusion of 1 M guanidine in the 1% TFA cleavage mixture suppressed non-specific release of the streptavidin monomer (Figure 4A, lane 6 vs Figure 4B, lane 4).</p><p>Next we tested the effect of guanidine concentration on release and elution. Two different concentrations, 1 M or 3 M guanidine hydrochloride, in combination with 1% TFA were tested. In both samples, suppression of streptavidin monomer release was observed. However, 3 M guanidine also releases some RNase A with the biotin probe still attached indicating that the RNase A release is due to protein denaturation rather than acetal cleavage (Figure 4C, lane 5), which results in reduced retention of biotinylated RNase A on the beads (Figure 4C, lane 7). We tested cleavage efficiency with BSA since other proteins can be sensitive to 1 M guanidine. 1 M guanidine did not affect the release of BSA from the streptavidin resin (Figure S1). Therefore, 1 M guanidine in 1% TFA is the preferred elution solvent for acetal linker release.</p><!><p>Aldehyde tags can be used to modify cell surface proteins specifically since the aldehyde functionality is not typically present in proteins.26 Aldehydes readily react with a variety of aminooxy or hydrazide-functionalized molecules.27, 28 The use of a capture linker that unmasks an aldehyde upon release can prove useful for further functionalization.29 Cleavage of the cyclic acetal linker 7 generates an aldehyde functionality on the tagged protein after purification. Therefore, we tested if the aldehyde is available for further modification of the protein. The cleaved BSA-7c was incubated with alkoxyamine-PEG-biotin for 4 hours, and the reaction mixture was directly analyzed by SDS-PAGE and streptavidin blot. After cleavage of the BSA-7c acetal, no BSA biotin signal remained (Figure 5, Lane 3). Upon reaction of the cleaved BSA with alkoxyamine-PEG-biotin, the biotin signal was restored (Figure 5, Lane 4). Likewise, RNase A-7c underwent the analogous reaction sequence (Figure S3). Therefore, the aldehyde funtionality generated through cleavage can be successfully conjugated with nucleophilic labeling reagents.</p><!><p>Existing methods to release biotinylated proteins from streptavidin bead matrices are harsh and often result in co-elution of high levels of contaminating proteins. The cyclic acetal, is cleaved under mildly acidic conditions, yet is sufficiently stable for use in cell lysates. Moreover, addition of guanidinium hydrochloride to the cleavage mixture suppresses release of monomeric streptavidin from the capture matrix under the acidic cleavage conditions. Lastly, cleavage of the cyclic acetal provides an aldehyde functionality on the captured protein, which conveniently can undergo further reaction to provide specific labeling of the captured protein.</p><!><p>Coupling reactions were performed under an Ar atmosphere using dry solvents. All commercially available reagents were purchased from Sigma-Aldrich and were used as received. NHS-LC-biotin and streptavidin agarose beads were purchased from Thermo Scientific. Spin ultrafilters (vivaspin 500) were purchased from GE Healthcare. 1H and 13C NMR spectra were recorded on Bruker instruments (400 or 500 MHz for 1H and 100 or 125 MHz for 13C). MS data were collected with AQUITY UPLC from Waters.</p><!><p>To a solution of 4-carboxybenzaldehyde (2.00 g, 13.3 mmol) in dry MeOH (40 mL) was added ammonium chloride (4.00 g, 74.8 mmol). The mixture was heated under reflux for 20 h. The solvent was evaporated under reduced pressure and the product was recrystallized from boiling hexane (2.0 g, 77 %): 199.5–200.0 °C, which was identical to the literature. 1H NMR (500 MHz, CD3OD) δ = 7.81 (d, J = 8.4 Hz, 2H), 7.51 (d, J = 8.1 Hz, 2H), 5.43 (s, 1H), 3.32 (s, 6H). 13C NMR (126 MHz, CDCl3) δ = 171.94, 143.86, 130.21, 129.61, 126.98, 102.27, 52.70; MS (m/z) [M-H]− calcd for C10H11O4: 195.07, found: 195.1.</p><!><p>Carbonyldiimidazole (88.7 mg, 0.54 mmol) and 1 (107.4 mg, 0.54 mmol) were dissolved in DCM. The mixture was stirred for 30 min at rt. To the solution was added azido-PEG8-amine (200 mg, 0.45 mmol). After 5 h, the solvent was evaporated under reduced pressure. Product 2 was obtained by gravity column chromatography (basic alumina, 0% – 5% MeOH/DCM) as an oil (210 mg, 75 %): 1H NMR (500 MHz, CDCl3) δ = 7.78 (2H, d, J = 8.6 Hz, 2H), 7.47 (d, J = 9.3 Hz, 2H), 6.92 (b, 1H), 5.43 (s, 1H), 3.73 – 3.56 (m, 34H), 3.42 – 3.35 (m, 2H), 3.32 (s, 6H); 13C NMR (126 MHz, CDCl3) δ = 167.21, 141.34, 134.71, 127.05, 126.92, 102.39, 70.72, 70.69, 70.66, 70.61, 70.59, 70.58, 70.55, 70.31, 70.07, 69.81, 52.65, 50.70, 39.83; MS (m/z) [M+H]+ calcd for C28H49N4O11: 617.34, found: 617.48.</p><!><p>To a solution of 3-amino-1,2-propanediol (250 mg, 2.74 mmol) in THF, ethyl trifluoroacetate (2.33 g, 16.46 mmol) was added drop-wise. After 4 h, the solvent was evaporated. DCM was added to the oil and evaporated. This step was repeated two more times. Benzene was added and evaporated. This step was also repeated two more times. The resulting product was used without further purification to yield compound 3: 1H NMR (500 MHz, CDCl3) δ = 4.85–4.7 (m, 1H); 3.6–3.3 (m, 4H); MS (m/z): [M-H]− calcd for C5H7F3NO3: 186.04, found: 185.98.</p><!><p>Trifluoracetamide 4 was prepared from serinol (250 mg, 2.74 mmol) and ethyl trifluoroacetate (2.33 g, 16.46 mmol) as described for 3 to yield compound 4: 1H NMR (500 MHz, DMSO-d6) δ = 8.99 (m, 1H),, 4.75 (t, J = 7.1 Hz, 1H), 3.86–3.477 (m, 1H), 3.52–3.38 (m, 4H); MS (m/z): [M-H]− calcd for C5H7F3NO3: 186.04, found: 185.98.</p><!><p>To a solution of 3 or 4 (191 mg, 1.022 mmol) in THF/toluene (3/7), 2 (210 mg, 0.34 mmol) and p-toluene sulfonic acid•H2O (13 mg, 0.068 mmol) were added. The mixture was heated to 100 °C. The solvent was distilled to remove H2O generated during the reaction and toluene added to maintain reaction volume as the reaction proceeded. After 4 h, the reaction was quenched with 50 µl of TEA. Product 5 was obtained by column chromatography (basic alumina, 0%–5% MeOH/DCM) as an oil: 5a (from 4, 180 mg, 72 %): 1H NMR (500 MHz, CD3OD) δ = 7.86 – 7.83 (m, 2H), 7.62 – 7.55 (m, 2H), 5.69, 5.55 (s, 1H, two isomers), 4.30 – 4.21 (m, 4H), 3.87 – 3.81 (m, 1H), 3.68 –3.56 (m, 34H), 3.36 (t, J = 5.0 Hz, 2H); 13C NMR (126 MHz, CD3OD) δ = 169.81, 169.78, 159.14, 159.12, 158.84, 142.88, 142.49, 136.15, 130.66, 129.15, 128.41, 128.33, 128.28, 128.25, 128.21, 128.03, 127.99, 127.97, 127.55, 127.53, 127.51, 127.44, 127.42, 118.56, 116.29, 102.13, 102.02, 101.79, 71.63, 71.57, 71.56, 71.55, 71.51, 71.33, 71.13, 70.51, 70.50, 69.98, 69.95, 69.67, 61.64, 57.32, 51.77, 47.06, 45.12, 44.29, 41.03, 36.51, 27.10, 26.50; MS (m/z): [M+H]+ calcd for C31H49F3N5O12: 740.36, found: 740.66. 5b (from 3, 160 mg, 64 %): 1H NMR (500 MHz, CD3OD) δ = 7.87 – 7.84 (m, 2H), 7.61 – 7.54 (m, 2H), 5.99, 5.82 (1H, two isomers), 4.48 – 4.43 (m, 1H), 4.22 4.12 (dd, J = 8.6, 6.4 Hz, 1H, two isomers), 3.94, 3.78 (dd, J = 8.5, 5.0 Hz, 1H, two isomers), 3.68 – 3.58 (m, 34H), 3.55 – 3.49 (m, 2H), 3.40 – 3.33 (m, 2H); 13C NMR (126 MHz, CD3OD) δ = 169.78, 169.75, 159.55, 159.50, 159.26, 159.20, 142.96, 142.11, 136.58, 136.39, 130.65, 129.14, 128.41, 128.37, 128.06, 127.73, 118.65, 116.37, 104.97, 104.10, 76.04, 76.01, 71.62, 71.56, 71.55, 71.54, 71.50, 71.32, 71.13, 70.50, 69.21, 69.12, 54.84, 51.76, 43.31, 42.58, 41.04; MS (m/z): [M+NH4]+ calcd for C31H52F3N6O12: 757.36, found: 757.55.</p><!><p>To a solution of 5a or 5b (160 mg, 0.22 mmol) in MeOH/H2O (7/3) K2CO3 (209.35 mg, 1.5149 mmol) was added. The reaction was heated at reflux for 2 h. After evaporating all the solvent, the product was purified by gravity column chromatography (basic alumina, 2%–10% MeOH/DCM) to yield 6 as an oil. 6a (from 5a, 134 mg, 82 %): 1H NMR (500 MHz, CDCl3) δ = 7.88 – 7.75 (m, 2H), 7.58 – 7.48 (m, 2H), 5.52, 5.40 (s, 1H, two isomers), 4.37 – 4.00 (m, 4H), 3.67 – 3.55 (m, 34H), 3.35 (q, J = 4.6 Hz, 2H), 3.30 – 3.10 (m, 1H); [M+H]+ calcd for C29H50N5O11: 644.34, found: 644.49. 6b (from 5b, 105 mg, 80 %): 1H NMR (400 MHz, CDCl3) δ = 7.82 – 7.78 (m, 2H), 7.57 – 7.46 (m, 2H), 6.90 (b, 1H), 5.95, 5.82 (s, 1H, two isomers), 4.32, 4.16 (m, 1H, two isomers), 4.09 (t, J = 10.0 Hz 1H), 3.89, 3.69 (m, 1H, two isomers), 3.66 – 3.57 (m, 34H), 3.35 (t, J = 5.0 Hz, 2H), 3.01 – 2.78 (m, 2H). MS (m/z): [M+H]+ calcd for C29H50N5O11: 644.34, found: 644.57.</p><!><p>d-Biotin or NHS-LC-biotin (50 mg, 0.205 mmol) and carbonydiimidazole (33 mg, 0.205 mmol) were dissolved in dried DMF. The mixture was stirred for 30 min. To the mixture, 6 was added and the reaction was stirred for 12 h at rt. The product was purified by gravity column chromatography (neutral alumina, 3%–7% MeOH/DCM) to yield 7 as an oil. Compound 7a (90 mg, 50 %): 1H NMR (400 MHz, CD3OD) δ = 7.86 – 7.83 (m, 2H), 7.66 – 7.53 (m, 2H), 5.67, 5.52 (s, 1H, two isomers), 4.64 – 4.38 (m, 1H), 4.33 – 4.10 (m, 4H), 3.81 (s, 1H), 3.55 – 3.70 (m, 35H), 3.36 (t, J = 4.9 Hz, 2H), 3.26 –3.12 (m, 1H), 2.96 – 2.85 (m, 1H), 2.70 (t, J = 12.4 Hz, 1H), 2.40 – 2.19 (m, 2H), 1.78 – 1.25 (m, 6H); 13C NMR (126 MHz, CD3OD) δ = 197.12, 176.01, 169.77, 143.12, 142.70, 142.43, 136.06, 128.21, 128.17, 127.54, 127.51, 102.09, 101.72, 71.61, 71.55, 71.53, 71.52, 71.49, 71.32, 71.30, 71.12, 70.96, 70.93, 70.51, 63.25, 61.57, 56.97, 54.82, 51.77, 41.02, 36.44, 29.47, 26.92; HRMS ESI (m/z): [M+H]+ calcd for C39H64N7O13S: 870.4277, found: 870.4295. Compound 7b (100 mg, 67%): 1H NMR (400 MHz, CD3OD) δ = 7.88 – 7.84 (m, 2H), 7.62 – 7.55 (m, 2H), 5.98, 5.81 (s, 1H, two isomers), 4.48 – 4.32 (m, 2H), 4.25 – 4.20 (m, 1H), 4.12 – 4.07 (m, 1H), 3.92 – 3.86 (m, 1H), 3.78 – 3.72 (m, 1H), 3.69 – 3.56 (m, 34H), 3.51 – 3.41 (m, 2H), 3.37 (t, J = 4.9 Hz, 2H), 3.19 – 3.09 (m, 1H), 2.92 – 2.84 (m, 1H), 2.70 (d, J = 12.8 Hz, 1H), 2.30 – 2.18 (m, 2H), 1.78 – 1.35 (m, 6H); 13C NMR (126 MHz, CD3OD) δ = 176.41, 169.69, 166.04, 143.15, 142.38, 136.54, 136.39, 128.41, 128.12, 127.81, 122.64, 104.78, 104.09, 76.91, 76.71, 71.63, 71.57, 71.56, 71.54, 71.33, 71.13, 70.52, 69.29, 63.28, 61.59, 57.00, 50.97, 50.78, 48.39, 41.06, 41.04, 36.74, 29.72, 29.67, 29.49, 29.48, 26.88, 26.86; HRMS ESI (m/z): [M+H]+ calcd for C39H64N7O13S: 870.4277, found: 870.4288. Compound 7c (70 mg, 61%): 1H NMR (500 MHz, CD3OD) δ = 7.88 – 7.84 (m, 2H), 7.62 – 7.55 (m, 2H), 5.98, 5.82 (s, 1H, two isomers), 4.49 – 4.46 (m, 1H), 4.39 – 4.33 (m, 1H), 4.31 – 4.26 (m, 1H),), 4.22 – 4.07 (m, 1H, two isomers), 3.89 – 3.73 (m, 1H, two isomers), 3.70 – 3.55 (m, 34H), 3.51 – 3.35 (m, 4H), 3.22 – 3.10 (m, 3H), 2.93 – 2.90 (m, 1H), 2.70 (d, J = 12.8 1H), 2.25 – 2.17 (m, 4H), 1.75 – 1.31 (m, 15H); 13C NMR (126 MHz, CD3OD) δ = 176.50, 175.93, 169.78, 166.07, 143.16, 142.39, 136.53, 136.36, 128.41, 128.39, 128.08, 127.75, 104.78, 104.07, 76.89, 76.66, 71.63, 71.56, 71.54, 71.51, 71.33, 71.14, 70.51, 69.27, 69.20, 63.36, 61.60, 57.00, 51.77, 42.78, 42.13, 41.05, 40.19, 40.14, 36.87, 36.85, 36.81, 34.66, 30.13, 30.12, 30.09, 29.79, 29.50, 27.59, 27.46, 26.92, 26.63, 25.68; HRMS ESI (m/z): [M+H]+ calcd for C45H75N8O14S: 983.5123, found: 983.5152.</p><!><p>Biotin-NHS (0.31 mmol, 107 mg) and O-(2-Aminoethyl)-O′-(2-azidoethyl)nonaethylene glycol (0.21 mmol, 110 mg) were dissolved in 1 mL dry DMF. DIEA (0.31 mmol, 56 µL) was added to the mixture, and the reaction was stirred for 16 h at rt. After evaporation of solvent, the product was precipitated with Et2O. Chromatography (MeOH:EtOAc/1:1) yielded product 8: 1H NMR (500 MHz, d6-DMSO) δ = 7.81 (t, J = 5.5, 1H), 6.40 (br s, 1H), 6.34 (br s, 1H), 4.30 (m, 1H), 4.12 (m, 1H), 3.60 (m, 2H), 3.53 (m, 38H), 3.39 (t, J= 5.1, 4H), 3.18 (q, J= 5.8, 2H), 3.09 (dd, J= 11.7, 7.3, 1H), 2.82 (dd, J= 12.4, 5.1, 1H), 2.58 (d, J= 12.4, 1H), 2.06 (t, J= 7.4, 2H), 1.62 (dd, J= 21.4, 7.9, 1H), 1.50 (dt, J = 14.4, 7.5, 3H), 1.30 (m, 2H); MS (m/z): [M+H]+ calcd for C32H61N6O12S: 753.4, found: 753.4.</p><!><p>To a solution of bovine serum albumine (20 µM) in PBS, N-(1-propynyl)-maleimide (2 mM) was added. The mixture was gently agitated for 12 h in the dark. The excess maleimide was removed using cold acetone.</p><!><p>Alkyne functionalized BSA (50 µM) was mixed with biotin probe 7a or 7b (100 µM), BTTP25 (200 µM), CuSO4 (100 µM), and sodium ascorbate (2.5 mM) for 1 h at rt. The reagents were removed using an ultrafiltration spin filter (MWCO = 3 kDa). The concentration of BSA was measured by Pierce Coomassie Plus protein assay, following the manufacturer's instructions.</p><!><p>After coupling of BSA to probe, the mixture (about 100 µg BSA) was incubated with streptavidin-ultralink resin or streptavidin-agarose resin (100 µL of 50% (v/v) slurry in PBS) for 1 h at rt. The beads loaded with biotinylated BSA were spun at 1000 g for 3 min. The pelleted beads were washed sequentially with 1% SDS in PBS, 6 M urea in 250 mM ammonium bicarbonate, 1 M NaCl in PBS, and two times with water. The beads were incubated in 1 mL 1% TFA for 1 h at 37 °C. The supernatant was collected by pelleting the beads. The beads were washed sequentially with 0.1% SDS/PBS, and two times with PBS and the supernatant was combined with the washes. The combined solutions were concentrated using an ultrafiltration spin filter (MWCO = 3 kDa) at 7000 g. After two more washes with PBS, the beads were boiled in sample loading buffer for 15 min.</p><!><p>Each protein sample (typically equivalent to about 20 µg BSA in the initial sample per lane) was separated by 12% or 15% SDS-PAGE gel and transferred onto a PDVF membrane (Bio Rad). The membrane was blocked with 4% BSA/TBST for 1 h at rt. After washing the membrane with TBST three times, streptavidin conjugated with Alexa-488 (20 µg/mL) was added and the solution was gently agitated for 1 h at 25°C. The membrane was washed with TBST three times and was visualized using a Typhoon 9400 scanner (GE Healthcare).</p>

PubMed Author Manuscript

1,2,5,6‐Tetrakis(guanidino)‐Naphthalenes: Electron Donors, Fluorescent Probes and Redox‐Active Ligands

AbstractNew redox‐active 1,2,5,6‐tetrakis(guanidino)‐naphthalene compounds, isolable and storable in the neutral and deep‐green dicationic redox states and oxidisable further in two one‐electron steps to the tetracations, are reported. Protonation switches on blue fluorescence, with the fluorescence intensity (quantum yield) increasing with the degree of protonation. Reactions with N‐halogenosuccinimides or N‐halogenophthalimides led to a series of new redox‐active halogeno‐ and succinimido‐/phthalimido‐substituted derivatives. These highly selective reactions are proposed to proceed via the tri‐ or tetracationic state as the intermediate. The derivatives are oxidised reversibly at slightly higher potentials than that of the unsubstituted compounds to dications and further to tri‐ and tetracations. The integration of redox‐active ligands in the transition‐metal complexes shifts the redox potentials to higher values and also allows reversible oxidation in two potentially separated one‐electron steps.

1,2,5,6‐tetrakis(guanidino)‐naphthalenes:_electron_donors,_fluorescent_probes_and_redox‐active_ligan

<!>Introduction<!><!>Introduction<!>Synthesis and characterisation<!><!>Synthesis and characterisation<!><!>Synthesis and characterisation<!><!>Redox properties<!><!>Redox properties<!>Isolation of the dicationic redox state<!><!>Isolation of the dicationic redox state<!><!>Isolation of the dicationic redox state<!>Aromatic substitution<!><!>Aromatic substitution<!><!>Aromatic substitution<!><!>Protonation‐induced fluorescence<!><!>Protonation‐induced fluorescence<!>Redox‐active coordination compounds<!><!>Redox‐active coordination compounds<!><!>Redox‐active coordination compounds<!>Conclusion<!>Materials and methods<!>Synthesis<!>Details of the quantum chemical calculations<!>Details of the structural characterisations<!>Conflict of interest<!>

<p>L. Lohmeyer, E. Kaifer, H. Wadepohl, H.-J. Himmel, Chem. Eur. J. 2020, 26, 5834.</p><!><p>There is a huge demand for redox‐active compounds that exhibit two (or more) stable redox states.1 The redox states generally differ in their colour,2 and the possibility to interconvert them electrochemically allows their use in electrochromic switching3 and memory devices.4 Also, applications in organic redox flow batteries depend on the stability of (at least) two redox states.5, 6, 7</p><p>In the last decade, we developed guanidino‐functionalised aromatics (GFAs) as a new class of redox‐active organic molecules,8, 9 starting with 1,2,4,5‐tetrakis(tetramethylguanidino)‐benzene (1, Figure 1).10 Compound 1 is a relatively strong organic electron donor with a redox potential E 1/2 (in CH3CN) versus ferrocenium/ferrocene (Fc+/Fc) of −0.73 V and also a strong Brønsted base with an estimated pK a value of 25.3 for (1+H)+ in CH3CN. A number of stable, storable salts of the oxidised compound, for example, 1(BF4)2, 1(PF6)2, 1(I3)2, 1{N(CN)2}2 11 and others, are known. GFA compounds with redox potentials (E 1/2) up to approximately −1 V have been since synthesised.8, 12 In their neutral state, they were applied in (photochemical) redox reactions13 and in their oxidised state in proton‐coupled electron transfer (PCET) reactions.14, 15, 16 Furthermore, they could be employed as redox catalysts for the oxidation of organic substrates with dioxygen as a terminal oxidant.14 A variety of mono‐ and bimetallic late‐transition‐metal complexes with redox‐active GFA ligands have been reported, and it is possible to switch by temperature, solvent polarity or by chemical reactions between redox‐isomeric CuI complexes with oxidised GFA ligands and CuII complexes with neutral GFA ligands.17</p><!><p>Lewis structures of the known redox‐active guanidines 1,2,4,5‐tetrakis(tetramethylguanidino)‐benzene (1), 1,4,5,8‐tetrakis(tetramethylguanidino)‐naphthalene (2) and 1,4,5,8‐tetrakis(N,N′‐dimethylethyleneguanidino)‐naphthalene (3), as well as the new guanidines 1,2,5,6‐tetrakis(tetramethylguanidino)‐naphthalene (4) and 1,2,5,6‐tetrakis(N,N′‐dimethylethyleneguanidino)‐naphthalene (5) that are isomers of compounds 2 and 3.</p><!><p>The two 1,4,5,8‐tetrakis(guanidino)‐naphthalene derivatives 2 and 3 (see Figure 1)18 are slightly weaker electron donors compared to 1 (e.g., E 1/2=−0.65 V vs. Fc+/Fc for 2 in CH2Cl2)9 but are double‐proton sponges with very large pK a values (estimated to be 27.4 for (2+H)+ in CH3CN). For (2+2H)2+, the positions of the protons in (asymmetric) N−H⋅⋅⋅N bridges were confirmed by NMR spectroscopic studies in solution and structural characterisation in the solid state.19 However, so far only a few stable, storable compounds of the dication 2 2+ are known, for example, the salt 2(I3)2 and the complex [2(CuBr2)2] (a dinuclear CuI complex with the dication 2 2+ as bridging ligand),20 hampering the use of the oxidised compound up to date.</p><p>In this work, the synthesis and chemistry of the redox‐active guanidines 1,2,5,6‐tetrakis(tetramethylguanidino)‐naphthalene (4) and 1,2,5,6‐tetrakis(N,N′‐dimethylethyleneguanidino)‐naphthalene (5) are reported (Figure 1). The properties and chemical reactivity of these new GFA compounds will be shown to differ significantly from those of their isomers 2 and 3.</p><!><p>The new tetrakisguanidines 4 and 5 were synthesised in a four‐step procedure according to Scheme 1. The synthesis of the direct precursor 1,2,5,6‐tetraamino‐naphthalene, starting with 2,6‐dibromo‐naphthalene, deviated from the previously reported synthesis21 and followed the synthesis reported by Stille et al.22 The commercially available 2,6‐dibromo‐naphthalene was nitrated analogous to that reported by Shepherd et al.23 by using concentrated HNO3 to obtain the required isomer in at least 60 % yield. The substitution of the bromine atoms through amino groups was achieved by using benzophenone imine as an ammonia surrogate in a Buchwald–Hartwig‐type amination, instead of gaseous ammonia in a high‐pressure, high‐temperature reaction, because of better yields and an easier work‐up. Conveniently, the cleavage of the benzophenone imine and the reduction of the nitro groups could be carried out simultaneously by using SnCl2 in concentrated hydrochloric acid, giving 1,2,5,6‐tetraaminonaphthalene tetrahydrochloride in 95 % yield. In the last step, this compound was reacted with an "activated urea" compound (from reaction between oxalyl chloride and a urea24), either 2‐chloro‐1,1′,3,3′‐tetramethylformamidinium chloride to give 4 (31 % isolated yield) or 2‐chloro‐1,3‐dimethylimidazolinium chloride to give 5 (52 % isolated yield).</p><!><p>Synthesis of the new redox‐active guanidines 4 and 5 (BINAP=2,2′‐bis(diphenylphosphino)‐1,1′‐binaphthyl; dba=dibenzylideneacetone).</p><!><p>Crystals of 4 were grown by diffusing diethyl ether into a saturated acetonitrile solution. Crystals of 5 were obtained from a saturated acetonitrile solution. Figure 2 illustrates the solid‐state structures of 4 and 5. The CN3 planes of the guanidino groups are highly twisted (almost perpendicular) with respect to the aromatic‐ring plane, and the C1‐N1‐C6/C11 and C2‐N4‐C11/C16 bond angles at the nitrogen atoms attached to the aromatic core (120.27(11)° and 119.69(11)° for 4 and 127.20(16)° and 122.95(15)° for 5) display significant deviation from a linear geometry. This "bent‐twisted" structure is in line with the structures of other GFA compounds.8, 25 The alternative "bent‐planar" structure, in which the guanidino CN3 plane is in the aromatic plane, is disfavoured by steric strain. As a consequence of the "bent‐twisted" conformation, there is no steric strain in molecules with two guanidino groups ortho to each other (in contrast to two ortho‐positioned dimethylamino groups).8a</p><!><p>Illustration of the solid‐state structures of compounds 4 and 5. Displacement ellipsoids are drawn at the 50 % probability level. Hydrogen atoms are omitted. Selected bond parameters can be found in Table 1 (see below).</p><!><p>Quantum chemical calculations using the B3LYP functional in combination with the def2‐TZVP basis set were carried out to compare the relative energy of isomers 2 and 4. The energy difference was calculated to be 32 kJ mol−1 in favour of 4. Figure 3 compares the isodensity surfaces of the frontier orbitals (HOMO‐1, HOMO and LUMO) of 2 and 4. For both molecules, the HOMO is localised primarily on the aromatic core. The relatively high HOMO energies (−3.93 eV for 2 and −4.05 eV for 4) are in line with the observed electron donor character (see below). In the case of the HOMO‐1 orbitals, the isodensity plots exhibit for both molecules one nodal plane but show a different electron‐density distribution. For 2, the nodal plane intersects the three parallel C−C bond axes of the naphthalene core. In the case of 4, the nodal plane intersects only the central C−C bond and includes two diagonal C atoms of the naphthalene core. These differences also affect the structures of the dicationic molecules (see discussion below).</p><!><p>Illustrations of the isodensity surfaces for some frontier orbitals of 2 and 4.</p><!><p>Cyclic voltammetry (CV) measurements of compound 5 (Figure 4) found one quasi‐reversible two‐electron wave at E 1/2=−0.47 V (E ox=−0.43 V) versus ferrocenium/ferrocene (Fc+/Fc) for the redox couple 5/5 2+ in CH2Cl2. For compound 4 the two‐electron oxidation and reduction waves seem to be slightly split, arguing for two one‐electron steps at potentials of E 1/2=−0.51 V (4/4 .+) and E 1/2=−0.40 V (4 .+/4 2+). Hence, the new compounds are slightly weaker electron donors than compounds 2 or 3 (E 1/2=−0.65 V vs. Fc+/Fc for 2/2 2+ and E 1/2=−0.71 V vs. Fc+/Fc for 3/3 2+).18, 19 We have previously found a simple correlation between the redox potentials and the HOMO energies of differently substituted GFA compounds, according to which more negative redox potentials directly correlate with higher HOMO energies.26 Here again, compound 2 has the more negative redox potential and the higher HOMO energy compared with that of 4 (see Figure 3). By measuring a CV curve for 4 and 5 to higher potentials (see Supporting Information), two more (reversible) one‐electron oxidation steps are visible. These are assigned to the redox couples 4 2+/4 .3+ (E 1/2=0.27 V, E ox=0.32 V) and 5 2+/5 .3+ (E 1/2=0.25 V, E ox=0.28 V), as well as the redox couples 4 .3+/4 4+ (E 1/2=0.40 V, E ox=0.45 V) and 5 .3+/5 4+ (E 1/2=0.48 V, E ox=0.51 V). In addition, broad shoulders towards higher potentials of the last oxidation step generating the tetracation are visible. As shown in the next section, the tetracation is barely soluble in standard organic solvents and is extremely reactive. Therefore, these additional features might arise from segregation effects at the electrode and/or decomposition of the highly reactive tri‐ or tetracationic compound. This would also explain the relatively low reduction currents.</p><!><p>CV curves for 4 and 5 in CH2Cl2 (Ag/AgCl reference electrode, 0.1 m N(nBu)4(PF6) as supporting electrolyte, scan rate=100 mV s−1). All curves are measured in the direction of oxidation. Potentials are given versus the Fc+/Fc couple.</p><!><p>The combination of electrochemical oxidation with UV/Vis spectroscopy (spectro‐electrochemistry) was used to obtain information about the oxidised species. These measurements show that the two‐fold oxidised compounds exhibit broad absorptions in the Vis/NIR region, with a maximum of absorption at 778 nm for compound 5 2+ in CH2Cl2 solution (see Supporting Information). Electronic excitation energies in the visible region signal the presence of an extended π‐conjugated system but removal of aromaticity. For comparison, upon two‐electron oxidation, compound 1 shows strong absorptions near 430 nm. The lower electronic excitation energy of 5 2+ compared with that of 1 2+ could be rationalised by the larger π‐conjugated system in 5 2+.</p><!><p>Motivated by the reversibility of the redox processes in the CV experiments, we next tried to oxidise compound 5 chemically by reaction with two equivalents of NO(BF4). Indeed, neutral 5 was oxidised cleanly to the dication 5 2+, and the product salt 5(BF4)2 was isolated in 84 % yield. In Figure 5, the UV/Vis spectra of neutral 5 (reduced form) and 5(BF4)2, dissolved in CH2Cl2, are compared. Oxidation leads to the appearance of a broad absorption in the visible region (λ max=778 nm) and a small absorption around λ max=458 nm, in line with the results from spectro‐electrochemical measurements. In combination with strong absorptions centred below 400 nm that extend into the visible region, a deep‐green colour results.</p><!><p>UV/Vis spectra for CH2Cl2 solutions of 5 and 5(BF4)2. Photos of solutions of 5 (left) and 5(BF4)2 (right) in CH2Cl2 are shown in the insets.</p><!><p>It was not possible to grow crystals of 5(BF4)2 suitable for a structural analysis by single‐crystal X‐ray diffraction (XRD) measurements. Therefore, we repeated the oxidation of 5 with Ag(SbF6) and were able to isolate suitable crystals of the product 5(SbF6)2 in 88 % yield (Figure 6). In Table 1, some bond lengths of neutral 5 and the dication 5 2+ in solid 5(SbF6)2 are compared. The variations in the C−C bond lengths within the central ring system indeed signal loss of aromaticity. Some of the C−C bond lengths are significantly elongated upon oxidation. For example, the C1−C2 bond length increases from 1.396(2) Å before oxidation to 1.475(2) Å after oxidation. The C1−N1 bond length decreases to 1.326(2) Å, and the C2−N4 bond length decreases to 1.355(2) Å. The structure is in line with the Lewis representation drawn in Scheme 2. The "bent‐twisted" conformation of the guanidino groups, which was already found for the neutral molecule, is preserved in the dicationic form; the CN3 planes of each guanidino group are highly twisted to the central ring plane, C1‐N1‐C6 and C2‐N4‐C11 bond angles of 119.8(8) and 126.7(9)°, respectively. In this conformation, the N1/N4 atoms could establish a π‐bond with the C1/C2 atoms and at the same time establish some π‐interactions with the C6/C11 atoms.25</p><!><p>Illustration of the structure of one dication 5 2+ and two SbF6 − counterions of 5(SbF6)2 in the solid state. Displacement ellipsoids are drawn at the 50 % probability level. Hydrogen atoms are omitted. Selected bond parameters can be found in Table 1.</p><p>Comparison between selected bond lengths [Å] in crystalline neutral 4 and 5, the salt 5(SbF6)2 of the two‐fold oxidised guanidine and the complex [5{Pd(OAc)2}2].</p><p>Parameter</p><p>4</p><p>5</p><p>5(SbF6)2</p><p>[5{Pd(OAc)2}2]</p><p>C1−C2</p><p>1.391(2)</p><p>1.396(2)</p><p>1.475(2)</p><p>1.371(8)</p><p>C1−C5/C10</p><p>1.419(2)</p><p>1.421(2)</p><p>1.432(2)</p><p>1.400(8)</p><p>C2−C3</p><p>1.433(2)</p><p>1.435(2)</p><p>1.410(2)</p><p>1.438(7)</p><p>C3−C3′/C8</p><p>1.429(2)</p><p>1.428(2)</p><p>1.413(2)</p><p>1.440(2)</p><p>C3′−C4</p><p>1.418(2)</p><p>1.423(2)</p><p>1.410(2)</p><p>1.407(7)</p><p>C4−C5</p><p>1.374(2)</p><p>1.371(3)</p><p>1.369(2)</p><p>1.367(7)</p><p>C1−N1</p><p>1.419(2)</p><p>1.411(2)</p><p>1.326(2)</p><p>1.410(7)</p><p>C2−N4</p><p>1.408(2)</p><p>1.418(2)</p><p>1.355(2)</p><p>1.424(7)</p><p>Oxidation of 5 with two equivalents of NO(BF4) or Ag(SbF6) to form 5 2+.</p><!><p>Preparative oxidation of 5 with four equivalents of Ag(SbF6) in dichloromethane led to a purple precipitate insoluble in dichloromethane and 1,2‐difluorobenzene. Attempts to dissolve the possible tetracation in acetonitrile, diethyl ether, tetrahydrofuran or dimethylformamide resulted in a green solution indicating reduction to the dication, as confirmed through NMR spectroscopy. Given that all our attempts to dissolve 5(SbF6)4 without conversion failed, it was not possible to isolate a pure compound.</p><!><p>Interestingly, reaction of 5 with N‐halogenosuccinimides (NXS) or N‐bromophthalimide (NBP) did not only lead to halogeno‐substituted compounds but also to selective substitution by succinimido/phthalimido groups. After reduction with hydrazine, the neutral 1,2,5,6‐tetrakis(dimethylethyleneguanidino)‐3,7‐dihalogeno‐4,8‐disuccinimido‐naphthalenes (6 and 7) and 1,2,5,6‐tetrakis(dimethylethyleneguanidino)‐3,7‐dibromo‐4,8‐diphthalimido‐naphthalene (8) were obtained in good yields (see Scheme 3), in which all aromatic hydrogen atoms are substituted. Only one of several possible isomers is formed in all reactions. By contrast, reaction of 1 with NXS compounds and subsequent reduction with hydrazine resulted in dihalogenated derivatives of 1.9 The difference can be rationalised by the steric shielding effect of the guanidino groups ortho to a C−H proton, allowing substitution only by sterically less‐demanding halogen atoms.</p><!><p>Reaction of 5 with N‐halogenosuccinimides (NXS) and N‐bromophthalimide (NBP) and subsequent reduction with hydrazine, leading to the new compounds 6–8.</p><!><p>Compounds 6–8 were crystallised and structurally characterised (Figure 7). Owing to the orientation of the substituents (i.e., all ring planes of the substituents are almost perpendicular to the central naphthalene ring plane), there is no steric constraint in these compounds.</p><!><p>Illustration of the structures of the compounds 6–8 and the crystallised product of the reaction of 5 with four equivalents of N‐iodosuccinimide (NIS) (i.e., the reaction intermediate) in the solid state. Displacement ellipsoids are drawn at the 50 % probability level for 7, 8 and the crystallised reaction intermediate and are drawn at the 30 % probability level for 6. Hydrogen atoms are omitted. Colour code: N=blue, C=grey, O=red, Br=green, I=pink.</p><!><p>In a preliminary effort to rationalise the pathway of these intriguing reactions, we carried out experiments with different amounts of N‐iodosuccinimide. By applying one equivalent, we could only observe an oxidation of 5 to the dication but no substitution of the aromatic hydrogen atoms. On the other hand, the doubly succinimido‐substituted oxidised product (see Figure 7) crystallised (together with two iodide counterions) from a reaction mixture of 5 with four equivalents of N‐iodosuccinimide. Hence, the positions 4 and 8 are substituted first, not with a halogen but with succinimide. Using more than six equivalents had no further impact on the reaction. To identify the reactive species, we tried to react 5(SbF6)2 with two equivalents of potassium phtalimide but observed no conversion. This result means that phthalimide substitution is possible only after substrate oxidation, given that it does not occur from the neutral form of the substrate. According to these results, we propose a multiple‐step pathway, starting with the oxidation of 5 to either 5 .3+ or 5 4+ by two equivalents of the used reagent, followed by substitution at the positions 4 and 8 by the produced succinimide/phthalimide anion and deprotonation. The resulting disubstituted compound is again oxidised by two more equivalents of the reagent, substituted by the corresponding halides at positions 3 and 7, and deprotonated. Final reduction with hydrazine gives the neutral end‐product. The preference for halide over succinimide/phthalimide substitution in the last substitution steps is presumably due to steric effects.</p><p>CV measurements in CH2Cl2 solutions were also carried out for compounds 6–8 (Figure 8). As expected, the substitution with electron‐withdrawing groups shifts the redox potential to slightly higher values (E 1/2=−0.30 V (E ox=−0.25), −0.25 V (E ox=−0.18) and E ox=−0.18 V vs. Fc+/Fc in CH2Cl2 for 6, 8 and 7, respectively). Similar to the free ligands 4 and 5, for all compounds two more (reversible) one‐electron oxidation steps were visible at higher potentials (see Supporting Information). These are assigned to the redox couples 2+/.3+ and .3+/4+. For all three compounds, the redox couples 2+/.3+ as well as the redox couples .3+/4+ are (as expected) shifted to higher potentials in comparison to that of 5 (see Table 2).</p><!><p>CV curves for 6–8 in CH2Cl2 (Ag/AgCl reference electrode, 0.1 m N(nBu)4(PF6) as supporting electrolyte, scan rate=100 mV s−1). All curves are measured in the direction of oxidation. Potentials are given versus the Fc+/Fc couple.</p><p>E 1/2 and E ox values [V vs. Fc+/Fc] for compounds 6–8 in CH2Cl2 solution.</p><p>Redox couple</p><p>6 E 1/2 (E ox)</p><p>7 [a] E 1/2 (E ox)</p><p>8 E 1/2 (E ox)</p><p>0/2+</p><p>−0.30 (−0.25)</p><p>(−0.18)</p><p>−0.25 (−0.18)</p><p>2+/.3+</p><p>0.67 (0.69)</p><p>(0.70)</p><p>0.69 (0.73)</p><p>.3+/4+</p><p>0.79 (0.85)</p><p>(0.84)</p><p>0.88 (0.93)</p><p>[a] For compound 7, the reduction wave splits into two components (see Figure 8); thus, no E 1/2 value is given.</p><!><p>The new GFAs 4 and 5 could easily be doubly protonated to give (4+2 H)(PF6)2 and (5+2 H)(PF6)2 by reaction with two equivalents of NH4PF6. With an excess of HCl⋅OEt2, they are tetra‐protonated to the salts (4+4 H)Cl4 and (5+4 H)Cl4. The bands in the electronic absorption spectra experience small hypsochromic shifts upon protonation. Hypsochromic shifts upon protonation have been reported previously for guanidines.27 Recently, such shifts and the influence of intermolecular hydrogen bonding to anions were analysed.28</p><p>In clear difference to all other GFA compounds, blue fluorescence is switched on by protonation. Although the neutral compounds are fluorescence silent, the protonated compounds show fluorescence signals, with maxima of emission at λ=471 nm for (5+2 H)(PF6)2 and λ=463 nm for (5+4 H)Cl4 (Figure 9). The quantum yields for protonated 5 were determined to be approximately 2 % for (5+2 H)(PF6)2 (λ ex=300 nm) and 15 % for (5+4 H)Cl4 (λ ex=280 nm). The drastic increase in the fluorescence intensity with the degree of protonation becomes evident from the photos in Figure 9 a. The increasing energetic separation between electronic ground and excited states might be responsible for a suppression/attenuation of thermal relaxation pathways, explaining the onset of fluorescence upon protonation. Changes in the fluorescence intensity with the degree of protonation (amplification as well as attenuation of fluorescence) have already been observed for guanidines that are fluorescent in their unprotonated form,29, 30 but such a drastic switching effect from fluorescent silent to fluorescent upon protonation has not been reported previously. In recent years, excited‐state intermolecular proton transfer (ESPT) has been evidenced for some protonated guanidines.31, 32 However, protonated 5 does not display ESPT, given that changes in the degree of protonation do not lead to shifts in the energy of the fluorescence signal.</p><!><p>a) UV/Vis and fluorescence spectra for compounds (5+2 H)(PF6)2 and (5+4 H)Cl4 in CH3CN solutions, measured with excitation wavelengths of λ ex=375 nm for (5+4 H)Cl4 and λ ex=385 nm for (5+2 H)(PF6)2. The inset shows photos of the fluorescence of solutions of (5+2 H)(PF6)2 (left) and (5+4 H)Cl4 (right) with excitation wavelength of 254 nm. b) Comparison between the Lewis structures of (3+2H)2+ and (5+2H)2+.</p><!><p>In the 1H NMR spectrum of (5+2 H)(PF6)2 in CD3CN, two of the guanidino methyl groups experience a strong shift relative to 5, indicating a localised protonation (terminal N−H protons) and the absence of N−H⋅⋅⋅N hydrogen bonds. By contrast, the 1H NMR spectrum of (2+2 H)(PF6) shows a singlet for all methyl protons of the four guanidino groups, indicating delocalised bonding of the protons in N−H⋅⋅⋅N hydrogen bonds, and (3+2 H)2+ certainly also exhibits this bonding (see Lewis structures in Figure 9 b).18 A picture in the Supporting Information (p. 20) illustrates a solid‐state structure that we measured for (2+2H)2+, also indicating protonation of two guanidino groups with terminal N−H bonds in the solid state. In contrast to the two‐fold protonated forms of 2 and 3, N−H⋅⋅⋅N hydrogen bonds are absent in the protonated compounds 4 and 5. Hence, they are not proton sponges.</p><!><p>Next, compound 5 was reacted with two equivalents of ZnCl2 or Pd(OAc)2, leading to the dinuclear complexes [5(ZnCl2)2] and [5{Pd(OAc)2}2] (see Scheme 4). Crystals of [5{Pd(OAc)2}2] were obtained through diffusion of diethyl ether in a saturated dichloromethane solution (see Figure 10). A characteristic feature of binuclear late‐transition‐metal complexes of ligands 2 and 3 is the massive displacement of the metal atom from the aromatic plane of the ligand.18, 19, 20 For example, in the complex [2(CoCl2)2], the cobalt atoms are 1.034 Å above and below this plane.18 Similar large displacements were also found for complexes of 1,8‐bis(tetramethylguanidino)naphthalene.33 By contrast, the two Pd atoms in [5{Pd(OAc)2}2] remain in the naphthalene ring plane. The differences in the coordination geometries can be rationalised by the different orientations of the lone‐pair imino N orbitals in 2/3 and 4/5.</p><!><p>Synthesis of dinuclear metal complexes of redox‐active 5.</p><p>Illustration of the solid‐state structure of [5{Pd(OAc)2}2]. Displacement ellipsoids are drawn at the 30 % probability level. Hydrogen atoms are omitted. Selected bond lengths [Å] and angles [°]: Pd−N1 1.998(5), Pd−N4 2.033(4), N1−C1 1.410(7), N1−C6 1.360(8), N4−C2 1.424(7), N4−C11 1.337(8), C1−C2 1.371(8), C1−C5 1.400(8), C2−C3 1.438(7), C3−C3′ 1.440(10), C3′−C4 1.407(7), C4−C5 1.367(7), N1‐Pd‐N4 82.14(18).</p><!><p>CV measurements show that oxidation of [5(ZnCl2)2] is an irreversible process (Figure 11). Complexation shifts the oxidation potential to higher values (E ox=0.09 V vs. Fc+/Fc for [5(ZnCl2)2] compared to E ox=−0.43 V for free 5). As in free 5, two electrons are removed at equal potential. Interestingly, the reduction seems to occur in two potentially separated waves. Overall, the redox processes are not reversible, indicating some kind of transformation of the oxidised complex (e.g., partial cleavage of the coordinative bond).</p><!><p>CV curves for [5(ZnCl2)2] (blue) and [5{Pd(OAc)2}2] (red) in CH2Cl2 solutions (Ag/AgCl reference electrode, 0.1 m N(nBu)4(PF6) as supporting electrolyte, scan rate 100 mV s−1). All curves are measured in the direction of oxidation. Potentials are given versus the Fc+/Fc redox couple.</p><!><p>Next, the redox properties of [5{Pd(OAc)2}2] were studied in CV measurements (Figure 11). Again, complexation shifts the redox potential to higher values, but now two potentially separated and reversible one‐electron redox events are obtained. These are located at E 1/2=−0.31 V (E ox=−0.26 V) for the redox couple [5{Pd(OAc)2}2].+/[5{Pd(OAc)2}2] and E 1/2=−0.01 V (E ox=0.04 V) for the redox couple [5{Pd(OAc)2}2]2+/[5{Pd(OAc)2}2].+. By using the formulas in Equations (1), (2):(1)ΔG0=F×ΔE1/2 (2)Kdisp=exp-FRTΔE1/2</p><p>(with E 0≈E 1/2), the Gibbs free energy change (ΔG 0) for disproportionation of the monocation [5{Pd(OAc)2}2].+ can be estimated to be 29 kJ mol−1 and the equilibrium constant K disp to be 8.49×10−6 (F=Faraday constant). Hence, by complexation three stable redox states of 5 (neutral, radical monocationic and dicationic) are assessible. We showed previously that complexation of 1 with metal acetate leads in some cases to two one‐electron waves, allowing the isolation of dinuclear complexes bearing a radical monocationic or dicationic GFA ligand.34 Radical monocationic dinuclear copper(II) acetate, nickel(II) acetate or palladium(II) acetate complexes with bridging redox‐active guanidine ligands were shown to exhibit strong ligand–ligand or metal–ligand charge‐transfer bands, and the special role of the acetate co‐ligands in the stabilisation of the radical monocation was evaluated.34a, 34b</p><p>Preliminary spectro‐electrochemical measurements for [5{Pd(OAc)2}2] show a clear UV/Vis band around 780 nm upon removal of two electrons, confirming the presence of a dicationic ligand unit. When the oxidation–reduction cycles were repeated several times, the absorption maintained its intensity showing that the ligand‐centred redox process is fully reversible.</p><!><p>In this work, two new redox‐active organic molecules, 1,2,5,6‐tetrakis(tetramethylguanidino)‐naphthalene (4) and 1,2,5,6‐tetrakis(N,N′‐dimethylethyleneguanidino)‐naphthalene (5), were synthesised and their chemistry studied. Two‐electron oxidation of the colourless compounds in CH2Cl2 solutions at redox potentials E 1/2 of −0.46 V versus ferrocenium/ferrocene (Fc+/Fc) for 4/4 2+ and −0.47 V for 5/5 2+ leads to persistent dark‐green dications, which display a broad electronic transition around 800 nm. Salts of the dication (5(BF4)2 and 5(SbF6)2) were synthesised and shown to be stable, storable compounds. Cyclovoltammetric measurements show that further reversible oxidation in two one‐electron steps to the tetracation is possible. The salt 5(SbF6)4 appears to be a stable species but an extremely reactive oxidant, which cannot be dissolved in standard organic solvents without being reduced.</p><p>The neutral compounds are fluorescence silent. On the other hand, strong blue fluorescence is switched on upon protonation. The quantum yield increases with the degree of protonation, reaching 15 % upon tetraprotonation. This behaviour could be used for fluorescence sensing. We are currently using the protonation‐induced fluorescence to probe the proton‐coupled electron transfer (PCET) reactivity, for example, 5 2++2 H++2 e−→(5+2 H)2+ (see work on PCET reactivity of other GFA compounds16, 28).</p><p>Aromatic substitution reactions with N‐halogenosuccinimides or N‐bromophthalimide led in chemo‐selective reactions to the introduction of two halogeno and two succinimido/phthalimido groups, providing convenient access to new fully substituted derivatives with varying redox potentials. A stepwise oxidation–substitution pathway for these highly selective reactions is proposed on the basis of experiments with varying equivalents of the N‐iodosuccinimide.</p><p>Finally, compound 5 was applied as a redox‐active ligand in binuclear late‐transition‐metal complexes. Palladium acetate coordination does not only shift the redox potentials to higher values, but it also causes the two‐electron redox process for the uncoordinated compound to split into two potentially well‐separated one‐electron steps. Spectro‐electrochemical studies show that the redox processes are ligand‐centred. Hence, for the compound embedded in a metal complex, three stable redox states (neutral, radical monocationic and dicationic) are assessible. The results of this work build the basis for applications of compounds 4 and 5 in coordination chemistry and organic synthesis.</p><!><p>All reactions were carried out under a dry argon atmosphere using standard Schlenk techniques or in a dinitrogen‐filled glove box (Mbraun LABmaster dp, MB‐20.G). The applied solvents were dried with an MBraun Solvent Purification System and degassed prior to use. 2,6‐Dibromonaphthalene, sodium‐tert‐butyloxide, tris(dibenzylideneacetone)palladium(0) [Pd2(dba)3], tin(II) chloride, 2,2′‐bis(diphenylphosphino)‐1,1′‐binaphthyl (BINAP) and benzophenone imine were purchased from abcr, Sigma–Aldrich or Alfa Aesar and used without further purification. The reagents 2,6‐dibromo‐1,5‐dinitronaphthalene,35 2‐chloro‐1,1′,3,3′‐tetramethylformamidinium chloride19 and 2‐chloro‐1,3‐dimethylimidazolinium chloride19 were synthesised according to the literature. Infrared spectra were recorded as KBr discs with a BIORAD Excalibur FTS 3000 spectrometer or as solids with an AGILENT Cary 630 FTIR spectrometer. NMR spectra were measured with BRUKER DPX 200, BRUKER Avance II 400 or BRUKER Avance III 600 instruments at a temperature of 298 K if not stated otherwise. Owing to the high sensitivity and low solubility of some of the compounds synthesised, a few 1H NMR spectra and most of the 13C NMR spectra measured are very dilute. As a result, some of the aromatic 13C signals could not be detected and therefore could not be assigned. Elemental analyses were performed at the Microanalytical Laboratory of the University of Heidelberg by using the vario EL and vario MICRO cube devices from Elementar Analysensysteme GmbH. Owing to the high sensitivity of the synthesised compounds to oxygen and water, some elemental analyses deviate by more than 0.4 %. ESI mass spectrometry was performed with a JEOL JMS‐700 magnetic sector and BRUKER ApexQe hybrid 9.4 T FT‐ICR apparatus at the MS Laboratory of the University of Heidelberg. UV/Vis spectra were measured with a VARIAN Cary 5000 UV‐VIS‐NIR spectrometer. CV measurements were carried out with an EG&G Princeton 273 apparatus with an Ag/AgCl reference electrode. All voltammograms were recorded at room temperature. CH2Cl2 was used as solvent for the individual compounds (c=10−3  m), and N(nBu)4(PF6) (electrochemical grade (≥99.0 %), Fluka) was employed as supporting electrolyte (c=0.1 m). Spectro‐electrochemical studies were recorded with an AvaSpec 2048×14 CCD spectrometer and a Metrohm Autolab PGSTAT 204 potentiostat/galvanostat in an RHD Instruments TSC spectro sealable cell (working electrode: Pt gauze, counter electrode: Pt disc, Ag as pseudo reference electrode). Fluorescence measurements were recorded with a Varian Cary Eclipse instrument. Quantum yields were estimated with a PTI Quantum Master 40 equipped with an Ulbricht sphere.</p><!><p>N,N′ ‐(1,5‐Dinitronaphthalene‐2,6‐diyl)bis(1,1‐diphenylmethanimine): Sodium‐tert‐butyloxide (961 mg, 10.00 mmol), BINAP (150 mg, 0.24 mmol) and Pd2(dba)3 (109 mg, 0.12 mmol) were dissolved in degassed toluene (40 mL). The deep‐red solution was stirred at room temperature for 10 min before 2,6‐dibromo‐1,5‐dinitronaphthalene (1.50 g, 4.01 mmol) was added in portions. After stirring for another 5 min, benzophenone imine (1.48 mL, 8.82 mmol) was added dropwise, leading to precipitation of a brown solid. The reaction mixture was stirred for 3 h at 80 °C. Then the solution was allowed to cool down to room temperature, and the solvent was removed under vacuum. The obtained dark‐brown solid was washed with absolute methanol (3×15 mL) and dried under vacuum to yield 1.52 g (2.64 mmol, 82 %) product as a brown powder. C, H, N analysis (%) for C36H24N4O4: calcd C 74.99, H 4.20, N 9.72; found C 74.24, H 4.52, N 9.23. 1H NMR (600 MHz, CD2Cl2): δ=7.61 (d, 2 J=8 Hz, 2 H, Hnaph), 7.33–7.56 (m, 20 H, Hbenz), 6.74 ppm (d, 2 J=12 Hz, 2 H, Hnaph).</p><p>1,2,5,6‐Tetraamino‐naphthalene tetrahydrochloride: N,N′‐(1,5‐Dinitronaphthalene‐2,6‐diyl)bis(1,1‐diphenylmethanimine) (1.00 g, 1.74 mmol) and SnCl2 (6.60 g, 34.81 mmol) were dissolved in concentrated HCl (30 mL). The red solution was stirred at 70 °C for 3 h. After cooling down the solution to 5 °C, the precipitated red‐brown product was filtered off and washed with degassed EtOH until the washing solution turned colourless. The product was obtained as a white‐grey solid in 95 % yield (530 mg, 1.60 mmol). 1H NMR (400 MHz, [D6]DMSO): δ=7.51 (d, 2 J=8 Hz, 2 H, Hnaph), 7.24 ppm (d, 2 J=8 Hz, 2 H, Hnaph).</p><p>1,2,5,6‐Tetrakis(tetramethylguanidino)‐naphthalene (4): 1,2,5,6‐Tetraamino‐naphthalene tetrahydrochloride (150 mg, 0.45 mmol) was suspended in degassed acetonitrile (15 mL) at 0 °C. A solution of 2‐chloro‐1,1′,3,3′‐tetramethylformamidinium chloride (338 mg, 1.99 mmol) in acetonitrile (5 mL) was added dropwise to the resulting grey suspension. The reaction mixture was stirred for 10 min at room temperature. Upon addition of triethylamine (1.00 mL, 7.22 mmol), the reaction mixture changed colour to yellow. The solution was stirred for 1 h at 0 °C and an additional 2 h at room temperature. After removing the solvent from the brown solution under vacuum, the residue was redissolved in 25 % NaOH solution and extracted with dichloromethane (3×20 mL). The combined organic phases were dried over potassium carbonate, and the solvent was removed under vacuum. The crude product was recrystallised from acetonitrile to yield 80 mg (31 %, 0.14 mmol) of a colourless solid. Crystals suitable for structural characterisation by single‐crystal X‐ray diffraction were grown by diffusion of diethyl ether into a saturated acetonitrile solution. C, H, N analysis (%) for C30H52N12: calcd C 62.04, H 9.02, N 28.94; found C 61.60, H 8.90, N 28.53. 1H NMR (200 MHz, CDCl3): δ=7.21 (d, 2 J=5 Hz, 2 H, Hnaph), 6.70 (d, 2 J=8 Hz, 2 H, Hnaph), 2.64 ppm (d, 2 J=8 Hz, 48 H, CH3). 1H NMR (200 MHz, CD2Cl2): δ=7.12 (d, 2 J=10 Hz, 2 H, Hnaph), 6.54 (d, 2 J=8 Hz, 2 H, Hnaph), 2.64 ppm (s, 48 H, CH3). 13C NMR (600 MHz, CD2Cl2): δ=122.60 (s, CH), 116.32 (s, CH), 39.90 (s, CH3), 39.49 ppm (s, CH3). MS (ESI+): m/z (%)=581.4 (100) [M+H]+, 536.4 (18) [M−NMe2]+, 491.3 (3) [M−(NMe2)2]+. UV/Vis (CH3CN): λ max (ϵ in L mol−1 cm−1)=257 (2.49×104), 348 nm (0.62×104). IR (KBr): ṽ=2924w, 2804w, 1617vs, 1600vs, 1560s, 1371m, 1307w, 1325w, 1136m, 990w, 898w, 613m, 480w cm−1.</p><p>1,2,5,6‐Tetrakis(dimethylethyleneguanidino)‐naphthalene (5): 1,2,5,6‐Tetraamino‐naphthalene tetrahydrochloride (150 mg, 0.45 mmol) was suspended in degassed acetonitrile (15 mL) at 0 °C. Then 2‐chloro‐1,3‐dimethylimidazolinium chloride (317.5 mg, 1.89 mmol) in acetonitrile (5 mL) was added dropwise to the resulting grey suspension, and the reaction mixture was stirred for 15 min at room temperature. Triethylamine (1.00 mL, 7.22 mmol) was added, resulting in a colour change to red. Subsequently, the solution was stirred for 1 h at 0 °C and for an additional 2 h at room temperature. After removing the solvent from the red solution, the residue was redissolved in 25 % NaOH solution and extracted with dichloromethane (3×20 mL). The combined organic phases were dried over potassium carbonate, and the solvent was removed under vacuum. The crude product was recrystallised from acetonitrile to yield 135 mg (52 %, 0.24 mmol) of a colourless solid. Crystals suitable for an X‐ray analysis were grown from a saturated acetonitrile solution. C, H, N analysis (%) for C30H44N12: calcd C 62.91, H 7.74, N 29.35; found C 62.18, H 7.12, N 29.89. 1H NMR (400 MHz, CD2Cl2): δ=7.30 (d, 2 J=8 Hz, 2 H, Hnaph), 6.81 (d, 2 J=12 Hz, 2 H, Hnaph), 3.20 (s, 16 H, CH2), 2.58 ppm (d, 24 H, CH3). 13C NMR (600 MHz, CD2Cl2): δ=122.60 (s, CH), 116.56 (s, CH), 49.18 (s, CH2), 49.07 (s, CH2), 35.23 (s, CH3), 34.59 ppm (s, CH3). MS (ESI+): m/z (%)=573.4 (100) [M+H]+. UV/Vis (CH2Cl2): λ max (ϵ in L mol−1 cm−1)=276 (2.19×104), 340 nm (0.56×104). IR (KBr): ṽ=2932w, 2842w, 1660vs, 1628vs, 1561s, 1485m, 1394m, 1275m, 1037m, 965w, 910s, 827m, 763w, 712w cm−1.</p><p>Compound 5(BF4)2: Compound 5 (20 mg, 0.03 mmol) was dissolved in THF (5 mL). Then NO(BF4) (7 mg, 0.06 mmol) was added in one portion. The solution was stirred for 2 h at room temperature. During this time, a green solid precipitated. The crude product was filtered off and washed with THF (3×3 mL) and Et2O (3×3 mL). The obtained solid was dried under vacuum to yield 22 mg (84 %, 0.03 mmol) of the product as a green solid. C, H, N analysis (%) for C30H44N12B2F8: calcd C 48.28, H 5.94, N 22.25; found C 48.30, H 6.27, N 22.10. 1H NMR (200 MHz, CD2Cl2): δ=7.77 (d, 2 J=10 Hz, 2 H, Hnaph), 6.00 (d, 2 J=10 Hz, 2 H, Hnaph), 3.79 (s, 8 H, CH2), 3.59 (s, 8 H, CH2), 2.80 (s, 12 H, CH3), 2.76 ppm (s, 12 H, CH3). 13C NMR (600 MHz, CD2Cl2): δ=48.73 (s, CH2), 48.24 (s, CH2), 33.45 (s, CH2), 33.30 ppm (s, CH2). UV/Vis (CH2Cl2): λ max (ϵ in L mol−1 cm−1)=230 (3.29×104), 277 (3.22×104), 778 nm (0.23×104).</p><p>Compound 5(SbF6)2: Compound 5 (20 mg, 0.03 mmol) was dissolved in dichloromethane (5 mL). Then Ag(SbF6) (17 mg, 0.06 mmol) was added in one portion. The solution was stirred for 2 h at room temperature. The dark‐green solution was filtered from the precipitated silver, and the solvent was removed under vacuum. The crude product was washed with THF (3×3 mL) and Et2O (3×3 mL). The obtained solid was dried under vacuum to yield 32 mg (88 %, 0.03 mmol) of the product as a green solid. Crystals suitable for structural characterisation by single‐crystal X‐ray diffraction were grown by diffusion of pentane into a saturated dichloromethane solution. C, H, N analysis (%) for C30H44N12Sb2F12: calcd C 34.51, H 4.25, N 16.10; found C 34.89, H 4.52, N 16.13. 1H NMR (200 MHz, CD2Cl2): δ=7.77 (d, 2 J=10 Hz, 2 H, Hnaph), 6.00 (d, 2 J=10 Hz, 2 H, Hnaph), 3.79 (s, 8 H, CH2), 3.59 (s, 8 H, CH2), 2.80 (s, 12 H, CH3), 2.76 ppm (s, 12 H, CH3). 13C NMR (600 MHz, CD2Cl2): δ=48.73 (s, CH2), 48.24 (s, CH2), 33.45 (s, CH2), 33.30 ppm (s, CH2).</p><p>Compound (5+2 H)(PF6)2: Compound 5 (20 mg, 0.03 mmol) was dissolved in tetrahydrofuran (5 mL). Ammonium hexafluorophosphate (13 mg, 0.06 mmol) was added in one portion to the light‐yellow solution. Then the reaction mixture was stirred at room temperature for 1 h. During this time, a white solid precipitated. The crude product was filtered and washed with tetrahydrofuran until the solution turned colourless. Remaining tetrahydrofuran was removed under vacuum to give the product as a white solid in 87 % yield (26 mg, 0.03 mmol). Crystals of (5+2 H)(BF4)2 suitable for structural characterisation by single‐crystal X‐ray diffraction were grown by diffusion of diethyl ether into a saturated solution of 5(BF4)2 in dichloromethane. C, H, N analysis (%) for C30H46N12F12P2 ⋅THF: calcd C 43.59, H 5.81, N 17.94; found C 43.22, H 6.31, N 17.25. 1H NMR (600 MHz, CD3CN): δ=7.65 (d, 2 J=12 Hz, 2 H, Hnaph), 7.18 (d, 2 J=6 Hz, 2 H, Hnaph), 3.54 (s, 8 H, CH2), 3.45 (s, 8 H, CH2), 2.69 (s, 12 H, CH3), 2.58 ppm (s, 12 H, CH3). 13C NMR (600 MHz, CD3CN): 49.44 (s, CH2), 49.34 (s, CH2), 34.63 (s, CH3), 34.32 ppm (s, CH3). UV/Vis (CH3CN): λ max (ϵ in L mol−1 cm−1)=211 (3.05×104), 257 (2.51×104), 301 nm (2.10×104). IR (KBr): ṽ=3357m, 2947m, 2882m, 2848m, 1629s, 1604s, 1574m, 1489w, 1443w, 1423w, 1405m, 1390m, 1323w, 1299m, 1287s, 1242m, 1202w, 1142w, 1123w, 1040m, 1021m, 965w, 914m, 879w, 833vs, 787m, 766m, 742w, 712w, 677w cm−1.</p><p>Compound (5+4 H)Cl4: Compound 5 (20 mg, 0.03 mmol) was dissolved in tetrahydrofuran (5 mL). HCl (2 m in diethyl ether, 0.1 mL) was added to the light‐yellow solution. After 1 h of stirring at room temperature, a white solid precipitated. The crude product was filtered and washed with tetrahydrofuran (3×5 mL) and diethyl ether (3×5 mL). Then the solid was dried under vacuum to give the product as a colourless solid in 95 % yield (24 mg, 0.03 mmol). The compound is extremely hygroscopic and quickly picks up H2O. C, H, N analysis (%) for C30H48N12Cl4 ⋅H2O: calcd C 48.92, H 6.84, N 22.28; found C 48.80, H 7.18, N 22.26. 1H NMR (200 MHz, CD3CN): δ=11.95 (s, 2 H, NH), 11.80 (s, 2 H, NH), 8.05 (d, 2 J=10 Hz, 2 H, Hnaph), 7.65 (d, 2 J=10 Hz, 2 H, Hnaph), 3.75 (d, 2 J=12 Hz, 16 H, CH2), 2.96 (s, 12 H, CH3), 2.78 ppm (s, 12 H, CH3). 13C NMR spectroscopic shifts are not given because of low solubility. UV/Vis (CH3CN): λ max (ϵ in L mol−1 cm−1)=268 (2.25×104), 304 nm (0.47×104). IR (KBr): ṽ=2930m, 2877m, 2801m, 2660m, 1591vs, 1479m, 1401m, 1373s, 1336w, 1296s, 1127w, 1080w, 1037m, 970m, 913m, 834w, 785w, 686m cm−1.</p><p>1,2,5,6‐Tetrakis(N,N′ ‐dimethylethyleneguanidino)‐3,7‐dibromo‐4,8‐disuccinimido‐naphthalene (6): Compound 5 (20 mg, 0.03 mmol) was dissolved in dichloromethane (5 mL). Then N‐bromosuccinimide (37 mg, 0.21 mmol) was added in one portion. The colour of the solution immediately changed to deep yellow. After 1 h of stirring at room temperature, the colour changed to green. The solvent was removed under vacuum, and the remaining solid was washed with tetrahydrofuran (2×5 mL). The oxidised product was dissolved in dichloromethane (8 mL), and hydrazine hydrate (80 % in H2O, 0.1 mL) was added. Gas development and a colour change to light yellow were observed. Subsequently aqueous NaOH solution (15 %, 4 mL) was added, and the mixture was stirred for an additional 20 min. The organic phase was separated and dried over MgSO4. The product was obtained, after solvent removal, as a yellow solid in 69 % yield (22 mg, 0.02 mmol). Crystals suitable for an X‐ray analysis were grown by diffusion of diethyl ether in a saturated dichloromethane solution. C, H, N analysis (%) for C38H48N14Br2O4: calcd C 49.36, H 5.23, N 21.21; found C 50.01, H 5.72, N 20.81. 1H NMR (200 MHz, CD2Cl2): δ=3.15 (m, 16 H, CH2gua), 2.76 (s, 8 H, CH2succ), 2.59 (s, 12 H, CH3), 2.45 ppm (s, 12 H, CH3). 13C NMR (600 MHz, CD2Cl2): δ=176.73 (s, CO), 49.02 (s, CH2), 48.69 (s, CH2), 34.75 (s, CH3), 33.98 (s, CH3), 29.35 ppm (s, CH2succ). MS (ESI+): m/z (%)=925 (100) [M+H]+, 846 (20) [M−Br]+. UV/Vis (CH2Cl2): λ max (ϵ in L mol−1 cm−1)=230 (3.69×104), 293 (4.81×104), 355 (0.88×104), 420 nm (1.07×104). IR (KBr): ṽ=2935m, 2846m, 1715vs, 1648vs, 1633vs, 1536w, 1489m, 1441w, 1420s, 1387s, 1338m, 1280s, 1243s, 1190s, 1114w, 1071w, 1045m, 1028m, 998w, 965w, 933s, 828m, 797m, 765m, 721m, 710m, 705m, 697m cm−1.</p><p>1,2,5,6‐Tetrakis(N,N′ ‐dimethylethyleneguanidino)‐3,7‐diiodo‐4,8‐disuccinimido‐naphthalene (7): Compound 5 (20 mg, 0.03 mmol) was dissolved in dichloromethane (5 mL). Then N‐iodosuccinimide (47 mg, 0.21 mmol) was added in one portion. The colour of the solution immediately changed to deep yellow. After 2 h of stirring at room temperature, the colour turned green. The solvent was removed under vacuum, and the remaining solid was washed with tetrahydrofuran (2×5 mL). The oxidised product was dissolved in dichloromethane (8 mL), and hydrazine hydrate (80 % in H2O, 0.1 mL) was added. Gas development and a colour change to light yellow were detected. Subsequently aqueous NaOH solution (15 %, 4 mL) was added, and the mixture was stirred for an additional 20 min. The organic phase was separated and dried over MgSO4. The product was obtained, after solvent removal, as a yellow solid in 65 % yield (24 mg, 0.02 mmol). Crystals suitable for an X‐ray analysis were grown by diffusion of diethyl ether in a saturated dichloromethane solution. C, H, N analysis (%) for C38H48N14I2O4: calcd C 44.80, H 4.75, N 19.25; found C 44.94, H 5.66, N 19.03. 1H NMR (200 MHz, CD2Cl2) δ=3.15 (m, 16 H, CH2gua), 2.76 (s, 8 H, CH2succ), 2.60 (s, 12 H, CH3), 2.44 ppm (s, 12 H, CH3). 13C NMR (600 MHz, CD2Cl2): δ=176.67 (s, CO), 49.01 (s, CH2), 48.68 (s, CH2), 34.76 (s, CH3), 33.99 (s, CH3), 29.53 ppm (s, CH2succ). MS (ESI+): m/z (%)=1019 (45) [M+H]+, 893 (100) [M−I]+, 767 (25) [M−2I]+. UV/Vis (CH2Cl2): λ max (ϵ in L mol−1 cm−1)=231 (2.98×104), 295 (3.47×104), 359 (0.62×104), 416 nm (0.69×104). IR (KBr): ṽ=3002w, 2927m, 2873m, 2798w, 1721m, 1570s, 1497m, 1495s, 1424m, 1370vs, 1299s, 1234m, 1179s, 1132s, 1064m, 1026m, 1004m, 946w, 918w, 906w, 895w, 850w cm−1.</p><p>1,2,5,6‐Tetrakis(N,N′ ‐dimethylethyleneguanidino)‐3,7‐dibromo‐4,8‐diphthalimide‐naphthalene (8): Compound 5 (20 mg, 0.03 mmol) was dissolved in dichloromethane (5 mL). Then N‐bromophthalimide (47 mg, 0.21 mmol) was added in one portion. The colour of the solution immediately changed to green. After 2 h of stirring at room temperature, the solvent was removed under vacuum, and the resulting solid was washed with tetrahydrofuran (2×5 mL). The oxidised product was dissolved in dichloromethane (8 mL), and hydrazine hydrate (80 % in H2O, 0.1 mL) was added. Gas development and a colour change to red were detected. Then aqueous NaOH solution (15 %, 4 mL) was added to the mixture and stirring continued for 20 min. The organic phase was separated and dried over MgSO4. The solvent was removed, giving the product as an orange‐red solid in 82 % yield (29 mg, 0.03 mmol). Crystals suitable for an X‐ray analysis were grown by diffusion of diethyl ether in a saturated dichloromethane solution. C, H, N analysis (%) for C46H48N14Br2O4: calcd C 54.13, H 4.75, N 19.21; found C 53.52, H 5.03, N 19.28. 1H NMR (200 MHz, CD2Cl2): δ=7.89 (m, 4 H, CHphthalimide), 7.74 (m, 4 H, CHphthalimide), 3.07 (m, 16 H, CH2), 2.57 (s, 12 H, CH3), 2.16 ppm (s, 12 H, CH3). 13C NMR (600 MHz, CD2Cl2): δ=167.60 (s, CO), 134.07 (s, Cphthalimide), 133.87 (s, Cphthalimide), 123.38 (s, Cphthalimide), 48.70 (s, CH2), 48.11 (s, CH2), 34.09 (s, CH3), 33.46 ppm (s, CH3). UV/Vis (CH2Cl2): λ max (ϵ in L mol−1 cm−1)=234 (4.56×104), 293 (1.77×104), 363 (0.78×104), 410 nm (0.93×104). IR (KBr): ṽ=2961w, 2921m, 2842m, 1713s, 1617vs, 1488m, 1440w, 1415m, 1388m, 1347w, 1317w, 1281m, 1253s, 1087m, 1039m, 1021s, 948m, 921w, 883m, 794s, 712vs cm−1.</p><p>Compound [5(ZnCl2)2]: Compound 5 (20 mg, 0.03 mmol) was dissolved in acetonitrile (5 mL). Then ZnCl2 (11 mg, 0.08 mmol) was added. The solution was stirred for 2 h. The precipitated white solid was filtered and washed with Et2O (3×6 mL) to remove any remaining ZnCl2. The product was dried under vacuum and was obtained as a white solid in 87 % yield (26 mg, 0.03 mmol). C, H, N analysis (%) for C30H44N12Cl4Zn2: calcd C 42.36, H 5.25, N 19.88; found C 42.40, H 5.57, N 19.88. MS (ESI+): m/z (%)=845 (7) [M+H]+, 808 (7) [M−Cl]+, 709 (20) [M−ZnCl2]+, 573 (100) [M−(ZnCl2)2]+. UV/Vis (CH3CN): λ max (ϵ in L mol−1 cm−1)=234 (0.59×104), 288 (0.95×104), 346 (0.39×104), 398 nm (0.09×104). NMR spectra were not measured because of low solubility.</p><p>Compound [5{Pd(OAc)2}2]: Compound 5 (20 mg, 0.03 mmol) was dissolved in acetonitrile (5 mL). Then Pd(OAc)2 (17 mg, 0.08 mmol) was added. The solution was stirred, and the colour of the reaction mixture changed to deep red. After 2 h of continuous stirring, the solvent was removed under vacuum, and the black residue was washed with tetrahydrofuran (3×5 mL). The product was dried under vacuum and obtained as a red‐black solid in 77 % yield (27 mg, 0.03 mmol). Crystals suitable for an X‐ray analysis were obtained through diffusion of diethyl ether in a saturated dichloromethane solution. C, H, N analysis (%) for C38H56N12O8Pd2: calcd C 44.67, H 5.52, N 16.45; found C 43.85, H 5.77, N 17.24. 1H NMR (200 MHz, CD2Cl2): δ=6.62 (d, 2 H, CHnaph), 6.44 (d, 2 H, CHnaph), 3.50 (s, 16 H, CH2), 2.95 (s, 12 H, CH3), 2.76 ppm (s, 12 H, CH3). 13C NMR (600 MHz, CD2Cl2): δ=177.95 (s, CO), 48.43 (s, CH2), 47.94 (s, CH2), 35.56 (s, CH3), 34.46 (s, CH3), 23.25 (s, CH3OAc), 23.04 ppm (s, CH3OAc). UV/Vis (CH2Cl2): λ max (ϵ in L mol−1 cm−1)=231 (3.82×104), 290 (2.44×104), 351 (0.88×104), 404 nm (0.43×104). IR (KBr): ṽ=3383m, 2932m, 1660vs, 1536vs, 1467m, 1371vs, 1284vs, 1047w, 1008m, 967m, 935m, 853w, 821m, 792m, 679m cm−1.</p><!><p>DFT calculations were carried out with the TURBOMOLE program package.36 The B3LYP functional37 was used in combination with the def2‐TZVP basis set.38 All structures are stationary points on the energy potential surface as confirmed by frequency computations.39</p><!><p>Suitable crystals for single‐crystal structure determination were taken directly from the mother liquor, immersed in perfluorinated polyether oil and fixed on a cryo loop. For compound 4, a full shell of intensity data was collected at low temperature with an Agilent Technologies Supernova‐E CCD diffractometer (MoKα radiation, microfocus X‐ray tube, multilayer mirror optics). Detector frames (typically w‐scans, occasionally j‐scans, scan width 0.4°) were integrated by profile fitting.40, 41 Data were corrected for air and detector absorption and for Lorentz and polarisation effects41, 42 and scaled essentially by application of appropriate spherical harmonic functions.41, 43 Absorption by the crystal was treated with a semiempirical multiscan method (as part of the scaling process) and augmented by a spherical correction.41, 44 An illumination correction was performed as part of the numerical absorption correction.42 The structure was solved by ab initio dual space methods involving difference Fourier syntheses (VLD procedure)44 and refined by full‐matrix least‐squares methods based on F 2 against all unique reflections.45 All non‐hydrogen atoms were given anisotropic displacement parameters. Hydrogen atoms were input at calculated positions and refined with a riding model.46</p><p>Full shells of intensity data were collected at low temperature with a Nonius Kappa CCD diffractometer (MoKα radiation, sealed X‐ray tube, graphite monochromator, compounds: N,N′‐(1,5‐dinitronaphthalene‐2,6‐diyl)bis(1,1‐diphenylmethanimine), 5, 6, 8 and 9) and a Bruker D8 Venture, dual source instrument (MoKα or CuKα radiation, microfocus X‐ray tube, Photon III detector, compounds: 5(SbF6)2, (5+2 H)(BF4), [5{Pd(OAc)2}2] and the intermediate of 7). Data were processed with the standard Nonius and Bruker (SAINT, APEX3) software packages.47 Multiscan absorption correction was applied using the SADABS program.48 The structures were solved by intrinsic phasing and refined using the SHELX software package.45, 49 Graphical handling of the structural data during solution and refinement were performed with OLEX2.50 All non‐hydrogen atoms were given anisotropic displacement parameters. Hydrogen atoms bound to carbon were input at calculated positions and refined with a riding model. Hydrogen atoms bound to nitrogen were located in difference Fourier syntheses and refined, either fully or with appropriate distance and/or symmetry. CCDC 1968384, 1968385, 1968386, 1968387, 1968388, 1968389, 1968390, 1968391, 1968392, and 1968393 contain the supplementary crystallographic data for this paper. These data are provided free of charge by The Cambridge Crystallographic Data Centre.</p><!><p>The authors declare no conflict of interest.</p><!><p>As a service to our authors and readers, this journal provides supporting information supplied by the authors. Such materials are peer reviewed and may be re‐organized for online delivery, but are not copy‐edited or typeset. Technical support issues arising from supporting information (other than missing files) should be addressed to the authors.</p><p>Supplementary</p><p>Click here for additional data file.</p>

PubMed Open Access

Interaction Energy Analysis of Monovalent Inorganic Anions in Bulk Water Versus Air/Water Interface

Soft anions exhibit surface activity at the air/water interface that can be probed using surface-sensitive vibrational spectroscopy, yet the statistical mechanics behind this surface activity remains a matter of debate. Here, we examine the nature of anion-water interactions at the air/water interface using a combination of molecular dynamics simulations and quantum-mechanical energy decomposition analysis based on symmetry-adapted perturbation theory. Results are presented for a set of monovalent anions including Cl − , Br − , I − , CN − , OCN − , SCN − , NO − 2 , NO − 3 , and ClO − n (n = 1, 2, 3, 4), several of which are archetypal examples of surface-active species. In all cases, we find that average anion-water interaction energies are systematically larger in bulk water although the difference (with respect to the interaction energy in the interfacial environment) is well within the magnitude of the instantaneous fluctuations. Specifically for the surface-active species Br − (aq), I − (aq), ClO − 4 (aq), and SCN − (aq), and also for ClO − (aq), the charge-transfer (CT) energy is found to be larger at the interface than it is in bulk water, by an amount that is greater than the standard deviation of the fluctuations. The Cl − ion also has a slightly larger CT energy at the interface but NO − 3 (aq) does not; these two species are borderline cases where consensus is lacking regarding their surface activity. However, CT stabilization amounts to < 20% of the total induction energy, for all of the ions considered here, and CT-free polarization energies are systematically larger in bulk water, again in all cases, so the role of these effects in soft anion surface activity remains unclear. This analysis complements our recent work suggesting that the short-range solvation structure around these ions is scarcely different at the air/water interface from what it is in bulk water. Together, these observations suggest that changes in first-shell hydration structure around soft anions cannot explain observed surface activities.

interaction_energy_analysis_of_monovalent_inorganic_anions_in_bulk_water_versus_air/water_interface

Introduction<!>Classical MD Simulations<!>Symmetry-Adapted Perturbation Theory<!>Polarization and Charge Transfer<!>Results and Discussion<!>5(b)]<!>Conclusions

<p>One of the earliest results of surface-sensitive vibrational sum-frequency generation (VSFG) experiments [1,2] was the observation that soft anions impact the vibrational lineshape in the O-H stretching region, but that hard anions do not [3][4][5][6]. The term "soft" is chosen carefully here, as an alternative to "polarizable", and can be roughly interpreted as "monovalent and polarizable", equivalent to having a low surface charge density [7]. (Such ions are sometimes called "chaotropic" [8].) Although the surface activity of certain anions is often discussed in terms of polarizability [9][10][11][12][13][14][15][16][17], it should be noted that polyvalent anions such as SO 2− 4 (aq) are quite polarizable [18], however the presence of such ions in solution does not affect the O-H lineshape measure in VSFG experiments [19]. Molecular dynamics (MD) simulations suggest that hard anions, including polyvalent species but also fluoride, are repelled from the air/water interface [20,21].</p><p>The nature of the surface activity in the soft anions remains a matter of debate. Whereas continuum electrostatics predicts that all ions are repelled from the air/water interface [13], early MD simulations using polarizable force fields suggested that soft anions are not only present at the interface but in fact partition preferentially there [9,13,20,22]. More recent work, however, has suggested that these concentration enhancements were exaggerated by the force fields in use at the time [23][24][25][26][27][28][29], which aligns with the interpretation of some of the early experiments [3]. According to this point of view, surface activity may simply reflect the absence of depletion of soft anions at the interface [30,31], rather than a concentration enhancement.</p><p>To this debate, the present authors have recently added the observation (based upon MD simulations using polarizable force fields) that the first-shell hydration structure around soft anions is hardly different at the air/water interface as compared to values computed for the same anions in bulk water [7]. Such similarities had been noted previously for I − (aq) [32] and for SCN − (aq) [33], and this is notable since iodide and thiocyanate are archetypal examples of ions that perturb the O-H lineshape in VSFG experiments [3,4,19,34,35]. Our work considered a much larger set of anions [7], and the close structural similarities that we observe (including both the average number as well as the orientation of the ion-water hydrogen bonds) suggest that the origins of anion-induced changes in the O-H vibrational lineshape must be rather subtle effects on water-water hydrogen bonds, perhaps due to ion-induced changes in local electric fields [36].</p><p>Our previous work [7] was limited to structural characterization of the ions in question, along with a detailed examination of their ionization energies in order to make contact with liquid microjet photoelectron spectroscopy [37]. The present work adds an energetic component to this analysis. Specifically, we compute anion-water interaction energies for the same set of monovalent anions that we considered previously: Cl − , Br − , I − , CN − , OCN − , SCN − , NO − 2 , NO − 3 , and ClO − n (n = 1, 2, 3, 4). Some of these are typical surfaceactive ions (e.g., Br − , I − , SCN − , and ClO − 4 ), whereas others such as CN − , OCN − , and NO − 2 visit the interface much less frequently, according to the MD simulations [7], and are not classified as surface-active. Intermediate cases where the surface activity is weak, or where experimental consensus is lacking, include Cl − and NO − 3 [19]. Amongst these ions, our simulations indicate that even the ones that are not considered surface active nevertheless spend enough time near the air/water interface that it is possible to assemble an interfacial data set for them, and these cases offer a useful comparison to the canonical surface-active anions.</p><p>We present a detailed analysis of the (ensemble-averaged) interaction between each of these ions and its short-range hydration sphere, in both bulk water and at the air/ water interface, using the quantum-chemical methods of symmetry-adapted perturbation theory (SAPT) [38][39][40][41]. The SAPT family of methods [38,39] is designed for accurate calculation of noncovalent interaction energies as well as a physically-motivated energy decomposition analysis of those energies [39,40]. Of key interest will be whether the interfacial environment engenders any discernible changes in the ion-water interactions, relative to what is observed for the same ion in bulk water.</p><!><p>MD simulations of the aforementioned ions, in a periodic slab configuration, were reported in Ref. 7 and the same set of simulations is used here to obtain snapshots for interaction energy analysis. These simulations were performed under NVT conditions at T = 298 K and a bulk density of 0.997 g/cm 3 , and the size of the periodic simulation cell (31.3 Å × 31.3 Å × 156.7 Å) was previously shown to afford converged results [7]. The simulation data were subsequently partitioned into bulk and interfacial parts depending on the position of the ion relative to the Gibbs dividing surface (GDS) that we take to define the air/water interface. For the snapshots classified as "interfacial", the ion's center of mass lies no more than 3 Å below the GDS. Anything beyond this cutoff is considered to be a bulk water environment, as this interior region of the periodic slab affords properties that are essentially indistinguishable from results performed in an isotropic simulation that has no interface [7]. Simulations were performed using the AMOEBA force field for water [42], whose parameterization includes some of the ions in question, such as the halides [43]. Parameters for the remaining ions were developed in Ref. 7 along similar lines, following an established protocol [44], and are included in the Supplementary Material. Energetic analyses with the AMOEBA force field were performed using the Tinker software, v. 8 [45].</p><p>Following an equilibration period, snapshots were extracted that include two solvation shells around the ion, according to the distance criteria described in Ref. 7. The number of water molecules varies from one snapshot to the next, with the average number N w depending on both the size of the ion and how tightly hydrated it is. In bulk water, these averages range from N w ≈ 28 for Cl − (aq) up to N w ≈ 43-44 for Br − (aq) and I − (aq), with N w = 35-37 for the remaining ions. The interfacial snapshots contain fewer water molecules, on average, as the water density is smaller in the interfacial region. In the analysis that follows, we consider interaction energies (E int ) between the ion and its first two hydration shells. The quantity E int is intensive with respect to system size and this insulates our analysis against the step-to-step fluctuations in the number of water molecules that are included in these calculations. Ensemble averages reported below represent 51 snapshots for each ion in bulk water as well as 51 snapshots for each ion at the air/water interface, with each individual snapshot separated by 5 ps in time in the corresponding MD simulation. Coordinate files for these data sets are provided in the Supplementary Material.</p><!><p>Quantum-mechanical values of E int were computed using SAPT based on Hartree-Fock (HF) wave functions for the monomers and second-order perturbation theory for the intermolecular Coulomb operators, a method that is usually called SAPT0 [39,46,47], and which is closely related to second-order Møller-Plesset perturbation theory (MP2). However, because second-order dispersion is far from quantitative [38,39,48], we replace it in these calculations with a many-body dispersion (MBD) model [39,49,50], in what we have termed a "hybrid" or "extended" form of SAPT [39]. This method will be designated as SAPT0 + MBD. At this level of theory, results for small-molecule data sets suggests that errors in E int are within ∼ 1 kcal/mol of the best-available benchmarks [47,50], provided that adequate basis sets are employed [47]. All electronic structure calculations were performed using the Q-Chem software, v. 5.4 [51].</p><p>The interaction energy computed using SAPT0 + MBD is naturally partitioned as [38,39] E</p><p>The terms on the right represent electrostatics (E elst ), meaning the Coulomb interaction between isolated-monomer charge densities; exchange or Pauli repulsion (E exch ), which is the cost to antisymmetrize the isolated-monomer wave functions; induction (E ind ); and dispersion (E disp ) [39,52]. In our approach,</p><p>and</p><p>in Eq. ( 1) are the first-order SAPT electrostatic and exchange energies, while E disp is the dispersion energy computed using the MBD model [50]. The induction energy comes from second-order SAPT but warrants additional discussion and is detailed in Section 2.3. Previous basis-set testing of SAPT0 + MBD reveals that polarized triple-ζ basis sets are both necessary and sufficient to obtain converged energetics [38,47]. This is a unique feature of our hybrid approach to SAPT [39], which replaces the very slow basis-set convergence of perturbative dispersion with a model (MBD) that converges quickly, with the density. Tests for Cl − (aq) in Fig. 1 demonstrate that interaction energies computed using the 6-311+G(d,p) basis set agree with SAPT0 + MBD/def2-TZVPD values to within an average of 2.0 kcal/mol, in a total interaction energy that averages −106 kcal/mol. Relative to the More important than these relatively small differences is the fact that instantaneous values of E int fluctuate from snapshot to snapshot in a similar way in either basis set. For these calculations, which involve Cl − (H 2 O) n with an average of n = 28 water molecules, SAPT0 + MBD/6-311+G(d,p) calculations are 17× faster than the corresponding calculations with def2-TZVPD. (This speedup results largely from the absence of diffuse functions on hydrogen but also benefits from Q-Chem's very efficient handling of sp shells in Pople basis sets.) In the present work, we are concerned with comparisons between bulk and interfacial behavior rather than absolute interaction energies, and the need for ensemble averaging requires high throughput. As such, 6-311+G(d,p) is used for all subsequent SAPT calculations.</p><p>Interaction energies defined in Eq. (1) do not include relaxation of the monomer geometries, so E int is an interaction energy in the "vertical" sense, not a binding energy or a solvation energy. In considering the ion-water clusters X − (H 2 O) n extracted from MD simulations, we treat the entire water cluster (H 2 O) n as a single monomer for the purpose of computing E int and its components, then average over the ensemble of snapshots. Even so, the value E int corresponds to vertical removal of the ion. It includes the change in (electronic) polarization of the water molecules upon removal of the ion but does not include the (orientational) reorganization energy of the water to fill the void left behind by the ion.</p><p>Unless otherwise specified, all of the SAPT0 calculations reported herein use HF wave functions for the monomers. However, we will report a few SAPT0(KS) calculations [39,47] in which Kohn-Sham (KS) molecular orbitals from density functional theory (DFT) are used in place of HF orbitals. These SAPT0(KS) calculations employ the long-range corrected (LRC) density functional LRC-ωPBE [53]. Previous work has emphasized the importance of using an asymptotically correct exchange potential in SAPT calculations [47,48,54,55], and this condition can be achieved in practice via monomer-specific tuning of the rangeseparation parameter (ω) in LRC-ωPBE functional. Although "optimal tuning" of LRC functionals [56,57] is sometimes accomplished using the ionization energy (IE) theorem of DFT, IE = −ε HOMO ,</p><p>a more robust procedure in the present context is the "global density-dependent" (GDD) or "ω GDD " procedure [47,48,55]. This approach, which adjusts ω based on the size of the exchange hole, mitigates the strong dependence on system size that is observed when using IE tuning [47], and which might otherwise be a problem when studying water clusters of varying size [58]. For water, we use ω = 0.277 a −1 0 , which represents an average over several cluster geometries. For the ions, we tune ω individually at the optimized gas-phase geometry of each, resulting in a range of values from ω = 0.248 a −1 0 for iodide and ω = 0.261 a −1 0 for bromide, where the tails of the anion's density are most diffuse, up to ω = 0.398 a −1 0 for cyanate and ω = 0.405 a −1 0 for cyanide, where the density is most compact. (Recall that LRC functionals switch from semilocal to HF exchange on a length scale of ∼ 1/ω.)</p><p>In previous work we have often used self-consistent charge embedding of the SCF monomer wave functions as a means to incorporate many-body polarization effects into a pairwise SAPT calculation, albeit implicitly [38,[59][60][61][62]. The present work does not make use of any charge embedding, however, and instead the X − (H 2 O) n system is treated as dimers, with (H 2 O) n as one monomer. In principle, charge embedding could be used to describe these clusters more efficiently as (n + 1)-body systems with monomers X − and H 2 O, but we have chosen not to do so here. The dimer approach makes the SAPT interaction energies more directly comparable to those obtained using the AMOEBA force field.</p><!><p>In our calculations, the induction term in Eq. ( 1) is defined as</p><p>where the first two terms are the second-order (SAPT0) induction and exchange-induction energies, and</p><p>elst + E</p><p>(1)</p><p>is the so-called "δHF" correction [39]. It uses a counterpoise-corrected, supramolecular HF interaction energy (∆E HF int ) to correct the SAPT0 interaction energy for induction effects beyond second order in perturbation theory, which is crucial for the accurate description of hydrogen bonds [39,47]. See Ref. 63 for a definition of the second-order response ("resp") energies that appear in Eq. ( 5).</p><p>As defined in SAPT, the induction energy in Eq. ( 4) contains both polarization and charge transfer (CT),</p><p>for reasons that are discussed in Ref. [64]. In the analysis of hydrogen bonding it is often of interest to separate these effects but that separation has historically been considered problematic, and not just within the SAPT formalism; many schemes for separating polarization from CT exhibit strong dependence on the choice of basis set [64]. To accomplish the separation in Eq. ( 6) in a robust way that converges rapidly with respect to basis set, we use the machinery of a charge-constrained self-consistent field (SCF) calculation [65] in order to define a CT-free reference state. Here, the monomers are allowed to polarize one another but their charge densities are constrained to integrate to integer numbers of electrons. Because the SCF procedure is variational, lifting of this constraint necessarily lowers the energy (to that of the fully-relaxed SCF solution), and this energy lowering is taken to define E CT . The CT energy thus obtained is then subtracted from the induction energy to obtain the CT-free polarization energy, E pol = E ind − E CT [64,[66][67][68]. CT energies defined in this way are very nearly converged already in double-ζ basis sets [64]. This approach has previously been used to demonstrate that E CT furnishes a driving force for formation of quasi-linear hydrogen bonds in binary halide-water complexes [52,68]. Implementation of the charge-constrained SCF procedure requires a method to count electrons, and Becke's multicenter partition scheme [69] is commonly used for this purpose [65]. This approach divides space into Voronoi cells [70], which are regions of space that are closest to a particular nucleus, and then Becke applies a smoothing function at the boundaries of these polyhedra. Alternatively, and specifically for the purpose of defining a CT-free reference state in order to effect the partition suggested in Eq. ( 6), a counting procedure based on fragment-based Hirshfeld (FBH) weighting has also been suggested [66,68]. In the latter approach, the number of electrons contained in fragment A is defined as</p><p>where ρ(r) is the supramolecular electron density, which is integrated subject to a weighting function w A (r). That function is defined as</p><p>where ρ 0 X (r) is the charge density of isolated fragment X. The denominator in Eq. ( 8) is thus a superposition of isolated-fragment densities.</p><p>The Becke scheme can also be conceptualized as a form of Eq. ( 7) in which w A (r) is a smoothed version of a Heaviside step function, which switches rapidly between w A (r) = 0 and w A (r) = 1 at the boundaries of the Voronoi polyhedra. In practice, our implementation of Becke's procedure uses the "atomic size adjustments" that are described in Ref. 69, in which a set of empirical atomic radii [71] are used to adjust the boundaries of the Voronoi cells away from the midpoints of the internuclear vectors. As discussed below, this adjustment is crucial for systems with substantial size mismatch between nearby atoms.</p><p>Even so, the FBH approach strikes us as the more reasonable one, especially where anions are involved, because Becke's approach depends only on the positions of the atoms (along with the empirical atomic radii), whereas the weight function defined in Eq. ( 8) respects the diffuseness of the isolated anions's wave function. In the present context, this almost inevitably means that the extent of anion → water CT is smaller when the FBH approach is used, because the tails of the X − wave function cause a larger region of space to contribute to that fragment's integrated electron number, N X . As an example, Fig. 2 presents E CT computed using both methods, for each snapshot of I − (aq) in bulk water. The results are considerably different depending on which method is used to count electrons, with the FBH approach compressing the CT energy into the interval 0 > E CT > −2 kcal/mol whereas the Becke procedure affords values of |E CT | as large as 20 kcal/mol. The latter value is comparable to the the average magnitude of the total SAPT0 induction energy, which is E ind = −22.3 kcal/mol for I − (aq) in bulk water. (Note that energy components corresponding to attractive interactions are negative.)</p><p>Figure 3 shows the polarization energy (E pol = E ind − E CT ) that is obtained using either the Becke or the FBH weighting function to define the charge constraint. (Both definitions of E pol start from the same SAPT0 induction energy, E ind .) It is apparent that the two definitions afford step-to-step fluctuations that do not seem to correlate with one another. In the Becke definition, the size and shape of the Voronoi cell that contains the iodide anion is sensitive to the instantaneous values of all iodide-water distances in the first solvation shell, whereas the FBH definition uses a spherically-symmetric charge density for the isolated anion in order to define the charge constraint; the latter definition is less sensitive to fluctuations in the atomic coordinates. (Note that FBH definition certainly remains sensitive to the presence of hydrogen bonds [52,68].) CT energies for snapshots of I − (aq) in bulk water, computed using a charge-constrained SCF procedure with the charge constraint defined either using fragment-based Hirshfeld (FBH) weights (scale at left), or else Becke's multicenter partitioning procedure (scale at right). Results using the Becke scheme include the "atomic size adjustments" that are described in Ref. 69, wherein Slater's set of atomic radii [71] are used to adjust the boundaries of the Voronoi cells based on atomic size. For I − (aq), it is consistently the case that the CT-free reference state defined using Becke partition results in CT energies that are larger in magnitude: |E CT (Becke)| > |E CT (FBH)|. This is evident from the rather different energy scales in Fig. 2, but the situation is not the same for all of the anions. As a second example we consider ClO − (aq), which exhibits the largest values of |E CT | of any of the ions studied here, at least when the FBH definition is used. Figure 4 considers both definitions and examines how E CT fluctuates from snapshot to snapshot. Becke's partition predicts very little CT for ClO − in bulk water ( E CT = −1.2 kcal/mol) whereas the FBH definition results in an average value of E CT = −6.2 kcal/mol. In either case, E CT is consistently larger for the interfacial snapshots.</p><p>We will use the FBH-based definition for the remainder of this work, and our main interest is in understanding how various energy components compare when the ion is in a bulk versus an interfacial environment. As noted in the examples presented above, the magnitude of E CT can depend strongly on the method that is used to count electrons. This observation suggests that in other applications of constrained DFT [65], which is the more common form of charge-constrained SCF calculation (in contrast to the constrained HF calculations employed here), the results should be checked carefully to ensure that conclusions are robust with respect to the details of how the constraints are implemented.</p><p>The SG-3 quadrature grid [72] is used to integrate the SCF constraint equations as well as Eq. (7). As a technical aside, we note that the atomic size adjustments mentioned above are crucial in order to obtain results that are even remotely sensible, when Becke partition is used to implement the charge constraint. However, the original implementation of the charge-constrained SCF procedure in the Q-Chem program did not include these corrections [73], for reasons that are unclear because the same algorithm with these size adjustments was implemented in the NWChem program, by the same authors at around the same time [74]. Atomic size corrections were later added to Q-Chem's version of Becke partition for the purpose of SAPT-based CT analysis [68]. Absent these corrections, the Voronoi cell boundaries are placed at midpoints of the internuclear vectors, which affords unreasonable results in cases where neighboring atoms have very different size. This includes covalent bonds to hydrogen, where the midpoint definition causes too much density to be assigned to the smaller hydrogen atom, often leading to a negative charge assigned to hydrogen [68]. In the present work, neglecting the atomic size corrections leads to a significant fraction of the iodide's charge being assigned to first-shell water molecules, resulting in completely unrealistic CT energies whose magnitudes exceed the total SAPT0 induction energy. In our view, constrained DFT based on Becke partition should probably never be used without the atomic size corrections.</p><!><p>Figure 5 presents ensemble-averaged interaction energies for the sets of X − (H 2 O) n structures that are considered here, where X − is one of 12 monovalent inorganic anions. Two solvation shells of surrounding water are treated as a single monomer for the purpose of the SAPT calculations. Results are presented both at the quantum-mechanical SAPT0 + MBD/6-311+G(d,p) level [Fig. 5(a)] and also using the AMOEBA force field [Fig.</p><!><p>, where the latter is the same force field that was used for the simulations from which these X − (H 2 O) n structures were extracted. Bulk and interfacial data are averaged separately, with the criterion GDS − 3 Å used to decide whether a particular snapshot represents a bulk or an interfacial solvation environment.</p><p>There are two interesting observations to be made from the interaction energy data in Fig. 5. Foremost is the fact that differences between the bulk and interfacial mean values E int for a given ion are small compared to the fluctuations in the instantaneous value of E int . Bulk values of E int are systematically (slightly) larger in magnitude than interfacial values, except for CN − , OCN − , and NO − 3 where the averages are essentially identical. In all cases, however, the difference between bulk and interfacial average values of E int is well within the standard deviation in either quantity; see the numerical values that are provided in Table 1. For the halides, the modest reductions in E int at the interface (up to 7-8 kcal/mol for bromide and iodide) are consistent with results from classical MD simulations indicating that the average ion-water interaction is reduced, for all of the halides, as the ion moves towards the interface [21]. It should be noted that the simulations reported in Ref. 21 indicate that the enthalpic portion of the potential of mean force is more favorable for the heavier halides at the interface, as compared to its value in bulk water. As such, the rather subtle differences between ion-water interactions that are documented in our quantum-mechanical calculations are more than compensated by ion-induced changes in the water-water interactions [21]. This is consistent with our detailed structural analysis of the ions [7], which indicates very little change in the first-shell structure at the interface as compared to that in bulk water.</p><p>A second interesting observation is the generally strong correlation between classical (AMOEBA) and quantum-mechanical (SAPT) values of E int , even if the former are systematically smaller than the latter, e.g., by 15-19 kcal/mol for the halide ions. (These systematic differences are smaller for the other ions except in the case of ClO − 3 , which is discussed below.) For the halide ions, we use AMOEBA parameters that were originally developed by Ponder and co-workers [43], and we note that the discrepancies between the force field and the quantum chemistry that are documented in Fig. 5 are much larger than b These data are plotted in Fig. 5(a). c These data are plotted in Fig. 7(b). those reported in Ref. 43 for binary X − (H 2 O) complexes. This underscores the importance of considering larger ion-water clusters, given the many-body nature of polarization in aqueous systems [75][76][77][78][79][80]. Simulation of the hydration free energy of Cl − using AMOEBA results in an error of 11.9 kcal/mol with respect to experiment [43], assuming that the reference value is defined using the proton solvation energy of Tissandier et al. [81], which has since emerged as the consensus value [82][83][84]. In view of this, the systematic difference of 17 kcal/mol between AMOEBA and SAPT0 + MBD values of E int in bulk water (see Table 1) is not so dissimilar from previous results. Improvements to the AMOEBA force field for ions, using SAPT energy components as benchmark data, is a topic of contemporary interest [85][86][87].</p><p>The chlorate (ClO − 3 ) ion represents the lone exception to an otherwise systematic correlation between classical and quantum-chemical interaction energies. This particular species is much more strongly solvated by AMOEBA ( E int = −126.6 ± 9.9 kcal/mol in bulk water) than it is by SAPT0 + MBD ( E int = −85.8 ± 10.0 kcal/mol). Considering the chlorine oxyanions as a group, the trend amongst the AMOEBA values of</p><p>The fact that perchlorate (ClO − 4 ) is an outlier is easy to rationalize in terms of its tetrahedral symmetry and vanishing dipole moment, but the trend amongst the other three chlorine oxyanions is more puzzling. Ensemble-averaged SAPT0 + MBD energy components for the four species ClO − n (aq) are listed in Table 2, and it is seen that E int , E elst , and E ind all follow the same trend exhibited by the gas-phase dipole moments of the ions in question. However, this means that the trend amongst total interaction energies is different from that predicted by AMOEBA. Instead, for the quantum-mechanical calculations the trend (from strongly to weakly interacting</p><p>In contrast to the AMOEBA results, the SAPT0 + MBD calculations afford similar ensemble-averaged interaction energies for both ClO − 3 and ClO − 4 , meaning that the ClO − 3 value seems anomalously small, given that all of the chlorine oxyanions except for ClO − 4 has a sizable dipole moment. As a sanity check, we recomputed interaction energies for all of the ions using SAPT0(KS) + MBD, which includes intramolecular electron correlation. These results are plotted in Fig. 6(a), which should be compared to the corresponding SAPT0 + MBD results in Fig. 5(a). Total interaction energies at either level of theory are quite comparable, and in particular both methods exhibit the same trend amongst the ClO − n ions, which differs from the trend predicted by AMOEBA.</p><p>To investigate this further, we consider the SAPT0 + MBD energy components. These are plotted for each of the ions in Fig. 7, again separating bulk and interfacial environments and ensemble-averaging over either data set. In considering the energy decomposition in Eq. ( 1), we have opted to group first-order electrostatics and exchange together,</p><p>because their sum approximates the electrostatic interaction between antisymmetrized monomer wave functions. This combination of "primitive" electrostatics (E elst , which is the Coulomb interaction between isolated-monomer charge densities) and Pauli repulsion (E exch ) has proven to be easier to interpret for halide-water systems as compared to electrostatics alone [52,68]. An example can be found in the ensemble-averaged energy components for the ClO − n (aq) species (Table 2), where the much less repulsive value of E exch for perchlorate at first seems at odds with the larger size of this ion. However, the reduced Pauli repulsion in this case is actually commensurate with a much less attractive value of E elst , suggesting a hydration sphere that is not as tight around the ion as it is in smaller (but electrostatically much more attractive) ClO − n ions. Statistical distributions of E elst+exch are shown in Fig. 7(a) for all of the ions, and immediately ClO − 3 stands out as the only ion for which E elst+exch > 0, meaning that the sum of first-order interactions is net repulsive in this case but is net attractive for each of the other ions. These observations are independent of whether one considers the bulk or interfacial data sets because differences between the bulk and interfacial mean values of E elst+exch are tiny in comparison to the instantaneous fluctuations, as was the case for E int . Furthermore, this anomalous prediction regarding ClO − 3 is not unique to the SAPT0 level of -</p><p>-10 theory that is used in Fig. 7. A similar anomaly is evident in the SAPT0(KS) results, which can been seen from the statistical distributions of E elst+exch at that level of theory [Fig. 6(b)]. We note that the largest values of E exch often correspond to the largest (most attractive) total interaction energies, as is seen for example in SAPT calculations of ClO − n • • • C 6 H 6 complexes (n = 1, 2, 3, 4) [88]. In the present case, ClO − 3 bucks this trend, according to the energy components listed in Table 2.</p><p>A possible explanation for the apparently anomalous behavior of ClO − 3 can be found by examining radial distribution functions (RDFs), g(r), obtained from the MD simulations. (These can be found in the Supporting Information for Ref. 7 but the salient details are described here.) Amongst the chlorine oxyanions, a unique feature of ClO − 3 is the appearance of two distinct peaks in the RDF for Cl• • • O w (where O w denotes water oxygen), one at r ≈ 3.5 Å and another at r ≈ 4.1 Å. For each of the other ClO − n species, the RDF consists of a single well-resolved feature at r ≈ 3.5-3.7 Å. The shorter-r feature for ClO − 3 does not appear to be present in simulations based on a hybrid quantum mechanics/molecular mechanics (QM/MM) formalism, which were used to interpret x-ray scattering results [89]. If the small-r feature for ClO − 3 is an indication of an extraneous water molecule present at short range, then this could explain the anomalously repulsive values of E elst+exch that we then compute at snapshots extracted from the MD simulations. The presence of such a water molecule in those simulations, however, suggests that something in AMOEBA's ion-water interaction is compensating for the short-range repulsion, or perhaps that the latter is simply not repulsive enough. Although polyvalent anions are not considered in the present work (because they are excluded from the air/water interface), it is notable that a short-r peak in the S• • • O w RDF is also observed in the simulations of SO 2− 3 (aq) that were reported in Ref. 7. These feature also appears to be absent from QM/MM simulations and x-ray scattering experiments [90]. In view of this, AMOEBA parameterizations for both of these ions ought to be revisited. This is beyond the scope of the present work, though it is interesting to note the way that SAPT analysis of ion-water clusters was able to detect an anomaly. Notably, vertical ionization energies computed for ClO − 3 (aq) and SO 2− 3 (aq) in Ref. 7 are no less accurate, as compared to experimental values [37], than what we obtain for other inorganic anions including other ClO − n ions [7]. The typical accuracy reported in Ref. 7 is ∼ 0.2 eV, considerably smaller than the widths of the corresponding photoelectron spectra.</p><p>Returning exclusively to the monovalent ions and examining the other energy components whose statistics are summarized in Fig. 7, another curiosity arises in regard to dispersion energies for the chlorine oxyanions. Dispersion is size-extensive, so that all else being equal it should scale in proportion to the number of electrons. For the ClO − n species, however, we observe that |E disp | decreases in the order ClO − 3 > ClO − 2 > ClO − > ClO − 4 . This time, perchlorate is the apparent anomaly. Dispersion energies in Fig. 7(c) were computed using the MBD model [50], so as a sanity check we recomputed E disp using the third-generation ab initio dispersion potential aiD3 [38], which consists of atom-atom C 6 and C 8 potentials fitted to dispersion-only data from high-level SAPT calculations. Dispersion energies obtain for the ClO − n species with both dispersion models are provided in Table 3 as ensemble averages. Both models afford rather similar dispersion energies, consistent with previous tests for cases where many-body effects on E disp are not significant [50]. (In the context of dispersion, "many-body" implies an effect that cannot be described by pairwise atom-atom potentials [48,91]. These typically arise in conjugated molecules where screening effects significantly modify the effective C 6 coefficients [92]. For small molecules, three-body dispersion effects are quite small [78].) Notably, in the aiD3 model the C 6 and C 8 coefficients depend only on atomic number and do not respond to the electronic structure of the monomers.</p><p>The sharp drop in dispersion between chlorate (ClO − 3 ) and perchlorate is a feature of both dispersion models, suggesting that this is not an artifact. A likely explanation is that in perchlorate, the addition of a fourth oxygen atom around the central (and more polarizable) chlorine atom screens the water molecules from this polarizable center, and thus significantly attenuates chlorine's contribution to the dispersion energy. In contrast, for the other ClO − n ions the chlorine atom remains solvent-exposed and the dispersion is much larger. This mechanism would be reflected in both dispersion models, if only as a function of increased chlorine-water distance in the aiD3 case. Also in support of this hypothesis are the data in Fig. 7(b) for SAPT0 + MBD induction energies, which also exhibit a pronounced drop in magnitude between ClO − 3 and ClO − 4 . As compared to dispersion interactions, polarization effects decay more slowly with distance, e.g., as r −4 for charge-dipole polarization, but this dependence is still rather steep. Polarization is often invoked in discussions of ions at the air/water interface [9][10][11][12][13][14][15][16][17], so it is interesting to note that induction energies are systematically smaller in the interfacial environment [Fig. 7(b)]. As with the total interaction energies, however, the difference between bulk and interfacial mean values E ind is small in comparison to the instantaneous fluctuations as measured by the standard deviation. The numerical data corresponding to Fig. 7(b) are provided in Table 1. Note that "polarization" as it is typically understood means strictly intramolecular redistribution of charge, with CT considered as a separate effect, and these two parts of the induction energy are separated in Fig. 8. Because the CT-free polarization energy (E pol ) is much larger than the CT energy (E CT ), the result is that E pol follows essentially the same trend from ion to ion as does the total induction energy, E ind . In particular, this means that the polarization energy is systematically smaller in the interfacial environment, for each of the ions considered here. Indeed, for the canonical surface-active anions Br − , I − , ClO − 4 and SCN − [19,34,35,93], the polarization energy is significantly smaller in the interfacial environment, by at least the standard deviation of E pol in bulk water; see Fig. 8(a).</p><p>That observation, in turn, is a direct result of CT energies that are systematically larger at the interface for precisely those four surface-active anions. Statistical distributions of E CT for all of the ions are plotted in Fig. 8(b). In contrast to other energy components, only for E CT do we observe pronounced difference between averages computed for the bulk and interfacial data sets. That said, the overall scale of the CT energies is a rather small part of either the total induction energy or the total interaction energy, with |E CT | 10 kcal/mol except in the case of interfacial ClO − . (Although CT energies smaller than 10 kcal/mol do play a pivotal role in establishing the directionality of hydrogen bonds [52,68], that kind of detailed analysis of a potential energy surface is not attempted in the present work, where we are interested in ensemble-averaged properties.) For Br − , I − , ClO − 4 , and SCN − , the average CT energy at the air/water interface is larger than its mean value in bulk water by at least one standard deviation in the bulk value. For Cl − (aq), the interfacial average value of E CT is larger in magnitude than the bulk value, though not quite by a full standard deviation. It is perhaps noteworthy that outliers for the CT energies tend to be larger at the interface, particularly towards negative (more stabilizing) values of E CT .</p><p>In the context of the Hofmeister series [94,95], the anions I − , ClO − 4 , and SCN − have especially large binding constants to protein [95,96], which is historically associated with the definition of chaotropes or "structure breakers" [8], whereas Cl − binds more weakly [96]. That said, NO − 3 is usually categorized as a structure-breaker on par with Br − in the Hofmeister series [94], and as weakly surface-active on the basis of VSFG measurements [19], yet the mean values of E CT that we obtain for NO − 3 are essentially identical in the bulk and interfacial environments, albeit with larger outliers in the interfacial case. The hypochlorite ion (ClO − ) stands out in this analysis, with a significantly larger mean value of |E CT | in the interfacial environment. This species is not typically discussed in the context of the Hofmeister series or in VSFG studies of the air/water interface, due to its limited stability in aqueous solution.</p><!><p>Detailed analysis of anion-water clusters extracted from MD simulations reveals that the total ion-water interaction energy (considering two solvation shells around the ion) is systematically larger for a given ion in bulk water than it is for the same ion near the air/water interface. The same is true for the CT-free polarization component of the total interaction energy, which is interesting given that polarization is often assumed to play a central role in surface activity [13], although this contention is disputed [23,24]. In any case, we observe systematically larger polarization energies in bulk water for both the "soft" anions with low surface charge density that are considered surface active (Br − , I − , ClO − 4 , and SCN − ) as well as for hard anions that are not considered surface active (CN − , OCN − , and NO − 2 ). That said, systematic differences in the mean values E int and E pol in bulk versus interfacial environments are rather small in comparison to the magnitude of the instantaneous fluctuations in E int and E pol .</p><p>Anion-to-water CT stands out as the only energy component whose magnitude is larger at the air/water interface for some of the ions. In fact, it is larger specifically for the traditional surface-active anions: Br − , I − , ClO − 4 , and SCN − . However, NO − 3 can also be detected in surface-sensitive vibrational spectroscopy [19], yet for that species E CT is essentially the same at the interface as it is in bulk water. The Cl − ion is a borderline case whose average CT energy is slightly more stabilizing at the interface, albeit by less than one standard deviation in the fluctuations. In all cases, the CT energy constitutes less than 20% of the total induction energy, meaning that it is at least 5× smaller than the CT-free polarization energy, the latter of which does not exhibit a surface preference and is in fact larger in bulk water. Nevertheless, the consequences of this "excess" CT for soft anions at the air/water interface interface seems worth pursuing in future work, especially in the context of VSFG experiments. Intermolecular CT mechanisms have been invoked in the past to explain the surface charge of liquid water that is inferred from electrophoretic measurements [97][98][99][100].</p><p>Considering the halide ions as a series that ranges from kosmotropic to chaotropic [8], or equivalently whose surface activities decrease in the order I − > Br − > Cl − F − , it has previously been noted that no single mechanistic explanation for this ordering can be gleaned from atomistic simulations [21,24]. Changes in the water-water interactions as the an ion approaches the interface appear to play a role [21]. The present analysis, based on accurate quantum-mechanical calculations of ion-water interaction energies, supports the notion that ion-water interactions alone do not readily afford any kind of a diagnostic (let alone a mechanism) to determine whether an ion resides in a bulk or interfacial environment. This null result complements our recent conclusion that short-range (firstshell) solvation structure is extremely similar in the bulk and interfacial environments [7]. The detailed mechanism of soft anion surface activity remains an open question.</p>

ChemRxiv

SAR and QSAR modeling of a large collection of LD50 rat acute oral toxicity data

The median lethal dose for rodent oral acute toxicity (LD50) is a standard piece of information required to categorize chemicals in terms of the potential hazard posed to human health after acute exposure. The exclusive use of in vivo testing is limited by the time and costs required for performing experiments and by the need to sacrifice a number of animals. (Quantitative) structure–activity relationships [(Q)SAR] proved a valid alternative to reduce and assist in vivo assays for assessing acute toxicological hazard. In the framework of a new international collaborative project, the NTP Interagency Center for the Evaluation of Alternative Toxicological Methods and the U.S. Environmental Protection Agency’s National Center for Computational Toxicology compiled a large database of rat acute oral LD50 data, with the aim of supporting the development of new computational models for predicting five regulatory relevant acute toxicity endpoints. In this article, a series of regression and classification computational models were developed by employing different statistical and knowledge-based methodologies. External validation was performed to demonstrate the real-life predictability of models. Integrated modeling was then applied to improve performance of single models. Statistical results confirmed the relevance of developed models in regulatory frameworks, and confirmed the effectiveness of integrated modeling. The best integrated strategies reached RMSEs lower than 0.50 and the best classification models reached balanced accuracies over 0.70 for multi-class and over 0.80 for binary endpoints. Computed predictions will be hosted on the EPA’s Chemistry Dashboard and made freely available to the scientific community.

sar_and_qsar_modeling_of_a_large_collection_of_ld50_rat_acute_oral_toxicity_data

Introduction<!><!>Datasets<!><!>Datasets<!><!>Balanced random forest (BRF)/random forest in regression (rRF)<!><!>Balanced random forest (BRF)/random forest in regression (rRF)<!>aiQSAR<!><!>aiQSAR<!>istkNN<!>SARpy<!>Random forest with hyper parameter tuning (HPT-RF)<!><!>Generalized linear model (GLM)<!><!>Integrated modeling<!>Model parameterization<!><!>Statistical performance<!><!>Integrated models evaluation<!><!>Integrated models evaluation<!>Discussion<!><!>Discussion<!>Conclusions<!><!>Conclusions<!><!>Supplementary information

<p>Over the past 25 years, synthetic organic chemical production world-wide has increased dramatically, from about 50 million tons to approximately 150 million tons [1]. This ever-growing increase of chemical substances represents a primary issue for the environment and human safety. Toxicological tests need to be performed to evaluate which of these chemicals are safe and which can potentially contaminate the environment and cause toxicity.</p><p>In the first stages of toxicological testing programs, acute toxicity studies are frequently used to categorize the agent in terms of the potential hazard posed to human health. Acute toxicity describes the adverse toxicological effects of a chemical that occur either from a single exposure or from multiple exposures in a short period of time (usually less than 24 h) [2].</p><p>The median lethal dose (LD50) is the basis for the toxicological classification of chemicals for various regulations concerning chemical hazard [3, 4]. The acute LD50 is the lethal dose of a substance that will kill 50% of the test animals/organisms within 24 h of exposure to the test substance [5–7]. Acute toxicity studies are conducted following various routes of exposure (e.g. oral, dermal and inhalation), and rodents are the most common animal model employed to estimate the lethal dose [8]. The estimation of rodent acute toxicity provides a baseline value when detailed toxicity data are unavailable for the chemical(s) of interest. In this case, LD50 values may be employed to make a first assessment of relative toxicity among chemicals [6].</p><p>However, the exclusive use of in vivo testing has obvious limitations, related to the high monetary and time cost of performing such experiments, the need to sacrifice a number of animals, and the number of chemicals requiring assessment. Indeed, it has been reported that toxicological and other safety evaluations represent 8% of the total number of animals used for experimental purposes in Europe, with rodents being the most commonly used specie [9, 10].</p><p>In light of this, recent laws are pushing the acceptance of alternative methods (e.g., in vitro and in silico methods) and their use by the regulatory and public health bodies in order to reduce the use of animals [11, 12]. Computational toxicology is a viable approach to reduce both the cost and the number of animals used for experimental toxicity assessment [13].</p><p>Structure–activity relationship and quantitative structure–activity relationship [(Q)SAR] models are in silico approaches to determine the toxicity of a large number of chemicals by analyzing their chemical structure. These methods are increasingly used to fill the toxicological data gaps for high-production volume chemicals (e.g., pharmaceutical, agrochemical, food additives and industrial) [3, 6, 14, 15] because they require a relatively small amount of resources and time.</p><p>Despite this, the reported number of (Q)SAR studies on mammalian toxicity is limited [3, [9], with the majority being restricted to particular classes of chemicals and based on small, focused datasets [16, 17, 18].</p><p>Recently, the NTP Interagency Center for the Evaluation of Alternative Toxicological Methods (NICEATM) in collaboration with the U.S. Environmental Protection Agency (EPA) National Center for Computational Toxicology (NCCT) compiled a large list of rat acute oral LD50 data on ~ 12 k chemicals. These data have been made available to the scientific community, to serve as the basis for an international collaborative modeling initiative. The modeling initiative was launched by the Interagency Coordinating Committee on the Validation of Alternative Methods (ICCVAM) Acute Toxicity Workgroup with the aim to develop new computational models for predicting five specific acute oral systemic toxicity endpoints required for regulatory purposes [19].</p><p>These five endpoints of regulatory significance for acute oral toxicity included the identification of (1) "very toxic" chemicals (i.e., LD50 less than 50 mg/kg); (2) "non-toxic" chemicals (LD50 greater than or equal to 2000 mg/kg); (3) point estimates for rodent LD50; (4) categorization of toxicity hazard using the U.S. EPA classification scheme [20]; (5) categorization of toxicity hazard using the United Nations Globally Harmonized System of Classification and Labelling (GHS) classification schemes (United Nations 2009).</p><p>The present article encompasses all the efforts that our research group contributed to this initiative. Both regression and classification computational models were developed for the five endpoints, by employing several statistical (i.e. QSAR) and knowledge-based (i.e. SAR) methods. External validation performance was provided to demonstrate the predictive capacity of the models. In the end, integrated modeling strategies were proposed to improve performance of single models. A multi-objective optimization based on the concept of Pareto optimum was proposed for identifying the best solution (i.e., individual or integrated model) for each modeled endpoint [21, 22].</p><!><p>LD50 single point estimates (continuous) expressed in mg/kgbw, and converted for modeling purposes in logarithm of mmol/kgbw.</p><p>"Very toxic" (vT) binary classification: LD50 < 50 mg/kg (positive class) and LD50 ≥ 50 mg/kg (negative class).</p><p>"Non-toxic" (nT) binary classification: LD50 > 2000 mg/kg (positive class) and LD50 ≤ 2000 mg/kg (negative class)</p><p>Category I (LD50 ≤ 50 mg/kg) is the highest toxicity category. Category II (moderately toxic) includes chemicals with 50 < LD50 ≤ 500 mg/kg. Category III (slightly toxic) includes chemicals with 500 < LD50 ≤ 5000 mg/kg. Safe chemicals (LD50 > 5000 mg/kg) are included in Category IV.</p><p>Category I includes chemicals with LD50 ≤ 5 mg/kg. Category II includes chemicals with 5 < LD50 ≤ 50 mg/kg. Category III includes chemicals with 50 < LD50 ≤ 300 mg/kg. Category IV includes chemicals with 300 < LD50 ≤ 2000 mg/kg). Category V includes chemicals with LD50 > 2000 mg/kg.</p><!><p>NICEATM and NCCT compiled and curated a rat acute oral systemic toxicity (LD50) inventory with values expressed as mg/kg of body weight (bw). This dataset was split semi-randomly, i.e. ensuring equivalent coverage with respect to LD50 distribution (and corresponding classes and categories for binary and categorical endpoints), by the organizers of the project into a list of compounds to be used for modeling (75%; 8994 chemicals) and validation (i.e., evaluation set, ES) (25%; 2895 chemicals). All the data and project information were made available to the cheminformatics community by NICEATM and NCCT at https://ntp.niehs.nih.gov/pubhealth/evalatm/test-method-evaluations/acute-systemic-tox/models/index.html.</p><!><p>Number of chemicals included in the TS and the ES for each toxicity class</p><p>Hazard categories for the two multi-class modeled endpoints (EPA and GHS classes) are sorted for decreasing toxicity, from 1 to 5. For the vT classification, class 1 corresponds to the positive (i.e., very toxic) compounds, while for the nT endpoint class 1 corresponds to the negative (i.e., toxic) compounds</p><!><p>The ES was initially imbedded into a larger prediction set by NICEATM and NCCT to facilitate a blind evaluation of all the models that were developed by the various institutions during the initiative. It was subsequently released with all the information relative to the five endpoints (https://ntp.niehs.nih.gov/iccvam/methods/acutetox/model/validationset.txt). While the TS used here were a result of a reworking of data provided for modeling, the ES was used as released for validating models presented here. Deduplication was performed by organizers of the project based only on CAS registration numbers. Consequently, a small degree of superimposition was observed between the TS and the ES due to the presence of different CAS numbers pointing to the same chemical structure. This overlap in chemistry is limited (i.e., about 8% of ES chemicals were also included in the TS) and it does not undermine statistical relevance of validation results. Table 1 summarizes, for each endpoint, the number of chemicals included in each toxicity category for TS and ES. In all cases, TS and ES showed a nearly analogous distribution of chemicals among the various classes.</p><p>The entire TS and ES are included in Additional file 2: Table S1 and Additional file 2: Table S2. The TS was analyzed by means of a principal component analysis based on CDK descriptors [27] implemented in KNIME analytical platform [31]. Eight chemicals identified as structural outliers based on score values on the first two principal components were removed from the TS.</p><!><p>Summary of modeling methods used</p><p>"Error" model</p><p>Confidence</p><p>Similarity</p><p>Mirror matrix</p><p>Isolation forest</p><p>For each method, the software, the descriptors used, the applicability domain definition and the modeled endpoints are specified. The methods listed are balanced random forest (BRF)/regression random forest (rRF); ab initio QSAR (aiQSAR); istKNN; hyper-parameter tuning random forest (HPT-RF); generalized linear model (GLM)</p><!><p>Data split For each compound, structural fingerprints were calculated using the Indigo toolkit implemented in KNIME (http://lifescience.opensource.epam.com/indigo/). K-means clustering was applied, taking into account the structural information (codified by the fingerprints) and the experimental values of each chemical. TS was split into an internal training set (iTS) (80%) and an internal validation set (iVS) (20%) based on a stratified sampling of the obtained clusters, to ensure structural and activity analogy between the two datasets. The iTS was used for the development of QSAR models, while the iVS was used for the tuning of model and applicability domain (AD) parameters. The number of chemicals included in iTS and iVS for each modeling endpoint is reported ("Results" section).</p><p>Algorithms Random Forest in regression (rRF) [28, 29] was applied for the derivation of LD50 single point estimate prediction models. For the categorical ones (i.e., nT, vT, GHS and EPA), balanced random forest (BRF) was used. This technique is a combination of under-sampling and the ensemble idea that artificially alters the class distribution so that classes are represented equally in each tree [30]. This allowed handling of unbalanced distributions of chemicals among classes for some of the categorical endpoints. The number of trees for each model was varied among 50, 100 and 150, then the best solution was selected based on performance on the iVS. All algorithms were implemented in the KNIME platform [31].</p><p>Descriptors Molecular descriptors were calculated for each compound using Dragon software [32]. Descriptors for iTS compounds were pruned by constant and semi-constant values (i.e., standard deviation < 0.01), then those having at least one missing value were removed. In case of highly correlated pairs of descriptors (i.e., absolute pair correlation higher than 95%), only one was retained and the descriptor showing the highest pair correlation with all the other descriptors was removed. Descriptors were normalized in the range of 0–1, then the same normalization scheme was applied to iVS descriptors.</p><!><p>Similarity A matrix containing pairwise Manhattan distances (based on Dragon descriptors used in the model) was calculated for iTS compounds. Chemicals were sorted based on the mean distance with respect to their first k neighbors and then the value corresponding to a given percentile of the distribution of distances was used as a threshold (TD). Chemicals with mean distances above TD were excluded from the AD. The same procedure was repeated for iVS chemicals with respect to their neighbors in the iTS, for identifying compounds outside of AD. Values assigned to k were 1 and 5; values assigned to TD were those corresponding to the 100th, the 97.5th, the 95th and the 90th percentiles of the iTS distance distribution. This method was applied on both continuous (LD50) and classification models.</p><p>Error model An "error model" predicts the uncertainty of the predictions coming from a classical "activity model". An error model was derived from the same iTS of the associated activity model, with the difference that the cross-validated absolute errors (previously generated by the activity model) represent the dependent variables while independent variables are represented by a series of AD metrics that reflect the accuracy of the predictions made by the activity model [33]. The RF algorithm was used for the error model derivation. iTS chemicals were sorted based on errors in prediction estimated by the error model, then the value corresponding to a given percentile of the distribution of predicted errors was used as a threshold (TE). Chemicals exceeding TE were excluded from the AD. The same TE was applied on predicted errors calculated for iVS chemicals. For the present work, values assigned to TE corresponded to the 100th, the 90th, the 75th and the 65th percentile of the iTS errors distribution. This method was applied only on the continuous LD50 point estimate models.</p><p>Confidence The percentage of trees within the RF yielding the same prediction (i.e., confidence) was estimated. This method was applied only for classification models. For binary classification models (i.e., vT and nT) a confidence threshold (TC) was gradually incremented by 0.05, from a minimum of 0.60 to a maximum of 0.75. For multi-class models (i.e., EPA and GHS), the confidence threshold was incremented by 0.10 from a minimum of 0.30 to a maximum of 0.70. Chemicals having confidence lower than this threshold were considered outside of the model AD.</p><!><p>For each model, the various parameters were evaluated in each possible combination, i.e. TD, k, TE for continuous models, TD, k, TC for classification models. The best combination of parameters was selected based on the best trade-off in terms of coverage and performance on the iVS (see "Results" section).</p><!><p>Data split The entire TS was used to derive aiQSAR models.</p><!><p>Workflow for aiQSAR-based model development</p><p>(Adapted from [34])</p><!><p>Similar compounds were selected on the basis of Tanimoto distances computed from the comparison of "PubChem" (ftp://ftp.ncbi.nlm.nih.gov/pubchem/specifications/pubchem_fingerprints.txt) and "Extended" fingerprints [35] between the target compound and all the TS compounds, as implemented in the "rcdk" R package [36]. A minimum of 20 and a maximum of 50 compounds were selected for the development of each model. First all compounds that are above the threshold value are selected (PubChem" similarity ≥ 0.80 and "Extended" similarity ≥ 0.70) Then, in case where the required number is not met, average ranks of compounds are considered to either add additional compounds, or to discard some of the selected ones.</p><p>Several mathematical models implemented in "caret" R package [37] were built from the local group of neighbors and each model was used to predict the value of the target compound. The type and the number of models varied based on the endpoint (i.e., regression, binary classification, multi-class classification) and is listed in Fig. 1. Further details on the methods used are reported in Vukovic et al. [34].</p><p>Finally, predictions from all methods were combined into a single output value for each target compound. For regression endpoints (i.e., LD50 point estimates), the final prediction was the average of predictions from individual methods, after discarding those more than 10% out of the range of experimental values of the TS. For classification endpoints (i.e., vT, nT, EPA and GHS), a majority vote approach was applied. In case of any tied votes, the class that is least represented in TS (out of the tied classes) was selected. In the present work, this always corresponded to the most toxic class in the tie being selected.</p><p>Descriptors Molecular descriptors were calculated using Dragon 7 software [32]. All available 1D and 2D descriptors (3839 overall) were considered for modeling purposes. Dragon descriptors with a missing value in any compound from the local group or in the target compound were iteratively discarded before each local model generation, as well as descriptors that were constant and near-constant within the local group of neighbors.</p><p>Applicability domain Applicability domain measure (ADM) of the target compound was computed based on average values of "PubChem" and "Extended" fingerprint similarities between the target compound and its local group of neighbors. The more similar is the local group of neighbors, the higher is the ADM score, that ranges between 1 (out of AD) and 5 (in AD). For the present work, chemicals with ADM ≥ 2 (i.e., "PubChem" similarity ≥ 0.60 and "Extended" similarity ≥ 0.30) were considered within the model's AD.</p><!><p>Data split The same data split described in "BRF/rRF" paragraph ("Data split") was used for selecting model's optimal parameters (see below). Once optimal parameters were selected, the global TS (iTS + iVS) was used to derive a new model that was validated on the ES.</p><p>Algorithm istkNN is a commercial tool [38] implementing a modified k-Nearest Neighbors (kNN) algorithm. kNN estimates the outcome of a sample in a dataset on the basis of read-across accounting for the k most similar samples (neighbors) in the TS for which the outcomes are known [39, 40]. If the algorithm is applied for predicting continuous endpoints (e.g., LD50 point estimates) the mean of the activities of neighbors is calculated [41].</p><p>Similarity Similarity between chemicals is described by an integrated similarity index (SI) ranging from 1 (maximum similarity) to 0 (minimum similarity), resulting from a weighted combination of a binary fingerprint array and three non-binary structural keys, as follow:\documentclass[12pt]{minimal} sepackage{amsmath} sepackage{wasysym} sepackage{amsfonts} sepackage{amssymb} sepackage{amsbsy} sepackage{mathrsfs} sepackage{upgreek} \setlength{\oddsidemargin}{-69pt} egin{document}$${ ext{SI}}\, = \,{ ext{S}}\left( { ext{FP}} ight)0.4*{ ext{S}}\left( { ext{CD}} ight)0.35*{ ext{S}}\left( { ext{HE}} ight)0.1*{ ext{S}}\left( { ext{FG}} ight)0.15$$\end{document}SI=SFP0.4∗SCD0.35∗SHE0.1∗SFG0.15where CD are structural keys with 35 constitutional descriptors (MW, nr of skeleton atoms, etc.), HE are structural keys with 11 hetero-atoms descriptors and FG are structural keys with 154 functional groups (specific chemical moieties) as implemented in Dragon (v. 7.0.8, Kode srl, 2017) (Talete srl, Milano, Italy). FP are the extended fingerprints, which comprise Daylight notation (http://www.daylight.com/dayhtml/doc/theory/theory.finger.html) and additional bits accounting for ring features. FP similarity is calculated with Maxwell-Pilliner index while CD, HE, and FG similarities are calculated with Bray–Curtis index [42].</p><p>Applicability domain istKNN refines the classical kNN algorithm by setting additional conditions that a sample (i.e. chemical) should fulfill to be considered reliably predicted. The k nearest neighbors used for prediction should have a similarity value with the target greater than a given threshold (Tsim1), otherwise they are not used for prediction. If no neighbors match the threshold, the model does not provide a prediction for the target compound. If only one neighbor matches the threshold, the similarity should be higher than a second stricter threshold (Tsim2) to return a prediction (which is equal in this case to the experimental values of this selected neighbor). If two or more neighbors fulfill the Tsim1, the range of experimental values of retained neighbors is considered. If the difference between the maximum and minimum experimental values of neighbors is lower than a threshold (Tmin–max), the target is predicted, otherwise the model does not return predictions. To calculate the prediction when more than one neighbor is selected, the experimental values of the similar compounds can be weighted differently on the basis of their similarity with the target (by setting an enhancement factor that increases the weight to the most similar compounds in the prediction).</p><p>In the present work, a batch process was used to optimize the settings of the five customizable parameters according to the following criteria: (a) number of neighbors from 2 to 5; (b) Tsim1 from 0.70 to 0.90 (step = 0.50); Tsim2 = 0.85 or 0.90; (c) enhancement factor from 1 to 3; (d) Tmin-max from 1.0 to 2.0 (step = 0.50).</p><p>The iTS was used for the development of (Q)SAR models using all the possible combinations of the parameters above. A restricted pool of valid models was pre-selected based on leave-one-out cross-validation on the iTS, seeking a good compromise in terms of coverage and performance. The final model was the one among the pool with the best performance in external validation on the iVS with a coverage of at least 0.85 (arbitrary threshold). Finally, the selected parameter settings were used to derive a new model on the global TS (iTS + iVS) and to use it for the prediction of the ES.</p><!><p>Data split The entire TS was used to derive SARpy models.</p><p>Algorithm The freely available SARpy software (https://www.vegahub.eu/portfolio-item/sarpy/) was used to build classification models for nT and vT classifications.</p><p>Given a TS of molecular structures described by SMILES, SARpy applies a recursive algorithm considering every combination of bond breakages working directly on the SMILES string. The software generates every structural fragment included in the TS that is then encoded into SMARTS (www.daylight.com/dayhtml/doc/theory/).</p><p>Each substructure is validated as a potential structural alert (SA) by verifying the existing correlation between the incidence of a particular molecular fragment and the class of activity of the molecules containing it. In this way, a reduced ruleset of relevant SAs is defined. Each SA was associated with an activity label (e.g., positive or negative) and a likelihood ratio, estimating the statistical relevance of the SA [43, 44].</p><p>Applicability domain If a chemical does not contain any fragment present in the rule set, it is not predicted and is considered outside of the model AD.</p><!><p>Data split The whole TS was used to derive HPT-RF models.</p><p>Algorithm "caret" [37] and "ranger" [45] R packages were used to develop regression (LD50) and multi-category classification (EPA and GHS) RF models [28]. Hyperparameter tuning (HPT) research was performed during the RF derivation, in order to select an "optimal" model across various parameters. Selection was made by evaluating the effect of model tuning on performance (i.e., accuracy) in internal validation (i.e., bootstrap). Three parameters were tuned by grid search: (1) mtry (number of randomly selected descriptors used in each tree of the RF), (2) splitrule (the rule used to choose descriptors for a single tree, i.e. "gini" or "extratrees" for classification; "variance" or "extratrees" for regression), (3) min.node.size (minimal node size of trees). The number of trees was equal to 500. The reader is referred to the user's guides of the above-indicated packages for further details.</p><p>A first model run served to evaluate the presence of response outliers within the TS. The isolation forest [46] method for anomalies isolation as implemented in the "isofor" R package was used to identify outliers. In addition, chemicals were flagged as outliers if they were characterized by a high variance and high error among iterations of bootstrap internal validation (100 iterations). In a second run, outliers were excluded from model's derivation.</p><p>Descriptors Calculation, pruning and normalization of descriptors were the same as in "rRF/BRF" paragraph ("Descriptors").</p><!><p>Similarity PubChem Fingerprints (ftp://ftp.ncbi.nlm.nih.gov/pubchem/specifications/pubchem_fingerprints.txt) were calculated for each compound starting from SMILES using the "rcdk" R package. Fingerprint-based similarity (Tanimoto) between a target compound and all the compounds in the TS was computed. If the mean similarity with the three most similar compounds among those flagged as outliers was higher than the similarity with three most similar compound from the TS cleaned from outliers, the compound was considered out of AD.</p><p>"Dummy" matrix Descriptors of TS chemicals were randomly permuted (vertical permutation) to create a mirror TS. The shuffled TS was merged with the original one. Samples of the original TS were flagged as "real", while those of the mirror TS were flagged as "dummy". A RF classification model was built to distinguish real from dummy samples. External chemicals classified as "dummy" were considered outside of the model's AD [47, 48].</p><!><p>Data split The same data split described in "rRF/BRF" paragraph ("Data split")  was used. The original iTS for the nT classification was further split by 20% of the iTS as an internal calibration set (iCS, 1330 chemicals) in order to evaluate the accuracy of the model during the building process. The iVS was used for validating the final model.</p><p>Algorithm The model was built using the H2O 3.16.0.3 (https://www.h2o.ai/download/) package for GLM in R v. 3.4.0 (https://www.R-project.org).</p><p>Logistic regression (LR) is used for binary classification problems when the response is a categorical variable with two levels. In this case, only the nT endpoint was modelled. This approach was not applied to the vT endpoint due to a strong bias arising from few chemicals having a vT designation, and the inadequacy of the method in modeling highly unbalanced datasets. LR-GLM model was fitted by finding a set of parameters that maximizes the probability of an observation belonging to its experimental category. Parameter tuning was performed on the iCS. Penalties were also introduced to the model building process to avoid over-fitting, reducing variance of the prediction error and handling correlated predictors. Penalties (elastic method) were controlled by parameters \documentclass[12pt]{minimal} sepackage{amsmath} sepackage{wasysym} sepackage{amsfonts} sepackage{amssymb} sepackage{amsbsy} sepackage{mathrsfs} sepackage{upgreek} \setlength{\oddsidemargin}{-69pt} egin{document}$$palpha$$\end{document}α and λ that were set to 0.09 and 0.02, respectively. A more in-depth description of the algorithm is reported by [49].</p><p>Descriptors Calculation, pruning and normalization of descriptors were performed as in "rRF/BRF" paragraph ("Descriptors").</p><p>Applicability domain AD was not implemented for this model.</p><!><p>Integrated model for predicting LD50 point estimates was obtained by averaging predictions of the four individual models (i.e., rRF, aiQSAR, istkNN, HPT-RF). Only predictions within the AD of each model were considered for integration. The integrated model's AD was implemented by applying the concept of "integrated prediction fraction" (PF) described by [50]. In particular, the number of models returning a prediction in AD and contributing to the final integrated prediction for a given sample was considered. The higher was the number of used models, the more reliable the integrated prediction. A threshold was set based on the PF and then integrated predictions with a PF lower than the threshold were considered out of the integrated AD.</p><p>Integrated models for categorical endpoints (vT, nT, EPA, GHS) were obtained by using a majority voting approach. Similarly to above, a consensus score (CS) was used as an index of prediction reliability. In this case, CS was calculated as the number of single models returning the same prediction as the integrated one, minus the number of models returning a prediction different from the integrated one. Only predictions within the AD of each model were considered for integration. For example, three out of four nT models returned predictions for a given sample, two of them being positive and one negative. In this case, the integrated prediction for the sample was "positive" with a CS of 1. No predictions were returned in case of ties.</p><!><p>For all the endpoints, the variation of integrated performance with respect to the PF/CS was evaluated.</p><!><p>Table 3 reports final settings for each model developed with the rRF and BRF method.</p><p>For the istkNN approach, the selected model was characterized by the use of 3 neighbors maximum, Tsim1 = 0.80, Tsim2 = 0.85, enhancement factor = 3 and Tmin-max = 2.0.</p><p>For the SARpy model, 349 fragments were identified for the vT endpoint (64 for the vT class, 285 for the not vT class), while 446 fragments were identified for the nT endpoint (228 for the nT class, 218 for the class of toxic compounds). The list of fragments encoded as SMARTS are reported in Additional file 3: Table S3a (vT model) and Table S3b (nT model).</p><!><p>rRF and BRF settings</p><p>For each model, the size of the internal training set (iTS) and internal validation set (iVS), the number of descriptors (#descrs), the number of trees (#trees) and the tuned parameters for AD definition are indicated</p><p>HPT-RF settings</p><p>For each model, the number of descriptors in each tree (mtry), the rule for descriptor selection for single trees (splitrule) and the minimal node size of trees (min.node.size) are indicated</p><p>External performance of single models for predicting single point logLD50 (mmol/kg)</p><p>For each model, the R2, the mean absolute error (MAE), the root-mean squared error (RMSE), the number (#AD) and the percentage (%AD) of predictions in AD are reported. The best values for each metric are italicized</p><p>External performance of single models for predicting classification endpoints (vT, nT, EPA, GHS)</p><p>For each model, the sensitivity (SEN), the specificity (SPE), the balanced accuracy (BA), the Matthew's correlation coefficient (MCC), the number (#AD) and the percentage (%AD) of predictions in AD are reported. For multi-category endpoints (EPA and GHS), SEN and SPE are the average of values computed separately for each class, while BA is the arithmetic mean of the average SEN and SPE. The best values for each metric are italicized</p><!><p>Internal validation is not discussed in the present article. The interested reader is referred to Additional file 1: Tables S4 and S5 for internal performance of each model.</p><p>Predictions of all models on the TS and ES are included in Additional file 2: Tables S1 and S2.</p><p>All continuous logLD50 (mmol/kg) predictive models showed good predictivity on ES. In particular, aiQSAR, istkNN and HPT-RF models showed a similar behavior with an R2 higher than 0.60 and RMSE near 0.54. On this basis, the istkNN method was particularly appealing for its intuitiveness and simplicity, compared with other more computationally-demanding methods. The rRF model was slightly less predictive, but with a higher percentage of predictions in AD (%AD) (i.e., 0.900) compared to other models (Table 5).</p><p>As far as classification models, Matthew correlation coefficient (MCC) [51] and balanced accuracy (BA) metrics were used for an overall estimation of classification performance for their ability to deal with unbalanced classes. Cooper statistics [52] were also calculated. For multi-category classifiers (EPA and GHS), the generalized formula of MCC reported by Ballabio et al. [53] was used, while BA was obtained as arithmetic mean of averaged sensitivities and specificities calculated for each separate category.</p><p>With respect to binary classification models, all the models showed good predictivity on the ES, with BAs close or higher than 0.80 and MCCs often higher than 0.50 (i.e., BRF and aiQSAR). In both cases, the best techniques was BRF, that returned BAs in external validation of 0.839 (for nT) and 0.880 (for vT), and MCCs of 0.674 (for nT) and 0.585 (for vT), despite a slightly lower %AD (i.e. 0.728) with respect to other methods (Table 6). This confirmed our previous experience [54] on the suitability of this technique to effectively handle highly unbalanced datasets.</p><p>As far as multi-categorical endpoints, EPA predictive models showed close performance across algorithms, with MCCs higher than 0.40 and BAs close to 0.73 in all cases. The HPT-RF method showed high performance, but at the cost of a slight loss in coverage with respect of other methods. aiQSAR was close in terms of performance, but with a gain in %AD (i.e., 0.891 compared to 0.763) (Table 6).</p><p>For GHS models, aiQSAR and HPT-RF performed better than BRF. However, coverage for GHS models was disappointing, being close to or lower than 0.50 (Table 6). This is reasonable due to the challenging nature of the GHS endpoint, characterized by a high number of (unbalanced) categories (n = 5) (Table 1).</p><!><p>External performance of the continuous integrated model for predicting single point logLD50 (mmol/kg)</p><p>The R2, the mean absolute error (MAE), the root-mean squared error (RMSE), the number (#AD) and the percentage (%AD) of predictions in AD are reported, with respect to the PF threshold for defining predictions in AD</p><p>Comparison of experimental and predicted (integrated) logLD50 (mmol/kgbw) for ES chemicals. Darkest circles indicate samples with a high prediction fraction (PF), while lightest circles indicate samples with lower PF. Dashed line represents the case of ideal correlation, while dotted lines delimit samples with an absolute error in prediction lower than 1.00 log unit</p><p>External performance of continuous models for predicting classification endpoints (vT, nT, EPA, GHS)</p><p>For each model, the sensitivity (SEN), the specificity (SPE), the balanced accuracy (BA), the Matthew's correlation coefficient (MCC) the number (#AD) and the percentage (%AD) of predictions in AD are reported, with respect to the CS threshold for defining predictions in AD. For multi-category endpoints (EPA and GHS), SEN and SPE are the average of sensitivities/specificities computed separately for each class, while BA is the arithmetic mean of the average SEN and SPE</p><!><p>The concept of "Pareto optimum" was applied to retrospectively confirm the improvement of consensus strategy with respect of single models, keeping into account both statistical performance and %AD. The approach allows solving of multi-objective optimization problems, in which no single solution exists that simultaneously optimizes multiple quality criteria. A solution is called "nondominated" (i.e., Pareto optimal) if no other solutions exist for which an evaluation function can be improved without degrading one of the other functions [21, 22].</p><!><p>Overview of Pareto optimal solutions for regression and classification models. a Performance of continuous LD50 models is described as root-mean squared error (RMSE) versus percentage of compounds in the AD (%AD). b Performance of classification (nT, vT, EPA, GHS) models are described as balanced accuracy (BA) versus percentage of compounds in the AD (%AD). All the parameters refer to the ES. Models in the bottom-left part of the plots are characterized by the best compromise in terms of performance and coverage, with dotted lines representing the Pareto front for a given endpoint. White indicators are single models (R = rRF; B = BRF; H = HPT-RF; K = istkNN; S = SARpy; Q = aiQSAR), while black indicators are integrated models, flagged with the corresponding PF (for regression) or CS (for classification) threshold</p><!><p>As shown in Fig. 3, integrated models always represented the optimal Pareto solutions for each endpoint, confirming the effectiveness of the integrated approach. Conversely, it was never possible to designate the best integrated model among those obtained varying the PF/CS value. Indeed, the increase of the threshold always resulted an increased performance, but at the cost of a systematic reduction of %AD.</p><p>For LD50 point estimate models, the integrated model with PF = 0.75 returned a coverage (i.e., 90%) analogous to best individual models, but with a relevant gain in performance (i.e., RMSE = 0.512 with respect of a mean RMSE = 0.542 for the best three individual models) (Tables 5 and 7).</p><p>For vT and nT endpoints, the BRF models were the closest to the Pareto front, however both solutions are dominated by the related integrated model with CS = 2.</p><p>As for multi-categorical endpoints, the aiQSARs were both close to the Pareto front and in particular to the integrated model with CS = 1. In this case, aiQSAR and integrated models have comparable performance, with a BA close to 0.730 for both endpoints. In both cases, integrated models showed a gain in %AD with respect to aiQSAR models, i.e. 3.7% additional predicted chemicals for the EPA model and 3.0% for GHS.</p><!><p>This work describes the modeling efforts our research group contributed in the development of new (Q)SAR models for predicting five endpoints (one continuous, four classification) related to acute oral toxicity in rats, as a result of our participation in the collaborative project launched by NICEATM and NCCT [19]. This endpoint is of utmost importance to several regulatory frameworks, being currently the basis for the toxicological classification of chemicals [4].</p><p>To date, the recent literature reports only a small number of successful attempts in modeling oral rat acute toxicity [3, 9]. The small number of models available can be explained by the nature of the endpoint and the lack of curated datasets prior to this effort. Indeed, compared with other widely modeled endpoints (e.g., (eco)toxicity) the modeling of mammalian toxicity is challenging, representing the sum of a plethora of toxicological mechanisms, each involving different biological pathways and molecular events concurring to the final effect [3, 55].</p><p>Individual (Q)SAR methods often showed inherent limitations in developing single models able to handle different mechanism of action, even more in case of a lack of complete understanding of some of the biological mechanisms contributing to the overall effect (e.g., death) [6, 41]. This issue can, however, be compensated with large enough datasets that adequately represent the diversity of the chemical universe.</p><p>An additional obstacle is often represented by the lack of reliable toxicity data in terms of quality, source of experimental data and organisms used [56]. The involvement of absorption, distribution, bioaccumulation, metabolism and excretion aspects further contributes to making the whole scenario even more complex and challenging [3, 56].</p><p>As a consequence, (Q)SAR models for this endpoint so far have been largely limited to small datasets restricted to a well-defined class of chemicals while global models are few and, often, not satisfactory in terms of predictive power [3, 9].</p><p>Despite this, the demand of in silico tools for such complex in vivo endpoints and (Q)SAR models continues to grow, due to objective resource and ethical limitations related to the execution of in vivo tests for a high number of chemicals. In this regard, the need of developing global models addressing a common endpoint such as acute toxicity have arisen as a primary need [3].</p><p>Within the project of NICEATM and EPA's NCCT Acute Toxicity Modeling Consortium, the availability of a large and high-quality rat oral acute toxicity database and the involvement of the entire scientific community represented an unprecedented chance to improve the current state of the art on the in silico modeling of this endpoint. To the best of our knowledge, this is the largest curated dataset of heterogeneous chemicals made available so far for the development of new (Q)SARs.</p><p>Integrated modeling was also applied in order to improve predictions of single models. As demonstrated previously and in the current work, integrated models had the highest external prediction power compared to any individual model used in the integrated prediction [57, 58, 59].</p><p>This is particularly true when the individual models have been developed using different techniques, have ADs differently defined and showed different behaviors based on the structural and activity profile of predicted chemicals. For example, an inspection of correct predictive rates that classification models have on specific categories of toxicity showed that some methods performed better on certain classes (e.g. more toxic or less toxic compounds) than others (for detailed statistics see Additional file 1: Tables S8–S11).</p><p>In this regard, the integrated method can compensate for and correct the limitations of individual techniques, as well as afford greater chemical space coverage [57, 59].</p><!><p>External performance of the continuous integrated model for separate PFs</p><p>The R2, the mean absolute error (MAE), the root-mean squared error (RMSE), the number (#) and the percentage (%) of predictions with a given prediction fraction (PF) are reported. In addition, the percentage of samples with a given PF value and an absolute error in prediction equal to or greater than 1.00 log unit with respect of the total number of samples with the same PF (%AE ≥ 1) are reported</p><!><p>Machine learning methods we used (e.g., RFs) proved to be valid in terms of predictive performance, but a mechanistic interpretation of these models is often more difficult than classical linear models. Indeed they can be based on thousands of different molecular descriptors and the relationship existing between the endpoint and each descriptor is often a complex, non-linear one that is only implicitly included in the model itself. With this in mind, we provided an analysis of most relevant features used in the descriptor-based global models here presented (rRF/BRF, HPT-RF, GLM). Models generated with aiQSAR and istKNN were not considered in this analysis as they are local models not capable of identifying features associated with the global trend of acute toxicity. In addition, istKNN is based on the similarity concept only and not on descriptors.</p><p>The top twenty Dragon descriptors for each of the above-cited models were listed in Additional file 1: Table S12. Details on how the importance of descriptors was determined for each model were included in the same Table. Descriptors belonging to the 2D Atom Pairs category (binary, frequency or weighted topological atom pairs) were the most frequent (79 out of 180 descriptors). They were followed by 2D autocorrelation (16), CATS2D (14), functional group counts (11) and P_VSA-like descriptors (11). P_VSA_s_1 (P_VSA-like on I-state, bin 1) was also the single most frequent descriptor (it was included in four out of nine models considered). P_VSA descriptors are related to van der Waals surface area of chemicals and, indirectly, with their size and lipophilicity. Low LD50 compounds in the TS were characterized by higher values of P_VSA descriptors, suggesting a role of molecular size and lipophilicity in the offset of acute toxicity (e.g., in absorption/excretion and bioaccumulation of chemicals in tissues).</p><p>Looking at 2D Atom Pairs descriptors, those referring to the presence of phosphorus (31), carbon (26), fluorine, nitrogen and sulfur (19) atoms were those more frequent. In particular, chemicals having a higher number of phosphorus, fluorine and sulfur atoms had lower LD50 values with respect to the mean of the distribution of values for the TS. The number of phosphorus (nP) was also one of the single most frequent descriptors, that was found in HPT-RF models for EPA and GHS and in BRF/rRF models for single point LD50 and vT classification. Binary/frequency phosphorous-based 2D Autocorrelation descriptors (B01[O-P], B04[C-P] and F02[C-P]) also appeared in multiple models (three models each).</p><p>Functional group counts are easier to be interpreted with respect to other theoretical descriptors, because they describe the presence of well-defined structural motif. For example, high values of nOHp (number of primary alcohols) were characteristic of low-toxicity chemicals. High numbers of hydroxyls flagged for high solubility of chemicals that influence the excretion rate, the capability to cross biological membranes and accumulate in tissues to exert toxicity. The presence of primary hydroxyls is important for phase II metabolism (conjugations) that contribute to detoxification of chemicals [17]. On the contrary, high number of aliphatic tertiary alcohols (nOHt) and aliphatic tertiary amines (nRNR2) was observed in high-toxicity compounds. It is possible that high counts of tertiary groups flags for bulky, lipophilic molecules that are easily to accumulate in the organism. Conversely, beta-Lactams, sulphonic and sulphuric acids, that are more hydrophilic, are only presents in safest toxicity categories.</p><p>Fragments identified by the SARpy model for vT class with LR = inf were evaluated too. Compared to the functional groups they can include larger fragments. The chemical moieties spotted with these fragments are halogenated 2-trifluoromethyl benzimidazoles, dioxins, phosphonothioates, organothiophosphates (including organothiophosphate aliphatic amides). They refer especially to chemical classes well represented among pesticides or former pesticides active ingredients.</p><!><p>In the present study, a series of computational models were developed as part of the NICEATM and EPA's NCCT collaborative project, for the prediction of five regulatory relevant endpoints describing rat acute oral toxicity (LD50). A series of different (Q)SAR methods were applied and the obtained models were validated on a large external dataset to assess their predictivity. Briefly, no single methods proved to be the best for all the endpoints, despite some of them constantly returning highly satisfactory predictive performance. In particular, results showed that some machine learning methods (e.g. RFs) were especially effective in modeling this kind of composite endpoint. These findings support the fact that machine learning approaches have often been indicated as promising tools in the field of computational toxicology [61, 62], and that they are able to handle multiple mechanisms of actions better than classical linear approaches [63, 64].</p><p>A review from Gonella-Diaza et al. [15] recently proposed an evaluation of the performance of existing models predicting LD50 implemented in a series of computational platforms. It was shown that only a few models were able to deliver acceptable predictive performance. Indeed, the best models among those evaluated returned RMSE values in external validation and within AD never lower than 0.55 for regression, while accuracy values for the five-class GHS classification ranged between 0.45 and 0.56. The models presented here showed robust performance in external validation, with RMSE values close to 0.50 for integrated models and BAs exceeding 0.80. In this regard, the models presented here can be easily considered an improvement on the current state-of-the-art for in silico modeling of this endpoint.</p><p>Another important outcome of this study is that integrated methods always returned improved performance with respect to single models. This confirms, as has already been widely reported, that the integration of multiple strategies and the application of a weight-of-evidence approach solves the limitations inherent to single methods and increases the confidence in the final toxicological prediction.</p><!><p>External performance of other published acute toxicity models developed within the NICEATM and EPA's NCCT collaborative project</p><p>R2 = 0.737</p><p>RMSE = 0.408</p><p>%AD = 0.347</p><p>SEN = 0.873</p><p>SPE = 0.915</p><p>MCC = 0.793</p><p>%AD = 0.263</p><p>SEN = 0.789</p><p>SPE = 0.998</p><p>MCC = 0.857</p><p>%AD = 0.320</p><p>MCC = 0.730</p><p>%AD = 0.159</p><p>MCC = 0.733</p><p>%AD = 0.223</p><p>SEN = 0.800</p><p>SPE = 0.840</p><p>BA = 0.820</p><p>%AD = 0.730</p><p>SEN = 0.850</p><p>SPE = 0.940</p><p>BA = 0.900</p><p>%AD = 0.770</p><p>For each method, performance for the five acute toxicity relevant endpoints are reported</p><!><p>Leveraging the collective expertise of the entire scientific community in a collaborative effort was the main aim of the NICEATM/NCCT initiative. Given the encouraging results of this first exercise, as well as the comparable results in validation of other research groups, authors strongly believe that the development of new, comprehensive integrated models will represent a further improvement to the already satisfactory results here presented. Indeed, the combination of several methods will mitigate the weaknesses of single models, towards a better collective consensus approach, as well as an enlarged chemical domain. Finally, all the predictions will be hosted on the EPA's Chemistry Dashboard and made freely available to the entire scientific community.</p><!><p>Additional file 1. Tables S4–S12 reporting additionals statistics for models, and details on dataset curation are included.</p><p>Additional file 2. Tables S1 and S2 including experimental values and predictions of models for TS and ES are included.</p><p>Additional file 3. Tables S3a and S3b with lists of SMARTS implemented in SARpy are included.</p><p>Additional file 4. The list of descriptors used for models derivation are included.</p><p>ab initio QSAR</p><p>applicability domain's</p><p>applicability domain measure</p><p>balanced accuracy</p><p>balanced random forest</p><p>consensus score</p><p>evaluation set</p><p>Environmental Protection Agency</p><p>Globally Harmonized System of Classification and Labelling</p><p>internal calibration set</p><p>internal training set</p><p>internal validation set</p><p>k-Nearest Neighbors</p><p>Matthew's correlation coefficient</p><p>median lethal dose</p><p>National Center for Computational Toxicology</p><p>Nontoxic</p><p>NTP Interagency Center for the Evaluation of Alternative Toxicological Methods</p><p>percentage of predictions in AD</p><p>prediction fraction</p><p>(quantitative) structure activity relationship</p><p>random forest with hyperparameter tuning</p><p>random forest in regression</p><p>root-mean squared error</p><p>structural alert</p><p>training set</p><p>very toxic</p><p>Publisher's Note</p><p>Springer Nature remains neutral with regard to jurisdictional claims in published maps and institutional affiliations.</p><!><p>Supplementary information accompanies this paper at 10.1186/s13321-019-0383-2.</p>

PubMed Open Access

2H,15N\xe2\x80\x93Substituted Nitroxides as Sensitive Probes for Electron Paramagnetic Resonance Imaging

Electron paramagnetic resonance imaging (EPRI) using nitroxides is an emergent imaging method for studying in vivo physiology, including O2 distribution in various tissues. Such imaging capabilities would allow O2 mapping in tumors, and in different brain regions following hypoxia or drug abuse. We have recently demonstrated that the anion of 3-carboxy-2,2,5,5-tetramethyl-1-pyrrolidinyloxyl (2) can be entrapped in brain tissue to quantitate O2 concentration in vivo. To increase the sensitivity of O2 measurement by EPR imaging, we synthesized 3-carboxy-2,2,5,5-tetra(2H3)methyl-1-(3,4,4-2H3,1-15N)pyrrolidinyloxyl (7). EPR spectroscopic measurements demonstrate that this fully isotopically-substituted nitroxide markedly improves signal-to-noise ratio and, therefore, the sensitivity of EPR imaging. The new isotopically-substituted nitroxide shows increased sensitivity to changes in O2 concentration, which will enable more accurate O2 measurement in tissues using EPRI.

2h,15n\xe2\x80\x93substituted_nitroxides_as_sensitive_probes_for_electron_paramagnetic_resonance_ima

Introduction<!><!>Results and Discussion<!>General Materials and Methods<!>4-Oxo-2,2,6,6-tetra(2H3)methyl-(3,3,5,5-2H4,1-15N)piperidine (3)<!>4-Oxo-2,2,6,6-tetra(2H3)methyl-1-(3,3,5,5-2H4,1-15N)piperidinyloxyl (4)<!>4-Oxo-2,2,6,6-tetra(2H3)methyl-1-(2H)hydroxy-(3,3,5,5-2H4,1-15N)piperidine (2H)hydrochloride (5)<!>3-Bromo-4-oxo-2,2,6,6-tetra(2H3)methyl-1-(3,5,5-2H3,1-15N)piperidinyloxyl (6)<!>3-(2H)Carboxy-2,2,5,5-tetra(2H3)methyl-1-(3,4,4-2H3,1-15N)pyrrolidinyloxyl (7)<!>3-Acetoxymethoxycarbonyl-2,2,5,5-tetra(2H3)methyl-1-(3,4,4-2H3,1-15N)pyrrolidinyloxyl (8)<!>EPR Spectroscopy<!>EPR Spectral Linewidth Measurements<!>

<p>Electron paramagnetic resonance imaging (EPRI) is an emergent magnetic resonance imaging modality that can be used to study a wide range of physiology.1 For example, molecular oxygen, being paramagnetic, broadens the electron paramagnetic resonance (EPR) spectral lines of other paramagnetic species, including trityl radicals2 and nitroxides3. Using these spin probes, one can make reliable, minimally invasive measurements of O2 concentration in vivo. Besides the significance of O2 generally in brain metabolism, the importance of O2 measurement in brain stems from the suggestion that oxygen-centered free radicals are responsible for methamphetamine-induced neurotoxicity,4 leading to altered O2 levels in the affected regions of the brain. Therefore, it is clinically important to map the regions of decreased O2. A major obstacle to using EPR imaging to quantify O2 in brain tissue is the difficulty of transporting O2-sensitive imaging probes across the blood-brain barrier.</p><!><p></p><!><p>In preparing isotopic-substituted nitroxide 7, 15N must be introduced in the first step, forming 4-oxo-2,2,6,6-tetra(2H3)methyl(3,3,5,5-2H4,1-15N)piperidine (3), prior to subsequent reactions, including ring contraction. We originally considered using the procedure of Rozantsev,6 which called for the condensation of acetone with ammonia. The reported low yield (19%) for this reaction, however, compelled us to seek an alternative synthetic pathway. A review of the literature uncovered the procedure of Lin et al.,7a in which heating perdeuteroacetone and 14ND4Cl with MgO in a sealed vessel gave the desired piperidine with 56% yield. 15ND4Cl is not commercially available. Therefore, it was prepared by isotope exchange: 15NH4Cl was dissolved in D2O and lyophilized; the process was repeated 4 times.</p><p>In our hands, heating perdeuteroacetone and 15ND4Cl with MgO in a sealed vessel yielded 3 in ~40% yield (Scheme 1). Whether 3 was purified by distillation made little difference in the yield of the next reaction; therefore crude 3 was used directly in the next step to ensure efficient use of isotopically-substituted material. The oxidation of 3 with D2O2 in D2O afforded nitroxide 4 in reasonable yield.</p><p>Although it has been reported8 that addition of HCl to isotopically-unmodified nitroxide 4 followed by bromination gave the corresponding 6 in good yield via the intermediate 5, Marc and Pecar9 found that reducing 4 by catalytic hydrogenation prior to bromination gave superior results. For the isotopically-unmodified 4, this procedure should give a nearly quantitative yield of 5. In the case of 2H-substituted nitroxide 4, however, the procedure could lead to hydrogen-deuterium exchange, owing to keto-enol tautomerization (Scheme 2). To eliminate this possibility, we initially considered using perdeuterated ammonium formate as the electron source for reduction,10 but ultimately decided to reduce piperidinyloxyl 4 to the hydroxylamine 5 with D2 over Pd/C in CH3OD.</p><p>Subsequent bromination of 5 at a single position α to the carbonyl, achieved by dropwise addition of Br2 in chloroform over 45 min, followed by NaNO2 oxidation, afforded the bromopiperidinyloxyl 6,8 with no evidence of dibromination.11 Importantly, piperidinyloxyl 4, which was recovered by chromatography, was recycled through the reaction to bring the total yield of 6 to 64%. Favorskii rearrangement of 6 with KOD8 gave the desired carboxylic acid 7.</p><p>Figure 1 shows EPR spectra for the K+ salts of nitroxides 2 and 7, each at 100 µM in air-equilibrated H2O ([O2] = 0.25 mM). Although the samples are at the same concentration, nitroxide 7 exhibits significantly larger and narrower EPR spectral peaks. Using relatively low magnetic fields permits EPR imaging with low-frequency electromagnetic radiation, which penetrates tissue well, but reduces SNR. Nitroxide 7 remedies this deficit with much narrower, and thus larger, spectral peaks. Using nitroxide 7 improves the limit of detection and enables imaging with higher contrast.</p><p>Figure 2 shows the dependence of the EPR linewidth on O2 concentration for nitroxides 7 and 2. As O2 concentration varies, the relative change in linewidth for nitroxide 7 is much larger than for nitroxide 2. Linear least-squares fits of the data show that the line-broadening effect of O2 is 2.74-fold greater for nitroxide 7 than for nitroxide 2. Thus, nitroxide 7 is more O2-sensitive and capable of resolving smaller changes in O2 than nitroxide 2. In EPR imaging, over-modulation is often used to improve SNR, at the cost of spectral line broadening. The threshold modulation amplitude at which significant line broadening occurs is lower for 7 than for 2. Therefore, over-modulation could reduce the difference in the line-broadening effect of O2 on the two nitroxides. The spectra in Figure 2 were acquired at a modulation amplitude of 0.125 G, at which line broadening was negligible for both nitroxides (Supplemental Figure S1). We measured the dependence of the linewidths of 7 and 2 on O2 at an increased modulation amplitude of 0.5 G (Supplemental Figure S2). At this larger modulation amplitude, the difference in the line-broadening effect of O2 is reduced from 2.74-fold to 2.60-fold. Thus even under typical conditions of over-modulation, nitroxide 7 is still far superior to nitroxide 2 as an O2 sensor.</p><p>Isotopic substitution of 15N and 2H in the pyrrolidinyloxyl ring is a significant advancement in using nitroxides as oxygen-sensitive probes in EPR imaging. We have shown previously that nitroxide 1, the isotopically-unmodified analogue of labile ester 8, is a pro-imaging agent that can cross the blood-brain barrier and be converted to nitroxide 2 in brain tissue.12 Therefore, the isotopically-substituted labile ester 8, like its isotopically-unmodified counterpart, is expected to cross the blood-brain barrier. When introduced into live animals, ester 8 will enable us to determine the actual extent to which isotopic substitution improves O2 imaging in vivo. These studies are underway.</p><!><p>Reagents and solvents from commercial vendors were used without further purification. Silica gel (230–400 mesh) was used for column chromatography.</p><!><p>This compound was prepared following the general procedure of Lin, et al.7a with minor modifications. To generate 15ND4Cl, 15NH4Cl (10 g) was dissolved in D2O (99.9%; 15 mL); the solution was evaporated to dryness under vacuum. This procedure was repeated a total of 4 times.</p><p>In a glove box under positive internal N2 pressure, 15ND4Cl (3.5 g, 60 mmol) was added to a 250-mL round bottom flask containing oven-dried anhydrous Na2CO3 (3.18 g, 30 mmol) and MgO (3.0 g, 75 mmol). Thereafter, perdeuteroacetone (12.5 mL, 150 mmol; 99.9%) was introduced into the flask by canula under N2 pressure. While still under N2 atmosphere, the flask was sealed with a rubber septum. The reaction mixture was kept for 3 days at 50°C in an oil bath, and then allowed to cool. Perdeuterocetone (20 mL) was added to the flask and the resulting mixture was filtered. The filter cake was crushed into a fine powder, washed with dry ether and perdeuteroacetone (1:1 mixture, 20 mL) and again filtered; this procedure was repeated three more times. The combined filtrates were concentrated on a rotary evaporator to give a red liquid (5.5 g), a portion of which was distilled to yield a yellow liquid (bp 60 – 64 °C at 12 mm Hg),7 which solidified upon cooling. In trial runs, we found that distillation of 3 did not significantly affect the yield of 4; therefore, crude 3 was used for the next reaction without further purification.</p><!><p>To a solution of crude 4-oxo-2,2,6,6-tetra(2H3)methyl-(1,3,3,5,5-2H5,1-15N)piperidine (3) (5.5 g, 34 mmol) dissolved in D2O (60 mL), oven-dried Na4EDTA (0.55 g, 1.5 mmol) and oven-dried Na2WO4 (0.55 g, 1.7 mmol) were added. Upon dissolution of the salts, D2O2 (30% in D2O, 6 mL) was added and the reaction was allowed to proceed in the dark for 10 days. The reaction mixture was filtered and extracted with ether (3 × 50 mL). The ether extract was first washed with cold dilute DCl (10% in D2O, 2 × 20 mL) and then with saturated Na2CO3 in D2O (10 mL). Thereafter, the solution was dried over anhydrous MgSO4, filtered, and reduced to dryness on a rotary evaporator. This residue was chromatographed (hexane:Et2O, 2:1) to yield 4-oxo-2,2,6,6-tetra(2H3)methyl-1-(3,3,5,5-2H4,1-15N)piperidinyloxyl 4 as a red oil, which solidified in the cold (2.8 g, 51%). IR (CHCl3): 1720 cm−1 (C=O). Anal. calculated for C92H1615NO2: C, 57.69; 2H, 8.61; 15N, 7.48. Found: C, 57.57; 2H, 8.58; 15N, 7.40.</p><!><p>The general procedure of Marc and Pecar9 was used with minor modifications. 4-Oxo-2,2,6,6-tetra(2H3)methyl-1-(3,3,5,5-2H4,1-15N)piperidinyloxyl (4) (2.8 g, 15 mmol) was dissolved in CH3OD (30 mL) and 5% Pd/C (50 mg) was added. Deuterium gas (99%) was gently bubbled into the reaction mixture for several min and the flask was sealed. The flask was periodically recharged with D2 over the next several hours. After stirring overnight, the reaction mixture was filtered through Celite. The colorless filtrate was acidified with 4 M DCl (in D2O, 2.5 mL) and reduced to dryness on a rotary evaporator. The residue was washed with dry ether (2 × 20 mL) to remove any remaining nitroxide and dried, in vacuo, to yield 4-oxo-2,2,6,6-tetra(2H3)methyl-1-(2H)hydroxyl-(3,3,5,5-2H4,1-15N)piperidine (2H)hydrochloride (5) as a white solid (2.7 g, 80%). Compound 5 is very hygroscopic and was therefore used immediately in the monobromination reaction to yield 6, as described below.</p><!><p>The procedure of Sosnovsky and Cai8 was used with minor modifications. Br2 (2.14 g, 11.9 mmol) in CHCl3 (10 mL) was added dropwise over 45 min at room temperature to a stirred solution of 4-oxo-2,2,6,6-tetra(2H3)methyl-1-(2H)hydroxyl-(3,3,5,5-2H4,1-15N)piperidine (2H)hydrochloride (5) (2.7 g, 11.9 mmol) in CHCl3 (25 mL); thereafter the reaction mixture was stirred for 2.5 h. Thereafter, a solution of NaNO2 (1.85 g, 27 mmol) in D2O (10 mL) was added dropwise over 10 min to the vigorously stirred reaction mixture; stirring was continued for another 15 min. The organic phase was washed with D2O, dried over anhydrous MgSO4, filtered, and evaporated to dryness under reduced pressure. Chromatography (hexane:Et2O, 2:1) yielded two fractions: 1) 3-bromo-4-oxo-2,2,6,6-tetra(2H3)methyl-1-(3,5,5-2H3,1-15N)piperidinyloxyl (6), and 2) nitroxide 4. Re-reducing the recovered piperidinyloxyl 4 with D2 over 5% Pd/C (25 mg) in CH3OD (30 mL) followed by bromination and then oxidation with NaNO2 led to a 64% overall yield of piperidinyloxyl 6 (2.0 g). Recrystallization from hexane yielded light orange crystals, whose purity was confirmed by HPLC (Supplemental Figures S3): mp = 81–82°C; IR (CHCl3): 1730 cm−1 (C=O). HRMS (ESI) calcd for C91H2H1515NO279Br [M + H]+ 265.12773, found 265.11910; calcd for C91H2H1515NO281Br [M + H]+ 267.12558, found 267.12953.</p><!><p>The general procedure of Sosnovsky and Cai8 was used, with minor modifications. KOD (1 M in D2O, 5 mL) was added to 3-bromo-4-oxo-2,2,6,6-tetra(2H3)methyl-1-(3,5,5-2H3,1-15N)piperidinyloxyl (6) (0.52 g, 2 mmol). With stirring over the next 2 h, nitroxide 6 dissolved completely. The alkaline solution was extracted with Et2O (3 × 20 mL), cooled in an ice bath, and adjusted to pH 3 with dilute DCl (10% in D2O). This acidic solution was extracted with Et2O (3 × 20 mL); the combined extract was dried over anhydrous MgSO4, filtered, and evaporated under reduced pressure to yield 3-carboxy-2,2,5,5-tetra(2H3)methyl-1-(3,4,4-2H3,1-15N)pyrrolidinyloxyl (7) as a yellow solid (0.25 g, 61%). Recrystallization from CHCl3/hexane gave a yellow powder, mp = 190 – 194 °C (with decomposition);13 IR (CHCl3): 3500 cm−1 (broad peak, OH), 1711 cm−1 (C=O).</p><!><p>The synthesis followed our reported procedure.14 Bromomethyl acetate (0.044 g, 0.03 mL, 0.33 mmol) was added to a solution of 3-carboxy-2,2,5,5-tetra(2H3)methyl-1-(3,4,4-2H3,1-15N)pyrrolidinyloxyl (7) (0.050 g, 0.25 mmol) and K2CO3 (0.070 g, 0.50 mmol) in acetonitrile (dried over CaH2, 10 mL); the mixture was stirred overnight at room temperature. Thereafter, this mixture was filtered through Celite, and the filtrate was evaporated to dryness. The oily residue was chromatographed (hexane:ethyl acetate, 5:1) to yield a thick oil, which was crystallized from hexane to yield 3-acetoxymethoxycarbonyl-2,2,5,5-tetra(2H3)methyl-1-(3,4,4-2H3,1-15N)pyrroldinyloxyl (8) as a yellow solid (0.050 g; 74%), mp = 77–78°C; IR (CHCl3): 1763 cm−1 (C=O). HRMS (ESI) calcd for C121H62H1515NO5 [M + H]+ 275.23316, found 275.23085. Anal. calculated for C122H151H515NO5: C, 52.52; 2H + 1H, 7.35; 15N, 5.10. Found: C, 52.76; 2H + 1H, 7.42; 15N, 5.12.</p><!><p>EPR spectra were recorded on an X-band spectrometer at the following settings: microwave power, 20 mW; microwave frequency, 9.55 GHz; field set, 3335 G for 14N or 3324 G for 15N; modulation frequency, 1 kHz; modulation amplitude, 0.125 G; field sweep, 4 G at 13.3 G min−1. These settings encompassed the central spectral peak of the 14N spectrum, or the first spectral peak of the 15N spectrum. For recording the complete spectrum of either nitroxide, the field set was 3335 G, and field sweep was 50 G at 13.3 G min−1. Digital acquisition of EPR spectra was through EWWIN software.</p><!><p>To assess the effect of O2 on the EPR linewidths of the nitroxides, deionized H2O (18.3 MΩ·cm resistivity) was sparged with N2, equilibrated with air, or sparged with O2 at 24°C for 30 min, to yield solutions containing O2 at 0.003 mM, 0.25 mM and 1.25 mM (O2-saturated15), respectively. Submillimolar O2 concentrations were determined using a dissolved O2 meter. Stock solutions (10 mM) of the K+ salt of nitroxide 2 or 7 were diluted 500-fold into the gas-equilibrated H2O samples to a final nitroxide concentration of 20 µM. Each solution was transferred into a flat quartz EPR cell previously purged with the appropriate gas. The quartz cell was sealed, and immediately positioned in the EPR spectrometer. Duplicate spectroscopic measurements on the samples were performed in random order. Reported linewidths are the peak-to-peak width of the central spectral line of 2, and the first spectral line of 7.</p><!><p>EPR spectra of nitroxides 2 and 7 (K+ salt, each at 100 µM in H2O equilibrated with air). Both spectra were acquired with identical spectrometer settings (see Experimental section) and are represented on the same intensity scale. The hyperfine splittings and linewidths are indicated.</p><p>Relative linewidths of nitroxides 2 and 7 at different O2 concentrations in water. For each nitroxide, the linewidths were normalized to the value measured at 0.003 mM O2 (in N2-sparged water). Solid lines are least-squares fits of the data; the slope of each line is indicated on the graph.</p><p>Reagents and conditions: (a) 15ND4Cl, MgO; (b) D2O2, D2O; (c) i. D2, Pd/C, CH3OD, ii. DCl, D2O; (d) i. Br2, CHCl3, ii. NaNO2, D2O; (e) i. KOD, ii. DCl; (f) CH3CO2CH2Br, K2CO3, CH3CN.</p>

PubMed Author Manuscript

Direct Conversion of Aldehydes and Ketones to Allylic Halides by a NbX5-[3,3] Rearrangement

Sequential addition of vinylmagnesium bromide and NbCl5, or NbBr5, to a series of aldehydes and ketones directly provides homologated, allylic halides. Transposition of the intermediate vinyl alkoxide is envisaged through a metalla-halo-[3,3] rearrangement with concomitant delivery of the halogen to the terminal carbon. The [3,3] rearrangement is equally effective for the conversion of a propargyllic alcohol to the corresponding allenyl bromide.

direct_conversion_of_aldehydes_and_ketones_to_allylic_halides_by_a_nbx5-[3,3]_rearrangement

<p>Allylic halides occupy a privaleged position as electrophiles.1 Activation of the carbon-halogen bond by adjacent π electrons dramatically facilitates nucleophilic displacement,2 allowing otherwise difficult bond constructions with a diverse range of carbon, heteroatom, and organometallic nucleophiles.3 Allylic halides prepared as intermediates during total syntheses are typically prepared by halogenation of allylic alcohols. These in turn are often obtained by olefination-reduction sequences4 because the direct "halo-olefination" of carbonyls with Wittig-type procedures is circumvented by elimination in the reagent.</p><p>A conceptually direct method for converting aldehydes and ketones to allylic halides is through a sequenced vinyl addition-metalla-halo-[3,3]-rearrangement (Scheme 1).5 Addition of vinylmagnesium bromide to an aldehyde or ketone generates the bromomagnesium alkoxide 2,6 potentially allowing transmetallation to a more oxophilic metal capable of weakening the carbon-oxygen bond for a metalla-halo-[3,3]-rearrangement.7 Particularly Lewis acidic halometal alkoxides 3 may cause competitive ionization and halogenation leading to a mixture of allylic halide regioisomers8 whereas metals bearing more nucleophilic halides would favor the rearranged allylic halide 4.</p><p>TiCl4 effectively promotes this sequence with aromatic aldehydes and with magnesium alkoxides derived from deprotonation of allylic alchohols.7 Results presented below show that NbCl5 and NbBr5 expand the scope of this vinyl addition-metalla-halo-[3,3]-rearrangement to include aromatic and aliphatic aldehydes and aromatic and aliphatic ketones. In addition, the [3,3] rearrangement is equally effective for the conversion of a propargylic alcohol to the corresponding bromoallene implying a concerted rearrangement in distinction to the related reactions with TiCl4.</p><p>Exploring the viability of a sequenced vinyl addition-metalo-halo-[3,3]-rearrangement initially focused on identifying an optimal metal and solvent combination. Using the potassium alkoxide 6 as a prototype, a diverse range of metal chlorides were evaluated for their effectiveness in providing the allylic chloride 4a (Scheme 2). Screening several oxophilic metals9 identified dry10 niobium pentachloride11 in 1,4-dioxane as the optimal combination. After only 10 minutes at room temperature the alcohol 5 was completely converted to the allylic chloride 4a in essentially quantitative yield. 1H and 13C NMR analysis of the crude product indicated the material to be pure.12</p><p>The use of niobium pentachloride is particularly intriguing because this strong Lewis acid13 is employed under basic conditions. These conditions contrast with many related reactions14 in which transition metal halides exhibit reactivity similar to that of HCl which might be liberated by contact of the reagent with moisture.15 Mechanistically attack of the alkoxide 6 on NbCl5 might trigger rearrangement from the niobiate 3a having an expanded coordination sphere (Scheme 2).16 Sequential ionization and chlorination of 3a is a mechanistic possibility, although the absence of regio- and stereoisomers suggests this to be a minor reaction manifold. Synthetically the NbCl5 rearrangement offers the advantage over related phosphonium-based reagents in facile isolation of (E)-allylic chlorides simply by an aqueous extraction to remove inorganic species.</p><p>Optimizing the metalo-halo-[3,3]-rearrangement with the allylic alcohol 5 provided a platform for directly converting aldehyde 1a to the allylic chloride 4a (Scheme 3). Partial ring opening of THF17 by NbCl5 led to a procedure in which vinylmagnesium bromide18 was added to a 10 °C, THF solution of aldehyde 1a followed by dilution with four volumes of 1,4-dioxane and addition of solid NbCl5 (1.2 equiv). After 30 minutes the solution was washed with aqueous HCl and concentrated to provide the crude chloride19 4a (99%).20</p><p>The naphthyl-substituted allylic chloride 4a is challenging to purify. Silica-based purification methods resulted in significant mass loss suggesting partial alkylation of the solid-phase.12 Consequently the crude chloride 4a was redissolved in THF and reacted with sodium phenylsulfenylate (Scheme 3). The procedure efficiently provided the sulfide 7a ratifying the efficiency of the niobium rearrangement in generating essentially pure allylic chloride 4a.</p><p>Performing the vinyl addition-niobium rearrangement-sulfenylate displacement with a series of aldehydes efficiently provides the corresponding sulfides (Table 1). Aromatic aldehydes and ketones smoothly rearrange in the presence of NbCl5 (Table 1, entries 1-5). The nitrile-containing aromatic aldehyde 1e was less reactive toward NbCl5 but was effectively halogenated with NbBr5 (Table 1, entry 5). Aliphatic aldehydes are converted to the corresponding allylic chlorides with NbCl5 but greater efficiency is obtained with NbBr5 (Table 1, entries 6-7). The aliphatic ketone cyclohexanone reacted sluggishly even with NbBr5 (Table 1, entry 8). In each case analysis of the intermediate allylic halide shows complete rearrangement prior to the sulfenylate displacement.</p><p>The metalo-halo-[3,3]-rearrangement is best suited to hydrocarbons. A conjugated aldehyde (Table 1, entry 4) and a nitrile (Table 1, entry 5) are readily tolerated whereas an acetal is not.21 Effort to further probe the functional group tolerance is in progress.</p><p>The metalla-halo-[3,3]-rearrangement is not limited to the addition of vinylmagnesium halide to aldehydes and ketones. The strategy was extended to halogenation of the propargylic alcohol 8a with the expectation that a concerted rearrangement would favor a haloallene (Scheme 4). Deprotonating 8a and adding NbCl5 afforded only a trace of the corresponding chloro allene at room temperature with full conversion requiring heating the reaction mixture to reflux. Under these thermal conditions considerable decomposition occurred whereas substituting NbBr5 triggerred a smooth rearrangement at room temperature. After 2 h the bromoallene 10a was obtained in 78% yield.</p><p>Direct transformation of aldehydes and ketones to the corresponding allylic halides is readily achieved through a vinyl addition metalo-halo-[3,3]-rearrangement strategy. Sequential addition of vinylmagnesium bromide and NbCl5 or NbBr5 to aromatic and aliphatic aldehydes and ketones provides essentially pure allylic chlorides for use in subsequent displacement reactions. The [3,3] rearrangement is equally effective in the case of a propargyllic alcohol, which provides the corresponding allenyl bromide. Synthetically, the addition-niobium halide rearrangement provides an efficient and direct conversion of aldehydes and ketones to allylic halides in one synthetic operation.</p>

PubMed Author Manuscript

Design, synthesis, and characterization of novel eco-friendly chitosan-AgIO3 bionanocomposite and study its antibacterial activity

This work reports a facile and green approach to preparing AgIO 3 nanoparticles decorated with chitosan (chitosan-AgIO 3 ). The bionanocomposite was fully characterized by Fourier transform infrared (FTIR), scanning electron microscopy (SEM) images, energy-dispersive X-ray spectroscopy (EDX), and X-ray diffraction analysis (XRD). The antibacterial effect of chitosan-AgIO 3 bionanocomposite was investigated for Pseudomonas aeruginosa, Klebsiella pneumoniae, Staphylococcus saprophyticus, Escherichia coli, and Staphylococcus aureus as pathogen microorganisms via the plate count method, disk diffusion method, and optical density (OD) measurements. The antibacterial performance of the bionanocomposite was compared with two commercial drugs (penicillin and silver sulfadiazine) and in some cases, the synthesized bionanocomposite has a better effect in the eradication of bacteria. The bionanocomposite represented great antibacterial properties. Flow cytometry was performed to investigate the mechanism of bionanocomposite as an antibacterial agent. Reactive oxygen species (ROS) production was responsible for the bactericidal mechanisms. These results demonstrate that the chitosan-AgIO 3 bionanocomposite, as a kind of antibacterial material, got potential for application in a broad range of biomedical applications and water purification. The design and synthesis of green and biodegradable antibacterial materials with simple processes and by using readily available materials cause the final product to be economically affordable and could be scaled in different industries.Regarding the spread of infectious disease attributable to pathogenic bacteria and the rise of antibiotic resistance, demand for the design and synthesis of unique antibacterial agents is increased 1 . Pseudomonas aeruginosa is a pernicious pathogen as a gram-negative aerobic bacillus isolated from soil, water, plants, and animals, including humans. Pseudomonas aeruginosa is also known to change its phenotype and attune to the environment and are multi-drug resistant bacterial species, affecting compromised immune system patients 2,3 . Klebsiella pneumoniae is a rod-shaped, gram-negative pathogen extensively found in the mouth, skin, intestines, in-hospital settings, and medical devices. The opportunistic pathogen Klebsiella pneumoniae mostly influences those with compromised immune systems or is weakened by other infections. Considering that Klebsiella pneumoniae has become increasingly resistant to antibiotics, successful eradication of this bacterium is very important 4,5 . Staphylococcus saprophyticus is related to uncomplicated urinary tract infection in humans 6 . Escherichia coli (gram-negative) and Staphylococcus aureus (gram-positive) cause diarrhea diseases in humans after infected water. A safe drinking water supply is a critical and essential aspect of human health 7 . Silver sulfadiazine (AgSD) has been used for the treatment of second-order burns since the early 1970s. It represents a more effective antibacterial dressing with improved stimulation of wound regeneration 8,9 . Penicillin is a β-lactam antibiotic that is useful against a wide range of bacteria. Penicillins are the drug of choice for upper and lower respiratory infections (Streptococcus pyogenes), meningococcal disease (Neisseria meningitides), syphilis (Treponema pallidum), and anaerobic infections 10 . The field of nanotechnology has been introduced as a high potential possibility for producing novel nanoscale materials which represent high surface to volume area and particular physical and chemical properties with wide applications 11 . Amongst them, a DNA nanostructure electrochemical biosensor was designed and

design,_synthesis,_and_characterization_of_novel_eco-friendly_chitosan-agio3_bionanocomposite_and_st

<!>Result and discussion<!>X-ray diffraction analysis of chitosan-AgIO 3 bionanocomposite.<!>Study the antibacterial activity of the chitosan-AgIO 3 bionanocomposite. Zone of inhibition<!>Conclusion

<p>synthesized for monitoring cyanazine herbicide in water and food samples 12 . Nitrogen and sulfur co-doped carbon dots were effectively synthesized from waste orange peels via a cost-natural and easy synthesis process as a nano-booster with high oxygen-reduction reaction activity 13 . Besides other oxygen-reduction reactions electrocatalyst in neutral media was designed and synthesized as an effective alternative to expensive metal-based nanocatalysts 14 . A novel carbon paste electrode modified with ZIF-8/g-C 3 N 4 /Co nanocomposite and 1-methyl-3-butylimidazolium bromide as an ionic liquid was utilized as an extremely sensitive electrochemical sensor for the detection of synthetic azo dyes 15 . A guanine-based DNA biosensor was designed and fabricated in a simple way for monitoring anticancer drugs during chemotherapy treatments 16 . Nanoparticles as an antibacterial agent with a large surface area to volume ratio that can provide better contact with bacterial cells. Considering that nanoparticles tend to aggregate, which could reduce the antibacterial property, utilizing support for nanoparticles will be very practical and helpful in producing a nanocomposite with high antibacterial efficacy 17,18 . Green nanotechnology supplies many superiorities in terms of process development, manufacturing, and product design. The synthesis of nanocomposite in the direction of green chemistry has many preponderances, including simple and mild reaction conditions that cause the process to be scaled up, economically affordable, and ecofriendly 19 . Bionanocomposite could be used in several fields, such as antibacterial agents 20 , bionanocatalysts 21 , photocatalytic activity 22 , adsorbents for heavy metals 23 , and drug delivery 24 . With a view to green chemistry, extending and utilizing biodegradable polymers is considered the most thorough method for designing bionanocomposites as an antibacterial agent. Polymers derived from natural resources, including starch, cellulose, chitin, chitosan, and lignin have emerged as promising candidates for synthesizing bionanocomposite with particular applications [25][26][27][28] . As a natural polymer derived from the marine environment, chitosan is an amino polysaccharide obtained by the deacetylation of chitin (poly-N-acetyl-D-glucosamine). Chitin is the second most abundant natural polymer after cellulose. Chitosan is the most beneficial derivative of chitin with numerous amine and hydroxyl groups in its structure, which enables the synthesis of biocomposite with various applications. Chitosan represented special features, including biodegradability, biocompatibility, nontoxicity, inexpensiveness, and availability 29,30 . Among the different antibacterial nanomaterials, silver nanoparticles and their compounds have substantial antimicrobial capabilities. Silver nanoparticles (AgNPs) attracted the attention of many researchers due to their antimicrobial activity against a wide variety of drug-resistant microorganisms 31 . Gold/silver-tellurium nanostructures (Au/Ag − Te NSs) has very good antimicrobial activity against different microorganism due to the generated ROS which destroys the bacteria membrane 32 . Silver nanoparticle anchored graphene oxide (GO-Ag) has shown good antibacterial activity 33 , Vancomycin capped with silver nanoparticles as an antibacterial agent 34 , Nanowires of silver-polyaniline bionanocomposite as an antibacterial agent 35 , cellulose/γ-Fe 2 O 3 /Ag bionanocomposite as an antibacterial agent 36 , are some of the examples of silver nanocomposite with antibacterial activity. AgIO 3 is an insoluble white crystal with an orthorhombic structure representing good photocatalytic activity for the decomposition of organic pollutants in the UV region because of its wide bandgap and high separation rate of photoexcited charge carriers [37][38][39] . Silver iodate nanoparticles are highly insoluble (AgIO 3 K sp = 3.1 × 10 -8 ) and could be used in water treatment as well as medical applications 40 . In connection with our previous research on the antibacterial activity of nanocomposite [41][42][43] , in this study, we present the preparation and identification of chitosan-AgIO 3 bionanocomposite as an antibacterial agent (Fig. 1). To the best of our knowledge, this is the first report on the antibacterial properties of bionanocomposite based on chitosan and silver iodate against several gram-negative and gram-positive bacteria. Chitosan-AgIO 3 bionanocomposite was introduced as a unique, cost-effective with high antibacterial activity. This paper opens a new approach in an antibacterial field involving economically and environmentally efficient nanoscale composite based on natural polymer. Moreover, the synthesized bionanocomposite could be employed in broad usages and could be scale-up due to its novel and specialproperties. Fabricating nanomaterials with available and green resources causes biodegradability and biocompatibility and reduces the cost of the synthesized products. Simple equipment and procedure and inexpensive and readily available materials without the use of any surfactants, external templates, and toxic solvents for synthesizing the bionanocomposite are of great importance. In order to acknowledge and emphasize on antibacterial activity of the chitosan-AgIO 3 bionanocomposite, the antibacterial performance of the bionanocomposite was compared with two commercial antibiotics against five human Synthesis of chitosan-AgIO 3 bionanocomposite. For the synthesis of chitosan-AgIO 3 bionanocomposite, 0.16 g AgNO 3 was dissolved in 10 ml of deionized water, then added to the 10 ml of 0.5% (m/v) chitosan solution and stirred for 30 min in a dark condition to adsorb Ag + ions. In the following 0.21 g, KIO 3 was dissolved in 10 ml deionized water, gradually added to the above solution, and stirred for 3 h. A milky white solid was obtained and washed with deionized water and ethanol and dried at 70 °C for 12 h.</p><p>Procedure for antibacterial studies with chitosan-AgIO 3 bionanocomposite. For the study, the antimicrobial performance of the synthesized bionanocomposite, standard agar diffusion test, colony count method, and antibacterial activity screening (OD study), was performed against various bacteria, including Pseudomonas aeruginosa (ATCC 27853), Klebsiella pneumoniae (ATCC 700603), Staphylococcus saprophyticus (ATCC 1440), Escherichia coli (ATCC 9637), Staphylococcus aureus (ATCC 12600). Besides, the antibacterial performance of chitosan silver iodate was compared with penicillin and silver sulfadiazine by utilizing a standard agar diffusion test. All instruments have been sterilized at 121 °C for 10 min in an autoclave before any process. Further, 0.5 McFarland turbidity standard to Nutrient Broth media was applied for antibacterial tests.</p><p>Antibacterial activity screening (ZOI study). Muller-Hinton agar was utilized as a base medium and a solid growth medium plus nutrients microorganisms; 20 ml of Muller-Hinton agar was added to the plate (8 cm) and 50 ml was added to the other plate (15 cm) in the sterile condition. Five different tube test which was filled with 10 ml of sterile physiological serum and each of the five microorganisms was adjusted with 0.5 McFarland turbidity, then with utilizing a sterile glass hockey stick, the culture (cell concentration was adjusted to 10 7 cells/ mL) was dispersed around the surface of the plate. In the following the 0.01 g of chitosan-AgIO 3 was added to the Muller-Hinton agar culture medium containing bacteria. Petri dishes containing bacteria and bionanocomposite were incubated for 24 h at 37 °C. To evaluate the antibacterial effects of the bionanocomposite against these five types of bacteria, and compared it with commercial drugs plates consisting of penicillin and silver sulfadiazine (with different concentrations) were studied in the separate plate which was prepared as mentioned above.</p><p>The growth inhibition zones of chitosan-AgIO 3 bionanocomposite against Pseudomonas aeruginosa, Klebsiella pneumoniae, Staphylococcus saprophyticus, Escherichia coli, and Staphylococcus aureus were measured.</p><p>Colony count method. One of the most practical and useful methods for studying the effect of bionanocomposite as an antibacterial agent against bacteria is the colony count method. Escherichia coli and Staphylococcus aureus were grown for 24 h in Muller-Hinton agar and were used for the colony count method. Then two tube tests containing these bacteria and 10 ml sterile physiological serum were adjusted with 0.5 McFarland turbidity standard. Each of these tube tests was diluted three times with sterile physiological serum. Then 0.1 mℓ of DMSO was added to four flasks of Nutrient Broth culture media. The obtained opacity solution was then divided into two portions for each bacterium and placed in two different flasks. Bionanocomposite (0.01 g) was added to one of the solutions of each bacterium in a flask, and the remaining flask has no bionanocomposite. Therefore two flasks have bacterium and bionanocomposite and two of them have no bionanocomposite, the flasks have just bacterium introduced as proof for comparison with flasks with bacterium and bionanocomposite. These four flasks were incubated at 37 °C for 2 h. Next, 0.1 mℓ of the content of each flask was added in Mueller Hinton agar, and the dishes were kept at 37 °C for 24 h. The antibacterial performance of bionanocomposite was investigated by counting the colonies on agar plates after 24 h. www.nature.com/scientificreports/ rial cultures with an approximate concentration of 10 6 -10 7 colony-forming units per milliliter (CFU/mL) were inoculated into a Nutrient Broth medium. To ensure optimum contact between chitosan-AgIO 3 bionanocomposite and bacterial cells, all experiments were performed in an incubator shaker at 37 °C and 180 rpm. The rate of bacterial killing was checked at various time intervals in terms of the UV-visible spectra. Two cultures of the bionanocomposite-free medium under the same growth bacteria conditions were used as controls. To avoid potential optical interference during optical measurements of the growing cultures caused by the light-scattering properties of the nanoparticles, the same liquid medium without microorganisms, but containing the same concentration of bionanocomposite cultured under the same conditions as blank controls.</p><!><p>One of the most critical and practical approaches in organic chemistry is the design and synthesis of nanocomposites with natural polymers derived from renewable resources and the investigation of their remarkable performance in medical fields. Among various areas in drug discovery design, synthesis and introduce novel antibacterial agents is extremely important. In this research, chitosan-AgIO 3 bionanocomposite was synthesized through an easy method with low cost and readily available materials. Briefly, upon the addition of AgNO 3 , the amino and hydroxyl groups on the structure of chitosan coordinated Ag + ions tightly to form a suspension of chitosan-Ag + . In the following with adding KIO 3 , Ag ions reacted to KIO 3 to generate AgIO 3 nanoparticles. Eventually, chitosan-AgIO 3 bionanocomposite was applied for antibacterial features against several bacteria. The structural corroboration of bionanocomposite is studied by utilizing analytical techniques including FT-IR spectra for detecting related functional groups, SEM for determination of morphology and structure, EDX analysis for elemental confirmation, and XRD pattern for study the crystal structure of bionanocomposite.</p><p>Investigation of chitosan-AgIO 3 bionanocomposite characteristics. Fourier transform infrared spectroscopy of chitosan-AgIO 3 bionanocomposite. According to Fig. 2, the fabrication of chitosan-AgIO 3 was determined by the FT-IR spectroscopy technique. As can be seen in Fig. 2a, the characteristic absorption bands for chitosan were shown at 3413 cm −1 as a strong peak for stretching vibrations for both O-H and N-H, overlapped, 2918 cm −1 is for C-H symmetric stretching vibrations, 2873 cm −1 is for C-H asymmetric stretching vibrations. The bands confirmed the presence of residual N-acetyl groups at around 1654 cm −1 for C=O stretching of amide and 1593 cm −1 for N-H bending vibrations. The peak at 1518 cm −1 is illustrated C-H bending vibrations and 1375 cm −1 is represented CH 3 symmetrical deformation. The band at 1315 cm −1 is indicated the C-N stretching vibration of amide, and 1259 cm −1 is related to the bending vibration of hydroxyl groups present in chitosan. The asymmetric stretching of the C-O-C bridge is determined by an absorption peak at 1153 cm −1 . The bands at 1072 and 1029 cm −1 correspond to the C-O stretching vibrations. Compared to the FT-IR spectrum of chitosan, the FT-IR spectrum of bionanocomposite demonstrates variation (Fig. 2b). The presence of absorption peaks at 710 cm −1 and 748 cm −1 correspond to AgIO 3 nanoparticles which represent the presence of AgIO 3 in the final bionanocomposite. On the basis of the data find from these two spectra, a reason for the shifting of wavenumbers and lower intensities of peaks is confirming interactions between Ag, O, and N atoms. In addition, C=O stretching band at 1654 cm −1 and N-H absorption band at 1593 cm −1 shifted to lower frequency owing to the bind between chitosan and AgIO 3 .</p><p>Energy dispersive X-ray spectroscopy of chitosan-AgIO 3 bionanocomposite. EDX analysis was performed to study the presence of elements in the structure of the bionanocomposite (Fig. 3a). EDS spectrum further represented C, O, N, Ag, and I elements in prepared bionanocomposite, and the atomic ratio between Ag and I was about 1:1, representing that chitosan-AgIO 3 bionanocomposite was successfully synthesized. In addition, the elemental mapping of EDX patterns shows the presence of C, O, N, Ag, and I elements in the bionanocomposite (Fig. 3b).</p><p>Scanning electron microscopy of chitosan-AgIO 3 bionanocomposite. To study the morphology and structure of chitosan-AgIO 3 bionanocomposite, the SEM analysis was done (Fig. 4). As seen in the figure, it is clear that www.nature.com/scientificreports/ AgIO 3 nanoparticles have covered the structure of chitosan. The presence and uniform distribution of AgIO 3 nanoparticles on the surface of chitosan is well observed. According to the SEM image, the average size of synthesizing AgIO 3 nanoparticles is less than 60 nm. To specify the size of nanoparticles, 70 particles were selected randomly. The average particle size of nanoparticles, is about 57 nm.</p><!><p>To investigate the structure of the inorganic nanoparticles to verify the formation of AgIO 3 nanoparticles in the bionanocomposite, the XRD pattern was prepared. As shown in Fig. 5, the bionanocomposite exhibited the main peaks, consistent with the characteristic peaks of AgIO 021), ( 041), ( 211), ( 230), ( 002), ( 231), ( 001), ( 232) and ( 271) crystal planes of the orthorhombic AgIO 3 , respectively 44 . Besides, the size of the</p><!><p>studies. The evaluation of the antibacterial performance of the synthesized chitosan-AgIO 3 bionanocomposite against Pseudomonas aeruginosa, Klebsiella pneumoniae, Staphylococcus saprophyticus, Escherichia coli, Staphylococcus aureus by in vitro study and agar diffusion method was done. Investigation of the inhibition zone in millimeters (mm) around the bionanocomposite (0.01 g) was measured to determine its antibacterial efficacy against these five types of microorganisms. The growth inhibition zone of bionanocomposite against all five bacteria was shown in Fig. 6. In addition, the details of the inhibition zone width of each bacteria are listed in Table 1. These results properly demonstrate the high performance of chitosan-AgIO 3 for killing various bacteria. In order to evaluate and expand the performance of the bionanocomposite against these five bacteria, the efficacy of the bionanocomposite was studied in comparison with penicillin and silver sulfadiazine as commercial antibiotics with different concentrations of the penicillin and silver sulfadiazine and constant amount of bionanocomposite (0.01 g). The inhibition zone in millimeters (mm) around the bionanocomposite (0.01 g) and penicillin (0.001 g/ml) (Fig. 7) and silver sulfadiazine (0.001 g/ml) (Fig. 8) were summarized (see Tables S1 and S2 in Supplementary Information). Inhibition zone measurements around the bionanocomposite (0.01 g) and penicillin (0.01 g/ml) and silver sulfadiazine (0.01 g/ml) were further conducted (see Figs. S1 and S2 and Tables S3 and S4 in Supplementary Information) to determine the influence of concentration in antibacterial activity. These findings represent that chitosan-AgIO 3 has a high potential for eradication of various microorganisms. Moreover, the antibacterial activity of the chitosan-AgIO 3 bionanocomposite was compared with the relevant antibacterial materials reported in the literature. Data about the ZOI (mm) toward the target bacteria were summarized in Table 2.</p><p>Plate-count method. The colony plate images of Staphylococcus aureus (ATCC 12600) and Escherichia coli (ATCC 9637) bacteria in the presence of chitosan-AgIO 3 bionanocomposite (0.01 g) are represented in Fig. 9.</p><p>As illustrated in the figure, all colonies of Staphylococcus aureus and Escherichia coli were killed by treatment with the chitosan-AgIO 3 bionanocomposite.</p><p>Optical density. The OD measurements of bacterial cultures were investigated in the presence of 0.01 g chitosan-AgIO 3 bionanocomposite, 0.5 McFarland turbidity standard, and Nutrient Broth media. Growing cultures were checked at various times, including 3, 6, and 18 h (Fig. 10). As can be seen in a bar diagram in Fig. 10a, www.nature.com/scientificreports/ the antibacterial property of bionanocomposite demonstrate a considerable inhibition of bacterial growth for Escherichia coli and Staphylococcus aureus. As represented in a chart in Fig. 10b, for Escherichia coli after 3 h, 71.96% of bacteria content decreased with the presence of bionanocomposite, and this growth inhibition was 84.37% after 6 h. After 18 h, the reduction in bacteria content is 85.1%. For Staphylococcus aureus after 3 h, 64.7% of bacteria content decreased in the presence of bionanocomposite. After 6 h, this reduction in bacteria content was about 71.22%, and at least after 18 h, 75.69% of bacteria content was reduced. Based on the present findings of this research, the chitosan-AgIO 3 bionanocomposite introduces as an antibacterial agent for killing and inhibiting bacterial growth in terms of the UV absorption spectra.</p><p>Mechanism for antibacterial activity of bionanocomposite. The mechanism of the antimicrobial efficacy of chitosan-AgIO 3 bionanocomposite was further studied on Escherichia coli by flow cytometry. It is reputable that ROS is a fundamental factor in the antibacterial activities of nanomaterials, which could directly damage the cell membrane, phospholipids, and/or membrane proteins. Reactive oxygen species as free radicals cause damage to a variety of pathogens 51,52 . Nanomaterials with ROS-producing abilities could be a beneficial strategy to fight against bacteria. To explore the antibacterial mechanism of bionanocomposite, the production of ROS was measured by DCFDA, a commercial fluorescent probe. As can be seen in Fig. 11, the fluorescence intensity in the diagram of bionanocomposite increased, and the stronger the fluorescence intensity, the greater the number of dead cells which means a good antibacterial activity of bionanocomposite. The fluorescence intensity of bionanocomposite was stronger than the control group and the MFI parameter of bionanocomposite is three times more than the control group which means the dead cells in bionanocomposite are highly increased. As a consequence, the bacterium was damaged due to the membrane damage caused by ROS production, thereby achieving antibacterial activity (Fig. 12).</p><!><p>Due to the increase in microbial resistance to existing antimicrobial agents, which unfortunately increases the percentage of disease and mortality, the introduction and synthesis of new antimicrobial agents are essential.</p><p>To introduce new antimicrobial agents in this study, chitosan-AgIO 3 nanocomposite was successfully designed and synthesized for the first time with a simple method and with utilizing readily available materials. Chitosan is a natural polymer and was used as the base of the bionanocomposite and AgIO 3 nanoparticles were immobilized on the structure of chitosan. The presence of chitosan in the bionanocomposite cause biodegradable and environmentally friendly nanocomposite. The structural features of bionanocomposite were studied with several techniques including FT-IR, EDX, SEM, and XRD analyses, and the results verified the effective immobilization of AgIO 3 on chitosan. The antibacterial efficacy was evaluated through the agar diffusion strategy, plate count method, and optical density study with different microorganisms. Besides the antibacterial efficacy of the bionanocomposite was compared with two commercial drugs and in some cases, the synthesized bionanocomposite has better performance against bacteria in comparison with common drugs. In a view of the difficult eradication of Pseudomonas aeruginosa due to its considerable capacity to resist antibiotics, presenting a new antibacterial agent which has a good operation against this bacteria is very important. In addition, chitosan-AgIO 3 bionanocomposite as an environment-friendly and efficient antibacterial agent is highly recommended in a water purification process, and biomedical applications due to its specific properties such as high antibacterial activity, low-cost, simple, and green synthesis method. The findings represented that this work could open a way for www.nature.com/scientificreports/ producing a novel low-cost, biodegradable bionanocomposite with high performance against different bacteria. Due to the effectiveness of the synthesized bionanocomposite compared with antibiotics, it could be introduced as a good suggestion in therapeutic applications as well as burn wounds. The main impediment of this work is the separation of the bionanocomposite which is designed in our future research as well as other applications of chitosan-AgIO 3 bionanocomposite.</p>

Scientific Reports - Nature

Site-Specific Mapping of Sialic Acid Linkage Isomers by Ion Mobility Spectrometry

Detailed structural elucidation of protein glycosylation is a tedious process often involving several techniques. Glycomics and glycoproteomics approaches with mass spectrometry offer a rapid platform for glycan profiling but are limited by the inability to resolve isobaric species such as linkage and positional isomers. Recently, ion mobility spectrometry (IMS) has been shown to effectively resolve isobaric oligosaccharides, but the utility of IMS to obtain glycan structural information on a site-specific level with proteomic analyses has yet to be investigated. Here, we report that the addition of IMS to conventional glycoproteomics platforms adds additional information regarding glycan structure and is particularly useful for differentiation of sialic acid linkage isomers on both N- and O-linked glycopeptides. With further development IMS may hold the potential for rapid and complete structural elucidation of glycan chains at a site-specific level.

site-specific_mapping_of_sialic_acid_linkage_isomers_by_ion_mobility_spectrometry

<!>Results and Discussion<!>Conclusion<!>Preparation of Glycopeptides<!>LC\xe2\x80\x93IMS-MS<!>Data Analysis

<p>Glycosylation plays a role in nearly all aspects of biology including cell–cell communication, differentiation and development, modulating enzymatic activity, and immune regulation and is integral to many host–pathogen interactions.1,2 Glycan chains are attached to proteins either through the side chains of asparagine (N-linked) or serine and threonine residues (O-linked). All glycan chains are composed of relatively few types of monosaccharides, but the large number of possible branching patterns, linkage isomers, and other modifications (e.g., sulfation) results in an enormous diversity of structural variants with different biological effects. For example, sialic acids moieties can be linked α2-3 or α2-6 to the outer sugars of the nonreducing end, which alter host recognition of the influenza viruses.3 Despite their importance to clinical diagnostics and therapeutic development, our knowledge of glycosylation has been hindered by the structural complexity of carbohydrates, which create difficult problems for data analysis.4</p><p>Unlike DNA or proteins, for which technologies have enabled rapid sequencing, glycan sequencing remains a very involved and specialized task, sometimes requiring several orthogonal methods. Strategies for full glycan structural elucidation generally involve glycan removal either enzymatically (N-glycanase) or chemically (e.g., hydrazine) and either liquid chromatography or capillary electrophoresis to resolve each glycan structure.4–7 In many cases complex glycans need to be further analyzed with stepwise treatment using specific exoglycosidases for full structural elucidation. While such methods have been established for rapid parallel processing, one major limitation is the reliance on glycan removal from the protein. With proteins containing multiple glycosylation sites, this approach does not resolve differences among the glycan structures at the individual sites. Site-specific glycan characterization typically necessitates additional prefractionation to isolate each glycosylation site on a glycopeptide level. This additional step can be labor intensive and generally requires significantly more starting material.</p><p>Modern glycoproteomics with mass spectrometry offer a rapid platform for site-specific glycan profiling, but the information content is limited by the inability to resolve linkage- and positional isomers.8 Additional information on the glycan structures requires specific exoglycosidase treatments or derivatization schemes in conjunction with mass spectrometry.9–14 Ion mobility spectrometry (IMS) adds a dimension capable of resolving isobaric species by their gas-phase collision cross-section (CCS), which has proven useful at resolving oligosaccharides15–28 and databases for carbohydrate CCS values have recently been established.29 However, all of these studies have been limited to synthetic standards or glycan chains isolated from glycoproteins. Additionally, because of gas-phase rearrangements within certain types of sugars during typical collision-induced dissociation (CID), mass spectrometry based glycan structural characterization has largely focused on metal adducted glycan chains.30–32 Here we demonstrate how incorporation of IMS into glycoproteomics platforms can provide additional information regarding glycan structure on a site-specific level, particularly in regard to differentiation of sialylation linkages on N-linked and O-linked glycopeptides.</p><!><p>As a first assessment of the potential of IMS to resolve sialic acid isomers the arrival time distributions (ATD) of 3′ and 6′ sialyl-N-acetyllactosamine (SLN) were compared (Figure 1A). The two trisaccharides have identical structures with the exception of the linkage of the sialic acid (NeuAc) to the galactose (Gal), which is either linked α2-3 or α2-6. The ATD of the [M − H2O + H]+ species (657.24 Da), one of the abundant fragments that is observed during typical CID of an intact glycan chain, showed remarkable separation of the two linkage isomers. While 6′ SLN had a clear Gaussian peak at ∼6 ms, 3′ SLN showed a major peak at ∼6.4 ms with a minor peak near 6 ms. The fragment corresponding to the NeuAc-Gal species (m/z 454.17) was also examined, but was very poorly detected, presumably due to poor fragmentation efficiency at the Galβ1-4GlcNAc glycosidic bond and therefore was not used for further analysis.</p><p>To test whether the additional peak in the 3′ SLN was due to some 6′ impurity, both isomers were analyzed by hydrophilic interaction chromatography (HILIC), which was able to resolve the two anomers of both 3′ and ′6 species (Figure S1A). From the ATDs, only one of the two anomers of the 3′ SLN shows a strong signal near 6 ms (Figure S1B). The presence of the second peak was not dependent on the collision energy used (Figure S2A). Structural rearrangements in the gas-phase have been reported to occur on several protonated glycan species even under mild CID conditions, albeit at relatively low levels.31–33 However, the CID spectra of both 3′ and 6′ SLN isomers showed only a trace peak (less than 0.1% base peak intensity) corresponding to a NeuAc-GlcNAc species (495.17 Da), that would arise from such rearrangements. The sodium adduct, which should not undergo any gas-phase rearrangements,32 also showed a second peak only in the 3′ SLN (Figure S2B), and its fragmentation pattern showed no evidence of any linkage rearrangements. A similar pattern was seen with the intact protonated ion, though for this case even the 6′ SLN showed two peaks, likely relating to the different anomers (Figure S2B). At present, we speculate that the minor peak in the 3′ SLN ATD is due to an alternate gas-phase conformation that is dependent on the anomeric configuration of the reducing sugar.</p><p>The same IMS separation was applied to resolve sialic acid linkage isomers on intact glycopeptides from a tryptic digest of human alpha-1-acid glycoprotein (α1AGP). Intact glycopeptides were resolved by LC, their precursor ions isolated in the quadrupole, fragmented by CID, and separated by IMS to resolve the glycan fragment isomers (LC–CID-IM-MS). Figure 1B shows the ATDs of the SLN fragment from one selected α1AGP glycopeptide with a biantennary, triantennary, or tetraantennary N-linked glycan chain. There is a clear signal for both the 3′ and 6′ isomers that matches the major peaks observed with the SLN standards. From the known structures of the N-linked glycans of α1AGP, the two peaks should correspond to NeuAcα2-3Galβ1-4GlcNAc and NeuAcα2-6Galβ1-4GlcNAc.34 To confirm these assignments, the tryptic digests were treated with Sialidase S to specifically cleave only α2-3 linked sialic acids. Accordingly, the ATD of the Sialidase S treated sample showed only the single peak at ∼6 ms (Figure 1B, bottom trace).</p><p>LC–CID-IM-MS was similarly performed on a tryptic digest of bovine Fetuin. The two peaks in the ATD for the SLN fragments matched those from α1AGP and the SLN standards (Figure 2A). An additional peak at ∼5.7 ms was evident when analyzing the fragments from the triantennary tetrasialo glycoform, which has been shown to contain a GlcNAc-linked sialic acid.35,36 After Sialidase S treatment the ∼6.5 ms peak disappeared, leaving only the 5.7 and 6 ms peaks. The tryptic digest was further treated with Sialidase A to remove all sialic acids. However, a signal for the triantennary glycoform bearing a single sialic acid remained, which has previously been observed to be resistant to sialidase treatment.35 The ATD of the SLN fragment from this glycoform showed a predominant peak at ∼5.7 ms, which therefore corresponds to the Galβ1-3(NeuAcα2-6)GlcNAc structure present on a fraction of triantennary N-linked glycans on Fetuin. Among the fragments from the tetrasialo glycoforms, there was also an abundant signal at m/z 495.2 and 948.34 that had a single peak in the ATD consistent with the fragments NeuAcα2-6GlcNAc and NeuAcα2-3Galβ1-3(NeuAcα2-6)GlcNAc, respectively (Figure 2B). These ions were not observed in the triantennary trisialo or bianternnary Fetuin glycopeptides or any of the α1AGP glycopeptides. Therefore, there is no detectable occurrence of gas-phase rearrangements between the sialic acid and the GlcNAc during CID of intact glycopeptides. Calibrated CCS values for all observed sialylated glycan fragments are reported in Table 1.</p><p>In the course of examining the various glycoforms of each glycopeptide, we observed a trend in which the higher degree of branching correlated with an increase in the content of α2-3 linked sialylation. Such correlations have been previously observed during comprehensive NMR structural elucidation of Fetuin N-linked glycans.35 However, with the ability to resolve variations in sialylation patterns on a site-specific level we can see that not all sites share identical patterns (Figure S3). For example, the biantennary bisialo glycopeptide 142-169 shows 41.3% α2-3 while glycopeptide 54-85 with the same glycoform contains only 23.2% (Table S1). These types of differences are also observed among the different biantennary glycopeptides in α1AGP, with the content of α2-3 linked sialylation ranging from 4.3 to 23.6%. As a complementary way to monitor relative sialylation levels, we examined the signal intensity of each glycoform before and after sialidase S treatment as a qualitative assessment of the degree of α2-3 sialylation. Overall, the trends are consistent with the IMS results in that biantennary glycans have the highest content of 6′ sialylation and tetraantennary glycans have the most 3′ (Figure S4). While several studies have examined the distributions of Fetuin and α1AGP glycoforms along with relative amounts of sialylation linkages,35,37 to our knowledge this is the first example of such a quantitative examination of sialylation at a site-specific level.</p><p>The Fetuin digest also provided a sample to test whether IMS can be used to obtain structural information for glycopeptides bearing O-linked glycans. We examined peptides 316-330 bearing a single glycosylation site, and 228-288 which bears 4 well-characterized glycosylation sites.38 The most abundant glycoforms with resolved precursor ions were examined by IMS. Though the overall signal was relatively poor the ATDs of the sialyl-N-acetyllactosamine fragments could still be monitored (Figure 2C). With both peptides there were three species with the predominant peak at 6 ms and minor peaks at 6.3 and 5.6 ms. The most prevalent O-linked glycan structure on Fetuin is α2-3 sialylated with galactose linked β1-3 to a N-acetylgalactosamine (NeuAcα2-3Galβ1-3GalNAc).39 Two other glycoforms have α2-6 sialylation on the GalNAc: NeuAcα2-3Galβ1-3(NeuAcα2-6)GalNAc and Galβ1-3(NeuAcα2-6)GalNAc); the later of which might correspond to the peak at 5.6 ms, as seen with the Galβ1-3(NeuAcα2-6)GlcNAc) on the Fetuin N-linked glycopeptides. Additionally there is also a minor glycoform containing a second N-acetyllactosamine chain attached to the GalNAc (NeuAcα2-3Galβ1-3(NeuAcα2-3Galβ1-4GlcNAcβ1-6)-GalNAc).40 This hexasaccharide can fragment to yield NeuAcα2-3Galβ1-4GlcNAc that would be consistent with the peak at 6.3 ms, as observed with the N-linked glycopeptides. Further work will be necessary for definitively establishing the drift times of structures relevant to O-linked glycosylation, but for now this data demonstrates the utility of IMS for resolving different structures of O-linked glycans.</p><p>To assess the overall quantitative accuracy of sialylation by IMS an additional LC–MS run was performed with a high cone voltage to induce fragmentation of all precursor ions and observe the combined ATD from all glycopeptides (Figure 3). This signal should be dominated by the N-linked glycans, as the Fetuin O-linked glycopeptide precursor ions were of relatively weak intensity. From the combined ATD the overall ratios of sialic acid linkage isomers observed on Fetuin is approximately 38.4% α2-3, 59.4% α2-6, and 2.2% attributed to sialylation at the GlcNAc. These overall ratios are consistent with previous quantitative measurements of sialylation of Fetuin that range from 38:62 up to 51:49 in α2-3:α2-6 sialylation, possibly reflecting differences in the methods used for glycan isolation, detection, and variations in sample preparations.35,37 Additionally, the overall content of sialylation at GlcNAc is similar to that observed by nuclear magnetic resonance (∼2.6%).35</p><p>As an additional control we examined the sialylation pattern of a HIV-1 Env gp140 glycoprotein purified from Chinese Hamster Ovary (CHO) cells. gp140 trimers contain 28 N-linked glycosylation sites on each of the three subunits, some of which bear biantennary monosialylated glycan structures.41 Expression in CHO cells produces proteins with only α2-3 linked sialylation as CHO cells lack α2-6 sialyltransferase.42 A tryptic digest of gp140 was analyzed using the same high cone voltage approach as described for α1AGP and Fetuin. The resulting combined ATD showed only a single peak at 6.3 ms (Figure 3, lower panel). Examining the individual patterns for each glycopeptide in such a complex glycoprotein was beyond the scope of this study. However, this data sufficiently demonstrates that the underlying cause for the multiple peaks observed in the ATD of the 3′ SLN standard (Figure 1A) is not confounding the analysis of sialyl-N-acetyllactosamine fragments from intact glycopeptides.</p><p>Lastly we examined the effect of the collision energy (CE) on the ATD profile of the glycan fragments. The m/z 657.24 peak was observed even with the lowest CE (4 V), became more abundant as the voltage was increased, and eventually diminished above 40 V, likely due to further fragmentation of the SLN fragment (Figure S5A). Even at the highest CE the ATD for m/z 657.24 fit well to two Gaussian distributions, unlike that observed in the 3′ SLN standard (Figures S1B and S5C). The relative intensities of the 3′ and 6′ SLN peaks are consistent at lower CE, but deviate above 20 V, with a drop in the observed 3′ SLN content (Figure S5B). The offsets at higher CE may be due to further fragmentation if one of the sialylation isomers is more susceptible to secondary fragmentation. Overall, this indicates that a lower CE is favorable for a more accurate quantitative measure of the sialylation linkage by IMS.</p><!><p>A number of studies have demonstrated the power of IMS to resolve linkage and positional isomers in synthetic oligosaccharides and differentiate intact glycan chains. However, the utility of IMS to obtain glycan information on a site-specific level during glycoproteomics analyses is only now becoming appreciated. Here, we report that the addition of IMS to conventional glycoproteomics platforms adds the ability to differentiate sialic acid linkage isomers on both N- and O-linked glycopeptides. During the revision stage of this article, another group reported a very similar strategy for differentiating sialylation linkage by IMS,43 further illustrating the utility of this approach. With further optimization of separation conditions for closely related glycans present on biologically relevant glycan structures, IMS has the potential to extract a higher level of information from glycoproteomics. Such a tool can elucidate fine glycan structural information on a site-specific level with minimal sample requirements and without additional sample preparation steps beyond current proteomic protocols.</p><!><p>In total, 4 mg of bovine Fetuin and human α1AGP (Sigma-Aldrich, St. Louis, MO) were resuspended in 100 μL of 8 M urea, 100 mM Tris pH 8.0, 50 mM dithiothreitol (DTT) and heated at 85 °C for 30 min. Cysteines were alkylated by the addition of 100 mM iodoacetamide (IAM) and incubation for 1 h in the dark, followed by addition of 50 mM DTT to quench remaining IAM. The samples were diluted 16-fold in 20 mM Tris pH 8.0 and proteins were digested with TPCK-treated trypsin (Sigma-Aldrich) at a 1:50 ratio of trypsin–substrate overnight at 37 °C and subsequently quenched with 1 mM PMSF. A volume of 100 μL of the digest was treated with 10 mU Sialidase A (Arthrobacter ureafaciens, Prozyme, Hayward, CA) at 37 °C for 4 h. A second 100 μL portion was acidified to pH 6.0 with sodium acetate and treated with 10 mU Sialidase S (Streptococcus pneumoniae, Prozyme) at 37 °C for 4 h.</p><p>HIV-1 Env gp140 was expressed in CHO cells and purified as described previously.44 A volume of 50 μL of Env gp140 (0.4 mg/mL in PBS pH 7.4) was reduced with 20 mM DTT at 85 °C for 20 min, acidified to pH 2.5 with 0.5% formic acid, and digested with 3 μg of pepsin (Worthington Biochemicals, MA) at 37 °C for 15 min. Samples were neutralized with 10 μL 1 M sodium acetate for a final pH of 5.0 to stop digestion.</p><!><p>Tryptic digests (2 μg per injection) of glycoproteins were resolved with a Waters Aquity HPLC over a 1 mm × 100 mm 1.7 μm BEH C18 column (Waters, Milford, MA) with a linear gradient of 3% to 37% B over 18 min at a flow rate of 70 μL/min (A, 0.1% formic acid; B, 0.1% formic acid, 100% acetonitrile). Eluting peptides were analyzed online by electrospray with a Synapt G2-Si mass spectrometer (Waters). Source and desolvation temperatures were 100 and 250 °C, and the capillary and cone voltages were set to 2.5 kV and 40 V. Initial runs were collected in a data-dependent fashion to select ions with a quadrupole resolution of 7 and perform MS/MS on precursor ions from 800 up to 2000 Da with a charge state of at least 2. Glycopeptides were identified by using fragmentation data and exact mass to assess the glycan composition and the peptide sequence.45</p><p>Targeted glycopeptide runs were subsequently performed by running identical injections in a data-independent manner with fragmentation in the trap prior to the IMS cell (premobility MS/MS). Glycopeptides were selected by retention times and narrow m/z windows (2 Da) for quadrupole selections. Two second acquisitions for each MS/MS were collected with the collisions energy ramped from 30 to 50 V. The IMS cell pressure was ∼2.85 mbar with a N2 gas flow of 90 mL/min and the traveling wave height and velocity were set at 40 V and 650 m/s, respectively. Six additional injections and LC runs were performed with a fixed trap CE value of 0, 5, 10, 20, 35, and 50 V. For untargeted IMS analysis of glycan fragments from all glycopeptides, an identical LC–IMS-MS run was performed with a cone voltage of 120 V. For CCS calibration, IMS-MS was collected for 5 minutes before and after the series of LC runs using a 0.01 mg/mL polyalanine ladder (Waters) in 0.1% formic acid infused at 40 μL/min with identical IMS and MS settings.</p><p>3′ and 6′ sialyl-N-acetyllactosamines (SLNs) were purchased from Prozyme (Hayward, CA). SLNs were resuspended in 0.1% formic acid for a final concentration of 10 μM and infused at 40 μL/min into a Waters Synapt GS-Si mass spectrometer with settings as described above. For HILIC separation, 4 μg of either 3′ or 6′ SLN was injected over a GlycoSep 1 mm × 50 mm XBridge Glycan BEH amide column (Waters) in 90% acetonitrile. A linear gradient of 95 to 65% B over 20 min was used to elute SLNs at 150 μL/min (A, 0.1% formic acid, B, 0.1% formic acid, 100% acetonitrile). The flow was coupled to a Synapt G2-Si, and spectra were collected over a mass range of 300 to 1200 every 2 s with ion mobility engaged as described above.</p><!><p>Data were analyzed in MassLynx 4.1 and DriftScope v2.8 (Waters). ATDs for sialylated glycan fragments (see Table 1) with a mass window of 0.05 Da were exported and copied into Excel (Microsoft) for fitting Gaussian distributions with custom macros. The corrected drift times of the 1+ ions for Ala3 up to Ala14 relative to their known collision cross section (CCS) values was used to calculate CCS values for all glycan fragments as described previously.46,47</p>

PubMed Author Manuscript

Silver nanoparticle-enhanced fluorescence in microtransponder-based immuno- and DNA hybridization assays

The aim of this study is to improve assay sensitivity in common solid-phase bioassay configurations as the result of using silver nanoparticles. The solid phase was provided by numerically indexed, silicon-based electronic chips, microtransponders (p-Chips) that have previously been used in multiplexed assays. Assay configurations investigated included an ELISA-type immunoassay and a DNA hybridization assay. The surface of p-Chips was derivatized with the silver island film (SIF) and a polymer, and then characterized with AFM and SEM. Silver nanoparticle sizes were in the range of 100 to 200 nm. Four fluorophores were tested for fluorescence enhancement; namely, green fluorescent protein, phycoerythrin, Cy3 and Alexa Fluor 555. We consistently observed significant fluorescence enhancement and sensitivity improvement in the p-Chip-based assays: the sensitivity in the cytokine IL-6 immunoassay was 4.3 pg/ml, which represented a 25-fold increase over the method not involving a SIF; and 50 pM in the hybridization assay, a 38-fold increase. The greatest enhancement was obtained for p-Chip surfaces derivatized first with the polymer and then coated with SIF. In conclusion, we show that the SIF-p-Chip-based platform is a highly sensitive method to quantify low-abundance biomolecules in nucleic acid-based assays and immunoassays.

silver_nanoparticle-enhanced_fluorescence_in_microtransponder-based_immuno-_and_dna_hybridization_as

Introduction<!>Fluorescent dyes and reagents<!>Coating p-Chips with polymer<!>Conversion of amino group to carboxylic acid<!>Deposition of SIF on p-Chips<!>Atomic force microscopy<!>Scanning electron microscopy<!>Measurement of fluorescence lifetimes<!>Conjugation of antibody to p-Chips<!>Immunoassay<!>DNA assay<!>Deposition of silver nanoparticles on p-Chips: surface properties<!>Fluorescence enhancement by SIF and lifetime measurements for various fluorophores<!>Immunoassay<!>DNA hybridization assay<!>Conclusions<!>

<p>Fluorescent labeling techniques are widely applied in nucleic acid and protein assays [1, 2] as they are easy to use, adaptable to different analytes, highly sensitive, and can be part of multiple labeling strategies. Given their widespread use, there is a desire to further improve the limit of detection in order to determine low-abundance biomolecules. In this paper, we report on a silver island film (SIF)-based approach that greatly improves the sensitivity of both a fluorescence-based immunoassay and a fluorescence-based DNA hybridization assay.</p><p>The effects of silver nanostructures on fluorescence enhancement have been theoretically and empirically studied during the past two decades [3–7]. A significant enhancement of fluorescence signal has been shown by many groups when fluorophores are placed in close proximity to a layer of metallic nanoparticles [8–12]. The enhancement is due to the interaction of excited-state fluorophores with mobile electrons on the surface of the metallic nanoparticles that results in an increase in the rate of radiative decay [6, 13], shorter fluorescence lifetimes and faster turnover of excited states [3, 4, 8, 9]. This phenomenon is known as "surface plasmon resonance". There have been many studies of potential bioanalytical applications [11, 14]. More recently, it was demonstrated that silicon wafers are a suitable solid phase for observing metal-enhanced fluorescence [15]. Thus, it was anticipated that the use of fluorescence with SIF on silicon chips would improve the assay sensitivity and signal-to-noise ratio.</p><p>In this study, we applied SIF to a microtransponder p-Chip-based platform in a DNA hybridization assay and a cytokine IL-6 immunoassay. This platform has been developed and is used in both genomics and proteomics research [16–18]. The key feature of the platform is its applicability to multiplex assays, in which concentrations of many analytes are determined simultaneously. In the recently described cystic fibrosis (CF) DNA assay [19], p-Chips were used to determine 50 mutations related to CF. Up to 300 p-Chips were used in a single assay for one DNA sample.</p><p>The p-Chip is a monolithic, integrated semiconductor device with standard dimensions of 500 μm×500 μm×100 μm. The electronic side of the p-Chip is composed of photocells, read-only memory (ROM), control electronics, and an antenna loop. The photocells, when illuminated, provide power for the logic circuitry. The circuitry accesses the ROM contents and modulates current through the antenna. The antenna transmits the ID digitally through a varying magnetic field in the vicinity of the chip that can be decoded by the radio frequency readout system to provide the specific serial number. The current p-Chip design allows 210 (1,024) unique IDs to be encoded; however, the ROM contains additional 48 unused bits and the encoding scheme can easily be expanded for up to 230 (~109) unique IDs. In bioassay applications, p-Chips are coated with a polymer layer containing both hydroxyl and amino groups that allow conjugation of oligonucleotide or protein probes to the surface. Probe identity is thus associated with a unique ID for each p-Chip. The p-Chip-based platform is designed to increase the flexibility and throughput of bioassays and has been successfully applied in multiplexed genotyping [19], proteomic and cytotoxicity assays [20].</p><p>There are two advantages associated with using p-Chips in a bioassay: first, p-Chips are individually identifiable, thus traceability of p-Chips throughout manufacturing is maintained and the manufacturing process itself is easier to control. Secondly, it is easy to modify the multiplex panel configurations by adding p-Chips derivatized with additional oligonucleotides or antibody probes (this task might be much more difficult in alternative approaches, e.g., microarrays).</p><p>The goal of this study was to demonstrate that silver nanoparticles can increase the sensitivity of p-Chip-based bioassays by enhancing the fluorescence signal. Towards this goal, we developed a method to coat p-Chips with silver nanoparticles. Coating small chips or beads requires a different approach than coating glass slides and involves many challenges including: uniform adherence to the chip surface (a silicon or silicon dioxide surface) under conditions during which the chips are suspended in a solution and are being agitated; controlling the size of silver nanoparticle being deposited; and assuring of the thickness and uniformity of the polymer and silver film layers on the chip; drying of the chips. It should be noted that spin coating techniques cannot be used with these applications. The two coating methods available to us: immersing the silver nanoparticles in the polymer coating of the p-Chips, or forming a separate silver nanoparticle layer on surface of the polymer layer, were tested, and physical measurements of the surface properties using atomic force microscopy (AFM) and scanning electron microscopy (SEM), were performed.</p><p>In this study, fluorescence enhancement for four commonly used fluorophores was investigated, namely green fluorescent protein (GFP), Alexa Fluor 555 (AF555), phycoerythrin (PE) and Cy3. In addition, to demonstrate potential applications for diagnostics in vitro, two types of biological assays were performed on p-Chips coated with silver nanoparticles: an interleukin IL-6 enzyme-linked immunosorbent assay (ELISA); and a DNA hybridization assay in which a fluorescently labeled DNA probe was allowed to bind to a DNA target synthesized on the p-Chip.</p><!><p>GFP was purchased from BioVision. AF555 labeled bovine serum albumin (BSA-AF555) and streptavidin-R-phycoerythrin (SA-PE) were purchased from Invitrogen. Cy3-labeled oligonucleotides were synthesized by MWG Biotech. All of the chemicals used for polymer coating and silver deposition were purchased from Sigma.</p><!><p>p-Chips were pretreated with 99.5% methyl alcohol at room temperature (RT) for 10 min, and repeated three times. The p-Chips were then rinsed with 0.01% distilled water and 0.9% aminopropyltriethoxysilane (APTS) in dry toluene/dimethylformamide (DMF) mixture at RT, and repeated four times. After rinsing, p-Chips were immediately treated with a coating solution (mixture of 0.01% distilled water, 0.9% APTS, and 0.3% 3-glycidoxypropyltrimethoxysilane (GPTS) in dry toluene and DMF) at 80 °C for 45 min and repeated once. After the coating reaction, p-Chips were washed once with toluene, three times with DMF, and three times with acetonitrile at RT, followed by air drying. The procedure placed both amino and hydroxy groups on the surface of p-Chips.</p><!><p>Amino-derivatized p-Chips were treated with 10% succinic anhydride in dry pyridine:DMF (1:9) on a tissue culture rotator at RT for 30 min. This step was repeated once using fresh reagents. After the reaction, the carboxylated p-Chips were washed with DMF four times and acetonitrile twice, followed by air drying.</p><!><p>SIF was deposited on the surface of carboxylated p-Chips as reported previously [14] with several modifications. Two drops of 5% NaOH was progressively added to 6 ml of 0.83% AgNO3 solution with intensive stirring at RT in a 15-ml reaction tube. A 0.2-ml of 30% NH4OH was subsequently added with intensive stirring at RT. The clear solution was incubated in an ice bath for 10 min, followed by the addition of 1.5 ml of a 4.8% fresh glucose solution with intensive stirring. Carboxylated p-Chips were incubated in this solution in an ice bath for 2 min, and then on a tissue culture rotator at RT for 20 min. After the silver deposition, p-Chips were immediately washed with distilled water three times followed by air drying.</p><p>To test the effectiveness of the SIF-polymer layer on p-Chip in fluorescence enhancement, the p-Chips were incubated with 50 μg/ml GFP for 1 h at RT and then washed with distilled water three times followed by air drying. The GFP fluorescence enhancement was measured using a fluorescent microscope (Nikon Eclipse E600 with Y-FL EPI fluorescence attachment and IP Lab v. 3.55 software [Scanalytics, Inc.]).</p><!><p>AFM images were collected by scanning the surface of a p-Chip on a TMX2100 Explorer SPM atomic force microscope (Veco). This instrument is equipped with a dry scanner calibrated using a standard calibration grid as well as 100 nm gold nanoparticles (Ted Pella). Images were analyzed using SPMLAB software.</p><!><p>Joel JSM-6510LV scanning electron microscope (Joel Ltd., Japan) was used for fast characterization and imaging of silver structures deposited on the p-Chips. All images were taken at high vacuum at the potential of 20 kV applied to the tungsten filament. Magnification was ×15,000 with the beam width 20 mm.</p><!><p>Fluorescence decay of AF555 deposited on p-Chips was measured on a FluoTime 200 fluorometer (PicoQuant) using excitation from a pulsed picosecond 475 nm solid state laser. The instrument was equipped with a microchannel plate photomultiplier ultrafast detector, a monochromator and a polarizer in the detection path. Two 550 nm long wave pass filters were used on the emission optics for observation at 600 nm to eliminate scattered excitation light. For measurements, p-Chip were placed between two cover slips and mounted in a front face attachment. Fluorescent lifetime data were analyzed with a FluoFit version 4 software (Picoquant) and fitted to a multi-exponential model.</p><!><p>In order to protect SIF from scratches, a thin layer of polymer (20 min polymer coating, see above) was deposited onto SIF-p-Chip and carboxylated (see the "Conversion of amino group to carboxylic acid" section above) prior to performing the immunoassays. The antibody was conjugated to p-Chips by incubating the carboxylated p-Chips with 50 mg/ml 1-ethyl-3-(3-dimethy-laminopropyl)-carbodiimide and 50 mg/ml N-hydroxysuccinimide in 0.1 M HEPES buffer (pH 7.5) on a rotator for 30 min at RT. The p-Chips were then washed with 60 μl PBS three times and incubated with 200 μg/ml monoclonal anti-human IL-6 antibody (R&D Systems) for 30 min at RT on a tissue culture rotator. The p-Chips were washed with PBS three times and blocked with SuperBlock solution (Thermo Scientific) for 5 min at RT on a rotator. This blocking step was repeated twice. The p-Chips were then washed with PBS three times (2 min each) and stored in PBS with 1% BSA at 4 °C.</p><!><p>Anti-IL-6-conjugated p-Chips were incubated with 50 μl recombinant human IL-6 standard (R&D Systems) in PBS with 1% BSA or 50 μl diluted normal human serum (United States Biological) spiked with recombinant human IL-6 for 1.5 h at RT on a rotator. After incubation, the p-Chips were washed with Tris-Buffered Saline Tween-20 (TBST) three times.</p><p>The detection antibody solution was prepared by diluting biotinylated anti-human IL-6 antibody (R&D System) to 5.0 μg/ml with 1% BSA in PBS. The p-Chips were then incubated with 50 μl of detection antibody for 1 h, followed by washing with TBST three times. The p-Chips were pooled and incubated with 50 μl of 8 μg/ml SA-PE in PBS for 15 min at RT in the dark. After incubation, the p-Chips were washed with TBST three times and distilled water twice, and air dried. PE fluorescence was confirmed using a fluorescent microscope (Nikon Eclipse E600 with Y-FL EPI fluorescence attachment) and quantified with Image J software (NIH).</p><p>The assay sensitivity is defined as the minimum IL-6 concentration producing a signal equal to three deviations (SD) from the standard zero. This represents the lowest value read from the standard curve that can be statistically distinguished from zero. To determine the assay sensitivity, a standard curve was generated using Prism 5.0 software: three standard SDs for the standard zero were added to the mean fluorescence for the standard zero replicates (six replicates for non-SIF and nine replicates for SIF) and the corresponding concentration was determined from the standard curve.</p><!><p>Carboxylated SIF or non-SIF-p-Chips were conjugated with 50 μg/ml avidin (see the "Immunoassay" section above) and incubated with 10 μM 5′-biotinylated probe oligonucleotides on a tissue culture rotator at RT for 1 h. The p-Chips were then washed with TE buffer three times at RT, and stored in TE buffer at 4 °C. The sequence of the oligonucleotide probe was: 5′-biotin-TTTTTTTTTGCTTTCCTTCACTG-3′. As a negative control, another probe with a point T/C mutation (underlined) was used: 5′-biotin-TTTTTTTTTCTTTCCTCCACTG T-3′. In the alignment of the two sequences, the 14 nt section within the negative control sequence is shifted 1 nt to the 3′ end compared to the probe oligo to ensure that the control and the probe oligo have similar Tm.</p><p>The hybridization process was done as previously reported [19]. In brief, oligonucleotide-linked p-Chips were incubated in 1× pre-hybridization buffer at 45 °C for 10 min. After removing the pre-hybridization buffer, the p-Chips were incubated in 1× hybridization solution that contained 5′-Cy3-labeled target oligonucleotides for 2 h at 45 °C in a hybridization oven (Bambino, Boekel Scientific), in the dark. The sequence of the target oligonucleotide was: 5′-Cy3-AATAACTTTGCAACAGTGAAGGAAAGCCTTTGG A-3′. The target and probe oligonucleotides contained a perfectly matched sequence of 14 nt (underlined). After hybridization, p-Chips were rinsed twice with a pre-warmed (45 °C) 1× washing buffer and incubated in a fresh 1× washing buffer at 45 °C for 30 min. Then the p-Chips were washed with distilled water three times at RT and air dried. Images of fluorescent p-Chips were taken with a digital camera attached to the fluorescent microscope and the intensity quantified.</p><p>The pre-hybridization buffer contained 50 mM Tris–HCl (pH 8.0), 150 mM NaCl, 0.1% SDS, 0.5% Ficoll (type 400), 5 mM EDTA, 200 μg/ml sheared, denatured salmon sperm DNA and 1 μg/μl BSA. The 1× hybridization solution contained 50 mM Tris–HCl (pH 8.0), 150 mM NaCl, 0.1% SDS, 5 mM EDTA (pH 8.0) and 50 μg/ml sheared, denatured salmon sperm DNA. Washing buffer was 50 mM Tris–HCl (pH 8.0) containing 150 mM NaCl and 0.1% SDS.</p><!><p>Silver nanoparticles were deposited on both sides of p-Chips using a reduction reaction with AgNO3, as described in the "Materials and methods" section. The surface of a SIF-deposited p-Chip appeared granular in bright field microscopy images (Fig. 1a). In contrast, the surface of a bare p-Chip was smooth showing only the fine lines resulting from the wafer thinning process. To determine the size of silver nanoparticles the surfaces of SIF on bare p-Chips and polymer-coated p-Chips were scanned by AFM and SEM (Fig. 1b–c). The size of most silver particles was between 100 and 200 nm. However, aggregated silver particles (larger than 1 μm) were also observed occasionally (Fig. 1b–c). Such inhomogeneity of the particles may limit the maximum fluorescence enhancement and cause a batch-to-batch variation (as discussed in the next section). This can be minimized in the future by further improvements of the method to coat p-Chips with SIF.</p><p>We also tried to form the SIF by mixing silver nanoparticles (~100 nm) with the APTS/GPTS polymer-forming solution during the p-Chip coating process. The latter approach was not satisfactory. The silver nanoparticles formed aggregates, especially on the edges of p-Chips and were not evenly distributed on the surface. As a result, the fluorescence enhancement was low (data not shown). Therefore, the chemical reduction method was used for all subsequent experiments.</p><!><p>Silver nanostructures have been shown to be able to enhance fluorescence of nearby fluorophores in many recent studies [3–12]. Such a phenomenon is due to the interactions between the excited-state fluorophores and free electrons in the metal, the so-called surface plasmon electrons, which lead to enhanced local fluorescence intensity and increased radiative decay rates. Geddes et al. estimated that the radiative rate could be theoretically increased 1,000-fold when a fluorophore is placed near a silver nanoparticle [21]. In practice the fluorescence enhancement has been observed in less than ideal conditions under which an aggregate fluorescence of a surface is measured. For instance, Matveeva et al. reported a four- to fivefold fluorescence enhancement in an immunoassay on flat glass [14]. Lukomska et al. reported an eightfold enhancement when SIF was deposited on quartz slides using Texas Red-labeled BSA [22].</p><p>To test the fluorescence enhancement by SIF on silicon surfaces, p-Chips were coated with the APTS/GPTS polymer and the SIF was deposited on the p-Chips. The purpose of polymer coating on the surface of p-Chips was to provide functional groups for conjugation of fluorescent dyes or charged groups to facilitate non-covalent protein binding to the chips. Different fluorescent probes were placed on the surface of the chips, namely GFP (Fig. 2a), AF555 (Fig. 2b), PE (Fig. 4), and Cy3 (Fig. 5). SIF and non-SIF-p-Chips were incubated with GFP (50 μg/ml) or AF555-labeled BSA (10 μg/ml) for 1 h at RT. We observed a 23.7-fold enhancement of the signal for GFP and 14.4-fold for AF555 on SIF-p-Chips compared with non-SIF-p-Chips (Fig. 2). In the test of PE (Fig. 4) and Cy3 (Fig. 5) in immunoassay and DNA hybridization assay, a fluorescence enhancement of 5.5 to 50-fold was consistently observed. All four fluorophores tested showed significant enhancement which varied from five to 50-fold. This means that the SIF-based enhancement is not dye-specific and, thus, has potential for broad use.</p><p>Batch-to-batch variation in fluorescence enhancement was noticed for p-Chips coated with polymer and having a SIF deposited on them. We implemented a routine test to evaluate the variation. The test involved incubating the derivatized p-Chips with GFP and measuring fluorescence enhancement. The observed enhancement was in the range from 8.1- to 23.7-fold (the number is based on three determinations). Despite the fact that the amount of enhancement varied, the enhancement was consistently observed. To minimize the effects of the batch-to-batch variation on the conclusions reached, the p-Chips used in the same experiment came from the same batch in our study. The current protocol can be further optimized to reduce the variation.</p><p>Previous studies have shown that the enhanced fluorescence intensity by SIF is associated with decreased fluorophore lifetimes [3, 4, 8, 9]. Therefore, we also measured the radiative decay rate of AF555 on the surface of SIF-p-Chips after the chips were incubated with AF555-labeled BSA (Fig. 3). We tested both high (50 μg/ml) and low (0.08 μg/ml) BSA-AF555 concentrations. In both cases, the lifetime of AF555 was significantly shorter on SIF-p-Chips compared to non-SIF-p-Chips (Fig. 3). When 50 μg/ml AF555-BSA was used, the half-life of AF555 was 0.19 ns on non-SIF-p-Chips compared to 0.02 ns on SIF-p-Chips, a 9.5-fold reduction. When 0.08 μg/ml AF555-BSA was used, the half-life of AF555 was 1.15 ns on non-SIF-p-Chips compared to 0.05 ns on SIF-p-Chips, a 23-fold reduction.</p><!><p>To test the effects of SIF in bioassays, we performed IL-6 immunoassays. The cytokine IL-6 is a pleiotropic protein that is critical in acute-phase reactions, inflammation, hematopoiesis, bone metabolism, and cancer progression [23–25]. In our test, a monoclonal anti-human IL-6 antibody was conjugated to SIF-p-Chips or non-SIF-p-Chips. The antibody conjugation involved a two-step carbodiimide reaction (see the "Materials and methods" section). The carboxyl groups on the surface of the polymer-coated p-Chips were first activated with a carbodiimide derivative prior to coupling with the primary amines on the antibody. The anti-IL-6 conjugated p-Chips were used to capture recombinant human IL-6. To detect IL-6, a biotinylated anti-human IL-6 antibody was incubated with the p-Chips, followed by incubation with SA-PE (Fig. 4a). We observed that the SIF-induced fluorescence enhancement ranged from 8.1 to 15.5-fold (Fig. 4b). The assay sensitivity was initially calculated using normalized fluorescence intensity units, then interpolated from the standard curve and expressed as pg/ml. The IL-6 curve indicates that the sensitivity for IL-6 standard is 4.3 pg/ml for SIF-p-Chips and 107 pg/ml for non-SIF-p-Chips, a 25-fold improvement. In recent studies, Liao et al. reported that the serum level of IL-6 in healthy people is 4.08±0.28 pg/ml [26] and Edgell reported a level of 31±9 pg/ml in healthy people [27]. In addition, the IL-6 level can be elevated up to tenfold in cancer patients [26, 27]. The assay sensitivity with SIF-p-Chips in our study is 4.3 pg/ml. Thus, it is in the range of detecting serum IL-6 in healthy individuals and even more suitable for cancer patients.</p><p>To test the SIF-p-Chip-based immunoassay in a complex environment, we used normal human serum spiked with 50 ng/ml recombinant human IL-6. The spiked serum was then serially diluted (1:5, 1:10, 1:20 and 1:40, respectively) in PBS. For both the non-SIF and SIF-p-Chip-based assays, the linearity of dilution was confirmed with a R2 greater than 0.99 (Fig. 4c). In addition, the fluorescence enhancement by SIF was consistently observed in the range from 10.7 to 14.9-fold (Fig. 4d). To determine the recovery rate after spiking, the IL-6 level was calculated based on the IL-6 standard curve (Fig. 4a) and compared to the expected level. For non-SIF chips, the recovery rate was in the range from 105% to 128%, while for SIF chips the recovery rate was between 97% and 114% (Table 1). The results suggest that the p-chip-based IL-6 assay works well in a complex environment such as spiked serum.</p><p>It should be noted that the assay sensitivity is influenced by signal strength, affinities of binding molecules used, fluorescence background, structure of the polymer on the solid phase, design of the analytical instrument to measure the fluorescence intensity, and others factors. In this report, we only concentrated on enhancing the fluorescence signal strength. Thus, it is anticipated that the sensitivity of the assays can be further improved by using a more sensitive detection method, reducing fluorescence background by optimizing the polymer surface on SIF-p-Chips and screening for antibodies demonstrating the highest affinity in p-Chip-based assays.</p><!><p>We also performed a simple DNA-based assay on SIF-derivatized p-Chips. The assay involved a determination of whether the target DNA (here: an oligonucleotide) carried a mutation in the human cystic fibrosis transmembrane conductance regulator gene. The target oligonucleotide contained the mutant allele of the W1282X mutation, one of the 23 mutations in the basic panel used in many CF tests [19]. The probe oligonucleotide-linked on p-Chips is specific to the target oligonucleotide. We also used a control oligonucleotide designed to be only specific to the wild type allele.</p><p>First, a protein layer of avidin was built by conjugation to the polymer surface of SIF-p-Chips or non-SIF-p-Chips. The biotinylated oligonucleotide probe was then bound to the avidin protein layer, and the Cy3-labeled target oligonucleotide hybridized to the probe-linked p-Chips. We tested different concentrations of the target oligonucleotide to derive a standard curve (Fig. 5a). The fluorescence enhancement was in the range from 5.5 to 50-fold (Fig. 5b). The assay sensitivity was determined to be 1.7 nM for the non-SIF-p-Chips and 0.05 nM for the SIF-p-Chips, a 38-fold enhancement (Fig. 5a).</p><p>The purpose of using the avidin layer was to create a short distance between the oligonucleotides and SIF to avoid quenching effects when the reporting fluorophores are too close to the SIF [28]. In our test, we observed that SIF-induced fluorescence enhancement is significantly reduced without the avidin layer (data not shown). We also noticed that the fluorescence enhancement is dependent on the concentration of target oligonucleotide with the greatest enhancement observed at 0.5 nM (Fig. 5b). The mechanism responsible for this observation is still under investigation.</p><!><p>The present study was focused on testing colloidal silver nanoparticles as a vehicle to enhance fluorescence intensity on a flat surface through plasmon resonance. We tested various fluorophores and showed that the application of SIF to the p-Chip-based assay enhanced fluorescence intensity from five to 50-fold. The increase of fluorescence intensity of the p-Chips led to a clear improvement in the assay sensitivity: a 25-fold increase over the method not involving a SIF in the IL-6 immunoassay, and a 38-fold increase in the DNA hybridization assay. The SIF-p-Chip-based assay has clearly great potential in extending multiplex assays to higher sensitivities. It is expected that the methods developed here are applicable to other types of multiplex assays, such as those based on microarrays or microbeads both in basic research as well as clinical applications.</p><!><p>Silver nanoparticles on the surface of bare p-Chip coated and not coated with SIF. Non-electronic sides of the p-Chips are shown. No polymer coating was applied to the p-Chips in panels a and b. a Bright field images of p-Chips. b AFM images of p-Chips. The sizes of most silver particles are in the range from 100 to 200 nm. Aggregates of silver particles can also be seen. c SEM image of an APTS/GPTS-coated p-Chip deposited with SIF</p><p>Fluorescence enhancement for different fluorophores. SIF was deposited on APTS/GPTS-coated p-Chips. The p-Chips were incubated with 50 μg/ml GFP (a) or 10 μg/ml AF555-labeled BSA (b). Fluorescence was greatly enhanced on SIF-p-Chips (p-Chips on the right) compared to non-SIF-p-Chips (p-Chips on the left). In the control (Ctrl) group, SIF-p-Chips were incubated in PBS. Panel a magnification 40×, exposure time 15 s. Panel b magnification 40×, exposure time 50 ms. Experiments were repeated at least three times; representative p-Chips are shown</p><p>Fluorescence intensity decay rates of AF555 on SIF-p-Chips. SIF-p-Chips were incubated with 50 μg/ml AF555-BSA (a) or (b) 0.08 μg/ml AF555-BSA. Left panels no SIF was applied. Right panels SIF was deposited on the p-Chips. Blue lines represent the fluorescence intensity decay curves. Black lines are the best multi-exponential fits. Red profile is an instrument response function (IRF) taken from the scatter at the 470 nm excitation wavelength</p><p>The application of SIF in p-Chip-based IL-6 immunoassay. a Standard curve for the IL-6 assay. Anti-human IL-6 monoclonal antibody was conjugated to the polymer surface of SIF-p-Chips and non-SIF-p-Chips. The anti-IL-6 conjugated p-Chips were used to capture the recombinant human IL-6. The background signal was determined using PBS in the capture step. Biotinylated anti-human IL-6 antibody was then used for detection, followed by incubation with SA-PE. The background signal (PBS control) from non-SIF or SIF-p-Chips was subtracted. Error bar: standard deviation from n=3 or 4 p-Chips. b The fluorescence enhancement by SIF. c The linearity of dilution in spiked serum. Normal human serum was spiked with 50 ng/ml recombinant human IL-6 and serially diluted (1:5, 1:10, 1:20, and 1:40) in PBS. The IL-6 immunoassay on non-SIF and SIF-p-Chip was performed as described in (a). The linearity of dilution was then determined. d The fluorescence enhancement by SIF using the spiked human serum. aFluorescence enhancement is the ratio of fluorescence intensity of SIF-p-Chips to that of non-SIF-p-Chips</p><p>The effect of SIF in a p-Chip-based DNA assay. Avidin was conjugated to the polymer surface of SIF-p-Chips and non-SIF-p-Chips. Biotinylated oligonucleotide probe was then attached to the avidin-conjugated p-Chips, and a Cy3-labeled target oligonucleotide was hybridized to the conjugated p-Chips. A probe that had a single mismatch to the target oligo was used as a negative control (data not shown as fluorescence intensities were very low, typically less than 10% of that found in corresponding experimental groups). As a control to determine the background signal, no target oligonucleotide was used in the hybridization step. The Cy3 fluorescence intensity (a) and fluorescence enhancement (b) were determined and plotted. The background signal (no target control) from non-SIF or SIF-p-Chips has been subtracted from fluorescence intensity data. Error bar indicated the standard deviation calculated from fluorescence intensity measurements for n=3 or 4 p-Chips. aFluorescence enhancement is the ratio of fluorescence intensity of SIF-p-Chips to that of non-SIF-p-Chips</p><p>Recovery of dilution experiment</p><p>Normal human serum was spiked with 50 ng/ml recombinant human IL-6. The observed value for the recovery of dilution was determined from the IL-6 standard curve (Fig. 4a)</p>

PubMed Author Manuscript

A small group of sulfated benzofurans induces steady-state submaximal inhibition of thrombin

Despite the development of promising direct oral anticoagulants, which are all orthosteric inhibitors, a sizable number of patients suffer from bleeding complications. We have hypothesized that allosterism based on the heparin-binding exosites presents a major opportunity to induce sub-maximal inhibition of coagulation proteases, thereby avoiding/reducing bleeding risk. We present the design of a group of sulfated benzofuran dimers that display heparin-binding site-dependent partial allosteric inhibition of thrombin against fibrinogen (\xce\x94Y = 55\xe2\x80\x9375%), the first time that a small molecule (MW <800) has been found to thwart macromolecular cleavage by a monomeric protease in a controlled manner. The work leads to the promising concept that it should be possible to develop allosteric inhibitors that reduce clotting, but do not completely eliminate it, thereby avoiding major bleeding complications that beset anticoagulants today.

a_small_group_of_sulfated_benzofurans_induces_steady-state_submaximal_inhibition_of_thrombin

<p>It is estimated that nearly 10% of the adult population will be treated with anticoagulants at least once in their lifetime. Orally bioavailable agents constitute frequently used anticoagulants these days but their risk of bleeding remains substantial. A number of patients on oral anticoagulants suffer from bleeding consequences,1 which raises considerable harm.</p><p>All current direct oral anticoagulants (DOACs) target the active site of a coagulation enzyme, e.g., thrombin or factor Xa.2 In this orthosteric inhibition mechanism, the only parameter available for regulation of clotting activity is the inhibitory potency (either IC50 or KI). In contrast, allosteric inhibition mechanism offers two parameters for regulating clotting activity, i.e., potency and efficacy (ΔY). Such dual parameter regulation is not possible for orthosteric inhibitors because of their all (ΔY = 100%) or none (ΔY = 0%) property.</p><p>Allosteric inhibition also promises to offer enhanced level of selectivity because allosteric sites typically tend to be less homologous.3,4 This has certainly been found to be true for a group of allosteric inhibitors of thrombin that we have been studying for the past few years. We had reasoned that appropriate sulfated non-saccharide glycosaminoglycan mimetics (NSGMs) could be designed to target exosite 2 of thrombin and induce allosteric inhibition (Fig. 1). To realize this, we first developed sulfated low molecular weight lignins5 as polymeric mimetics of the highly sulfated polysaccharide, heparin, which binds to an exosite of certain serine proteases with 10–20 μM affinity.6 However, the sulfated lignins inhibited several coagulation proteases.7 To enhance selectivity, we designed sulfated benzofuran monomers (SBMs), which preferentially inhibited thrombin and factor Xa,8 albeit displaying a high IC50. Further improvement in design led to sulfated benzofuran dimers (SBDs, Fig. 1) that selectively inhibited thrombin by interacting with Arg173 on the periphery of exosite 2.9,10 This was a major advance. Yet, the allosteric SBDs displayed efficacies ≥90%, which does not truly afford an opportunity for efficacy-based regulation (i.e., based on ΔY).</p><p>Allosteric agents that display efficacies, or alternatively responses, in the intermediate range (e.g., 30–70%) have been well-known for receptors and oligomeric proteins. Such agents are referred to as partial antagonists. However, partial allosterism has been extremely challenging to discover for monomeric proteins or proteases. We recently discussed the first example of an allosteric partial inhibitor of a monomeric protease, a sulfated coumarin analog, which displayed sub-maximal efficacy.11 The analog reduced thrombin's hydrolysis of a small chromogenic substrate with a maximal efficacy of only about 50%. However, this partial allosterism was lost for fibrinogen, thrombin's natural macromolecular substrate. We reasoned that alternative small, allosteric partial inhibitors of thrombin that function even against the macromolecular substrate should be possible to design/discover considering that exosite 2 displays multiple hydrophobic sub-sites that are coupled with its catalytic triad.12</p><p>To realize such agents, we focused on a prototypic SBD, 1 (Fig. 1), which had been designed earlier but displayed ΔY of 75%.9 We posited that modifications in aromatic and ester substituents of 1 may lead to a more favorable partial inhibition characteristics (ΔY ~30 – 70%). We synthesized a library of 16 SBDs with variations in R, R', R1 and R1′ substituents (Fig. 1). We decided that whereas predecessor 1 carries methyls or ethyls at these positions, the analogs would carry larger lipophilic substituents, which could possibly engage the hydrophobic sub-domains present in exosite 2 better.13 The synthesis of these SBDs was achieved in 7–14 steps involving protection–deprotection, nucleophilic substitution, free radical bromination and sulfation reactions (see Supplementary Material and Schemes S1–S7). We also synthesized 12 new SBMs (see Supplementary Table S1), which were also studied.</p><p>Thrombin inhibition was studied using chromogenic substrate hydrolysis assay at pH 7.4 and 25 °C in the presence of 2.5 mM CaCl2, as described elsewhere.9-12 Of the 13 SBDs studied, 11 were found to be 2–3-fold more potent than 1 identified earlier (Table 1).10 Of these, 2c carrying a PhCH2CH2 group at the R′ position, instead of a CH3 present in prototype 1, displayed the best potency (IC50 = 1.8 μM, Fig. 2). In contrast, 2o carrying sulfate groups and 2p at the R′ position failed to inhibit thrombin at concentrations lower than 300 μM suggesting a sensitive structure–activity relationship (SAR).</p><p>Although the discovery of higher potency was interesting, the most exciting finding was the property of partial inhibition for this series of agents. Of the 11 active inhibitors, nine displayed ΔY in the range of 55 to 75% (Table 1, Fig. 2a). Inhibitor 2c inhibited thrombin with an efficacy of 58% at saturation, a characteristic of possibly major consequences with regard to regulation of enzyme function. Such partial inhibition profile is not possible for orthosteric agents. Interestingly, two agents, i.e., 2h and 2i, inhibited thrombin with ~90% efficacy suggesting intricate SAR involving inhibition efficacy too.</p><p>To confirm that this phenomenon arises from direct binding, we measured thrombin affinity of SBDs using spectrofluorimetry. The fluorescence of FPRCK-thrombin, monitored as a function of two representative SBDs 2c and 2i at pH 7.4 and 25 °C decreased in a classic hyperbolic manner (Fig. 2b) to yield KDs of 3.7±0.3 and 1.0±0.1 μM, respectively. In a similar manner, the affinity of wild-type plasma thrombin for 2c, determined using intrinsic fluorescence, was found to be 2.8±0.3 μM. The similarity of the KDs and IC50s is expected on the basis of allosteric inhibition mechanism.</p><p>To ascertain the mechanism of thrombin inhibition, we resorted to Michaelis-Menten kinetic studies. For both 2c and 2i, the VMAX of Spectrozyme TH hydrolysis decreased nearly 2-fold in a dose-dependent manner, as expected (see Supplementary Fig. S1 and Table S2). However, their effect on KM was distinctly different. For 2c, KM remained essentially constant at 8.2±1.1 μM, whereas for 2i, it increased from 8.6 μM to 61 μM as the inhibitor concentration increased to 40 μM (Table S2). Thus, although 2c and 2i are structural analogs, they exhibit slightly different mechanism of inhibition. SBD 2c functions as a noncompetitive inhibitor, while 2i displays mixed inhibition. Yet, both 2c and 2i do not compete with the substrate alluding to an allosteric inhibition process.</p><p>To identify the allosteric site engaged by 2c and 2i, we utilized prototypic ligands hirudin peptide (HirP) and unfractionated heparin (UFH) that bind to exosites 1 and 2, respectively. Figure 3 shows the thrombin inhibition profiles at varying levels of HirP and UFH. The profiles further clarify the difference between the two related allosteric inhibitors. The potency of thrombin inhibition by both 2c and 2i remains essentially unaffected by the presence of HirP (IC50 = 1.8±0.6 μM (2c) or 2.8±0.6 μM (2i), see Table S3) suggesting that these molecules do not engage exosite 1. For competition by UFH, the potency of 2c decreases by a factor of 11-fold, whereas that of 2i remains constant at 2.5±0.4 μM (Table S3). This implies that 2c competes with UFH, while 2i does not.</p><p>To further pinpoint the site of binding, we utilized a group of 11 recombinant thrombin mutants containing replacement of a single electropositive residue of exosite 2 by alanine. These mutants have been studied earlier in the Rezaie laboratory14,15 and cover nearly all lysines and arginines known to interact with UFH.6 For 2c, there was virtually no difference in IC50 between wild-type and mutant thrombin except for Arg233Ala and Lys235Ala proteins (Fig. 4a). This implies that whereas the prototypic SBD 1 was found to engage Arg173, 2c behaved in a variant fashion. Considering that 1 and 2c differ only in the type of substituent at the R′ position, these differences bring forth additional subtle facets of thrombin allostery.16,17</p><p>For 2i, the potency changed significantly for several mutants including Arg165Ala, Lys169Ala, Arg175Ala, Arg233Ala and Lys236Ala (Fig. 4b). This is an unusual result if one takes into account the absence of competition with UFH (Fig. 3c, Table S3). Yet, the result can be explained by noting that 2i cannot possibly engage each of these residues simultaneously if it binds in a highly selective manner. Most probably it samples many different binding modes in which each of these residues are engaged part of the time. Such a non-selective interaction process not only explains non-competition with UFH but also helps understand the mixed inhibition mechanism identified above (Fig. S1). These differences also help rationalize the considerable difference in efficacy of inhibition observed between 2c and 2i (see Fig. 2).</p><p>In terms of drug discovery, the above results present 2c and 2i, two small molecules with significant difference in efficacy of inhibition, as tools to evaluate the concept of allosteric partial inhibition. As stated in the introductory section, partial allosterism against small chromogenic substrate may not necessarily hold against a macromolecular substrate.11 In fact, it is to be expected that the binding energy gained upon complexation with a macromolecule (e.g., fibrinogen), which is typically high, may help disengage the small molecule from its binding site, thereby releasing the partial allosteric conformation. This is the key reason behind the difficulty of discovering such small inhibitors of monomeric proteases.</p><p>So to assess whether the submaximal inhibition, observed against the chromogenic substrate, would be transferred to the primary thrombin substrate in vivo, we utilized polyacrylamide gel electrophoresis (PAGE) for monitoring fibrinogen cleavage. Thrombin cleavage of fibrinogen results in cleaved fibrinogen of ~320 kDa and two fibrinopeptides of much lower molecular weights (which cannot be followed easily using PAGE). We performed densitometry analysis of bands at 320 kDa to measure levels of uncleaved fibrinogen in the presence and absence of 2c and 2i (Fig. 5). The results show that cleavage of fibrinogen by thrombin reduces as the concentration of inhibitors increase, as expected. However, at saturating levels of 2c, inhibition of fibrinogen cleavage reached a maximum of about 70±8% (Fig. 5B), whereas in the presence of inhibitor 2i, inhibition was found to be greater than 90%.</p><p>These results with small molecules can be compared with allosteric peptide-based inhibitors of proteins. For example, thrombin's activity is known to be modulated by allosteric peptides and substrates18-20 as well as by nucleic acids.21 However generally, partial allosteric inhibition has not been documented to date. Thus, the sulfated benzofurans being presented in this work as partial inhibitors represent a novel class of allosteric agents.</p><p>Another point worth mentioning is that the results imply that a molecule as small as 2c, which is less than 800 Da in size, induces a conformational change in thrombin that cannot be reversed by the binding of a large macromolecule (320 kDa) such as fibrinogen. To the best of our knowledge, this is the first observation of induction of conformational rigidity in a monomeric protease by a small molecule that even a 400-fold larger molecule fails to undo.</p><p>Most probably, partial allosterism arises from as yet-identified specific coupling between the allosteric site and the protease active site. This is supported by the observation that 2c and 2i appear to bind to different sites on thrombin, although being structurally very similar. The results also suggest that SBD inhibition of thrombin is likely to be a thermodynamic property, and not a kinetic property. This is important because if allosteric inhibition is a kinetic property, it is unlikely to benefit in terms of drug development. Finally, whereas this concept of partial allosterism has been known and easily applied for multimeric proteins, e.g., receptors, realizing it for monomeric, soluble proteins has been extremely difficult.</p><p>A key outcome of the concept being demonstrated here is that it may help solve the problem of bleeding risk associated with DOACs. Current anticoagulants completely shut down clotting at saturation and inadvertent use of higher doses leads to bleeding. In contrast, partial allosteric inhibition of coagulation proteases would only reduce clotting signal even at saturation, or in the event of an inadvertent excessive dose. For example, in the case of 2c, approximately 30% fibrinogen cleavage continues to proceed even in the presence of saturating levels of inhibitor.</p><p>In summary, we present the first SBDs that display partial allosterism against fibrinogen (ΔY = 55–70%), the first time a small molecule (MW <800) has been found to thwart macromolecular cleavage of a monomeric protein in a regulated manner. We believe that partial allosterism of thrombin is a valuable concept because it affords the possibility of maintaining basal level of a pro-clotting signal. Such basal level of fibrinogen cleavage can be expected to retard bleeding and thereby reduce risk of major bleeding, which beset all current DOACs. This work also indicates that this promising concept should be possible to explore for other monomeric coagulation proteases, e.g., factor Xa.</p>

PubMed Author Manuscript

Glycoconjugated Metallohelices have Improved Nuclear Delivery and Suppress Tumour Growth In Vivo

AbstractMonosaccharides are added to the hydrophilic face of a self‐assembled asymmetric FeII metallohelix, using CuAAC chemistry. The sixteen resulting architectures are water‐stable and optically pure, and exhibit improved antiproliferative selectivity against colon cancer cells (HCT116 p53+/+) with respect to the non‐cancerous ARPE‐19 cell line. While the most selective compound is a glucose‐appended enantiomer, its cellular entry is not mainly glucose transporter‐mediated. Glucose conjugation nevertheless increases nuclear delivery ca 2.5‐fold, and a non‐destructive interaction with DNA is indicated. Addition of the glucose units affects the binding orientation of the metallohelix to naked DNA, but does not substantially alter the overall affinity. In a mouse model, the glucose conjugated compound was far better tolerated, and tumour growth delays for the parent compound (2.6 d) were improved to 4.3 d; performance as good as cisplatin but with the advantage of no weight loss in the subjects.

glycoconjugated_metallohelices_have_improved_nuclear_delivery_and_suppress_tumour_growth_in_vivo

<!>Introduction<!><!>Introduction<!>Selection of metallohelix system<!><!>Synthesis and characterization<!><!>Synthesis and characterization<!>Antiproliferative activity and cell studies<!><!>Antiproliferative activity and cell studies<!><!>Antiproliferative activity and cell studies<!><!>Antiproliferative activity and cell studies<!><!>Antiproliferative activity and cell studies<!><!>Biophysical studies in vitro<!>In vivo studies<!><!>Conclusion<!>Conflict of interest<!>

<p>H. Song, S. J. Allison, V. Brabec, H. E. Bridgewater, J. Kasparkova, H. Kostrhunova, V. Novohradsky, R. M. Phillips, J. Pracharova, N. J. Rogers, S. L. Shepherd, P. Scott, Angew. Chem. Int. Ed. 2020, 59, 14677.</p><!><p>We have developed several structurally distinct ranges of metallohelices comprising three organic ligands that encapsulate two metal ions,1 such as that shown in Scheme 1 a. Unlike conventional helicates,2 these water‐stable FeII compounds self‐assemble as optically pure architectures, principally a result of inter‐ligand steric and secondary interactions including hydrophobic π‐stacks.3 There is mounting evidence that as a result of their charge, shape, size and amphipathic structures, these compounds emulate some of the functional properties of short cationic α‐helical peptides. Oriented binding to various nucleic acid structures is observed.1a, 4 One class1b inhibits ice recrystallization apparently as a result of the facially amphipathic architecture that is also present in natural antifreeze peptides.5 A similar structure binds to the central hydrophobic α‐helical region of an amyloid β protein and attenuates toxicity.6 Perhaps most convincingly, we showed recently that a class of antimicrobial metallohelix in our library1e rapidly penetrates the formidable cell envelope of a clinically‐relevant Gram negative microbe and causes a peptide‐like genomic and transcriptomic response.</p><!><p>Synthesis of new sugar‐functionalised metallohelices, using CuAAC post‐assembly modification of self‐assembled triplex metallohelices.</p><!><p>Cell‐penetrating peptides (CPPs) are usually relatively short (5–50 residues)7 and contain an excess of cationic amino acids (lysine and arginine).8 It is proposed that they pass through the plasma membrane via an ion exchange mechanism9 using negatively charged species such as anionic lipids and glycosaminoglycans. Since these components are in excess in cancer cell and microbial outer‐leaflets,10 a generalized source of selectivity over other cells is provided. Nevertheless, such polycationic molecules may also have non‐specific affinity for a number of biomolecular structures7, 11 and the modification of CPPs with biocompatible fragments has been used in an attempt to modulate the attendant toxicity.11b, 12 In particular, glycoconjugation has been used extensively for the modification of potential therapeutics of a number of kinds.13 In nature, glycosylation is one of the most common post‐translational modifications14 and glycopeptides are involved in cell signalling,15 providing cell surface markers for recognition, and immune response.16 From a drug‐design perspective, monosaccharide‐conjugated analogues have been reported in the literature since the early 1990s,17 improving the water solubility and serum stability of their cargo,17b as well as altering drug metabolism and pharmacokinetics (DMPK)18 including some literature precedent for exploiting the Warburg effect in cancer therapy.19</p><p>Several groups have also shown that glycosylation of a peptide increases membrane penetration, including through the blood–brain barrier.20 In recent work, Montenegro and co‐workers developed a strategy for the glycosylation of short peptides, and have systematically characterized the uptake efficiency and distribution in various cell lines.21</p><p>Our recent success in CuAAC derivatization of metallohelices using relatively simple functionality,1d and an in cellulo click staining protocol,1e gave us confidence to attempt the rather more ambitious glycoconjugations. We report here that this chemistry, giving rise to some of the most complex functionalized metallosupramolecular structures known, proceeds smoothly and efficiently, leading to improved cancer‐cell targeting in vitro, and improved efficacy in vivo.</p><!><p>The position of hydrophobic regions within a peptide is conventionally assessed by a simple residue‐based approach, but this is not applicable here. Instead, analysis5 (Figure S1 in the Supporting Information) of the position of counter‐anions in the solid state molecular structure reveals a favorable charge distribution for one of our so‐called triplex1b architectures (Scheme 1) in that the two π‐stacked arene rings, colorized pink in Figure 1, shield the cationic charge, leading to the creation of a relatively hydrophobic upper ridge. A third π‐stack is hidden at the rear of this view. The yellow colorized atoms correspond to the positions of groups R in Scheme 1; they will surround a relatively hydrophilic face and hence by adding sugar units at these latter positions we retain the amphipathic architecture. This, we considered, was the approach most likely to allow retention of the kinds of biological activity we have seen from the core structure, while allowing us to test the idea that glycosylation may lead to improvements in delivery, selectivity and tolerance.</p><!><p>Schematic representation of [M2 L1 3]4+ architecture. Space‐fill model based on a previously reported structure.1b Note the two π‐stacked arenes colorized in pink on the upper hydrophobic edge. The H atoms colorized in yellow correspond to the positions of the R groups attached on the lower hydrophilic face.</p><!><p>The starting materials for new synthesis were assembled: the previously‐reported1d enantiomerically pure triplex metallohelix [Fe2 L2 3]Cl4 with alkynyl groups at the positions colorized yellow in Figure 1 was prepared on a multi‐g scale via a one‐pot highly diastereoselective self‐assembly reaction; the range of monosaccharide azides of Scheme 1 b, including acylated analogues, were synthesized by literature procedures.22</p><p>The subsequent CuAAC glycosylation was not initially straightforward. The conventional copper sulfate/sodium ascorbate catalyst led to difficulties in isolation in this rather polar system, while the heterogeneous catalyst copper‐in‐charcoal23 failed to complete the reaction. We considered the copper free click reaction24 but the requirement for cyclooctyne groups would significantly increase the synthetic challenge and restrict versatility. Eventually we found to our surprise that that while copper(I) iodide catalyst required elevated temperatures, this was not deleterious, the reactions were complete, and the work‐up was trivial. This gave us access to the glycoconjugated triplex metallohelices [Fe2 L3a‐g 3]Cl4 as optically pure isolated compounds.</p><p>The success of this post‐assembly CuAAC is apparent from the 1H‐NMR spectra (Figure 2 A and B; for all spectra see Figure S2–S9). For example, the singlets Hj at ca 3 ppm corresponding to the three inequivalent alkyne units in the starting material are cleanly replaced by three new singlets at 8.06, 8.17, and 8.28 ppm (Hm) for the triazole rings in the product. It is also noteworthy that the two bipyridine protons involved in inter‐strand hydrogen bonds, and thus giving rather low field resonances (ca 9.2 ppm), are present in both starting material and product, confirming that the asymmetric triplex architecture is unperturbed by the presence of the sugars. High resolution electrospray mass spectra were readily obtained; Figure 2 C shows the expected tetracationic molecular ion pattern for Sc,ΛFe‐HHT‐[Fe2 L3a 3]Cl4. The circular dichroism (CD) spectra of the diastereoisomers (Figure 2 D) Λ‐[Fe2 L3a 3]Cl4 and Δ‐[Fe2 L3a 3]Cl4 in H2O display peaks of opposite molar differential extinction coefficients, and mimic the features of the enantiomeric pairs of [Fe2 L1 3]Cl4 and [Fe2 L2 3]Cl4.1d</p><!><p>Characterization of glucose‐functionalized triplex metallohelices. A) 1H NMR (500 MHz, D2O, 298 K) of the precursor complex Δ‐[Fe2 L2 3]Cl4 (cyan), and B) of the product Δ‐[Fe2 L3a 3]Cl4 (red) following CuAAC. C) High resolution ESI mass spectrum of Λ‐[Fe2 L3a 3]Cl4 showing the observed z=+4 charge (top), compared to the theoretical isotope pattern (bottom). D) Circular dichroism spectra of Λ‐[Fe2 L3a 3]Cl4 (black) and Δ‐[Fe2 L3a 3]Cl4 (blue) (40 μm in H2O).</p><!><p>The glycoconjugated compounds were found to be extraordinarily stable under aqueous conditions; no decomposition was observed on monitoring the absorption at the MLCT band in aqueous solution over many months, and even when dissolved in KCl/HCl buffer at pH 1.5 (at 8 mm) no decomposition was observed over one month (Figure S18).</p><!><p>The whole panel of FeII compounds of Scheme 1 were evaluated alongside cisplatin for potency against the human colorectal cancer cells with wild‐type p53 (HCT116 p53+/+) and non‐cancerous human epithelial retinal pigment cells (ARPE‐19) (Figure 3).</p><!><p>Antiproliferative activity of triplex metallohelices in cancer and non‐cancer cells. The half maximum inhibitory concentration (IC50) values are measured in triplicate by MTT assay, dosing for 96 h against HCT116 p53+/+ and ARPE‐19 cells. A) Λ‐triplex metallohelices; B) Δ‐triplex metallohelices. The selectivity index C) defined as [mean IC50(ARPE‐19)]/[mean IC50(HCT116 p53+/+)] for the clinical drug cisplatin (cisPt), the "parent" triplex [Fe2 L1 3]Cl4 and CuAAC‐derived sugar systems [Fe2 L3a‐g 3]Cl4.</p><!><p>We observe that the sugar‐appended triplex systems all inhibit HCT116 p53+/+ cell proliferation in the 2–30 μm concentration range (96 h IC50), and for all examples the Λ‐diastereoisomers are more potent than Δ. The selectivity indices (SI, defined as IC50 [ARPE19]/IC50 [HCT116 p53+/+]) vary from 1.4 to 17, with greater selectivity observed most often with the Δ‐diastereoisomers. With SI of 17, Δ‐[Fe2 L3a 3]Cl4 is the most selective compound in the panel for this pair of cells. Since this indicates a potential therapeutic window, we chose to focus on this compound for more detailed study.</p><p>We compared the antiproliferative activity of Δ‐[Fe2 L3a 3]Cl4 in both glucose‐rich and glucose‐free media and observed no difference in IC50 (Table S2). We further incubated the drug with GLUT‐1 overexpressing MCF‐7 breast cancer cells and compared the IC50 with wild type MCF‐7 cells and found that rather than being more sensitive to the glucose derivative, the GLUT‐1 overexpressing cells are actually ca three‐fold more resistant (Table S3). Firstly, this suggests that the cellular entry of these compounds is not (or not mainly) GLUT‐mediated; given the specificity of binding of this receptor this is perhaps unsurprising, but the addition of glucose units to large molecules has been nevertheless been described as a cancer cell‐targeting strategy.17b, 25 Secondly, we note that the resistance we observed may be beneficial in that normal cells that have high GLUT‐1 expression (e.g. red blood cells) will be less adversely affected.</p><p>The conjugation of sugars with therapeutic peptides and other drug candidates can alter pharmacokinetic properties, and has been demonstrated to improve physiological properties and bioavailability,26 such as enhancing biodistribution in tissues,27 improving membrane penetration28 and targeted delivery.29 We therefore firstly compared the effects of Δ‐[Fe2 L1 3]Cl4 and the glycosylated analogue Δ‐[Fe2 L3a 3]Cl4 on the cell cycle in HCT116 p53+/+ cells, which were treated at different concentrations for 24 h and then evaluated via flow cytometry.</p><p>As shown in Figure 4, Δ‐[Fe2 L1 3]Cl4 induces a decrease in the proportion of cells in the G2/M phase (green), whereas in cells treated with Δ‐[Fe2 L3a 3]Cl4 this remains unchanged even up to 20 μm. Correspondingly, Δ‐[Fe2 L1 3]Cl4 causes a slight dose‐dependent increase in the proportion of cells in the G1 and S phases of the cell cycle. In distinct contrast, Δ‐[Fe2 L3a 3]Cl4 induces a dose‐dependent loss of the number of cells in G1 phase in favor of S phase. These findings indicate a change in mechanisms of action upon attaching the glucose unit to the triplex metallohelix. The counts associated with the sub‐G1 phase were also analyzed; the increasing amount of cell material indicates a growing number of cells undergoing cell death, with Δ‐[Fe2 L1 3]Cl4 inducing greater cell death than Δ‐[Fe2 L3a 3]Cl4.</p><!><p>Cell cycle analysis in HCT116 p53+/+ cells. Effects of Δ‐[Fe2 L1 3]Cl4 and Δ‐[Fe2 L3a 3]Cl4 on cell‐cycle profiles of HCT116 p53+/+ cells treated for 24 hours. A) Percentages of counts allocated to individual populations, G1 (red), S (blue dashed) and G2/M (green). B) Percentages (of total) of cells associated with sub‐G1 phase. The dashed lines show the average value (with SD) of non‐treated control. C) Cell cycle profiles. Effects of Δ‐[Fe2 L1 3]Cl4 and Δ‐[Fe2 L3a 3]Cl4 on cell‐cycle profiles of HCT116 p53+/+ cells treated for 24 hours. (i) control, non‐treated cells, (ii) 20 μm Δ‐[Fe2 L1 3]Cl4, (iii) 40 μm Δ‐[Fe2 L1 3]Cl4, (iv) 10 μm Δ‐[Fe2 L3a 3]Cl4 and (v) 20 μm Δ‐[Fe2 L3a 3]Cl4.The cells were stained with propidium iodide and assessed by FACS analysis. Red represents G1 phase, blue dashed S phase and green G2/M phase. Data were gained using FSC Express software.</p><!><p>We also compared the cellular accumulation of Δ‐[Fe2 L3a 3]Cl4 with that of Δ‐[Fe2 L1 3]Cl4; HCT116 p53+/+cells were incubated with metallohelix (5 μm) for 16 h, and Fe content was determined using ICP‐MS, with Fe counts for untreated control cells subtracted as a baseline from all samples. In addition, we determined the nuclear uptake of Δ‐[Fe2 L1 3]Cl4 and Δ‐[Fe2 L3a 3]Cl4 under the same conditions using a Nuclei EZ Prep (Sigma–Aldrich) nuclei isolation kit.</p><p>Accumulation of 21.9±2.1 pmol Fe/106 cells was observed following incubation with Δ‐[Fe2 L1 3]Cl4 and 15.9±2.7 pmol Fe/106 cells with Δ‐[Fe2 L3a 3]Cl4 (Figure 5 A).Despite the lower cellular uptake of Δ‐[Fe2 L3a 3]Cl4 compared to Δ‐[Fe2 L1 3]Cl4 (ca 73 %), 2.5 times more Fe was localized in the nucleus; only 4 % of the total ion uptake was associated with the nuclei for the parent triplex Δ‐[Fe2 L1 3]Cl4, whereas 12 % was observed for the sugar‐conjugate Δ‐[Fe2 L3a 3]Cl4.</p><!><p>Cellular uptake, distribution and single‐cell gel electrophoresis. A) Cellular and nuclear uptake of Δ‐[Fe2 L1 3]Cl4 and Δ‐[Fe2 L3a 3]Cl4 in HCT116 p53+/+cells treated for 16 h at 5 μm concentrations. Fe content was measured by ICP‐MS, and Fe content measured in untreated control cells was subtracted from each measurement. Nuclear were isolated using a Nuclei EZ prep kit. B) Cellular distribution of Fe in HCT116 p53+/+ cells treated under the same conditions and processed into sub‐cellular components using a FractionPREP cell fractionation kit. C) Single‐cell gel electrophoresis (Comet assay) analysis. Top panels: analysis of DNA strand break induction in HCT116 p53+/+ cells untreated (i) or exposed to 20 μm Δ‐[Fe2 L1 3]Cl4 (ii) and 10 μm Δ‐[Fe2 L3a 3]Cl4 (iii) for 18 h. Bottom panels: analysis of DNA crosslink induction in untreated (iv) or cells treated with 20 μm Δ‐[Fe2 L1 3]Cl4 (v) and 10 μm Δ‐[Fe2 L3a 3]Cl4 (vi) for 18 h; after treatment the cells were exposed to hydrogen peroxide.</p><!><p>To confirm this observation, the intracellular compartmentalization of Δ‐[Fe2 L1 3]Cl4 and Δ‐[Fe2 L3a 3]Cl4 in HCT116 p53+/+ was also investigated using a FractionPREP™ Cell Fractionation kit (BioVision) to isolate four sub‐cellular fractions: (i) cytoskeletal fraction (total cellular insoluble proteins) plus genomic DNA), (ii) nuclear fraction (nuclear soluble proteins, including nuclear membrane proteins), (iii) membrane fraction(organelles and organelle membrane proteins, but excluding nuclear membrane proteins), and (iv) cytosolic fraction (total cytoplasmic soluble proteins). The cells were grown and treated as above and Fe content was again determined by ICP‐MS. As shown in Figure 5 B, the localization of Δ‐[Fe2 L3a 3]Cl4 in the nuclear fraction (13.6 %) was more pronounced in comparison with Δ‐[Fe2 L1 3]Cl4 (4.4 %), and was consistent with the data observed in Figure 5 A. Both Δ‐[Fe2 L1 3]Cl4 and Δ‐[Fe2 L3a 3]Cl4 distribute most predominantly in the membrane fraction at 16 h (62.0 % and 55.7 % respectively), whereas the localization of Δ‐[Fe2 L1 3]Cl4 (25.3 %) in the cytoskeleton fraction is more significant than for Δ‐[Fe2 L3a 3]Cl4 (19.6 %). There are several reports of glycosylation‐dependent nuclear import of proteins and plasmids,30 which could be related to the cytosolonuclear lectins shuttling between the cytosol and the nucleus.30c</p><p>Single‐cell gel electrophoresis studies (Comet Assay) in HCT116 p53+/+ cells treated with Δ‐[Fe2 L3a 3]Cl4 revealed an absence of single‐ or double‐ strand DNA breaks due to the lack of a "comet" tail (Figure 5 C). In addition, Δ‐[Fe2 L3a 3]Cl4 does not retard the formation of the "comets" in cells treated with DNA damaging peroxide, indicating that it does not form DNA cross‐links. The parent compound Δ‐[Fe2 L1 3]Cl4 behaves similarly.1b Thus if these metallohelices interact with DNA in the nucleus, they do not cause irreversible changes leading to cell death, as does cisplatin.31</p><p>Notwithstanding these findings, we compared the antiproliferative activity of these complexes in the pair of Chinese Hamster Ovary Cell lines CHO‐K1 and MMC‐2 (Table 1); a system previously used to identify the DNA damage involvement of cytotoxic agents. MMC‐2 is a CHO‐K1 mutant carrying the ERCC3/XPB mutation, which renders this cell line deficient in DNA nucleotide excision repair (NER).32</p><!><p>Antiproliferative data (IC50) determined by MTT test for CHO‐K1 (wild‐type) and MMC‐2 (NER‐deficient).[a]</p><p>Compound</p><p>CHO‐K1</p><p>MMC‐2</p><p>F[b]</p><p>Δ‐[Fe2 L1 3]Cl4</p><p>20±3</p><p>6.5±0.9</p><p>3.1</p><p>Δ‐[Fe2 L3a 3]Cl4</p><p>13±2</p><p>2.2±0.2</p><p>5.7</p><p>cisPt</p><p>25±4</p><p>2.6±0.4</p><p>9.7</p><p>[a] The treatment was 72 h. The results are expressed as mean values ± SD (μm) from three independent experiments (p<0.002). [b] F: the factor is defined as IC50 (NER efficient, CHO‐K1)/IC50 (NER‐deficient, MMC‐2).</p><!><p>The factor F (Table 1), which compares IC50 for Chinese Hamster Ovary cells (wild type) and the NER deficient system, is rather lower for Δ‐[Fe2 L1 3]Cl4 and Δ‐[Fe2 L3a 3]Cl4 than it is for the DNA damaging agent cisplatin, but there is a three or six‐fold difference between the response of the two cell lines; this prompted us to study DNA interactions in vitro (below). We further compared the antiproliferative activity of Δ‐[Fe2 L1 3]Cl4 and Δ‐[Fe2 L3a 3]Cl4 against A2780 ovarian cancer cells, and the cisplatin‐resistant strain A2780cisR (Table 2). No cross‐resistance with cisplatin was detected. We also compared the response of p53‐deficient and wild type HCT116 cells. Whilst p53‐deficient cells were less responsive to cisplatin, there was no significant difference between the response of HCT116 p53+/+ and p53−/− cells (p>0.05) in the case of Δ‐[Fe2 L3a 3]Cl4, with Δ‐[Fe2 L1 3]Cl4 demonstrating significantly (p<0.01) enhanced activity against p53 deficient cells. Together these data are consistent with both Δ‐[Fe2 L1 3]Cl4 and Δ‐[Fe2 L3a 3]Cl4 inducing their antiproliferative effects on the cells via a different mechanism to cisplatin, whilst indicating a non‐destructive interaction with DNA, more so for Δ‐[Fe2 L3a 3]Cl4.</p><!><p>Antiproliferative activity data (IC50) determined by MTT test for A2780 (wild‐type), A2780cisR, HCT116 (wild‐type, p53+/+) and HCT116 p35−/−.</p><p>Cell line</p><p>Δ‐[Fe2 L1 3]Cl4</p><p>Δ‐[Fe2 L3a 3]Cl4</p><p>cisPt</p><p>A2780[a]</p><p>15±3</p><p>1.4±0.3</p><p>3.3±0.2</p><p>A2780cisR[a]</p><p>13±3</p><p>1.2±0.1</p><p>20±3</p><p>HCT116 p53+/+  [b]</p><p>21±1[c]</p><p>7±1</p><p>3.3±0.4[c]</p><p>HCT 116 p53−/−  [b]</p><p>8±4[c]</p><p>11±2</p><p>7.5±0.2[c]</p><p>[a] The drug exposure time was 72 h. [b] The drug exposure time was 96 h. [c] Data previously published in reference 1d. The results are expressed as IC50 mean values ± SD (μm) from three independent experiments.</p><!><p>Given the above observations, we investigated the in vitro DNA‐binding of Δ‐[Fe2 L1 3]Cl4 and Δ‐[Fe2 L3a 3]Cl4 via a fluorescence competition assay.33 The behavior was very similar for both compounds (Figure S19) with log K app=6.3±0.1 and 6.1±0.1 for Δ‐[Fe2 L1 3]Cl4 and Δ‐[Fe2 L3a 3]Cl4 respectively. Thus DNA‐binding affinity is not responsible for the higher accumulation of Δ‐[Fe2 L3a 3]Cl4 in the nucleus.</p><p>Further, linear dichroism (LD) studies indicate that the complexes bind to naked calf thymus DNA in a specific orientation, probably the major groove.1a These results, alongside the negative comet assays suggest that the DNA interactions are non‐covalent, and probably reversible, akin to those of peptide α‐helices and zinc fingers.34, 35</p><!><p>Based on their potency and selectivity, Λ‐[Fe2 L1 3]Cl4 and Δ‐[Fe2 L3a 3]Cl4 were selected for initial in vivo evaluation. They were administered as a single intravenous (IV) injection in HCT116 p53−/− bearing athymic nude mice. Prior to these studies, the maximum tolerated dose (MTD) was determined for both compounds; the glucose‐appended metallohelix Δ‐[Fe2 L3a 3]Cl4 (MTD=1.75 mg kg−1) was far better tolerated than Λ‐[Fe2 L1 3]Cl4 (MTD=0.3 mg kg−1). Statistically significant tumour growth delays compared to the negative control group was seen for both compounds Λ‐[Fe2 L1 3]Cl4 (p<0.05), Δ‐[Fe2 L3a 3]Cl4 (p<0.01). A single injection of the parent triplex system Λ‐[Fe2 L1 3]Cl4 inhibited the tumour growth by 2.6 d, whereas the glycosylated metallohelix Δ‐[Fe2 L3a 3]Cl4 led to a growth delay of 4.3 d, that is, very similar to the clinical drug agent cisplatin (Table 3, Figure 6). Importantly, no weight loss effects were observed following treatment with Λ‐[Fe2 L1 3]Cl4 or of Δ‐[Fe2 L3a 3]Cl4, whereas cisplatin induced a showed 6 % loss of body weight in the first day following injection.</p><!><p>In vivo tumour studies. Tumour growth (top) and relative body weight (bottom) curves for [HCT116 p53−/‐]‐tumour‐bearing mice, administered with either nothing (control), 6 mg kg−1 cisplatin (positive control), 0.3 mg kg−1 Λ‐[Fe2 L1 3]Cl4, or 1.75 mg kg−1 Δ‐[Fe2 L3a 3]Cl4. Mice were administrated with a single dose on day 0 by intravenous injection. Mean relative tumour volumes (A) and Mean relative bodyweight B) were measured at different time points, plotted, and expressed with ± standard error; the significance p value <0.01 was considered to be statistically significant (n=8).</p><p>Efficacy study results.</p><p>Group</p><p>Relative tumour</p><p>doubling time</p><p>(days)</p><p>Growth delay</p><p>(days)</p><p>Significance</p><p>Maximum % weight</p><p>loss (day)</p><p>Λ‐[Fe2 L1 3]Cl4</p><p>(0.3 mg kg−1)</p><p>6.8</p><p>2.6</p><p>p<0.05</p><p>0</p><p></p><p></p><p></p><p></p><p></p><p>Δ‐[Fe2 L3a 3]Cl4</p><p>(1.75 mg kg−1)</p><p>8.5</p><p>4.3</p><p>p<0.01</p><p>0</p><p></p><p></p><p></p><p></p><p></p><p>Cisplatin</p><p>(6 mg kg−1)</p><p>8.9</p><p>4.7</p><p>p<0.01</p><p>6.0 (2)</p><p></p><p></p><p></p><p></p><p></p><p>Untreated</p><p>controls</p><p>4.2</p><p>–</p><p>–</p><p>2.0 (6)</p><!><p>We have developed a very efficient method for the conjugation of triplex metallohelices with sugar units. The highly complex products have amphipathic structures, are optically pure, water‐soluble, and extremely stable in water and biological media.</p><p>The addition of the carbohydrate units leads to substantial changes in the antiproliferative activity. Most strikingly, for the Δ‐configured (right‐handed helix) compounds, the apparent selectivity for cancer cells is greatly increased. In a mouse model, the drug tolerance and effect, as measured by MTD and tumour growth delay, are substantially improved versus the parent system. Encouragingly, no weight loss was recorded in the subjects following the dose.</p><p>The triplex metallohelix system is also shown to be a rare example of a class of DNA‐binding/aligning metallohelix. The parent and glycosylated compounds bind and align with DNA with very similar strength, thus validating our structural strategy of appending these polar units to the hydrophilic face of the helix, leaving the relatively hydrophobic ridge unperturbed.</p><p>In mechanistic terms, the addition of the glucose units leads to drug‐like dose‐dependent cell cycle effects, and the response observed in the cell cycle differs significantly between diastereoisomers of the metallohelices. Further, while the glucose derivative was found to be the most selective for the chosen cancer cell system, we conclude that this is not due to GLUT receptor targeting. Indeed, the cellular uptake is actually attenuated by addition of the sugars. Interestingly however, intranuclear transport is overall increased, perhaps by a sugar‐mediated process.30c Notably, the intranuclear transport, and the presumed DNA binding events in cellulo, do not lead to DNA damage.</p><p>Overall it would appear that the modification of triplex metallohelices in this way is worthy of investigation as a strategy for improvement of targeting and efficacy in this system, just as it is for the natural α‐helical systems. Also, we can add this behaviour to a growing list of evidences that this class of molecule, with its many variants, share features with cationic antimicrobial and anticancer peptides.</p><p>In vivo evaluation was performed under contract at the Institute of Cancer Therapeutics UK under Home Office licence PPL 40/3670. Local ethical approval was obtained on 07 April 2016 by the Animal Welfare and Ethical Review Body (AWERB) of the University of Warwick (reference AWERB.26/15‐16).</p><!><p>The authors declare no conflict of interest.</p><!><p>As a service to our authors and readers, this journal provides supporting information supplied by the authors. Such materials are peer reviewed and may be re‐organized for online delivery, but are not copy‐edited or typeset. Technical support issues arising from supporting information (other than missing files) should be addressed to the authors.</p><p>Supplementary</p><p>Click here for additional data file.</p>

PubMed Open Access

A texaphyrin\xe2\x80\x93oxaliplatin conjugate that overcomes both pharmacologic and molecular mechanisms of cisplatin resistance in cancer cells\xe2\x80\xa0

A texaphyrin\xe2\x80\x93oxaliplatin conjugate, oxaliTEX, was designed to test the concept that a platinum analog can overcome defects in drug accumulation and p53-dependent DNA damage response in a tumor model expressing multifactorial mechanisms of cisplatin resistance. Cytotoxic studies resulted in a resistance factor of only 1.2, which essentially indicated complete reversal of resistance in 2780CP cells expressing a factor of 22 with cisplatin. Unlike cisplatin, oxaliTEX accumulated and formed DNA adducts, stabilized and activated p53 at similar levels in both sensitive and resistant cells, and induced apoptosis in both models. The ability and importance of a designer drug, such as oxaliTEX, to overcome cisplatin resistance by targeting two dominant resistance mechanisms is discussed.

a_texaphyrin\xe2\x80\x93oxaliplatin_conjugate_that_overcomes_both_pharmacologic_and_molecular_mechan

Introduction<!>Conjugate design<!>Cytotoxicity, comparison of resistance factors, and reactivity with plasma proteins<!>Intracellular platinum uptake and DNA\xe2\x80\x93Pt lesion quantification<!>FACS analysis<!>Western blot<!>Discussion<!>Conclusion

<p>Cisplatin and carboplatin (Fig. 1) are two of the most important antitumor agents in the clinic. They have demonstrated significant activities against several cancers and are critical components of front line therapeutic regimens in medical oncology.1 These drugs are curative against testicular cancer, but their clinical utility against other cancers remains limited by intrinsic or acquired resistance, which in the case of NSCLC, mesothelioma, and ovarian cancer is reflected in low 5-year survival rates of only 5–20%.2 The biochemical and pharmacologic mechanisms of platinum resistance (e.g., decreased drug uptake, increased glutathione levels, and increased DNA adduct repair) are well established.1 These mechanisms co-exist with molecular mechanisms of resistance and present treatment-limiting barriers to the effective application of these two platinum agents in the clinic.</p><p>Among molecular mechanisms inducing cisplatin or carboplatin resistance, the most formidable involves loss of function of the tumor suppressor p53, often through missense gene mutation.3,4 However, over a half of advanced cancers (e.g., ovarian cancer, NSCLC and mesothelioma) that retain wild-type p53 are also resistant to cisplatin, and this is ascribed to loss of p53 regulation.3–5 Data from our group has established that Pt analogs, such as DAP [1,2-diaminocyclohexane-diacetato-dichloro-Pt(IV)], can restore stabilization and activation of wild-type p53 and effectively circumvent the molecular mechanism of resistance.5 This has led us to propose that oxaliplatin (Fig. 1), which contains the same diaminocyclohexane (DACH) ligand found in DAP, should also activate p53 function in resistant cells.</p><p>Unfortunately, restoring p53 function solves only part of the problem associated with clinical Pt resistance. Among the most important of the biochemical/pharmacologic mechanisms of resistance is reduced drug accumulation. However, finding an effective means to overcome resistance that occurs as a result of diminished Pt drug accumulation has proved elusive. Approaches that have been tested to address this limitation include conjugation to carrier molecules, such as folate, poly-(ethyleneglycol) (PEG), porphyrins, peptides, and nanoparticles.6–13 While in selected cases increased delivery of a Pt drug was achieved, the drug in question did not circumvent other mechanisms of resistance (including diminished p53 function). For an effective strategy, therefore, it is necessary not just to deliver the increased Pt payload to resistant tumor cells, but the Pt drug must have the properties needed to also overcome multifactorial mechanisms of resistance, particularly those associated with reduced p53 function, thereby restoring its dominant apoptotic activity.</p><p>Texaphyrins are unique tumor-localizing agents that are ideally suited for creating Pt conjugates.14 Motexafin gadolinium (1, MGd, cf. Fig. 1) has been extensively studied as a water soluble gadolinium(III) texaphyrin complex, and its ability to localize within cancerous lesions in vivo makes it an attractive choice for the design of conjugates to deliver a Pt payload.15–17 Herein, we report the design, synthesis, and biological evaluation of a novel texaphyrin–oxaliplatin like conjugate, and demonstrate as proof-of-principle that this conjugate is capable of overcoming both pharmacologic and molecular mechanisms associated with cisplatin resistance.</p><!><p>We have recently reported the synthesis of a texaphyrin–platinum conjugate (2, cisTEX, cf. Fig. 1) that produced equivalent levels of intracellular Pt and DNA–Pt adducts in both sensitive A2780 and isogenic multifactorial-resistant 2780CP cells upon exposure to conjugate 2.18,19 This conjugate was designed to form qualitatively identical adducts to those induced by cisplatin or carboplatin.18 Consequently, it displayed potent cytotoxicity in the A2780 human ovarian tumor model. However, its activity in the ~20-fold resistant cell line was improved by only a modest amount (i.e., a reduction in the resistance factor to ~10-fold).19 Thus, cisTEX is able to overcome the pharmacologic transport-related mechanism of resistance but is unable to modulate the molecular mechanisms of resistance. On the other hand, we have demonstrated that the DACH-containing analog DAP was potently effective at circumventing molecular mechanisms of resistance, although in this latter case residual resistance from defects in drug uptake remained.20 This has led us to consider that combining both approaches might be particularly effective. Specifically, we postulated that incorporation of a DACH–platinum within our conjugate design would provide a texaphyrin–platinum congener (3, oxaliTEX, cf. Fig. 1) that would release the DACH–Pt moiety within the cell and circumvent both pharmacologic and molecular mechanisms of resistance. The present study was designed to test this hypothesis.</p><!><p>Proof-of-principle insights into whether a particular conjugate design has the potential to circumvent cisplatin resistance may be garnered from the relative activities of the conjugate in a sensitive/resistant pair of intimately related cell lines. Therefore, we compared the activity of conjugate 3 in the A2780 human ovarian cancer model and its ostensibly isogenic cisplatin-resistant 2780CP model.</p><p>Cytotoxic studies were conducted following exposure of tumor cells to the Pt agents, and the activity of conjugate 3 (oxaliTEX) in inhibiting cell proliferation was compared to that of conjugate 2 (cisTEX), oxaliplatin, and cisplatin. Typical dose–response curves are shown for conjugates 2 and 3 in Fig. 2. The two curves for conjugate 2 in A2780 and 2780CP models are substantially separated, with the curve for 2780CP cells shifted to higher concentrations, which is an indication of resistance to this particular conjugate in the case of these resistant tumor cells. In contrast, the dose–response curves for conjugate 3 in the two models are almost superimposable. This finding is taken as evidence that this newly designed conjugate is equi-effective against both sensitive and resistant tumor cells and a strong indication that it can fully circumvent resistance in 2780CP cells.</p><p>The IC50 values generated from these curves are presented in Table 1. This table also includes values for oxaliplatin and cisplatin for the sake of comparison, together with associated resistance factors in the 2780CP cell line relative to parent A2780 cell line. Conjugate 3 (IC50 = 0.55 ± 0.06 μM) provided a dose potency in the A2780 cell line that was nearly 3-fold greater than conjugate 2 (IC50 = 1.63 ± 0.2 μM). The FDA-approved platinum complexes oxaliplatin and cisplatin provided IC50 values of 0.15 ± 0.05 and 0.33 ± 0.02, respectively. Against 2780CP cells, oxaliTEX (3) and oxaliplatin maintained their potent activities, with IC50 values of 0.65 ± 0.09 and 0.30 ± 0.05 μM, respectively. In contrast, our first generation conjugate 2 and cisplatin produced significantly greater IC50 values of 17.0 ± 1.5 and 7.3 ± 0.2 μM, respectively, values that reflect a 11- to 26-fold lower potency relative to conjugate 3.</p><p>The above results lead us to suggest that conjugate 3 may have a role to play in overcoming the resistance observed in cisplatin resistant wild type p53 ovarian cancer. Although conjugate 3 displays lower potency (higher IC50) than oxaliplatin in the A2780 and 2780CP cell lines, this does not negate its clinical potential. This key point is underscored by carboplatin, which has substantially lower potency than cisplatin (1.6 vs. 0.31 μM in A2780 human ovarian cancer cell line),19 but nevertheless enjoys widespread clinical use.21 The fact that conjugate 3 also overcomes resistance pathways makes it attractive for further study.</p><p>A better understanding of relative activity of complexes against resistant cells can be garnered from the resistance factor, which is a ratio of IC50 of an agent in sensitive and resistant cells (Table 1). As expected, 2780CP cells demonstrated 22.1-fold resistance to cisplatin22 and a lower 10.4-fold cross-resistance to conjugate 2. In contrast, 2780CP cells were only 2-fold resistant to oxaliplatin, but were almost devoid of cross-resistance to conjugate 3 (cross-resistance factor, 1.2). This is consistent with essentially complete circumvention of resistance.</p><p>The high potencies of conjugate 3 in the A2780 and 2780CP cell lines, as shown in Table 1, is consistent with the design expectation that the conjugate is labile and rapidly converts to a reactive product that is able to interact covalently with cellular macromolecules, including DNA. Based on the underlying chemistry this conversion was expected to involve release of a free DACH–Pt aquo species. To determine the activation rate of conjugate 3 relative to cisplatin and oxaliplatin, samples of fetal bovine serum (FBS) were incubated with the Pt-based agents and the amount of free (unbound) Pt was determined over time. It is presumed that the irreversible reaction of platinum with FBS results in inactivation of the platinum drug. The half-life (t1/2) of the initial logarithmic decay phase was determined by linear regression analyses (Fig. 3). Cisplatin and oxaliplatin reacted rapidly with serum proteins, and this corresponded to an accelerated decrease in free Pt (t1/2 = 5.1–5.3 hours). Conjugate 3 did become activated but its binding to proteins was comparatively slower (t1/2 = 8.3 hours). However, conjugate was found to react more quickly than the clinically approved agent carboplatin (t1/2 = 36–45 hours) and conjugate 2 (t1/2 = 16 hours).19,23</p><!><p>Cisplatin resistance is multifactorial and is reflective of a variety of biochemical pharmacologic mechanisms, including reductions in cellular uptake and retention, decreased DNA adduct formation and a greater tolerance for DNA damage.24–27 The minimal resistance factor obtained for conjugate 3 reflects an ability to overcome most forms of resistance present in the 2780CP model. To gain insights into the basis of this high activity against resistant cells, the intracellular uptake of 3 was analyzed, along with that of the control complexes oxaliplatin and cisplatin. The decrease (>50%) in cisplatin uptake in 2780CP cells relative to A2780 cells was readily apparent and statistically significant, as was the reduced accumulation of oxaliplatin in resistant cells (Fig. 4). Such findings are consistent with the reported 2-fold reduction in uptake seen in resistant cells exposed to platinum anticancer agents.28 Interestingly, no difference in uptake was observed for conjugate 3 between A2780 and 2780CP tumor cells. Relative to oxaliplatin, conjugate 3 demonstrated a ≥2- and ≥4-fold increase in platinum uptake for the A2780 and 2780CP cell lies, respectively. Additionally, it is notable that conjugate 3 also demonstrated a ≥2-fold higher uptake than cisplatin in 2780CP cells.</p><p>The reduction in platinum uptake in resistant cells generally translates into reduced Pt-adducts formed with DNA.29 This, however, is not always the case and is not seen with every platinum drug. Because this correlation is not certain, the Pt–DNA adducts formed by conjugate 3, as well as the Pt control complexes oxaliplatin and cisplatin, were quantified in both the A2780 and 2780CP cell lines. With cisplatin and oxaliplatin, a statistically significant 50% reduction in Pt–DNA adducts was seen in the 2780CP cells (Fig. 5). This is considered to reflect the corresponding differences in drug uptake noted in Fig. 4, and is fully consistent with what was expected for these known agents. Also in accord with expectations with oxaliplatin,30 the observed adduct levels of conjugate 3 in A2780 cells were lower than those seen for cisplatin, even though the uptake of these two platinum species is comparable (see Fig. 4). This is attributed to the differences in Pt coordination provided by the two species in question. Finally, and in contrast to cisplatin and oxaliplatin, the adduct levels provided by conjugate 3 were not reduced in resistant cells (Fig. 5).</p><p>It is evident from an inspection of Fig. 5 and Table 1 that adduct levels may not reflect the potency of the molecule. This is clearly demonstrated by the fact that oxaliplatin adducts are formed at a lower level than those of cisplatin in sensitive A2780 cells, and yet, oxaliplatin is the more potent drug. To generalize this finding and expand it to conjugate 3 in both sensitive and resistant cells, the DNA damage tolerance ratio was calculated. This tolerance is determined from the level of adducts in Fig. 5 and extrapolating to the IC50 concentration; that is, the level of adducts required to kill 50% of tumor cells. The DNA damage tolerance data in Fig. 6 demonstrate that resistant 2780CP cells require 10-fold greater adduct levels as compared to A2780 cells for equivalent antiproliferative activity. As reported prevously,19 cisplatin resistant 2780CP cells were capable of similarly tolerating ca. 10-fold the number of Pt–DNA lesions formed by conjugate 2. Conjugate 3 or oxaliplatin, on the other hand, induced cytotoxicity in the 2780CP model at substantially lower adduct levels.</p><!><p>Apoptosis is a desirable endpoint for antitumor agents in cancer chemotherapy. To determine whether such a feature was inherent to conjugate 3, and also to ascertain the underlying basis for its antiproliferative activity in Table 1, we conducted apoptotic studies using FACS analysis to detect Annexin V-positive cells. We included the parent drug oxaliplatin as a positive control and MGd (1) as a model for the Pt-free conjugating texaphyrin moiety present in conjugate 3. A2780 and 2780CP cells were exposed to these agents for 24, 72, and 120 hours and the fraction of apoptotic cells determined. The findings are presented in Fig. 7. The data in Fig. 7A demonstrate that the 14–16% of drug-induced apoptotic A2780 cells at 24 hours were not significantly different from controls (12%). No gross changes were observed at 72 and 120 hours in Annexin V-positive control cells or those treated with MGd (1). In contrast, the level of apoptotic cells following exposure to conjugate 3 and oxaliplatin increased 2- to 4-fold with time at 72 (~40%) and 120 (55–60%) hours. Similar temporal profiles for apoptosis were observed with the cisplatin-resistant 2780CP model. Such findings are consistent with the notion that 3 is capable of inducing a similar degree of apoptosis in both A2780 and 2780CP cell lines. Moreover, conjugate 3 is capable of inducing apoptosis to the same extent as oxaliplatin. Finally, the observed apoptosis is attributable to the Pt-bearing conjugate as a whole and not to the texaphyrin carrier.</p><!><p>Facile apoptosis following DNA damage, as observed with conjugate 3 and oxaliplatin in Fig. 7, can occur in a p53-dependent or an independent manner. Both A2780 and 2780CP models have wild-type p53 function, but only in A2780 cells can p53 be induced and activated by cisplatin through post-translational modification events via, e.g., Ser-15 phosphorylation.22,31 To investigate the effect of conjugate 3 and oxaliplatin on p53 induction and Ser-15 phosphorylation, A2780 and 2780CP cells were exposed to these agents at concentrations equal to 2.5× and/or 5× the IC50 values for 6-, 12-, and 24 hours. At these time points, cells were collected, washed with cold PBS, and cell lysates prepared for subsequent analysis by Western blotting. Equal protein loading was confirmed by β-actin immunoblots. The protein bands in immunoblots were quantified by densitometry and the data presented in Fig. 8. Induction of p53 in A2780 cells increased 2-fold (relative to basal p53 levels) upon exposure to 3 and oxaliplatin and peaked after 12 hours of exposure (Fig. 8A and C). A corresponding increase in Ser-15-p53 over the time course was also evident. The exposure of cisplatin resistant 2780CP cells to these two Pt-complexes led to similar increases in p53 and Ser-15-p53, with total p53 increasing progressively to over 2-fold at 24 hours (Fig. 8B and D).</p><p>Induction of p53 and its functional activation are separate events. It was therefore necessary to assess p53 function by monitoring the transcriptional activation of p21 as a downstream target of p53. Both conjugate 3 and oxaliplaitin induced p21, which increased in a time-dependent manner to 3- to 5-fold above controls by 24 hours (Fig. 9). Such a finding supports the conclusion that the p53–p21 pathway is similarly upregulated by conjugate 3 and oxaliplatin and that this occurs in both the Pt-sensitive A2780 and Pt-resistant 2780CP cell lines.</p><!><p>While active in several cancer types and included in front line therapy by oncologists, platinum anticancer agents display acquired resistance in many cancers, which limits their clinical utility. The cause of this resistance is multifactorial and includes both pharmacologic mechanisms (e.g., decreased drug uptake, increased glutathione, and increased DNA adduct repair) and molecular mechanisms of resistance (e.g., loss of p53 function, increase in survivin, and increase in Bcl2).1,3,4 In this study, we have focused on two major cisplatin resistance mechanisms, reduced drug uptake and attenuated wild-type p53 function. Specifically, we sought to target these mechanisms via a novel platinum drug design. With this goal in mind, we designed conjugate 3 (oxaliTEX). As detailed below, this new agent not only circumvents these two mechanisms, it also demonstrates antiproliferative activity in a multifactorial-resistant in vitro tumor model (the Pt-resistant 2780CP ovarian cancer cell line) that makes the cells in question as responsive as the corresponding Pt-sensitive parental (A2780) cell line to this DNA-damaging agent.</p><p>The design of oxaliTEX was based on literature data and our earlier studies involving conjugate 2. The focus on targeting the tumor suppressor p53 derived from an appreciation that cisplatin has a greater curative rate in ovarian cancer when p53 is present in its wild-type state than in the mutant form.3,4 Paradoxically, about a half of advanced ovarian cancers that harbor wild-type p53 are resistant, primarily as a result of failure of upstream DNA damage signaling to stabilize and activate the p53. Furthermore, in these resistant cancers, the presence of wild-type p53 can lead to a "gain-of-resistance" phenotype, where the resistance is greater than those with mutant p53.3,4 Thus, loss of function of wild-type p53 is one of the most formidable molecular mechanism of resistance. However, we have reported that a panel of resistant ovarian tumor models respond to DACH-based platinum drug (e.g., DAP) through distinctly different DNA damage signaling processes that serve to restore p53 function and cellular apoptotic activity.20,32,33 Such a restoration of activity was considered likely to hold in the case of DACH-based oxaliplatin, and was specifically confirmed in the present study using the resistant 2780CP cell line.</p><p>Unfortunately, circumvention of resistance by DAP or oxaliplatin is not complete and at least a 2-fold level of resistance remains in the test 2780CP cell line, as seen with oxaliplatin in the present study. This residual resistance is ascribed to resistant cells displaying reduced accumulation of DAP or oxaliplatin.20,34 On the other hand, we have previously demonstrated in an independent study with conjugate 2, which has a cisplatin-like diammine-Pt configuration around the Pt(II) center, that Pt accumulation in 2780CP cells was not diminished, and that this translated into a concomitant 2-fold reduction in cross-resistance.19 Collectively, these previous investigations provided support for the hypothesis that conjugating texaphyrin with a DACH–Pt structure, such as that present in DAP and oxaliplatin, would serve to circumvent both pharmacologic and molecular mechanisms of resistance. Our present study, involving the new conjugate 3 (oxaliTEX) provides proof-of-principle support that this goal can be accomplished in vitro.</p><p>That activation of wild-type p53 is sufficient to overcome multifactorial molecular mechanisms of resistance is intriguing. Normally, wild-type p53 plays a critical role in drug-induced apoptosis. However, this activity becomes compromised when p53 is mutated, which leads to cisplatin/carboplatin resistance and, in the specific case of advanced ovarian cancer for which statistics are available, a 4- to 5-fold reduction in the 5-year survival rate compared to the wild-type p53 cancer sub-group.3,4 Advanced cancers other than ovarian cancer (e.g., NSCLC and mesothelioma) that retain wild-type p53 also demonstrate resistance to cisplatin,4 an observation ascribed to a number of mechanisms, including the critical post-translational modifications of p53 to release p53 from its inhibitory interaction with Mdm2.35,36 Based on reports from molecularly engineered mouse models,37 it appears that activation of wild-type p53 for apoptosis is a dominant DNA-damaging effect, and is sufficient to override the potential negative influence of other molecular defects that may co-exist in multifactorial resistant tumor cells. The 2780CP tumor cells used as a model for Pt resistance in ovarian cancer have been characterized as having a multifactorial cisplatin-resistant phenotype.20 The ability of oxaliTEX to restore platinum sensitivity in this model highlights the potential of this agent to enhance in due course both the initial response and the long-term cure rate in patients with cancers harboring wild-type p53.3,4</p><p>In addition to circumventing molecular mechanisms of resistance, we have demonstrated that oxaliTEX is capable of delivering the DACH–Pt payload at similar levels in both sensitive and resistant tumor cells. This was ascertained in the present study from examinations of both total intracellular Pt and DNA adduct levels. The similar delivery of Pt is likely due to the of the expanded porphyrin, texaphyrin, which has been shown to localize selectively within tumors.14–17 That the effective delivery of Pt is due to the conjugating texaphyrin carrier and not the DACH–Pt moiety can be inferred from the knowledge that uptake and DNA adduct data with oxaliTEX (conjugate 3) mirror those reported by us for cisTEX (conjugate 2), which has an alternate diamine-Pt coordination environment.19</p><p>It is further apparent from apoptotic investigations using Annexin V as a biomarker in the present study that the texaphyrin control, MGd (1), is devoid of antiproliferative effects at concentrations that were equivalent to those employed in the studies of oxaliTEX. This is taken as evidence that the cellular effects are predominantly due to the Pt moiety and that the texaphyrin is functioning primarily as a delivery vehicle. Support for this conclusion comes from noting the close similarities between oxaliTEX and oxaliplatin in the level of adducts formed, the phosphorylation of p53, the induction of the p53/p21 pathway, and the extent and temporal profile of apoptosis in both sensitive and resistant tumor cells. However, there is an apparent difference in drug potency between oxaliTEX and oxaliplatin in both the A2780 and 2780CP cell lines. This difference likely reflects the slow release of the DACH–Pt moiety from the conjugate, which serves to increase the time for Pt activation by ~50%.</p><!><p>The ability of oxaliTEX and oxaliplatin in the present study, and of DAP in an earlier study,20 to activate p53 stands in direct contrast to our previous finding that cisplatin fails to activate the p53/p21 pathway in 2780CP cells.20,31 The ability to upregulate p53 function in resistant cells augurs well for the eventual application of 3 in treating refractory cancers that harbor wild-type p53. The ability to enhance the uptake of a DACH–Pt payload into resistant 2780CP ovarian cancer cells distinguishes oxaliTEX (3) from other Pt delivery strategies, and identifies this conjugate as the first small molecule platinum complex capable of overcoming both pharmacologic and molecular mechanisms of resistance in vitro.</p><p>It is normally presumed that success of chemotherapy in resistant disease may require addressing each of the many mechanisms of resistance. However, it is evident from the present study that targeting only a few critical cellular impediments with a designer drug may be sufficient to achieve a markedly improved therapeutic response. The fact that texaphyrin is known to localize selectively to tumors, leads us to predict that suitably designed conjugates, such as 3, may prove effective in treating resistant cancers. Further tests of this paradigm, including in vivo studies to determine the ability of 3 to selectively localize to tumor, are currently in progress.</p>

PubMed Author Manuscript

Identification of a New Allosteric Binding Site for Cocaine in Dopamine Transporter

Dopamine (DA) transporter (DAT) is a major target for psychostimulant drugs of abuse such as cocaine that competitively binds to DAT, inhibits DA reuptake, and consequently increases synaptic DA levels. In addition to the central binding site inside DAT, the available experimental evidence suggests the existence of alternative binding sites on DAT, but detection and characterization of these sites are challenging by experiments alone. Here we integrate multiple computational approaches to probe the potential binding sites on the wild-type Drosophila melanogaster DAT and identify a new allosteric site that displays high-affinity for cocaine. This site is located on the surface of DAT, and binding of cocaine is primarily dominated by interactions with hydrophobic residues surrounding the site. We show that cocaine binding to this new site allosterically reduces the binding of DA/cocaine to the central binding pocket, and simultaneous binding of two cocaine molecules to a single DAT seems infeasible. Furthermore, we find that binding of cocaine to this site stabilizes the conformation of DAT but alters the conformational population and thereby reduces the accessibility by DA, providing molecular insights into the inhibitory mechanism of cocaine. In addition, our results indicate that the conformations induced by cocaine binding to this site may be relevant to the oligomerization of DAT, highlighting a potential role of this new site in modulating the function of DAT.

identification_of_a_new_allosteric_binding_site_for_cocaine_in_dopamine_transporter

Introduction<!>Modeling of the Wild-Type dDAT<!>Molecular Docking<!>MD Simulations<!>Conformational Energy and Binding Energy<!>Cocaine and DA display different docking patterns.<!>Cocaine binds to a new site S3 on dDAT in an outward-open conformation.<!>S3 is a high-affinity binding site for cocaine and F122A mutation lowers cocaine binding affinity.<!>Binding of cocaine to S3 reduces binding of DA to dDAT.<!>Discussion

<p>Dopamine (DA) transporter (DAT) is a member of the neurotransmitter:sodium symporter (NSS) transporters, which plays an important role in regulating neurotransmission by sodium-driven reuptake of neurotransmitters from the synapse into the presynaptic neuron.1 The dysfunction of DAT and other related NSS members has been implicated in the neuropsychiatric disorders including depression, attention-deficit hyperactivity disorder, epilepsy, and Parkinson's disease.2, 3 The NSS transporters are also the primary targets for many psychostimulants such as cocaine and amphetamines.4 The architecture of the NSS involves 12 transmembrane (TM) helices, which displays significant sequence identities (~50%‒70%) among the eukaryotic NSS members but only about 20% sequence identity in the prokaryotic amino acid transporters such as the bacterial leucine transporter (LeuT).5 Despite the varying sequence identity, DAT shares a conserved fold as the other NSS transporters, containing two helical bundles, TMs 1‒5 and TMs 6‒10.5 DAT works through the proposed alternating access mechanism, in which the transporter switches between the outward-open and inward-open conformational states, to alternate the substrate access between the extracellular and intracellular sides of the membrane.6 In addition, the alternating accessibility may be related to the switching between alternative conformations of the two helical bundles (TMs 1‒5 and TMs 6‒10).6 The transition from the outward-open to the inward-open state may also involve the outward-occluded state.7</p><p>DAT efficiently removes neurotransmitter DA from the synaptic space.3 The widely abused psychostimulant cocaine is a competitive inhibitor of DAT and enhances extracellular DA concentrations by locking the transporter in an inactive conformation.8 One of the fundamental questions regarding the function of DAT is the identification of the binding sites for the substrate and the inhibitor as well. The X-ray crystal structures of the Drosophila melanogaster DAT (dDAT) in complex with DA and cocaine in an outward-open conformation show that DA and cocaine bind to the central binding site, surrounded by TMs 1, 3, 6, and 8.8 This observation confirms that the binding site for DA and cocaine in DAT overlaps, a finding obtained based on the structural model of bacterial homologue LeuT.9 In addition to the primary binding site, a secondary binding site for leucine located in the extracellular vestibule (~11 Å above the primary binding site) has been suggested for LeuT.10 This secondary binding site overlaps with the binding site of tricyclic antidepressant clomipramine in the cocrystal structure of LeuT where the substrate leucine occupies the primary binding site.11 In addition to clomipramine, binding of a tryptophan,12 a sertraline,13 and an octylglucoside detergent molecule14 in the secondary site of LeuT was also observed in the crystal structures. Like DAT and LeuT, the serotonin transporter (hSERT), and multi-hydrophobic amino acid (MhsT) transporters are also NSS homologues. The binding of S-citalopram molecules in both the primary and secondary sites of hSERT has been observed in the crystal structure,15 and the simultaneous binding of two tryptophan substrates in MhsT were also indicated.16 Although these evidence support the existence of a secondary substrate binding site in NSSs, further experiments using a variety of binding measurements,17 as well as solid-state nuclear magnetic resonance (NMR) technique18 suggested that only one leucine binds to the central/primary site of LeuT. This apparent controversy could be ascribed to the different experimental conditions—the binding of leucine in the secondary site can be impaired by the protein preparation procedures used for crystallography and for some functional studies.19, 20</p><p>The X-ray crystal structures of DA- and cocaine-bound dDAT suggest the presence of a single, high-affinity substrate/inhibitor binding site in dDAT.8 However, the existence of a secondary binding site in DAT has also been proposed. Using the X-ray crystal structure of the bacteria LeuT21 as a template, a human DAT structure (23% sequence identity) was homolog-modeled, and a second DA binding site surrounded by residues from TMs 1, 3, 10, and extracellular loops (ELs) 2 and 4 was identified by performing steered molecular dynamics (MD) simulations.22 In another computational study of human DAT, a model based on the crystal structure of dDAT (52% sequence identity)8 indicated that the DA transiently binds to the secondary site from the extracellular medium before entering into the primary binding site, but cocaine and other inhibitors were precluded from this secondary binding site.23 Interestingly, experimental study of human DAT mutants identified a phenylalanine-to-alanine mutant in TM3 F154A (F154 corresponds to F122 in dDAT) that significantly lowers cocaine affinity, implying that cocaine may not bind to the primary site as shown in the crystal structure of dDAT.24, 25 Another experimental study using mouse DAT indicated that F155Y (F155 corresponds to F123 of the dDAT) showd different effects on the binding of cocaine and cocaine analogs.26 Together, these mutation studies seem to support the presence of a binding site for cocaine in TM3. However, the exact location remains elusive. Moreover, the study of binding of bivalent compounds to DAT suggested the existence of multiple substrate binding sites in a single DAT.27 Identification of these potential binding sites is challenging by experiments alone considering different methods and conditions applied to characterize proteins.20 In addition, the NSS transporters have been shown to oligomerize at the plasma membrane,28, 29, 30 which may also affect the recognition of substrates/inhibitors by DAT.</p><p>In this study, we investigated the potential binding sites for cocaine on dDAT by performing molecular docking, MD simulations, and binding energy calculations. In addition to the primary binding site, our random docking results show different binding patterns between cocaine and DA. MD simulations of cocaine-bound dDAT complexes obtained from docking provide evidence for the binding of cocaine in a site formed by TMs 3, 9, 10, and 12 of dDAT. The predicted binding affinity of cocaine in this site is even higher than that in the central binding site, and the binding of cocaine in this site allosterically reduces the binding affinity of cocaine in the central site. Our results show that the dDAT mutant F122A lowers the binding affinity of cocaine in both binding sites, in agreement with previous experimental study.24 Contrary to the increased stability of dDAT when accommodating two DA molecules in the central and extracellular site simultaneously, the stability of dDAT decreases when two cocaine molecules bind to the central and the new site respectively, suggesting that the binding of cocaine to this new site significantly alters the conformation of dDAT. As a consequence, dDAT becomes less favorable for the binding of DA.</p><!><p>The X-ray structure of DA-bound dDAT (PDB ID: 4XP1)8 was used to construct the wild-type dDAT structure. Mutations in the crystal structure were reverted to the wild-type amino acids. The missing residues Ser162‒Val202 were not determined because of high flexibility, and modeled using MODELLER 9.23.31 After model building, the region Ser162‒Val202 in the best model was subjected to loop refinement by using the newer DOPE-based loop modeling protocol.32 The N-terminal (residues 1‒24) and C-terminal (residues 601‒645) were not added in the above modeling procedure, but were capped by an acetyl and amide group, respectively. The two sodium ions (Na+) and one chloride (Cl−) ion found in the crystal structure were kept in the final modeled structure. The best model after loop refinement was selected for subsequent molecular docking and simulations. The position of the DA in the crystal structure was used as the binding site of DA in the central binding pocket. The position of the cocaine in the model structure was determined by the alignment of cocaine-bound crystal structure (PDB ID: 4XP4) with the DA-bound crystal structure (PDB ID: 4XP1). These two structures are almost identical, with the root-mean-square deviation (RMSD) of heavy atoms being about 0.25 Å. The structures of DA and cocaine are positively charged in the quaternary ammonium motif.33</p><!><p>Molecular docking was carried out using Autodock 4.2 program.34 Instead of using the default partial charges for protein and ligand, the partial charges from the CHARMM36m force field parameters35 were applied to the structure of dDAT. The partial charges for the cocaine and DA molecules were derived using the RESP method36 (Fig. S1). In each docking, the structure of dDAT was treated as rigid but the ligand was considered as flexible. Random docking was performed, that is, the grid space used to define the searching space of a protein includes the whole surface of the protein, and the ligand (cocaine or DA) randomly searches the potential binding sites available to the protein. A longer search with 25 million evaluations was used to provide adequate searching. Lamarckian genetic algorithm37 was applied to search 100 ligand conformations in each docking experiment. Our docking results show that the above docking protocol is capable of predicting the binding conformation of DA and cocaine in the central site as found in the crystal structures (Fig. 1). The stability of the docked conformations is further evaluated by performing MD simulations.</p><!><p>To model the ligand-bound dDAT in membrane environment, the cholesteryl hemisuccinate molecule found at the interface among TMs 2, 7, and 11 of the crystal structure was modeled by a cholesterol (CHOL) molecule, together with another CHOL molecule (among TMs 1a, 5, and 7) and water molecules found in the crystal structure were included in the model dDAT structure. The binding conformation of cocaine and DA in the central site of dDAT was taken from the crystal structure, and other binding conformations were obtained from the docked conformations with the best predicted binding affinity using Autodock 4.2.34 The structure of dDAT was represented with the CHARMM36m force field parameters,35 and the CHARMM General Force Field (CGenFF)38 was used to generate force field parameters for the ligands, with the partial charges replaced by the RESP charges as used in docking studies. The above model of dDAT was embedded into a mixed lipid bilayer consisting of total 420 lipids (POPC:POPE:POPG:CHOL = 3:1:1:1).39 This lipid bilayer has been shown to be a suitable membrane mimic for the study of the conformational dynamics of dDAT.39 A different type of lipid bilayer consisting of 420 lipids (POPE:CHOL = 5: 1) was also used to reproduce the results of some systems. The orientation of the dDAT model relative to the membrane was defined by aligning to the dDAT structure (PDB ID: 4M48) in the Orientation of Protein in Membrane database.40, 41The CHARMM36 lipid force field parameters42 were used to represent all lipids. The system was solvated by adding a 20 Å thick water layer (TIP3P water molecules) below and above the lipid bilayer, and each system contains 0.15 M NaCl. The CHARMM-GUI web server43 was used to generate the starting structures and configuration files for MD simulations.44</p><p>MD simulations were performed using NAMD2.13 program.45 Each system was first subjected to energy minimization for 10, 000 steps, followed by six stages of equilibration with the harmonic constraints exerted on lipid head groups, protein and ligand heavy atoms. The force constant was gradually decreased from 50 to 0.5 kcal/(mol·Å2) during the equilibration procedures. The temperature was kept constant at 300 K using the Langevin dynamics method with a damping coefficient of 5 ps, the pressure was controlled at 1.0 bar by applying the Nosé-Hoover Langevin piston method.46 The use of flexible cell was enabled to allow the system to fluctuate independently in three dimensions, along with the ratio of the system cell in the x-y plane kept constant. A cutoff of 12 Å was applied for the van der Waals interactions and the long-range electrostatic interactions were treated using the PME method. The integration time step is 2 fs and the trajectory was saved every 10 ps. The production dynamics were performed at the constant temperature (300 K) and pressure (1.0 bar) without any positional constraints. The simulation time varies from 250 to 300 ns for each system after equilibration, and the last 50-ns trajectory of each system was used for analysis.</p><!><p>The relative stability of dDAT conformations (including two bound Na+ and one Cl− in the central binding pocket) was evaluated by the conformational energy, which was calculated using the Generalized Born using Molecular Volume (GBMV) implicit solvent model implemented in the CHARMM (v44b1) program.47 The single point energy was calculated after a 200-step minimization of each conformation using the GBMV II algorithm.48, 49 Other energy terms including bonded energy, van der Waals energy, electrostatic energy, and solvation energy were also obtained with the GB implicit solvent model. The block-average method was used to estimate the mean value (500 frames) and standard deviations.</p><p>The binding free energy of cocaine/DA to dDAT was calculated using the g_mmpbsa program,50 which implements the Molecular Mechanics Poisson–Boltzmann Surface Area (MM-PBSA) method51, 52 to predict the binding affinity for protein-ligand complex. The entropy contribution was not included in the current binding energy calculations. To be consistent, the block average method was used to estimate the mean value (1,000 frames) and standard deviations rather than applying the default bootstrap analysis method.</p><!><p>In the docking experiment, 100 conformers of each ligand were obtained and clustered according to their position on the structure of dDAT. The distribution of 100 docked conformations of cocaine and DA on dDAT is shown in Fig. S2. For each cluster, the docked conformer with the highest binding affinity was selected and shown in Fig. 1. Six distinct binding sites of cocaine on dDAT were predicted by docking analysis. The site 1 (S1) surrounded by TMs 1, 3, 6, and 8 corresponds to the central binding site of cocaine as observed in the cocaine-bound dDAT crystal structure.8 The S2 is exposed to the extracellular side and is about 12 Å (the distance of center of mass between two cocaine molecules in the S1 and S2, respectively, Fig. 1B) away from the S1. In S2, the docked cocaine interacts with Trp51, Gly385, Pro386, Phe471, His472, Asp475, Phe468, and Tyr547 (Fig. S3), similar to the proposed secondary substrate binding site in LeuT.10 Among the remaining binding sites, the hydrophobic binding pocket S3 is particularly interesting as the docked cocaine interacts with Phe122 in TM3, a residue relevant for the binding affinity of cocaine in a previous mutation study of DAT.24 In addition to Phe122, the cocaine in S3 also interacts with Phe118 and Tyr119 in TM3, Leu454 and Val458 in TM9, Phe470 and Leu474 in TM10, and Ile563 in TM12 (Fig. S4). Different from cocaine, results from docking of DA to dDAT show that DA mainly occupies the central and secondary binding sites, and the space between these two sites (Figs. 1B and S2B). The binding sites for DA in the extracellular loops were excluded because the protein was treated as rigid in docking, but in fact the loop regions are rather flexible during MD simulations. As a result, the extracellular loops are not expected to be able to accommodate DA stably.</p><!><p>The multiple cocaine binding sites predicted by docking need further validation by MD simulations because of the limitations in docking such as rigid receptor and fixed bond angles in the ligand, as well as the simplified scoring function based on empirical free energy of binding.53 MD simulations were performed on dDAT with the cocaine-bound conformations taken from docked complexes. Results from MD simulations show that if one cocaine initially binds to S2, S4, S5, and S6, the docked cocaine dissociates from these sites either quickly or slowly, depending on the specific site to which cocaine binds. Specifically, cocaine moves away quickly from S2 (within 2 ns) to extracellular loops, but is transiently trapped there for 350 ns before entering into solution (Fig. S5A). When a cocaine initially binds to S4, S5 or S6, it is observed to diffuse into solution after 4 ns, 30 ns, and 200 ns, respectively (Figs. S5B, S5C, and S5D). In contrast, the binding of cocaine to S3 is stable over the 300-ns MD simulations. Taken together, these results indicate that S3 could be an alternative binding site for cocaine on dDAT.</p><p>The effect of binding of cocaine to S3 on the conformation of dDAT was compared with the binding of cocaine to S1. It is found that the two Na+ ions, the Cl− ion, cocaine, and the cholesterol molecules taken from the crystal structure are stable in their binding sites over the 300-ns MD simulations. The RMSDs of the dDAT relative to the starting dDAT crystal structure are below 2.5 Å, suggesting that the cocaine-bound dDAT is stable in the lipid bilayer (Fig. 2A). The simultaneous binding of two cocaine molecules to S1 and S3 slightly enhances the conformational dynamics of the dDAT. The calculated root-mean-square fluctuations (RMSFs) reveal no significant structural dynamics in the TM domains of cocaine-bound dDAT (Fig. 2B), which display the lowest RMSF values. The large fluctuations in the EL2 provide a rationale for the truncation of segment from Ser162 to Val 202 in the crystal structure.</p><p>To further examine the influence of binding of cocaine to S3 on the conformational change of dDAT, we calculated the distance between two pairs of residues: Arg52-Glu475 and Tyr124-Phe319. The former pair is able to form a salt-bridge, and the latter aromatic pair serves as the extracellular gate. The binding of cocaine decreases the distances between residues Arg52 and Glu475 from the initial 11.5 Å to an average distance of 6.5 Å, 7.0 Å, and 4.6 Å for cocaine in S1, S3, and in both S1 and S3, respectively, suggesting that the formation of the salt-bridge is feasible only when cocaine occupies S1 and S3 on dDAT simultaneously. However, this salt-bridge seems unstable as the distance between Arg52 and Glu475 fluctuates significantly (Fig. 2C). On the other hand, the distance between the gating residues Tyr124 and Phe319 decreases slightly from 14.4 Å to an average value of 13.7 Å, 12.3 Å, and 13.8 Å for cocaine in S1, S3, and in both S1 and S3, respectively (Fig. 2D), implying that the structure of dDAT remains in the outward-open state and the binding of cocaine may not trigger the conformational transition during MD simulations.</p><p>For comparison, we also simulated ligand-free and DA-bound dDAT under the same conditions. The structure of dDAT remains stable in the simulations of one DA in S1 and two DAs in both S1 and S2 initially. In fact, it appears that the structural dynamics of dDAT decrease slightly with the increasing number of bound DA molecules (Fig. S6), suggesting an overall increase in the stability of dDAT upon DA binding. Similar to cocaine-bound systems, the DA-bound dDAT also remains in the outward-open conformation. Different from the dynamics of cocaine in S2 where cocaine escapes from S2 and dissociates from dDAT eventually (Fig. S5A), the DA in S2 moves closer to the central pocket and is trapped in a location above S1 (Fig. S7) over the 250-ns MD simulations. The time scale required for the DA to reorient and occupy the same binding site as observed in the crystal structure seems far beyond the time scale of simulations presented here. Interestingly, when one DA binds to S1, a second DA bound to S2 initially was also observed to move toward S1 and interact with the DA bound to S1 (Fig. S8, and discussion below). To test if this phenomenon is independent of simulation conditions, we inserted the DA-bound dDAT into a different lipid bilayer consisting of POPE and CHOL (POPE:CHOL = 5: 1). DA was also found to diffuse toward S1 in the absence or presence of another DA in S1 (data not shown). Collectively, these results imply that S2 may not provide a high-affinity binding site for DA, but most likely act as a pathway connecting S1 and the extracellular space.</p><!><p>The interacting residues with the bound cocaine in S1 is shown in Fig. 3B, including Phe43 (100%, interaction frequency over the last 50-ns trajectory), Ala44 (97%), Asp46 (100%), Ala48 (80%), Ala117 (83%), Val120 (100%), Asp121 (99%), Tyr123 (69%), Tyr124 (100%), Phe319 (100%), Ser320 (100%), Gly322 (100%), Phe325 (90%), Ser421 (100%), Ser422 (94%), Gly425 (76%), Ser426 (50%), and Ile429 (78%). Except Gly322, Ser422, Gly425, and Ile429, the other residues interfacing with cocaine are the same as those identified in the crystal structure,8 confirming the stable binding of cocaine to S1. Compared with the initial orientation when binding to S3 (Fig. S4), the cocaine molecule changes its orientation and displays enhanced interactions with its surrounding residues over the 300-ns MD simulations (Fig. 3A). In addition to Tyr119 (99%), Phe122 (76%), Phe470 (100%), and Ile563 (100%), more hydrophobic residues are involved, including Ile229 (81%), Leu473 (100%), Ala478 (81%), Trp555 (82%), Ala556 (99%), and Leu559 (76%). In addition, strong interactions with Tyr477 (93%) and Gly560 (94%) were also identified.</p><p>To investigate the mutual effects of simultaneous binding of two cocaine molecules to S1 and S3, residues interacting with cocaine in S1 or S3 in the presence of a second cocaine are also shown in Fig. 3. The binding of cocaine to S3 appears to show little effect on the binding of cocaine to S1. Most interacting residues in S1 are preserved, and only interactions with Tyr123 and Ser426 are weakened to a negligible extent (Fig. 3D). By contrast, the binding of cocaine to S1 has notable impacts on the interactions of cocaine in S3 with dDAT (Fig. 3C), which diminishes interactions with Ala478, Ala556, and Gly560, weakens interactions with Tyr119 (64%), Leu473 (81%), Trp555 (68%), and Ile563(77%), but enhances interactions with Phe122(97%) and Leu559(96%), and establishes new interactions with Val458 (88%). Therefore, the binding of cocaine to S1 seems to have more significant influence on the binding of cocaine to S3.</p><p>The binding affinity of cocaine in different binding sites of dDAT was predicted in terms of the MM-PBSA approach implemented in the tool g_mmpbsa,50 and results were summarized in Table 1. The most striking finding is that the binding affinity of S3 for cocaine is higher than that of the central binding site S1 when only one cocaine binds to dDAT. In the case of two cocaine molecules separately binding to S1 and S3, the binding energy increases for both sites, indicating that binding of two cocaine molecules becomes less favorable. Also, the binding energy of cocaine in S1 increases 19%, and the binding energy of cocaine in S3 increases 54%, suggesting that binding of cocaine to S3 is more prone to be allosterically modulated by binding of cocaine to S1, in line with the above analyses of interacting residues.</p><p>The binding affinity of DA with respect to dDAT was also calculated (Table 1). As expected, the binding affinity of cocaine to S1 is higher than that of DA, supporting a competitive mechanism for cocaine inhibition of DA uptake. As shown in Figs. S7 and S8, DA moves from the more extracellularly positioned site S2 toward S1 regardless of the absence (Fig. S7) or presence (Fig. S8) of a second DA in S1. When one DA occupies S1, the secondary incoming DA does not alter the stability of dDAT (Fig. S6) by interacting with residues in the centrally located S1. (Fig. S9) Direct interactions between the two bound DA molecules were also found, but such interactions are not intense as the frequency of occurrence is about 39%. Because the binding energy for individual DA was calculated in the presence of another DA, direct comparison with the result of only one DA binding to dDAT is not straightforward. Therefore, the binding energy was also calculated without considering the presence of a second DA (Table 1), and a lower binding energy was obtained. However, the binding energy calculated in this way may not reflect the real binding environment as the effect of binding of a second DA was not taken into account. Nevertheless, the finding that the binding affinity of the second DA is on the same magnitude as that of the first one may suggest the possibility of simultaneous binding of two DA molecules on a single dDAT.</p><p>To further explore the effect of cocaine binding on DA binding, MD simulations of dDAT inserted in the same lipid bilayer were performed (Fig. S11), along with one DA bound to S1 and one cocaine bound to S3 (Fig. 4). Compared with DA-bound dDAT system (Fig. S7), in addition to strong interactions with Phe43 (99%), Asp46 (100%), Tyr124 (97%), Ser421 (100%), and Gly425 (92%), the frequency of interactions with Ala117 and Ser422 by DA is increased by 37% and 97%, respectively, whereas interaction of DA with Phe325 is decreased by 26%. The greatly enhanced interactions with Ser421 and Ser422 might imply a shrink of the binding pocket S1. On the other hand, in the presence of DA in S1, cocaine largely maintains the initial docking orientation, different from those systems where one cocaine binds to S1 (Fig. 3A) or two cocaine molecules bind to S1 and S3, respectively (Fig. 3C). As a result, interactions of cocaine with dDAT become weaker in general. In particular, interactions with Phe122, Ile229, Leu473, Tyr477, Ala478, Trp555, Ala556, and Gly560 were not observed. Instead, the cocaine establishes stable interactions with Val458 (96%), Ala461 (85%), and Ser462 (96%). Taken one DA- or cocaine-bound dDAT as a reference, the calculated binding energy of DA to S1 and cocaine to S3 increases by 3.52 kcal/mol and 8.28 kcal/mol, respectively (Table 1), suggesting a relatively lower binding affinity of DA with respect to S1 and cocaine with respect to S3. In addition, the greater change in the binding energy of cocaine could indicate that the binding of DA to S1 affects the binding of cocaine to S3 more significantly.</p><p>The human DAT F154A mutant was reported to lower cocaine affinity.24 The corresponding F122 in dDAT is located in S3 and strongly interacts with cocaine (Fig. 3). To examine the impact of F122A mutant on cocaine binding, we performed MD simulations on F122A mutant bound with one cocaine in S1 or S3, and in both S1 and S3. No significant structural changes were observed and dDAT remains in the outward-open conformation (Fig. S12). The notable difference in the ligand-protein interactions is the decreased number of residues that interfere with cocaine bound to S3 and S1 simultaneously, compared to the interacting residues when only cocaine binds to S3 (Fig. S13). Because the contribution of residue F122/A122 to the binding energy is less than 4% in all simulated systems (Table S2), it seems applicable to direct compare the change in the binding energy between the wild-type and the F122A mutant. Because of the mutation, the binding energy of cocaine to S1 decreases by −3.89 kcal/mol, suggesting that the F122A mutation enhances the binding affinity of cocaine to S1. On the other hand, the binding energy of cocaine to S3 increases by 10.63 kcal/mol, resulting in a net increase in the binding energy and thereby a net decrease in the binding affinity of cocaine to dDAT. Similar results were also obtained for the binding energies calculated using ε(membrane) = 3. Moreover, the binding of two cocaine molecules reduces the binding affinity of both sites in the F122A mutant.</p><!><p>The influence of ligand binding on the conformational stability of dDAT was estimated based on the calculated conformational energy of protein structure (Fig. 5). The ligand was removed from the binding site but the bound Na+ and Cl− ions were kept in all calculations. For comparison, all-atom MD simulations were also performed on the ligand-free dDAT embedded in the same lipid bilayer, and the calculated conformational energy of the dDAT was used as a reference. Relative to the ligand-free dDAT, binding of one or two DA molecules decreases the conformational energy of dDAT by about 45 kcal/mol and 126 kcal/mol, respectively, suggesting that binding of DA increases the stability of dDAT. However, the conformational energy of dDAT with one cocaine bound to S1 is comparable to that of dDAT in free state, indicating no obvious energy barrier to the conformational transition from cocaine-free to cocaine-bound state. Interestingly, binding of one cocaine to S3 results in a decrease of about 27 kcal/mol in the conformational energy of dDAT, but binding of two cocaine molecules to S1 and S3 leads to an increase of about 47 kcal/mol in the conformational energy of dDAT. A similar trend was also found in the dDAT with F122A mutation. Such a destabilizing effect reveals that simultaneous binding of cocaine to both S1 and S3 seems unstable and infeasible. This finding is in line with the results of binding energy, showing that the binding affinity of both sites decreases when cocaine molecules bind to S1 and S3 simultaneously. Another interesting finding is that the dDAT displays stabilization upon binding of DA to S1 and cocaine to S3. The conformational energy decreases by about 31 kcal/mol, comparable to the system where one cocaine binds to S3 of dDAT (~27 kcal/mol). This result may indicate the coexistence of DA bound to S1 and cocaine bound to S3, but such a state appears unstable and tends to shift toward the more stable state in which only one DA binds to S1. This conclusion is also consistent with the calculated binding energy, which shows a decrease in the binding affinity of DA to S1 and cocaine to S3. However, the calculated binding energy implies that the binding affinity of cocaine to S3 is higher than that of DA to S1, in that the calculations are based on the ligand-bound states.</p><p>To study the impact of binding of cocaine to S3 on the binding of DA, the change in the volume of S1 during simulations was calculated using MDpocket program,54 and results are shown in Fig. 6. No significant fluctuations were observed for those systems where S1 is occupied by DA and cocaine. The volume of S1 varies between 418 to 438 Å3 for cocaine-bound systems, and a smaller value of ~291 Å3 was obtained for DA-bound system. By contrast, in the wild-type dDAT and F122A mutant, the binding of cocaine to S3 results in striking fluctuations in the volume of S1, and the averaged volume of S1 decreases to ~100 Å3 and ~190 Å3 for the wild-type and the F122 mutant, respectively, which may reduce the binding DA to dDAT as the volume of S1 is smaller than that needed for accommodating DA (~291 Å3), or a large conformational change of the protein induced by DA binding is required. Of interest, no significant alterations in the helical structures were observed along with the large fluctuations in the pocket volume (Fig. S14), suggesting that the changes in the pocket volume may involve the movement of domain that consists of several transmembrane helices.</p><!><p>The bacterial homologues LeuT is a model transporter used to study the molecular mechanism of substrate binding and transporting in NSS family. The crystal structures of LeuT in complex with one substrate and one other ligand provide evidence to support the two binding site model of NSS transporters.55 However, in addition to the primary binding site at the center of the protein, the existence of an alternative substrate binding site in LeuT remains controversial.10, 17, 19 The DAT is also a member of NSS family, but the X-ray crystal structures of dDAT in complex with DA and cocaine determine that both ligands bind to the central binding site S1,8 excluding the existence of a secondary high-affinity binding site in the protein. In addition to the binding sites in the interior of a protein, binding pockets on the surface of a protein may also be suitable for binding of a ligand.56 However, characterization of such binding pockets on the surface of membrane proteins such as DAT is challenging because most of cavities on the protein surface could be buried by surrounding lipid or cholesterol molecules. Identification of potential binding sites on the surface of DAT allows us to investigate the difference in the binding propensity between substrate DA and inhibitor cocaine and, accordingly, provide molecular insights into the conformational dynamics induced by ligand binding.</p><p>In the present study, the combination of molecular docking and MD simulations enables us to identify different binding modes of DA and cocaine on dDAT (Fig. 1). In addition to S1, an alternative DA binding site S2, which has been proposed by early steered MD simulations,22 was also found by the present docking study. Further MD simulations show that DA diffuses from S2 toward S1 (Figs. S7 and S8), suggesting that S2 is not a high-affinity binding site for DA. In a previous study of human DAT modeled based on the structure of dDAT, DA was also found to transiently bind to S2.23 The finding that cocaine escapes from S2 and translocates into solution (Fig. S5A) is also consistent with the above MD simulations of human DAT which showed that S2 precluded cocaine to penetrate into inner extracellular vestibule.23 The residues Arg52 and Asp475 were found to form a salt-bridge intermittently, but this salt-bridge is not sufficient to close the extracellular gate because the distance between the pair of gating residues Tyr124 and Phe319 is reduced by only ~1–2Å. Therefore, the structure of dDAT remains in the outward-open conformation (Fig. 2). In addition, the present simulations reproduced the interactions between DA/cocaine and dDAT as observed in the crystal structures (Fig. 3 and Fig. S9A), confirming that S1 is a high-affinity binding site for both DA and cocaine. Note that the calculated binding energy seems very sensitive to the alterations of conformations of dDAT. For example, the same residues that interact with DA in S1 were observed for the system where only one DA binds to S1, as well as for the system where two DA molecules bind to S1 and S2, respectively (Figs. S9 and S10, and Table S3). However, the residue contributions to the corresponding binding energy vary significantly, which could be attributed to the different conformational spaces sampled by dDAT. A large difference in the conformational energy between these two dDAT systems in complex with DA was observed (Fig. 5).</p><p>Our docking studies identified multiple cocaine binding pockets on the surface of dDAT (Fig. S2A), but only one site S3 located between TMs 3, 9, 10, and 12 (Fig.S4) is capable of binding cocaine stably. Interactions with more hydrophobic residues from different TMs contribute to the association of cocaine with S3 on dDAT (Fig. 3). Extending our current 300-ns MD simulations of cocaine-bound dDAT to 1 μs, we did not observe any dissociation of cocaine from S3 (Fig. S15), and the dDAT still remains in the outward-open state, suggesting that S3 might be an alternative high-affinity binding site for cocaine on the surface of dDAT. Moreover, the calculated binding energies imply that the binding affinity of S3 to cocaine is even higher than that of S1. However, different from S1 that is in the interior of the protein, S3 is on the surface of dDAT, and cocaine is positively charged, the actual binding affinity could be even higher than the calculations in which a dielectric constant of ε=80 was used to model solvent effect in the solvation energy calculations.50 To test this, we calculated the binding energy for several systems with cocaine bound to S3 using a dielectric constant of ε=3 to mimic the lipid bilayers in aqueous solutions57 (Table 1). Indeed, the calculated binding energies decreased by ~9–20 kcal/mol, depending on the systems, suggesting an increase in the binding affinity of cocaine to S3 in the membrane environment.</p><p>Our results show that upon simultaneous binding of two cocaine molecules to S1 and S3, the binding affinity of S3 reduced, independent of environment. This finding indicates a coupling effect between these two sites. Because S1 is buried inside the protein and S3 is located on the surface of the protein, their distinct property prevents us from directly comparing the binding affinity between the two sites. The calculated binding energy in an aqueous solution predicted that S1 is a more favorable binding site, but on the other hand, S3 was expected to have a lower binding energy and accordingly a higher binding affinity than in solution. Nevertheless, our results showed a greater change in the binding energy of S3, implying that the binding of cocaine to S1 has a more significant impact on the binding of cocaine to S3. The experimental finding that the F122A mutant lowers cocaine affinity24 could be due to a net reduction in the binding affinity of cocaine to different sites. Compared to the wild-type dDAT, the increase in the binding affinity of S1 was overwhelmed by the decrease in the binding affinity of S3. Similar to the wild-type dDAT, a larger change in the calculated binding energy of cocaine to S3 was observed when two cocaine molecules bound to the F122A mutant when compared to the corresponding systems where only one cocaine molecule binds to the protein. Overall, these results indicate that the binding of cocaine to S1 destabilizes the binding of cocaine to S3 significantly.</p><p>Cocaine acts as a competitive inhibitor of DA uptake by locking the DA transporter in the outward-open conformation. In this way, the binding of cocaine reduces the availability of the inward-open conformation that facilitates the transport of DA from an extracellular space to an intracellular space.8, 58 Our results show that the binding energy of cocaine to S1 is much lower than that of DA to S1, confirming the higher binding affinity of cocaine than that of DA to the same central binding pocket. Of interest, our results show that the binding of two DA molecules in a single dDAT seems possible, and the interactions between the two DA molecules may accelerate the conformational transition and therefore facilitate the efficient transport of DA. For cocaine-bound dDAT, the stabilization effect is more prominent when cocaine binds to S3 only, whereas the finding that binding of a second cocaine to S1 destabilizes the conformation of dDAT, along with the decreased binding affinity, collectively suggests that cocaine either binds to S1 or S3, but simultaneous binding to S1 and S3 appears infeasible. Furthermore, the binding of cocaine to S3 could also affect the binding affinity of DA to S1, providing an alternative mechanism on the competitive inhibitor of DA by cocaine. In addition, the binding of cocaine to S3 induces significant fluctuations on the volume of S1, but no significant changes in the helical structures were found. Although the plasticity of S1 allows it to accommodate ligands of varying sizes,8 but on the other hand, the large fluctuations in the binding pocket would make ligand recognition by dDAT less favorable as ligand-protein association is modulated by the relation between the time scale of accessibility of the binding site and the time scale for ligand binding.56 Large pocket dynamics could inevitably influence the binding of ligands with high specificity. Therefore, in addition to directly compete for binding to S1 and inhibit DA binding, our results imply that binding to S3 could also allosterically reduce the binding of DA to S1, providing an alternative mechanism on the multiple inhibitory patterns of cocaine.</p><p>In the DA- or cocaine-bound crystal structures, a CHOL molecule is found at the interface of TMs 1a , 5, and 7 and a CHOL analogue, cholesteryl hemisuccinate, is found at the interface between TMs 2 and 7.8 Because no cholesterol molecules occupy this allosteric binding site, S3 seems free for cocaine binding. However, the binding of cocaine to S3 could not be identified by crystallization because the bound cocaine could dissociate from dDAT during the preparation of the crystals such as washing out the impurities. Moreover, the dDAT has been suggested to form oligomers,30 and cocaine has been reported to induce the formation of DAT oligomers in mouse neuroblastoma cells,59 resulting in cocaine addiction. Our results show that the binding of cocaine to S3 results in a more stable conformation of the dDAT, suggesting that this conformational state is most likely to associate and form dimer and other oligomers. Because no structures of DAT oligomers are solved to date, we used the LeuT dimer as a template (PDB ID: 3TT1)7 to model a dDAT dimer with the dimer interface involving TMs 9 and 12 (Fig. 7). Note that this interface overlaps with the cocaine binding site S3. Although a clear relation between the formation of DAT dimer and the cocaine binding to S3 remains to be determined, our work highlights the important role of S3 in cocaine binding and oligomerization of DAT. Overall, the present work provides novel insights into the molecular mechanism on the inhibitory effect of cocaine on DAT.</p>

PubMed Author Manuscript

Cyclic siloxanes in air, including identification of high levels in Chicago and distinct diurnal variation

The organosilicon compounds octamethylcyclotetrasiloxane (D4), decamethylcyclopentasiloxane (D5), and dodecamethylcyclohexasiloxane (D6) are high production volume chemicals that are widely used in household goods and personal care products. Due to their prevalence and chemical characteristics, cyclic siloxanes are being assessed as possible persistent organic pollutants. D4, D5, and D6 were measured in indoor and outdoor air to quantify and compare siloxane concentrations and compound ratios depending on location type. Indoor air samples had a median concentration of 2200 ng m\xe2\x88\x923 for the sum of D4, D5, and D6. Outdoor sampling locations included downtown Chicago, Cedar Rapids, IA, and West Branch, IA, and had median sum siloxane levels of 280, 73, and 29 ng m\xe2\x88\x923 respectively. A diurnal trend is apparent in the samples taken in downtown Chicago. Nighttime samples had a median 2.7 times higher on average than daytime samples, which is due, in part, to the fluctuations of the planetary boundary layer. D5 was the dominant siloxane in both indoor and outdoor air. Ratios of D5 to D4 averaged 91 and 3.2 for indoor and outdoor air respectively.

cyclic_siloxanes_in_air,_including_identification_of_high_levels_in_chicago_and_distinct_diurnal_var

1. Introduction<!>2.1. Materials<!>2.2. Sampling locations<!>2.3. Sampling and extraction method<!>2.4. Instrumental analysis<!>2.5. Quality assurance/quality control<!>2.6. Boundary layer height<!>3.1. Indoor air concentrations<!>3.2. Outdoor air concentrations<!>3.3. Emissions from indoor air<!>3.4. Comparison of measured and modeled concentrations and D5/D4 ratios<!>4. Conclusions

<p>Siloxanes are anthropogenic chemicals used in personal care and household products, and in the production of the silicon polymer polydimethylsiloxane (Horii and Kannan, 2008). Siloxanes in cyclic, linear, and/or polymer forms are used in an extremely wide range of applications. They are commonly found in antifoaming agents, in automotive care products, as coatings and sealants in construction, and as main ingredients in personal care products (Kaj et al., 2005). The cyclic siloxane compounds (commercial name cyclomethicone) are valued as solvents or carriers in personal care products because of their inert, odorless, and colorless qualities in addition to their volatility and smooth feel. In a recent survey of siloxanes in personal care and household products octamethylcyclotetrasiloxane (D4), decamethylcyclopentasiloxane (D5), and dodecamethylcyclohexasiloxane (D6) were found in more than 50% of the items analyzed (Horii and Kannan, 2008). In a similar study that measured siloxane levels in personal care products only, cyclic siloxanes made up as much as 68% of the total product mass (Wang et al., 2009). D4, D5, and D6 are categorized as high production volume chemicals (greater than 1 million pounds produced or imported annually in the U.S.) (Canada, 2008a,b,c).</p><p>Cyclic siloxanes are of concern because of their high rate of use and their physical properties (Howard and Muir, 2010). They are extremely volatile, have a relatively long half-life in air (6–11 d), and a high Kow (Kaj et al., 2005). Accordingly, D4, D5, and D6 were recently identified as substances requiring further study because of their potential for long range transport and bioaccumulation (Howard and Muir, 2010). Environmental risk assessments have been conducted for cyclic siloxanes in Canada (Canada, 2008a,b,c), the UK (Brooke et al., 2009a,b), Sweden (Kaj et al., 2004), and for a consortium of Nordic countries (Kaj et al., 2005). The assessment of siloxanes in the Nordic environment focused on measurements of siloxane concentrations in a number of different media. Siloxanes were found to be ubiquitous in the environment, with measurable amounts in nearly all media analyzed, including air, water, sediment, sludge, and biota, with D5 found predominantly (Kaj et al., 2005). The screening assessment conducted by Environment Canada did not include environmental concentrations, but based on fugacity modeling and a review of studies relating to environmental toxicology, they concluded that D4 and D5 are "…entering or may be entering the environment in a quantity or concentration or under conditions that have or may have an immediate or long-term harmful effect on the environment or its biological diversity" (Canada, 2008a,c). Furthermore, D4 was recently included in a list of 18 chemicals to be reviewed by the U.S. EPA in 2013 and 2014 (C and EN, 2012). The evaluation will determine whether or not the compound should be regulated under the Toxic Substances Control Act (C and EN, 2012).</p><p>If regulation is to be considered, then it is important to learn more about the sources and distribution of siloxanes in the environment. Publications of indoor air measurements which could aid in source identification are especially lacking. Currently, one indoor (Shields et al., 1996) and six outdoor air studies have been published (Wang et al., 2001; Kierkegaard and McLachlan, 2010; McLachlan et al., 2010; Cheng et al., 2011; Genualdi et al., 2011; Krogseth et al., 2012). Several studies have been conducted that have modeled concentration distributions of one or more of the cyclic siloxanes (Mueller et al., 1995; McLachlan et al., 2010; Navea et al., 2011). However, increased monitoring is necessary in order to more accurately assess the influence siloxanes may have on humans and the environment. This is especially key in areas of high population density (i.e. Chicago) where emissions of anthropogenic compounds are much higher, and therefore the probable impact on the local population and environment will be correspondingly high (i.e. Lake Michigan). The purpose of this work is to measure and compare concentrations of D4, D5, and D6 in outdoor air in areas with varying population densities, as well as determine indoor concentrations and provide an estimate for per-person emissions of cyclic siloxanes. Concentrations, emissions, and compound ratios will be contextualized through a discussion of current modeling and measurement data for siloxanes in air.</p><!><p>Octamethylcyclotetrasiloxane, decamethylcyclopentasiloxane, and dodecamethylcyclohexasiloxane (>98%) were purchased from TCI America Portland, OR. Tetrakis(trimethylsiloxy)silane was purchased from Gelest Inc. in Morrisville, PA. Solvents dichloromethane, methanol, acetone, and n-hexane (all pesticide grade) were purchased from Fischer Scientific Pittsburgh, PA. 10 mg Isolute ENV+ cartridges (Biotage AB) were employed for all samples.</p><!><p>Indoor samples were collected in the Seamans Center for the Engineering Arts and Sciences at the University of Iowa: ten samples in one laboratory, one in a student office (Office A), and three in a second office (Office B). Both offices are used during regular daytime work hours (between 9 am and 5 pm) by up to ten people. The laboratory has between one and seven occupants during daytime hours. Eight of the samples ran overnight and five samples ran during the day. The laboratory ventilation flow rate varies between 775 and 2270 cubic feet per minute depending on laboratory hood use and heating or cooling requirements. Additional details can be found in the Supplemental information.</p><p>Two outdoor locations in Iowa were chosen as representative of medium and low population density sites. The rural site is approximately three miles north of West Branch, IA (population 2400) at the base of a NOAA tall tower for carbon cycle gas sampling (Andrews et al., 2013). The mid-sized city selected for the transition site was Cedar Rapids, IA (population 122000) at the Linn County Public Health Department's air quality monitoring station. The urban samples were taken at the EPA's Integrated Atmospheric Deposition Network (IADN) site at the Illinois Institute of Technology (IIT) campus in Chicago.</p><!><p>Air sampling was conducted by drawing air through an extraction cartridge using a diaphragm pump (GAST MAA-V109-HD) connected with polyethylene tubing (Watts 1/4″ OD, .170″ ID) (Kierkegaard and McLachlan, 2010). Two lines were used in order to have duplicates for each sampling period, and the flow rates for each line were measured separately. Flow rates varied from 3 to 6 liters per minute and were measured with Toptrak mass flow meters (822-2-OV1-V1 and 822-2-OV1-PV1-V1). One flow meter did not include a volume totalizer, so flow rate and time were recorded with a USB data logger from Lascar Electronics (EL-USB-3). The SPE cartridges were installed with the open side down and were shielded from the elements with funnel attachments. Sampling times varied depending on the concentration ranges expected at each site. Indoor samples and samples taken at IIT were run for 12–16 h. Cedar Rapids samples were collected over 24 h and West Branch about 36 h in order to ensure that detectable masses accumulated. After sampling, the cartridges were wrapped in aluminum foil and stored in glass amber jars in a cooler or freezer until extracted. The extraction process involved running approximately 1.5 mL of n-hexane through the SPE cartridge directly into a GC vial. Internal standard tetrakis trimethyl-siloxysilane (M4Q, 100 ng) was then added to each vial.</p><!><p>The samples were analyzed on a Hewlett Packard gas chromatograph mass spectrometer (HP 5973) in select ion monitoring mode. The column used was a Restek RTX-5MS. The injector temperature was 200 °C. Injection volume was 2 μL. The flow rate of helium gas was held at 1.0 mL/min. The temperature gradient was: 60 °C (2 min), to 150 °C at 10 °C/min, to 300 °C at 30 °C/min (2 min), with a detector temperature of 250 °C. The ions monitored were m/z 281 (D4 and M4Q), 355 (D5), and 341 (D6). The samples were quantified by the internal standard method. A 500 ng/mL standard containing D4, D5, D6, and M4Qwas used to calculate the relative response factors for each run.</p><!><p>All glassware was combusted overnight at 450 °C before use. Temperature-sensitive sampling components were triple-rinsed with pesticide grade methanol, hexane, and acetone. SPE cartridge cleaning practices entailed soaking the cartridges overnight in n-hexane followed by a run-through three times with dichloromethane and three times with n-hexane before drying with a nitrogen stream. Cartridges were then wrapped in aluminum foil and frozen. Use of siloxane-containing products was avoided on days when lab or field work was conducted.</p><p>Cartridge performance was measured by sampling through two cartridges in series. Less than 1% of each D4, D5, and D6 was found on the second cartridge. Extraction efficiency was evaluated by spiking the top frit of a cartridge. Average recoveries were 99, 114, and 110 percent for D4, D5, and D6, respectively.</p><p>Field blanks and duplicates were included with all the samples except for the five indoor air samples taken in 2011 (Lab1–Lab4 and Office A). All outdoor air samples and the remaining indoor air samples had a field blank attached to a capped length of tubing installed alongside the sample cartridges for the duration of each sampling period. The overall average field blank mass for indoor and outdoor air was 3.2, 10, and 12 ng per cartridge for D4, D5, and D6 respectively. A field blank concentration was determined by dividing the mass of the field blanks for each site by the average volume sampled at each site. The average sampling volumes were 9.4, 7.1, 3.2, and 2.7 m3 for West Branch, Cedar Rapids, Chicago, and indoors respectively. A specific limit of detection (LOD) and limit of quantification (LOQ) was then calculated for each combination of site and compound. The LOD was determined by the average of the field blank concentration plus three times the standard deviation, and the LOQ is the average plus ten times the standard deviation. Duplicate samples exhibit an average relative percent difference of 13% for 34 sample pairs. The results were not blank corrected.</p><p>We found little evidence of decay or loss during storage. Krogseth et al. (2012) reported that D5 may convert to D4 within the cartridge during storage. We found that D4 concentrations were positively correlated with storage time (p = .023), but that D5 concentrations were not. There was no correlation between storage time and the ratio of D5/D4. Therefore, no corrections were made for storage, although we recognize that some D4 may be present due to transformation after sampling.</p><!><p>The atmospheric boundary height during each of the Chicago sampling periods were retrieved from the National Oceanic and Atmospheric Administration (NOAA) archived meteorological simulations for the latitude and longitude of the Illinois Institute of Technology (41°50′4.10″ N; 87°37′25.59″ W). The North American Mesoscale (NAM) model 12 km Forecast Data Archive product (boundary layer depth variable) was used and was accessed at http://ready.arl.noaa.gov/READYamet.php. This product has 3 h time resolution. To produce an average height during a sampling period, the heights were linearly interpolated to a 15 min time basis, and then averaged. Since the conceptual model we are working with is that emissions are diluted throughout and well-mixed within the boundary layer, we utilize an inverse weighted mean boundary layer height hI¯=Δt[∫h(t)−1dt]−1 which has the useful property that hI¯C¯=h(t)⋅c(t)¯ under the assumption of a constant loading of pollutant under the boundary layer. In these formulas h(t) is the time varying boundary layer height, ∆t is the sampling interval, c(t) is the time varying concentration, and c¯ is the time averaged sample concentration measured in this work. hI¯ is lower than the arithmetic mean boundary layer height.</p><p>As an example, if the boundary layer height was 100 m for 1 h and 1000 m for 1 h, the simple arithmetic mean boundary layer height would be 550 m while hI¯ would be 182 m. A well-mixed pollutant with loading within the boundary layer of 1000 mg m−2 would have concentration 10mg m−3 during the first hour and 1 mg m−3 during the second hour, and c¯ of 5.5 mg m−3. The column loading of 1000 mg m−2 can be recovered by 5.5 mg m−3 times hI¯ but would be overestimated by multiplying c¯ by the arithmetic mean boundary layer height. This is of course an extremely simplified model of the boundary layer processes, but it is suitable for the purposes of this paper, which are to contrast daytime and nighttime concentration measurements with an adjustment for the effect of the changes in boundary layer height.</p><!><p>Indoor concentrations of individual cyclic siloxanes ranged from not detected (D6), to as much as 56000 ng m−3 for D5 (Table 1). High indoor air concentrations were expected because of the prevalence of siloxane-containing products and their subsequent accumulation in small indoor areas. Concentrations of D4 and D5 have been shown previously to be reflective of occupant density (Shields et al., 1996). The influence of occupancy is best illustrated by the variation in D5 concentrations depending on the time of day the sample was taken (Fig. 1).</p><p>D5 is the dominant cyclic siloxane in indoor air, accounting for an average of 97% of the mass found in the fourteen samples. Interestingly, the only previous study of indoor air (conducted in 1991) reported levels of D4 that were on the same order of magnitude as D5 (Shields et al., 1996). This difference from our study may be a reflection of formulation changes in personal care products over the last 20 years. Two recent studies of cyclic and linear siloxanes in common products report that D5 was the largest contributor to siloxane mass in the personal care items surveyed (Horii and Kannan, 2008; Wang et al., 2009). Based on product use profiles for women in the U.S., and the concentrations measured, Horii and Kannan estimated that the average use rate of D5 is 233 mg (person-day)−1 (2008). Wang et al. estimated a D5 exposure of 306 mg (person-day)−1 from body lotion alone also using data for use rates of lotion for women in U.S. (2009).</p><!><p>For outdoor air, the median concentrations of the cyclic siloxanes increased with population density (Table 2). The lowest concentrations were found in West Branch, IA with a median of 29 ng m−3. Siloxane levels were slightly higher in Cedar Rapids at 73 ng m−3, and the IIT concentrations were 280 ng m−3. Average sum concentrations were significantly different in Chicago compared to both Cedar Rapids and West Branch (p-value = .005 and .012), with the Chicago median 4 and 10 times greater than Cedar Rapids and West Branch respectively.</p><p>A diurnal trend is apparent in the samples collected at the IIT site in Chicago. The samples were collected over consecutive 12-h periods with cartridges exchanged at approximately 7 am and 7 pm daily. The trend is most clearly illustrated by the fluctuations in the compound with the highest concentrations, D5 (Fig. 2). D5 concentrations were significantly different during the day versus the night (p = .0014) with a median concentration of 138 ng m−3 in the daytime, and 338 ng m−3 at night. D4 day and nighttime concentrations were also significantly different (p = .024), and the D6 concentrations were generally above the LOQ during the day, and below the LOD at night. This is contrary to what we would expect if the majority of the siloxane concentration comes from volatilization during morning or daytime personal care product use. The major degradation pathway for siloxanes is by hydroxyl radical (OH) attack, and concentrations of OH are higher in the presence of sunlight, which could explain, in part, the diurnal variation (Navea et al., 2011). However, the predicted magnitude of the OH effect is small, ±10% of the mean concentration.</p><p>We investigated the effect of the variation in the average planetary boundary layer height on the day and nighttime concentrations with a regression analysis. D5 concentration was found to be significantly correlated with boundary layer height (p = .003, R2 = 0.49). D4 concentrations were not significantly correlated with the height of the boundary layer, and the correlation of D6 concentrations with boundary layer height were not analyzed because of the number of samples that were below the LOQ. This analysis included all samples except for IIT6. No obvious explanation is apparent for the high concentrations found in sample 6, as the NAM boundary layer height was not particularly low for that nighttime sampling period, nor was any unusual activity observed that could have caused a spike in siloxane concentration. Nevertheless, sample 6 is an outlier, with a standardized residual of 3.52 so was removed from the data set for the analysis.</p><p>A rough estimate of column abundance in the mixed layer was calculated by multiplying the sample concentrations by the boundary layer heights with the temporal averaging of boundary layer heights as described above. The daytime column abundances of D4 and D5 are higher than their nighttime abundances, with day/night ratios of 1.9 and 1.4, respectively. But the differences are not consistent enough to be statistically significant given the small sample size. Our interpretation of this is that emissions occur throughout both the day and nighttime sampling periods; that the relative day and night emission rates remain uncertain; and that there is inconclusive evidence that daytime emission rates are larger than nighttime emissions.</p><!><p>Emissions were calculated using the data from the laboratory because sufficient information about the ventilation systems for Office B was not available. The laboratory however, is ventilated directly from an HVAC system that receives 100% outside air. Per-person mass emission rates are therefore given by (1)E=(Cindoor−Coutdoor)∗Qnwhere E is the emission rate per person-day, Q is the volumetric flow rate of the ventilation system, and n is the number of occupants of the lab. Building design data constrain Q to between 775 and 2270 cubic feet per minute depending on laboratory hood use and heating or cooling requirements, and n ranges from 1 to 7 people. The flow rate and number of people, while not known exactly during the sampling periods, are varied across these bounds to give minimum and maximum values for E. As we do not have simultaneous measurement of Cindoor and Coutdoor, we approximate the Coutdoor by the indoor nighttime concentration, when the room is not occupied and the indoor and outdoor concentrations should be equal.</p><p>Per-person emissions were calculated using Eq. (1) and the average day and nighttime concentrations found in the laboratory air. The minimum emission rate, using seven people and a flow rate of 22 m3 min−1 for D4 and D5 is 0.0090 and 29 mg (person-day)−1 respectively. The maximum emissions rate (1 person and 64 m3 min−1) is 0.027 and 590 mg (person-day)−1 for D4 and D5. Because the sample involved a small n, it is not meant to be representative of population averaged emission rates. However, due to the potential for wider applicability of the method (using indoor air mass balance approaches to estimate emission rates) and the scarcity of per capita emission data, we report and compare the values here. The measured D5 emission rate (590 mg (person-day)−1) is within a factor of 3 of the value found by Horii and Kannan (230) and larger than the value used in modeling by Navea et al. (140). This seems consistent with the higher outdoor measured concentrations (18–210 ng m−3) in this study when compared with the modeled concentrations by Navea et al. (3–45 ng m−3) – which will be discussed further in Section 3.4.</p><p>For D4, the conclusions to draw from comparison of this work with Horii and Kannan and Navea et al. are perhaps more intriguing and worthy of future research. Both the mass emission rate from this work (<0.027 mg (person-day)−1) and the value reported by Horii and Kannan based on personal care product usage (1.1 mg(person-day)−1) are much lower than the value used by Navea et al. in modeling (90 mg (person-day)−1). However, as discussed in the section below, the Navea et al. D4 emission rate better explains observed atmospheric measurements.</p><!><p>The D4 concentrations found in our study and a measurement study by Genauldi et al. fall within a factor of two when compared to Navea et al.'s predictions in high and medium population density locations (Table 3). On the other hand, the measured concentrations of D5 in this work are five to ten times higher than those predicted by Navea et al. As the ambient measurements are the strongest constraint on emissions, this suggests that the per person emission rate of D4 reported in this work and the emission rate of Horii and Kannan are not representative of the average per capita emission rate from all sources. This further suggests that the emission rates measured in this work and the per-person per-day usage estimates (from personal care products) of Horii and Kannan for D5 are likely more accurate than the 140 mg (person-day)−1 value used in the Navea simulations.</p><p>It is interesting to note the variations in the measured and modeled D5/D4 ratio between the urban, suburban, and rural locations. The measured average ratio is 4.5, 3.1, and 2.1 for Chicago, Cedar Rapids, and West Branch respectively. Half-lives of D4 and D5 with respect to atmospheric oxidation are (under average conditions) 11 and 7 d, respectively. Because of this, D5/D4 ratios will be highest at the points of common emission of the two chemicals, and will decrease as air masses move away from source regions and undergo oxidative aging. Evidence of this phenomenon is apparent in the measured data, although the difference is statistically significant only between West Branch and Chicago (p-value = .002). This is also predicted in the Navea model, although with much smaller variation than that measured in this study. The modeling assumption in Navea et al. – that the majority of airflow was coming from a remote background- is likely not correct for the Midwestern United States. Future work utilizing 3D chemical transport modeling of D4 and D5, and comparison to other similar "chemical clock" ratios such as the benzene to toluene ratio (Warneke et al., 2007), could be used to explore and exploit geographic patterns in siloxane ratios.</p><!><p>In this study we found that cyclic siloxane concentrations in outdoor air to be in the ng m−3 range and to vary diurnally. Concentrations in the indoor environments reached into the μg m−3 range, demonstrating that emissions from indoor air can be a significant contributor to outdoor air concentrations. The dominance of D5 mass (which is especially notable in indoor air) indicates that personal care product use is one of the primary sources of cyclic siloxanes to the environment. Comparisons between this study and others support the conclusion that the major aspects of degradation and transport of siloxanes in air are well understood. However, significant sources of uncertainty remain in the magnitudes, diurnal pattern, and compound ratios in emissions estimates, which is an area for further study.</p>

PubMed Author Manuscript

A cell-permeable gadolinium contrast agent for magnetic resonance imaging of copper in a Menkes disease model\xe2\x80\xa0

We present the synthesis and characterization of octaarginine-conjugated Copper-Gad-2 (Arg8CG2), a new copper-responsive magnetic resonance imaging (MRI) contrast agent that combines a Gd3+-DO3A scaffold with a thioether-rich receptor for copper recognition. The inclusion of a polyarginine appendage leads to a marked increase in cellular uptake compared to previously reported MRI-based copper sensors of the CG family. Arg8CG2 exhibits a 220% increase in relaxivity (r1 = 3.9 to 12.5 mM\xe2\x88\x921 s\xe2\x88\x921) upon 1 : 1 binding with Cu+, with a highly selective response to Cu+ over other biologically relevant metal ions. Moreover, Arg8CG2 accumulates in cells at nine-fold greater concentrations than the parent CG2 lacking the polyarginine functionality and is retained well in the cell after washing. In cellulo relaxivity measurements and T1-weighted phantom images using a Menkes disease model cell line demonstrate the utility of Arg8CG2 to report on biological perturbations of exchangeable copper pools.

a_cell-permeable_gadolinium_contrast_agent_for_magnetic_resonance_imaging_of_copper_in_a_menkes_dise

Introduction<!>Design and synthesis of a Gd contrast agent bearing a polyarginine tail<!>Spectroscopic properties and response to Cu+<!>Arg8CG2 exhibits greater cellular uptake and in cellulo relaxivity than CG2<!>Arg8CG2 can report on labile copper pools in a disease model<!>Conclusions

<p>Copper is an essential element for life, acting as a cofactor in a number of important enzymes that govern redox processes within the body.1 On the other hand, genetic disorders such as Wilson's and Menkes diseases arise from mutations in copper handling proteins, and disruption of copper homeostasis is associated with a number of neurodegenerative pathologies including Alzheimer's, Parkinson's and prion diseases. A number of fluorescent sensors have been developed to enable visualization of biological copper pools, based on nucleic acids,2,3 proteins,4,5 and small-molecule fluorophores.6–13 Gaining a better understanding of how copper contributes to healthy and disease states, however, requires the development of methods for imaging copper distribution in the whole body rather than in cultured cells or tissue samples. To this end, X-ray fluorescence microscopy,14,15 positron emission tomography (PET),16,17 and near infrared imaging18 have been used to study copper distribution in tissue and animal models of copper mishandling.</p><p>Our laboratory has been interested in the application of magnetic resonance imaging (MRI) for studying copper biology, as this technique is a powerful imaging modality in which whole body images of live specimens can be obtained at high resolution without the use of ionizing radiation.19,20 The observed contrast in MR images can be increased by the use of contrast agents containing paramagnetic ions. Gd3+ (S = 7/2) is most commonly employed for this purpose, owing to its large number of unpaired electrons and its ability to increase the proton relaxation rate of water molecules interacting with the metal center. A number of elegant responsive MR contrast agents have been developed,21,22 in which a relaxivity change can report on pH,23,24 gene expression,25 enzyme activity,26–28 or metal ion concentration.22,29–32 Notably, a number of recent studies have reported MR-based probes that respond to Cu2+.33–36 We are also interested in studying Cu+, which is the primary copper species in the reducing intracellular environment. We have previously developed a series of MR-based probes that exhibit increases in longitudinal relaxivity (r1) in response to Cu+ and/or Cu2+.36–38 Of particular note is CG2, a Cu+-sensitive contrast agent that displays a 360% increase in r1 upon addition of one equivalent of Cu+ (r1 = 1.5 to 6.9 mM−1 s−1; Scheme 1).22,38</p><p>Despite the promising in vitro behaviour of the many metal-responsive Gd-based contrast agents,22,39 their ability to image changes in metal concentration in cellulo remains insufficiently explored. Previous biological, MR-based metal ion sensing has largely focused on Zn physiology, including elegant studies by Sherry and co-workers on the use of a Gd-based zinc sensor for the in vivo detection of extracellular zinc released by pancreatic beta cells31 and Lippard, Jasanoff, and colleagues on the application of a cell-permeable Mn-based contrast agent to sense zinc levels in vivo.30,40 Encouraged by the large dynamic range of CG2, we sought to exploit its properties in the design of an MR contrast agent for the imaging of Cu+ in living systems. The +1 oxidation state of copper dominates in intracellular spaces; we expected that CG2, like other typical Gd3+-containing MR contrast agents such as Gd-DTPA and Gd-DOTA, would be largely confined to the extracellular space due to a poor ability to cross cellular membranes.41 We therefore considered various strategies to elicit enhanced cellular uptake, which include electroporation,42 encapsulation into liposomes43,44 and conjugation to various macromolecules such as peptides,45–48 dendrimers,49 dextrans,50 and TiO251 or gold52 nanoparticles, and settled on the use of a polyarginine tag as a promising method to this end.53–56 In this context, Weissleder and coworkers have exploited the TAT peptide sequence from the HIV virus to render nanoparticle-based MR contrast agents cell-permeable,57–59 and Meade and coworkers have reported modified Gd3+-DTPA and Gd3+-DOTA complexes containing octaarginine (Arg8) tails.41,45,51 For the latter, cellular uptake of these complexes was confirmed by inductively coupled plasma mass spectrosmetry and X-ray fluorescence techniques. Based on these precedents, we reasoned that appending an Arg8 tag onto our CG2 scaffold should readily deliver our Cu+-responsive agent to the interior of the cell. In this edge article, we now report the design and synthesis of Arg8CG2, and our comparisons of the cellular behaviour of this complex to the parent CG2. We then detail our studies of a murine cell line bearing a mutant form of the copper efflux protein, ATP7A, using Arg8CG2, demonstrating that this molecular imaging agent can distinguish between normal and aberrant labile copper pools in a disease model, and reporting the in cellulo application of a metal-responsive Gd-based MR contrast agent.</p><!><p>The synthesis of Arg8CG2 is detailed in Scheme 2. Bis-protected cyclen 1 was synthesized by a previously-reported literature procedure in quantitative yield.60 Alkylation of 1 with tert-butyl bromoacetate furnished compound 2. The benzyl carbamate protecting groups were removed by hydrogenation over Pd/C to yield 3 in nearly quantitative yield. The trisubstituted cyclen 5 was prepared by slow addition of electrophile 438 to a heated mixture of 3 and K2CO3. By this method, compound 5 was produced in 42% yield, with the tetrasubstituted by-product also observed. Triester 6 was obtained by alkylation of 5 with ethyl bromoacetate at room temperature in THF. The mono-carboxylate 7 was formed by the selective deprotection of the ethyl ester in 0.1 M NaOH in a water–dioxane (1 : 3) mixture at 50 °C. Arg8-functionalized Wang resin was obtained using standard solid phase peptide synthesis techniques. Coupling of 7 to the resin-bound polyarginine using HATU and DIPEA in DMF, followed by cleavage from the resin using TFA, gave ligand 8 following HPLC purification. Metalation of 8 was achieved with a slight excess of Gd(OH)3 in H2O (pH 6), followed by purification via reversed-phase chromatography using a C18 SepPak cartridge.</p><!><p>The relaxivity properties of Arg8CG2 were characterized in phosphate buffered saline (PBS; pH 7.4) at 37 °C and Gd3+ concentrations were determined by ICP-OES. In the absence of Cu+, the relaxivity of Arg8CG2 is 3.9 mM−1 s−1 (Fig. S1†). This r1 is significantly higher than that of the parent complex CG2 (r1 = 1.5 mM−1 s−1), which is likely to reflect an increased rotational correlation time (τR) due to the large octaarginine group, and increased second-sphere hydration due to the presence of additional hydrogen-bonding groups from the arginine residues. Upon addition of Cu+, the r1 of Arg8CG2 increases to 12.5 mM−1 s−1, a 220% increase in relaxivity (Δr1 = 8.6 mM−1 s−1; Fig. 1a). Although the observed turn-on response is a smaller percent increase in r1 than for our original CG2 platform (360%), this value is competent for in vivo applications. We note that the presence of the polyarginine tail does not appear to greatly affect the ability of our platform to respond to Cu+ ions.</p><p>Since our polyarginine-functionalized Gd3+ complex responds to Cu+, we sought to characterize the binding interaction between Arg8CG2 and Cu+. A Job plot analysis confirms a 1 : 1 Arg8CG2/Cu+ binding stoichiometry (Fig. S2†). The binding affinity of our complex for Cu+ was determined by monitoring changes in the absorbance spectrum of the Arg8CG2–Cu+ complex in the presence of varying amounts of thiourea, which is known to be a competitive ligand for Cu+.61 Analysis of the resulting data gave an apparent Kd of 8.3 × 10−15 M (Fig. 1b), signifying even tighter binding than the parent CG2 complex (Kd = 2.6 × 10−13 M). Moreover, the presence of the polycationic tail on the Gd core does not affect the metal ion selectivity of the CG2-based scaffold (Fig. 2). Arg8CG2 selectivity responds to Cu+ over a range of biologically relevant metal ions including 10 mM Na+, 2 mM K+, Mg2+, and Ca2+, 0.2 mM Fe2+, Fe3+, and Cu2+, and Zn2+ at both 0.2 mM and 2 mM concentrations. Arg8CG2 does have some response to Cu2+, but we do not anticipate this behavior to be a drawback for the in cellulo use of Arg8CG2, since the reducing environment of the cell favors Cu+ over Cu2+. Phantom MR images of Arg8CG2 in PBS showed that the probe displays different levels of contrast in samples with and without added Cu+ at clinically-relevant field strengths (Fig. S3†).</p><!><p>We anticipated that the polyarginine tail would afford Arg8CG2 much greater cellular uptake compared to the parent complex, CG2. This behavior was tested using HEK 293T as a model cell line. Cells were treated for one hour with either of the complexes at a range of concentrations, washed with PBS, lysed and digested prior to the determination of Gd content by ICP-MS. For each sample, protein content was also determined using a BCA assay and Gd uptake per cell was determined by calibration of the BCA assay to total cell number. These results show a ten-fold greater uptake of Arg8CG2 than CG2 at a dosing concentration of 500 μM (Fig. 3), confirming that the octaarginine tail contributes to better transport across the cell membrane. Notably, the calculated number of complexes per cell is within the range of values observed by Allen et al. for their series of polyarginine-containing complexes.41 An analysis of the uptake of Arg8CG2 after 1 h and 15 h indicates greater cellular uptake after the shorter time period (Fig. S4†), which may be due to cellular clearance mechanisms. This observation is consistent with the findings of a broad-scale study of cell-penetrating peptides, which reported that maximal uptake was achieved between 30 min and 2 h.62 On the basis of these results, all subsequent experiments were performed with an incubation time of 1 h and a dosing concentration of 500 μM.</p><p>Having established the superior cellular uptake of Arg8CG2, we then sought to investigate whether this difference translated into a change in in cellulo relaxivity. We pretreated HEK cells with CuCl2 or bathocuproine disulfonate (BCS), a membrane-impermeable copper chelator, which is commonly used to globally deplete intracellular copper. After washing, we then treated the cells with 500 μM CG2 or Arg8CG2 for 1 h. The cells were again washed, trypsinized, pelleted, resuspended in PBS, and immediately placed in a relaxometer tube, in which the T1 was measured. Cu-treated cells incubated with CG2 exhibited a relaxivity that was significantly higher than control cells (p < 0.05, Fig. 4), confirming that the probe is able to detect increases in the cellular copper pool. On the other hand, BCS-treated cells, in which there is a global depletion of labile copper, exhibited a significantly lower relaxivity, showing that CG2 is sensitive to basal levels of copper and can detect depletions in this pool. The relaxivity values obtained for this pool, however, were not much greater than those measured for cells alone, in the absence of probe. Cells treated with Arg8CG2, on the other hand, exhibited markedly higher relaxivities than those cells with no probe, and while they showed the same trends as CG2 for Cu- and BCS-treated cells, these differences were much more pronounced. The collective results in the model HEK 293T cell line indicate that, compared to CG2, Arg8CG2 exhibits in cellulo relaxivities that are much more suitable for the study of perturbations in labile cellular Cu levels.</p><!><p>Since Arg8CG2 can report on basal labile copper levels, and on increases and decreases in this pool, we sought to apply this probe to study the effects of mutations in the Atp7a gene on intracellular copper availability. This gene encodes the copper-transporting ATPase 1, ATP7A, which is the major copper efflux protein, and is therefore essential in the maintenance of cellular Cu homeostasis.63 In humans, mutations in the X-linked Atp7a gene are responsible for Menkes syndrome. On a cellular level, mutations in Atp7a give rise to elevated copper levels, exacerbated in cases of Cu supplementation.12,64 To investigate whether our probe could detect differences between healthy and disease states, we used the Menkes model WG1005 fibroblast line, which bears mutations in Atp7a, in comparison to control fibroblasts, MCH58, which express wild-type Atp7a.12</p><p>We initially sought to evaluate the toxicity of our two Gd complexes in the fibroblast cell lines, as well as in HEK 293T cells, by using the WST-1 assay (Table 1). No appreciable toxicity for CG2 was observed in any cell line at concentrations up to 2 mM over a 24 h period. Arg8CG2, on the other hand, decreased cellular viability over 24 h in the three cell lines, particularly HEK 293T and WG1005. This observation is consistent with reports of the toxicity of octaarginine-bearing compounds.62 The results of the 4 h incubations, however, demonstrate much better tolerance for the complex, and indicate that our dosing conditions of 500 μM for 1 h are unlikely to elicit significant cell death.</p><p>Uptake studies show a 9-fold greater accumulation of Arg8CG2 than CG2 in both cell lines, with similar intracellular Gd concentrations measured to those observed for HEK cells (Fig. S5†). Egress studies were also performed in these cell lines; following 1 h incubation with complex, the medium was removed, and replaced with fresh medium for 30 min, after which time intracellular Gd concentrations were measured. These results show that Arg8CG2 is retained very well, with a greater than 80% retention in both cell lines, while only 20% of CG2 is retained in the cell. These data provide further support for the use of a polyarginine tag for an in cellulo probe. Since all in cellulo relaxivity and phantom imaging studies reported here were performed within 30 min of cell collection, we can be confident that the localization of Arg8CG2 is primarily intracellular.</p><p>The 1/T1 values of the two cell lines upon treatment with CG2 and Arg8CG2 are shown in Fig. 5. As observed for the HEK 293T cells, there is a significant increase in 1/T1 for cells treated with Cu and then Arg8CG2, and a significant decrease for cells treated with BCS, compared to control and copper-treated cells. Notably, there is a significantly higher relaxivity, corresponding to a higher copper concentration, for the WG1005 cell line than for MCH58 cells. This observation is consistent with a mutation in Atp7a perturbing the ability of the cell to excrete copper. The same trend can also be observed for CG2-treated cells, although only the control cells exhibit a statistically significant difference in relaxivity, highlighting the superiority of Arg8CG2 for the in cellulo assessment of labile copper levels.</p><p>Finally, we collected phantom images of the two cell lines treated in the same manner as for the relaxivity studies (Fig. 6). Cells treated with CG2 or Arg8CG2 exhibit higher contrast than cells alone, with copper-treated cells displaying the highest signal. It is also possible to observe a difference in the contrast between the two cell lines for the copper- and Arg8CG2-treated samples. The phantom images also highlight the enhanced contrast of cells treated with Arg8CG2 compared to CG2-treated cells. These results indicate the potential of this probe for use in MRI studies in higher living model systems.</p><!><p>In summary, we have described the synthesis, relaxivity behavior and cellular characterization of Arg8CG2, a Cu+-responsive MRI contrast agent that exhibits improved cellular uptake. This work demonstrates the utility of the octaarginine group to promote cellular uptake and retention, and the great advantage that such behavior lends to in cellulo experiments. Using this new probe, we have been able to detect differences in copper accumulation between cells bearing a mutant copper transporter and wildtype cells. The clear differences in relaxivity between Atp7a mutant and wildtype cells as detected by Arg8CG2, and the resulting variation in contrast in phantom imaging experiments, points to the applicability of this probe for in vivo imaging in disease states. We have identified a number of animal models of copper mishandling diseases, to which we can apply Arg8CG2, including the toxic milk65 and Atp7b knockout mice,66 which are models for Wilson's disease, and which exhibit markedly different copper distributions from wildtype mice.17,18 We are currently pursuing the use of Arg8CG2 in a variety of whole animal imaging studies.</p>

PubMed Author Manuscript

Generation of Alkyl Radicals: From the Tyranny of Tin to the Photon Democracy

Alkyl radicals are key intermediates in organic synthesis. Their classic generation from alkyl halides has a severe drawback due to the employment of toxic tin hydrides to the point that “flight from the tyranny of tin” in radical processes was considered for a long time an unavoidable issue. This review summarizes the main alternative approaches for the generation of unstabilized alkyl radicals, using photons as traceless promoters. The recent development in photochemical and photocatalyzed processes enabled the discovery of a plethora of new alkyl radical precursors, opening the world of radical chemistry to a broader community, thus allowing a new era of photon democracy.

generation_of_alkyl_radicals:_from_the_tyranny_of_tin_to_the_photon_democracy

Introduction<!><!>Introduction<!><!>Introduction<!>Formation of a C(sp3)-C Bond<!>Addition to C–C Double Bonds: Hydroalkylations<!>Heteroalkylation of C–C Double Bonds<!>Allylation<!>sp3–sp3 Cross-coupling<!>Other Reactions<!>Alkenylation<!>Acylation<!>Minisci-Like Reactions<!>Ipso-Substitution Reactions<!>Cyanation<!>Alkynylation<!>C–B Bond<!>C–N Bond<!>C–O Bond<!>C-Halogen Bond<!>C–S or C–Se Bonds<!>C–H Bond<!>Three/Four-Membered Rings<!>Five-Membered Rings<!>Six-Membered or Larger Rings<!>Conclusions and Outlook<!><!>Conclusions and Outlook<!>Author Contributions<!>

<p>Among all the open-shell species, carbon-centered radicals are intriguing neutral intermediates that find extensive use in organic synthesis, despite the initial distrust about their possible application.1−5 In particular, the generation of unstabilized alkyl radicals under mild conditions granting the controlled and selective outcome of the ensuing reactions has been a challenge for many years. The first and more obvious way to form such species is the homolytic cleavage of a labile C–X bond; alkyl halides appeared as the ideal choice in this respect. The real breakthrough in radical chemistry was the discovery of Bu3SnH to promote radical chain reactions as reported about 60 years ago in the reduction of bromocyclohexane.6 Reduction of an organotin halide by lithium aluminum hydride formed the reactive tin hydride in solution. In subsequent modifications of the protocol, both sodium borohydride7 and sodium cyanoborohydride8 acted as effective reducing agents. Alkyl radicals generated via tin chemistry were then used for C–C bond formation mainly via the addition to (electron-poor) olefins, the well-known Giese reaction,9−12 an evolution of the original process which made use of organomercury compounds.12,13</p><p>As illustrated in Figure 1a, tributyltin hydride has the double role of allowing the formation of Bu3Sn• as the radical chain carrier and as a hydrogen donor to close the catalytic cycle. The unique features of this catalytic cycle are attributed to the forging of stronger Sn–X and C–H bonds at the expense of the cleavage of the more labile Sn–H and C–X ones. A more quantitative aspect of this reaction can be appreciated comparing the different bond dissociation energies (BDE) associated with the steps mentioned above (see Figure 1a).14,15 More recent applications showcase the crucial role of tin intermediates in controlling the outcome of different reactions. Sn–O interactions direct the regioselective addition of the radical in the radical stannylation of the triple bond in propargyloxy derivatives,16 whereas tin radicals induced the synthesis of stannylated polyarenes via double radical peri-annulations, increasing the solubility of the products.17</p><!><p>(A) Thermal generation of radicals from alkyl halides in the Giese reaction. (B) LD50 values for selected organotin compounds. (C) Thermal generation of radicals from alcohols via xanthates (I). (D) Thermal and photochemical generation of radicals from carboxylic acids via Barton esters (II).</p><!><p>The performance of Bu3SnH was so competitive9,18−20 that more than 20 years ago it was claimed that it was improbable to have "flight from the tyranny of tin" in radical processes,21,22 a hard statement that subtly introduces the problem of the substantial toxicity and high biological activity of triorganotin compounds.23 The LD50 of 0.7 mmol/kg in murine species (see Figure 1b) combined with the long half-lives in aquatic environment represent the biggest concerns for the application of these otherwise extremely versatile species, especially in the absence of viable alternatives.24 Indeed, O-thiocarbonyl derivatives like xanthates (I, obtained from alcohols) were considered an alternative to the alkyl halides, albeit the radical generation required in most cases the use of tin hydrides (Figure 1c).3,22,25,26</p><p>Efforts in substituting toxic tin derivatives with other hydrides such as (TMS)3SiH27 or lauroyl peroxides and xanthates met some success, however, only in limited cases.28−30 Other initiators to promote tin-free radical chain reactions were organoboranes,31 thiols,32 P–H-based reagents,32 and 1-functionalized cyclohexa-2,5-dienes,32−34 but nowadays they are not commonly used in synthetic planning.</p><p>The use of metal oxidants (MnIII acetate)35 or metal reductants (TiIII catalyst36 or SmII iodide37) were sparsely used, but only in the latter case unstabilized alkyl radicals were formed from alkyl iodides.</p><p>The introduction of the Barton esters II in 1985 represented a step forward in solving the conundrum of the tyranny of tin: the conversion of the strong O–H bond of an acid into the (photo)labile O–N bond of the corresponding thiohydroxamate ester (Figure 1d).38,39 Barton esters have the advantage of being slightly colored, allowing the use of visible light irradiation to induce the cleavage of the O–N bond. The last point is significant, demonstrating the formation of alkyl radicals in a very mild way under tin-free conditions with no need of further additives, albeit Barton esters have currently a limited application. On this occasion, photochemistry showed an attractive potential for the development of novel synthetic strategies based on radical chemistry. However, Barton esters remained for several years an isolated niche. In most cases, the photochemical generation of radicals required harmful UV radiation and dedicated equipment.40 Since the milestone represented by the development of the chemistry of Barton esters, new photochemical ways were sought toward more efficient ways to generate radicals. The photon appears to be the ideal component for a chemical reaction, assuming the form of a traceless reagent, catalyst, or promoter that leaves no toxic residues in the final mixture.41−44 The breakthrough that would allow moving forward from the "tyranny of tin" to a greener "photon democracy" can be associated with the use of solar or visible photons, freely available from the sun that shines throughout the scientific world. The renaissance of the photocatalyzed processes that we have witnessed in the last years represents a significant step toward this direction.45−58</p><p>The multifaceted use of photoredox catalysis and photocatalyzed hydrogen transfer reactions expanded the range of possible radical precursors and unconventional routes for the generation of several carbon (or heteroatom based) radicals, including the challenging formation of unstabilized alkyl radicals.59−65 Consequently, in this review, we aim to present a summary of the novel ways to generate alkyl radicals by photochemical means that, in the last years66 have revolutionized the way to carry out radical chemistry. This work will focus exclusively on the reactions promoting the formation of unstabilized alkyl radicals, and not the stabilized ones, e.g., α-oxy, α-amino, benzylic, or allylic.</p><p>Figure 2 collects the main paradigmatic approaches to the photogeneration of alkyl radicals (either photochemically or photocatalyzed). The more classical, although the less employed, path to generate alkyl radicals consists of the introduction of a photoauxiliary group which renders a bond labile to a direct photochemical cleavage (Figure 2A).67 The Barton esters are the archetypal moiety belonging to this class.39 A conspicuous body of literature have been focusing on the development of suitable alkyl substituents able to facilitate redox reactions making the derivatives more oxidizable or reducible. The strategy that is followed in Figure 2B consists in the conversion of a common functional group (e.g., OH or COOH), which in most cases is tethered to the alkyl group, into a different electroauxiliary group68 (Figure 2B). As a result, the interaction of the activated species with an excited photoredox catalyst (PCSET) able to induce a single electron transfer (SET) process generates the corresponding radical ions, either by an oxidative pathway or a reductive pathway. The desired alkyl radical is then formed by fragmentation of these radical ion intermediates. The oxidative pathway is efficient when the radical precursor is negatively charged (see further Figure 3) as in the case of alkyl carboxylates and alkyl sulfinates causing the CO2 or SO2 loss, respectively, despite the fact that the exothermicity of the process is verified only in the C–C cleavage rather than the C–S cleavage.69 On the contrary, positively charged Katritzky salts were ideal candidates for the releasing of radicals via the reductive pathway (Figure 3).70</p><!><p>Different approaches for the photogeneration of alkyl radicals (A) by photochemical means through the introduction of a photoauxiliary group (B) via fragmentation of a radical cation (oxidative pathway) or anion (reductive pathway) formed by photoredox catalysis (C) via a halogen atom transfer reaction (XAT) with a photogenerated radical (D) through the photocatalyzed cleavage of a C–H bond via direct (d-HAT) or indirect (i-HAT) hydrogen atom transfer (E) by the remote-controlled C–H activation via a photogenerated heteroatom based radical (F) by a ring-opening via a photogenerated heteroatom (nitrogen) based radical.</p><p>On the left, substrates used to promote the photochemical formation of alkyl radicals divided according to the C–Y bond cleaved. The oxidation potentials (Eox, in orange) or the reduction potentials (Ered, in green) of the precursors as well as the BDE values of the bond that is broken (highlighted in gray) by direct photocleavage are reported. On the right, a selection of common photoredox catalysts with their main redox features are collected.</p><!><p>A viable alternative is the photogeneration (often from a photoredox process) of a reactive radical on a heteroatom like a silyl radical, which can exploit a halogen atom transfer reaction to afford an alkyl radical through the smooth Si-X bond formation (XAT, Figure 2C).71,72 This strategy provides an elegant way to overcome the Giese conditions in the tin mediated activation of alkyl halides. Recently, an α-amino radical was used in the same strategy, promoting the formation of alkyl radicals via C–X bond cleavage.73</p><p>A more challenging approach requires the photocatalyzed selective cleavage of a strong alkyl-H bond, via a direct hydrogen atom transfer reaction (d-HAT, Figure 2D) operated by an excited photocatalyst (PCHAT).72,74−76 The indirect version of the latter path exploits the photogeneration of a stable heteroatom based radical (i-HAT, Figure 2D) that will become the competent intermediate in the abstraction of the H atom from the alkyl moiety.76 An indirect HAT (i-HAT) may also take place by intramolecular hydrogen transfer thus releasing an alkyl radical (Figure 2E).76−81 As an alternative, the photochemical radical generation may induce a ring-opening in strained structures like cyclobutanes, to form a substituted alkyl radical (Figure 2F).82</p><p>Figure 3 showcases a collection of the main alkyl radical precursors devised for the generation under photochemical conditions of unstabilized alkyl radicals. In this figure, the radical precursors were collected depending on the C(sp3)-Y bond cleaved during the radical release. As apparent, the photochemically triggered cleavage of several C-heteroatom bonds like C–X,71,83−88 C–O,89−98 C–B,99−102 C–S,103−106 or C–N70 (Figure 3) affords carbon-centered radicals. The alkyl radical generation is granted by the very versatile photochemical tool. This feature includes particular cases such as C–Se (in alkyl selenides),107,108 C–Te [in (aryltelluro)formates,109,110 for a previous thermal generation of alkyl radical from diorganyl tellurides, see ref (111)], and C–Si (in tetra alkyl silanes and bis-chatecolates)112−114 to be added to C–Sn (in alkyl stannanes).112,115 Interestingly, even the more resilient C–H15,76 or C–C38,83,94,116−130 bonds may be cleaved for alkyl radical generation, opening up new exciting possibilities for the synthetic (photo)chemist (Figure 3).</p><p>For the clarity of the reader, each radical precursor is accompanied by its oxidation potential (EOX, in orange) or its reduction potential (ERED, in green) to guide the feasibility on the generation of the radical via the oxidative or reductive pathway (type B, Figure 2B), respectively. Since the redox potentials may vary with the nature of the alkyl group, the values reported are referred to known structures. In alternative, the BDE values of the bond that is broken by direct photocleavage (type A, Figure 2A) or by photocatalyzed hydrogen abstraction (type D, Figure 2D) are reported. Figure 3 (right part) likewise collects the redox properties of commonly used photoredox catalysts including metal-free photoorganocatalysts (POC) to be used in the oxidative/reductive pathways.131−137</p><p>The reactions collected and commented on in this review are primarily divided according to the type of the bond formed, namely the forging of C–C or C–heteroatom bonds, along with the construction of rings of different sizes. When possible, in each section, we will further categorize the reactions depending on the mechanism of the radical generation, ascribing them to the six types (A–F) described in Figure 2.</p><!><p>Photochemically generated alkyl radicals have been employed to forge C(sp3)-C(spn) bonds (n = 1–3) in an intermolecular fashion following different strategies. In most cases, a conjugate addition onto a Michael acceptor or a Minisci-like reaction occurred, albeit alkenylations, acylations, or oxyalkylations are likewise used.</p><!><p>Many reactions belonging to this class involve the nucleophilic alkyl radical addition onto an electrophilic Michael acceptor, resulting in a formal hydroalkylation of the double bond viz. the incorporation of an alkyl group (in position β with respect to the EWG group in the starting unsaturated compound) and a hydrogen atom (in position α). This reaction is usually one of the first that many authors would test during the discovery process of a new radical precursor, as testified by the plethora of reagents that are used in this transformation. Photoredox catalysis is by far the preferred approach here, especially by using the oxidation of a negatively charged precursor (oxidative pathway in Figure 2B).</p><p>A typical example is the oxidation of carboxylates138,139 that releases an alkyl radical via CO2 loss from the carboxyl radical intermediate (Scheme 1). Adamantylation of both acrylonitrile (Scheme 1a)140 and dimethyl 2-ethylidenemalonate starting from adamantane carboxylic acid 1-1 (Scheme 1b)141 were carried out following this strategy. In the former case, the authors employed 1,4-dicyanonaphthalene (DCN) as the POC under UV light irradiation, while visible light and an IrIII complex in the latter case. The approach used in Scheme 1b was also useful for the three steps preparation of the medicinal agent (±)-pregabalin.141 Also, the Fukuzumi catalyst (9-mesitylene-10-methylacridinium perchlorate, [Acr+Mes]ClO4) can promote this Giese-type reaction,142 allowing the alkylation of α-aryl ethenylphosphonates for the synthesis of fosmidomycin analogues.143</p><p>A variation of this procedure is the decarboxylative-decarbonylative process occurring on an α-keto acid 2-1 under sunlight-driven photoredox catalyzed reaction conditions (Scheme 2).125</p><p>Oxalates are another class of electron-donors having two carboxylate moieties that can be lost upon photocatalyzed oxidation. These species may be introduced in situ by reaction of the alcohol with oxalyl chloride. The process induced the cleavage of a C–O bond, and the resulting radical could be trapped by butenolide 3–1 to form the menthyl derivative 3–2 used for the enantioselective preparation of cheloviolene A (3–3, Scheme 3).144 An IrIII-based photocatalyst efficiently promoted the reaction also in this case, allowing the synthesis of quaternary centers89 and the total synthesis of trans-clerodane diterpenoids.145</p><p>Alkyl trifluoroborates stand out as another important class of easily oxidizable moieties.146 The photocatalyzed oxidation of these salts (e.g., 4–1) causes the smooth release of BF3 and the formation of the reactive alkyl radical. Such a reaction was employed to functionalize Michael acceptors under sunlight irradiation (Scheme 4a) exploiting Acr+Mes as the POC.147 Complexation of 4–4 by a chiral rhodium complex (Λ-RhS, Scheme 4b) delivered 4–5 in good yields with 97% ee.148</p><p>This synthetic strategy can be extended to neutral boronic acids or esters, upon in situ activation by a Lewis base (LB). The so formed negatively charged species is consequently more prone to oxidation, which eventually will provide the formation of the alkyl radicals. A typical example is illustrated in Scheme 5 where the boronic acid 5–1 was activated by 4-dimethylaminopyridine (DMAP) and then oxidized by an IrIII complex. The resulting cyclobutyl radical was trapped by methyl vinyl ketone to access the substituted ketone 5–2 in a good yield.100 This reaction was later scaled up under flow conditions by using the Photosyn reactor. In such a way, the authors could synthesize gram amounts per hour of the analogues of some drugs belonging to the GABA family.149</p><p>Following the examples of the carboxylate derivatives, the electron-donating species may be generated in situ by deprotonation, as in the case of sulfonamides, employed in the desulfurative conjugate addition of alkyl radicals onto Michael acceptors (Scheme 6). Again, the process is based on a photocatalyzed oxidation pathway. The starting sulfonamide (6–1) was first deprotonated by a mild base (K2HPO4), and the resulting anion was easily oxidized to a N-centered radical. Loss of N-sulfinylbenzamide generates the desired radical that gave the adduct 6–3 upon reaction with 6–2 in 75% yield.103</p><p>In some instances, the radical precursor is a neutral compound. This situation is possible only when the derivative contains a highly oxidizable or reducible moiety. 4-Alkyl-1,4-dihydropyridines (alkyl-DHPs) under PC-free conditions act as radical precursors when combined with photoexcited iminium ion catalysis (Scheme 7). Here, enal 7–1 formed a chiral iminium ion 7–4+ by reaction with amine 7–3. Cation 7–4+ upon visible light excitation oxidized the alkyl-DHP 7–2 that in turn released the radical 7–5• upon fragmentation, along with radical 7–4•. Radical recombination followed by hydrolysis gave the desired alkylated dihydrocinnamaldehyde 7–6 in a satisfactory yield with a good enantiomeric excess (Scheme 7).150 A similar Giese reaction was later proposed, where the alkyl-DHP was excited and a SET reaction with Ni(bpy)32+, acting as an electron mediator, took place. The alkyl radical derived from the radical cation of alkyl-DHP readily attacked a series of Michael acceptors.151</p><p>Looking at the other edge of the redox spectrum, easily reducible compounds were devised as radical precursors via a photocatalyzed process. As an example, the incorporation of a N-phthalimidoyl moiety in an organic compound helps its photocatalyzed reduction, ultimately leading to the release of the alkyl radical. A typical case is represented by N-(acyloxy)phthalimides.126 A stereoselective variant of this reaction was applied to the synthesis of (−)-solidagolactone (8–4, Scheme 8). Thus, the photocatalyzed reduction of phthalimide 8–1 by a RuII complex released a tertiary carbon radical. Attack to the terminal carbon of the unsaturated core present in β-vinylbutenolide 8–2 yields 8–3 with a very high diastereomeric excess. Further elaboration of compound 8–3 gave 8–4 in a single step.152 This reaction emerges as a very interesting tool to construct quaternary carbons153 and to synthesize biologically active derivatives, e.g., (−)-aplyviolene.154</p><p>Interesting results were also obtained using N-phthalimidoyl oxalates such as 9–1 in place of the N-(acyloxy)phthalimides for the generation of alkyl radicals starting from tertiary alcohols (Scheme 9).92,97 The similarity of this reaction to the one presented in Scheme 8 is striking, despite a less atom economical radical generation.</p><p>Reduction of an organic compound may be carried out even on organic iodides by using cyanoborohydride anion as the reducing agent. The reaction is chemoselective, since no alkyl bromides or chlorides could be activated following this way. Giese adducts were formed by irradiation with a Xe lamp of the reaction mixture in good yields as illustrated by the formation of 10–2 from 10–1 in Scheme 10.155 This is another interesting example to circumvent the use of tin hydrides in the activation of alkyl halides.</p><p>Alkyl chlorides can be activated using Ir(dtbby)(ppy)2PF6 in the presence of micelles. The micellar environment stabilizes the photogenerated [Ir(dtbby)•–(ppy)2] species (−1.51 V vs SCE), unable to directly reduce the alkyl chlorides (ca. −2.8 V vs SCE). A second excitation of this long-lived intermediate allows the electron transfer to the halide, which could react with different electron-poor olefins, forging a novel C–C bond. The micellar system allowed intramolecular cyclizations to form five-membered cycles.156</p><p>Reduction of the alkyl halide 11–1 could be avoided applying a halogen transfer reaction. In fact, due to the strong BDE of the Si-halogen bond, an alkyl radical is formed thanks to the action of a purposely generated silyl radical (from (Me3Si)3SiH, TTMSS) by a photoredox catalytic step. Radical addition onto an unsaturated amide (11–2) gave the 1,8-difunctionalized derivative 11–3, a key compound in the preparation of Vorinostat 11–4, a histone deacetylase (HDAC) inhibitor active against HIV and cancer (Scheme 11).157 This is the typical case where the radical is formed by the cleavage of an Alk-Br bond without the assistance of tin derivatives. It is interesting that the reaction requires a substoichiometric amount of silane to proceed. Indeed, with higher loadings the product yield decreases, possibly due to the presence of competing nonproductive pathways. A chain reaction mechanism could be envisaged; however, the quantum yield for this reaction (Φ = 0.45) does not fully clarify the mechanistic details of the transformation.</p><p>In many instances the formation of the alkyl radical arose from a direct or indirect photocatalyzed C–H homolytic cleavage. The excited state of the decatungstate anion in its tetrabutylammonium salt form (TBADT) promoted in several cases the direct chemoselective cleavage of a C–H bond.75,158Scheme 12 depicts two examples involving the hydroalkylation of acrylonitrile. Unsubstituted cycloalkanes were suitable hydrogen donors under flow conditions (yielding 12–1Scheme 12a).159 Moreover, the chemoselective cleavage of the methine hydrogen in isovaleronitrile allowed the preparation of dinitrile 12–2 in 73% yield (Scheme 12b).160</p><p>Similarly, the presence of a tertiary hydrogen was the driving force of the chemoselective TBADT-photocatalyzed C–H cleavage in several derivatives, as depicted in Scheme 13. As an example, alkylpyridine 13–1 was selectively functionalized and gave derivative 13–2 as the exclusive product in the reaction with a vinyl sulfone (Scheme 13a).161 Interestingly, the labile benzylic hydrogens present in 13–1 remained untouched under these reaction conditions. Noteworthy, steric and polar effects cooperatively operated in the derivatization of lactone 13–3. As a result only the methine hydrogen of the isopropyl group was selectively abstracted and afforded 13–4 in very high yields by reaction with fumaronitrile (Scheme 13b), albeit the seven different types of hydrogens present in 13–3.162 The C–H cleavage may also take place in branched alkanes as witnessed by the derivatization of 13–5 to form the succinate derivative 13–6 (Scheme 13c).163</p><p>In rare instances, the hydroalkylation reaction may be applied to olefins different to the usual Michael acceptors. Thus, substituted vinylpyridines were functionalized by TBADT-photocatalyzed addition of cycloalkanes. Scheme 14 showed the smooth synthesis of 14–2 starting from 14–1 simply by irradiation of the reaction mixture containing a slight excess of cyclohexane in the presence of a catalytic amount of the decatungstate salt.164</p><p>Recently, alternative PCs have been developed for the direct photocatalyzed activation of C–H bonds in cycloalkanes, namely uranyl cation165 and Eosin Y,166 both having the advantage of absorbing in the visible light region. The alkyl radical formation may be induced by a photogenerated stable radical which acts as a radical mediator. An IrIII based photoredox catalyst oxidized the chloride anion (being the counterion of the Ir complex) to the corresponding chlorine atom, which abstracted a hydrogen atom from cyclopentane, thus forming adduct 15–1 in 69% yield upon addition onto a maleate ester (Scheme 15).167</p><p>Another intriguing way to induce the cleavage of unactivated C(sp3)-H bonds is by a photocatalyzed intramolecular hydrogen abstraction. Usually a photoredox or a proton-coupled electron transfer (PCET)47 step induced the formation of a heteroatom centered radical that abstracts a tertiary C–H bond intramolecularly in a selective fashion, following a 1,5-HAT process mimicking the Hoffmann-Löffler-Freytag reaction (Scheme 16).168−170</p><p>When the reaction was applied to compound 16–1, an oxidative PCET generated a neutral amidine radical that promotes the 1,5-hydrogen atom abstraction forming a tertiary radical which is able to functionalize olefin 16–2 in a complete regioselective fashion affording 16–3 (Scheme 16a).171 The reaction was also applied to medicinally relevant molecules such as the steroid-derived trifluoroacetamide 16–4 (Scheme 16b). Despite the fact that this compound has several labile C–H bonds including tertiary C–H bonds and C–H bonds adjacent to heteroatoms, the intramolecular hydrogen abstraction followed by conjugate addition onto 16–5 gave 16–6 as the sole product.172</p><p>The remote activation of the C–H bond in the δ-position following this approach is a general reaction as demonstrated in related systems applied to amides protected with a carbamate group173 or in simple benzamide derivatives.174 In the latter case, the reaction was carried out in the presence of a chiral Rh-based Lewis acid catalyst that allowed the asymmetric alkylation of α,β-unsaturated 2-acyl imidazoles.174</p><p>The abstracting species could be likewise a photogenerated iminyl radical as illustrated in Scheme 17. Here a carbonyl group is converted in an oxime derivative (e.g., 17–1) by reaction with an α-aminoxy acid. Photocatalyzed oxidation followed by fragmentation of the resulting carbonyloxy radical gave an iminyl radical prone to a 1,5-HAT to afford a tertiary radical that upon addition to acrylate 17–2 gave compound 17–3 in 77% yield.175</p><!><p>An interesting variation of the functionalization of a double bond is the formation of a C–C bond (upon an alkylation step) followed by the formation of another C–Y bond (Y ≠ H) on the adjacent carbon. As an example, alkyl diacyl peroxides were reduced photocatalytically and the fragmentation released an alkyl radical and a carboxylate anion both incorporated in the structure of the product. Thus, 2-vinylnaphthalene 18–2 was converted into compound 18–4 in a very good yield upon reaction with lauroyl peroxide 18–1 upon an oxidative quenching process by consecutive C–C and C–O formation (Scheme 18).176 The reaction was made possible by the oxidation of the resulting radical adduct 18–3• (by RuIII, the oxidized form of the PC) that generated the cation 18–3+ that was easily trapped by the carboxylate anion previously released.</p><p>N-(acyloxy)phthalimide 19–1 as radical precursor found use in a similar multicomponent oxyalkylation of styrenes. The addition of the alkyl radical onto the vinylarene followed by the incorporation of water present in the reaction mixture afforded derivative 19–2 in 72% yield (Scheme 19).177 Noteworthy, the labile C–Br bond in 19–1 remained untouched in the process.</p><p>The use of water as the oxygen source was likewise used in the difunctionalization of aryl alkenes where the carbon-centered radical was formed by an intramolecular 1,5-HAT of a photogenerated iminyl radical.178</p><p>Performing the reaction in DMSO, allows for the use of the solvent as an oxygen donor adopting the Kornblum oxidation. The intermediate benzyl radical formed after the alkylation step reacts with the solvent and eventually forming a carbonyl group in place of a simple C–O bond. An elegant example is shown in (Scheme 20) for the synthesis of ketonitrile 20–4.179 A cycloketone oxime ester (20–1) was photocatalytically reduced, inducing a ring opening on the resulting iminyl radical. The resulting cyano-substituted alkyl radical reacted with styrene 20–2, and the addition with DMSO formed the intermediate 20–3, that, upon Me2S loss, afforded the product.</p><p>A related oxyalkylation of styrenes made again the use of the Kornblum oxidation as the last step in the synthesis of substituted acetophenones. Indeed, N-hydroxyphthalimides (e.g., 21–1) were employed as the radical source, and an IrIII complex was used as the PC, obtaining good yields even on a 7 mmol scale (64% of 21–3, Scheme 21a).180,181 Ester 21–1 was also adopted for the preparation of the aryl alkyl ketone 21–5 in 61% yield (Scheme 21b). In this case, however, the decarboxylative alkylation was applied to silyl enol ethers having the carbonyl oxygen already incorporated in the initial structure such as 21–4.182 The same process described in Scheme 21b can be carried out under uncatalyzed conditions under blue LED irradiation in the presence of an excess of NaI (150 mol %) and PPh3 (20 mol %). The reaction was based on the photoactivation of a complex formed by N-(acyloxy)phthalimide with NaI and PPh3 through Coulombic and cation-π interactions. In this case, the excitation caused the reduction of the phthalimide by a SET reaction within the complex.183</p><p>Alkylated ketones 22–3a–d were likewise obtained by the IrIII-photocatalyzed reaction between a 2-mercaptothiazolinium salt (22–1, as alkyl radical precursor) and silyl enol ethers 22–2a–d (Scheme 22).106</p><p>Lauryl peroxide (LPO, see Scheme 18) was adopted for the Ru-catalyzed three-component carbofluorination of styrenes as illustrated in Scheme 23a. The vinylic double bond of compound 23–1 derived from estrone was functionalized twice by using triethylamine trihydrofluoride Et3N·HF as the fluoride anion source to deliver the desired alkyl-fluorinated olefin 23–2 in 61% yield. The reduction of LPO is mediated by the presence of a copper salt in the role of a cocatalyst in a dual catalytic process.184 The carbofluorination was later applied to dehydroalanine derivative 23–4 by using alkyltrifluoroborates and an excess of Selectfluor as an electrophilic fluorine source (Scheme 23b). The use of a visible light POC ([Acr+Mes]ClO4) allowed for the synthesis of a wide range of unnatural α-fluoro-α-amino acids including F-Leu (23–5).185</p><p>In rare instances, two C–C bonds could be formed in the adjacent position of the double bond as in cyanoalkylations. The enantioselectivity of the reaction was controlled exploiting the capability of a copper catalyst to form complexes with chiral Box ligands. Thus, a methyl radical was obtained by IrIII-photocatalyzed reduction of phthalimide 24–1 that readily attacked styrene (Scheme 24). Meanwhile, the CuI salt incorporated Box 24–2 as the ligand, and the resulting complex reacted with the adduct radical in the presence of TMSCN. As a result, cyanoalkylated 24–3 was obtained in a good yield and in good ee.186</p><p>A particular case of cyanoalkylation was later reported in the photocatalyzed reaction between cyclopropanols and cyanohydrins having a pendant C=C bond. Oxidative ring opening of the three-membered ring followed by addition onto the double bond and cyano migration gave a series of multiply functionalized 1,8-diketones incorporating the cyano group.187</p><!><p>Allylation reactions can be easily performed by reaction of an alkyl radical with substituted allyl sulfones (mainly with 1,2-bis(phenylsulfonyl)-2-propene 25–1, Scheme 25a). The alkyl radical was generated under visible light irradiation by hydrogen abstraction from cycloalkanes by an aromatic ketone, e.g., 5,7,12,14-pentacenetetrone 25–2. Addition of a cycloalkyl radical onto 25–1 followed by sulfonyl radical elimination gave access to vinyl sulfones 25–3a–b in good yields (Scheme 25a).188</p><p>Other related reactions were designed to forge C(sp3)–allyl bonds following this simple scheme. The alkyl radical was formed by photocatalytic oxidation of hypervalent bis-catecholato silicon compounds as shown in Scheme 25b. Thus, compound 25–4 upon oxidation released the desired substituted alkyl radical, and addition onto allyl sulfone 25–5 gave the corresponding allylated derivative 25–6 in 70% yield.113 Olefin 25–5 was also used in a decarboxylative allylation of alkyl N-acyloxyphthalimides under RuII photocatalysis. The great advantage of the process was the reaction time since the allylation was completed in a few minutes at room temperature.189</p><p>A particular class of phthalimides could be employed with no need of a photocatalyst to promote the reaction. N-alkoxyphthalimide (26–1) is able to form a donor–acceptor complex with electron donor compounds, such as the Hantzsch ester HE. Upon excitation, an electron transfer occurred within the complex generating a radical anion, which released an alkoxy radical upon N–O bond cleavage. Loss of formaldehyde formed the desired alkyl radical which reacted with 26–2, to obtain product 26–3 in 60% yield (Scheme 26).127</p><p>Photocatalyzed reduction of Katritzky salts 27–1a–c obtained from the corresponding amines (Scheme 27) gave access to the allylated compounds 27–3a–c. Thus, the monoelectronic reduction of pyridinium salts 27–1a–c caused the release of the corresponding pyridines along with the substituted cyclohexyl radicals than upon trapping by 27–2 efficiently afforded acrylates 27–3a–c.190</p><p>A remote allylation under visible light irradiation was devised starting from amide 28–1 making use of eosin Y (EY) as the PC (Scheme 28). The excited EY is able to reduce 28–1 thanks to the electron-withdrawing capability of the substituted phenoxy group on the nitrogen of the amide. The amidyl radical formed upon fragmentation of 28–1•– gave rise to a tertiary radical upon 1,5-HAT, allowing the remote allylation, forming 28–2 in 75% yield.191</p><p>A different approach involved the use of trifluoromethyl-substituted alkenes (e.g., 29–1) that upon addition of the alkyl radical gave access to valuable gem-difluoroalkenes such as 29–2a–b (Scheme 29). The oxidation of alkyltrifluoroborates was here assured by the organic photocatalyst 4CzIPN, leading to nonstabilized primary, secondary, and tertiary radicals. The defluorinative alkylation resulted from the reduction of the radical adduct, followed by an E1cB-like fluoride elimination.192</p><p>A dual catalytic approach was designed for valuable allylation using vinyl epoxides as allylating agents (Scheme 30). The mechanism was investigated by quantum mechanical calculations [by DFT and DLPNO–CCSD(T)] and supported an initial complexation of Ni0 to 30–2 that quickly underwent a SN2-like ring opening, followed by the incorporation of the alkyl radical formed by DHP-derived compounds 30–1a,b into the metal complex. Allyl alcohols 30–3a,b were then formed by inner sphere C(sp2)-C(sp3) bond formation from the resulting NiIII complex.193</p><!><p>Another intriguing possibility offered by the photochemical approach to alkyl radicals is the formation of a C–C bond by a sp3–sp3 cross-coupling reaction. The transformation could lead to novel pathways to interesting targets, as represented by the synthesis of the drug tirofiban in only four steps, starting from easily available compounds. The protocol made use of two consecutive photocatalyzed reactions applying a metallaphotoredox strategy (Scheme 31). The key step is the coupling between carboxylic acid 31–1 and alkyl halide 31–2. The halide is first complexed by a Ni0 catalyst and the resulting NiI complex trapped the alkyl radical (obtained by photocatalyzed decarboxylation) to yield a NiIII complex that in turn released the sp3–sp3 coupled product 31–3 after desilylation with TBAF. The desired tirofiban was then obtained by elaboration of 31–3 in two subsequent steps.194</p><p>Another example of a C(sp3)–C(sp3) cross-coupling process is the reaction between alkylsilicates and alkyl halides. As in the previous case, a dual catalytic Ir/Ni system was required.195 The alkyl radical may be likewise generated from an alkyl halide by a halogen atom transfer with a photogenerated silyl radical (from a silanol). The radical that is hence formed could be coupled with another alkyl bromide, e.g., methyl bromide, using a Ni0 catalyst to perform valuable methylation reactions.196 Aliphatic carboxylic acids were used to form alkyl-CF3 bonds via a photocatalyzed reaction making use of Togni's reagent as the trifluoromethylating agent. The reaction was promoted under visible light irradiation employing an IrIII photocatalyst coupled with a CuI salt. This process tolerates various functionalities including olefins, alcohols, heterocycles, and even strained ring systems.197</p><p>The alkylation of a benzylic position in N-aryl tetrahydroisoquinoline 32–1 was reported following two different approaches (Scheme 32). The first allowed the reaction of an unactivated alkyl bromide (32–2) by the excitation of a Pd0 complex. Compound 32–1 was oxidized in the catalytic cycle, and the resulting α-amino radical coupled with the isopropyl radical to form 32–4 in 81% yield (Scheme 32a).198 An alternative heavy-metal-free route catalyzed by a dye-sensitized semiconductor is depicted in Scheme 32b. Excitation of an inexpensive dye (erythrosine B) caused the reduction of titanium dioxide that in turn was able to reduce phthalimide 32–3 that eventually yielded quinoline 32–4.199</p><!><p>In particular cases, a C=N bond can be made sufficiently electrophilic to undergo alkyl radical addition as in the case of N-sulfinimines, exploited for the preparation of protected amines. A high degree of diastereoselectivity can be obtained when starting from chiral N-sulfinimines (33–2a–c, Scheme 33). Thus, the asymmetric addition of an isopropyl radical (formed from derivative 33–1) onto 33–2a–c allowed the isolation of sulfinamides 33–3a–c in good yields.200</p><p>The alkylation of related imines can be carried out by using ammonium alkyl bis(catecholato)silicates as the radical precursors under metal-free conditions adopting 4CzIPN as the POC201 or by using potassium alkyltrifluoroborates in the alkylation of N-phenylimines.202</p><p>Another particular case is the alkylative semipinacol rearrangement devised for the synthesis of 2-alkyl-substituted cycloalkanones. The reaction involved the photocatalytic reaction between TMS protected α-styrenyl substituted cyclic alcohol 34–2 and the unactivated bromoalkane 34–1 (Scheme 34a). The reaction was promoted by the dimeric gold complex [Au2(dppm)2]Cl2. This complex is able to reduce 34–1 (ca. −2.5 V vs SCE) despite having an oxidation potential in the excited state considerably lower for the reaction to occur (ca. −1.63 V vs SCE). This can be explained by the formation an inner sphere exciplex between the excited dimeric catalyst and 34–1 that promotes the otherwise thermodynamically unfeasible redox process, generating an AuI–AuII dimer and 34–4•. The combination of the latter species formed an AuIII complex that induces a semipinacol rearrangement coupled with C(sp3)–C(sp3) reductive elimination, which furnished 34–3 in 84% yield (Scheme 34b).203</p><p>A similar reaction was later developed starting from cycloalkanol-substituted styrenes and N-acyloxyphthalimides under IrIII photocatalysis.204</p><!><p>The reaction between an alkyl radical with a cinnamic acid followed by loss of the COOH group is one of the more common approaches to promote an alkenylation reaction. Thus, the radical formed from salt 35–3 attacked the benziodoxole adduct 35–2, synthesized from acrylic acid 35–1. The reaction yielded 83% of the diphenylethylene derivative 35–4 upon a deboronation/decarboxylation sequence (Scheme 35).205 The benziodoxole moiety gave efficient results in promoting the radical elimination step, while other noncyclic IIII reagents were ineffective.</p><p>Different decarboxylative alkenylations have been reported by changing the radical source and the photocatalyst (Scheme 36). The homolytic cleavage of an alkyl-I bond has been promoted by a CuI complex and the resulting cyclohexyl radical afforded styrene 36–3 in 68% yield upon addition onto cinnamic acid 36–1 (path a).206 The same product may be formed as well starting from the same substrate by using phthalimide 36–2 under visible light irradiation with the help of an IrIII207 (path b) or a RuII photocatalyst.208 As an additional bonus, the formation of adduct 36–3 was obtained with a preferred E configuration.</p><p>The alkenylation may mimic a Heck reaction as in the visible light-induced Pd-catalyzed reaction between a vinyl (hetero)arene and an α-heteroatom-substituted alkyl iodide or bromide (see Scheme 37). Here, the generation of the radical is induced by the reduction of the TMS-derivative 37–1 by the excited Pd0 species. Radical addition onto 37–2 followed by β-H-elimination from the adduct radical delivered allyl silane 37–3 in 81% yield.209 Noteworthy, the same reaction did not take place under usual thermal Pd-catalysis.</p><p>Alkylation of styrenes could be carried out using an inexpensive palladium source (Pd(PPh3)4) with no need of any base or classical photocatalyst. The reaction was promoted by visible light, adopting N-hydroxyphthalimides as radical sources.210 Other visible light Pd promoted alkenylations include the reaction of vinyl arenes with carboxylic acids211 or tertiary alkyl halides212 as radical precursors. Other metal catalysts, however, were helpful for the substitution of a vinylic hydrogen atom with an alkyl group. In this respect, a dinuclear gold complex was employed for the activation of an alkyl bromide to promote a photocatalyzed Heck-like reaction.213 The synergistic combination of a POC and a cobaloxime catalyst promoted the photocatalyzed decarboxylative coupling between 38–1 and styrene 38–2 to give the alkenylated product 38–3 in 82% yield and with a complete E/Z selectivity as illustrated in Scheme 38.214</p><p>The addition of the alkyl radical may take place even on substituted alkenes via an ipso-substitution reaction. An example is shown in Scheme 39 where a vinyl iodide (39–2) is used for an alkenylation by the reaction with a radical generated from silicate 39–1, obtaining compound 39–3. The RuII photocatalyst in the dual catalytic system has the role of generating the radical, while the Ni0 catalyst activates the C(sp2)-I bond.215</p><p>Alkenylation of alkyl iodide 40–1 can also take place starting from an alkenyl sulfone (40–2). Also in this case, an ipso-substitution is central to the novel bond formation and the Pd0 catalyst formed the radical by a SET reaction with 40–1. After the addition of the radical onto 40–2, the sequence is completed by the elimination of a sulfonyl radical affording 53% yield of 40–3 (Scheme 40).216</p><p>Alkyl bromides were used in alkenylations by reaction with vinyl sulfones made possible by the photocatalytic generation of silicon centered radical that in turn formed the alkyl radical by a halogen atom transfer reaction.217</p><!><p>Acylation owes its importance to the possibility to convert an alkyl radical into a ketone, a reaction that proceeds in most cases with the intermediacy of an acyl radical.218,219 A classical approach is based on the homolytic cleavage of an alkyl-I bond followed by carbonylation with CO and reaction with electrophiles of the resulting nucleophilic acyl radical. Scheme 41 illustrates the concept. Irradiation of iodide 41–1 with a Xe lamp in the presence of CO (45 atm) and a Pd0 complex led to an electron transfer reaction which formed an alkyl radical that, upon carbonylation and addition onto phenyl acetylene, gave ynone 41–2 in 63% yield.220 The reaction is supposed to proceed via a photoinduced electron transfer from the Pd0 catalyst to the iodoalkane, furnishing a PdII species and the alkyl radical. The carbon-centered radical promptly reacts with CO to generate an acyl radical. The PdI catalyst intervenes here again to couple the acyl derivative with the alkyne, preserving the triple bond in the final product. This reaction was later applied to the acylation of styrenes to give the corresponding enones.221 The electrophilic nature of the alkyne could be exploited if the moiety is placed in the same reagent bearing the iodide. In this case, the first reaction observed was an intramolecular cyclization forming an alkenyl radical which eventually reacted with CO, furnishing an α,β-unsaturated ketone.222</p><p>A reductive step induced the generation of the alkyl radical through an IrIII-photocatalyzed C–N bond activation in pyridinium salt 42–1 (Scheme 42). Trapping of the alkyl radical with CO followed by addition onto 1,1-diphenylethylene gave access to the Heck-type product 42–2 with no interference by the 2,4-dioxo-3,4-dihydropyrimidin-1-yl ring.223</p><p>The alkyl radical to be carbonylated was likewise formed starting from a cycloalkane for the preparation of unsymmetrical ketones via radical addition onto Michael acceptors. The reaction proceeded via a photocatalyzed decatungstate hydrogen atom transfer reaction224 When cyclopentanones were subjected to the photocatalyzed C–H activation, a regioselective β-functionalization occurred. Thus, 1,4-diketones 43–3a–c were smoothly formed by reaction of the photogenerated acyl radical 43–1• onto Michael acceptors 43–2a–c (Scheme 43).225</p><p>Unsymmetrical ketones have been likewise formed by carbonylation of alkyl radicals generated from organosilicates by using 4CzIPN as POC under visible-light irradiation.226</p><p>Potassium alkyltrifluoroborates were extensively used for acylation reactions having recourse to a dual photocatalytic system. The unstabilized alkyl radical was generated from trifluoroborate 44–1 with the help of an IrIII PC (Scheme 44). Meanwhile, the acid 44–2 was converted in situ into a mixed anhydride (by reaction with dimethyl dicarbonate, DMDC) that was activated by a Ni0 complex. Addition of the alkyl radical onto the resulting complex led to the acylated product 44–3.227 In a similar vein, Ir-photoredox/nickel catalytic cross-coupling reactions were devised by using acyl chlorides228 and N-acylpyrrolidine-2,5-diones229 as acylating reagents.</p><p>A Ni/Ru, dual-catalyzed amidation protocol was possible thanks to the coupling between an alkylsilicate and an isocyanate. Even in the latter case, the alkyl radical attacked the complex formed between the isocyanate and a Ni0 species and, as a result, the mild formation of substituted amides took place.230</p><p>The acylation of the radical was also exploited for the synthesis of esters. This elegant approach involves the generation of radicals from unactivated C(sp3)–H bonds (e.g., in cycloalkanes). The hydrogen abstraction on cycloalkanes was induced by a chlorine atom released from the photocleavage of the complex formed between chloroformate 45–1 and a Ni0 complex, allowing one to synthesize scaffolds with different ring sizes (45–2a–d in Scheme 45).231</p><!><p>A fundamental transformation for the construction of C(sp2)-C(sp3) bond is the Minisci reaction, where the functionalization of heteroaromatics took place by substituting a H atom with an alkyl group. The reaction was widely investigated in the last years and mainly involves the functionalization of a nitrogen-containing heterocycle.232 An interesting example is the methylation reported in Scheme 46.94 A methyl radical was formed by using a peracetate such as 46–1. The protonation of 46–1 by acetic acid facilitates a PCET reduction of the peracetate by the IrIII PC. A double fragmentation ensued, and the resulting methyl radical may attack the protonated form of biologically active heterocycles (e.g., fasudil 46–2) in a mild selective manner to afford 46–3 in 43% yield.94</p><p>Another approach made use of an alkyl boronic acid as the radical precursor. The process is initiated by the RuII-photocatalyzed reduction of acetoxybenziodoxole (BI-OAc) that liberated the key species ArCOO• (Scheme 47). Upon addition onto an alkyl boronic acid, this ortho-iodobenzoyloxy radical made available the alkyl radical that in turn functionalized pyridine 47–1 in position 2 in 52% yield (47–2, Scheme 47).233</p><p>The generation of the alkyl radical from boron-containing derivatives was made easier starting from alkyltrifluoroborates. A POC (Acr+Mes) is, however, required, but in all cases, the regioselective functionalization of various nitrogen-containing heterocycles was achieved.234 A related chemical oxidant-free approach process was later developed where alkyl radicals were formed by merging electro and photoredox catalysis.235</p><p>Alkyl halides are versatile substrates for the photoinduced functionalization (e.g., butylation) of lepidine 48–1 (Scheme 48). An uncatalyzed redox process is a rare occurrence here, since alkyl halides reduction is more demanding. This drawback can be overcome by the adoption of a dimeric AuI complex (see Scheme 34) that upon excitation coordinates an unactivated haloalkane promoting an inner sphere PET. This interaction pushes the activation of R-Br despite its larger Ered with respect to the PC (Scheme 48a).236 A different approach promoting the homolytic cleavage of the R-I bond is shown in Scheme 48b. Decacarbonyldimanganese Mn2(CO)10 was cleaved upon visible light irradiation, and the resulting Mn-based radical was able to abstract the iodine atom from an alkyl iodide thus generating the desired butyl radical. This route was smoothly applied to the late-stage functionalization of complex nitrogen-containing substrates.237 Moreover, the activation of alkyl halides may be obtained by the photogeneration of a silyl based radical derived by TTMSS (Scheme 48c, see also Scheme 11). The robustness and the mildness of this approach was witnessed by the broad substrate scope and the compatibility of several functional groups present in the radical.238 The use of acidic conditions (required to make the nitrogen heterocycle more electrophilic) may however be avoided. Excited [Ir(ppy)2(dtbbpy)]PF6 was sufficiently reducing to convert alkyl iodides to alkyl radicals under basic conditions by combining conjugate and halogen ortho-directing effects.239</p><p>In general, lepidine is the preferred substrate to test new ways for the C–H alkylation of heteroarenes. Accordingly, adamantane carboxylic acid 49–1 served for the visible light induced synthesis of 49–3 starting from lepidine 49–2 (Scheme 49). An IrIII PC was adopted to alkylate various nitrogen heterocycles, making use of a large excess of persulfate anion as the terminal oxidant (path a).240 The presence of a PC is not mandatory for the adamantylation reaction with (bis(trifluoroacetoxy)-iodo)benzene as starting material. This compound in the presence of a carboxylic acid gave the corresponding hypervalent iodineIII reagent that upon irradiation generates the alkyl radical. The TFA liberated in the process was crucial for the activation of the nitrogen heterocycle and adduct 49–3 was isolated in 95% yield (path b).241</p><p>Very recently, an interesting approach for the generation of alkyl radicals from the C–C cleavage in alcohols was reported making use of a CFL lamp as irradiation source. The combination of 2,2-dimethylpropan-1-ol (50–1) with benziodoxole acetate (BI-OAc) gave adduct 50–3. Photocatalytic reduction of compound 50–3 released and alkoxy radical that upon fragmentation formed a tbutyl radical that reacted with N-heteroarene 50–2 to form 50–4 in 57% yield (Scheme 50).130</p><p>The use of hypervalent iodineIII in promoting the decarboxylation of R-COOH was effective in the derivatization of drugs or drug-like molecules. As a result, the quinine analogue 51–2 was formed in a 76% yield from quinine 51–1, utilizing Acr+Mes as the POC (Scheme 51).242</p><p>Azoles can be adamantylated starting from adamantane carboxylic acid by a dual catalytic approach (Acr+Mes as the POC and [Co(dmgH)(dmgH2)Cl2] as the cocatalyst)243 or simply C2-alkylated under photoorganocatalyzed conditions.244</p><p>The photocatalyzed reduction of N-(acyloxy)phthalimide 52–1 induced by an IrIII* complex is an alternative approach for the functionalization of N-heterocycles such as 2-chloroquinoxaline 52–2 to form the cyclopentenyl derivative 52–3 (Scheme 52).245</p><p>The reductive pathway is feasible even when the generation of the alkyl radical was carried out starting from the redox-active pyridinium salt 53–1. In this case, the obtained cycloalkyl radical gave a regioselective addition onto 6-chloroimidazo[1,2-b]pyridazine 53–2 to yield 53–3 under mild conditions (Scheme 53).70</p><p>The alkyl radical could be formed even from simple hydrocarbons via hydrogen atom transfer reaction. A valuable example is reported in Scheme 54. The hypervalent iodine oxidant PFBI–OH is reduced by an excited RuII complex generating a carbonyloxy radical that acted as hydrogen atom abstracting agent. Functionalization of isoquinoline 54–2 by the resulting radical (derived from 54–1) afforded adduct 54–3 in 65% yield (>15:1 dr).246 The high selectivity observed in the functionalization of 54–1 was ascribed to the slow addition of the tertiary alkyl radical possibly formed onto 54–2.246 The direct (rather than indirect) C–H cleavage in cycloalkane was possible by using decatungstate anion as PC. Various nitrogen-containing heterocycles were then easily derivatized even under simulated solar light irradiation.247</p><p>PFBI–OH was likewise used for the remote C(sp3)–H heteroarylation of alcohols (Scheme 55). As an example, the reaction of pentanol with PFBI–OH gave adduct 55–1 that was reduced by the photocatalyst releasing the alkoxy radical 55–2•. 1,5-HAT and addition onto protonated phthalazine 55–3 afforded adduct 55–4 and the functionalized heterocycle 55–5 from it in 72% yield after sequential oxidation and deprotonation.248</p><p>As previously stressed, an acid is often required for an efficient Minisci-like reaction. To overcome this problem the alkylation may be carried out on the corresponding N-oxide derivatives as it is the case of pyridine N-oxides (56–2, Scheme 56). The radical is generated from a trifluoroborate salt (56–1) and the alkylation is regioselective in position 2 (forming compound 56–3).249 The process is efficient thanks to the photocatalytic degradation of BI-OAc that promoted a hydrogen abstraction, operated by the resulting carbonyloxy radical, on the Minisci radical cation adduct.</p><p>On the other hand, the pyridine N-oxide 57–1 can be acylated in situ with suitable acyl chlorides to furnish the electron-poor 57–2a–c+ derivatives. Photocatalytic reduction of these intermediates leads to the generation of alkyl radicals prone to attack the pyridine nucleus itself in the ortho position resulting in a decarboxylative alkylation (57–3a–c, Scheme 57).122</p><!><p>The forging of an alkyl-sp2 bond (e.g., an alkyl-Ar bond) is undoubtfully one of the most crucial goals pursued by a synthetic organic chemist. Alkyl radicals generated via different mild routes can be successfully employed for the arene ipso functionalization, given the presence of a suitable group X on the (hetero)aromatic ring that directs the selective formation of a new Ar–C bond at the expense of an Ar-X bond. Dual catalysis (with the help a Ni-based complex) is one of the preferred approaches.</p><p>In a recent example, the hydrogen atom transfer ability of the excited TBADT catalyst (see also Scheme 12) is used to form an alkyl radical starting from different aliphatic moieties (see Scheme 58).250 The combined action of the tungstate anion and the nickel catalyst (Ni(dtbbpy)Br2) allowed the coupling of (hetero)aromatic bromides with unactivated alkanes, overcoming their high bond dissociation energies (ca. 90–100 kcal/mol) and low acidities. Both linear (41–56% yield) and cyclic (57–70% yield) alkanes could be functionalized with a vast range of competent partners. Interestingly, the radicals are generated preferentially on the less sterically demanding secondary carbons in alkanes, affording a remarkable selectivity.</p><p>The scope of this method could be proved by the functionalization of natural products and drugs, such as in the preparation of the bicyclic derivative 58–3a (61% yield) and the N-Boc protected epibatidine alkaloid 58–3b (28%, Scheme 58).250 A very similar approach was later reported for the dual photocatalytic formation of an Ar–C bond starting from aryl bromides and cycloalkanes.251</p><p>Another dual-catalytic approach allowed the coupling reaction of aryl bromides (59–2, Scheme 59) and alkyl sulfinates (59–1), in the presence of Ni(COD)2 and tetramethylheptanedione (TMHD, Scheme 59a) to give 59–3 in 84% yield under air.104 The photogenerated radical was trapped by the Ni complex that mediated the coupling with the aryl halide 59–2. The method was then applied to the synthesis of 59–5, selective ATP-competitive inhibitors of the casein kinase 1δ, an enzyme related to the regulation of the circadian rhythm (Scheme 59b).104</p><p>A very similar strategy to access C(sp3) radicals involves the photoredox induced cleavage of alkyl oxalate 60–1, starting from the corresponding alcohols (see Scheme 60, see also Scheme 3).252 The rapid in situ formation of the oxalate (without purification) was followed by the metallaphotoredox sequence based on Ni catalysis, allowing to obtain the C(sp2)-C(sp3) coupling to give derivatives 60–3a–e in good yields.</p><p>The advantage of the use of potassium and ammonium bis-catecholato silicates relies in the smooth generation of unstabilized primary and secondary alkyl radicals to be engaged in dual catalysis.253,254 An example is the consecutive functionalization of bromo(iodo)arene 61–2 (Scheme 61, see also Scheme 25) for the preparation of 61–4 where the radical (from 61–1) is trapped by Ni0 (stabilized by a phenanthroline ligand). The synthesis of 61–3 can be achieved in high yields on 10 mmol scale with reduced effect on yield (75%) and selectivity (98%). The crude bromide 61–3 was further functionalized by a second Ni/photoredox cross-coupling of the alkylsilicate 61–5, affording product 61–4 in 66% yield.255 The procedure was extended successfully to alkyl triflates, tosylates and mesylates,256 and to brominated borazaronaphthalene cores.257 The latter approach was crucial to access previously unknown isosteres of azaborines.</p><p>The action of a silyl radical on an alkyl bromide 62–1 forms an alkyl radical that, again with the help of a Ni based catalyst, reacted with aryl bromides 62–2 (Scheme 62, see also Scheme 11). The scope of products 62–3 that can be obtained is varied and includes both aromatic and heteroaromatic substrates, along with cycloalkanes of different size.71</p><p>A peculiar case is when the ipso-substitution took place via a radical rearrangement such as shown in Scheme 63.258 Thus, the heteroaromatic sulfonamide 63–1 was subjected to the Finkelstein reaction, obtaining the corresponding iodide 63–2 that acted as the source of radical able to induce a Smiles rearrangement via 63–5•. Intermediate 63–5• has lost its aromaticity; however, the radical has become tertiary, gaining further stabilization from the ester group nearby. Restoration of the aromaticity is followed by a presumable hydrogen atom transfer to obtain compound 63–6 in 95% yield.258</p><p>Dual photoredox/nickel catalysis was successfully applied to couple β-trifluoroboratoketones 64–1 with aryl bromides 64–2a–f (Scheme 64, see also Scheme 4). Arylated compounds 64–3a–f were efficiently prepared with substituents of different electronic nature on the aryl ring.259</p><p>Potassium tetrafluoroborate salts have been applied to generate secondary alkyl radicals via Ir photocatalysis coupled with Ni.260 However, they were found to be likewise suitable for cross-coupling reactions devoted to the forging of quaternary carbon centers (Scheme 65) without the need of using reactive organometallic species.261 In the adamantylation of bromides 65–1a–d better yields were obtained when the aryl ring was substituted with electron-withdrawing groups.</p><p>An interesting application of this synthetic strategy is the functionalization of 7-azaindole pharmacophores with cycloalkyl scaffolds to improve the drug likeness of the azaindole core structure. Different potential drug candidates (66–3a–c, Scheme 66) were prepared via dual photocatalysis in a flow setup varying the dimension and substitution of the ring.262</p><p>In a similar way, a DHP-functionalized cyclohexene 67–1 was used to generate a secondary alkyl radical. In this case, the authors promoted the oxidation of 67–1 by using the strongly oxidizing 4CzIPN photocatalyst. Coupling with bromopyridine 67–2 gave substituted pyridine 67–3 in moderate yields (Scheme 67, see also Scheme 7).118</p><p>DHP-derivatives (68–1) may be used in ipso-substitution reaction even in the absence of a photocatalyst (Scheme 68). Violet light LED illumination directly excited didehydropyridine 68–1, fueling electrons to the NiII species which formed the catalytic competent Ni0 along with the desired radical by the fragmentation of 68–1•+. Noteworthy, the alkyl aromatic 68–3 was then formed where the use of electron-withdrawing groups on the ring contributes to the good yields of the process.263</p><!><p>Alkyl radicals have been used for the synthesis of alkyl nitriles by using different cyanide sources. The cyanation may be carried out in the presence of cyanide anion as tetrabutyl ammonium salt (TBACN). The C–C bond formation here may be carried out under very mild conditions by using the inexpensive precatalyst CuI, starting from unactivated alkyl chlorides (e.g., 69–1a–c, Scheme 69). Probably, a CuI-cyanide adduct is the species that was excited and engaged an electron transfer reaction with the alkyl halide to form a CuII-cyanide adduct. This intermediate combines with the alkyl radical formed to release nitriles 69–2a–c. The CuI-halide complex formed in the reaction restores the initial photocatalyst by exchange with the cyanide anion.264</p><p>TMSCN was instead used for the remote δ-C(sp3)-H cyanation of alcohols under Ir/Cu-photocatalyzed conditions. The reduction of an N-alkoxypyridinium salt generated an alkoxy radical that upon intramolecular 1,5-HAT formed an alkyl radical that is cyanated with the help of the copper catalyst.265</p><p>A typical cyanation procedure, however, makes use of tosyl cyanide as cyanating agent. Thus, the radical obtained by oxidation of trifluoroborate 70–1a (by excited Acr+Mes)266 or acid 70–1b (by riboflavin tetraacetate RFTA)267 was trapped by tosyl cyanide to afford nitrile 70–2 by a substitution reaction (Scheme 70).</p><p>A related RuII-photocatalyzed cyanation employing Ts-CN starting from alkyl trifluoroborates but requiring BI-OAc as a mild oxidant has been likewise reported.268</p><p>An elegant way to forge an alkyl-CN bond required the photocatalyzed elaboration of cyanohydrines 71–1a–d. At first, the interaction of the OH group with the sulfate anion (generated by the decomposition of persulfate anion) allowed its oxidation by a proton-coupled electron transfer (PCET) process promoted by an in situ formed IrIV species. Alkoxy radicals 71–2•a–d were then formed and promoted a regioselective cyanation of remote C(sp3)–H bonds by a 1,5-HAT followed by cyano migration to form cyanoketones 71–3a–d (Scheme 71).269</p><!><p>Direct alkynylation of photogenerated alkyl radicals could be accomplished utilizing a reagent or catalyst that activates the alkyne moiety, making it more prone to the forging of a novel C(sp3)–C(sp) bond. One of the first strategies that were employed made use of benziodoxole-functionalized alkynes to promote the reaction with the alkyl radical.270 A representative case is illustrated in Scheme 72. [Ru(bpy)3](PF6)2 promoted the alkyl radical formation from trifluoroborate salt 72–1 that upon addition onto the alkynyl derivatives 72–2a–d induced the alkynylation via the intermediacy of vinyl radicals 72–3•a–d. This deboronative alkynylation strategy could be performed in neutral DCM:water (1:1) at room temperature, giving access to the alkynylation of primary, secondary, and tertiary derivatives. To further prove the mildness of the conditions used, the authors carried out the reaction in PBS at pH 7.4 in the presence of biomolecules such as amino acids, but also single-stranded DNA and proteins (e.g., bovine serum albumin), obtaining satisfactory yields ranging from 68 to 86% of selectively alkynylated product.270</p><p>The nature of the substituents on the alkynylbenziodoxole reagent were proved to determine the outcome of the alkynylation process. The electron-rich compounds performed better in the photocatalyzed transformation, both as radical acceptor and oxidative quencher of the RuII* photocatalyst.271</p><p>A similar strategy to the one mentioned before consists in the IrIII-photoredox-catalyzed alkynylation of carboxylic acids 73–2 (see Scheme 73a, path a).272,273 In this case benziodoxole derivatives 73–1 were again used to activate the sp carbon of the alkyne to the radical attack, affording good yields of products 73–3. Following these results, they developed a reaction to synthesize ynones 73–4 utilizing the same reaction conditions in the presence of gaseous CO (see Scheme 73a, path b and Section 2.2.2). Gram-scale reactions and late-stage functionalization of natural terpenoids such as ursolic acid (73–5, Scheme 73b) were likewise reported.273</p><p>Alkynyl sulfones were extensively employed as alkynylating reagents, with a mechanism like the one described in Scheme 72. Alkynyl phenyl sulfone was used in combination with N-acyloxyphthalimide derivatives as radical precursors in a RuII-photocatalyzed reaction that gave direct access to TIPS-substituted alkynes.274N-Phthalimidoyl oxalates and tolyl alkynyl sulfones were found to be competent for the reaction (even for the preparation of internal alkynes having quaternary carbons),275,276 the latter even in combination with pyridinium salts as radical precursors.277 The consecutive photoredox decarboxylative coupling of doubly functionalized adipic acid derivatives with alkynyl phenyl sulfones induced the cascade formation of interesting cyclic derivatives with an exo double bond (Scheme 74). In this case, compound 74–1 underwent two efficient consecutive photoredox decarboxylative couplings leading first to alkyne 74–3 that it was subjected to radical cyclization to form radical 74–4• and styrene 74–5 from it.278 The authors reported the formation of five-membered rings via the consecutive formation of two C–C bonds, along with one example showing the application to the synthesis of six-membered derivatives (31% yield).278</p><p>In rare instances alkynyl bromides could be used as sp counterpart in the radical addition of alkyl derivatives obtained from the oxidative decomposition of various Hantzsch esters under visible light conditions promoted by 4CzIPN.279</p><p>The versatility of the photocatalytic method, however, allowed to obtain functionalized alkynes starting from terminal alkynes (Scheme 75). The first approach involves the UV light induced cleavage of the C–I bond in iodide 75–1 (used in large excess) in basic aqueous media. Addition of the cyclohexyl radical onto alkyne 75–2 followed by the incorporation of the iodine atom gave vinyl iodide 75–3. The strong basic conditions used (NaOtBu) coupled with heating (up to 50 °C) favored an elimination of HI to yield the desired alkyne 75–4 under metal-free conditions (Scheme 75a).280 Visible-light (450 nm) was used in the copper-catalyzed coupling of an alkyl iodide (75–5) and again a terminal alkyne (75–6, Scheme 75b). The success of the reaction was ascribed to the use of terpyridine ligand 75–8 that avoided the photoinduced copper-catalyzed polymerization of the starting substrates. Probably, the reaction started by the excitation of the first formed copper acetylide that upon SET with 75–5 promoted the synthesis of alkyne 75–7 in high yields.281</p><!><p>Borylation of an alkyl derivative to access differently substituted boron containing compounds can be carried out under mild conditions, employing different photochemical approaches. Thus, the alkyl radical formed from an N-hydroxyphthalimide 76–1 (derived from dehydrocholic acid) may be trapped either by bis(pinacolato)diboron (B2pin2) to give the corresponding alkyl pinacol boronates 76–2 (Scheme 76, path a) or by tetrahydroxydiboron (B2(OH)4) followed by treatment with KHF2 to give alkyl tetrafluoroborates (Scheme 76, path b).282</p><p>A variation of the previous methodology involves the irradiation of N-hydroxyphthalimide esters 77–2 in the presence of B2cat2 (77–1) with the help of N,N-dimethylacetamide (DMAc) as the solvent under uncatalyzed conditions (Scheme 77). These components formed a heteroleptic ternary complex able to be excited by blue light and ultimately leading to the corresponding benzo[1,3,2]dioxaborole 77–4 that upon treatment with pinacol and TEA released the desired pinacol boronic ester 77–3. The functionalization of a series of drugs and natural products, such as pinonic acid and fenbufen were likewise effective, underlying the broad scope and functional group tolerance of the method.283</p><p>Two related approaches were later developed and involve the irradiation of the ternary complex formed by differently substituted N-alkyl pyridinium salts, B2cat2 and DMAc. The reaction gave again pinacol boronic esters in what is considered a deaminative protocol for the borylation of aliphatic primary amines since the latter compounds were used for the synthesis of the pyridinium salts.284−286</p><p>Interestingly, 2-iodophenyl thionocarbonates were later adopted as radical precursor for the preparation of boronic ester via photocatalyzed reaction with B2cat2 (Scheme 78).95 The strategy is based on the photoinduced reduction of compound 78–1 that upon iodide anion elimination formed aryl radical 78–2• that underwent a 5-endo-trig cyclization causing the release of benzo[d][1,3]oxathiol-2-one 78–3 and alkyl radical 78–4•. Usual borylation gave boronic ester 78–5 in 85% yield.</p><!><p>Diverse structural motifs based on the C–N bond such as hydrazine and hydrazide cores were accessed by the photochemical addition of alkyl radicals onto the N=N of azodicarboxylates. TBADT-photocatalyzed HAT was applied to synthesize hydrazines by the coupling of cycloalkyl radicals with diisopropyl azodicarboxylate (DIAD). A synthetically challenging three component reaction can be achieved in the presence of CO, allowing the synthesis of the corresponding hydrazides.287</p><p>The C–H amination can be smoothly achieved even starting from light hydrocarbons, such as methane (79–1, Scheme 79), with ditert-butylazodicarboxylate (DBAD, 79–2) in the presence of CeIII salts. This inexpensive photocatalyst furnished the desired product 79–3 in 63% yield, with a turnover number up to 2900. The authors proposed a ligand-to-metal charge transfer excitation between the cerium salt and trichloroethanol as the source of alkoxy radicals that acted as hydrogen atom transfer agents.288</p><p>Aminated alkanes can be obtained by reacting aliphatic carboxylates with DIAD making use of Acr+Mes as a photoredox catalyst.289 A cerium catalyst was adopted for the generation of several alkyl radicals starting from carboxylic acids, under basic conditions, allowing for the functionalization of a broad range of substrates, including natural products such as drugs like gemfibroxil (80–2) and tolmetin (80–1, Scheme 80).290</p><p>The N=N bond of differently substituted azobenzenes (81–1a–e) can be functionalized on both nitrogens with a tandem N-methylation and N-sulfonylation, by cleavage of DMSO by UV irradiation of the Fenton reagent (FeSO4/H2O2,Scheme 81).105</p><p>Synthesis of amides can be achieved recurring to copper photocatalysis. Secondary alkyl bromide 82–1 could be efficiently coupled with cyclohexane carboxyamide 82–2 in 90% yield using CuI in catalytic amounts (Scheme 82a). The authors were able to isolate the catalytic species (a copper–amidate complex), formed by the assembly of four copper ions and four amides.291</p><p>The same group reported the functionalization of carbamates with secondary alkyl bromides by shifting the wavelength of irradiation in the visible region by developing a tridentate carbazolide/bisphosphine ligand 82–4 for the copper catalyst thus able to prepare Boc-pregnenolone 82–5 in 90% yield (Scheme 82b).292 A variation of this protocol was applied to the synthesis of amines, using secondary unactivated alkyl iodides and CuI/BINOL as the catalytic system.293</p><p>Several reagents can be used as an azide source to synthesize synthetically valuable C–N3 bonds. Tertiary aliphatic C–H bonds can be selectively functionalized via Zhdankin azidoiodane reagent 83–2. Visible light was used to excite Ru(bpy)3Cl2 that cleaves the labile I–N3 bond, triggering the cascade of radical reactions that leads to the product formation (Scheme 83). The selectivity and compatibility of this reaction with different groups is underlined by the conversion of the dipeptide 83–1 to 83–3 in 30% yield.294 A related C–H azidation was performed by using tosyl azide as an alternative azide source with the help of 4-benzoylpyridine to promote the photocatalytic C–H cleavage in various cycloalkanes.295</p><p>Another example of the functionalization of unactivated C–H bonds is depicted in Scheme 84 making use of tosyl azide 84–2. The reaction needs the intermediacy of an oxygen radical center on a phosphate group, previously oxidized by the action of the mesityl acridinium photocatalyst 84–3. This allows the C–H to C–N3 conversion in menthol benzoate 84–1 to give azide 84–4 in a satisfying yield, with a regioselectivity favoring the more electron-rich tertiary position.296</p><p>The synthesis of amines is undoubtedly more challenging to be dealt with, relying on radical chemistry. However, several strategies were developed to effectively forge this fundamental functional group. A classic reaction for the synthesis of amine is the Curtius reaction that has the drawback in handling of potentially dangerous azides. A dual copper/photoredox catalytic approach mimicked this process for the obtainment of N-protected amines from the N-hydroxyphthalimide ester of cholic acid triacetate 85–1 (Scheme 85, see also Scheme 8). The alkyl radical was again formed by CuI-photocatalyzed reduction of 85–1, but this recombine with the CuII–phthalimide complex formed to release 85–2 (52% yield) by a formal decarboxylation process. A great variety of functional groups are compatible with this reaction including steroidal structures.297</p><p>A very interesting approach to synthesize β-aminoalcohols from the unfunctionalized alcohol 86–1 relies on the introduction of a radical relay chaperone to direct the C–H functionalization of the β position of the OH group (Scheme 86). Imidate radicals can be accessed via the photodecomposition of PhI(OAc)2. A transient sp2N-centered radical is generated from 86–2, which allows a 1,5-hydrogen atom transfer. A source of iodine promotes the formal transfer of an iodine radical to the β-position to the imidate, followed by cyclization to obtain 86–3 which can be promptly hydrolyzed to 86–4. The nature of the substituents on acetimidate 86–2 may affect the overall yield.298</p><p>Direct cross-coupling between alkyl carboxylic acids and nitrogen nucleophiles can be achieved by dual copper/photoredox catalysis through iodonium activation. The scope of the transformation is broad and applicable to a diverse array of nitrogen nucleophiles such as heterocycles, amides, sulfonamides, and anilines to give the corresponding C–N coupling product in excellent yields on short time scales (5 min to 1 h). The high regioselectivity obtained in late stage functionalization of complex pharmaceuticals such as Skelaxin 87–2 (to give 87–3 in 90% yield from 87–1, Scheme 87, see also Scheme 48) gave an idea of the importance of the approach.299</p><p>Similar strategies were explored for the synthesis of amines via C(sp3)–N cross-coupling combining a copper catalyst and the action of a photoredox catalyst by using anilines300 or benzophenone imines301 as nitrogen source. Hydroxylamines were instead formed under photoorganocatalytic conditions by reaction of carboxylic acids and nitrosoarenes.302</p><!><p>The C–O bond formation is without doubt a prerogative of polar chemistry. However, there are examples of photochemically driven reactions making use of an alkyl radical for the introduction of different oxygen-containing functional groups. In Scheme 88, the nonenolizable ester 88–1 is transformed into 88–2 via a photochemically promoted decarboxylation of the NPhth-ester (see also Scheme 8) in the presence of Hantzsch ester to yield a tertiary radical. The intermediate is promptly quenched by TEMPO, affording 88–2 in 91% yield, in a multigram scale reaction.303</p><p>A similar reaction was employed to synthesize alkyl aryl ethers, given their importance in medicinal and agricultural chemistry. A tandem photoredox and copper catalysis approach allows the decarboxylative coupling of alkyl N-hydroxyphthalimide esters (NHPI) with phenols (89–2Scheme 89). Various NHPI esters of different drugs and natural products easily underwent a late-stage decarboxylative etherification. As an example, the chlorambucil derivative 89–1 was converted into the corresponding 2-MeO phenyl ether 89–3 in 49% yield.304</p><p>Following a similar strategy, carboxylates are converted into alcohols via a photocatalytic decarboxylative hydroxylation mediated by the mesityl acridinium salt. In this case, molecular oxygen is used as the oxidant, to promote the formation of the desired C–O bond. Since the reactions gave mainly a mixture of ketones and hydroperoxides, reduction in situ by sodium borohydride allowed the synthesis of alcohols in good yields.305 A decarboxylative hydroxylation may be carried out with the intermediacy of Barton esters that upon irradiation in oxygen-saturated toluene followed by treatment with P(OEt)3 afforded an alcohol intermediate for the total synthesis of Crotophorbolone.306 The more challenging oxidation of unactivated alkanes to alcohols or ketones can be achieved through a photoelectrochemical approach, as testified by the C–H bond activation of cyclohexane to prepare a mixture of cyclohexanone and cyclohexanol (the so-called KA oil) with high partial oxidation selectivity (99%) and high current utilization ratio (76%). The highest current ratio was obtained illuminating the solution with 365 nm wavelength.307 Decatungstate photocatalysis was efficiently applied to oxidize activated and unactivated C–H bonds. Taking advantage of a microflow reactor setup, a late stage regioselective CH2/C=O conversion in several natural compounds, such as artemisinin 90–1 to form artemisitone-9 90–2 was readily pursued even in a 5 mmol scale (Scheme 90).308</p><!><p>Halogenation of alkanes through a radical reaction under UV irradiation is one of the core pathways to chemically activate a paraffin. Industrially, chlorine gas is used to functionalize methane. A major drawback of the classical chain reaction using either Cl2 or Br2 under direct irradiation is the formation of di or polyhalogenated products. The application of microflow technology in combination with visible light irradiation (with an absorption maximum in the near UV at ca. 350 nm) allowed the monobromination of different alkanes with molecular bromine. High selectivity for the monobrominated compound and excellent overall yields (between 60 and 99%) could be achieved for secondary and tertiary alkanes, along with primary benzylic positions.309</p><p>Chlorination with molecular chlorine, on the other hand, suffers from the low yields of the reaction, typically around 50%, from the high concentrations of HCl generated in the process and from the toxicity of the chlorine gas itself. However, when Cl2 was generated by mixing NaClO with HCl and the chlorination took place under flow conditions, efficient C–H to C–Cl conversion resulted.310,311 A photochemical alternative using NaCl as chlorine source was developed.312 In the reaction, Cl2 was formed in situ by oxidation of the chloride anion with oxone. The monochlorination of cyclohexane 91–1 to give 91–2 could be obtained in 93% isolated yield thus overcoming the limitation of the classical chlorination process with chlorine gas (Scheme 91, see also Scheme 15).</p><p>Fluorination is essential to modern medicinal chemistry, both as a viable way to insert radiotracers or to deactivate specific degradation pathways in drugs. Photochemistry is a reliable tool to achieve the fluorination of C–H bonds, following different strategies. Excited TBADT may formed a radical intermediate (from unactivated alkanes) that abstracts the fluorine atom from the labile N–F bond of the fluorinating agent N-fluorobenzenesulfonimide (NFSI). An N-centered radical resulted which closes the radical cycle oxidizing the reduced photocatalyst. Acetate 92–1 was fluorinated in 40% yield following this procedure to yield 92–2 (Scheme 92). The reaction applied to sclareolide, however, was not selective and gave a mixture of fluorinated regioisomers (68% overall yield).313</p><p>Following a similar reaction scheme, uranyl acetate was employed in combination with NFSI to promote the fluorination of secondary alkanes but poorly on the benzylic positions. Indeed, in the absence of an aromatic scaffold, the excited U=O abstract a hydrogen atom through HAT, while the presence of an aromatic ring deactivated the excited state of the catalyst via exciplex formation preventing the fluorination to occur.314 Acetophenone in its excited state promoted the hydrogen abstraction of secondary alkanes, with the advantage that a common CFL housebulb can be used to promote an efficient conversion, using Selectfluor as the fluoride source.315 In this case, the authors irradiated the tail of the n-π* absorption band of the ketone which can be found in the visible region due to the high concentration of the photocatalyst present in solution. N-Alkyl phthalimides having an alkyl chain linked to the nitrogen was fluorinated by using Selectfluor under photocatalyst-free conditions. An exciplex was supposed to be formed between the reagents and it was proposed that the C–F bond formation took place concomitantly with hydrogen atom abstraction with the nitrogen radical of the fluorinating agent.316</p><p>A considerable regioselectivity in the fluorination reaction can be achieved using carboxylates as alkyl radical precursors and again Selectfluor as a fluorinating reagent. The reaction is possibly initiated by reduction of Selectfluor 93–2 by means of Ir[dF(CF3)ppy]2(dtbbpy)PF6. Fluorination of different carboxylic acids can be achieved in a very high yields (between 70 and 99%), and 93–1 was readily converted into 93–3 in 90% yield (Scheme 93).317 In case of unactivated primary substrates, a prolonged irradiation (12–15 h) was mandatory to achieve a high conversion of the substrate.</p><p>An interesting case is the fluorination of compounds having the MOM group to direct the halogenation event. In this case, the PC oxidized an imidine base (DBN) that acted as hydrogen atom abstractor of the dioxolanyl group in compound 94–1 (Scheme 94). The resulting α,α-dioxy radical 94–2 released an alkyl radical (upon formiate loss) that was fluorinated by Selectfluor. This metal-free approach again used visible light and is particularly successful when applied to tertiary alkyl ethers to give sterically hindered alkyl fluorides (e.g., 94–3).318</p><p>Interestingly, fluorination of carboxylates with Selectfluor was also reported to occur under heterogeneous photocatalytic conditions, using titania as photocatalyst to promote the oxidation of the carboxylate anion.319 Fluorination and chlorination of nitriles and ketones could be obtained starting from oximes, using Selectfluor and NCS as halogen sources, respectively. With this methodology, γ-functionalization of ketones and a complex photoinduced ring-opening/halogenation of oximes via the intermediacy of an iminyl radical was pursued. The C=N moiety of the reagent (e.g., 95–1) was preserved in the products (95–2a,b) in its oxidized nitrile form (Scheme 95). Several natural products could be functionalized following this methodology, such as androsterone (95–3a,b) and camphor (95–4a,b) derivatives (Scheme 95, see also Scheme 20).320</p><p>Alcohols were converted into their corresponding pyruvates that upon irradiation in the presence of an IrIII photocatalyst with blue LEDs released an alkyl radical prone to be chlorinated by 2,2,2-trichloroacetate as the chlorine atom source. A series of secondary and tertiary chlorides could be obtained in good to excellent yields.321</p><p>An Ir-based photocatalyst was used to promote bromination of carboxylic acid (96–1) with bromomalonate as brominating agent (Scheme 96).322 The acids used in this work were likewise converted into the corresponding alkyl chlorides and iodides in the presence of the corresponding N-halosuccinimides.322</p><p>A radical relay strategy was employed to synthesize gem-diiodides through successive intramolecular 1,5-HAT processes and iodine trapping. Indeed, the excitation with visible light of an N–I imidate group, formed in situ from the reaction of a trichloroacetimidate with PhI(OAc)2 as an iodine source, allowed the synthesis of a small library of gem di-I compounds in good yields. As an example, the cholic acid derivative 97–1 has been converted to its corresponding di-iodo derivative 97–2 in 71% yield (Scheme 97). Moreover, the authors could also achieve a dibromination using NaBr and TBABr and visible light, while only monochlorination is reported when NaCl, TBACl, and UV light were adopted.323</p><!><p>Alkyl radicals were sparsely used for the unusual introduction of a SCF2X (X = F, H) or an SAr moiety in an organic compound. The introduction of a SCF2X group has recently sparked attention due to the remarkable hydrogen donor nature of the group when X = H, making it the lipophilic surrogate for OH or NH groups.324 On the other hand, the trifluoromethylthio group increases the metabolic stability and the lipophilicity of drugs.</p><p>One strategy for the introduction of a SCF2X group is the photocatalyzed (by IrIII PC) oxidation of alkyl carboxylates via visible light irradiation in the presence of PhthN-SCF2H (98–2) as the sulfur donor. Indeed, 98–1 was converted into 98–3 in high yields (Scheme 98). The reaction was sustained by the stability of the imidyl radical liberated in the process, that was able to promote a chain reaction oxidizing a further carboxylate group. Indeed, the quantum yield for the reaction was found to be 1.7.325 Bis-methyltiolation was observed in different cases, possibly due to HAT triggered by an intermediate of the reaction, presumably PhthN• and following transfer of SCF2X from the reactant. To avoid the formation of byproducts either mesitylene or 3-(methyl) toluate were added as sacrificial hydrogen donors.</p><p>An interesting follow-up for this methodology from the same group made use of the hydrogen atom transfer process previously reported as detrimental for the reaction yield. In fact, when using an aryl carboxylate instead of an aliphatic one, the carboxyl radical that is formed upon electron transfer with the excited Ir catalyst is now stable enough to act as a hydrogen abstractor, selectively targeting secondary or tertiary H in alkyl chains. Also, in this case, PhthN-SCF2X acted as the sulfur source. The conversion of ambroxide 99–1 to its trifluorothiomethyl derivative 99–2 proceeded with 95% yield (Scheme 99).326</p><p>A photocatalyst-free decarboxylative arylthiation took place by mixing an N-acyloxyphthalimide (e.g., 100–2) in the presence of an aryl thiol (100–1) under basic conditions (by Cs2CO3) upon visible light irradiation. In this case, a SET between 100–1 and 100–2 caused the formation of 100–2•– along with thiyl radical 100–3• (that easily dimerized to disulfide 100–4). Trapping of the resulting cyclohexyl radical (by loss of PhthN– from 100–2•–) with 100–4 afforded alkylaryl sulfide 100–5 in 89% yield (Scheme 100).327</p><p>The most widely used reaction for the C–S bond synthesis requires the incorporation of sulfur dioxide by using DABSO (DABCO(SO2)2) as its surrogate as depicted in Scheme 101.328 Thus, excited mesityl acridinium salts promoted the oxidation of an alkyl-BF3K salt that generated a nucleophilic radical able to react with DABSO. The sulfonyl radical intermediate formed has been employed in a three-component reaction with electron poor olefins (e.g., a vinyl piridine 101–1, Scheme 101a)329 or an alkyne (phenyl acetylene, Scheme 101b),330 affording alkyl sulfone (101–2) or (E)-vinyl sulfone (101–3), respectively.</p><p>Alternatively, alkyl iodides can be used to react with olefins decorated with EWGs and DABSO to generate a broad range of alkyl sulfones.331 A very similar strategy was implemented by the same authors using differently substituted Hantzsch esters as alkyl radical precursors, upon irradiation in the presence of Eosin Y.332 In the latter case, the sulfonyl radical added onto vinyl azides and, after releasing of molecular nitrogen, an imidyl radical resulted which reacted with the reduced photocatalyst, forming an anion that is easily protonated. After a tautomeric equilibrium, (Z)-2-(alkylsulfonyl)-1-arylethen-1-amines were formed, with good regioselectivity and complete control over the configuration of the double bond.332</p><p>Cyclobutanone oximes can be reduced via photocatalytic means in the presence of Ir(dtbbpy)(ppy)2PF6 to form γ-cyanoalkyl radicals after radical fragmentation. In this process, a vinyl sulfone was used having the dual role of radical acceptors and SO2 source, allowing the synthesis of β-ketosulfones or allylsulfones through a radical transfer mechanism.333</p><!><p>Classical radical reductive dehalogenation is one of the most successful reactions based on tin chemistry.21 Photocatalysis and photochemistry propose a milder and more environmentally friendly alternative to this process, via different strategies. As an example, fac-Ir(ppy)3 was used to convert alkyl iodides 102–1 into their corresponding alkyl radicals using Hantzsch ester or HCO2H as the hydrogen atom source for the HAT process that drives the reaction to the formation of 102–2 (Scheme 102). The authors optimized their procedure by using tributylamine as the sacrificial electron donor to reduce the oxidized form of the catalyst and restore the catalytic cycle.334 A variation of this protocol using p-toluenethiol, DIPEA, and fac-Ir(mppy)3 was used to synthesize D-albucidin.335</p><p>Other catalytic systems were proved to be competent in the reduction of halides. In particular, unactivated aryl and alkyl bromides could be reduced using [Ir(ppy)2(dtbbpy)]PF6 in combination with TTMSS as a reducing agent. The mild conditions typical of the reaction were critical to obtain both the mono and the bis reduction of a gem-dibromocyclopropane in a selective fashion.336</p><p>Alkyl iodides and bromides were reduced under metal-free conditions via irradiation of 4-carbazolyl-3-(trifluoromethyl)-benzoic acid as the photocatalyst and 1,4 cyclohexadiene as sacrificial hydrogen donor.337 The reduction of C–X bonds to C–H bonds can take place under photocatalyst-free conditions by PET reactions between the halide and an amine as sacrificial reductant. In this way, adamantane was obtained in 95% yield by photochemical reduction of 1-bromoadamantane.338 Borohydride-mediated radical photoreduction of alkyl halides (iodides, bromides, and chlorides) is another valuable tool for the formation of a C–H bond.339</p><p>The C–H bond formation could be achieved via a hydrodecarboxylation of carboxylic acids. In fact, carboxylic acid 103–1 could be reduced in 97% yield to 103–2 by generating the corresponding carboxyl radical through excitation of an acridinium photocatalyst with 450 nm LEDs, in the presence of 10% mol of (PhS)2 (Scheme 103). The authors achieved good yields in the decarboxylation of different carboxylic acids. Most notably they succeeded in the double reduction of doubly substituted malonic acids, although with the necessity of longer irradiation times and higher catalyst loading to compensate for the increased amount of substrate to be reduced.340</p><p>The challenging reduction of alcohols to the corresponding alkane can take place via functionalization of the OH group to form an O-thiocarbamate. This compound is the substrate of a photocatalyzed Barton-McCombie deoxygenation in combination with Ir(ppy)3 and DIPEA under an oxidative quenching. Accordingly, the xylofuranose derivative 104–1 was cleanly reduced to 104–2 in 70% yield by maintaining the benzoyl group in position 5 (Scheme 104).93 The reaction was studied mostly on secondary alcohol derivatives being another interesting alternative to the usual tin-mediated reaction.22</p><p>An alternative pathway to reduce the hydroxy function required a more sophisticated functionalization of the OH group making use of two consecutive photochemical reactions. Conducting the reaction in CBr4 under UVA irradiation, the hydroxy groups of a series of primary alcohols were converted into their bromides and then subjected to a one-pot photoreduction mediated by the dimeric gold complex [Au2(dppm)2]Cl2 in the presence of DIPEA.341</p><!><p>In this last section, selected examples will be given when a photogenerated alkyl radical is used for the construction of a ring. Scheme 105 shows one example of formation of a three-membered ring. 1,1-Disubstituted cyclopropanes 105–3a–d were obtained through the addition of an alkyl radical (from silicate 105–1) onto homoallylic tosylates 105–2a–d. The trick here is a radical/polar crossover process where the reduction of the benzyl radical adducts to benzyl anions (by SET with the reduced form of the photoorganocatalyst 4-CzIPN) followed by intramolecular substitution gave the three-membered ring (Scheme 105).342 The versatility of the method was demonstrated by using alkyl trifluoroborates or 4-alkyldihydropyridines as radical precursors and a good tolerance of various functional groups.</p><p>A related approach was adopted for the construction of cyclobutanes.102 Here, the alkyl radical was formed by easily oxidizable electron-rich alkyl arylboronate complexes and added to an iodide-tethered alkene such as methyl 5-iodo-2-methylenepentanoate. However, changing the length of the chain in the haloalkyl alkenes led to the synthesis of three-, five-, six-, and seven-membered rings.102</p><!><p>Five-membered ring is one of the privileged structures accessible via photogenerated alkyl radicals. A common approach is the cyclization onto an alkyne to form an exocyclic double bond as exemplified in Scheme 106. In most cases, an alkyl halide is reduced by an excited photocatalyst and the resulting radical cyclizes in a 5-exo dig fashion to form the desired alkene. When using a dimeric gold complex 106–5 the reaction of alkyl bromide 106–1 generates diester 106–2 in 93% yield (Scheme 106a).343 Cyclopentanes were likewise formed starting from an unactivated alkyl iodide that underwent an intramolecular radical closure by using a strong reductant in the excited state (Ir(ppy)2(dtb-bpy)PF6). The iodine atom, however, was incorporated in the final product forming an alkenyl iodide.344 The same metal-based photocatalyst was effective to induce a visible light-promoted preparation of five-membered heterocycles (Scheme 106b). The cyclization step was applied on a Ueno–Stork reaction starting from 2-iodoethyl propargyl ethers (e.g., 106–3a,b) to construct a tetrahydrofuran ring (in 106–4a,b).345 The examples described in Scheme 106 required an amine as a sacrificial donor. However, amines can be used as efficient reducing agents by a PET reaction with excited alkynyl halides. The resulting photocyclization may then be carried out under metal-free conditions and in a flow photomicroreactor providing the preparation of five-membered rings in a 4 g scale.346</p><p>The dehalogenation/cyclization strategy was explored even under heterogeneous conditions by using platinum nanoparticles on titania (PtNP@TiO2) as the photoredox catalyst. The pyrrolidine scaffold was then obtained by reaction of N-(2-iodoethyl)-4-methyl-N-(prop-2-yn-1-yl)benzenesulfonamide under irradiation (DIPEA as sacrificial donor).347</p><p>As an alternative, a biphasic system may be adopted (Scheme 107). In fact, a polyisobutylene-tagged fac-Ir(ppy)3 complex (Ir(ppy)2(PIB-ppy)) soluble in heptane was prepared. The substrate 107–1 along with the reagents were soluble in a MeCN phase. However, heating at 85 °C allowed the two phases to mix. Preparation of tetrahydrofuran derivative 107–2 was then accomplished in continuous flow in a satisfying yield with an automatic recovery and reuse of the catalyst (Scheme 107).348</p><p>Alkyl N-hydroxyphthalimide esters were used as alkylation reagents in the functionalization of alkenoic acid 108–2 (Scheme 108). The alkyl radical added onto the double bond, and the resulting benzyl radical was oxidized to a benzyl cation readily trapped by water and cyclization of the resulting hydroxy acid gave alkyl-substituted lactones 108–3a–e in moderate yields.349</p><p>Alkyl N-hydroxyphthalimide esters were exploited for the photocatalyzed (by a RuII complex) alkylation of N-arylacrylamides that caused the cyclization of the adduct radical onto the phenyl ring to afford 3,3-dialkyl substituted oxindoles.350 Moreover, the same radical precursors have been used for the derivatization of alkynylphosphine oxides under metal- and oxidant-free conditions to form benzo[b]phospholes in very good yields.351</p><p>A five-membered ring may be accessed via late-stage C(sp3)-H functionalization in N-chlorosulfonamides 109–1 (Scheme 109a). The IrIII-photocatalyzed reduction of 109–1 induced the elimination of the chloride anion along the formation of a N-centered radical prone to abstract a hydrogen atom from a remote position to afford an alkyl radical. Oxidation of this radical to the cation followed by incorporation of the chloride anion gave the corresponding chloride 109–2 that upon treatment with solid NaOH formed pyrrolidine 109–3 by an intramolecular nucleophilic substitution.352 The mildness of the process allowed the application onto biologically important (−)-cis-myrtanylamine and (+)-dehydroabietylamine derivatives.</p><p>A related intramolecular 1,5-HAT transfer was induced in N-tosyl amides 109–4a–e. In this case, the haloamide is formed in situ by iodination by reaction of 109–4a-e with iodine (obtained by oxidation of iodide anion by PhI(OAc)2). The excess of iodine allowed for tuning the amount of iodine released in solution by forming the triiodide anion. Then, visible light irradiation of the mixture induced the cyclization to give N-tosylpyrrolidines 109–5a–e (Scheme 109b).353</p><p>Even primary nonactivated sp3-hybridized positions were functionalized again by a remote intramolecular radical 1,5-hydrogen abstraction in γ-bromoamides to produce several γ-lactones in a one-pot fashion.354 Trifluoroethyl amides were found useful as the directing group increasing the efficiency of the hydrogen abstraction process.</p><p>It is also possible to incorporate more than one heteroatom in the ring starting from benzyl amine 110–2 and unactivated bromides 110–1a–e (Scheme 110). Compound 110–2 incorporates CO2 (with the help of the base TBD), and the resulting carbamate underwent attack by an alkyl radical photogenerated by reaction of 110–1a–e and an excited Pd0 photocatalyst (Pd(PPh3)4). Ring closing yielded valuable 2-oxazolidinones 110–3a–e under very mild conditions and easy scalability.355</p><!><p>Different approaches were devised to form a six-membered ring even cointaining heteroatoms. A cyclohexane ring was constructed by ring opening of an iminyl radical by IrIII-photocatalyzed reduction of a 3-phenyl O-acyl oxime (e.g., 111–1) to give radical 111–2• that upon addition onto unsaturated esters 111–3a–e and ensuing cyclization led to cyanoalkylated 1,2,3,4-tetrahydrophenanthrenes (111–4a–e, Scheme 111).356</p><p>Six-membered rings have been likewise obtained by ring expansion in cycloalkanone derivatives. This expansion was caused by the photocatalyzed decarboxylation of α-(ω-carboxyalkyl) β-keto esters, followed by an exo-trig cyclization of the resulting radical onto the carbonyl group that ultimately led to the one-carbon expanded cycloalkanones by β-cleavage.357</p><p>Reduction of indoles having an unactivated haloalkane chain is a useful approach to construct a ring. As an example, bromo derivatives 112–1a,b were reduced by a AuI photocatalyst and radical cyclization onto the heteroaromatic ring afforded 6,7,8,9-tetrahydropyrido[1,2-a]indoles 112–2a,b in excellent yield (Scheme 112a).358 Interestingly, changing the reaction conditions and starting from N-(2-iodoethyl)indoles 112–3a,b in place of 112–1a,b in the presence of Michael acceptors 112–4a–c caused a dearomatizative tandem [4 + 2] cyclization to deliver tri- and tetracyclic benzindolizidines 112–5aa–bc with high diastereoselectivity and yield (Scheme 112b).359</p><p>The phenanthridine core is one of the elective scaffolds to be prepared by using a cyclization step induced by photogenerated alkyl radicals. Scheme 113 illustrated a representative case where an alkyl radical added onto a vinyl azide 113–2, and after nitrogen loss the resulting iminyl radicals 113–3a-c• yielded phenanthridines 113–4a-c by ring closure.360 The method has several advantages including metal-free conditions (a dye as a POC) an excellent functional group tolerance and a broad substrate scope.</p><p>An alternative way to prepare phenanthridines is by having recourse to photoredox gold catalysis employing bromoalkanes as alkyl radical source. In this case, radicals attack a biaryl isonitrile thus forming a sp2-hybridized radical that readily cyclizes upon the pendant arene.361 Aryl isocyanides (e.g., 114–2a–d) were largely used for the construction of heterocycles such as pyrrolo[1,2-a]quinoxalines 114–4a–d. PhenyliodineIII dicarboxylate 114–1 was used for the incorporation of the cyclohexyl group both in batch and flow under IrIII-photocatalyzed conditions (Scheme 114, see also Scheme 49).362</p><p>The photocatalyzed insertion of SO2 into an unactivated C(sp3)-H bond was designed to prepare 1,2-thiazine 1,1-dioxide derivatives under uncatalyzed conditions. In fact, visible light irradiation of the complex between an electron-poor O-aryl oxime and DABCO·(SO2)2 releases an iminyl radical that upon 1,5-HAT, SO2 incorporation and cyclization gave the hoped-for heterocycle in a satisfying yield.363</p><p>In rare instances a ring larger than six may be constructed. By using the approach depicted in Scheme 115, it was possible to pursue a late stage functionalization on ursolic acid (a compound having excellent pharmaceutical activity). Accordingly, the NHPI ester of ursolic acid acetate (115–2) underwent a radical addition cascade by a photocatalyzed reaction with acrylamide-tethered styrene (115–1) with the intermediacy of radical 115–3•. As a result, the benzazepine unit was incorporated in the end compound 115–4 combining two privileged bioactive scaffolds.364</p><!><p>This review provides a concise and up-to-date selection of modern methods to generate alkyl radicals via photochemistry and photocatalysis. The effort and interest of the chemical community in developing and applying these new methods is witnessed by the rapid increase in the number of articles devoted to this topic that appeared in the literature in the last two decades. Indeed, the rediscovery of photocatalysis and the renaissance of visible light-driven processes have contributed to elevate radical chemistry from the isolated (yet efficient) niche of the tyrannical organotin compounds to a vast plethora of methodologies that relies on more environmental benign compounds. The facile synthesis of the precursors necessary for these transformations, along with the readily available setups (a vast number of reactions can occur by simple irradiation with visible LEDs), made radical chemistry approachable, appointing the photon as the agent of this revolutionary democracy.</p><p>Photocatalysis has reached the stage of maturity; however, we are still far from the statement of Ciamician envisioning "industrial colonies without smoke [···] forests of glass tubes [···]; inside of these will take place the photochemical processes that hitherto have been the guarded secret of the plants, but that will have been mastered by human industry which will know how to make them bear even more abundant fruit than nature, for nature is not in a hurry and mankind is".365 New practical methods and theoretical assumptions are needed to foster the revolution that has just started. A promising approach makes use of the upconversion of reductants to generate strongly reductive species, but the method was not applied so far to alkyl radicals.366 This phenomenon can be exploited, for example, if the reaction of a radical anion R•– to give P•– is less exoergonic (see the ΔG•– value in Figure 4A) than its neutral counterpart (ΔG, referred to R → P conversion). The difference between these two free energies defines the upconversion energy (ΔGup = ΔG•– – ΔG). The high quantum yields associated with the transformation of R into P in Figure 4A (Φ = 44) were attributed to the presence of electrocatalytic cycles propagated by P•–, which is able to transfer an electron to the reactant, closing the catalytic cycle. This phenomenon is attributed to P•– being a better reductant than R•–, due to the diminished conjugation (Figure 4A).</p><!><p>(A) Upconversion of the reducing power of the intermediates in a photocatalytic/photoinitiated cyclization. (B) Two pathways to employ the photoelectrocatalytic strategy: either promoting a single electron transfer with photocatalysis first and a second one with electrocatalysis or vice versa.</p><!><p>The novel approach granted by the merging of homogeneous photocatalysis with electrocatalysis (see Figure 4B) is surfacing as the new challenge in this constantly evolving topic.367−369</p><p>Joining the almost unlimited potential of these two interchangeable fields of research would open unprecedented scenarios in chemical synthesis, allowing one to tweak the reactivity of intermediates and excited state species at will, walking on the path carved by the institution of the photon democracy.</p><!><p>S.C. and M.F. discussed and contributed to the final manuscript. M.F. conceived the original idea.</p><!><p>The authors declare no competing financial interest.</p><p>Maurizio Fagnoni is currently an Associate Professor at the PhotoGreen Lab (Department of Chemistry, University of Pavia, Italy). His academic and professional background is in organic photochemistry and his activity has always been focused on the exploration of the photochemistry of organic molecules and the attending applications in various fields. The photochemical generation of intermediates, e.g., radicals and cations and radical ions by photochemical means, is the main topic of his research. Particular attention has been given to the significance of such mild synthetic procedures in the frame of the increasing interest for sustainable/green chemistry. He was the recipient in 2019 of the "Elsevier Lectureship Award" from the Japanese Photochemical Association. He was recently coeditor of the book Photoorganocatalysis in Organic Synthesis (World Scientific, 2019). Since 2019, he has been the President of the Didactic Council in Chemistry of the University of Pavia.</p><p>Stefano Crespi received his Ph.D. in 2017 at the University of Pavia (Italy) under the supervision of Maurizio Fagnoni. He won a two-year fellowship as a Post-Doc in the same University focusing on the study of novel heteroaryl azo photoswitches. He joined the workgroup of Burkhard König at the University of Regensburg, where he studied new scaffolds based on heteroaryl azo dyes and novel photocatalytic transformations. In 2019, he moved to Groningen to work on molecular motors in the group of Ben Feringa as a Marie Skłodowska-Curie fellow. His research interests lie in the combination of reaction design in organic (photo)chemistry with computational models.</p>

PubMed Open Access

Evaluation of Silicon Phthalocyanine 4 Photodynamic Therapy Against Human Cervical Cancer Cells In Vitro and in Mice

Background Cervical cancer is the second most common cancer in women worldwide [1]. Photodynamic therapy has been used for cervical intraepithelial neoplasia with good responses, but few studies have used newer phototherapeutics. We evaluated the effectiveness of photodynamic therapy using Pc 4 in vitro and in vivo against human cervical cancer cells. Methods CaSki and ME-180 cancer cells were grown as monolayers and spheroids. Cell growth and cytotoxicity were measured using a methylthiazol tetrazolium assay. Pc 4 cellular uptake and intracellular distrubtion were determined. For in vitro Pc 4 photodynamic therapy cells were irradiated at 667nm at a fluence of 2.5 J/cm2 at 48 h. SCID mice were implanted with CaSki and ME-180 cells both subcutaneously and intracervically. Forty-eight h after Pc 4 photodynamic therapy was administered at 75 and 150 J/cm2. Results The IC50s for Pc 4 and Pc 4 photodynamic therapy for CaSki and ME-180 cells as monolayers were, 7.6\xce\xbcM and 0.016\xce\xbcM and >10\xce\xbcM and 0.026\xce\xbcM; as spheroids, IC50s of Pc 4 photodynamic therapy were, 0.26\xce\xbcM and 0.01\xce\xbcM. Pc 4 was taken up within cells and widely distributed in tumors and tissues. Intracervical photodynamic therapy resulted in tumor death, however mice died due to gastrointestinal toxicity. Photodynamic therapy resulted in subcutaneous tumor death and growth delay. Conclusions Pc 4 photodynamic therapy caused death within cervical cancer cells and xenografts, supporting development of Pc 4 photodynamic therapy for treatment of cervical cancer. Support: P30-CA47904, CTSI BaCCoR Pilot Program.

evaluation_of_silicon_phthalocyanine_4_photodynamic_therapy_against_human_cervical_cancer_cells_in_v

Introduction<!>Materials<!>Cell Culture<!>Prevention of Phototoxicity<!>PDT in vitro<!>Cytotoxicity of Pc 4 and Pc 4 PDT in cell monolayers<!>Cytotoxicity of Pc 4 and Pc 4 PDT in spheroids<!>Pc 4 uptake studies<!>Intracellular distribution of Pc 4<!>Intracellular organelle markers for endoplasmic reticulum and mitochondria<!>Animals and tumor model<!>Pc 4 in plasma and tissues by HPLC<!>PDT in vivo<!>Growth and cytotoxicity of Pc 4 and Pc 4 PDT in monolayers<!>Cytotoxicity of Pc 4 and Pc 4 PDT in spheroids<!>Pc 4 uptake in monolayers<!>Pc 4 uptake in spheroids<!>Intracellular localization of Pc 4 in CaSki and ME-180 cells<!>Tumor growth in mice<!>Pc 4 concentration in xenografts and normal tissues<!>Pc 4 PDT in vivo<!>Discussion

<p>Cervical cancer is the second most common cancer in women worldwide behind breast cancer [1]. In developing countries, cervical cancer is the most common cancer in women and the leading cause of cancer death [2]. In the United States, according to the American Cancer Society there will be approximately 12,900 new cases and 4,100 deaths in 2015 [3]. Cervical cancer can be categorized by extent of the spread of disease: localized tumors up to 4 cm are early stage disease; tumors larger than 4 cm or with paracervical involvement are locally advanced disease; involvement of other pelvic organs or distant metastases is advanced disease [4]. In the United States and Canada, concomitant chemoradiation is the standard of care for locally advanced disease and advanced disease, while neoadjuvant chemotherapy followed by radical surgery is used in Europe, Asia and Latin America [4]. Treatment of locally advanced disease with concomitant chemoradiation is often suboptimal and leaves approximately 40% of patients with residual disease [5]. There are limitations on the total radiation dose that can be administered so patients with persistent or recurrent disease after chemoradiation have few therapuetic options other than radical pelvic surgery that often entails removal of the rectum, bladder and vagina en bloc.</p><p>For cervical intraepithelial neoplasia (CIN), a preinvasive disease, the treatment is excisional or ablative therapy: LEEP (loop electrosurgical excision procedure), CKC (cold knife conization) or local destruction with either CO2 laser or cryotherapy [6]. These excisional and ablative treatments can negatively impact the reproductive health of the women by potentially increasing the risk of preterm premature rupture of membranes and preterm labor [7–9]. There has been little progress in the last 20 years examining novel therapeutic modalities for CIN or cervical cancer. Photodynamic therapy (PDT) is an attractive treatment modality for both preinvasive disease and invasive cervical cancer because its therapeutic agents, a photosensitizer, light and oxygen, are safe by themselves but lead to selective tumor destruction when combined. Advantages of PDT in the treatment of CIN and invasive cervical cancer include: the ability to localize treatment with the application of light; the sparing of intracellular matrix allowing regeneration of normal tissue; the potential for repetition of therapy without accumulation of toxicity; the potential of combining PDT with chemotherapy and, or radiation therapy to improve their efficacies; and the enhancement of antitumor immunity which may contribute to long term tumor control [8,9]. Finally, PDT does not require anesthesia and generally does not cause bleeding which makes its use applicable to the outpatient setting.</p><p>The most widely used PDT therapeutics are Photofrin and various forms of 5-amino levulinic acid (ALA). Most recently, topical formulations of ALA have been used directly in the cervix with remission rates of 71%, 50% and 71% for CIN 1, 2 and 3 [8]. To date, PDT has not been used in the treatment of invasive cervical cancer.</p><p>Silicon phthalocyanine 4 (Pc 4) (Figure 1), is a second generation photosensitizer developed at Case Western Reserve University in coordination with the Drug Decision Network of the National Cancer Institute. Among Pc 4's desirable features are: its chemical purity; its high extinction coefficient, 0.23cm−1 μM−1 at 670nm in aqueous solution, a wavelength allowing deep tissue penetration of light; and its rapid clearance from skin which limits the extent and duration of cutaneous photosensitivity after intravenous administration [10–12]. Pc 4 PDT induces apoptosis in vitro, due to damage to mitochondria and endoplasmic reticulum [13]. Pc 4 PDT has also demonstrated promise in preclinical studies of squamous cell cancers of the head and neck, and has been used topically to treat cutaneous t-cell leukemia [14,16].</p><p>In this study, we evaluated the effectiveness of Pc 4 PDT in inhibiting the growth of human cervical cancer cells in vitro when grown as monolayers and spheroids, and in vivo. We measured the uptake of Pc 4 by the cells over time and the Pc 4 intracellular localization. We also examined the growth of cervical cancer cells when implanted in the cervix of immunodeficient mice and the uptake of Pc 4 by cervical cancer xenografts as well as other tissues after IV administration of Pc 4. We attempted intravaginal and subcutaneous delivery of irradiation after treatment of the mice with Pc 4. This preliminary data will be used to support further in vivo studies of Pc 4 PDT in the treatment of CIN and cervical cancer.</p><!><p>Pc 4 (NSC 676418) was obtained from the Cancer Treatment Evaluation Program of the National Cancer Institute (Bethesda, MD). Sterile water and 0.9% sodium chloride injection solution (saline) were purchased from Baxter Healthcare Corp (Deerfield, IL). RPMI medium, fetal bovine serum, trypsin-EDTA (10X), penicillin-(100 U) streptomycin μg/ml in 0.85% saline and PBS (pH 7.4, without calcium or magnesium) were purchased from Invitrogen (Carlsbad, CA). Intralipid 10% IV fat emulsion was purchased from Fresenius Kabi Clayton, L.P. (Clayton, NC). Cremophor El, dimethyl sulfoxide,microtetrazolium dye (MTT, 3-(4,5-dimethylthiazol-2-yl)-2,5-diphenyltetrazolium bromide) and nile blue were purchased from Sigma Aldrich Corp (St. Louis, MO). Ethanol was purchased from Pharmco Products, Inc. (Brookfield, CT). HPLC grade ethyl acetate and glacial acetic acid were purchased from J. T. Baker (Phillipsburg, NJ). HPLC grade methanol was purchased from Fisher Scientific (Pittsburgh, PA). Nembutal and Buprenex (Buprenorphine) were purchased from Henry Schein Animal Health, Inc. (Melville, NY), isoflurane from Hospira, Inc. (Lake Forest, IL) and heparin sodium for injection (10,000 U mL) was purchased from American Pharmaceutical Partners, Inc. (Schaumburg, IL).</p><!><p>The human squamous cell carcinomas of the uterine cervix, CaSki and ME-180, were obtained from the American Type Culture Collection (CaSki:ATCC® CRL-1550™ and ME-180:ATCC® HTB-33™, Manassas, VA) and cultured in RPMI 1640 medium (Fisher Scientific) with L-glutamine, containing 10% fetal bovine serum, penicillin (100IU), streptomycin (100μg/mL) and maintained in an incubator at 37˚C, 5% CO2 and 95% humidity. Monolayers were grown in standard 96-well flat bottom plates. Spheroids were created using Perfecta3D® Hanging Drop Plates (3D Biomatrix, Ann Arbor, MI) and were cultured on Corning™ Matrigel™ Membrane Matrix (Fisher Scientific) 1:1 with complete media.</p><!><p>In order to prevent phototoxicity due to room lighting, cells removed from the incubator were processed in a vented laminar flow hood without lights and surrounded by green theatrical gels (Lee filter, 124) prior to and after introduction of Pc 4. Simiarly, mice were housed in cages surrounded by these same theatrical gels 1 week following the administration of Pc 4 to prevent skin phototoxicity.</p><!><p>Cells in logarithmic growth, 24h after the addition of Pc 4, were irradiated through a microlens diffuser fiber (P/N5470RevD; 400micron core, Pioneer Optics Co., Bloomfield, CT) on a diode laser (HPD 670nm laser, 7404-D-EXT; High Power Devices, Inc., New Brunswick, NJ) at 670nM. The laser was calibrated using a power meter (PM100; Thorlabs, Newton, NJ). Fluence was maintained at 2.5 J/cm2 because it was the most effective over a wide range of Pc 4 concentrations in monolayers than fluences of 0.1, 0.2, and 1 J/cm2 (our unpublished data; to vary the fluence, time was adjusted as power was constant). To maintain fluence of 2.5 J/cm2 between experiments, we varied time based on power as detected by the power meter (t=Fluence*(π*radius2)/Power) and was approximately twenty-four minutes to irradiate a 96-well plate from above within a circle with a diameter of 12cm with an median power of 190mW.</p><!><p>CaSki cells or ME-180 cells (1 × 105 cells) in logarithmic growth were plated into each well of 96-well flat bottom, black sided culture plates and allowed to acclimate for 24h. Pc 4 was added at a final concentration of 1nM-10μM in medium containing 0.1% DMSO in every well under reduced lighting. Pc 4 was removed, cells were washed with PBS and the medium was replaced with complete RPMI after 48h. For the Pc 4 alone effects, the cells were allowed to grow for an additional 24 hours in Pc 4 free medium before the addition of 50μl of 1 mg/ml MTT to each well. The plates for Pc 4 PDT were washed and medium was replaced with complete medium without Pc 4. The laser component (670nm with a fluence of 2.5 J/cm2 as described above) of PDT was administered to the monolayers. Following laser exposure, the cells were incubated for an additional 24h prior to the addition of 1 mg/ml MTT. After the addition of MTT, the cells were incubated for 4h at which time medium containing MTT was removed from each well, 100μl DMSO was added and the plate was shaken for 5 minutes. The absorbance at 570nm was read on a Tecan Safire microplate reader (Tecan, San Jose, CA). Results were compared to wells containing vehicle-treated cells and expressed as % inhibition. The IC50 was calculated using the Hill equation in the computer program ADAPT V from experiments with a minimum of triplicate samples at each concentration (D'Argenio, Schumitzky and Wang, 1997; https://bmsr.usc.edu/software/adapt/). The experiments were performed at least 3 times.</p><!><p>Spheroids of CaSki or ME-180 cells (1 × 104 cells) were prepared as described above. The spheroids were allowed to attach for 24h and Pc 4 was added at a final concentration of 1nM-10μM in medium containing 0.1% DMSO in every well under reduced lighting. After 24h, the Pc 4 containing medium was removed; the spheroids were washed with PBS and new complete medium was added. The laser component (670nm with a fluence of 2.5 J/cm2 as described above) of PDT was administered to the spheroids. After 24h, the spheroids were exposed to 1 mg/ml MTT for 4h. The spheroids were then examined for the insoluable formazan using light microscopy and the percentage of cells surviving in each spheroid was estimated. The IC50 was calculated using the Hill equation in the computer program ADAPT V from experiments with a minimum of triplicate spheroids at each concentration (D'Argenio, Schumitzky and Wang, 1997; https://bmsr.usc.edu/software/adapt/). The experiements were performed at least three times.</p><!><p>Monolayers or spheroids of CaSki or ME-180 cells were prepared as described above. For monolayers, after 24h, Pc 4 at 0.3μM was added in complete medium to each well; medium was removed at time points between 0 – 48h and stored for analysis. After removal of medium, cells were washed with PBS and the wash removed. DMSO was then added to wells to solubilize the membranes and extract intracellular Pc 4 at time points between 0 – 48h. The medium and DMSO from wells was measured on the Tecan Saphire as fluorescence (ex: 350nm, em: 667nm) and values were back calculated from linear Pc 4 standard curves in medium and DMSO. Each time point was run in at least triplicate and with 3 separate assays. For spheroids, after 4 days when they had reached sizes of between 300 and 650 microns, kinetics of Pc 4 uptake in CaSki and ME-180 spheroids were determined. Pc 4 at 0.3μM was added to spheroid medium. Spheroids of uniform and equal sizes were chosen for imaging at ten Z positions between 0 and 150 microns from the top of the spheroids (step-size = 15 or 25 microns). After the addition of Pc 4, confocal images were collected at intervals between 3 and 9h over a 36h period using an inverted Zeiss 510 Meta confocal microscope. The spheroids were kept at 37 C, 100% humidity and 5% C02 using a Tokai stage incubation chamber in the dark except for when confocal images were recorded.</p><!><p>CaSki and ME-180 cells were cultured on glass chamber-slides (Ibidi, μ-Slide 8 Well, 80826). When cells were 50% confluent, adenovirus containing the gene coding for a cerulean-endoplasmatic reticulum (ER) marker fusion protein or a cerulean-mitochondria marker fusion protein was added to the cells at an MOI of 25; 48h later, Pc 4 (at 0.3μM) was added and the cells were incubated for another 6h. To visualize lysosomes, LysoTracker Yellow protocol (LysoTracker® Yellow HCK-123, L-12491, ThermoFisher Scientific) was added (75nM) to the cell medium 3h prior to incubation with Pc 4. After washing with PBS, the cells were examined by confocal microscopy (Inverted Zeiss 510 Meta) and confocal micrographs were acquired.</p><!><p>Expression plasmids containing ER or mitochondrial localized mCerulean3 were generously provided by Dr. Marcel Bruchez, Carnegie Mellon University [10]. The coding sequences from these plasmids were excised and inserted into a pENTR1A no ccdB Gateway shuttle vector using the following restriction enzymes: for the ER construct, BamHI/XbaI; for the mitochondria construct, BamHI/XhoI. Using Gateway LR Clonase II (Life Technologies, Waltham, MA), the resulting plasmids were recombined with the pAd⁄CMV⁄V5-DEST Gateway Vector to produce pAd⁄CMV-DEST-ER and pAd⁄CMV-Dest-Mitochondria, respectively. These constructs were then utilized to produce adenovirus using the ViraPower™ Adenoviral Gateway Expression Kit according to the manufacturer's recommendations (Life Technologies, Waltham, MA).</p><!><p>Female C.B-17 SCID mice (specific pathogen-free, 4–6 weeks of age) were purchased from Charles River Laboratories (Wilmington, MA) and allowed 1 week to acclimate to the animal facilities at the University of Pittsburgh. Mice were handled in accordance with the Guide for the Care and Use of Laboratory Animals (National Research Council, 2012) and on a protocol approved by the University of Pittsburgh Institutional Animal Care and Use Committee. A day before the implantation of cervical cancer cells, the mice were anesthetized with isoflurane using a nose cone and their abdomens were shaved with any additional fur removed using a topical application of Nair™ for Men Cream. The abdomens were washed with warm water and the mice were allowed to recover from the anesthesia. On the day of surgical implantation, CaSki or ME-180 cells in logarithmic growth were harvested from culture. Mice to receive the cells were administered buprenorphine (0.05 mg/kg) 2h prior to surgery and the mice were anesthetized with Nembutal (50 mg/kg IP). Under anesthesia, the shaved site was cleaned with 70% ethanol followed by betadine scrub. Aseptic techinique was used throughout the procedure. The skin over the linea alba was cut to create a longitudinal incision of approximately 1.5cm. Using mosquito hemostats the skin was retracted from the underlying muscle. A longitudinal incision of approximately 1cm was made in the abdominal musculature directly on the midline. Using a sterile saline soaked 2×2 gauze pad the urinary bladder was expressed and the inguinal fat pads were moved out of the abdomen onto a saline soaked sterile gauze. Using a small curved forceps, the bifurcated uterus was gently pulled upward to expose the cervix. Cells (1 × 105 cells in 10μL of 1:1:: complete medium:matrigel) were implanted intracervically using a 27g needle and 0.5cc syringe. The injection was confirmed by the formation of a small weal at the site of injection. After injection the uterus, cervix, fat pads, and urinary bladder were returned to the peritoneal cavity. The peritoneum and musculature was sutured with dissolvable sutures (3 sutures using X-1, 22mm ½cm reverse cutting needle and 3-0 coated vicryl plus antibacterial sutures, Ethicon Inc. Cincinnati, OH) and the skin was sealed with surgical staples that were removed on post-operative day 5–7. Buprenorphine was administered twice daily for 2 days post-operatively for pain relief. Tumor volumes were estimated by palpation twice weekly beginning after staples were removed, and scored as +1 to +5 depending on the size of the tumors.</p><!><p>When the intracervical tumors reached a score of +1 to +4 by palpation the mice were administered Pc 4 (2 mg/kg in 1:1:10:: crempphor: ethanol: saline) intravenously via tail vein and one mouse in each group was euthanized with CO2 at each of the following times: 6, 24, 48 and 72h after dosing. Blood was collected by cardiac puncture and the select tissues (tumor, uterus, liver, kidneys, lungs, skeletal muscle, fat, spleen, colon) were removed, weighed and frozen in liquid nitrogen. The blood was centrifuged at 12,000 xg for 4 minutes at which point the plasma was separated from the RBCs, was transferred to cryovials, and stored, as the tissues were, at −80 ◦ C until analyzed. Pc 4 concentrations were measured by HPLC as described previously [10,15]. Tumors and tissues were homogenized in 3 volumes of PBS, and 200μL of each plasma sample or tissue homogenate was transferred to a separate tube containing 10μL of 10mg/mL of Nile blue (internal standard). Samples were processed. Pc 4 standards (0.03, 0.1, 0.3, 1, 3, 10, 30μM) were prepared in tumor homogenate or plama from untreated mice. The Beckman HPLC Programmable Gradient System (BeckmanCoulter, Fullerton, CA) consisted of a model 508 autosampler, a model 128 gradient solvent delivery module, and a model 166 UV detector. Compounds were separated on a IBondapak C18 column (10μm, 3.9μm, 300mm, Waters Corp., Milford, MA) fitted with a Brownlee C18 guard column (Perkin-Elmer, Boston, MA). The isocratic mobile phase, consisting of methanol:dH2O:glacial acetic acid (80:20:1, vol ⁄ vol ⁄ vol), was pumped at 1 mL min). The column eluent was monitored at 658nm. Under the above conditions, the retention times of Pc 4 and Nile blue were 4.5 and 5.3 minutes, respectively. Recovery was 81%. The coefficient of variation was <15% at low and medium high concentrations. Standard curves of Pc 4 were constructed by plotting the Pc 4 to internal standard area ratio versus the known concentration of Pc 4 in each sample. Standard curves were fit by linear regression and followed by back calculation of concentrations. Concentrations in unknown samples were calculated by comparison with the appropriate standard curve.</p><!><p>Five SCID mice bearing ME-180 cells implanted intracervically were used for intravaginal PDT. Additionally, one SCID mouse bearing CaSki cells and one SCID mouse bearing ME-180 cells implanted subcutaneously were also used. When the intracervical tumors reached an estimated 100 mm3 based on palpation, the mice were administered Pc 4 at 2 mg/kg IV. Two days later, the mice were anesthesized with 50 mg/kg sodium pentobarbital. The mice were placed in the supine position on a platform and gently restrained so that the laser could be positioned intravaginally. The tumors were irradiated through a microlens diffuser fiber (P/N5470RevD; 400micron core, Pioneer Optics Co., Bloomfield, CT) on a diode laser (HPD 670nm laser, 7404-D-EXT; High Power Devices, Inc., New Brunswick, NJ) at 670nM. The microfiber was inserted intravaginally approximately 0.4cm, this would ensure the fiber would be approximately 0.6cm away from the vaginal apex, yielding a laser radius of 0.225cm at the depth of the cervix and tumor. This laser radius was chosen to encompass the entire tumor, while decreasing exposure of normal tissues to PDT. The laser was calibrated using a power meter (PM100; Thorlabs, Newton, NJ). Irradiation was initially at a fluence of 150 J/cm2 for two mice with intracervical tumors [15]. Secondary to toxicity, fluence was decreased to 75 J/cm2 for 3 mice with intracervical tumors. The fluence was adjusted, and the time was kept constant at 4 min, so the laser power was reduced appropriately from 100 to 50mW (t=Fluence*(π*radius2)/Power). The subcutaneous tumors (diameter 1.25 cm) were irradiated at a fluence of 150 J/cm2 with a split microfiber for two mice for approximately 16 min (t=Fluence*(π*radius2)/Power)15. The mice were followed for recovery from anesthesia and twice daily after PDT.</p><!><p>When CaSki cells and ME-180 cells were grown as monolayers, their doubling times were 56 and 25h respectively (data not shown). Pc 4, added under reduced lighting was relatively non-toxic and its IC50 in CaSki cells after 48h of exposure was 7.6μM, while after Pc 4 PDT, the IC50 was 0.016μM (Figure 2A). ME-180 cells were less sensitive to Pc 4 and the IC50 for Pc 4 alone was >10μM. Higher concentrations of Pc 4 could not be evaluated because at higher concentrations the Pc 4 came out of solution in the medium. For ME-180 Pc 4 PDT, the IC50 was similar to that of the CaSki cells, 0.026μM (Figure 2B).</p><!><p>When CaSki cells and ME-180 cells were grown as spheroids, Pc 4 by itself was relatively non-toxic and its IC50 for each cell line could not be calculated using concentrations up to 10μM (Figures 3A and B). For Pc 4 PDT, the IC50 in CaSki spheroids was much higher than had been observed in the CaSki cells grown as monolayers, 0.26 vs. 0.016μM (Figure 3A). In the case of ME-180 spheroids, the IC50 was no higher than that of Pc 4 PDT for cells grown as monolayers (0.011 vs. 0.026μM). This difference in response to Pc 4 PDT between cell types in response may be due to the structure of the spheroids; the ME-180 spheroids are much less tightly packed as is seen in Figure 3. The phase contrast images of insoluble formazan within the CaSki and ME-180 spheroids exposed to Pc 4 alone or after Pc 4 PDT are shown in Figures 3, C and D. Because we could not extract the insoluable formazan from the spheroids using DMSO and a number of other solvents, we sectioned the spheroids at their largest diameter (~300μm) to determine the depth of penetration of formazan and confirm that the viable spheroids contained formazan throughout. An ME-180 spheroid treated with Pc 4 only is shown in Figure 3E. As expected, there were no formazan crystals formed in the Pc 4 PDT treated spheroids beyond 0.1μM Pc 4 in CaSki spheroids and beyond 0.01μM in ME-180 spheroids.</p><!><p>To confirm the uptake of Pc 4 in the CaSki and ME-180 cells, we measured both the concentration of Pc 4 in the medium and in the cells over time when the cells were continuously exposed to a Pc 4 concentration of 0.3μM. As is apparent in Figure 4, A and B, Pc 4 concentrations within the medium for both cell lines decreased with time, while the concentration of Pc 4 within the cells increased with time. By 6h after addition of Pc 4 to the medium, the accumulation within the cells slowed and by 48h represented approximately only 25% of the initial concentration added to the medium.</p><!><p>Pc 4 accumulated at a slower rate in these cervical cell lines when they were grown as spheroids on matrigel. The rate of accumulation into the spheroids is shown in Figures 4, C and D, as fluorescence units (measured in confocal images taken at various times after addition of 0.3μM Pc 4 at a depth of 105μm into CaSki spheroids and a depth of 175μm into the ME-180 spheroids). Uptake in CaSki spheroids was linear between 6 and 24h, and then remained constant, while the uptake in the ME-180 spheroids continued to increase over the 35h of exposure (Figure 4E).</p><!><p>Pc 4 was widely distributed within intracellular organelles with a similar distribution in both CaSki and ME-180 cells. Pc 4 could be found in the cytosol, lysosomes, endoplasmic reticulum and in the mitochondria of both cell lines (Figure 5). In CaSki cells, the lysosomes appeared to show the most overlap between Pc 4 and the marker, followed by the endoplasmic reticulum; the mitochondria appeared to have less overlap between Pc 4 and the organelle marker (Figure 5A). A similar distribution was observed in the ME-180 cells grown as monolayers, but with more overlap between the mitochondrial marker and Pc 4 (Figure 5B).</p><!><p>In the mice after cervical implantation of either 1×105 CaSki or ME-180 cells, the tumors initially grew as small nodules in the cervix. As the disease progressed, however, it spread up the uterine horns and tumors were also localized as nodules attached to the peritoneal wall or mesentery. Several nodules were located attached to mesentery in peritoneal cavity just below the xypoid process. Select mice also had more advanced disease: metastases in liver parenchyma; or pelvic side wall disease eventually encasing the ureters and leading to post-renal obstruction. The tumors did not appear to spread distally to affect the vagina or anus. In the SCID mice, the tumor growth was much slower for the CaSki implanted cells than for the ME-180 implanted cells; CaSki tumors reached a volume of 500 mm3 by 3 months while the ME-180 tumors reached similar volumes in only 7–8 weeks.</p><!><p>Four female SCID mice were implanted with CaSki cells and 5 with ME-180 cells subcutaneously on the right flank. One mouse bearing each xenograft was euthanized at 6, 24, 48 and 72h after IV administration of Pc 4 at 2 mg/kg. One mouse bearing an ME-180 xenograft served as control. As is apparent in Table 1, the tissue weights were similar between the mice, however the tumor weights were variable between mice receiving the same tumor cell line. ME-180 tumors varied between 0.059g and 1.6g, while the CaSki tumors were slower growing and ranged between 0.09g and 0.5g.</p><p>The concentrations of Pc 4 in the tissues are shown in Fig. 6. The highest concentrations are present in liver, kidneys and lungs, while skeletal muscle concentrations are in the same concentration range as the tumors. The ME-180 tumor Pc 4 concentrations decreased from 0.9μM at 6h to 0.56μM at 48h, while in the CaSki xenografts, the concentration of Pc 4 rose between 6 and 24 h, and then remained relatively constant between 24 and 72h. Plasma concentrations are not detectable after 48h and are at least a log lower than tumor concentrations. Intestinal concentrations (after removal of feces and PBS wash) and fat concentrations of Pc 4 are not detectable beyond 24h. The concentrations of Pc 4 in the uterus and in the cervical tumors were similar. Pc 4 distributes widely to tissues and is retained in the cervical tumors as well as in the uterus of the mice, but not at concentrations higher than the well perfused tissues of liver, kidneys and lungs.</p><!><p>Five mice with intracervical ME-180 xenografts were treated with intravaginal irradiation of the cervical tumors. Two mice were irradiated at 150 J/cm2 and 3 were irradiated at 75 J/cm2. All recovered from treatment and anesthesia within 1–2 h and appeared active. Of the 2 mice who received Pc 4 PDT with a fluence of 150 J/cm2, one died 5 days after Pc 4 PDT, and one became scruffy and lethargic 2 days after Pc 4 PDT administration and was euthanized. Two of the 3 mice with intracervical tumors treated with Pc 4 PDT at a fluence of 75 J/cm2 were found dead in their cages 3 days post-Pc 4 PDT. The other mouse had tumor shrinkage based on palpation after intravaginal Pc 4 PDT but was euthanized over a month later secondary to regrowth of tumor over 2g at time of necrosy. Upon necrosy of all 4 mice who died after Pc 4 PDT, the tumors were reduced in size and were necrotic, but the intestinal tract was transparent and contained dark clots which suggested damage to the intestinal mucosa when the microfiber irradiation traveled beyond the small cervical tumors. One of the two mice who underwent treatment of its subcutaneous tumor with Pc 4 PDT at 150 J/cm2 was found dead 5 days post-Pc 4 PDT. Both subcutaneous tumors had decreased in size and were necrotic after Pc 4 PDT.</p><!><p>In this study we have evaluated the use of a second generation phototherapeutic in the treatment of cervical cancer. For both the in vitro and in vivo studies presented here, we chose cell lines that would represent the HPV status of most cervical cancers. Both CaSki and ME-180 uterine cervical cancer cell lines are epithelial in origin and were derived from patients infected with high risk HPV. CaSki cells were derived from a 40 year old Caucasian woman from a small intestinal metastasis and have an integrated human papillomavirus type 16 genome (HPV-16, about 600 copies per cell) as well as sequences related to HPV-18. ME-180 cells were derived from a 66 year old Caucasian women from a metastatic site in the omentum and contain HPV DNA with greater homology to HPV-39 than HPV-18. These cell lines were chosen because they contain DNA of the high risk HPV types (HPV 16, 18 and 39) which are considered to be carcinogenic and are associated with the development of cervical cancer. HPV types 16 and 18 as well as 45, 31, 33, 52, 58, and 35 are associated with 95% of squamous-cell carcinomas of the cervix[17]. Both CaSki and ME-180 cells grow well in culture both as monolayers and as spheroids.</p><p>In our studies, we compared the response to Pc 4 PDT between monolayers and spheroids grown on Matrigel. Spheroids appear to be more representative of tumors in their response to therapeutic agents [18]. Further, the cervical cancer spheroids were placed on a matrigel:complete medium:1:1 (v/v) foundation because cervical cancer appears to spread by local invasion to adjacent tissue structures and spheroids grown on extracellular matrix appear to be more representative of tumors [18,10]. Tumors are better represented by spheroids because: morphology and proliferation rates are different than in monolayers; there is more phenotypic herterogeneity; gene expression is altered compared to monolayers and gradients for nutrients, catabolites, oxygen; and drugs exist and there are 3D tensional forces between cells [19]. In the case of both the CaSki and ME-180 spheroids, between 200 and 1000μm in diameter, the growth rate was much slower than in the monolayers. Further, as expected, Pc 4 did not appear to be toxic to the CaSki or ME-180 spheroids and much higher concentrations of Pc 4 were required with the same laser dose to cause cytotoxicity in the tighter CaSki spheroids. The IC50 for CaSki spheroids was 0.26μM while the IC50 for the ME-180 spheriods was 0.011μM. This difference in response to Pc 4 PDT correlates with their 3D sturctures: the CaSki cells grown as spheroids were true spheres and very tightly compacted; the ME-180 spheroids were looser and grew in random 3D shapes. It was difficult to disrupt both types of spheroids with DMSO and other solvents; because of this, aqueous insoluble formazan crystals from the metabolism of MTT could not be completely extracted from CaSki nor ME-180 spheroids. The inability to extract the insoluable formazan crystals meant the assessment of the assays measuring cytotocity in the spheroids required microscopic imaging.</p><p>Because higher concentrations of Pc 4 were required for PDT to kill the spheroids, we investigated the intracellular concentrations of Pc 4 in the monolayers and spheroids. When the cells were grown as monolayers starting with 1×104 cells in 200μl of complete medium containing a concentration of 0.3μM Pc 4, by 6h the cells had taken up more than 20% of the Pc 4 introduced into the wells and the intracellular concentrations were at least 50-fold higher than the medium concentrations. As expected, the uptake into the spheroids was slower due to the diffusion gradients, and at a depth of 70μm into the spheroids, penetration of Pc 4 throughtout the spheroid was not complete until 24h of exposure. As expected, at earlier times, only the outer layers of the spheroid contained significant Pc 4 fluorescence (Figure 4).</p><p>The intracellular distribution of Pc 4 within the cervical cancer cells was similar to what has been observed in other tumor cell lines [12,13]. In both cervical cancer cell lines, Pc 4 was localized within the cytoplasm, lysosomes, endoplasmic reticulum and mitochondria. This wide intracellular distribution of Pc 4 allows for widespread damage throughout the cell when the laser is applied to activate Pc 4. Pc 4 PDT results in activation of many of the cell death pathways including apoptosis and necrosis as well as reactive oxygen species [20].</p><p>Prior to conducting PDT studies in mice we needed to establish the tumor models and understand the tumor growth characteristics when the cells were injected intracervically. In the mice the tumors grew in the cervix and in the peritoneum near the opening of the fallopian tubes above one horn of the bifurcated uterus. Only in advanced disease were metastases detected within the liver. Most lesions remained within the peritoneal cavity and firmly attached to the peritoneal wall. Mice had as many as 10 different tumor nodules in late stage disease, reflecting the metastatic nature of the tumor cell lines examined. The CaSki tumor xenografts grew more slowly within the peritoneum than did the ME-180 xenografts, in agreement with their doubling times when grown as monolayers. With intracervical implantation, the spread of these cells in SCID mice resembled the spread of human cervical squamous cell carcinomas in that it involved the paracervical and parauterine tissue. In advanced disease in mice it differed in that there was intraperitoneal spread. In humans cervical squamous carcinoma grows to involve the endometrium superiorly and the upper vagina inferiorly. Parametrial involvement results from extension through the cervical stroma. From the parametrium the tumor may extend laterally to the pelvic sidewall, anteriorly to the bladder base or posteriorly to the rectum. Squamous cell carcinoma also spreads to the endometrium, fallopian tubes, and ovaries [21–23]. Vascular and lymphatic invasion can occur leading to pelvic lymph node metastasis [23].</p><p>When Pc 4 concentration in xenografts and normal tissues was measured with time in the SCID mice bearing either CaSki or ME-180 xenografts, Pc 4 distributed rapidly to normal tissues with high perfusion rates, such as the liver, lungs and kidneys; these tissues also had the highest concentrations of Pc 4. At the earliest time point examined in this study, concentrations in these tissues were already at least a magnitude higher than plasma, and retained the highest concentrations out to the final observation time of 72h; plasma concentrations fell between 6 and 48 hours and could not be detected at 72h. Similarly, the concentrations in the colons and inguinal fat of the mice could not be detected beyond 24h. Uterus, skeletal muscle and tumor concentrations of Pc 4 were detected at 6h and remained relatively constant at approximately 0.5 to 1μM throughout the 72h time point. Because only a single mouse was used for each time point determination, the variability in the concentrations could not be determined. Nonetheless, this data is consistent with our previous observations of the concentrations of Pc 4 in normal CD2F1 female mice [10]. Based on the data presented here and our data from previous studies with other xenograft models in mice[15], we chose to apply the laser irradiation at 48h after IV administration of Pc 4 at 2 mg/kg.</p><p>Our attempt at laser irradiation through the vaginal opening to treat the intracerivcal lesions was not successful. Although the mice survived the initial irradiation and the tumor appeared to be killed, 4 of the 5 mice with intracervical tumors, and 1 of 2 mice with subcutaneous tumors, had to be euthanized or died between 2 and 5 days after irradiation. It was apparent on necropsy that the irradiation had affected the small intestines as they were transparent and filled with black clots suggesting severe damage to the mucosal lining. We had considered the emptying time of the gastrointestinal tract of mice, (estimated at 6h in mice when the major route of elimination was through the feces [24]) and had demonstrated that Pc 4 concentrations could not be detected in the colonic tissue past 24h; we should have considered, however, the slow release and redistribution of Pc 4 from storage sites within other tissues. Most probably, Pc 4 was present in the contents of the gastrointestinal tract at high enough concentrations to result in injury and death of the mice when 75 J/cm2 of irradiation at 670nm was applied through the microfiber.</p><p>Pc 4 has advantages over a number of other sensitizers as it is a defined compound with known spectra, while the spectra of tissue synthesized porphyrins after application of ALA or its related compounds are not known. Photofrin, Porfimer sodium, is a mixture of compounds as well, and has been used to treat CIN. Japanese studies have shown that colposcopic-assisted cercival canal illumination after intravenous Photofrin administration can achieve a high CR (<94%) [25]. A newer agent Radachlorin® has been shown to be an effective PDT agent against cervical cancer cells, [26, 27] however, the cell line, TC-1, was derived from epithelial cells of the lungs of C57BL/6 mice and transformed by HPV-16 E6 and E7. Photochlor (HPPH), like Pc 4, is a second generation photosensitizer that is currently in clinical trials. Both Pc 4 and Photochlor are single compounds with absorption spectra in the high 600's, have good tissue distribution and have better pharmacokinetics than Photofrin, with shorter periods of photosensitivity when administered intravenously [28].</p><p>This is the first study to investigate PDT utilizing Pc 4 and its effects on cervical cancer cells both in-vitro and in-vitro. We have demonstrated in vitro that Pc 4 itself in relatively non-toxic and Pc 4 PDT is effective at killing cervical cancer cells grown as either spheroids or monolayers. We have demonstrated in vivo that intracervical tumors in mice became necrotic after Pc 4 PDT. Previous studies using PDT with other photosensitizers to treat CIN have had promising results [8, 29]. We are encouraged therefore by the possibility of continuing these studies further in animals and eventually to women for the treatment cervical neoplasia. We feel the most clinically relevant application of Pc 4 PDT may be in its topical application in the treatment of CIN where light can be specifically directed at localized cervical lesions. The fact that PDT used in the treatment of CIN could potentially avoid the morbidity associated with current excisional or ablative techniques makes Pc 4 PDT worth investigating further. Pharmacokinetic studies would need to be performed in patients to identify the optimal drug dose for treatment of invasive cervical carcinoma due to the promixity of the lesion to vital organs. In addition, the application and dose of light would have to be focused more directly within the cervical lesion, either by direct application or interstitial therapy.</p>

PubMed Author Manuscript

Simultaneously Quantifying Ferrihydrite and Goethite in Natural Sediments Using the Method of Standard Additions with X-ray Absorption Spectroscopy

The presence of ferrihydrite in sediments/soils is critical to the cycling of iron (Fe) and many other elements but difficult to quantify. Extended X-ray absorption fine structure (EXAFS) spectroscopy has been used to speciate Fe in the solid phase, but this method is thought to have difficulties in distinguishing ferrihydrite from goethite and other minerals. In this study, both conventional EXAFS linear combination fitting (LCF) and the method of standard-additions are applied to the same samples in attempt to quantify ferrihydrite and goethite more rigorously. Natural aquifer sediments from Bangladesh and the United States were spiked with known quantities of ferrihydrite, goethite and magnetite, and analyzed by EXAFS. Known mineral mixtures were also analyzed. Evaluations of EXAFS spectra of mineral references and EXAFS-LCF fits on various samples indicate that ferrihydrite and microcrystalline goethite can be distinguished and quantified by EXAFS-LCF but that the choice of mineral references is critical to yield consistent results. Conventional EXAFS-LCF and the method of standard-additions both identified appreciable amount of ferrihydrite in Bangladesh sediments that were obtained from a low-arsenic Pleistocene aquifer. Ferrihydrite was also independently detected by sequential extraction and 57Fe M\xd3\xa7ssbauer spectroscopy. These observations confirm the accuracy of conventional EXAFS-LCF and demonstrate that combining EXAFS with additions of reference materials provides a more robust means of quantifying short-range-ordered minerals in complex samples.

simultaneously_quantifying_ferrihydrite_and_goethite_in_natural_sediments_using_the_method_of_standa

1. INTRODUCTION<!>2.1 Natural Sediments<!>2.2 Iron Oxide Minerals<!>2.3 Preparation of Standard-Additions and Mineral Mixtures<!>2.4 Iron EXAFS Spectra Collection and Processing<!>2.5 Additional Mineralogical Analyses<!>3.1 Iron Mineral References Included in EXAFS-LCF<!>3.2 Iron Mineral Composition of Unspiked Sediments<!>3.3 Performance of EXAFS-LCF on Known Binary Mineral Mixtures<!>3.4 Iron Mineral Composition Quantified By the Method of Standard-Additions<!>4.1 Comparison of Methods of Quantifying Ferrihydrite<!>4.2 Adequacy of EXAFS in Distinguishing Between Different Goethites<!>5. CONCLUSIONS AND IMPLICATIONS

<p>Ferrihydrite is a nanocrystalline/amorphous Fe(III) oxyhydroxide common in the natural environment, a critical adsorbent, and particularly susceptible to changes in redox conditions (Childs, 1992; Cornell and Schwertmann, 2003; Drits et al., 1993; Hiemstra, 2013; Michel et al., 2007; Willett et al., 1988). Ferrihydrite has a very large surface area up to 800 m2 g−1, meaning that if present in sediments and soils, it is often a key carrier for various metal(loid)s and nutrients, such as arsenic and phosphorus, and affects the turnover of organic carbon (Childs, 1992; Hiemstra, 2013; Johannesson et al., 2013; Michel et al., 2007; Sun and Bostick, 2015; Torn et al., 1997; Willett et al., 1988; Zhu et al., 2013). Ferrihydrite is also the most bioavailable Fe(III) mineral for dissimilatory Fe(III)-reducing bacteria (Hansel et al., 2005; Postma et al., 2010). The transformation of ferrihydrite, even to its more crystalline analogue, goethite, can significantly decrease surface area and liberate adsorbed species (Postma et al., 2010; Robinson et al., 2011; Willett et al., 1988). Differentiating ferrihydrite from other Fe minerals is therefore necessary to understand the biogeochemical cycles of Fe and numerous other elements associated in the environment.</p><p>Quantifying Fe mineral composition in complex sediments and soils remains an analytical challenge. Bulk X-ray diffraction (XRD) is insensitive to poorly crystalline minerals such as ferrihydrite or to any trace constituent (Houben and Kaufhold, 2011; Johnston and Lewis, 1983). Scanning and transmission electron microscopy can characterize ferrihydrite, but they are qualitative and not suited to quantify ferrihydrite at low concentrations (Akai et al., 2004; Childs, 1992; Johnston and Lewis, 1983). Mӧssbauer spectroscopy can detect low concentrations of Fe minerals including ferrihydrite, but responds to magnetic domains, which can contain multiple phases for intergrown and highly-substituted structures (Ginn et al., 2017; Johnston and Lewis, 1983; Postma et al., 2010). Other common and easily available methods for studying ferrihydrite include Brunauer–Emmett–Teller (BET) surface area analysis, infrared analysis, differential thermal analysis, cation exchange capacity analysis, and chemical extraction (Houben and Kaufhold, 2011; Poulton and Canfield, 2005). However, these methods are often not sufficiently quantitative to interpret Fe mineral composition in sediments and soils.</p><p>Extended X-ray absorption fine structure (EXAFS) spectroscopy measures backscattering photoelectrons with energies above the absorption edge of an element of interest and allows characterization of the atomic number, near-neighbor distances, coordination number, and less directly, bond angles (Bertagnolli and Ertel, 1994; O'Day et al., 2004). It is element-specific, sensitive to diluted phases, and provides direct measures of structure even in non-crystalline phases. EXAFS, therefore, has been used to speciate Fe in sediments and soils, typically through linear combinations of reference spectra or theoretical shell-by-shell fitting (Hansel et al., 2005; Karlsson et al., 2008; Schroth et al., 2009; Sparks, 2004; Sun and Bostick, 2015). However, the capability of EXAFS analysis to distinguishing between minerals having similar structures, for example, ferrihydrite and goethite, has been questioned (O'Day et al., 2004; Schroth et al., 2009).</p><p>One approach to assess the reliability of EXAFS on quantifying Fe mineral composition is to examine the spectra of Fe mineral mixtures with known composition. Using this approach, O'Day et al (2004) verified that EXAFS could accurately interpret mixtures of Fe sulfide and non-sulfide (phyllosilicate ± oxide) minerals. However, few studies have assessed the performance of EXAFS using minerals mixed within the same structural class, let alone using complex natural sediments and soils. The method of standard-additions is widely used in liquid-phase analyses to increase confidence in quantification when interferences are a potential concern. In this approach, the linear regression of the instrument's response to a known added amount of analyte is used to back-calculate the original concentration of the analyte by correcting for matrix effects and spectral interferences (Ellison and Thompson, 2008; Harris, 2010). Homogenization is a potential issue for using the standard-additions method on solids. In the solid phase, this method has been used for quantitative XRD (Harris, 2010; Hughes et al., 1994; Steenbruggen and Hollman, 1998), but, to our knowledge, has not been applied to quantify ferrihydrite by EXAFS.</p><p>This study applies both conventional EXAFS linear combination fitting (LCF) on a single sample and the method of standard-additions to quantify Fe mineral composition in natural sediments and mineral mixtures. Minerals used for spiking include ferrihydrite, goethite, and magnetite, all of which are Fe oxides but differ in arrangement of the basic octahedral structural units. Mineral compositions determined using the method of standard-additions were compared with those determined by conventional EXAFS-LCF method on unspiked samples, and with those determined independently by sequential extraction and 57Fe Mӧssbauer spectroscopy. Binary mineral mixtures were also prepared and examined.</p><!><p>Two aquifer sediments were selected for this study: one from the fluvial floodplain of central Bangladesh, and the other from the Dover Municipal Landfill Superfund Site in New Hampshire, USA. The two sediments are referred to hereafter as Bangladesh sediments and Dover sediments, respectively. Bangladesh sediments were obtained from a freshly drilled borehole in the village of Purinda in Araihazar upazila (23.8541°N, 90.6354°E) (Mihajlov, 2014). Immediately following drilling using the traditional hand-flapper method, a 30 cm long, 1.8 cm ID sediment core was recovered at ~18 m below ground surface (BGS) from the borehole with a manual push corer (AMS 424.45). The sediment core, with top and bottom several centimeters discarded, was then capped, wrapped with electrical tape, sealed and refrigerated in a nitrogen-flushed airtight Mylar bag with oxygen adsorbents (Sorbent Systems). Bangladesh sediments were also sampled from this borehole into polypropylene microcentrifuge tubes, coated with glycerol to prevent exposure to oxygen and to preserve reduced Fe minerals, sealed and refrigerated in another nitrogen-flushed airtight Mylar bag with oxygen adsorbents. Dover sediments from the Superfund site were obtained between 9 and 12 m BGS in the southeast corner of the landfill perimeter by sonic vibration drilling. Immediately following retrieval, the sediments were sealed in a steel can with epoxy liners and refrigerated. Bangladesh sediments were composed of orange-colored sand, whereas Dover sediments were composed of a mixture of gray-colored fine sand, silt and clay. X-ray fluorescence (XRF) analysis using an InnovX Delta Premium instrument indicate that bulk Fe concentrations are 2.0% in both samples (concentrations reported in this study are all on a dry mass basis). Bulk Fe analyses of four certified references (National Institute of Standards and Technology NIST 2709, 2010, 2711, and Chinese Geochemical Standard GSS 1) and one internal standard (Standard Lamont Observatory Sediment from the Hudson SLOSH III) containing 2.9 – 4.3% Fe on the same XRF instrument were consistent within 93 – 102% of the certified or accepted values.</p><!><p>Ferrihydrite and goethite were synthesized following the procedures of Schwertmann and Cornell (2000): Ferrihydrite was prepared by precipitating ferric nitrate with potassium hydroxide at pH 7–8; goethite was prepared by oxidizing ferrous sulfate in the presence of carbonate at pH 6–7, which according to Schwertmann and Cornell (2000), has particle size and morphology close to various natural goethites. Ferrihydrite and goethite were freeze dried and stored as powder. Mineral identities were confirmed as 2-line ferrihydrite and microcrystalline goethite (micro-goethite), respectively, by XRD analysis (Supplementary Material Figure SA1). Magnetite was obtained from Ward's Science and ground with an agate mortar-and-pestle to powder before use. The surface areas of the ferrihydrite, goethite and magnetite were 228, 24.9 and 2.77 m2 g−1, respectively, as determined by BET isotherm using nitrogen gas as adsorbate, although the actual surface area of ferrihydrite is likely underestimated by the BET method (Dzombak and Morel, 1990; Gustafsson, 2003). When calculating the concentrations of mineral additions, formulas Fe10O14(OH)2·0.74H2O, FeOOH and Fe3O4 were used for ferrihydrite, goethite and magnetite, respectively (Cornell and Schwertmann, 2003; Hiemstra, 2013).</p><!><p>Bangladesh sediments from the core were freeze dried and mortar-and-pestle ground for an hour before use, producing homogenized dry powder. A 1 g aliquot of the powder was used as an unspiked Bangladesh sample. To assess the reliability of using EXAFS to quantify ferrihydrite, a 20 g aliquot of the powdered Bangladesh sediments were spiked with five increments of known ferrihydrite mass and homogenized by grinding. After each increment homogenization, 1 g of the mixture was removed and labeled as one of the standard-additions. This process achieved an added Fe concentration ranging from 2000 to 9000 mg kg−1 (added Fe fraction ranging from 9 to 32% of total Fe, Table SA1) and thus a bulk Fe concentration from 2.2 to 2.9%. To test the effectiveness of EXAFS analysis in separating potentially interfering species, Bangladesh sediments were also spiked with both ferrihydrite and goethite. Another 20 g aliquot of the powdered Bangladesh sediments and three combinations of ferrihydrite and goethite were used to achieve an added Fe concentration ranging from 6000 to 9000 mg kg−1 (from 23 to 31% of total Fe, Table SA1) and thus a bulk Fe concentration from 2.6 to 2.9%. The Dover samples were prepared with additions of ferrihydrite and in some cases also magnetite. To enable comparison of preparation methods, ~100 g of moist Dover sediments were used without prior drying or grinding, and Dover sediments and spiked mixtures were homogenized by stirring with a polypropylene spatula. A 1.5 g aliquot of the homogenized moist Dover sediments was used as an unspiked Dover sample. A 30 g aliquot of sediments was spiked with five increments of ferrihydrite with 1.5 g removed after each increment, achieving an added Fe concentration ranging from 2000 to 8000 mg kg−1 (from 8 to 29% of total Fe, Table SA2) and thus a bulk Fe concentration from 2.2 to 2.8%. Another 30 g aliquot of sediments was used for spiking with three combinations of both ferrihydrite and magnetite, with an added Fe concentration ranging from 4000 to 6000 mg kg−1 (from 15 to 22% of total Fe, Table SA2) and thus a bulk Fe concentration from 2.4 to 2.6%. Each of the Dover samples was coated with glycerol. The remainder (~38.5 g) of the 100 g Dover sediments was weighed before and after oven drying, to determine the water content (23.6%) and correct dry weight factor. Five binary mixtures in known ratios, either ferrihydrite-goethite as dry samples or ferrihydrite-magnetite as glycerol-coated samples, were also prepared (Table SA3). All the samples were sealed in polypropylene microcentrifuge tubes and refrigerated prior to analysis, and analyzed by EXAFS within 48 hours following preparation.</p><!><p>Iron K-edge EXAFS spectra were collected at the Stanford Synchrotron Radiation Laboratory (SSRL) on Beamlines 11-2 and 4-1, which were equipped with 100- and 32-element Ge detectors, respectively. An aliquot of each sample was sealed in Kapton tape, and analyzed in fluorescence mode. The monochromator crystal used was Si(220) with phi angle of 90 degrees. Soller slits and a 6 μx Mn filter were used to minimize the effects of scattered primary radiation. The beam was detuned as needed to reject higher-order harmonic frequencies and prevent detector saturation. Scans were calibrated by setting Fe metal foil edge inflection to 7112 eV. EXAFS spectra of many commonly encountered Fe reference compounds were previously collected at SSRL in a consistent fashion.</p><p>EXAFS spectra were processed using the SIXpack interface (Webb, 2005) unless mentioned otherwise. For each sample/reference, parallel EXAFS scans were averaged, normalized with linear pre-edge and quadratic post-edge functions, and converted to k3-weighted chi function with a threshold energy (E0) of 7124 eV. Relevant Fe mineral references to be included in linear combination fitting (LCF) were selected based on previously published studies on Bangladesh and Dover aquifers (Aziz et al., 2016; Jung et al., 2012; Mihajlov, 2014; Sun et al., 2016a; Sun et al., 2016b). To be specific, relevant references included ferrihydrite, micro-goethite, hematite, magnetite, Fe-bearing silicates, mackinawite, and siderite (reference spectra are in Figure SA2). Additionally, SPOIL values from target transform analysis were used as a statistical criteria to evaluate whether the references selected were suitable. The SPOIL value indicates whether the vector of the tested reference spectrum (i.e., target) fits well or instead increases the error in the matrix of sample spectra reproduced. Targets having SPOIL values < 6 could be potential references to be included in fitting (Beauchemin et al., 2002; Strawn and Baker, 2009). To further evaluate the ferrihydrite and micro-goethite references selected, EXAFS of ferrihydrite and four goethites in our spectral library were compared (details are in Table SA4). Then, least-squares LCF was performed over k-range of 2 to 13 Å−1, to quantify the fractions (mol% Fe) of individual references in the sample. Uncertainties for EXAFS-LCF fits were obtained by SIXpack, which include error propagation from fitting, spectral noise in sample and reference spectra, and similarities between reference spectra (Webb, 2005). For standard-additions, the fractions (mol% Fe) were converted to concentrations (mg Fe per kg sediments, i.e., mg kg−1), by simply multiplying bulk Fe concentrations. "EXAFS-LCF determined concentration" in this study thus refers to product of bulk Fe concentration and EXAFS-LCF determined fraction.</p><p>For each set of standard-additions, a linear regression model was generated between the added concentrations of the analyte (ferrihydrite, goethite or magnetite), xi, and the apparent (i.e., the sum of original and added) concentrations determined by EXAFS-LCF, yi. The original concentration in the unknown sample, which corresponds to the absolute value of the x-intercept of this regression, was back-calculated according to the formula (Harris, 2010): (1)Original Concentrationstandard−additions=bm where m and b are the slope and y-intercept. For the method of standard-additions, uncertainty on the original concentration was calculated according to the formula (Harris, 2010): (2)Uncertaintystandard−additions=1m×[∑(yi−mxi−b)2N−2]×[1N+y¯2m2∑(xi−x¯)2] where N is the number of samples, m and b are the slope and y-intercept, and x̅ and y̅ are the average x- and y-values.</p><!><p>Sequential chemical extraction and 57Fe Mössbauer spectroscopy were used to provide independent measurements of Fe mineral composition that could be compared with EXAFS analysis. The sequential extraction procedure (see Table SA5 for details) was based on Poulton and Canfield (2005) and used in our previously published studies (Poulton and Canfield, 2005; Sun et al., 2016a; Sun et al., 2016b). The procedure includes four main steps to distinguish pools of Fe: (1) a 24 hr acetate extraction targeting carbonates, including siderite; (2) a 48 hr hydroxylamine-hydrochloride extraction targeting short-range-ordered oxides, including ferrihydrite; (3) a 2 hr dithionite-citrate extraction targeting crystalline oxides, including bulk goethite and hematite; and (4) a 6 hr ammonium oxalate extraction targeting recalcitrant oxides, including magnetite and presumably residual hematite from the previous step. Each extraction step was repeated once before proceeding to the next. Unspiked and spiked Bangladesh sediments were subjected to extraction after their EXAFS spectra were collected. The Dover samples prepared above were not used because they were glycerol-coated or oven-dried. Instead, fresh moist Dover sediments from the same container were freeze-dried, ground to powder, and subjected to extraction. Extractions were conducted at room temperature in constantly agitated polyethylene centrifuge tubes. Dissolved Fe concentrations in the extractions were determined by inductively coupled plasma mass spectrometry (Thermo Fisher Scientific Element XR) using previously published procedures (Sun et al., 2016a; Sun et al., 2016b).</p><p>57Fe Mӧssbauer spectroscopy was used, on unspiked Bangladesh sediments, to further assess Fe mineral composition. The analysis and spectra fitting routine were consistent with previously published procedures (Tishchenko et al., 2015). Mӧssbauer spectroscopy was performed with a variable temperature He-cooled system with a 1024 channel detector. A 57Co source embedded in a Rh matrix was used at room temperature. Velocity (gamma-ray energy) was calibrated using α-Fe foil at 298 K. The transducer was operated in constant acceleration mode and folding was performed to achieve a flat background. Each Fe mineral (site population) was quantified by the spectral fitting as a fraction of the total Fe spectral area. Quantification in this manner assumes equal Mӧssbauer recoilless fractions of all detected minerals, which should be valid at cryogenic temperatures and also be a good approximation at room temperature with dry samples (Tishchenko et al., 2015). Mӧssbauer analysis was also conducted on the synthesized ferrihydrite and goethite minerals. Additional details are contained in the Supplementary Material Section B.</p><p>Powder X-ray Diffraction (XRD) was also used, to determine sediment bulk mineralogy, including seeking to detect any possible Fe minerals. XRD analysis was carried out using a PANalytical X'pert3 Powder diffractometer, equipped with a PIXcel1D detector and a rotating sample stage. The diffractometer used Cu K-alpha radiation and scanned over 2θ range from 4° to 80°, with a step size of 0.013° and a counting time of 1 min per step.</p><!><p>For unknown multi-mineral assemblages, EXAFS-LCF requires proper identification of the minerals present and inclusion of their spectra in fitting. To be consistent with previously published studies (Aziz et al., 2016; Jung et al., 2012; Mihajlov, 2014), ferrihydrite, micro-goethite, hematite, magnetite, Fe-bearing silicates, mackinawite, and siderite were selected as references for the samples derived from the Bangladesh sediments. These references were then evaluated based on their SPOIL values. Ferrihydrite and micro-goethite, which were used in spiking Bangladesh sediments, had SPOIL values of 2.91 and 4.45, respectively; the other selected references also had SPOIL values < 6 and thus were acceptable references in EXAFS-LCF (Table SA6). To be consistent with previously published studies on the Dover Superfund site (Sun et al., 2016a; Sun et al., 2016b), the minerals selected for the Bangladesh sediments were also selected as references for the samples derived from the Dover sediments. Ferrihydrite and magnetite, which were used in spiking Dover sediments, had SPOIL values of 1.45 and 1.92, respectively; the other references also had SPOIL values < 6 (Table SA6).</p><p>In addition, the spectral signatures of ferrihydrite, micro-goethite and three other environmentally relevant goethites, which vary with particle size and morphology, were compared (Figure 1 and Figure SA3). Variations between their EXAFS spectra were mostly observed over the k-range of 5 – 10 Å−1, where Fe backscattering has the highest amplitude. Although the interatomic Fe-Fe distances in these minerals are roughly the same, the Fe-Fe shells in micro-goethite have higher amplitudes than those in ferrihydrite and in nanocrystalline goethite (nano-goethite) (Figure 1B). Micro-goethite also show high spectral similarity with the two natural goethites available (Figure 1 and Figure SA3). Because micro-goethite closely resembles natural goethites in soils and sediments (Bertsch and Seaman, 1999; Schwertmann and Cornell, 2000) and was the goethite used for spiking the sediment samples, micro-goethite was regarded as the most appropriate goethite reference.</p><!><p>For unspiked Bangladesh sediments, EXAFS-LCF (Table 1 and Table SA1) indicated a ferrihydrite concentration of 12700 ± 1700 mg kg−1 (64% ± 9% of total Fe) and a goethite concentration of 1740 ± 660 mg kg−1 (9% ± 3% of total Fe). EXAFS-LCF also reported similar results on glycerol-coated sediment samples collected from the same borehole (Figure SA4). Consistent with EXAFS-LCF, sequential extraction and 57Fe Mӧssbauer spectroscopy identified short-range-ordered Fe phases (e.g., Ferrihydrite and nano-goethite) as the major Fe phases. Sequential extractions (Table 1 and Table SA7) indicated an amorphous Fe oxide concentration of 7770 mg kg−1 (39% of total Fe, 58% of extractable Fe) and a crystalline Fe oxide concentration of 3620 mg kg−1 (18% of total Fe, 27% of extractable Fe). 57Fe Mӧssbauer spectroscopy (Table 2, Figure 2 and Table SB1) indicates a ferrihydrite concentration of 5370 ± 990 mg kg−1 (27% ± 5% of total Fe), a nano-goethite concentration of 4180 ± 200 mg kg−1 (21% ± 1% of total Fe), and an unidentified, highly disordered nano-scale Fe(III) oxyhydroxide concentration of 3780 ± 600 mg kg−1 (19% ± 3% of total Fe), which likely represents highly-substituted ferrihydrite or nano-goethite phases. As expected, XRD analysis could not identify ferrihydrite or any other Fe mineral in the Bangladesh sediments (Figure SA5).</p><p>For unspiked Dover sediments, EXAFS-LCF (Table 1 and Table SA2) indicated a ferrihydrite concentration of 8140 ± 1690 mg kg−1 (41% ± 9% of total Fe) and a magnetite concentration of 0 ± 470 mg kg−1 (0% ± 2% of total Fe). Compared to EXAFS-LCF fits, sequential extraction (Table 1, Table SA7 and Figure SA6) indicated a much lower concentration of amorphous Fe oxide, 1420 mg kg−1 (7% of total Fe, 18% of extractable Fe), and a higher concentration of recalcitrant Fe oxide, 2130 mg kg−1 (11% of total Fe, 27% of extractable Fe). Again, XRD analysis failed to identify any Fe mineral in the Dover sediments (Figure SA5).</p><!><p>To determine if EXAFS-LCF can quantify ferrihydrite in the presence of other Fe minerals, known ferrihydrite-goethite mixtures and ferrihydrite-magnetite mixtures were examined. Fits were performed with spectra of the two known end-members, and also with spectra of all the environmentally relevant minerals that were used to fit natural sediments (Figure 3 and Table SA3). In either case, EXAFS-LCF fits agreed with known composition (Figure 4). When extra mineral references were used, EXAFS-LCF incorrectly reported a low concentration, 4% on average, of Fe-bearing silicates but correctly excluded four minerals that were not present.</p><!><p>EXAFS-LCF combined with the method of standard-additions was used to increase confidence in the quantification of Fe minerals in natural complex samples (Figure 5). For Bangladesh sediments, EXAFS-LCF fits on samples with ferrihydrite added in known concentrations (ferrihydrite-additions, Figure 6A) indicated an original ferrihydrite concentration of 13200 ± 2000 mg kg−1 (66% ± 10% of total Fe), and fits on goethite-additions (Figure 6B) indicated an original goethite concentration of 2680 ± 910 mg kg−1 (13% ± 5% of total Fe). When the concentrations of goethite were changing, the EXAFS-LCF fits on ferrihydrite stayed constant; for sediments unspiked and spiked with ferrihydrite/goethite, EXAFS-LCF fits of the other Fe minerals including Fe-bearing silicates stayed nearly identical (Table SA1). The method of standard-additions was also combined with sequential extractions. The concentration of Fe solubilized in hydroxylamine-HCl extraction step increased proportionally with increments of ferrihydrite spiked into Bangladesh sediments (Figure 7A). Hydroxylamine-HCl extractions on ferrihydrite-additions indicated an original amorphous Fe oxide concentration of 6800 ± 1360 mg kg−1 (34% ± 7% of total Fe, 49% ± 10% of extractable Fe). Responding to the additions of ferrihydrite, EXAFS-LCF and extractions both correctly indicated increased ferrihydrite concentrations, whereas XRD was insensitive as expected (Figure SA7).</p><p>For Dover sediments, EXAFS-LCF fits on ferrihydrite-additions, on the whole, indicated lower original ferrihydrite concentration than fits on the single unspiked sample (Figure 8 and Figure 9A). If the unspiked Dover sample was excluded, EXAFS-LCF fits on ferrihydrite-additions indicated an original ferrihydrite concentration of 1020 ± 1000 mg kg−1 (5% ± 5% of total Fe). Compared to Bangladesh ferrihydrite-additions, the data on Dover ferrihydrite-additions showed more scatter (Figure 6A versus Figure 9A). EXAFS-LCF fits on Dover magnetite-additions (Figure 9B) indicated an original magnetite concentration of 130 ± 300 mg kg−1 (1% ± 2% of total Fe).</p><!><p>To determine if the "ferrihydrite" detected by EXAFS-LCF was truly ferrihydrite, this method was tested on known binary mineral mixtures (Figure 4). The result indicated that ferrihydrite could be differentiated from micro-goethite or other Fe oxides using EXAFS-LCF, even though their spectral similarity can complicate such differentiation (O'Day et al., 2004). EXAFS-LCF fits with extra mineral references agreed with fits with the two end-members, which further indicated the robustness of EXAFS-LCF and, to some extent, the uniqueness of the reference spectra. However, when extra references were used, as much as 4% of a component, often Fe-bearing silicates, might be included when it is not present (Table SA3). This indicates that the practical detection limit of EXAFS-LCF is on the order of 3 – 5%, similar to what was observed in other studies (O'Day et al., 2004).</p><p>For complex sediments/soils, if references are well defined, conventional EXAFS-LCF is effective at quantifying ferrihydrite; if Fe-bearing silicates or other minerals are variable in structure or not representatively described by a few references, then conventional EXAFS-LCF alone might be less accurate. An advantage of combining EXAFS-LCF with the method of standard-additions is that fitting error(s) caused by the choice of references is distributed uniformly over unspiked and spiked samples, at least when the addition does not appreciably change the bulk Fe concentration (in this study, the addition of Fe oxide minerals increased bulk Fe concentration from 2.0% to 2.2~2.9% in Bangladesh sediments and from 2.0% to 2.2~2.8% in Dover sediments). As such, the accuracy of quantifying ferrihydrite instead depends on the sensitivity of EXAFS to measure changes in the abundance of ferrihydrite. For Bangladesh sediments, the regression for ferrihydrite showed good linearity and a slope close to 1, and was not affected by the samples simultaneously spiked with goethite (Figure 6A). This indicates that EXAFS-LCF is able to respond systematically to additions of ferrihydrite in multi-mineral assemblages, even in the presence of goethite. Furthermore, the fits of the other minerals including Fe-bearing silicates stayed constant (Table SA1), indicating that they did not bias fits. For Bangladesh sediments, ferrihydrite quantified by EXAFS-LCF standard-additions agreed with the conventional EXAFS-LCF method on a single unspiked sample, and also had comparable uncertainty (Table 1). Given detailed prior work identified all the key references to include in the LCF model (Aziz et al., 2016; Jung et al., 2012; Mihajlov, 2014), it is expected that the EXAFS-LCF standard-additions and single sample methods agree. In cases where no prior work on Fe mineralogy is available, the concentration and uncertainty determined by EXAFS-LCF standard-additions should be more reliable given that it is less affected by the choice of other references.</p><p>Ferrihydrite in the Bangladesh sediments was also quantified by 57Fe Mӧssbauer spectroscopy and sequential extraction. Mössbauer is extremely sensitive to the degree of short-range ordering or crystallinity — much more so than EXAFS or XRD — such that the most highly substituted or disordered Fe oxyhydroxide phases exhibit only partial ordering (a collapsed sextet) at 5K (Figure 2). These highly-disordered phases may have angstrom-level atom spacing consistent with ferrihydrite or goethite, but regardless, the phases do not exhibit crystal ordering beyond a few nanometers. For Bangladesh sediments, nearly half of the Fe magnetically orders was consistent with either ferrihydrite or these highly disordered, nanocrystalline Fe(III) phases, which agreed with the abundance of ferrihydrite detected using EXAFS (Tables 1 and 2). Mӧssbauer spectroscopy also indicated broadly consistent composition of the other Fe minerals with EXAFS-LCF and validated the choice of mineral references included in fitting. As for sequential extraction, although it has certain limitations, this method is unique in that each extraction step corresponds to the reactivity of specific mineral class (Poulton and Canfield, 2005). Compared to ferrihydrite concentrations determined by EXAFS-LCF, the concentrations of hydroxylamine-HCl extractable Fe were consistently lower (Figure 7B and 7C). This can be attributed to incomplete extractions, something that is commonly observed in extraction experiments (Bacon and Davidson, 2008; Gleyzes et al., 2002). Nevertheless, hydroxylamine-HCl extraction, both conventional and combined with standard-additions, indicated significant concentration of amorphous Fe oxides in Bangladesh sediments (Table 1).</p><p>Conventional EXAFS-LCF also detected ferrihydrite in the Dover sediments (Table 1). However, this result disagrees with previous studies on the Dover Superfund site (Sun et al., 2016a; Sun et al., 2016b). The Dover sediments were gray-colored sediments from underneath a landfill that should contain limited quantity of oxidized reactive Fe. EXAFS-LCF fits on Dover ferrihydrite-additions successfully revealed the lack of ferrihydrite in the original sample (Figure 9A), which was further supported by hydroxylamine-HCl extraction (Table 1). The difference between EXAFS-LCF on a single sample and standard-additions, therefore, most likely reflects sediment heterogeneity or potentially sample preservation issue. For complex sediment/soil samples, it is hard to envision small aliquots being perfectly representative all the time. Such issues are not something that EXAFS-LCF itself or other analytical techniques can overcome. Furthermore, the Dover ferrihydrite-additions were less linear than Bangladesh ferrihydrite-additions (Figure 6A versus Figure 9A). The poorer fit is likely a result of difficulty with homogenizing small volumes of unpowdered solid materials, especially when they contain mixtures of sand, silt and clay. The comparison implies that grinding is a better method for homogenization.</p><!><p>For the Bangladesh sediments, conventional EXAFS-LCF quantified a lower concentration of goethite than the EXAFS-LCF standard-additions method (Figure 6B). This apparent disagreement is likely because the synthetic micro-goethite used for spiking was not perfectly representative of the natural goethite in the Bangladesh sediments. Although in theory such micro-goethite is close to natural goethites (Bertsch and Seaman, 1999; Schwertmann and Cornell, 2000), in reality soil and sedimentary goethites are variable in composition and structure. [Note that ferrihydrite may also exhibit structural variations, partially due to interferences from aluminum and silicon (Adra et al., 2013; Wang et al., 2015).] Independent evidence from the Mossbauer analysis indicates that the Fe(III) oxides in the unspiked Bangladesh sediments at 295K are superparamagnetic or near their blocking temperature (Supplementary Material Section B). This contrasts with the synthetic micro-goethite used in the standard-additions experiment, which yielded a full sextet at 295K (Table SB2 and Figure SB1). Thus, natural goethite in the Bangladesh sediments is more disordered or of smaller particle size than the synthesized micro-goethite and similar to nano-goethites typically found in highly weathered soils and sediments (Ginn et al., 2017; Thompson et al., 2011; Tishchenko et al., 2015; van der Zee et al., 2003).</p><p>While ferrihydrite and micro-goethite have sufficiently distinct EXAFS spectra, it appears that the EXAFS spectra of ferrihydrite and nano-goethite are similar (Figure 1 and Figure SA3). If micro-goethite and nano-goethite were both present in the sample, including only the micro-goethite reference in fitting would possibly result in an underestimation of goethite and an overestimation of ferrihydrite concentrations. One potential solution is to include both micro- and nano-goethite in fitting. The goethite-like phases detected by Mössbauer (21% of total Fe, Table 2) were best approximated by nano-goethite standards and those phases exhibited a continuum of crystallinity based on ordering temperature that is consistent with multiple populations of goethite (Supplementary Material Section B.3.2). This suggests using two goethite references in EXAFS-LCF could be effective to differentiate goethite based on crystallinity. To test this idea, EXAFS-LCF were redone on Bangladesh goethite-additions with including both micro- and nano-goethite as references (Table SA8). From a statistical perspective based on the goodness-of-fit parameter, reduced χ2, including nano-goethite did not significantly improve the individual fits. This is not surprising considering that the spectral signature of nano-goethite can be described (or say masked) by the combination of ferrihydrite and micro-goethite. Nevertheless, using the method of standard-additions, the regression combining micro- and nano-goethite had a slope of 0.82 (Figure 6C), much closer to 1 than fits using micro-goethite alone (a slope of 0.57, Figure 6B). The new fits indicated an original goethite concentration of 3430 ± 1170 mg kg−1 (17% ± 6% of total Fe) from conventional EXAFS-LCF and 3530 ± 2960 mg kg−1 (18% ± 15% of total Fe) from standard-additions. The new fits suggested a nano-goethite concentration of about 2000 mg kg−1 (10% of total Fe, Figure 6C), which was previously fit as ferrihydrite. Using nano-goethite alone (no micro-goethite) in fitting resulted in unstable fits for goethite and other Fe minerals (Table SA8).</p><p>The inclusion of nano-goethite (in addition to ferrihydrite and micro-goethite) in EXAFS-LCF, especially when combined with standard-additions, may provide more mineralogical information than has previously been possible. However, using spectroscopically similar references in fitting produces larger uncertainty (Figure 6C) and increases complexity in data interpretation.</p><!><p>This study verified the capability of EXAFS analysis to distinguishing ferrihydrite from other Fe minerals including micro-goethite, and verified the accuracy of conventional EXAFS-LCF. This is a timely, important verification as EXAFS analysis has been becoming one of the most popular analytical geochemical research tools. Furthermore, this study represented an initial attempt to apply the method of standard-additions to EXAFS-LCF analysis (and also to sequential extraction) using real sediments. Such application improves our ability to quantify Fe mineral composition in complex natural samples. In addition to Fe minerals, our general observations regarding quantification could be transferable to EXAFS-based analysis of other mineral phases. Due to limited time available on synchrotron-based EXAFS techniques, increased number of analyses required for standard-additions, as well as effort required for sample preparation including homogenization, it is probably difficult to involve standard-additions in routine EXAFS analysis. Nevertheless, such application provides a means of more conclusively detecting short-range-ordered minerals in unknown matrices. Data from the method of standard-additions are also less biased by any single sample when heterogeneity is present (as is often the case with natural sediments and soils). To evaluate the performance of EXAFS-LCF standard-additions more comprehensively, continued efforts are required and could be put into studying samples with unusually high or low bulk Fe concentrations, more variable degrees of crystallization, aluminum-substituted ferrihydrite and goethite, and variable contents of organic Fe species etc.</p><p>Another finding of this study is that all applied methods reveal the presence of ferrihydrite in Bangladesh sediments (Figure 10). This finding is significant because this short-range-ordered mineral is highly reactive with respect to microbial reduction, metal(loid) retardation and other processes. The Bangladesh sediments were obtained from a low-arsenic Pleistocene aquifer that provides critical drinking water resources in South and Southeast Asia, because groundwater of Holocene aquifers is contaminated with arsenic (Harvey et al., 2002; Horneman et al., 2004; Mihajlov, 2014; Zheng et al., 2005). Despite the broad consensus that Fe(III) oxides are more prevalent in Pleistocene aquifers than in Holocene aquifers, the presence of easily reducible ferrihydrite in Pleistocene aquifers is controversial, and as a result, ferrihydrite is often assumed to be absent (Jessen et al., 2012; Polizzotto et al., 2006; Stollenwerk et al., 2007). The multiple approaches applied in this study consistently indicate the presence of ferrihydrite (and nano-goethite) in Pleistocene aquifers and minimize the chances of a fitting artifact or incorrect attribution. This points out the need to better understand and document the distribution of ferrihydrite in these Pleistocene aquifers, to ensure more robust predictions of the long-term fate of arsenic and to design suitable remediation measures.</p>

PubMed Author Manuscript

The role of EP-2 receptor expression in cervical intraepithelial neoplasia

Prostaglandin induced signalling is involved in different cancers. As previously described, the EP3 receptor expression decreases with increasing stage of cervical intraepithelial lesions (CIN). In addition, in cervical cancer EP3 is an independent prognosticator for overall survival and correlates with FIGO stages. Currently the role of Prostaglandin 2 receptor 2 (EP2) in CIN is unknown. The aim of this study was to analyse the expression of EP2 for potential prognostic value for patients with cervical dysplasia. EP2 expression was analysed by immunohistochemistry in 33 patient samples (CIN1–3) using the immune-reactivity scoring system (IRS). Expression levels were correlated with clinical outcome to analyse prognostic relevance in patients with CIN2. Data analysis was performed using non parametric Kruskal–Wallis and Spearman rank sum test. Cytoplasmic expression levels of EP2 correlated significantly (p < 0.001) with different grades of cervical dysplasia. Median EP2-IRS in CIN1 was 2 (n = 8), 3 in CIN2 (n = 9) and 6 in CIN3 (n = 16). Comparing regressive (n = 3, median IRS = 2) to progressive (n = 6, median IRS = 4) CIN2 cases the median IRS differed significantly (p = 0.017). Staining intensity (p = 0.009) and IRS (p = 0.005) of EP2 and EP3 correlate inversely. EP2 expression level significantly increases with higher grade of CIN and could qualify as a potential prognostic marker for the regressive or progressive course in CIN2 lesions. These findings emphasize the significant role of PGE2 signalling in CIN and could help to identify targets for future therapies.Electronic supplementary materialThe online version of this article (10.1007/s00418-020-01909-2) contains supplementary material, which is available to authorized users.

the_role_of_ep-2_receptor_expression_in_cervical_intraepithelial_neoplasia

Introduction<!>Tissue samples<!><!>Immunohistochemistry<!>Statistical analysis<!>EP2 expression increases with progressing grade of cervical dysplasia<!>Cytoplasmic IRS of EP2 positive cells is higher in CIN 2 lesions with a progressive course of the dysplasia<!><!>Intensity and IR-Score of EP2 correlates negatively with EP3 IRS and intensity<!>Discussion<!><!>Funding<!>Data availability<!>Conflict of interest<!>Ethical approval<!>Informed consent

<p>After breast-, colorectal- and lung cancer, cervical cancer represents the fourth most common malignant tumour in women worldwide (Wallis 2014; Watson et al. 2014). Approximately 500,000 women worldwide are newly diagnosed with cervical cancer per year. 260,000 women die from the disease each year (Gottlieb 2016; Jiang et al. 2015; Landy et al. 2016). Incidence and mortality of cervical cancer correlates negatively with the Human Development Index and varies extremely in geographic contexts (Wentzensen 2016). Regarding Germany, 4500 women were diagnosed with cervical cancer in 2014 and 1500 of these patients died tumour associated (Zentrum fur Krebregisterdaten 2019). After the implementation of Pap smear screening, which detects precursor lesions of cervical epithelium, incidence dropped considerably (Hester et al. 2019). The persistent infection with specific types of high-risk papillomaviruses is considered the main risk of intraepithelial neoplasia and especially in the development of cervical cancer (Schiffman et al. 2011). The precursor lesions were formerly called cervical intraepithelial neoplasia (CIN) and ranged from CIN1 to CIN3 (Santesso et al. 2016). In 2014 the histological WHO classification has been altered, and cervical intraepithelial neoplasia is referred to as squamous intraepithelial lesion (SIL) since 2014 (Lu and Chen 2014). The lesions are divided in low grade (LSIL) and high grade squamous intraepithelial lesions (HSIL). CIN2 and CIN3 are now combined in HSIL (Lu and Chen 2014). However, pathologists still specify their diagnosis with CIN2/CIN3 due to the risk of progression to a cervical carcinoma that may differ between CIN2 and CIN3 (Luo et al. 2018; Papoutsis et al. 2017). Consequently, the therapy options also vary from conservative approaches to surgical treatments (Saah-Briffaut et al. 2006). Young women in childbearing age could especially profit from a watchful waiting strategy as conization increases the appearance of pregnancy complications such as cervical insufficiency and preterm labour (Wilkinson et al. 2015). However, apart from the size of the lesion there is no established marker for the prediction of progression or remission of CIN2 lesions (Kühn et al. 2015).</p><p>Heidegger et al. previously indicated that the prostaglandin E2-receptor EP3 is an independent negative prognostic factor in cervical cancer patients. The expression levels and the clinical outcome were proven to correlate with tumour stage (Heidegger et al. 2017). In addition, Hester et al. demonstrated that EP3 receptor expression levels correlate inversely with grades of CIN (Hester et al. 2019). Our aim was to further investigate the role of prostaglandin receptors in cervical intraepithelial neoplasia. This study is focussed on the EP2 receptor, as it is unique among all EP receptors. The fact that it is not desensitized by Prostaglandin E2 (PGE2) sets it apart from other EP receptors and highlights its role in the deferred phases of cellular response (Kalinski 2012).</p><!><p>The cervical tissue samples used in this study were collected from patients treated between 2007 and 2014 in the Department of Gynaecology and Obstetrics from Ludwig-Maximilians-University of Munich, Germany. This cohort was analysed in previous studies from our group (Hester et al. 2019; Kolben et al. 2016; Vogelsang et al. 2020). Due to multiple sections the CIN lesions got lost on the slides in many cases, which therefor had been excluded from the present study.</p><p>In total, 38 tissue samples of cervical dysplasia were immunohistochemically stained with anti-EP2-antibody; the staining was successful in 33 cases. Of these, 8 were classified as CIN1, 9 as CIN2 and 16 as CIN3. On their first visit all patients were tested positive for high risk Human Papillomavirus (Hybrid Capture 2, Quiagen). Histopathological grade of dysplasia and diagnosis were confirmed by a second gynaecological pathologist. Regarding the CIN2 collective, only cases with either a histologically confirmed progress (n = 6) or regress (n = 3) were used. The follow-up interval for patients with CIN2 ranged from 5 to 14 months. The cases that were classified as CIN2 at the latest possible date and had been ranked as CIN3 previously, were defined as regress. CIN2, which had progressed from a former CIN1 were also defined as progress.</p><p>The tissue samples were eligible for this study after all routine histopathological diagnostic procedures were completed. The data of the patients were completely pseudonymized. All analytic procedures complied with the Helsinki Declaration guidelines (Reference No. 167-14). Informed consent of the patients was guaranteed before study participation. The Ethics Committee of the Ludwig-Maximilians-University (Munich, Germany) accepted the design of the study.</p><!><p>The expression of EP2-receptors in the cytoplasm increased significantly with increasing grade of cervical dysplasia, displayed by boxplots. The median value is stated above the median-line within the boxes (a). The images show representative microphotographs of EP2 staining in CIN1 (b IRS 2), CIN2 (c IRS 6) and CIN3 (d IRS 9). 200× magnification was used for picture b, c and d. Scale bars refer to 100 µm. Asterisk represents statistically significant differences in the staining results of CIN1-3</p><p>The median IR-score in regressive cases and in progressive cases differs significantly as shown by the boxplots (a). The different staining results with the anti-EP2-antibody in regressive (b IRS 1) versus progressive (c IRS 6) CIN2 samples (p = 0.017). 200× magnification was used for picture b and c. Scale bars refer to 100 µm. Asterisk represents statistically significant differences comparing regressive and progressive CIN2 cases</p><!><p>Immunohistochemistry regarding the EP3 were derived from a previous study performed by our group (Hester et al. 2019).</p><!><p>For statistical analysis SPSS 25 (PASW Statistic, SPSS Inc., IBM, IL, USA) was used. To compare the expression of EP2 in varying levels of the cervical dysplasia the non-parametric Kruskal–Wallis rank-sum test was applied. The correlation between levels of EP3 and EP2 was tested with the non-parametric Spearman rank correlation test. p values ≤ 0.05 were considered as statistically significant. Figures were configured with SPSS 25 and Microsoft Power Point 2016 (Microsoft, Redmond, WA, USA).</p><!><p>We compared the EP-2-IR-scores between the groups of CIN1-3 to analyse differences in EP2 expression levels. The expression of EP2-receptors in the cytoplasm increased significantly in correlation with increasing grade of cervical dysplasia as shown in Fig. 1a. This difference was statistically significant when each grade of dysplasia was compared to the next higher one. In CIN1 the median EP2-IRS in the cytoplasm was 2, in CIN2 incidents the value was 3 and in CIN3 cases the median EP2-IRS was 6 (p < 0.001).</p><p>Exemplary staining for all grades of CIN are shown in Fig. 1b–d.</p><!><p>To determine if EP2-receptor expression might serve as a prognostic factor in regard to a progressive or regressive course in cervical dysplasia, we compared EP2 expression between CIN2 cases with histologically confirmed regress or progress. Although the number of cases was little (n = 3 for regress, n = 6 for progress) the study revealed statistically significant differences between the cytoplasmatic IRS of EP2-receptor expressions. In regressive cases the median IR-score was 2, while it was 4 in progressive cases (p = 0.017) as shown in Fig. 2a. Figure 2b and c display the different staining in regressive (Fig. 2b) versus progressive (Fig. 2c) CIN2 samples (p = 0.017).</p><!><p>Correlation between EP3 and EP2 staining results</p><p>**represent statistically significant differences at (p < 0.001)</p><p>Sig. significance, N number of cases, IRS immunoreactive score</p><p>Correlation diagram for IR-score of EP2 and EP3 representing the inverse correlation of the two prostaglandin receptors in CIN tissues (a).The comparison of the staining results in a tissue sample of the same patient with CIN1 for a staining with the anti-EP2-antibody (b) and the anti-EP3-antibody (c) represents this inverse correlation. EP2 was not detected in the staining (IRS 0) whereas EP3 was seemingly highly expressed (IRS 12). 200× magnification was used for picture b and c. Scale bars refer to 100 µm</p><!><p>Figure 3b and c show the comparison of the staining results in the same tissue sample of CIN1 for a staining with EP2 (IRS 0) and EP3 (IRS 12), representing the inverse correlation of the receptor types.</p><!><p>Herein we analysed the expression of the EP2 receptor in CIN samples for potential prognostic information for patients with cervical dysplasia. First, the level of EP2 receptor expression was compared to the grade of the dysplasia. In addition, we correlated the receptor expression to the clinical course of CIN2 samples. The analysis revealed that the median IR score of EP2 increases significantly with increasing grade of dysplasia (CIN1 = 2, CIN2 = 3, CIN3 = 6). CIN2 patients with a regressive clinical course had significantly lower EP2 levels compared to those with a progressive course. Therefore, increasing EP2 expression might indicate a progression of CIN towards cervical cancer.</p><p>The small number of CIN samples analysed (n = 33) is a critical limitation of our study. The study group proved to be adequate powered and well-reviewed by previous studies of our work group (Hester et al. 2019; Kolben et al. 2016; Vogelsang et al. 2020). However, due to several sections of the cervical biopsies, cases with missing CIN on the slide had to be excluded in the present study. The possibility of a colposcopy sampling error in the follow up check might represent an additional potential problem. In general, larger patient cohorts and further studies are needed to validate our findings.</p><p>The EP2 receptor is a G-protein coupled receptor with seven transmembrane domains bound to a heterotrimeric G-protein comprising the stimulatory Gαs and Gβγ subunits (Gilman 1987). It is physiologically activated by PGE2, a proinflammatory factor with immunosuppressive function (Phipps et al. 1991). PGE2 derives from arachidonic acids, which is firstly converted to prostaglandin H2 by cyclooxygenase 1/2 (COX 1/2) enzymes and further processed by PGE2 synthases (Lambeau and Lazdunski 1999). PGE2 is known to operate in many processes such as apoptosis, angiogenesis, chronic inflammation, tumour immunity, proliferation, migration and invasion (Kalinski 2012). Compared to other EP receptors, EP2 is interestingly not desensitized by PGE2 and therefore may contribute to deferred phases of cellular response (Nishigaki et al. 1996).</p><p>The Gα activation of the EP2 receptor can result in increased cAMP levels and activation of protein kinase A which regulates downstream transcription factors such as cAMP response element-binding protein (Fujino et al. 2005). Direct binding of Gα to regulator of G protein signalling promotes the release of glycogen synthase kinase-3β (GSK-3β) resulting in the activation of β-catenin pathway, which triggers the transcription of genes such as c-myc, cyclin d1 and vascular endothelial growth factor (Vaid et al. 2015). However, activation of serine/threonine-specific kinase (Akt) via Gβγ and phosphoinositide-3-kinase (PI3K) results in the inactivation of GSK-3β (Castellone et al. 2005). As a consequence, accumulated β-catenin can migrate to the nucleus to stimulate gene transcription via TCF/LEF family of transcription factors (Prasad and Katiyar 2014). When EP2 forms a complex with β-arrestin it can also function in a G protein-independent manner (Chun et al. 2009). With β-arrestin as a regulator EP2 can inaugurate pathways of PI3K, Akt, proto-oncogene tyrosine-protein kinase Src, extracellular signal-regulated kinases, c-Jun N-terminal kinases and epidermal growth factor receptors (Sun and Li 2018).</p><p>To this point, very little knowledge has been identified of the prostaglandin receptors in cervical intraepithelial neoplasia. Hester et al. showed that EP3 expression significantly decreases with higher grades of cervical intraepithelial neoplasia (Hester et al. 2019) and the expression levels of EP3 correlate with tumour stage as well as clinical outcome as Heidegger et al. could confirm (Heidegger et al. 2017). However, currently comparable studies analysing the expression of EP2 in cervical dysplasia are missing.</p><p>The role of EP2 has been studied in many malignancies as most of the induced pathways play a major role in cell proliferation, migration and angiogenesis (Bonanno et al. 2016; Sobhani et al. 2018). For instance, aberrant expression of EP2 has been found to be associated with chronic inflammation, deregulation of the immune system, angiogenesis, metastasis as well as multidrug resistance and has been observed in cancer of the colon, liver, breast and cervix (Asting et al. 2017; Cui et al. 2017; Gong et al. 2017; Huynh 2017). Besides the impact of EP2 activation on cell proliferation in cervical squamous intraepithelial lesions the immunosuppressive effect of EP2 seems of interest, as only HPV infections which are not cleared by the immune system can cause SILs and cervical cancer (Westrich et al. 2017).</p><p>HPV infections have to evade the host immune defence to persist (Schiffman et al. 2011). Incidence of HPV infections and HPV associated cancer is increased in patients with natural killer cell (NK) deficiencies (Orange 2013). Moreover, a strong cytotoxic T cell (CTL) response correlates with the regression of SILs (Woo et al. 2008). PGE2 contributes to an acute local inflammation. However, its prolonged immune response can shift cytotoxic T helper cell 1 (Th1), CTL and NK cell mediated type 1 immunity towards a Th2, Th17 and a regulatory T cell mediated immunity (Walker and Rotondo 2004). Thereby PGE2 prevents damage of lung or reproductive tissue (Huang et al. 2010; Vancheri et al. 2004). Although the limitation of type 1 immunity is pivotal for host self-preservation, it contributes to the establishment of infections with intracellular organisms and cancer, as they both depend on immunosuppression (Kalinski 2012).</p><p>den Boon et al. analysed the changes in gene expression patterns from HPV infected cervical tissue to cervical cancer. The study displayed that in early lesions, mostly genes functioning in DNA replication and cell division were upregulated. In transition from CIN3 to cancer the expression of genes serving the mitochondrial electron chain is reduced (den Boon et al. 2015). This suggests a switch from oxidative phosphorylation toward anaerobic glycolysis, and is known as the "Warburg effect" (Hsu and Sabatini 2008). As other DNA viruses, HPV sustain hypoxia inducible factor 1 alpha (HIF1α), possibly also endorsing the Warburg effect (Mazzon et al. 2013; Stover 2009). PGE2 also takes part in the induction of HIF1α (Jung et al. 2003).</p><p>Grabosch et al. revealed in a systemic review that non-steroidal anti-inflammatory drugs (NSAIDs) and selective COX2 inhibitors (celecoxib, rofecoxib) are not effective in the treatment of CIN (Grabosch et al. 2018). Other structures within the COX downstream signalling pathway like EP receptors might serve as alternative drug targets (Ganesh et al. 2018). Apart from that, levels of EP receptors such as EP2 and EP3 (Hester et al. 2019) might serve as potential prognostic biomarkers for patients with CIN2 lesions. In particular women in child bearing age, who might suffer from pregnancy complications after conization could benefit from additional prognostic information (Kühn et al. 2015).</p><!><p>Supplementary file1 (DOCX 12 kb)</p><p>Publisher's Note</p><p>Springer Nature remains neutral with regard to jurisdictional claims in published maps and institutional affiliations.</p><!><p>SM received research support, advisory board, honoraria and travel expenses from AstraZeneca, Clovis, Me- dac, MSD, PharmaMar, Roche, Sensor Kinesis, Tesaro and Teva. All other authors declare that they have no conflict of interest. Open Access funding provided by Projekt DEAL.</p><!><p>The datasets generated and analysed during the current study are available from the corresponding author on reasonable request.</p><!><p>AH has received research Grants from the "Walter Schulz Stiftung" and a speaker and advisory board honorarium from Roche, Germany. TMK is employed at Roche at the time of manuscript submission.</p><!><p>All procedures performed in studies involving human participants were in accordance with the ethical standards of the institutional research committee and with the 1964 Helsinki declaration and its later amendments or comparable ethical standards. The current study was approved by the Ethics Committee of the Ludwig-Maximilians-University, Munich, Germany (167-14).</p><!><p>Informed consent was obtained from all individual participants included in the study.</p>

PubMed Open Access

Identification of inhibitors of Plasmodium falciparum phosphoethanolamine methyltransferase using an enzyme-coupled transmethylation assay

BackgroundThe phosphoethanolamine methyltransferase, PfPMT, of the human malaria parasite Plasmodium falciparum, a member of a newly identified family of phosphoethanolamine methyltransferases (PMT) found solely in some protozoa, nematodes, frogs, and plants, is involved in the synthesis of the major membrane phospholipid, phosphatidylcholine. PMT enzymes catalyze a three-step S-adenosylmethionine-dependent methylation of the nitrogen atom of phosphoethanolamine to form phosphocholine. In P. falciparum, this activity is a limiting step in the pathway of synthesis of phosphatidylcholine from serine and plays an important role in the development, replication and survival of the parasite within human red blood cells.ResultsWe have employed an enzyme-coupled methylation assay to screen for potential inhibitors of PfPMT. In addition to hexadecyltrimethylammonium, previously known to inhibit PfPMT, two compounds dodecyltrimethylammonium and amodiaquine were also found to inhibit PfPMT activity in vitro. Interestingly, PfPMT activity was not inhibited by the amodiaquine analog, chloroquine, or other aminoquinolines, amino alcohols, or histamine methyltransferase inhibitors. Using yeast as a surrogate system we found that unlike wild-type cells, yeast mutants that rely on PfPMT for survival were sensitive to amodiaquine, and their phosphatidylcholine biosynthesis was inhibited by this compound. Furthermore NMR titration studies to characterize the interaction between amoidaquine and PfPMT demonstrated a specific and concentration dependent binding of the compound to the enzyme.ConclusionThe identification of amodiaquine as an inhibitor of PfPMT in vitro and in yeast, and the biophysical evidence for the specific interaction of the compound with the enzyme will set the stage for the development of analogs of this drug that specifically inhibit this enzyme and possibly other PMTs.

identification_of_inhibitors_of_plasmodium_falciparum_phosphoethanolamine_methyltransferase_using_an

Background<!>Materials<!>Expression and purification of recombinant PfPMT and BsAda<!>PfPMT enzyme-coupled spectrophotometric assay<!>Inhibition studies<!>Yeast growth assays<!>Yeast phospholipid analysis<!>NMR experiments<!>Coupling PfPMT activity to that of SAH nucleosidase and adenine deaminase<!><!>Application of the PfPMT spectrophotometric assay for screening inhibitors of the enzyme<!><!>Inhibition of PfPMT activity by amodiaquine in vitro and in yeast<!><!>Inhibition of PfPMT activity by amodiaquine in vitro and in yeast<!><!>Structural analysis of the interaction between PfPMT and amodiaquine<!><!>Structural analysis of the interaction between PfPMT and amodiaquine<!>Discussion<!>Conclusions<!>Abbreviations<!>Competing interests<!>Authors' contributions<!>Additional file 1<!><!>Additional file 2<!><!>Acknowledgements

<p>Malaria is one of the most important parasitic diseases worldwide responsible for over 200 million clinical cases with an estimated 1 million deaths annually [1]. Malaria is caused by intraerythrocytic protozoan parasites of the genus Plasmodium. Of the five species infective to humans, Plasmodium falciparum is responsible for the highest level of mortality and morbidity [1]. The lack of an effective vaccine and the rapid rise of resistance to the most potent and affordable antimalarial drugs creates an urgent need for new therapies to prevent malaria pathogenesis and transmission. Novel strategies are now necessary to limit the appearance and spread of drug resistant malaria strains. One such a strategy involves targeting metabolic pathways proven to play an essential function in the parasite's infection and transmission. Recent studies indicate that the metabolic pathways for the synthesis of the major P. falciparum phospholipids are excellent targets for the development of lipid-based antimalarial therapies [2-4].</p><p>Studies in P. falciparum demonstrated that the synthesis of phosphatidylcholine during the intraerythrocytic life cycle of the parasite occurs via two pathways, the serine-decarboxylase phosphoethanolemine methyltransferase (SDPM) pathway and the CDP-choline pathway [2-4]. The SDPM pathway uses serine either transported from human serum or resulting from degradation of host hemoglobin as a starting precursor. Serine is first decarboxylated by a parasite serine decarboxylase to form ethanolamine. Ethanolamine is next phosphorylated by a parasite ethanolamine kinase to form phosphoethanolamine (P-EA). A parasite S-adenosylmethionine (SAM)-dependent methyltransferase, PfPMT, catalyzes a three-step methylation of P-EA to form phosphocholine [3,5,6], which is converted into phosphatidylcholine (PtdCho) via the activity of two parasite enzymes PfCCT and PfCEPT. The CDP-choline pathway uses choline transported from the host as a precursor. Choline is phosphorylated by a parasite choline kinase PfCK to phosphocholine, which is subsequently modified by PfCCT to CDP-choline and by PfCEPT to PtdCho.</p><p>The 266 amino acid PfPMT is a member of a new class of SAM-dependent methyltransferases that acts on P-EA [3]. Homologs of this enzyme are found in plants, nematodes, frogs, fish and other protozoa but not in mammals [7-11]. While phosphoethanolamine methyltransferases (PEAMT) share significant homology in their primary structure, the organization of their catalytic domains differ. The plant PEAMTs have two tandem catalytic domains, with the N-terminal domain catalyzing the methylation of P-EA to monomethylphosphoethanolamine and the C-terminal domain acting in the last two methylation reactions to form phosphocholine [7,9,10]. C. elegans expresses two PEAMT enzymes each containing only one methyltransferase domain located at either the N-terminus of the protein, in the case of Pmt1, or at the C-terminus of the protein in Pmt2 [8,11]. Pmt1 catalyzes only the first methylation reaction, whereas Pmt2 catalyzes the last two methylation reactions. The Plasmodium PfPMT is only half the size of the plant and nematode proteins, possesses a single catalytic domain, and catalyzes all three methylation steps [3]. No crystal or solution structures of these enzymes are available.</p><p>Confocal and immunoelectron microscopy studies have shown that PfPMT is expressed in the Golgi apparatus of the parasite [12]. Biochemical and genetic analyses using yeast as a surrogate system allowed identification of residues in this enzyme that play a critical role in PfPMT catalysis and substrate binding [6]. The finding that PfPMT has no homologs in mammalian databases suggested that this protein could be an ideal target for the development of novel antimalarial inhibitors targeting lipid metabolism. Interestingly, biochemical studies revealed that PfPMT activity was inhibited by S-adenosyl homocysteine (SAH) as well as by phosphocholine and its analog and anticancer drug, hexadecylphosphocholine (Miltefosine) [3]. When tested against P. falciparum, hexadecylphosphocholine was found to inhibit the growth of the parasite within human erythrocytes with 50% inhibitory concentrations in the low micromolar range [3]. Genetic studies in P. falciparum demonstrated that PfPMT plays an important role during the intraerythrocytic life cycle of the parasite [13]. Parasites lacking PfPMT display severe alterations in development, replication and survival within human red blood cells [13]. These defects were only partially complemented by choline supplementation.</p><p>Here we provide biochemical and biophysical data indicating that PfPMT activity is inhibited by amodiaquine (AQ). Using yeast as a surrogate system we show that AQ inhibits PfPMT activity and blocks PtdCho biosynthesis.</p><!><p>S-adenosylmethionine (SAM), S-adenosylhomocysteine (SAH), Phosphoethanolamine (P-Etn), Cholorquine (CQ), Amodiaquine (AQ), Hexadecyltrimethylammonium bromide (HDTA), Dodecyltrimethylammonium bromide (DDTA), 2,3-Dichloro- a- methylbenzylamine (DCMB), Chlorpromazine, and Diphenhydramine were purchased from Sigma. Miltefosine (HePC) was purchased from Cayman chemicals. SKF91488 and Tacrine were purchased from Tocris Bioscience. Recombinant SAH nucleosidase (SAHN) was purified from an E. coli strain generously given by K. Cornell (VAMC, Portland, OR, USA), as previously described [14].</p><!><p>P. falciparum PfPMT cDNA cloned in the expression vector pET-15b and expressed in E. coli BL21-CodonPlus strain was purified as described earlier [3,6]. Expression strains were grown at 37°C in LB medium containing 100 μg/ml ampicillin and 35 μg/ml chloramphenicol. A 1 L culture of E. coli was grown until an A600 ~ 0.6, and PfPMT expression was induced by addition of 1 mM isopropyl-β-D-thiogalactopyranosidase (IPTG). The cells were harvested 3 h after induction by centrifugation, and resuspended in 25 ml lysis buffer (50 mM NaH2PO4, 300 mM NaCl, 10 mM imidazole, pH 8.0) containing a protease inhibitor cocktail (complete Mini, EDTA-free, Roche Diagnostics). The cells were lysed by sonication using 10 s discontinuous cycles for three minutes on ice, and the cell debris and unbroken cells were removed by centrifugation at 10,000 rpm at 4°C for 30 min. The supernatant was directly applied to a 5 ml Ni-NTA (Qiagen) column pre-equilibrated in lysis buffer. The column was subsequently washed with 40 ml of buffer (50 mM NaH2PO4, 300 mM NaCl, 20 mM imidazole, pH 8.0) and the protein was eluted from the column in a single 15 ml fraction of buffer (50 mM NaH2PO4, 300 mM NaCl, 250 mM imidazole, pH 8.0). The protein was incubated at 4°C for 1 h in the presence of 2 mM Na2EDTA, followed by dialysis and concentration against HEPES (N-(2-hydroxyethyl)-piperazine-N'-2-ethanesulfonic acid) assay buffer (100 mM HEPES-KOH, pH 7.5) using Amicon Ultra centrifugal filter devices. The purity of the recombinant enzyme was examined by SDS-PAGE, and the protein concentration was measured by the method of Bradford using bovine serum albumin as a standard.</p><p>The gene encoding Bacillus subtilis adenine deaminase (BsAda) was amplified from genomic DNA by PCR and ligated into a pET15b plasmid vector between XhoI and BamHI sites to yield an N-terminal 6His-epitope-tagged encoding construct. The resulting plasmid was then transformed into E. coli BL21-CodonPlus strain for enzyme expression. The purification protocol for BsAda is similar to PfPMT purification except that the 1 hr incubation step with 2 mM Na2EDTA was omitted.</p><!><p>Assays were performed in 96-well UV-transparent plates (acrylic, non-sterile, Costar) at 37°C. Manganese sulfate (MnSO4) was added to a final concentration of 1 mM [15]. The assay mixture in 1X HEPES assay buffer (100 mM HEPES-KOH, pH 7.5) contained 200 μM SAM, 200 μM P-EA, 1 mM MnSO4, 0.5 μM BsAda, 4.72 μM SAHN and 2.5 μM PfPMT in a total volume of 200 μl. A mixture without PfPMT was pre-incubated at 37°C for 10 min and the reaction was initiated with the addition of PfPMT. Absorbance at 265 nm was continuously recorded with a UV, visible plate-reader (Synergy HT Multi-Mode Microplate Reader, Biotek) under kinetic mode. A negative control containing all the reaction components except for PfPMT was conducted on the same plate as the experimental reactions. The absorbance value of the control reaction was subtracted from the experimental absorbance values to eliminate the background signal.</p><!><p>All inhibitor containing stock solutions were prepared in H2O, except for HDTA, which was prepared in DMSO. The final concentration of DMSO in the assays did not exceed 10%. Control experiments containing reaction components and inhibitors were performed in the absence of PfPMT to quantify the intrinsic absorbance of inhibitors at 265 nm. These inhibitors were added before the pre-incubation step. The absorbance at 265 nm for a PfPMT assay without inhibitor was assigned as 100% PfPMT activity. The absorbance values obtained for the various inhibitor concentrations were then converted into a PfPMT activity percentage and plotted against inhibitor concentration. Control experiments were performed to ensure that the inhibitors did not affect the coupling enzymes. To measure coupling enzyme sensitivity to inhibitors, 100 μM SAH was used in place of SAM and P-EA in a 200 μl reaction in the presence of the inhibitor at the highest concentration used in PfPMT assay. For Ki determination enzymatic assays were performed with increasing concentrations of P-EA (50, 75, 100, 125, 150, 175, and 200 μM), 200 μM SAM, 1 mM MnSO4, 0.5 μM BsAda, 4.72 μM SAHN and 2.5 μM PfPMT in a HEPES assay buffer, along with the inhibitor AQ (0, 2, 4, 10 μM). A reciprocal plot was generated from the absorbance readings at 265 nm versus the P-EA concentrations for each concentration of AQ. The fitted line was extrapolated through the X axis to determine the estimated Ki.</p><!><p>BY4741-pYes2.1 (wild-type) and mutant pem1Δpem2Δ-pYes2.1-PfPMT yeast strains were pre-grown overnight in uracil dropout synthetic galactose (4%) (SG-ura) medium supplemented with 10 μM ethanolamine. The next day, cells were harvested by centrifugation, washed twice by resuspension in water and diluted to an A600 = 0.005 in fresh SG-ura medium supplemented with 100 μM ethanolamine, in the absence or presence of 10, 50, or 100 μM AQ. In a second set of experiments BY4741-pYes2.1, pem1Δpem2Δ-pYes2.1 and pem1Δpem2Δ-pYes2.1-PfPMT yeast strains were pre-grown overnight in SG-ura medium supplemented with 2 mM ethanolamine. The next day, cells were harvested by centrifugation, washed by resuspension in water and diluted to an A600 = 0.005 in fresh SG-ura medium lacking or supplemented with 1 mM choline in the presence of 0 or 200 μM AQ. In both sets of experiments, cells were grown at 30°C and monitored by measuring the A600.</p><!><p>pem1Δpem2Δ-pYes2.1 and pem1Δpem2Δ-pYes2.1-PfPMT yeast strains were pre-grown in SG-ura medium supplemented with 2 mM ethanolamine and 2 mM choline for 24 hours. The cells were harvested by centrifugation, washed twice by resuspension in water and diluted to an A600 = 0.03 in the SG-ura medium supplemented with 2 mM ethanolamine and grown overnight to an A600 ~ 1.5. The cells were next harvested and the lipids extracted for two dimensional thin layer chromatography using previously described methods [5,6,16]. Lipids were stained with iodine vapor and excised from the plate for quantification by measuring phosphorus [17]. The results are shown as the percentage of total phosphorus in each phospholipid fraction.</p><!><p>For NMR analysis, E. coli cells expressing 6his-epitope-tagged PfPMT were grown in minimal medium (M9) containing 15 N-ammonium chloride. The purification of the recombinant enzyme followed the same procedure described above except for the addition of a second step of purification through a Superdex G75 gel filtration column in a buffer containing 50 mM HEPES, 50 mM NaCl, 5 mM DTT pH 6.9. The sample was concentrated and D2O was added to a final concentration of 7% (v/v). Amodiaquine stock solution was prepared in ddH2O due to its low solubility. Serial dilutions were prepared and 5 μl of the diluted samples were added to the NMR sample containing 310 μM of purified PfPMT. The AQ concentration was determined spectrophotometrically using extinction coefficients ε238 = 27 × 103M-1cm-1, ε251 = 23 × 103M-1cm-1 and ε341 = 19 × 103M-1cm-1 [18]. The PfPMT concentration was determined using the Edelhoch method [19-21]. Standard 1H-15N HSQC experiments were performed on 15N-labeled PfPMT using a Varian Inova spectrometer operating at a 1H frequency of 600 MHz. Data were processed using nmrPipe [22] and analyzed using Sparky [23] or the Rowland NMR Toolkit http://rnmrtk.uchc.edu.</p><!><p>To identify new inhibitors of PfPMT we adapted an assay that couples the SAM-dependent transmethylation of P-EA by PfPMT to the activity of two enzymes SAH nucleosidase (SAHN) and adenine deaminase (BsAda) (Fig. 1). PfPMT activity results in the production of SAH and phosphocholine. The second enzyme SAHN catalyzes the conversion of SAH into adenine and S-ribosylhomocysteine. Adenine is converted into hypoxanthine by the third enzyme, BsAda, thus producing a decrease in absorbance at 265 nm (Fig. 2A). As shown in Fig. 2A, addition of PfPMT resulted in a time-dependent decrease in absorbance at 265 nm over a 30 min period, whereas no change in absorbance occurred when the enzyme was omitted. To demonstrate that the measured rate of the coupled reactions was determined solely by PfPMT, the enzyme activity was investigated in the presence of a fixed PfPMT concentration but with varying concentrations of SAHN or BsAda. As shown in Fig. 2B, changing the concentrations of the coupling enzymes had little or no effect on the overall rate of the reactions. The effect of increasing concentrations of the substrate P-EA and co-substrate SAM on PfPMT activity was also measured. PfPMT activity increased proportionally to the concentration of these substrates, with optimal activity measured when the substrate and co-substrate concentrations were at 200 μM (Fig. 2C). Higher concentrations of P-EA only slightly affected the activity of the enzyme, whereas high SAM concentrations inhibited the enzyme (data not shown). Under optimal conditions, we also examined the effect of PfPMT concentration on the rate of the reaction. As shown in Fig. 2D, increasing the concentration of PfPMT resulted in a proportional increase in activity reflected by a decrease in absorbance over time. Above this range, the enzyme activity started to deviate from linearity (data not shown). We also tested the effect of different buffers at pH 7, 7.5 and 8 on PfPMT activity. These factors were found to have little effect on PfPMT activity over time (Fig. 2E).</p><!><p>Schematic description of the enzyme-coupled assay for measuring PfPMT activity. PfPMT catalyzes the conversion of 1 molecule of P-EA and 3 molecules of SAM to form 1 molecule of phosphocholine and 3 molecules of SAH. SAH is hydrolyzed to adenine and S-ribosylhomocysteine via SAHN nucleosidase. The deamination of adenine into hypoxanthine by adenine deaminase is associated with a decrease in absorbance at 265 nm that can be monitored continuously using UV plate reader.</p><p>Spectrophotometric analysis of phosphoethanolamine methyltransferase activity. (A) PfPMT-catalyzed methylation of P-EA. Reaction mixtures contained 200 μM SAM, 200 μM P-EA, 1000 μM MnSO4, 0.5 μM BsAda, 4.72 μM SAHN and 0 μM (red) or 2.5 μM (blue) of purified PfPMT enzyme in 100 mM HEPES assay buffer pH 7.5. The decrease in absorbance was monitored at 265 nm. The reactions components were kept the same as stated above in panels B-E, except when noted. (B) Dependence of the rate of PfPMT-catalyzed reaction on different concentrations of the coupling enzymes. Reactions contained BsAda 0.25 μM and SAHN 2.13 μM (orange), 0.5 μM BsAda, 4.72 μM SAHN (blue) and without PfPMT (red), and BsAda 1 μM and SAHN 8.5 μM (green). (C) Dependence of the rate of PfPMT-catalyzed reaction on different concentrations of SAM and P-EA. Reactions contained P-EA 50 μM and SAM 150 μM (orange), P-EA 100 μM and SAM 100 μM (black), p-Etn 150 μM and SAM 150 μM (green), P-EA 200 μM and SAM 200 μM (blue), and without PfPMT (red). (D) Effect of PfPMT concentration on its activity. Reactions contained 0 μM PfPMT (red), 312.5 nM PfPMT (orange), 625 nM PfPMT (green), 1.25 μM PfPMT (black), and 2.5 μM (blue). (E) Effect of the pH and buffer composition on PfPMT activity. Reaction mixtures contained 200 μM SAM, 200 μM P-EA, 1000 μM MnSO4, 0.5 μM BsAda, 4.72 μM SAHN and 0 μM (red) or 2.5 μM (black), of purified PfPMT enzyme in 100 mM HEPES assay buffer pH 7.5. Reactions also contained 100 mM HEPES assay buffer pH 7 (green), and pH 8 (cyan), as well as 100 mM Tris-HCl assay buffer pH 7 (orange), pH 7.5 (purple), and pH 8 (yellow). Results are representative of three independent experiments.</p><!><p>To assess the applicability of the enzyme-coupled assay for screening inhibitors of PfPMT, the effects of some known or potential inhibitors of this enzyme were examined. As most drug screens are performed in the presence of DMSO, we first tested the effect of this agent on PfPMT activity. DMSO up to 10% had no effect on PfPMT activity (data not shown). The hexadecylphosphocholine analog, miltefosine (HePC), was previously reported to inhibit PfPMT activity in vitro and P. falciparum proliferation inside red blood cells with similar efficacy [3]. Using the enzyme-coupled spectrophotometric assay, HePC was found to inhibit PfPMT activity with 100 μM of HePC reducing PfPMT activity by ~60% and 150 μM of the compound resulting in a complete loss of activity (Fig. 3A). As a control, an assay in which PfPMT, SAM and P-EA were omitted and replaced by their product SAH was performed in the absence or presence of 200 μM of HePC. Under these conditions HePC had no effect on the production of hypoxanthine over time, suggesting that neither SAHN nor BsAda was inhibited by this compound. Two compounds that are structurally related to HePC, hexadecyltrimethylammonium (HDTA) and dodecyltrimethylammonium (DDTA) that are known to have potent antimalarial activity in vitro and in vivo [24-26] were also tested to determine their inhibitory activity against PfPMT. HDTA was found to have an inhibitory activity similar to that of HePC with ~50% inhibition of PfPMT activity observed at 100 μM of the compound (Fig. 3B). DDTA, however, had no effect against the enzyme when used at concentrations of up to 100 μM and only modest inhibitory activity at 500 μM, inhibiting 25% of PfPMT activity at this concentration (Fig. 3C).</p><!><p>Inhibition of PfPMT by quaternary amines. Effect of increasing concentrations of hexadecylphosphocholine (HePC) (A), hexadecyltrimethylammonium bromide (HDTA) (B) and dodecyltrimethylammonium bromide (DDTA) on PfPMT activity. The assay was performed as described in Experimental Procedures. The data are the means +/- S.D. for three independent experiments. Statistically significant data with a P < 0.05 is indicated with an asterisk.</p><!><p>In order to identify possible lead inhibitors of PfPMT, we examined the effect of drugs known to inhibit other SAM-dependent methyltransferases. The 2,3-dichloro-methylbenzyl-amine (DCMB), a known inhibitor of the human phenylethanolamine-N-methyltransferase enzyme [27], which catalyses the terminal step in epinephrine biosynthesis had no effect on PfPMT activity even at 500 μM (Fig. 4A). Conversely, AQ, a known inhibitor of histamine methyltransferases (HNMT) and a potent antimalarial drug [28], was found to be a good inhibitor of the enzyme with 60% inhibition at 5 μM (Fig. 4B). None of these concentrations of AQ inhibited the coupling enzymes used in the PfPMT assay (data not shown). The finding that AQ inhibits PfPMT led us to investigate the effect of four other HNMT inhibitors, SKF91488, Tacrine, Chlorpromazine and Diphenhydramine. None of these compounds had a significant effect on PfPMT activity (Fig. 5A). Furthermore, none of the compounds inhibited the growth of P. falciparum within human erythrocytes at a concentration as high as 1 μM (data not shown). The specificity of inhibition of PfPMT by AQ was further demonstrated using its analog and antimalarial drug, chloroquine (CQ). At 200 μM concentration, CQ had no effect on PfPMT activity (Fig. 5B). Other aminoquinolines and amino alcohols, quinacrine, quinidine and quinine, known for their potent antimalarial activity, had no effect on PfPMT activity at concentrations up to 200 μM (Fig. 5B), suggesting that AQ inhibition of PfPMT is specific. In order to examine AQ inhibition of PfPMT in vivo we used yeast as a surrogate system. Yeast is a particularly attractive system for this analysis because, unlike P. falciparum, it is not sensitive to AQ and lacks phosphoethanolamine methyltransferases. We have previously shown that a codon-optimized P. falciparum PfPMT gene complements the choline auxotrophy of the yeast pem1Δpem2Δ mutant, which lacks the two phospholipid methyltransferases, Pem1p and Pem2p, and thus is unable to synthesize PtdCho from PtdEtn [29,30]. In the complemented strain, PfPMT restores PtdCho by providing phosphocholine following P-EA transmethylation [5,6]. Examination of the growth of wild-type yeast cells in media lacking or containing choline, and supplemented with either 100 μM or 2 μM ethanolamine (Fig. 6A and 6C), and in the absence or presence of AQ demonstrated no effect of this compound at concentrations up to 200 μM. Unlike pem1Δpem2Δ, which did not grow on medium containing ethanolamine but lacking choline (Fig. 6D, curve 5 & 6), pem1Δpem2Δ strains complemented with PfPMT grew on media lacking choline and their growth rate was significantly influenced by the availability of ethanolamine with the highest cell density reached in the presence of 2 μM ethanolamine (Fig 6E, curve 5 & 6). Interestingly, the growth of pem1Δpem2Δ+PfPMT was dramatically inhibited when AQ was added to the culture medium (Fig. 6B). AQ inhibited the growth of pem1Δpem2Δ+PfPMT strains in a concentration dependent manner with 100 μM drug reducing growth by 76% in medium containing 100 μM ethanolamine after 60 h (Fig. 6B). These results thus demonstrate a direct inhibition of PfPMT by AQ in vivo. Addition of choline to the culture medium of pem1Δpem2Δ-PfPMT cells resulted in complete resistance of these cells to AQ (Fig. 6E, curve 7 & 8), suggesting that the inhibition of growth was dependent on the essential function of PfPMT for survival in the absence of exogenous choline. As a control, the pem1Δpem2Δ mutant harboring an empty vector did not grow in the absence of choline and was resistant to AQ when choline was added (Fig. 6D).</p><!><p>Amodiaquine inhibits purified PfPMT activity. Effect of increasing concentrations of DCMB (A) and amodiaquine (AQ) (B) on PfPMT activity. The assay was performed as described in Methods. The data are the means +/- S.D. for three independent experiments. Statistically significant data with a P < 0.01 is indicated with an asterisk.</p><p>Effect of HNMT inhibitors and antimalarial aminoquinolines and amino alcohols on PfPMT activity. (A) Effect of the HNMT inhibitors SKF91488 (SKF), tacrine (Tac), diphenhydramine (Dip) and chlorpromazine (Chl) on PfPMT activity. (B) Effect of chloroquine (CQ), quinacrine (QC), quinidine (QD) and quinine (QN) on PfPMT activity. The data are the means +/- S.D. for three independent experiments.</p><p>Amodiaquine inhibits PfPMT function in yeast. Growth curves of wild-type (BY4741-pYes2.1) (A) and pem1Δpem2Δ-PfPMT (B) strains grown in minimal medium containing 4% galactose and 100 μM ethanolamine in the presence of 0 μM (1), 10 μM (2), 50 μM (3), or 100 μM (4) AQ. (C-E) Growth curves of wild-type (BY4741-pYes2.1) (C), pem1Δpem2Δ-pYes2.1 (D) and pem1Δpem2Δ-PfPMT (E) yeast strains grown in minimal medium containing 4% galactose and 2 mM ethanolamine in the presence of 0 μM AQ (5), 200 μM AQ (6), 200 μM AQ and 1 mM choline (7), or 1 mM choline (8).</p><!><p>To demonstrate that the inhibition of the growth of pem1Δpem2Δ+PfPMT by AQ was due to the inhibition of the synthesis of PtdCho from ethanolamine, the synthesis of the major phospholipids PtdCho and PtdEtn of yeast membranes was examined in the absence or presence of AQ. Consistent with previous findings [5,6], pem1Δpem2Δ cells harboring an empty vector produced ~3% of total phospholipid as PtdCho after 5 to 6 generations of growth in the choline deficient medium, whereas those expressing PfPMT produced ~18% PtdCho (Fig. 7). Addition of AQ to pem1Δpem2Δ cells expressing PfPMT resulted in a concentration dependent decrease in PtdCho levels with ~15% produced at 10 μM and ~5% produced at 200 μM AQ (Figs. 7A and 7B). These findings further demonstrate the specific inhibition of PtdCho biosynthesis by this compound. The depletion of PtdCho effected by genetic manipulation or AQ treatment was partially compensated by increased levels of PtdIns (Fig. 7).</p><!><p>Amodiaquine reduced PfPMT-dependent PtdCho levels in yeast. (A) Phospholipid analysis of pem1Δpem2Δ-pYes2.1 and pem1Δpem2Δ-PfPMT strains grown in minimal medium containing 4% galactose and 2 mM ethanolamine. The lipids were extracted, separated by 2-D TLC and stained with iodine vapor. (B) Each lipid was recovered from the TLC plate and quantified by measuring phosphorous. The graph is the percentage of total lipid phosphorous in each lipid fraction. PtdCho-phosphatidylcholine; PtdEtn- phosphatidylethanolamine; PtdSer- phosphatidylserine; PtdIns- phosphatidylinositol. The data are represented as the means +/- S.D. of three independent experiments.</p><!><p>Co-crystallization studies of human HNMT with AQ indicated that two molecules of AQ were bound per HNMT molecule [31]. One occupies the active site pocket (Site 1; Figs. 8A and 8B) and was proposed to competitively inhibit histamine binding, and the other occupies a deep pocket representing an uncompetitive component (Site 2; Figs. 8A and 8B) [31]. To characterize the nature of the inhibition of PfPMT by AQ and to calculate the inhibition constant, PfPMT activity was determined in the presence of increasing concentrations of P-EA and increasing concentrations of the inhibitor. These studies, however, did not allow distinction between competitive and noncompetitive inhibition. To further explore the interaction of AQ with PfPMT, we performed NMR studies of the enzyme with varying concentrations of AQ. The 1H-15N HSQC spectra revealed that of the 266 residues of PfPMT only about 20 residues showed a substantial change in chemical shift as a function of increasing concentrations of AQ (Fig. 9A and Additional file 1, Fig. S1). This demonstrates direct and site-specific binding of AQ to PfPMT. We were able to tentatively assign 9 of the 20 resonances exhibiting shifts upon AQ binding. Mapped onto corresponding residues in the HNMT structure (via ClustalW sequence alignment), 8 of the 9 residues were found to be proximal to the two AQ binding sites in HNMT. The 9th residue is proximal to residues that are in contact with an AQ binding site. Nonspecific changes involving many residues that are observed at high AQ concentrations appear to reflect biophysical changes in solution properties due to a concentrated co-solute.</p><!><p>Prediction of amodiaquine-interacting residues on PfPMT using HNMT structure. (A) Molecular surface representation of HNMT. Red solvent-accessible surfaces identify invariant residues between HNMT and PfPMT. Phe190 is colored in orange. Two AQ molecules are shown as stick with the carbon atoms depicted in cyan, nitrogen in blue, oxygen in red, and chlorine in green. (B) The Cα trace of HNMT in a similar orientation as (A). The AQ-interacting residues are shown as stick forms. The HNMT residues, corresponding to the nine residues of PfPMT whose chemical shifts are perturbed in response to AQ binding, are shown in green. Figures were prepared by using Pymol [56].</p><p>NMR analysis of PfPMT-amodiaquine interaction. The glycine region of the 1H-15N HSQC spectra of PfPMT (0.3 mM), titrated with AQ (A) or CQ (B). The different colors indicate the inhibitor concentrations as follows: red- no inhibitor, orange- 1:4, yellow- 1:2, green- 1:1, blue- 2:1, violet- 3:1, black- 10:1. (C and D) Inhibitor titration plots for Gly32 and Gly68 (derived from A and B) showing the difference in binding between AQ and CQ. The overlay of the full HSQC spectrum in the absence or presence of AQ is shown in Additional File 1, Fig. S1.</p><!><p>We also performed NMR titrations using CQ (Fig. 9B). While the effects of CQ are similarly restricted to about 20 residues, indicating site-specific binding, the affinity of CQ for PfPMT is much lower than that of AQ. Complete characterization of CQ binding by NMR is hampered by nonspecific effects that begin to dominate at concentrations above 3 mM (molar ratio of 10 in Fig. 9). Detailed comparison of the concentration dependence of the chemical shifts for two residues, tentatively assigned to Gly32 and Gly68, revealed the difference in the affinities of AQ and CQ for PfPMT. For AQ, the chemical shift changes approached an asymptotic value approximately exponentially, consistent with all binding sites occupied around a 3:1 molar ratio (Figs. 9C and 9D). In contrast, the chemical shifts of Gly32 and Gly68 changed only linearly up to a molar ratio of 10:1. The fact that the same residues are perturbed in response to both AQ and CQ suggest that they interact with the same site on PfPMT, but the different concentration dependence suggests that the affinity of CQ for PfPMT is at least an order of magnitude lower than that of AQ.</p><!><p>Methyltransferases represent a large group of enzymes divided into 160 classes (EC 2.1.1.1 to EC2.1.1.160). These enzymes catalyze the methylation of a large number of different substrates including DNA, RNAs, phospholipids, fatty acids, inositol, phosphoethanolamine, phenylethanolamine, tocopherol, catechol and histamine. Although some methyl donor molecules such as S-methylmethionine, betaine and folate can be used, the majority of biologically active methyltransferase enzymes utilize SAM as a methyl donor [32]. SAM-methyltransferases are recognized by the primary structure and topology of the core fold containing the SAM binding domain [32]. These enzymes play important functions in various biological systems and have been linked to different human diseases such as Alzheimer's disease, attention deficit disorder and pre-eclampsia [33-35]. Phosphoethanolamine methyltransferases (PMTs) represent a recently identified class of enzymes found mostly in plants, nematodes, frogs, fish and some species of protozoa. Their absence in humans and the finding that these enzymes play important physiological functions make them good targets for the development of novel inhibitors to treat worm and protozoan parasitic diseases [3,8,11]. The use of radiolabeled SAM to assay PMTs in vitro makes it difficult to perform large-scale screenings of available chemical libraries to search for inhibitors of these enzymes. Here we have modified a continuous coupled assay previously described by Dorgan et al [36] for the rat arginine N-methyltransferase 1 (PRMT1) to measure PfPMT activity in vitro and demonstrated its feasibility for drug inhibition assays using HePC, which was previously shown to inhibit PfPMT with a radioactivity-based assay [3]. In this report, PfPMT activity was inhibited ~60% at a concentration of 100 μM of HePC and 150 μM of this compound resulted in a complete loss of activity. The enzyme-coupled assay was also useful in identifying HDTA as an inhibitor of PfPMT activity. HDTA was found to have an inhibitory activity similar to that of HePC with ~50% inhibition of PfPMT activity observed at 100 μM of the compound. HDTA inhibits the growth of P. falciparum in culture with an IC50 of 2.1 μM [24]. Using this assay we have also identified AQ as an inhibitor of PfPMT. In addition to its antimalarial activity, AQ is also known as a potent inhibitor of the human histamine N-methyltransferase (HNMT) [31]. Other more potent inhibitors of HNMT such as SKF91488, tacrine, diphenhydramine and chlorpromazine [37-39] had no effect on PfPMT, suggesting that the interaction between PfPMT and AQ is unique.</p><p>Amodiaquine is a structural analog of CQ. Analysis of uncomplicated falciparum malaria cases in Africa suggested that AQ is more effective in clearing parasites, including CQ resistant strains, and resulted in better clinical recovery compared to CQ [40-42]. Amodiaquine is used mostly in combination therapy with artesunate and sulfadoxine-pyrimethamine to treat malaria infections, especially those caused by CQ resistant strains [43]. Following oral administration, AQ is detected in the plasma between 30 min and 8 h, but is rapidly converted to desethylamodiaquine (AQm) with a peak plasma concentration of 181 ± 26 ng ml-1 [44]. In the liver AQ is converted to AQm by the polymorphic P450 isoform CYP2C8 [45]. Up to 96 hours after administration, AQm could be detected in the plasma, a property that made AQ an ideal compound in combination therapy [46]. In vitro both AQ and AQm inhibit parasite intraerythrocytic growth with IC50 values of 18.2 and 67.5 nM, respectively [47]. The modes of action of AQ and AQm are unknown. Field studies suggest that low levels of resistance to AQ associate with mutations in the PfCRT and PfMDR1 genes, whereas high AQ resistance involves unknown mechanisms [48]. These findings suggest that AQ might have CQ like properties in the food vacuole as well as novel activities that remain to be elucidated. Interestingly, whereas AQ inhibited PfPMT activity, CQ did not. PfPMT activity was also not affected by the antimalarials quinine, quinidine and quinacrine at concentrations as high as 200 μM. The specificity of inhibition by AQ was demonstrated in vivo by using yeast as a surrogate system. In yeast, the growth of mutants that rely on PfPMT for survival was dramatically altered in the presence of this compound. Consistent with this phenotype, the synthesis of PtdCho in these complemented cells was strongly inhibited by AQ. Surprisingly, the amplitude of inhibition of pem1Δpem2Δ+PfPMT cells by AQ was dependent on the concentration of ethanolamine in the culture medium. This is likely due to the fact that these complemented cells grow faster on medium containing high concentrations of ethanolamine as a result of increased substrate availability for PfPMT.</p><p>Recent genetic studies have shown that PfPMT plays an important function during P. falciparum intraerythrocytic development, multiplication and survival [13], and an essential role in sexual differentiation (Bobenchik et al., unpublished data). Interestingly, silencing of the A. thaliana PEAMT results in multiple morphological phenotypes, including pale-green leaves, early senescence, temperature-sensitive male sterility and increased sensitivity to salts [49]. Furthermore, characterization of an A. thaliana mutant with the T-DNA inserted at the At3g18000 locus (XIPOTL1), which encodes PEAMT revealed severe alterations in root developmental and induces cell death in root epidermal cells [50]. In C. elegans, studies aimed at silencing the PEAMT genes PMT-1 and PMT-2 demonstrated an essential function of these genes in worm, growth, development and survival [8,11]. Altogether, these genetic studies validate PEAMT enzymes as possible targets for the development of inhibitors that could be used for the treatment of specific human and veterinary protozoa and nematode related infections as well as in agriculture.</p><p>P. falciparum parasites lacking PfPMT were, however, equally sensitive to AQ as wild-type parasites in medium containing choline. This result suggests that PfPMT is not the sole target of AQ in choline-containing medium. Our finding that AQ inhibits PfPMT activity, and the fact that this compound also inhibits other eukaryotic methyltransferases, suggest that it may exert its antimalarial activity by targeting several parasite methyltransferases. Several studies have suggested that aminoquinolines may exert their antimalarial activity by interfering with biological functions other than heme detoxification in the digestive vacuole. Famin and Ginsburg reported inhibition of parasite's 6-phosphogluconate dehydrogenase by CQ [51]. Sharma and Mishra reported inhibition of parasite's tyrosine kinase by this compound [52]. Recently, CQ was reported to inhibit vesicular trafficking within the parasite [53], and to induce proteome oxidative damage [54]. More recent studies aimed at investigating the stage specificity of CQ and comparing the effect of continuous vs bolus dosing on P. falciparum strains revealed that rings (prior to the formation of the digestive vacuole) and schizont stage parasites (after hemoglobin degradation and heme detoxification activities have plateaued) were equally sensitive to this compound [55]. The bolus dosing concentrations of CQ used by Gligorijevic and colleagues are in the range of the compound's peak plasma concentration (low micromolar range) [55]. Together these studies suggest that heme detoxification, although important, is unlikely to be the only target of CQ. Although, the stage specificity and the effect of bolus dosing of AQ on P. falciparum have not been investigated, our results indicate that at its peak plasma concentration, AQ is likely to inhibit PfPMT and possibly other parasite methyltransferases. Future studies will aim to identify Plasmodium methyltransferases inhibited by AQ and to characterize the mechanism of sensitivity and resistance to AQ. From a chemistry standpoint, AQ represents an excellent lead compound for the rational design of better inhibitors of PfPMT and other parasite methyltransferases as well as plant and nematode PMTs.</p><p>Horton and coworkers have previously shown that AQ acts both as a competitive and a noncompetitive inhibitor of human HNMT [31]. The crystal structure of the HNMT-AQ complex [31] revealed that one AQ molecule binds to the histamine-binding site (Site 1), while the second one tucks into an adjacent pocket on the outer surface of the protein (Site 2) (Fig. 8A). Both quinoline rings fit into a sandwich-like structure formed by the aromatic side chains of Tyr15 and Phe19, and Phe190 and Tyr198, respectively (Fig. 8B). The side chains of Trp183 and Phe243 also contribute to the stabilization of the branched alkylamino tail of the AQ in the histamine-binding site [31]. It should be noted that the branch structure of the second AQ is disordered in the outer-surface pocket, suggesting a weaker binding interaction. The strong hydrophobic interactions between AQ and the aromatic side chains of the enzyme have been proposed to account for the affinity and specificity of the inhibitor for HNMT [31]. Sequence alignment analysis of HNMT and PfPMT shows that Tyr15, Trp183, and Phe243 are conserved in PfPMT, while Phe190 is substituted by tyrosine (Fig. 8B and Additional file 2, Fig. S2). This structural conservation is significant, as the two proteins have an overall 14% pairwise sequence identity. The majority of the other invariant residues are within the classic SAM-dependent methyltransferase fold in HNMT (Fig. 8B and Additional file 2, Fig. S2). Our inhibition studies revealed that whereas AQ inhibited PfPMT activity, CQ did not. PfPMT activity was also not affected by the aminoquinolines, quinine, quinidine and quinacrine at concentrations as high as 200 μM, or by the histamine methyltransferase inhibitors SKF91488, diphenhydramine and tacrine at concentrations as high as 100 μM. Unlike AQ, diphenhydramine and tacrine have been shown to interact only with Site 1 of HNMT. If this property is also valid in the case of PfPMT, it may account for the difference in inhibition of PfPMT activity between AQ and other aminoquinolines. The NMR studies reported here confirmed the site-specific binding of AQ to PfPMT, as less than 10% of residues in PfPMT showed a significant change in chemical shift in the presence of the compound. Nine of these residues have thus far been assigned. Six of those (corresponding to Glu28, Ser26, Ala63, Glu65, Ile66 and Met36 in HNMT) are proximal to the histamine-binding site (Site 1), while the 7th (Leu261 in HNMT) is proximal to Site 2. The 8th (Cys196 in HNMT) is located between the two sites. The last residue (Leu108 in HNMT) is further from Site 1, but is proximal to residues in contact with this site, and could be a relayed effect. The three residues (Gly68, Gly32 and Gly40 corresponding to Glu65, Glu28 and Met36 in HNMT) with the largest shift changes are closest to site 1. The binding of AQ to the free enzyme suggests that the inhibition is not uncompetitive, i.e. via binding to the enzyme-substrate complex, and therefore either competitive or noncompetitive. The NMR experiments at this stage do not definitively reveal whether AQ binds competitively to the substrate binding site. NMR studies to determine the structure of PfPMT alone and in combination with AQ are underway. They will address the question of noncompetitive vs. competitive inhibition, and should shed new light on the interaction between the enzyme and the inhibitor. However, based on our preliminary assignment of two of the nine assigned residues, Gly32 and Gly68, and the known structure of HNMT, we can postulate a structural hypothesis consistent with the NMR results for Gly32 and Gly68. Gly32 and Gly68 of PfPMT correspond to Glu28 and Glu65 of HNMT (Fig. 9). Glu28 lies at the bottom of the cleft comprising site 1 of HNMT. Glu65 also lines the surface of the cleft, closer to the enzyme surface. The similarity of the chemical shift perturbation curves (to within the experimental uncertainty of ~ 0.006 ppm) for Gly32 and Gly68 is thus consistent with binding at site 1. Despite the low sequence homology between PfPMT and HNMT, these data suggest that PfPMT and HNMT nevertheless exhibit structural homology, and are consistent with two AQ binding sites. The proximity of the two binding sites suggests that titration data might reflect coupling between sites. Indeed, we do not observe partitioning of the titration data into two distinct classes, as would be expected for independent binding sites.</p><!><p>The enzyme-coupled assay adapted for PfPMT is a simple and reliable method for measuring PfPMT activity and screening for specific inhibitors of this enzyme. The identification of AQ as an inhibitor of PfPMT may help in the future design of compounds that specifically inhibit this enzyme and possibly other PMTs.</p><!><p>AQ: amodiaquine AQm: desethylamodiaquine CQ: chloroquine PtdCho: phosphatidylcholine; PtdEtn: phosphatidylethanolamine; PtdSer: phosphatidylserine; P-EA: phosphoethanolamine; PMT: phosphoethanolamine methyltransferase; SDPM: serine decarboxylase phosphoethanolamine methyltransferase; CDP: citidylyldiphosphate; PfCCT: CDP-choline cytidylyl-transferase; PfCEPT: CDP-diacylgylcerol-choline phosphotransferase; SAM: s-adenosylmethionine; TLC: thin layer chromatography; SAH: s-adenosylhomocystein; HNMT: histamine methyltransferase; NMR: nuclear magnetic resonance; SAHN: SAH nucleosidase.</p><!><p>The authors declare that they have no competing interests.</p><!><p>AB optimized the enzyme coupled assay, performed the enzyme inhibitor studies, prepared all the figures and assisted in the writing and editing of the manuscript and the revisions. AM contributed to the initial development of the enzyme coupled assay and cloning of the BsADA used in the study. She also helped in the design of experiments and writing of the manuscript. JC and DV conducted the yeast growth assays and the yeast phospholipid analyses. IR, BH and JH contributed to the structural modeling and NMR analysis of PfPMT-amodiaquine interaction and helped in the writing and editing of the manuscript. CBM conceived, established the experimental framework of the study, analyzed the data and contributed to the writing and editing of the manuscript. All authors read and approved the final manuscript.</p><!><p>Fig. S1. Overlay of the full 1H-15N HSQC spectra of PfPMT in the absence (red) or presence of 0.06 (orange), 0.12 (yellow), 0.25 (green), 0.5 (blue) and 1 mM (purple) of AQ.</p><!><p>Click here for file</p><!><p>Fig. S2. Sequence alignment of HNMT and PfPMT. Residues that are identical, conserved, and semi-conserved are indicated by asterisk, colon, and period, respectively. The AQ-interacting residues are colored as in Fig. 8B. Phe19 and Tyr198 of HNMT and their corresponding residues in PfPMT are shown in italics and bold.</p><!><p>Click here for file</p><!><p>We thank Dr. Peter Setlow for providing Bacillus subtilis DNA. We are grateful to Dr. Ken Cornell, Boise State University, for providing the SAH nucleosidase expression vector, and to Dr. Zhaohui Sunny Zhou, Washington State University, for helpful discussions about the assay. We thank Oksana Gorbatyuk, Li Luo and William Witola for help with the expression and purification of PfPMT. This research was supported by NIH and DOD grants [AI51507], [PR033005] and BWF award [1006267] to CBM. CBM is a recipient of the Burroughs Wellcome Award, Investigators of Pathogenesis of Infectious Disease. JCH acknowledges support from NIH [RR020125].</p>

PubMed Open Access

Chirality memory of α-methylene-π-allyl iridium species

Chirality is one of the most important types of steric information in nature. In addition to central chirality, axial chirality has been catching more and more attention from scientists. However, although much attention has recently been paid to the creation of axial chirality and the chirality transfer of allenes, no study has been disclosed as to the memory of such an axial chirality. The reason is very obvious: the chiral information is stored over three carbon atoms. Here, the first example of the memory of chirality (MOC) of allenes has been recorded, which was realized via an optically active alkylidene-p-allyl iridium intermediate, leading to a highly stereoselective electrophilic allenylation with amines. Specifically, we have established the transition metal-mediated highly stereoselective 2,3-allenylation of amines by using optically active 2,3-allenyl carbonates under the catalysis of a nonchiral iridium(III) complex. This method is compatible with sterically bulky and small substituents on both amines and 2,3-allenyl carbonates and furnishes the desired optically active products with a high efficiency of chirality transfer. Further mechanistic experiments reveal that the isomerization of the optically active alkylidene-p-allyl iridium intermediate is very slow.Scheme 1 Reaction of optically active a-alkylidene-p-allylic metallic species.

chirality_memory_of_α-methylene-π-allyl_iridium_species

Introduction<!>Optimization of reaction conditions<!>Allenylation of amines<!>Memory of chirality<!>Mechanistic studies<!>Conclusions

<p>Memory of chirality (MOC) is a topic of ever-lasting interest in the storage/transmission of steric information, chiral recognition, and asymmetric syntheses. 1 In addition to enolate chemistry, 2 some examples of transition-metal catalyzed stereospecic functionalizations of optically active allylic derivatives have been recorded. 3 However, reports on the memory of chirality of a-alkylidene-p-allyl metallic species are very limited: the rapid racemization of such chiral intermediate via s-p-s-rearrangements has been reported by Tsukamoto (Scheme 1a). 4 Herein, we wish to report the observation of an excellent memory of chirality (MOC) in the highly stereoselective reaction of optically active allenylic methyl carbonates with amines affording 2,3-allenylic amines in high ee via aalkylidene-p-allyl iridium species (Scheme 1b).</p><!><p>In the beginning, due to the importance of amines, 5 we were studying the synthesis of racemic 2,3-allenyl amines. Aer trial and error, we observed some very attractive results with iridiumcatalysis for the allenylation of benzylamine 2a with racemic methyl dodeca-2,3-dienyl carbonate 1a. With [Ir(COD)Cl] 2 as the catalyst, a variety of phosphine ligands was screened. PPh 3 (Table 1, entry 1) afforded a 13% yield of 3aa, 1% of bis-allenylation product 4aa, and 85% recovery of 1a. Further screening of other phosphine ligands led to poor results (Table 1, entries 2-10). To our delight, the utilization of a known cationic Ir(III) precatalyst A developed by Hartwig and co-workers 6 yielded 26% of 3aa and 25% of 4aa with no recovery of 1a (Table 1, entry 11).</p><p>Further optimization was focused on the selectivity of 3aa/ 4aa. We rstly increased the amount of Ir pre-catalyst A to 7 mol%, but no improvement was observed (Table 2, entry 2). As expected, the loading of benzylamine greatly affected the selectivity of 3aa/4aa: the addition of 4 equiv. of benzylamine provided 41% of 3aa, 6% of 4aa, and 38% recovery of 1a (Table 2, entry 4). By running the reaction at 50 C, the yield was improved to 70% and no recovery of 1a was detected (Table 2, entry 7). The reaction at a higher temperature resulted in an erosion of yield (Table 2, entry 8).</p><p>The reactions in other ethers (Table 3, entries 1-3) and chlorinated solvents (Table 3, entries 4 and 5) led to a lower a The reaction was conducted using 1a (0.2 mmol) in 0.4 mL of THF. b Determined by 1 H NMR analysis of the crude reaction mixture using 1,3,5trimethylbenzene as an internal standard. c 7 mol% Ir pre-catalyst A was used. d The reaction was conducted using 1a (0.2 mmol) in 1.0 mL of THF.</p><p>selectivity of 3aa/4aa. Increasing the polarity of aprotic solvents displayed decreasing yields (Table 3, entries 6-8). Toluene provided a very similar yield and selectivity (Table 3, entry 9). Scaling the reaction up to 1.0 mmol furnished similar results with 3aa isolated in a 71% yield together with 9% of the bisallenylation product (Table 3, entries 10 and 11). Thus, 1a (1 equiv.), 2a (4 equiv.), and Ir pre-catalyst A (3.5 mol%) in THF at 50 C was dened as the optimized reaction conditions for further study.</p><!><p>With the optimal conditions in hand, diverse 2,3-allenyl carbonates and amines were investigated to demonstrate the scope of this reaction (Scheme 2). Terminal 2,3-allenyl carbonates showed slightly lower reactivities (products 3bb-3be) (part I). 4-Mono-substituted 2,3-allenyl carbonates were then examined (part II): R 1 may be a 1 -alkyl group, such as n-C 8 H 17 (1a) , n-C 5 H 11 (1d), n-C 6 H 13 (1t), or a 2 -alkyl group, i-Pr (1e) and Cy (1f), smoothly affording products 3aa, 3af, 3ag, 3ah, 12, 3tx, 3dj, 3dk, 3el, 3fm, and 3fn with the corresponding amines. Benzoxy (1c) and phenyl group (1u) were tolerated (products 3ci, 3uy, and 4uz). 4-Aryl-2,3-allenyl carbonates were next exposed to the optimized reaction conditions (part III): 4-phenyl-2,3-allenyl carbonate 1g furnished a similar result as compared with 1a (product 3ga). Substrates with a wide range of functional groups, such as p-Me (1h), m-OMe (1i), p-F (1j), p-Cl (1k), p-Br (1l), p-COOMe (1v), p-CF 3 (1w), and p-CN (1x), all exhibited a decent reactivity under the standard conditions (products 3hp, 3im-3ir, 3js, 3kt, 3lb, 3vf, 3wf, and 3xr). 2-Naphthyl (1m) and 3-thienyl (1o) were also accommodated to afford the target products 3mf, 3mu, 3ov, and 3oa in 61-90% yields; 3-furyl substituted allenyl carbonate 1p turned out to be less reactive and reacted with p-methoxybenzylamine 2w to yield product 3pw in 48% isolated yield at 60 C with 7 mol% Ir catalyst. 4,4-Disubstituted allenyl carbonate 1q also worked to afford product 3qf.</p><p>A wide range of amines were also tested. Acyclic amines bearing different alkyl (2e, 2o, 2q, and 2r), phenyl (12, 5, 6, and 2r), benzyl (2a, 2f, 2k, and 2w), and synthetically useful functional groups, such as the allylic (2g and 2o), propargyl (2d and 2u) and ester (2c) groups, all furnished the corresponding products 3be, 3go, 3iq, 12, 5, 6, 3ir, 3xr, 3aa, 3ga, 3vf, 3wf, 3mf, 3oa, 3pw, 3ag, 3go, 3bd, 3mu, and 3bc in moderate to quantitative yields (48-99%). Tryptamine 2j, which contains 3 potential reaction sites, was detected to be allenylated exclusively on the primary amino group as judged by 1 H NMR analysis of the crude product(s). Under standard reaction conditions, it is not necessary to protect the hydroxyl group (2k) to afford product 3dk.</p><p>Cyclic amines act as widespread skeletons in drug molecules and natural products. Cyclic amines, such as morpholine 2b, pyrrolidine 2m, pyrroline 2t, piperidine 2i, piperazine 2h, 1,2,3,6-tetrahydropyridine 2s, tetrahydroisoquinoline (2l and 2p), and tetrahydroquinoline 2v, may all be modied with the allene unit furnishing the desired products 3bb, 3lb, 3fm, 3im, 3kt, 3ci, 3ah, 3js, 3el, 3hp, 3ol, and 3ov smoothly in 52-96% yields. Quaternary amine hydrochlorides exist widely in medical and pharmaceutical science due to their superior water solubility, absorption, and ease of formulation. Both nortropinone (2x) and 4-piperidinone (2y) furnished the desired products 3tx and 3uy in the presence of NaHCO 3 . Hydroxylamine hydrochloride 2z could also be allenylated, affording bis-allenylation product 4uz.</p><p>In order to demonstrate the scope of the current Ir-catalyzed 2,3-allenylation reaction, we applied this strategy for the latestage modication of drug molecules and derivatives of natural products (part IV). Two commercial drugs for cancer a The reaction was conducted using 1a (0.2 mmol) and BnNH 2 (0.8 mmol) in 1.0 mL of THF. b Determined by 1 H NMR analysis of the crude reaction mixture using 1,3,5-trimethylbenzene as an internal standard. c The reaction was conducted using allene 1a (1.0 mmol), BnNH 2 (4.0 mmol) in 5 mL of THF. d Isolated yield.</p><p>Scheme 2 Substrate scope of Ir-catalyzed 2,3-allenylation of amines. Standard conditions: Ir pre-catalyst A (3.5 mmol%), allene (1.0 mmol) and amine (4.0 mmol) in 5 mL of THF. a 5.3 mol% Ir pre-catalyst A was used. b 32% of 1b was recovered. c 10% of 1b was recovered. d 9% NMR yield of 4aa. e 5.3 mol% Ir pre-catalyst A was used and the reaction was conducted at 60 C. f 17% NMR yield of 4dk. g 7% NMR yield of 4fn. h The reaction was conducted using allene (1.0 mmol), amine$HCl (1.2 mmol), and NaHCO 3 (1.2 mmol) in 5 mL of THF. i 7 mol% Ir pre-catalyst A was used. j The reaction was conducted using allene (1.0 mmol), NH 2 OH$HCl (4.0 mmol), and NaHCO 3 (4.0 mmol) in 5 mL of THF and 6% NMR yield of 3uz was detected. k 6% NMR yield of 4ga was detected. l 15% of 1o was recovered and 3% NMR yield of 4oa was detected. m 7 mol% Ir pre-catalyst A was used and the reaction was conducted at 60 C.</p><p>treatment, Getinib and Erlotinib, 7 could be directly modied with an allene unit (products 7 and 8). Furthermore, the ATA reaction 8 of (AE)-citronellal and cholesterol with propargyl alcohol conveniently gave the desired allenols, which were treated with chloroformate to afford the carbonates 1r and 1s. Subsequent Ir-catalyzed reactions with tetrahydroisoquinolines and desloratadine 9 afforded 2,3-allenyl amines for 3rl, 9, and 3sp with 85-90% yields.</p><!><p>Nowadays, optically active 2,3-allenols are readily available from the EATA reaction. 10 Thus, aer unveiling the scope of this protocol, we were anxious to investigate the memory of chirality by applying optically active 2,3-allenyl carbonates. Our rst attempt was the reaction of (S)-1m with dibenzylamine 2f at room temperature with 3.5 mol% Ir pre-catalyst A (eqn ( 1)). The target product could be obtained with 95% yield and an excellent efficiency of chirality transfer (99% es), i.e., the axial chirality of the allene unit was passed to the axial chirality in the allenylic amines. To our knowledge, this is the rst example of such transition metal-involved chirality memory.</p><p>(1)</p><p>Encouraged by this result, we further examined the scope of memory of chirality (Scheme 3). For 4-alkyl-substituted allenyl carbonate (S)-1a, reactions afforded (S)-3aa in 71% yield and 98% es on a gram-scale with benzylamine 2a and (S)-12 in 85% yield and 95% es with 4-ethoxycarbonyl aniline, respectively. The more sterically hindered (S)-1f yielded (S)-3fn in 95% es. Various 4-aryl-2,3-allenyl carbonates were next investigated: different substituents on benzene ring, including F ((S)-1j), Cl ((S)-1k), and Br ((S)-1l), were intact under the reaction conditions (products (S)-3js, (S)-3kt, and (S)-3lb). Different types of acyclic (2a, 2n, 2o, and 2f) and cyclic amines (2s, 2t, and 2b) showed little inuence on the yields and stereoselectivity. Paroxetine hydrochloride, a drug used against depression and social phobia, 11 could also be modied with optically active 2,3allenyl carbonate ((S)-1d) efficiently to afford the desired product (S a ,S,R)-10 in high es. Late-stage modication of optically active derivatives of natural products, (S)-citronellal ((S a )-Scheme 3 Ir-catalyzed 2,3-allenylation of optically active 2,3-allenyl carbonates with amines. Reaction procedure: to a flame-dried Schlenk tube (25 mL) were added Ir pre-catalyst A (0.035 mmol)/THF (3 mL), amine (4.0 mmol)/THF (1 mL), and allene (1.0 mmol)/THF (1 mL) sequentially under an Ar atmosphere. The resulting mixture was stirred at r.t. a Enantioselectivity of the transformation based on the ee of the staring material. The reaction was conducted using allene (6.0 mmol) and amine (24.0 mmol) in 30 mL of THF at 40 C with a 10% NMR yield of (S,S)-4aa. c The reaction was conducted at 0 C. d 8% NMR yield of (S,S)-4fn was detected. e The reaction was conducted using (S)-1d (1.0 mmol), paroxetine$HCl$ 1 2 H 2 O (1.2 mmol), and NaHCO 3 (1.2 mmol) in 10 mL of THF.</p><p>1r) and cholesterol ((S a )-1s), were sequentially performed, furnishing the corresponding products (S a )-3rl and (S a )-3sp smoothly in high yields and excellent d.r. (Scheme 3).</p><!><p>The absolute conguration of (S)-3mf was conrmed by preparing this same product by following a known procedure. 12 To gain insight into the mechanism, a series of experiments was carried out. Firstly, we le a period of time for the racemization of the optically active alkylidene-p-allyl iridium intermediate as (S)-1m was treated with different amounts of Ir pre-catalyst A for as long as 12 h followed by the addition of the nucleophile dibenzylamine 2f. As showed in Scheme 4a, the level of racemization is linear with the loading of the Ir pre-catalyst A. With the increased loading of the Ir pre-catalyst A, the decarboxylation product, 2,3-allenyl methyl ether 11 was also generated with the same level of ee as compared to that of (S)-3mf. It is notable that the treatment of (S)-1m with 27% of precatalyst A for 12 h followed by the addition of dibenzylamine still formed (S)-3mf in 74% ee, which may be attributed to the much slower rate of racemization (Scheme 4b). Malcolmson et al. 13 had reported that the rapid racemization of optically active 2,3-allenyl amine happened even in the presence of a chiral palladium catalyst (Scheme 4c and d). However, the treatment of the optically active product (S)-3mf under the standard reaction conditions did not lead to racemization (Scheme 4e). Moreover, no racemization of (S)-3mf or generation of (S)-3mr was observed in the scrambling experiment, indicating that the nucleophilic attack of amine is irreversible under the current Ir-catalyzed reaction conditions (Scheme 4f).</p><p>Based on the mechanistic experiments, a catalytic cycle is proposed as shown in Scheme 5. At rst, the oxidative addition Scheme 4 Mechanistic studies. The reaction procedure for the data in (a): to a flame-dried Schlenk tube were added Ir pre-catalyst A (x mol%)/ THF (1.5 mL), and (S)-1m (0.5 mmol)/THF (0.5 mL) sequentially under an Ar atmosphere. The resulting mixture was stirred at room temperature for 12 h. Dibenzyl amine 2f (2.0 mmol)/THF (0.5 mL) were then added sequentially under an Ar atmosphere. The resulting mixture was stirred at room temperature.</p><p>Scheme 5 Proposed catalytic cycle.</p><p>of Ir(I)-h 2 -(S)-1 I leads to the formation of optically active alkylidene-p-allyl Ir(III) II, accompanied by the release of a carbonate anion. The chirality is memorized in this intermediate since the isomerization to form the allylic Ir intermediate V is very slow. The subsequent irreversible attack of amine results in the formation of Ir(I)-h 2 -2,3-allenyl aminylium ion III, which is much faster than the attack of methoxy in the carbonate anion aer releasing carbon dioxide. The subsequent coordination of another (S)-1 with Ir(I) in III regenerates I and releases the 2,3allenyl ammonium ion VI to complete the catalytic cycle. With the aid of a carbonate anion, the nal product (S)-3 could be afforded by the deprotonation of VI, accompanied by the release of carbon dioxide and methanol.</p><!><p>In conclusion, although racemizations of the optically active alkylidene-p-allylic transition metal complexes are very common, leading to the erasure of the chiral information in the starting materials and have been extensively applied to the asymmetric syntheses of optically active allenes, 14 we have recorded here the rst example of chirality memory involving alkylidene-p-allylic transition metal species and developed a robust highly stereoselective approach for asymmetric allene synthesis. Such a chirality memory involving iridium will be critical for the storage and transmission of the axial chirality in optically active allenols, which are readily available from propargylic alcohols and aldehydes. Further studies with other nucleophiles in this area are being actively pursued in our laboratory.</p>

Royal Society of Chemistry (RSC)

Metabolism and PK characterization of metarrestin in multiple species

Purpose: Metarrestin is a first-in-class pyrrolo-pyrimidine-derived small molecule targeting a marker of genome organization associated with metastasis and is currently in preclinical development as an anti-cancer agent. Here, we report the in vitro ADME characteristics and in vivo pharmacokinetic behavior of metarrestin. Methods: Solubility, permeability, and efflux ratio as well as in vitro metabolism of metarrestin in hepatocytes, liver microsomes and S9 fractions, recombinant cytochrome P450 (CYP) enzymes, and potential for CYP inhibition were evaluated. Single dose pharmacokinetic profiles after intravenous and oral administration in mice, rat, dog, monkey, and mini-pig were obtained. Simple allometric scaling was applied to predict human pharmacokinetics. Results: Metarrestin had an aqueous solubility of 150 \xce\xbcM at pH 7.4, high permeability in PAMPA and moderate efflux ratio in Caco-2 assays. The compound was metabolically stable in liver microsomes, S9 fractions, and hepatocytes from six species, including human. Metarrestin is a CYP3A4 substrate and, in mini-pigs, is also directly glucuronidated. Metarrestin did not show cytochrome P450 inhibitory activity. Plasma concentration-time profiles showed low to moderate clearance, ranging from 0.6 mL/min/kg in monkeys to 47.9 mL/min/kg in mice and moderate to high volume of distribution, ranging from 1.5 L/kg in monkeys to 16.9 L/kg in mice. Metarrestin has greater than 80% oral bioavailability in all species tested. The excretion of unchanged parent drug in urine was <5% in dogs and <1% in monkeys over collection periods of \xe2\x89\xa5 144 hr; in bile-duct cannulated rats, the excretion of unchanged drug was <1% in urine and < 2% in bile over a collection period of 48 hr. Conclusions: Metarrestin is a low clearance compound which has good bioavailability and large biodistribution after oral administration. Biotransformation is the major elimination process for the parent drug. In vitro data suggests a low drug-drug interaction potential on CYP-mediated metabolism. Overall favorable ADME and PK properties support metarrestin\xe2\x80\x99s progression to clinical investigation.

metabolism_and_pk_characterization_of_metarrestin_in_multiple_species

Introduction<!>Solubility assay.<!>PAMPA.<!>Caco-2 assay.<!>Plasma protein binding.<!>In vitro metabolic stability studie<!>CYP-mediated metabolism and CYP inhibition potential assessment<!>Metabolite identification<!>Pharmacokinetic studies<!>Sample processing and bioanalytical analysis<!>Calculation of pharmacokinetic parameters<!>Multispecies allometric scaling of pre-clinical in vivo clearance and volume of distribution<!>Solubility, plasma protein binding, and permeability of metarrestin<!>In vitro metabolic stability<!>CYP-mediated metabolism and CYP inhibition<!>In vitro metabolite identification<!>Pharmacokinetics of metarrestin across species<!>Allometric scaling<!>Discussion<!>

<p>Metastasis involves complex, multistep processes that allow for the selection of cancer cells with the ability to detach from primary tumors, enter vascular and lymphatic spaces to propagate, infiltrate, and colonize distant organ sites1,2. Metastasis, or complications of metastasis including cancer cachexia, are in over 80 percent of patients afflicted by cancer either a direct cause of death, or a major contributing factor in cancer-related mortality3,4. There is a void of effective therapies targeting the metastatic disease stage and the identification of effective interventions for stage IV cancer patients3,5,6.</p><p>Metarrestin (7-benzyl-4-imino-5,6-diphenyl-4,7-dihydro-3H-pyrrolo[2,3-d]pyrimidin-3-yl) is a novel pyrrolopyrimidine-based small molecule inhibitor specifically developed against the metastatic disease stage7,8. Metarrestin was derived after medicinal chemistry optimization of a series of small molecules from a high content drug screen disassembling the peri-nucleolar compartment (PNC), a small nuclear body adjacent to the nucleolus and a marker of genome organization associated with the metastatic disease stage9–11. Metarrestin was developed under the hypothesis that complex phenotypic markers that reflect the malignant capacity of cancer cells and are specific for the metastatic disease stage are useful targets for the discovery of novel treatments12. The PNC is formed almost exclusively in cancer cells, and PNC prevalence (number of cells with one or more PNC bodies) is closely associated with metastatic dissemination and death due to metastasis in vivo, as well as disease outcomes in various solid organ cancers (breast, colorectal, ovarian)12–14. Therefore, PNC as a marker of genome organization may reflect the complex characteristics specific to metastasis better than a single gene or gene product2,8.</p><p>Metarrestin, as a first in class PNC disassembler, selectively modulates the phenotype of metastatic cancer cells, reducing invasion and migration, while having very little effects on normal cells8. In vivo, the molecule suppresses metastasis in multiple mouse xenograft models including prostate, breast and pancreatic cancer. In preclinical models of pancreas cancer, metarrestin led to a dramatic increase in survival due to the suppression of metastasis usually not observed for systemic cytotoxic chemotherapies or targeted small molecule inhibition of single genes. The promising and novel mechanism of action is supported by a favorable pharmacological profile of the molecule studied in rodents and in Ras-driven, autochthonous KPC (Pdx1-Cre;LSL-KrasG12D/+;Tp53R172H/+) mice15. Intratumoral metarrestin exposure levels at efficacious, non-toxic doses surpass in vitro IC90 measures several-fold and initial PK studies suggest that metarrestin is a low clearance compound with high tissue/plasma AUC0–24hrs ratios in several organs. Hence, for cancers with a dense, desmoplastic stroma like pancreas cancer which limits drug delivery and lowers intratumoral exposure levels, metarrestin is an attractive candidate.</p><p>Here we present the in vitro ADME properties of metarrestin including CYP-mediated metabolism, metabolite identification as well as in vivo pharmacokinetic studies in mouse, rat, monkey, dog and mini-pig to predict human pharmacokinetics parameters and support metarrestin's advancement into clinical trials.</p><!><p>Metarrestin is a compound with low total polar surface area (tPSA=60 ang2), high lipophilicity (CLogP = 4.34), and a MW = 474.61 g/mol (Fig. 1). It is characterized by a low number of rotational bonds and hydrogen donors in the pyrimidine ring7. 3D single-molecule imaging of metarrestin was performed with the drug discovery software platform Molecular Operating Environment (Chemical Computing Group ULC, Montreal, Canada). Aqueous solubility was measured using the shake-flask method16. Metarrestin was incubated at 27 °C at 200 μM (from freshly prepared 10 mM DMSO stock solution) in phosphate buffer, pH 7.4 for 24 hours with 200 rpm agitation. The mixture was vacuum filtered in 0.45 μM polycarbonate filter plate. Determination of soluble fraction of metarrestin was performed via chemiluminescent nitrogen detection (CLND) as described in16, where samples are injected by flow injection analysis by an HPLC pump and CTC-PAL autosampler with a VICI/VALCO 6-port injection valve to the workstation Antek Model 8060. Data analysis was performed by Analiza's proprietary Automated Discovery Workstation (ADW) considering 4 measurable nitrogens for metarrestin.</p><!><p>Parallel Artificial Membrane Permeability Assay (PAMPA) was used to determine the passive diffusion permeability of metarrestin (PappPAMPA)17. Stirring double-sink PAMPA method (Pion Inc., Billerica, MA) was employed to determine the permeability of compounds via PAMPA according to the manufacturer's instructions.</p><!><p>Caco-2 monolayer permeability system was used to assess metarrestin permeability and impact of efflux transporters. Caco-2 cells (1.2 × 105 cells/mL) were cultured in 96-well Transwell® insert filters (Corning Incorporated, New York, NY) for 21 days. Metarrestin (3 μM) permeability was measured in both directions: apical to basolateral (Pappcaco-2(A-B)) and basolateral to apical (Pappcaco-2(B-A)) to evaluate the efflux ratio. Experiments were carried out in HBSS with 10 mM HEPES buffer and adjusted to pH 7.4. Single point evaluation was performed after 2 hours of incubation at 37°C.</p><!><p>Protein binding was assessed by equilibrium dialysis method. The 96-well Equilibrium Dialysis Plate and HTD 96a/b Dialysis Membrane Strips (MWCO 12–14K) were purchased from HTDialysis, LLC (Gales Ferry, CT). Pooled human plasma (male and female) and pooled CD-1 mouse (male and female) were used for protein binding assay on the plasma side and 100 mM PBS buffer (pH 7.4) was used on the buffer side. Samples were analyzed by UPLC-MS/MS using the previously described bioanalytical method for metarrestin15. The fraction unbound was calculated as % Free = (Peak Area Ratio buffer chamber / Peak Area Ratio plasma chamber) *100%. Analogously, the % bound was calculated as % Bound = 100% − % Free. Recovery was also evaluated to account for unspecific binding using the equation of %Recovery = (Peak Area Ratio buffer chamber + Peak Area Ratio plasma chamber) / Peak Area Ratio initial plasma sample*100%.</p><!><p>Metarrestin (at 1 μM concentration) was incubated with human and mouse liver S9 and microsome fractions at 37°C for 1 hour with or without 5 mM NADPH; verapamil was used as incubation control. Samples were collected at 0, 15, 45 and 60 minutes in both liver fractions.</p><p>Hepatocyte metabolic stability was evaluated in cryopreserved hepatocytes from CD-1 mouse (Lot# SQW), Sprague-Dawley (SD) rat (Lot# CAN), Beagle dog (Lot# USY), Cynomolgus monkey (Lot# UCH), Gottingen mini-pig (Lot# EUN) and human (Lot# RBR) obtained from Bioreclamation IVT, LLC (Baltimore, MD). Metarrestin (1 μM) was incubated in duplicate in a 24-well plate containing mouse, rat, monkey, dog, minipig or human hepatocyte suspension (1 million cells/mL) at 37°C for 4 hours in controlled atmosphere (5% CO2, 60% humidity). Aliquots (50 μL) were removed at 0, 15, 30, 60, 120, 180 and 240 min and the samples were treated with acetonitrile (150 μL) containing IS (200 ng/mL tolbutamide). The sample mixtures were centrifuged (4000 rpm for 10 min) at 4°C and aliquots (100 μL) of the supernatant were taken and mixed with water (100 μL). A 5 μL aliquot was analyzed by an LC/MS/MS system consisting of a Shimadzu LC pump coupled with a Sciex API4000 triple quadrupole mass spectrometer equipped with an electrospray ionization source. Data obtained from metarrestin metabolic stability assays were used to calculate intrinsic clearance (CLint) as described in18.</p><!><p>The metabolic stability of metarrestin was evaluated against 7 recombinant CYP enzymes: CYP1A2, 2B6, 2C8, 2C9, 2C19, 2D6 and 3A4. Incubations took place in 0.1 M phosphate buffer (pH 7.4) containing 3 mM MgCl at 37°C. Reactions were started by the addition of NADPH (1 mM), sampling occurred at 0, 5, 10, 15 and 25 minutes. Phenacetin (CYP1A2), diclofenac (CYP2C9) and amitriptyline (CYP2C8, CYP2C19, CYP2D6 and CYP3A4) were used as positive controls. Results presented in percentage of drug remaining in incubation starting at 100% at time 0.</p><p>Metarrestin was evaluated for cytochrome P450 inhibition potential with marker substrate metabolism by monitoring CYP1A2 conversion of phenacetin to acetaminophen, CYP2B6 conversion of bupropion to hydroxybupropion, CYP2C8 conversion of amodiaquine to N-deethyl amodiaquine, CYP2C9 conversion of diclofenac to hydroxydiclofenac, CYP2C19 conversion of (s)-mephenytoin to hydroxymephenytoin, CYP2D6 conversion of dextrometorphan hydrobromide to dextrorphan and CYP3A4 conversion of both midazolam to 1-hydroxymidazolam and testosterone to 6β-hydroxytestosterone. Metarrestin was tested at 0,0.001, 0.005, 0.01, 0.05, 0.25, 1, 10 and 50 μM in 1 mg/mL human liver microsomes containing 1 mM NADPH. IC50 determination was evaluated in: A. 0-min pre-incubation; B. 30-min pre-incubation in the absence of NADPH; C. 30-min pre-incubation in the presence of NADPH.</p><!><p>The metabolite profiles of metarrestin (at 10 μM concentration) were evaluated in CD-1 mouse, SD rat, Beagle dog, Cynomolgus monkey, Gottingen mini-pig and human cryopreserved hepatocytes obtained from Bioreclamation IVT, LLC (Baltimore, MD). Midazolam (5 μM) was used as the positive control for the assay. The incubation was started by adding the compounds to wells containing 1 million cells/mL hepatocytes; the reaction was carried out at 37°C using gentle stirring in an incubator with controlled atmosphere (5% CO2, 60% humidity) for 4 hours. At the end of the incubation, the reaction was stopped with the addition of 3 volumes of acetonitrile. The mixtures were centrifuged to separate precipitated proteins and the supernatants were transferred to clean tubes that were dried under a stream of N2. Residues were reconstituted in 30% ACN in water for HPLC-MSn analysis. Chromatographic separation, detection and structure identification was performed by Agilent 1100 HPLC pumps, autosampler and PDA (Agilent Technologies, Palo Alto, CA) interfaced to LTQ ion trap mass spectrometer (Thermo Finnigan, San Jose, CA, USA) operating on Full Scan (m/z 250 – 2000) and Data Dependent MSn (n=4) analysis.</p><!><p>All animal protocols were reviewed and approved by the Animal Care and Use Committee (ACUC) of respective institutes for in-life studies, and studies were conducted in compliance with institutional animal use and welfare guidelines.</p><p>Six-to-eight week-old male C57B/L6 mice were purchased from the Mouse Repository, Frederick National Laboratories for Cancer Research (FNLCR, Frederick, MD) or Charles River Laboratory (Frederick, MD). In-life PK studies in rats, dogs and monkeys were conducted at Charles River Labs. Mini-pigs were obtained from Sinclair Bio Resources LLC (Auxvasse, Missouri), studies were conducted at NIH animal facility.</p><p>Mouse pharmacokinetics studies were previously reported in15. In summary, mice (body weight ranged from 21 to 30 g) were dosed with metarrestin solutions prepared in 30% PEG-400 and 70% (20% w:v HP-β-CD in water) at 3 mg/kg IV via tail vein injection or 3 mg/kg orally via gavage. The dosing volume was 3 mL/kg for IV route and 10 mL/kg for PO route. All dosing solutions were freshly prepared on the day of administration. Blood samples were collected over 24 hrs (n=3 per time point).</p><p>Male Sprague-Dawley rats (n=3 per treatment group, ~250–300 g at dose initiation) were used for 3 dosing groups. In Group 1, a single oral dose of 2 mg/kg was administered via gavage to three rats with blood sampling from jugular vein over 120 hr. In Group 2, a single IV dose of 2 mg/kg was administered to three rats with catheters at femoral (for dosing) and jugular vein (for sampling), plasma and urine samples were collected over 120 hr. In Group 3, a single IV dose of 1 mg/kg was administered to bile duct cannulated (BDC) rats that also had catheters at femoral and jugular vein for dosing and sampling. Blood, urine and bile samples from Group 3 were collected over 48 hr. The vehicle used was 30% PEG-400/70% of a 20% w/v solution of 2-hydroxypropyl-β-cyclodextrin (HP-β-CD) in water, dosing solutions were freshly prepared on the day of administration.</p><p>Male cynomolgus monkeys (n=3 per treatment group, 2.76–3.90 kg at dose initiation) received IV and PO single doses of metarrestin in 5% EtOH/30% PEG-400/65% of 20% HP-β-CD in water. The IV treatment group received a single dose of 0.5 mg/kg with a dosing volume of 0.5 mL/kg. For the PO treatment group, a single dose of 1.0 mg/kg with a dosing volume of 1 mL/kg was administered. Blood samples were obtained from each monkey up to 240 hr after IV and PO administration. Urine samples were collected over 144 hours and stored at −80℃ until analysis. Male Beagle dogs (n=3 per treatment group, 9–11 kg) received IV and PO single dose of metarrestin in 20% PEG-400/80% of 20% HP-β-CD in water. One group received a single IV dose of 0.5 mg/kg with a dosing volume of 0.5 mL/kg administered via a catheter placed in a saphenous vein. The PO group received a dose of 1.0 mg/kg with a dosing volume of 1.0 mL/kg administered via oral gavage. Plasma and urine samples were collected over a collection period of 216 hours.</p><p>Female Gottingen mini-pigs (n=3 per treatment group, 23–33 kg in PO experiment, 35–47 kg in IV experiment) received a single PO dose of 2.5 mg/kg of metarrestin, following 2–3 weeks of washout period, the same animals received a single IV dose of 0.25 mg/kg. The IV dose was administered via a centrally placed indwelling catheter in the jugular vein. Plasma samples were collected over 192 hours.</p><!><p>Blood samples from all groups and species were collected in K2EDTA tubes and stored on wet ice pending processing. Blood was centrifuged at 2200 x g at 5℃ for 10 minutes to isolate plasma. Isolated plasma, bile and urine samples were stored at −80℃ until analysis.</p><p>Metarrestin was extracted from plasma, bile and urine samples by acetonitrile. Samples were prepared by adding a 10.0 μL aliquot of plasma samples, QCs and standards into the appropriate wells of a 96-well plate. A 200 μL aliquot of 1.0 ng/mL D7-metarrestin (internal standard, IS) in acetonitrile solution was added to the samples. All samples were then thoroughly shaken for 5 minutes. The samples were centrifuged at 3000 rpm and 4°C for 30 minutes and 150 μL supernatant was transferred to a 96-well plate. A 0.2 μL aliquot of plasma, urine and bile supernatant were injected for UPLC-MS/MS analysis as reported before15.</p><!><p>Phoenix WinNonlin, version 6.4 (Certara, St. Louis, MO) was used to perform pharmacokinetic analysis with the non-compartmental approach (Models 200 and 201 for PO and IV datasets). The area under the plasma concentration vs time curve (AUC) was calculated using the linear trapezoidal method. The terminal rate constant (λ) was obtained from the slope of at least three data points of the apparent terminal phase of the log linear plasma concentration vs time curve. AUC0−∞ was estimated as the sum of the AUC0-t (where t is the time of the last measurable concentration) and Ct/λ. The apparent terminal half-life (t½) was calculated as 0.693/λ. Bioavailability (F) was calculated using ratios of AUC obtained from IV and PO pharmacokinetic studies corrected by each route's dose.</p><!><p>Experimental pharmacokinetic parameters were obtained for mouse, rat, monkey, dog and minipig using the protocols described above. Clearance (CLp) and volume of distribution at steady-state (Vdss) data obtained from the IV administrations were plotted against each species body weights and log-log regression analysis was performed with the equation Y =aWb, where Y is the experimental CLp or Vdss from each species, "a" is the intercept and "b" the exponent of the log-log line regression and W is the experimental animal body weight. The Rule of Exponents (ROE) was used to classify whether any corrections to the equations would be applied for metarrestin's profile19,20.</p><!><p>Metarrestin aqueous solubility was >150 μM or > 71.2 μg/mL at pH 7.4 and room temperature. The plasma protein binding for metarrestin was assessed in mouse and human plasma using ketoconazole as positive control. Ketoconazole was measured to be bound 98.37 and 99.24% to plasma proteins in human and mouse respectively, similar to previous reports21; metarrestin was 99.41 and 98.99% bound to human and mouse plasma proteins, respectively. Both compounds showed recovery higher than 90% in above protein binding studies. Metarrestin showed high permeability (Papp) in the PAMPA permeability assay (PappPAMPA). The measured PappPAMPA for metarrestin was 1185.3 10−6 cm s−1, which indicates high passive diffusion through the artificial lipid membrane. In the Caco-2 assay, the compound showed a modest absorptive permeability (A->B) with a PappCaco-2(A-B) value of 0.2 10−6 cm s−1. However, in the secretory direction studies, metarrestin showed a higher PappCaco-2(B-A) value of 2.0 10−6 cm s−1, leading to an efflux ratio of 10. Therefore, metarrestin is considered as a compound that highly permeates membranes through passive diffusion but is a moderate substrate to efflux transporters.</p><!><p>The metabolic stability of metarrestin was evaluated in liver fractions from mouse and human. Metarrestin was stable in human and mouse liver microsomal and S9 fractions, presenting >120 min t1/2. When incubated with hepatocytes from different species, metarrestin was stable for most species, only rat and mini-pig showed enough parent depletion to enable intrinsic clearance (CLint) estimation. Percent remaining of metarrestin after 240 min incubation in hepatocytes was 103 % in mouse (midazolam control 0.3%), 44% in rat (0.3%), 119% in monkey (4%), 118% in dog (3%), 71% in mini-pig (42%), and 86% in human (midazolam 26%). The estimated CLint for rat was 19±7.1 mL/min/kg and 9±4.2 mL/min/kg for mini-pig, respectively.</p><!><p>To investigate enzymes involved in the metabolism of metarrestin, the compound was incubated with 7 recombinant human cytochrome P450 isoenzymes (Table 1). Metarrestin was stable against most CYP isoforms except for CYP3A4, in that 30% of the compound was converted. This data suggests that metarrestin is a substrate for CYP3A4. To examine the risk of potential drug-drug interactions, metarrestin was tested on CYP inhibition with seven concentrations from 1 nM to 50 μM. An example of the CYP inhibition assay is shown in Fig. 2a, in that the depletion of the substrate midazolam, converted to 1-hydroxymidazolam by CYP3A4, was not inhibited by metarrestin whereas the known inhibitor mifepristone (a positive control), reduced metabolism of midazolam, in particular when pre-incubated for 30 minutes with NADPH. Overall, metarrestin did not show inhibitory activity against any of the tested CYP isoenzymes, including CYP1A2, CYP2B6, CYP2C8, CYP2C9, CYP2C19, CYP2D6 and CYP3A4 as the metabolism of marker substrates was not altered in the presence of metarrestin (Suppl. Fig. 1).</p><!><p>For metabolic profiling, metarrestin (10 μM) was incubated with hepatocytes from CD-1 mouse, Sprague-Dawley rat, Beagle dog, Cynomolgus monkey, and human. Metarrestin was found to be stable in all species. Trace metabolites were found in most of the species (Fig. 2b,c). M490a, M490b and M490c were mono-hydroxylated products, and M472 was proposed to be the keto derivative of the secondary alcohol of metarrestin. M650, a glucuronidation product of the secondary alcohol, was exclusively observed in mini-pigs. The metabolite profile of metarrestin has been partially described in rat and mouse hepatocytes previously15. Here, we show that M490a, M490b and M472 were identified in all six species including human. Although it is found in trace amounts and should not be critical for metarrestin activity, M472 is known to carry PNC disassembling activity7,15. Two novel metabolites were observed; M490c was observed only in monkey hepatocyte incubation, while M650 formed exclusively in mini-pig hepatocytes as mentioned above. Reverse-phase LC/HRMS analysis showed that metarrestin eluted at 24.5 minutes with m/z of 475.25 (Fig. 2b). The collision induced fragmentation of metarrestin produced two distinctive ions that split the structure in three regions. The ion with m/z 377.3 refers to the loss of the cyclohexanol moiety, isolating the 7-benzyl-4-imino-5,6-diphenyl-4,7-dihydro-3H-pyrrolo[2,3-d]pyrimidin-3-ium group. MS3 spectra provided the fragment with m/z 286.3 which is the further loss of methylbenzene from fragment m/z 377.3. The metabolite M490a was a mono-hydroxylation product eluting at 21.5 minutes that produced the parent ion at m/z 491.4, its further CID provided the m/z 393.2, an addition of sixteen mass units to m/z 377 observed in the parent. MS3 analysis showed that m/z 393 further fragmented to m/z 302, suggesting that the hydroxylation site was in the 7-benzyl-4-imino-5,6-diphenyl-4,7-dihydro-3H-pyrrolo[2,3-d]pyrimidin-3-ium group located in center portion of the parent drug molecule (Fig. 2c). Analogously, M490b eluting at 22.1 minutes also showed the loss of cyclohexanol, producing fragment m/z 393; but it further fragmented to m/z 287, comparable with the parent MS3 spectra, suggesting that the hydroxylation site was in the methylbenzene portion of metarrestin. The metabolite M472 eluting at 25.04 minutes represented the loss of two hydrogens. Based on MS3 spectra being similar to parent, producing 472>377>286, the dehydrogenation occurred most likely in the primary alcohol of the cyclohexanol moiety. M490c, found in monkey hepatocyte incubation, eluted at 22.16 minutes. It fragmented to the ions with m/z 377>286, indicating that the hydroxylation occurred in the cyclohexanol ring.</p><p>The major metabolite in mini-pig hepatocytes was the glucuronide product M650 which appeared to occur at the secondary alcohol of the compound. The M650 metabolite eluted at 21.6 minutes with m/z of 651.3, its fragmentation showed the loss of glucuronic acid generating metarrestin with m/z 475, which was confirmed by further fragmentation to metarrestin characteristic fragments 377>286. The HPLC chromatograms and proposed metabolic pathways for metarrestin from each species are depicted in Fig. 2b and Fig. 2c, respectively. Of note, since no formal standard curves of identified metabolites were established, differences in metabolic profiles upon cross-species comparisons should be regarded qualitative rather than quantitative. In summary, apart from CYP3A4, direct glucuronidation of metarrestin appears an alternative metabolism pathway for the elimination of the compound.</p><!><p>The pharmacokinetic profiles of metarrestin across species are shown in Fig. 3a, PK parameters obtained after IV administration of metarrestin derived from different species are summarized in Table 2. The non-compartmental pharmacokinetic analysis (NCA) was found fit to obtain total CLp and Vdss for each animal which are critical for the allometric scaling across species. The plasma clearance (CLp) of metarrestin was moderate in mouse and rat, however, in monkey, dog and mini-pig the CLp was low comprising less than 5% of the hepatic blood flow of each species. Following oral administration, metarrestin reaches systemic circulation with bioavailability greater than 80% in all species and extensively distributes to tissues producing large Vdss.</p><p>The renal elimination of metarrestin was assessed by collecting urine samples from rat, dog and monkey. Less than 1% of dose was excreted unaltered in the urine from rat and monkey after IV administration; while in dog, around 3% of the administered dose was found in urine (Fig. 3b). Biliary excretion of metarrestin was assessed in bile duct-cannulated rat after intravenous administration of metarrestin. Only 1.6% of dose was detected unaltered. Collectively, these data suggest that excretion of metarrestin via urine and bile is low.</p><!><p>Allometric scaling analysis was conducted by plotting the experimental PK parameters (i.e. CLp and Vdss) of each species vs the body weights (mean values for mice and individual values for other species) and a log-log plot was drawn (Fig. 4). Cynomolgus monkey appeared to be an outlier; when present, the linearity of the PK parameters decreased significantly. When monkey data were excluded from the regression, the r2 value increased from 0.46 to 0.89 for CLp. The log-log equation (without monkey) for plasma clearance is CLp=5.86xW0.290 with a r2 value of 0.89 (Fig. 4a). Similar results were obtained for Vdss. When Vdss was calculated using simple allometric (SA) scaling, the r2 was 0.83 using data from all species and 0.95 when excluding monkey data. The regression equation for volume of distribution is Vdss=10.30xW0.873 (Fig. 4b). The calculated CLp and Vdss based on simple allometric scaling for a 70 kg human are 20.19 mL/min (13.81 mL/min when considering monkey data) and 421.4 L (273.6 L when considering monkey data), respectively. This information has been used to calculate human dosing; assuming similar potency of metarrestin in humans and the mouse model and similar AUCtumor : AUCplasma ratios of 19 (obtained after single dose oral administration of metarrestin and using noncompartmental PK analysis previously reported in15), as well as taking into account similar protein binding between both species and that human bioavailability is 80%, the predicted dose would be 1.25 mg of metarrestin in a 70 kg patient using monkey data and 1.82 mg without monkey data over a dosing interval of 48 hours. To estimate intratumoral exposure levels, under above assumptions, an oral dose of 1.82 mg metarrestin will translate into an AUCplasma of 1,198 ng*h/mL and based on the previously determined AUCtumor : AUCplasma ratios of 19 an AUCtissue of 22,762 ng*h/mL, or a mean concentration through 48 hours of exposure of 475 ng/mL or 1 μM metarrestin.</p><!><p>In this study we characterized in vitro and in vivo pharmacokinetic profiles of the first-in-class small molecule metarrestin. Metarrestin was developed under the premise that the PNC, a marker of genome organization which is highly correlated with the metastatic phenotype in solid organ cancers based on several series of well-annotated clinical specimens, is a bona fide drug target, and that compounds that disassemble the PNC harness anti-metastasis activity. Previous work of ours has shown that at doses given to achieve effective suppression of metastasis and extension of survival metarrestin achieves intratumoral exposure levels sufficient to disassemble the PNC in the absence of clinical or organ toxicity. This study aimed to determine ADME characteristics of metarrestin and its pharmacokinetic behavior in multiple animal species as well as use these preclinical PK profiles to estimate human clearance as metarrestin progresses to clinical studies. Metarrestin has high permeability and is metabolically stable in most species tested in vitro, a finding in line with the low clearance, large volume of distribution and good oral bioavailability of the compound observed in vivo. While permeability measured in the PAMPA assay was high with Papp(PAMPA) > 1000 10−6 cm s−1, permeability in the Caco-2 assay demonstrated lower PappCaco-2(A-B) value of 0.2 10−6 cm s−1. Together with the measured efflux ratio of 10 in the basolateral to apical (B-A) Caco-2 assay, these findings suggest that, despite the high passive permeability of metarrestin, the compound is a substrate for efflux transporters. While cancers are well known to upregulate efflux transporters to protect themselves from xenobiotica22, it is important to notice that (1) metarrestin administered orally achieves plasma exposure levels comparable to metarrestin given by IV administration, and that (2) metarrestin has shown appropriate tumor distribution in KPC mouse model with autochthonous pancreatic tumor suggesting that the identified moderate efflux has a limited impact on biodistribution of the molecule due to its high passive permeability15.</p><p>Metabolic stability studies revealed that metarrestin is stable in liver S9 and microsomal fractions. After 4 hours hepatocyte incubation, metarrestin showed an estimated CLint of 19 and 9 mL/min/kg in rat and mini-pig, respectively. To calculate plasma clearance from CLint, additional factors, such as unbound concentration available to enzymes, blood flow permeating eliminating organs and blood/plasma partition, need to be taken into consideration23–25.</p><p>Metarrestin showed no significant CYP inhibition in a panel of human major CYP enzymes at a concentration range of 1 nM to 50 μM, suggesting drug-drug interaction potential on the metabolism of a co-administered drug is low. On the other hand, metarrestin metabolism appeared dependent on a single CYP isoenzyme, CYP3A4; therefore co-administration of other drugs that are either CYP3A4 substrates or CYP3A4 inhibitors (e.g. amiodorone, clarithromycin, telithromycin, nefazodone, itraconazole, ketoconazole, protease inhibitors, in particular anti-retrovirals) may affect metarrestin elimination. Also, patients with CYP3A4 genetic polymorphisms may be at risk of enhanced accumulation of metarrestin26. In addition, irrespective of CYP status, disposition of metarrestin is associated with an inherent elevated risk of drug accumulation after multiple-dose treatment.</p><p>Following absorption, metarrestin shows extensive distribution to tissues, marked by a high volume of distribution and high tissue penetration15. For monkey, dog and mini-pig which doses for IV and PO were at different levels, F was sometimes higher than 100% probably due to the lack of data on pharmacokinetic linearity between doses, a potential limitation of the study. Nonetheless, F for metarrestin is high, probably larger than 80% in all species. The low CLp was not unexpected considering the in vitro metabolic stability of metarrestin and low extrahepatic clearance. The low CLp, combined with high Vdss produced long terminal half-life (t1/2) ranging from 5 to 150 hours across species (Table 2). The maximum plasma concentration (Cmax) was achieved at 2 hours (Tmax) after oral administration in mouse and rat. In dog and monkey, the Cmax was observed around 15 hours post administration. Metarrestin plasma concentrations in mini-pigs were sustained oscillating around Cmax for a long period, producing an artifact Tmax of 105 hours after oral administration. In our experimental design, we covered up to >216 hours of metarrestin exposure in large animals (i.e. monkey, dog and mini-pig); however, it proved insufficient to ideally assess metarrestin elimination phase in dog and mini-pig due to slow decline of plasma concentrations in these species. The AUC0-t/AUC0−∞ ratio was <60%, therefore caution must be employed when interpreting the PK parameters due to a prolonged low plasma concentrations observed over time in these species. It is noteworthy to mention that the AUC0-t/AUC0−∞ for mouse, rat and monkey were all larger than 95%, which provides more accurate calculation of PK parameters with the non-compartmental approach.</p><p>Due to the low metabolic clearance and high volume of distribution, metarrestin's elimination t1/2 is quite long in large mammals, such as mini-pig with elimination t1/2 of ~80 hours or dog with elimination t1/2 ~150 hours. This risk of accumulation might be mitigated by a less than daily dosing schedule with additional safety measures, such as PK-guided or reduced dose escalation schemes during phase I clinical testing27.</p><p>Initial five species allometric scaling showed that the monkey does not fit well into the simple allometry model. The log-log plot of CLp versus body weight of mouse, rat, monkey, dog and mini-pig generates r2 values of 0.46. However, when monkey is removed from the plot, the r2 increases to 0.89. For that reason, the simple allometry (SA) was performed with and without the monkey data. To account for the possibility that monkey dose might have been above a linear pharmacokinetics range, a repeat PK study was performed at a lower IV dose of 0.2 mg/kg of metarrestin, but r2 values of simple allometric scaling did not change. There has been an increased recognition in the field on the limitations of simple allometric scaling approaches and the reliability on r2 to assess the predictability of human CLp from simple allometric models28. For example, r2 in allometric scaling relationships has been reported to be artificially high29. To improve prediction of human CLp, Mahmood and Balian developed the Exponent rule-corrected allometry (ROE) algorithm which included maximum life-span potential (MLP) and brain weight adjustment (BRW) to improve the prediction of allometric relationships via evaluation of simple allometric equations in order to make the appropriate corrections to the data19. The ROE proposes that when the exponent of the simple allometric log-log curve is between 0.55 and 0.7, SA can reliably be used to calculate CLp. If the exponent is between 0.7 and 1.0, the CLp should be multiplied by maximum lifespan potential (MLP) of each species before being plot against animal body weight in order to improve the regression and prediction. When the exponent is larger than 1.0, CLp will be multiplied by each species' brain weight (BW) to improve the regression and prediction. However, when the exponent is below 0.55 or larger than 1.3, neither model would help to generate a reliably predicted CLp. In general, it would be expected that models with exponent below 0.55 would greatly under-predict the data, whereas models with exponent larger than 1.3 would over predict. The simple allometry for metarrestin using mouse, rat, dog and mini-pig showed an exponent of 0.29, which falls under the lower exception region of the ROE which means that the human CLp would be under-predicted by simple allometric scaling. Even when monkey data was included in the dataset, the exponent for the SA is 0.28. By using the ROE's assumptions, it can be expected that the human CLp could be greater than 20 mL/min or 0.29 mL/min/kg19,30.</p><p>The Vdss was also scaled from animal intravenous PK data. Vdss scaling is usually well assessed by SA20. Again, the monkey data was excluded from the calculations to keep consistency with the clearance calculation. When monkey Vdss is excluded from the curve, the r2 increases from 0.83 to 0.94. The human Vdss was calculated to be 273.6 L or 3.9 L/kg including monkey data and 421.4 L or 6.0 L/kg in absence of monkey data.</p><p>The drug exposure in terms of AUC in human tumor tissue can be predicted based on intratumoral concentration in the transgenic pancreas cancer mouse model (KPC model). It has been previously determined in cell-based models that metarrestin IC50 is 0.4 μM and in vivo KPC mice found excellent response to intratumoral concentrations of 4 μM. To a first in human study, the target tissue concentration should be at least 2-fold the IC50, at 1 μM. The allometric scaling of human parameters herein presented allowed the estimation of the range for the first human dose as 1.25 to 1.82 mg of metarrestin to a 70 kg patient to afforda 1 μM intratumoral tissue concentration. It is important to highlight in this context that these predictions assume similar potency of metarrestin in humans and the mouse model, similar AUCplasma : AUCtumor ratios in rodents and humans, and similar protein binding between the two species. Since dose linearity studies were conducted in mice (dose level AUC0–2hrs correlation, r2=0.96) but not in the other species, bioavailability interpretations might be limited for species which had IV and PO doses at different levels (dog, monkey, mini-pig).</p><p>In summary, metarrestin is a first-in-class metastasis inhibitor with therapeutic efficacy in preclinical models of cancer. The in vitro ADME characteristics and in vivo pharmacokinetic findings on metarrestin are in line with a low clearance compound which has good bioavailability and large biodistribution after oral administration supporting metarrestin's progression to clinical investigation. The observed extended elimination t1/2 of metarrestin likely constitute an increased risk of accumulation which might warrant early development of pharmacovigilance criteria.</p><!><p>Suppl. Fig. 1. Impact of metarrestin on CYP2 and CYP3 metabolism. Examined CYP isoform shown on top, known substrate and metabolite depicted on bottom. Impact by known inhibitor on CYP function (used as positive control) shown in blue curves, impact on CYP metabolism by metarrestin indicated by orange and red dose response curves (in triplicates).</p>

PubMed Author Manuscript

Differential COX-2 induction by viral and bacterial PAMPs: Consequences for cytokine and interferon responses and implications for anti-viral COX-2 directed therapies☆

Highlights•We report interactions of Toll-like receptors (TLRs) with COX enzymes in vivo.•COX-2 was broadly induced by LPS (TLR4) but more locally by poly(I:C) (TLR3).•COX-1/2 deletion modified the response to TLR activation in a TLR-specific manner.•COX-2 deletion enhanced interferon responses to viral-type TLR3/7/9 ligands.•COX-2 inhibition could provide a novel anti-viral therapeutic strategy.

differential_cox-2_induction_by_viral_and_bacterial_pamps:_consequences_for_cytokine_and_interferon_

Introduction<!>Animals<!>Bioluminescent imaging<!>Luciferase activity assays<!>Animal condition, body temperature and blood counts<!>Plasma cytokine levels<!>Isolated blood assays<!>Statistics and data analysis<!>Cox2 gene induction by prototypical viral and bacterial PAMPs<!>Modulatory role of COX-2 on physical features of sepsis induced by LPS in vivo<!>PAMPs produce specific patterns of cytokine and interferon response in vivo<!>Complex role of COX-2 in cytokine and interferon responses to viral and bacterial PAMPs<!>Discussion<!>Role of funding sources<!>Supplementary data 1<!>

<p>Cyclooxygenase (COX) enzymes catalyze the two-step conversion of arachidonic acid to the unstable prostaglandin (PG) intermediate PGH2, which is then further converted to a range of prostanoid mediators that include PGE2, prostacyclin (PGI2) and thromboxane (TXA2). Two COX isoforms, which catalyze identical reactions, exist. COX-1 is constitutively expressed in many tissues, and generally plays a role in homeostatic function [1,2]. In contrast, COX-2 is generally not expressed in most healthy tissues but is rapidly induced in response to mitogens [3] and cytokines [4] and is often present in elevated levels at sites of inflammation. COX-2 produces prostanoids that contribute to vasodilation, increased vascular permeability, leukocyte chemotaxis, fever and that potentiate nociception. Indeed, COX-2 is the target of both traditional non-steroidal anti-inflammatory drugs (NSAIDs) such as aspirin and ibuprofen, which inhibit both COX-1 and COX-2, and of newer COX-2 selective inhibitors such as celecoxib. These agents are widely used for the symptomatic control of pain and inflammation, particularly in patients with arthritis, reflecting COX-2 induction in the inflamed arthritic joint.</p><p>In addition to its expression in sterile inflammation, it is clear that COX-2 is induced by pathogens. For example, gram-negative bacteria, or lipopolysaccharide (LPS), which activates the prototypical bacterial pattern recognition receptor Toll-like receptor (TLR)4, can induce COX-2 expression both in isolated cells [5,6] and in vivo [5,7]. Viruses or viral mimetics such as poly(I:C) can also induce COX-2 in cultured cells [8–10]. However, no systematic comparison of the ability of bacterial and viral pathogen-associated molecular patterns (PAMPs) to induce COX-2 in different tissues has been described. Moreover, whilst prostaglandins are recognized as important modulators of the immune system [11], there is no study that compares the roles of COX-1 versus COX-2 in inflammatory responses induced by viral and bacterial PAMPs. Here, we use a "knock-in" Cox2 luciferase reporter mouse[7], in which luciferase activity expressed from the endogenous Cox2 promoter reflects Cox2 expression, to both visualize and quantify the tissue distribution of LPS- and poly(I:C)-induced Cox2 gene expression. We then explore the role of COX-1- and COX-2-derived prostanoids in modulating the inflammatory response to specific PAMPs, using mice deficient in either COX isoform.</p><!><p>Cox1−/− [1], Cox2−/− [12] and Cox2fLuc/+ mice [7] were back-crossed onto a C57Bl/6J background and identified by genomic PCR. Experiments were performed on 10–12 week old male and female mice. Animal procedures were conducted in accordance with Animals (Scientific Procedures) Act 1986 and after local ethical review by the Imperial College Ethical Review Panel or the UCLA Animal Research Committee, as appropriate for individual experiments.</p><!><p>Cox2fLuc/+ mice were injected with poly(I:C) (8 mg/kg; i.p.; Sigma, UK), LPS (from Escherichia coli serotype 055:B5; 0.1 or 10 mg/kg; i.p.; Sigma, UK) or vehicle (saline) under brief isoflurane anesthesia. 4 h later, d-luciferin (125 mg/kg; i.p.; Xenogen, USA) was administered. After a further 15 min, the dorsal skin was shaved and bioluminescent emission from this area was recorded over 3 min, using an IVIS imaging system (Xenogen, USA). Animals were then euthanized by isoflurane overdose and tissues were rapidly dissected and arranged on cell culture dishes, following which luciferase activity was imaged ex vivo. In both cases, collected photon number and images were analyzed using Living Image software (Perkin Elmer, USA) and quantified as the peak photon release/pixel detected from each tissue.</p><!><p>After bioluminescent imaging, tissues were snap frozen for biochemical measurement of luciferase activity. Tissues were dissociated using a Precellys24 bead homogenizer and the supernatants loaded into white 96 well microtitre plates. The luminescence was then read after injection of Luciferase Assay Reagent (Promega, UK). Protein concentration of homogenates was determined using the bicinchoninic acid method (Perbio, UK) and luciferase activity was normalized for protein concentration.</p><!><p>Ligands to TLR2/1 (Pam3CSK4; 2 mg/kg; Invivogen, UK), TLR2/6 (MALP2; 60 μg/kg; Enzo Lifesciences, UK), TLR3 (poly(I:C); 8 mg/kg), TLR4 (LPS; 0.1 or 10 mg/kg), TLR7 (R848; 2 mg/kg; Enzo Lifesciences, UK), TLR9 (CpG ODN 1826; 2 mg/kg; Invivogen, UK) or vehicle (saline) were administered to wild-type, Cox1−/− and Cox2−/− mice by intraperitoneal injection, under brief isoflurane anesthesia. Immediately prior to injection, core body temperature was measured using a rectal thermometer. After 4 h, the gross physical condition of mice was scored (0–5) by two independent observers based on coat condition (e.g. piloerection), behavior (e.g. activity, response to investigator) and posture (e.g. hunching). Mice were then euthanized by CO2 narcosis, and body temperature immediately measured. Blood was collected, from the vena cava, into heparin (10 U/ml final; Leo Laboratories, UK). Whole blood cells counts were obtained using a commercial veterinary biochemistry service (IDEXX Laboratories, UK) and plasma separated from the remainder by centrifugation for cytokine analysis.</p><!><p>Plasma levels of IL-1β, IL-4, IL-5, IL-10, IL-12, IFNγ, TNFα and KC were determined using a multiplex immunoassay system (Meso Scale Diagnostics, USA). Plasma levels of IP-10 and IFNλ (R&D Systems, USA) and IFNα and IFNβ (PBL InterferonSource, USA) were determined by individual ELISAs.</p><!><p>Untreated wild-type, Cox1−/− and Cox2−/− mice were euthanized by CO2 narcosis and whole blood collected as above then treated with ligands to TLR2/1 (Pam3CSK4; 1 μg/ml), TLR2/6 (FSL-1; 1 μg/ml; Invivogen, UK). TLR3 (poly(I:C); 10 μg/ml), TLR4 (LPS; 1 μg/ml) or TLR7 (imiquimod; 10 μg/ml; Invivogen, UK), the NOD1 ligand C12-iE-DAP (1 μg/ml), mouse IL-1β (10 ng/ml) or saline. Treated blood was incubated for 24 h at 37 °C, after which the plasma fraction was separated by centrifugation and IP-10 and KC levels determined.</p><!><p>Data were analyzed using Prism 5.01 (GraphPad software, USA). Statistical significance was determined by two-way ANOVA with Dunnett's post-hoc test unless otherwise stated, and data sets considered different if p < 0.05. Each n value represents a data point from a separate animal.</p><!><p>Cox2 gene expression is increased globally in Cox2fLuc/+ animals treated with LPS, a TLR4 ligand [7]. In the current study, we extended our earlier observations and used two doses of LPS to observe a graded, dose-dependent increase in gene expression (Fig. 1A). When surface luminescence values for intact tissues were measured, the following tissues displayed a robust response (⩾2-fold increase versus vehicle-treated mice) to LPS (0.1 mg/kg): aorta, heart, liver, lung and spleen (Fig. 1A). When a larger LPS dose (10 mg/kg) was administered, all studied tissues measured displayed ⩾2-fold increase in Cox2 gene expression (versus vehicle) with the following rank order spleen ≫ heart ⩾ liver > lung ⩾ aorta > skin ≈ kidney ≈ stomach ≈ thymus ≈ gut ≈ brain. In contrast to results obtained with LPS, the viral PAMP/TLR3 ligand, poly(I:C), at a dose (8 mg/kg), close to the maximum tolerated dose [13], produced only a modest induction of Cox2 gene expression; significant induction was only apparent in the spleen (Fig. 1B).</p><p>Surface luminescence values do not always reflect the tissue activity in this model, since tissue absorbance of emitted bioluminescence and penetration of substrate in vivo influence measured luminescence. For this reason, we also measured luciferase activity, driven from the Cox2 promoter, in tissue homogenates. These data (Supplementary Fig. 1) confirm those obtained by surface luminescence measurements and also reveal the stomach as a site of modest poly(I:C)-induced Cox2 induction.</p><!><p>Wild-type mice treated with LPS displayed observable 'physical condition' responses, including hunching, inactivity and piloerection (Fig. 2A). The physical condition response to LPS was accompanied by a decrease in body core temperature, reflecting loss of thermoregulation (Fig. 2B), and reduced circulating lymphocyte and platelet counts (versus vehicle treatment), suggesting tissue lymphocyte sequestration and intravascular coagulopathy, respectively (Supplementary Fig. 2). These LPS-induced behavioral and body temperature changes were COX-2-dependent, but not COX-1 dependent, since they were strongly suppressed in Cox2−/−, but not Cox1−/− mice (Fig. 2A and B).</p><p>In contrast to the results for LPS, other PAMPs tested did not affect the gross physical condition or body core temperature of the mice (Fig. 2A and B). Nonetheless, like LPS, Pam3CSK4, poly(I:C) and R-848 produced lymphocytopenia (Supplementary Fig. 2), whilst MALP-2, poly(I:C), R-848 and CpG ODN produced an increase in circulating neutrophils, probably reflecting mobilization from the bone marrow (Supplementary Fig. 2). Cox1 or Cox2 deletion did not affect any of these features.</p><!><p>Each tested PAMP produced a specific pattern of plasma cytokine response, in agreement with reports from previous in vivo studies [14] and reflecting the distinct tissue distribution and signaling pathways of each TLR [15]. The pro-inflammatory cytokines IL-1β (Fig. 2C), TNFα and KC and the anti-inflammatory cytokine IL-10 (Supplementary Fig. 3) were increased in plasma of mice treated with LPS, but not in mice treated with the other PAMPs. IL-5 was also increased by LPS, as well as Pam3CSK4, whilst plasma IL-4 was elevated in mice treatedwith Pam3CSK4, CpG ODN and R-848 (Supplementary Fig. 3). Type I IFNs are intrinsically associated with viral infection [15]. Therefore, as expected, plasma IFNα levels were increased in mice treated with the viral PAMPs poly(I:C) and R-848. IFNα levels also tended to be increased in mice treated with the bacterial PAMPs LPS and CpG ODN (Fig. 3A). In wild-type mice, LPS and R-848 produced a significant increase in plasma IFNγ levels (Fig. 3B), but only MALP-2 produced a strong increase in the type III interferon, IFNλ (Fig. 3C).</p><!><p>To understand how prostanoids produced by COX-1 and COX-2 modify the cytokine responses to specific viral and bacterial PAMPs, we performed studies using mice deficient in either isoform. In line with what was seen with physical condition and body temperature, in mice-treated with LPS, the IL-1β (Fig. 2C) IL-5 (Supplementary Fig. 3) and IFNγ responses were markedly suppressed by Cox2 gene deletion (Fig. 3B; wild-type + LPS: 296 ± 43 pg/ml; Cox2−/− + LPS: 109 ± 13 pg/ml). By contrast, deletion of Cox1 did not alter the cytokine response to LPS, perhaps reflecting an overwhelming Cox2 induction by this stimulus (Fig. 1). More modest and variable roles for COX enzymes were noted in cytokine responses to other bacterial PAMPs, with effects of Cox1 and Cox2 deletion noted on specific IL-5 (Supplementary Fig. 3), IFNλ (Fig. 3) and IFNβ responses (Supplementary Fig. 4).</p><p>Cytokine responses to viral PAMPs exhibited a distinctly different pattern of response to Cox1 and Cox2 deletion. The most pronounced interactions was an increase of upto 6-fold in the IFNα, IFNγ and IFNλ (Fig. 4) and IFNβ (Supplementary Fig. 4) responses to poly(I:C) in Cox2−/− mice. This suggests that COX-2-derived prostanoids act to limit TLR3-mediated IFN release, and contrasts sharply with the pro-inflammatory role of COX-2 in LPS cytokine responses. In agreement, levels of the IFN-associated cytokine IL-12, and the IFN response cytokine IP-10 were also increased by Cox2 deletion in poly(I:C)-treated mice, as were basal levels of IL-12 in vehicle-treated animals (Supplementary Fig. 4). Whilst there was a tendency for IFNα to be increased in poly(I:C) treated Cox1−/− mice, this effect was less robust than that seen in Cox2−/− mice and was in opposition to reduced R-848 and CpG ODN IFNα responses in Cox1−/− mice (Fig. 3). This further illustrates the specificity of the interactions between COX-1/2 activity and PAMP/TLR pathways. These effects of Cox1/2 gene deletion on IFNs and related cytokines were associated with a systemic effect, since no change in IP-10 was noted in whole blood stimulated with PAMPs in culture. This contrasts to the clear increase in KC seen in whole blood stimulated with LPS (Supplementary Fig. 5).</p><!><p>COX-2 is an inducible enzyme that regulates the production of prostaglandins in inflammation and infection. It is increasingly recognized that COX-2 has a complex role in immune responses, the extent of which we do not yet fully understand. In the current study we used a mouse model in which expression from the Cox2 gene can be imaged directly, using a luciferase reporter "knocked in" to the coding region of the endogenous Cox2 gene, to compare expression after treatment with bacterial and viral PAMPs. We present data showing that bacterial LPS caused a pan-COX-2 induction across all studied tissues. In contrast, the viral PAMP, poly(I:C), induced a much more tissue-specific and limited induction from the Cox2 gene, with increases seen only in the spleen and stomach. These results likely reflect the limited distribution of TLR3 to specific immune cells, compared to broadly expressed TLR4.</p><p>In the current study, we show, for the first time, how deletion of Cox1 or Cox2 affects responses to a broad range of PAMPs. These data, summarized qualitatively in Fig. 4, clearly highlight distinct roles of COX-1 and COX-2-derived prostanoids in the modulation of specific PAMP/cytokine responses and suggest a potentially important role for COX-2 in anti-viral interferon responses.</p><p>In mouse models, the TLR4 ligand, LPS, induces a systemic inflammatory response that includes pronounced changes in the physical condition of the animal. Typical responses include piloerection, inactivity and loss of thermoregulation. In our study these effects were prevented by Cox2 gene deletion, consistent with previous reports [16], but not by Cox1 deletion. Importantly, these observations reflect what we know about the consequences of COX-2 inhibition in man; NSAIDs, including COX-2 selective inhibitors, are effective anti-pyretic drugs [17].</p><p>Whilst the other tested PAMPs did not induce notable changes in physical condition or thermal deregulation, they did produce cytokine changes. These effects were not ubiquitous and each PAMP elicited a relatively specific cytokine response profile (Fig. 4). Cox1 and Cox2 gene deletion studies suggested distinct interactions between COX-1, COX-2 and specific TLR responses. These differences likely reflect both the targeting of unique cell populations by each PAMP and the specific prostanoid pathways associated with these targets. Consistent with the anti-inflammatory effects of COX-2 inhibitors in man and in previous animal studies [16] we noted that Cox2 gene deletion attenuated IL-1β, IL-5 and IFNγ responses to LPS. In contrast, the response to other tested bacterial PAMPs were, for the most part, not consistently altered by deletion of either the Cox1 or Cox2 gene.</p><p>TLR3, the receptor for poly(I:C) is highly expressed on dendritic cells, which secrete IFNα and IL-12. These ligands, in turn, stimulate IFNγ production in natural killer cells [14]. Interestingly, Cox2 gene deletion enhanced poly(I:C)-induced release of type I (IFNα, IFNβ), type II (IFNγ) interferons and IL-12, consistent with an augmented dendritic cell response, as well as release of type III interferon (IFNλ) and the IFN response protein, IP-10. Indeed, elevated basal IL-12 levels in Cox2−/− mice suggest dendritic cells may be primed for activation in these animals, consistent with previous reports that prostanoids can limit dendritic cell function and survival [18]. TLR7 and TLR9, the receptors for the viral PAMPs R-848 and CpG ODN, respectively, have a broader expression in the immune system than TLR3 and couple to the MyD88 rather than TRIF adapter protein. Unlike TLR3-mediated IFNα responses, TLR7/TLR9 stimulated-release of IFNα was not enhanced by Cox2 gene deletion and was suppressed by Cox1 gene deletion, illustrating a complex interaction between viral PAMPs and the two COX enzymes.</p><p>In addition to the predicted anti-inflammatory role of COX-2 deletion, our observations showing enhanced anti-viral interferon responses in Cox2−/− mice to poly(I:C), suggesting that COX-2 inhibitors could have disease-modifying activity in viral infections. Interactions between COX-2 and viral infection has been suggested in a limited number of studies; COX-2 can be induced by whole virus in isolated cells [8–10] and COX-2 protein is increased in biopsy specimens of viral target tissues from patients with active viral infection [19,20]. Moreover, data from isolated cells [8,10] and animal models suggest that COX-derived products may play an important role in the host response. Cox2 deletion reduces mortality in mice infected with influenza A [21]. Moreover, in the same model, Cox1 deletion is associated with worsening of infection, consistent with our data demonstrating that COX-1 can limit TLR7/9 interferon responses. Protective effects of COX-2 inhibition have also been described in rodent models of respiratory syncytial virus [22] and vesicular stomatosis virus (VSV) [23]. Whilst there is little consensus as to the mechanism(s) responsible for this effect, limited previous data point to a role for IFN. For example, COX-2 inhibition in VSV-infected mice increases plasma IFNγ and IL-12 levels [23], and treatment of hepatitis B or C patients with the COX-1/COX-2 inhibitor indomethacin increases serum levels of the IFN response product 2′5′-oligoadenylate synthetase-1 [24]. Our data may, therefore, provide a mechanistic link by which COX-2 regulates viral infection as a result of modulation of innate immune recognition and subsequent interferon response.</p><p>Taken together, this study is the first to provide a systematic analysis of the reciprocal interactions between COX enzyme induction and inflammatory responses to a range of bacterial- and viral-like stimuli. The ability of Cox2 deletion to suppress physical and cytokine responses to LPS has been previously reported, and contrasts strongly with our new data showing that Cox2 deletion enhances IFN responses to viral PAMPs, particularly poly(I:C). These data suggest a role for COX-2 in limiting the anti-viral cytokine/interferon response to infection, and may provide a plausible explanation for the previously published data showing that Cox2 deletion/COX-2 inhibition is beneficial in animal models of viral infection. If a similar mechanism is present in man, COX-2 inhibitors might be a potential anti-viral therapy, able to boost the endogenous anti-viral response when given soon after infection.</p><!><p>This research was supported by a Wellcome Trust program grant (0852551Z108/Z; J.A.M.) and NIH-NCI P50 award CA086306 (H.R.H.). A.K.Z. is the recipient of an American Society of Hematology Scholar Award. No funding source was involved in the study design; data collection, analysis or interpretation; writing of the report or the decision to submit for publication.</p><!><p>This article contains supplementary material.</p><!><p>This is an open-access article distributed under the terms of the Creative Commons Attribution License, which permits unrestricted use, distribution, and reproduction in any medium, provided the original author and source are credited.</p><p>Supplementary data associated with this article can be found, in the online version, at http://dx.doi.org/10.1016/j.bbrc.2013.07.006.</p><p>Expression from the Cox2 gene induced in mice by prototypical bacterial and viral PAMPs. Cox2 promoter-driven gene expression was measured, by bioluminescent imaging of skin in vivo and isolated tissues ex vivo, in Cox2fLuc/+ luciferase reporter mice (A) and quantified as the fold-change in luminescence (B). The bacterial PAMP, LPS, produced a dose-dependent induction of Cox2 gene expression across a broad range of tissues. The viral PAMP, poly(I:C), in contrast, had little effect on Cox2-driven gene expression in most tissues, but produced a selective Cox2 induction in the spleen. Data are expressed as means ± s.e.m. from n = 4 to 5 individual animals per treatment. ∗p < 0.05 vs. vehicle.</p><p>Effect of Cox1 and Cox2 gene deletions on the physiology of mice treated with bacterial and viral PAMPs. Wild-type, Cox1−/− and Cox2−/− mice were treated with a range of bacterial and viral PAMPs. After 4 h "animal condition" was scored (A), and core body temperature change (B) and plasma IL-1β levels measured (C). Only LPS produced an apparent of loss of condition in mice; this phenotype was accompanied by a hypothermic response and increased plasma IL-1β levels. Cox2 gene deletion protected mice against each of these responses. Data are expressed as means ± s.e.m. from n = 6 to 12 individual animals per treatment. ∗p < 0.05 vs. wild-type; #p < 0.05 vs. vehicle-treated wild-type.</p><p>Effect of Cox1 and Cox2 gene deletions on circulating interferon levels in mice treated with bacterial and viral PAMPs. Wild-type, Cox1−/− and Cox2−/− mice were treated with a range of bacterial and viral PAMPs. After 4 h, the levels of IFNα (A), IFNγ (B) and IFNλ (C) were measured in plasma. In general, the viral PAMPs, poly(I:C), R-848 and CpG ODN, were better able to stimulate interferon production than their bacterial counterparts. In wild-type mice, poly(I:C) increased IFNα levels and tended to increase IFNγ levels, however, in each case the response was markedly increased in Cox2−/− mice. A similar pattern was true for poly(I:C)-induced IFNλ. In contrast, the IFNγ response to LPS was suppressed by Cox2 deletion, whilst Cox1 gene deletion limited the IFNα response to R-848 and CpG ODN. Data are expressed as means ± s.e.m. from n = 6 to 12 individual animals per treatment. ∗p < 0.05 vs. wild-type; #p < 0.05 vs. vehicle-treated wild-type.</p><p>Heatmap summarizing the effect of Cox1 and Cox2 gene deletions on the physiology and cytokine response of mice treated with bacterial and viral PAMPs. Wild-type, Cox1−/− and Cox2−/− mice were treated with a range of bacterial and viral PAMPs. After 4 h animal condition, circulating blood cell counts and plasma cytokine levels measured. Data are presented as mean (n = 6–12) fold-changes from wild-type vehicle. For clarity, color change is only shown for parameters differing significantly from vehicle. Because ranges of change differ greatly for different ligands and/or responses, scale bars are not shown and individual columns/rows are normalized internally.</p>

PubMed Open Access

Human Platelets Utilize Cycloxygenase-1 to Generate Dioxolane A3, a Neutrophil-activating Eicosanoid*

Eicosanoids are important mediators of fever, pain, and inflammation that modulate cell signaling during acute and chronic disease. We show by using lipidomics that thrombin-activated human platelets generate a new type of eicosanoid that both stimulates and primes human neutrophil integrin (Mac-1) expression, in response to formylmethionylleucylphenylalanine. Detailed characterization proposes a dioxolane structure, 8-hydroxy-9,11-dioxolane eicosatetraenoic acid (dioxolane A3, DXA3). The lipid is generated in nanogram amounts by platelets from endogenous arachidonate during physiological activation, with inhibition by aspirin in vitro or in vivo, implicating cyclooxygenase-1 (COX). Pharmacological and genetic studies on human/murine platelets revealed that DXA3 formation requires protease-activated receptors 1 and 4, cytosolic phospholipase A2 (cPLA2), Src tyrosine kinases, p38 MAPK, phospholipase C, and intracellular calcium. From data generated by purified COX isoforms and chemical oxidation, we propose that DXA3 is generated by release of an intermediate from the active site followed by oxygenation at C8. In summary, a new neutrophil-activating platelet-derived lipid generated by COX-1 is presented that can activate or prime human neutrophils, suggesting a role in innate immunity and acute inflammation.

human_platelets_utilize_cycloxygenase-1_to_generate_dioxolane_a3,_a_neutrophil-activating_eicosanoid

Introduction<!>Materials<!>Isolation of Human and Murine Platelets<!>Isolation and Activation of Human Neutrophils<!>Culturing and Activation of RAW 264 Cells<!>Isolation of Human Serum<!>Lipid Extraction<!>Generation of DXA3 through Oxidation of 11-HPETE or by Purified or Recombinant COX Isoforms<!>Reversed Phase LC/MS/MS and LC/MS3 of DXA3 and Platelet Eicosanoids<!>Generation of a Quantitative Assay for DXA3<!>Purification and Derivatization of DXA3 and GC/MS Analysis<!>2,3,4,5,6-Pentaflourobenzyl Bromide Derivatization of Carboxyls<!>Methyloxime Derivatization of Carbonyl Groups<!>Trimethylsilane Derivatization of Hydroxyl Groups<!>Acid Hydrolysis of DXA3<!>Tin(II) Chloride Reduction of DXA3<!>Statistics<!>Platelets Generate DXA3<!><!>Platelets Generate DXA3<!><!>Platelets Generate DXA3<!>DXA3 Activates Neutrophil Surface Integrin Expression<!>Human Platelets Acutely Generate DXA3 on Thrombin Activation via COX-1<!><!>Human Platelets Acutely Generate DXA3 on Thrombin Activation via COX-1<!><!>Elucidating the Mechanism of DXA3 Generation by COX Isoforms<!><!>Elucidating the Mechanism of DXA3 Generation by COX Isoforms<!><!>Generation of DXA3 Is Independent of Thromboxane Synthase<!>Generation of DXA3 by RAW 264 Cells and Human Serum<!><!>Discussion<!>Author Contributions<!>

<p>Emerging evidence indicates that platelets influence innate immunity during acute infection and injury through their interactions with leukocytes (1–3). Platelets generate soluble lipid-signaling mediators that include eicosanoids such as thromboxane A2 (TXA2)4 and 12-hydroxyeicosatetraenoic acid (12-HETE) and small amounts of the prostaglandins (PGs) PGE2 and D2. Currently the effects of platelet-derived lipids on leukocytes are not fully known. In this study, we sought to discover whether platelets release leukocyte-regulating lipids using a lipidomic approach. Analogous methodologies have recently been used for discovery of lipids in diabetes, cardiovascular disease, and hemostasis (4–6).</p><p>Herein, we show that thrombin-activated human platelets generate a novel eicosanoid from endogenous substrate, proposed to be a dioxolane (DX), that elevates Mac-1 (CD11b/CD18) on neutrophils at nanomolar concentrations. We also present the detailed cellular and enzymatic mechanisms of formation along with characterization of its proposed covalent structure and established a quantitative assay. These data demonstrate a new platelet-derived leukocyte-activating eicosanoid and the first DX lipid to originate from mammalian cells, suggesting a novel mechanism for promoting neutrophil activities in the early stage of tissue damage/wounding responses.</p><!><p>Lipids and lipid standards were purchased from Avanti Polar Lipids (Alabaster, AL) or Cayman Chemical (Ann Arbor, MI). Deuterated standards are as follows: arachidonic acid-d8, 5Z,8Z,11Z,14Z-eicosatetraenoic-5,6,8,6:54 PM 5/12/20169,11,12,14,15-d8 acid, ≥99% deuterated forms; PGE2-d4, 9-oxo-11α,15S-dihydroxy-prosta-5Z,13E-dien-1-oic-3,3,4,4-d4 acid, ≥99% deuterated forms; and PGD2-d4, 9α,15S-dihydroxy-11-oxo-prosta-5Z,13E-dien-1-oic-3,3,4,4-d4 acid, ≥99% deuterated forms. HPLC grade solvents were from Thermo Fisher Scientific (Hemel Hempstead, Hertfordshire, UK). PAR-1 and PAR-4 agonists were from Tocris Biosciences (Bristol, UK). COX-1 inhibitor (Sc-560) was from Cayman Chemical. Platelet signaling inhibitors (PP2, oleyloxyethylphosphocholine (OOEPC), bromoenol lactone, cytosolic phospholipase A2α (cPLA2α) inhibitor (N-{(2S,4R)-4-(biphenyl-2-yl-methyl-isobutyl-amino)-1-[2-(2,4-difluorobenzoyl)-benzoyl]-pyrrolidin-2-ylmethyl}-3-[4-(2,4-dioxothiazolidin-5-ylidenemethyl)-phenyl]acrylamide, HCl), U73112, wortmannin, and p38 mitogen-activated protein kinase (MAPK) inhibitor were from Calbiochem (United Kingdom). Anti-human CD11b-Alexa Fluor 647 was from eBioscience. All other reagents were from Sigma unless otherwise stated. [14C]Arachidonic acid was from PerkinElmer Life Sciences; ovine COX-1 was from Cayman Chemical or purified as described (7, 8). Recombinant COX-2 was generated as described (9). N-Methyl benzohydroxamic acid (NMBHA) and 2,2′-azobis(4-methoxy-2,4,dimethyl valeronitrile) were kind gifts from Ned Porter (Vanderbilt University).</p><!><p>Human blood donations were approved by the Cardiff University School of Medicine Ethics Committee, were with informed consent (SMREC 12/37 and SMREC 12/10), and according to the Declaration of Helsinki. For cPLA2-deficient samples, samples were approved by St Thomas's Hospital Research Ethics Committee, reference 07/Q0702/24: patient samples; South East NHS Research Ethics Committee. For studies on isolated platelets, whole blood was collected from healthy volunteers free from non-steroidal anti-inflammatory drugs for at least 14 days and added to acid/citrate/dextrose (ACD; 85 mmol/liter trisodium citrate, 65 mmol/liter citric acid, 100 mmol/liter glucose) (blood/ACD, 8.1:1.9, v/v) then centrifuged at 250 × g for 10 min at room temperature. Platelet-rich plasma was collected and centrifuged at 900 × g for 10 min, and the pellet was resuspended in Tyrode's buffer (134 mmol/liter NaCl, 12 mmol/liter NaHCO3, 2.9 mmol/liter KCl, 0.34 mmol/liter Na2HPO4, 1.0 mmol/liter MgCl2,10 mmol/liter Hepes, 5 mmol/liter glucose, pH 7.4) containing ACD (9:1, v/v). Platelets were centrifuged at 800 × g for 10 min and then resuspended in Tyrode's buffer at 2 × 108·ml−1. Platelets were activated at 37 °C in the presence of 1 mmol/liter CaCl2 for varying times, with 0.2 units·ml−1 thrombin, 10 μg/ml collagen, 10 μmol/liter A23187, 20 μmol/liter TFLLR-NH2, or 150 μmol/liter AY-NH2 before lipid extraction as below. Experiments involving signaling inhibitors (1 mmol/liter aspirin, 1 μmol/liter SC-560, 10 μmol/liter indomethacin, 2 μmol/liter oleyloxyethylphosphocholine, 50 nmol/liter bromoenol lactone, 50 nmol/liter cPLA2αi, 75 μm thimerosal, 1 mm EGTA, 10 μm 1,2-bis(2-aminophenoxy)ethane-N,N,N′,N′-tetraacetic acid tetrakis(acetoxymethyl ester), 100 nm wortmannin, 100 nmol/liter Gö 6850, 50 μmol/liter PP2, 100 nmol/liter p38 MAPK inhibitor, 50 μm picotamide, 1–10 mm iodoacetate, and 5 μm U-73122) included a 10-min preincubation at room temperature. In some experiments, calcium was omitted from buffers. For separation of cells from microparticles, platelets were centrifuged at 970 × g for 5 min, and the supernatants were re-spun at 16,060 × g for 5 min. For aspirin supplementation, blood samples were first obtained following a 14-day nonsteroidal anti-inflammatory drug-free period for baseline determinations of eicosanoids. Subjects were administered 75 mg/day aspirin for 7 days, and they then provided a second blood sample. Platelets were isolated and activated in vitro using 0.2 unit/ml thrombin, as described above, and then lipids were extracted as described below. Exclusion criteria was a known sensitivity to aspirin. For studies on isolated murine platelets, whole blood was collected using cardiac puncture (mice were 28 weeks old) into 150 μl of ACD (85 mmol/liter trisodium citrate, 71 mmol/liter citric acid, 100 mmol/liter glucose). 150 μl of 3.8% sodium citrate and 300 μl of Tyrode's buffer (145 mmol/liter NaCl, 12 mmol/liter NaHCO3, 2.95 mmol/liter KCl, 1.0 mmol/liter MgCl2,10 mmol/liter Hepes, 5 mmol/liter glucose, pH 7.35) were added, and the blood was centrifuged at 150 × g for 5 min at room temperature. Platelet-rich plasma was collected, and 400 μl of Tyrode's buffer was added to the red cells and centrifuged again at 150 × g for 5 min at room temperature. Platelet-rich plasma was combined and centrifuged at 530 × g for 5 min at room temperature. Platelets were resuspended in Tyrode's buffer at 2 × 108 ml−1. All animal experiments were performed in accordance with the 1986 United Kingdom Home Office Animals Act (Scientific Procedures). 12/15-LOX knock-out mice were generated as described previously (10), and wild-type male C57BL/6 mice (25–30 g) from Charles River, UK, were kept in constant temperature cages (20–22 °C) and given free access to water and standard chow.</p><!><p>Human neutrophils were isolated from 20 ml of citrate anticoagulated whole blood and resuspended in Krebs buffer. Briefly, blood was mixed 1:3 with 2% trisodium citrate (w/v) and HetaSep (Stemcell Technologies) and allowed to sediment for 45 min at 20 °C. The upper plasma layer was recovered and underlaid with ice-cold LymphoprepTM (2:1 for plasma/LymphoprepTM) and centrifuged at 800 × g for 20 min at 4 °C. The pellet was resuspended in ice-cold PBS and 0.4% sodium tricitrate (w/v) and centrifuged at 400 × g for 5 min at 4 °C. Contaminating erythrocytes were removed using up to three cycles of hypotonic lysis. Finally, cells were resuspended in a small volume of Krebs buffer (100 mmol/liter NaCl, 50 mmol/liter Hepes, 5 mmol/liter KCl, 1 mmol/liter MgCl2, 1 mmol/liter NaH2PO4, 1 mmol/liter CaCl2, and 2 mmol/liter d-glucose, pH 7.4), counted, and kept on ice. Neutrophils were diluted to 2 × 106 cells/ml and incubated with or without DXA3 for 10 min at 37 °C. In some experiments, 10 μm fMLP was then added, and neutrophils were incubated for a further 10 min at 37 °C. Cells were blocked using 5% mouse serum in PBS (containing 0.5% BSA, 5 mmol/liter EDTA, and 2 mmol/liter sodium azide) for 1 h on ice and centrifuged at 320 × g for 5 min at 4 °C. Anti-human CD11b-Alexa Fluor 647 (0.0625 μg, eBioscience) or isotype control were added and incubated for 30 min on ice. Neutrophils were washed twice with ice-cold PBS (containing 0.5% BSA, 5 mmol/liter EDTA, and 2 mmol/liter sodium azide) and dissolved in the same buffer for flow cytometric analysis. Neutrophils were analyzed on a cyan ADP flow cytometer (Beckman Instruments) and identified by forward and side scatter and Alexa Fluor 647. DXA3 used for these experiments was purified from COX-1 incubations. Other lipids were not detectable in these preparations using MS.</p><!><p>RAW 264 cells were cultured in DMEM (10% FBS, 1× penicillin/streptomycin) at 37 °C and 5% CO2.</p><p>To determine PG synthesis, cells were incubated in serum-free DMEM (with 1× penicillin/streptomycin), and the cells were incubated for 1 h at 37 °C, 5% CO2. Where used, 200 ng/ml LPS was added, and cells were incubated for 24 h. Cells (8 × 106 ml−1) were treated with 10 μm ionophore at 37 °C for 10 min, and lipids were extracted and analyzed as described below.</p><!><p>Whole blood from healthy volunteers was clotted at 37 °C for 15 min in glass and centrifuged (1500 rpm, 10 min, 4 °C). Serum was re-spun (2900 rpm, 10 min, 4 °C), and 3 volumes of MeOH/water (20:80 v/v) were added. Protein precipitates were spun down (13,000 rpm, 10 min, 4 °C), and supernatants were applied to preconditioned Waters C18 Sep-Pak columns. These were washed with 10 ml of water, 6 ml of hexane, and the eicosanoids were then eluted using 7 ml of methyl formate into tubes containing 6 μl of MeOH/glycerol (70:30 v/v) (11). Lipid were redissolved in methanol, chilled (−80 °C, 60 min), and re-spun (13,000 rpm, 10 min, 4 °C) before LC/MS/MS analysis for DXA3.</p><!><p>Lipids were extracted by adding a solvent mixture (1 mol/liter acetic acid, isopropyl alcohol, hexane (2:20:30, v/v/v)) to the sample at a ratio of 2.5 ml to 1 ml of sample, vortexing, and then adding 2.5 ml of hexane (12). Where quantitation was required, 5–10 ng of PGE2-d4, PGD2-d4, and 12-HETE-d8 were added to the samples before extraction, as internal standards. After vortexing and centrifugation, lipids were recovered in the upper hexane layer. The samples were then re-extracted by addition of an equal volume of hexane. The combined hexane layers were dried and analyzed for DXA3 using LC/MS/MS as below.</p><!><p>Arachidonic acid was oxidized using N-methylbenzhydroxamic acid (NMBHA) and 2,2′-azobis(4-methoxy-2,4-dimethylvaleronitrile) (MeOAMVN) as detailed below. To a 6.5 μm arachidonic acid solution in chlorobenzene, 3.5 eq of NMBHA and 0.1 eq of MeOAMVN were added, and the mixture was stirred at 37 °C for 5 h under O2. After drying under N2, the sample was dissolved in methanol and stored at −80 °C until purification. Isolation of positional isomers used a Spherisorb ODS2 column (5 μm, 150 × 4.6 mm; Waters) with a gradient of 50–90% solvent B (acetonitrile, 0.1% formic acid) in solvent A (water, 0.1% formic acid) for 60 min, 90% solvent B for 4.5 min, and then re-equilibrating to 50% solvent B over 9.5 min with a flow rate 1 ml·min−1. Elution was monitored at 205 nm (unoxidized lipid) and 235 nm (HPETE). Fractions were collected and positional isomers identified using MS/MS transitions for the free HPETEs as follows: m/z 317.2 → 115.1 (5-HPETE); m/z 317.2 → 155.1 (8-HPETE); m/z 317.2 → 151.1 (9-HPETE); m/z 317.2 → 167.1 (11-HPETE); m/z 317.2 → 179.1 (12-HPETE); and m/z 317.2 → 219.1 (15-HPETE). Next, purified 11-HPETE was oxidized using 0.1 eq of MeOAMVN in 5 ml of chlorobenzene by stirring at 37 °C for 5 h under O2, and the hydroperoxides were then reduced to corresponding hydroxides using tin chloride (SnCl2). Lipids were extracted using the hexane/isopropyl alcohol extraction method, as described earlier.</p><p>Apo-COX-1 was stored in 80 mm Tris, pH 7.8, at −80 °C. In some experiments, a commercial preparation was used (Cayman Chemical). Wild-type murine COX-2 (recombinant) was at 10.61 mg·ml−1. For heme reconstitution, apo-COX-1 or -2 (35 μg) was preincubated on ice for 20 min with 2 molar equivalents of hematin in phosphate buffer (100 mm potassium phosphate buffer, pH 7.4). Then, 3.5 μg of the reconstituted enzyme was added to 1 ml of phosphate buffer and 500 μmol/liter phenol and incubated for 3 min at 37 °C in the presence of 150 μm arachidonate (AA or AA-d8). Where [14C]AA was used, 9.8 μg of enzyme was incubated with 70 μm AA (259 kBq). In some experiments, wild-type COX-2 was compared with active site mutants (V349A and W387F). The reaction was stopped by using ice-cold lipid extraction solvent and immediate extraction of lipids after addition of 5 ng each of PGE2-d4 and PGD2-d4 as internal standards, when required. In some experiments, 10 μm diethylenetriaminepentaacetic acid was added just before holo-COX-1. DXA3 was analyzed using reverse phase LC/MS/MS as described below.</p><p>In some experiments, AA was replaced with 1-stearoyl-2-arachidonyl-phosphatidylethanolamine. Free and esterified DXAs3 were analyzed using reverse phase LC/MS/MS as below.</p><!><p>Several different LC separations were used on a 4000 Q-Trap platform. For high resolution mass analysis and fragmentation of free DXA3, a reversed-phase UPLC Fourier Transform MS method was used (Thermo Scientific Orbitrap Elite) using a Spherisorb ODS2 column (5 μm, 150 × 4.6 mm; Waters) with a flow rate of 1 ml·min−1. Solvent B was increased from 20 to 42.5% over 50 min, then increased to 90% over 10.5 min, held for 4 min, and then returned to 20% over 1 min. Equilibration time between runs was 4.5 min. Analysis was performed using heated ESI in negative ion mode at sheath, auxiliary and sweep gas flows of 70, 20, and 0, and capillary and source heater temperatures at 300 and 350 °C, respectively. LC/MS of parent ions was monitored using accurate mass in Fourier MS mode. Negative MS/MS spectra were acquired using higher energy collision-induced dissociation. Data-dependent MS3 of m/z 351 was carried out in ion trap-MS mode on the LTQ ion trap.</p><p>For MS/MS of m/z 351 or m/z 359, collision-induced dissociation (CID) was used with a resolving power of 30,000 in negative FTMS mode. Data-dependent MS3 of m/z 351 or m/z 359 from DXA3-d8 was carried out in negative FTMS mode with a resolving power of 15,000.</p><!><p>[14C]AA was oxidized using COX-1 as described for unlabeled AA above. The amount of [14C]DXA3 was determined by comparison with a [14C]AA standard curve analyzed using LC separation with radiochemical detection (Berthold Technologies) using a Spherisorb ODS2 column (5 μm, 150 × 4.6 mm; Waters) with a gradient of 20–42.5% solvent B (acetonitrile, 0.1% formic acid) in solvent A (water, 0.1% formic acid) over 50 min, 42.5–90% solvent B from 50 to 60 min, 90% solvent B from 60 to 64.5 min, 90 to 20% from 64.5 to 65.5 min and 20% solvent B from 65.5 to 75 min with a flow rate of 1 ml·min−1, and fractions were collected at 30-s intervals for LC/MS/MS confirmation of [14C]DXA3. The same gradient was also used for LC/MS/MS detection of [14C]DXA3 (m/z 353.2 → 165.1).</p><!><p>DXA3 was purified from lipid extracts of thrombin-activated platelets or COX-1 reactions using HPLC/UV on a Spherisorb ODS2 column (5 μm, 150 × 4.6 mm; Waters) with a gradient of 20–42.5% solvent B (acetonitrile, 0.1% formic acid) in solvent A (water, 0.1% formic acid) over 50 min, 42.5–90% solvent B from 50 to 60 min, 90% solvent B from 60 to 64.5 min, 90 to 20% from 64.5 to 65.5 min, and 20% solvent B from 65.5 to 75 min with a flow rate of 1 ml·min−1, and fractions collected at 30-s intervals. DXA3-containing fractions were identified using MS using m/z 351.2 → 165.1, and then H2O was removed using Sep-Pak C18 cartridge purification (Waters). DXA3 was stored in methanol at −80 °C, prior to derivatization and GC/MS analysis.</p><!><p>Lipid was dried under N2, and 25 μl of 1% 2,3,4,5,6-pentaflourobenzyl bromide and 25 μl of N,N-diisopropylethylamine, both in acetonitrile, were added. The mixture was vortexed and incubated for 30 min at 20 °C. The sample was dried under N2.</p><!><p>Lipids were dried under N2 in a glass vial. In a second vial, 1 ml of 1 n NaOH was combined with a few grains of methyloxime. The tubes were connected with the dry lipid in the uppermost tube and solvent in the lower tube to separate and prevent solvation and incubated for 2 h at 60 °C.</p><!><p>Lipid was dissolved in 50 μl of N,O-bis(trimethylsilyl)trifluoroacetamide and 50 μl of acetonitrile, vortexed, and incubated for 1 h at 60 °C. The lipid was dried under N2, and 2 ml of ethyl acetate and 1 ml of H2O were added. The sample was vortexed, and the ethyl acetate layer was recovered, dried, and then dissolved in 1:2 H2O/methanol for LC/MS analysis or isooctane acetonitrile for GC/MS. GC/MS was carried out on a DSQ Thermo Finnigan as follows: source temperature, 200 °C; reagent gas, methane; gas flow, 1.8 ml/min; negative polarity, full scan 50–600. Column was a Phenomenex 30 m ZB-1.</p><!><p>Semi-purified DXA3 generated by COX-1 was solubilized in acetonitrile (1 ml) before addition of 1% acetic acid (4 ml). Samples were left at room temperature for 30 min before extraction using a C18 solid phase extraction cartridge.</p><!><p>DXA3 generated via oxidation of 11-HPETE was reduced using 95 μg of SnCl2 in water for 10 min at room temperature. Lipids were re-extracted as above using hexane/isopropyl/acetic acid.</p><!><p>Data on platelets are representative of at least three separate donors, with samples run in triplicate for each experiment. Data are expressed as mean ± S.E. of three separate determinations. Statistical significance was assessed using an unpaired two-tailed Student's t test. Where the differences between more than two sets of data were analyzed, one-way ANOVA was used followed by Bonferroni multiple comparisons test, as indicated in the figure legends. p < 0.05 was considered statistically significant.</p><!><p>We initially sought to discover esterified eicosanoids by scanning for precursors of m/z 351.2 in negative ion mode and of lipid extracts from thrombin-activated platelets. This work is published as the characterization of phospholipid esterified PGE2 and is described elsewhere (13). During precursor scanning for 351.2, we uncovered an unknown lipid also attached to phospholipids that was also generated as a free acid. This is visible when analyzing free acid lipids at m/z 351.2, where two lipids are seeded, including PGE2, and a more prominent ion at 48 min (marked by *, Fig. 1A) (13). MS/MS demonstrated a complex spectrum with major ions at m/z 163.2 and 165.2 that did not match any known eicosanoids in the LipidMaps database (Fig. 1B). However, a number of ions were indicative of prostaglandins, specifically m/z 333, 315, 289, and 271. The daughter ion at m/z 165.2 was then used to selectively detect the lipid in multiple reaction monitoring mode. A single lipid was visible at 48 min (Fig. 1C). The high resolution m/z [M − H]− of 351.2177 suggests an elemental composition C20H31O5, corresponding to arachidonic acid plus three oxygen atoms, and the presence of 5 rings/double bonds (Fig. 1B). Its elution on reverse phase LC/MS/MS, considerably later than PGE2, indicated a less polar lipid (Fig. 1A).</p><!><p>DXA3 is generated by human platelets, characterization using MS/MS and MS3 fragmentation. A, LC/MS/MS of DXA3 generated by thrombin-activated platelets. LC/MS/MS separation of lipids from thrombin-activated (0.2 units/ml, 30 min) platelets, using m/z 351.2 → 351.1, as described in the supplemental material, using a Q-Trap 4000. The later peak, labeled by (*), was identified as DXA3. B, MS/MS spectrum of DXA3. MS/MS of m/z 351.2 was acquired at the apex of the peak in A on an Orbitrap Elite. C, LC/MS/MS of DXA3. LC/MS/MS monitoring m/z 351.2 → 163.1 demonstrated a single peak for DXA3. D, LC/MS/MS of DXA3. Analysis was undertaken on the Orbitrap Elite in FTMS mode, separated by using reverse phase LC, isolated at m/z 351.2 in the Velos Pro, then fragmented by using CID at 50 V, with resolution 15,000 ppm, as described under "Experimental Procedures." E, MS3 of daughter ion at m/z 333.2, with CID at 30 V. C, MS3 of daughter ion at m/z 315.2, with CID 30 V. F, proposed fragmentation pathway for m/z 351.2 generating m/z 333.2071, which fragments to m/z 271.2067 via m/z 315.1967. DXA3 loses H2O forming 333.2071. Following ring opening, leaving a keto group at C9, H2O, and CO2 are lost, generating m/z 271.2067 via a m/z 315.1967 intermediate, as shown.</p><!><p>Extensive structural characterization was undertaken using GC/MS, derivatization, MS/MS, and MS3 and suggested the lipid as DXA3. GC/MS presented in the supplemental material and demonstrates one hydroxyl and no carbonyls (supplemental Fig. 1). For further structural confirmation, DXA3 or DXA3-d8 generated by COX-1 was analyzed using high resolution MSn on an Orbitrap Elite, during LC elution. MS3 of daughter ions at m/z 333.1 and 315.2 indicate the origin of m/z 271.2 (Fig. 1, D–F), whereas MS3 of m/z 225.1 and 207.1 shows the origin of m/z 163.1 (Fig. 2). Of note, m/z 155 is a prominent ion generated on CID fragmentation of 8-HETE, thus supporting the position of the –OH group at C8. To further confirm, analogous experiments were undertaken using deuterated AA as substrate for DXA3 generation. MS3 of DXA3-d8 showed the same fragmentation pattern; however, many ions showed an additional ion at 1 atomic mass unit lower (e.g. m/z 340 and 321, as well as the expected 341 and 322), supporting our proposed fragmentation mechanisms as shown (supplemental Figs. 2 and 3). The lower m/z ions represent ring opening with addition of –H and concomitant loss of a single deuterium, as shown. Finally, a UV spectrum was acquired during purification of low nanogram amounts of COX-1-derived DXA3, showing a λmax at 238 nm, similar to that reported for a similar DX by Teder et al. (14). This confirms the presence of a UV chromophore and is consistent with a conjugated diene at C12–15 of the lipid backbone (Fig. 3, A and B).</p><!><p>Characterization of DXA3 MS/MS and MS3 fragmentation, using high resolution FTMS. A, MS3 of daughter ion at m/z 207.1, with CID 30 V. B, MS3 of daughter ion at m/z 225.1, with CID 30 V. Bottom panel, proposed fragmentation pathway for m/z 351.2 generating m/z 333.2, then via fragmentation of 225.1, both at 207.1 and 163.1 are formed. Following ring opening, with keto group at C11, H2O is lost, followed by two 1[5]-sigmatropic shifts generating m/z 333.2071 Following loss of a conjugated triene, m/z 225.1132 is generated, which then loses H2O, and via an intermediate fragments to m/z 207.1027 and last 163.1128.</p><p>DXA3 contains a UV chromophore and primes and activates neutrophil integrin expression. A, DXA3 can be detected at 235 nm, during LC elution. DXA3 was generated using COX-1 as described under "Experimental Procedures" and then purified using LC/UV. B, DXA3 contains a UV chromophore. A UV spectrum was acquired at the apex of the peak at 46.7 and shows a λmax at 238 nm. C, DXA3 activates neutrophil Mac-1 expression. Neutrophils were incubated with fMLP, DXA3, or both, before addition of anti-human CD11b (Mac-1)-Alexa Fluor 647 and flow cytometry analysis as under "Experimental Procedures." A representative experiment repeated with three individual donors is shown (n = 3, mean ± S.E.). D–F, representative histograms depicting increased Mac-1 expression following activation with fMLP, DXA3, or fMLP/DXA3. Line represents the Mac-1-positive neutrophil gate, as set using untreated neutrophils. G, DXA3 primes neutrophil responses to fMLP. Bar chart showing activation of Mac-1 expression in three donors by fMLP with/without DXA3 priming for 10 min (n = 3 mean ± S.E.). H, representative histogram showing increased fMLP-stimulated Mac-1 expression following priming by DXA3. Bar represents the Mac-1 positive neutrophil gate, as set using untreated neutrophils. Statistical significance used Mann-Whitney U test, ****, p < 0.0001; ***, p < 0.001; **, p < 0.01. n = 3 donors.</p><!><p>DXA3 is named based on the dioxolane structure, "A" for the first member of the class discovered, and 3 for the number of double bonds, as per traditional eicosanoid naming conventions (15). Further work is required to confirm the structure using NMR and to determine enantiomeric/geometric isomer composition, using synthetic standards once these become available.</p><!><p>Neutrophils incubated with purified DXA3 elevated surface Mac-1 (integrin, CD11b/CD18), comparable with fMLP activation (Fig. 3, C and D). Together, fMLP and DXA3 caused an additive effect on Mac-1 expression, suggesting they activate neutrophils by distinct pathways (Fig. 3, C–F). However, at 10 nm, DXA3 effectively primed for fMLP activation after a 10 min pre-incubation (Fig. 3, G and H).</p><!><p>As a purified standard is not yet available, we synthesized and purified a biogenic standard by COX-1 oxidation of [14C]AA oxidation in vitro, using COX-1. This was quantified using radiochemical detection and HPLC-purified. The radiolabeled standard was then utilized in LC/MS/MS assays, to quantify cold DXA3 generated by COX-1, which was then used as a primary standard for quantitation, against PGE2-d4 as internal standard. Fig. 4, A–D, shows the LC/MS/MS and MS/MS spectrum of purified [14C]DXA3 along with standard curves for both radiochemical detection of [14C]DXA3 and DXA3 versus PGE2-d4.</p><!><p>Setting up a quantitative assay for DXA3 and determining its levels in human platelets. [14C]DXA3 was generated, using COX-1 and purified using HPLC with radiochemical detection, as described in the supplemental material. [14C]DXA3 was quantified by comparing the radiochemical response to [14C]AA. A, LC/MS/MS analysis of m/z 353.2 → 165.1 showing [14C]DXA3 eluting at 49.51 min, undertaken on the 4000 Q-Trap as described under "Experimental Procedures." B, MS/MS spectrum of [14C]DXA3 showing m/z 353.2 as parent ion. Experiment was performed on the 4000 Q-Trap platform in enhanced product ion mode, as described under "Experimental Procedures." C, standard curve for [14C]DXA3 using LC/MS/MS detection used for quantification of unlabeled DXA3. D, standard curve for quantitation of DXA3 in biological samples. A standard curve was generated with varying DXA3 but keeping PGE2-d4 levels constant, and responses were plotted as shown. E, time course of eicosanoid generation by thrombin-activated platelets. Washed platelets were activated for varying times, using 0.2 unit·ml−1 thrombin, then lipids were extracted and analyzed using reverse phase LC/MS/MS, monitoring parent m/z 351.2 → 165.1 (DXA3), 319.2 → 179.1 (12-HETE), m/z 351.2 → 271.1 (PGE2), and m/z 369.2 →169.1 (TXB2), as described under "Experimental Procedures." Data are representative of experiments repeated three times on different donors (n = 3, mean ± S.E.). F, levels of eicosanoids generated by genetically unrelated volunteers. Data are shown as Tukey boxplots, where whiskers represent 1.5 the lower and upper interquartile range, data not included within the whiskers are displayed as an outlier. Statistical significance used Mann-Whitney U test, ****, p < 0.0001; ***, p < 0.001; **, p < 0.01. n = 7–10 donors.</p><!><p>DXA3 was undetectable basally with levels rising by 10 min of thrombin activation (n = 7–10 separate donors, mean ± S.E.). Levels were higher than PGE2, but lower than TXB2 or 12-HETE (a representative donor is shown in Fig. 4E, data for all donors as Fig. 4F). Levels varied between genetically unrelated donors for all eicosanoids. DXA3 formed on activation by collagen, ionophore, or collagen/thrombin of platelets, with high levels already apparent 2 min post-activation (Fig. 5A). Its thrombin-dependent formation was blocked by the selective COX-1 inhibitor, SC560, aspirin, or indomethacin in vitro or in vivo following administration of 75 mg/day aspirin for 7 days in healthy donors (Fig. 5, B–D). DXA3 was absent in thrombin-activated platelets from a patient with genetic deficiency of cPLA2 (16) or in the presence of the cPLA2 inhibitor, cPLA2i (Fig. 5, E and F). In contrast, neither iPLA2 nor sPLA2 appeared significantly involved (data not shown). Pharmacological inhibitors/agonists implicated PAR-1 and -4 receptors, src-tyrosine kinase, p38 MAPK, intracellular calcium, and PLC, but ruled out phosphatidylinositol 3-kinase, although PKC played an inhibitory role (Fig. 6, A–D). Murine platelets also generated DXA3 at levels similar to human cells; however, levels were significantly higher in platelets genetically deficient in a second arachidonate-oxidizing enzyme, 12-lipoxygenase (LOX) (Fig. 6E). Similarly, other COX-1-derived lipids were elevated in these platelets, and 12-HETE-was absent (Fig. 6, F–H). Collectively, these data show that thrombin-stimulated DXA3 generation depends on a highly coordinated signaling pathway, culminating in cPLA2-dependent hydrolysis of AA from phospholipids, prior to its oxygenation by COX-1 (Scheme 1).</p><!><p>Time course of agonist-stimulated generation and demonstrating the requirement for COX-1 and cPLA2 in DXA3 formation. A, generation of DXA3 by human platelets. Washed platelets were activated for varying times, using 0.2 unit·ml−1 thrombin, 10 μg/ml collagen, 10 μmol/literA23187, and then lipids were extracted and analyzed using reverse phase LC/MS/MS, monitoring parent m/z 351.2 → 165.1 as described under "Experimental Procedures." Levels are expressed as analyte/internal standard. Data are representative of experiments repeated at least three times on different donors (n = 3, mean ± S.E.). B and C, requirement for COX-1 for DXA3 formation. Platelets were incubated with inhibitors 10 min prior to thrombin activation (0.2 units/ml for 30 min at 37 °C). Lipids were extracted and analyzed using LC/MS/MS monitoring m/z 351.2 → 165.1, as described under "Experimental Procedures." Data are representative of experiments repeated at least three times on different donors (n = 3, mean ± S.E.). ***, p < 0.001 versus thrombin, using ANOVA and Bonferroni post hoc test. Inhibitors used were aspirin, SC-560 (COX-1 selective), or indomethacin/aspirin (non-selective COX inhibition). D, in vivo aspirin blocks platelet DXA3 generation. Lipids were analyzed following thrombin activation of washed platelets, before or after supplementation with 75 mg/day aspirin for 7 days. Data are representative of five independent donors (n = 5, mean ± S.E.); ***, p < 0.001 versus thrombin alone, using ANOVA and Bonferroni post hoc test. E, cPLA2 is required for DXA3 formation. Platelets were incubated with 50 nm cPLA2 inhibitor (cPLA2i) 10 min prior to thrombin activation (0.2 units/ml for 30 min at 37 °C). Lipids were extracted and analyzed using LC/MS/MS monitoring m/z 351.2 → 165.1, as described under "Experimental Procedures." Data are representative of experiments repeated at least three times on different donors (n = 3, mean ± S.E.). ***, p < 0.001 versus thrombin, using ANOVA and Bonferroni post hoc test. F, platelets genetically deficient in cPLA2 do not generate DXA3. Washed human platelets from a patient genetically deficient in cPLA2 or a healthy control were activated using thrombin (0.2 units/ml for 30 min at 37 °C) before lipid extraction and analysis using reverse phase LC/MS/MS, monitoring m/z 351.2 → 165.1 as described under "Experimental Procedures."</p><p>Demonstration of a coordinated signaling pathway leading to DXA3 formation in thrombin-activated platelets, its elevated formation in murine platelets deficient in 12-LOX. A–C, effects of signaling inhibitors on DXA3 formation. A, PP2, 50 μm (src family tyrosine kinase), p38 inhibitor, 100 nm (p38 MAPK), or U-73112, 5 μm PLC. B, wortmannin, 100 nm (PI 3-kinase), Gö 6850 (PKC), 100 nm (PKC), or vehicle (DMSO, 0.5%). C, EGTA 1 mm (extracellular Ca2+) or 1,2-bis(2-aminophenoxy)ethane-N,N,N′,N′-tetraacetic acid tetrakis(acetoxymethyl ester) 10 μm (intracellular Ca2+). D, DXA3 is generated via PAR-1 and PAR-4 receptor stimulation. Washed platelets were activated with a PAR-1 agonist, TFLLR-NH2 (20 μm), and/or a PAR-4 agonist, AY-NH2 (150 μm), for 30 min at 37 °C and then analyzed as described under "Experimental Procedures." ***, p < 0.001 versus control, using ANOVA and Bonferroni post hoc test. E, generation of DXA3 by murine platelets is enhanced in 12-LOX deficiency. Murine platelets were activated using 0.2 units/ml thrombin for 30 min before lipids were extracted and analyzed using LC/MS/MS. F–H, generation of eicosanoids by murine platelets. Murine platelets were activated using 0.2 units/ml thrombin for 30 min before lipids were extracted and analyzed using LC/MS/MS.</p><p>Summary of cellular synthesis pathway for free DXA3 by human platelets. Thrombin triggers platelet activation via PAR1 and PAR4 and then intracellular signaling via Src tyrosine kinases, MEK, MAPK, PLC, and intracellular Ca2+. Activation of cPLA2 leads to AA release, which is then oxidized via COX-1 forming DXA3. Free DXA3 is secreted to interact with neighboring cells, including neutrophils.</p><!><p>To test whether DXA3 could be formed by COX turnover, we examined the incubations of COX-1 or -2 with AA and also synthesized the free acid form of DXA3, through oxidation of 11-HPETE, as described by Porter and co-workers for generating cholesteryl-esterified dioxolane lipids (17, 18). In all reactions, an ion with same retention time and MS/MS spectrum as the platelet lipid was formed (Fig. 7, A–H). However, for either COX or 11-HPETE oxidation-generated DXA3, two additional ions with m/z 351.2 →165.1 were seen eluting just before and after DXA3, which may represent enantiomers or positional isomers, e.g. at C8, C9, or C11. Monocyclic isomers contain three chiral centers, thus eight possible stereoisomers or four pairs of enantiomers): RRR, RRS, RSS, SSS, SSR, SRR, SRS, and RSR (19). These additional ions show identical MS/MS spectra to DXA3 (data not shown) suggesting these peaks to be isomers. With enantiomers not being separated by our chromatography, up to four peaks of isomers would be expected. The absence of these isomers in platelet extracts (Fig. 1C) indicates a higher degree of control over the cellular biosynthesis of DXA3, preventing generation of stereoisomers. This may indicate that additional unknown enzymatic pathways exert control of DXA3 generation in platelets and will be subject to further study.</p><!><p>DXA3 is generated by purified COXs and via oxidation of 11-HPETE. A and B, COX isoforms generate DXA3. 3.5 μg of apo- or holo-COX-1 or COX-2, or hematin (control for reconstituted enzyme), or DMSO (vehicle for hematin) was incubated with 150 μm of AA for 3 min at 37 °C, before lipid extraction and analysis as described under "Experimental Procedures." Levels are expressed as ratio analyte to internal standard/3.5 μg of enzyme generated over 3 min (n = 3, mean ± S.E.). Data are representative of ≥3 separate experiments. C and D, LC/MS/MS of DXA3 formed in vitro via COX-1 or -2. Lipid extracts were separated using reverse phase LC/MS/MS, monitoring m/z 351.2 → 165.1, with reactions as described under "Materials and Methods." E, LC/MS/MS of DXA3 formed in vitro via 11-HPETE oxidation. Purified 11-HPETE was oxidized as described under "Experimental Procedures" and separated using LC/MS/MS. F and G, MS/MS spectra of DXA3 formed in vitro via COX-1 or -2. Lipid extracts were separated as in C and D. MS spectra were acquired at the apex of elution of DXA3. H, MS/MS spectra of DXA3 formed in vitro via 11-HPETE oxidation. Lipid extracts were separated as in C and D. MS spectra were acquired at the apex of elution of DXA3. * shows position of additional isomers with identical MS/MS spectra to DXA3 eluting either before or after lipid.</p><!><p>Platelets generate significant amounts of 11-HETE via COX-1 turnover (20, 21). This likely results from 11-hydroperoxyl radical intermediates (11-LOO•) exiting the catalytic site, then being reduced to form 11-HETE. Similarly, we reasoned that DXA3 could form by COX via rearrangement of an enzyme-generated intermediate exiting the active site early, before full prostanoid ring formation. This could occur either at the 11-LOO• or 9,11-dioxolane radical stage (e.g. just before or after formation of the DX ring). To examine this, we measured DXA3 formation by two COX-2 mutants that generate more 11-HETE and less PGH2 than wild-type enzyme (22). Thus, these enzymes favor escape of lipid radicals prior to DX/prostanoid ring formation. Both mutants were found to generate less DXA3, indicating that the DX ring forms before DXA3 leaves the active site (Fig. 8, A–C, and Scheme 2). Following escape of a 9,11-DX radical, oxygen addition at C8 is expected, followed by peroxidase-dependent reduction. This could be mediated by COX-1 peroxidase or GSH peroxidase. In support, DXA3 formation was inhibited by 1–10 mm iodoacetate, a thiol-alkylating reagent (Fig. 8D).</p><!><p>DXA3 exits the COX active site downstream of 11-LOO• radical escape, although peroxidase activity but not thromboxane synthase is involved in platelet DXA3 generation. A–C, COX-2 mutants that generate more 11-HETE form less DXA3 during turnover. DXA3, PGE2, and 11-HETE generated by COX-2 wild type (WT) or mutants (V349A and W387F). Following reconstitution with hematin, 30 μm arachidonate was oxidized using 10.2 μg of enzyme at 37 °C for 5 min under O2 atmosphere. n = 5–7, ***, p < 0.05 (single factor ANOVA followed by two-tailed t test). D, peroxidase turnover is required for DXA3 generation. Platelets were treated with 1–10 mm iodoacetate before thrombin activation and analysis of DXA3 using LC/MS as described under "Experimental Procedures." One representative donor, triplicates ± S.E., single factor ANOVA followed by Bonferroni ***, p < 0.005 are shown. E–G, thromboxane synthase is not involved in platelet DXA3 generation. Platelets were incubated with 50 μm picotamide 10 min prior to thrombin activation (0.2 units/ml for 60 min at 37 °C). Lipids were extracted and analyzed using LC/MS/MS monitoring m/z 351.2 → 165.1, as described under "Experimental Procedures." Data are representative of experiments repeated at least three times on different donors (n = 5, mean ± S.E.). ***, p < 0.001 versus thrombin, using ANOVA and Bonferroni post hoc test. NS, not significant.</p><p>Proposed mechanism of DXA3 formation by COX. During COX turnover, a dioxolane ring forms between C9 and C11, prior to prostanoid ring formation, resulting in a carbon-centered radical at C8. Leakage of this lipid intermediate from the active site, then addition of oxygen followed by reduction to LOOH, and then LOH leads to formation of DXA3.</p><!><p>To determine the role of enzymatic activities downstream of COX-1, an inhibitor of thromboxane synthase was added to platelets during activation. Picotamide led to inhibition of TXB2 generation and a corresponding elevation in PGE2, because less PGH2 was being converted by thromboxane synthase (Fig. 8, E and F). However, DXA3 formation was unaffected (Fig. 8G). Last, there was no correlation between TXB2 and DXA3 levels, further supporting the idea that thromboxane synthase is not involved in DXA3 generation (data not shown).</p><!><p>To determine generation of DXA3 in other cell types, RAW 264 macrophages were treated using LPS for 24 h, with/without ionophore activation. Under basal conditions, these cells express only COX-1, although following LPS treatment, they up-regulate COX-2. We found that cells required ionophore for robust PG generation. DXA3 formation paralleled that of PGD2, being present basally, but unaffected by inflammatory activation. Thus, the lipid was most likely generated by COX-1 but not COX-2 in these cells (Fig. 9, A–D). Human blood was harvested and allowed to clot. Analysis of serum demonstrated a significant DXA3 peak, indicating that physiological coagulation forms this lipid (Fig. 8E). In contrast, DXA3 was absent from plasma (data not shown).</p><!><p>DXA3 is generated by RAW cells and during physiological coagulation. A–D, RAW cells generate DXA3 under basal non-inflammatory conditions. RAW cells were incubated in serum-free DMEM for 1 h at 37 °C, 5% CO2. Where used, 200 ng/ml LPS was added for 24 h. Cells (8 × 106 ml−1) were activated using 10 μm A23187 at 37 °C for 10 min, and lipids were extracted and analyzed using LC/MS/MS. DXA3 was monitored using m/z 351.2 to 165.1 and PGE2/D2 using m/z 351.2 to 271.1 utilizing a 4000 QTrap. A and B, LC/MS/MS of basal RAW cells with/without 10 μm A23187. C and D. LC/MS/MS of LPS-treated RAW cells with/without 10 μm A23187. E, DXA3 is generated during physiological blood clotting. Whole blood was clotted, serum was harvested, and lipid was extracted as described under "Experimental Procedures." LC/MS/MS was performed as for free DXA3 on a Q-Trap platform. Note that retention time of serum and RAW cell DXA3 differs slightly because these were analyzed several months apart on different columns. The identities have been confirmed through co-elution with platelet DXA3 (data not shown).</p><!><p>Herein, we used a lipidomic approach to identify and characterize a new neutrophil-activating lipid, proposed to be DXA3, formed endogenously by agonist-activated platelets in a COX-1-dependent manner, by a macrophage cell line, and during blood clotting. At this time, we present a proposed structure based on strong and consistent UV, GC/MS, and LC/MSn data. Once the sufficient synthetic standard is available, full NMR analysis will be undertaken. We note that many other biologically relevant lipids, including thromboxane, leukotrienes, protectins, etc., were first published as proposed structures in a similar manner to our study.</p><p>The mechanism of DXA3 formation in vitro by COX enzymes is described, as well as its detailed cellular biosynthesis pathway in human platelets. DXA3 represents the first DX eicosanoid isolated and characterized within cells. To date, these have only been demonstrated to form via chemical oxidation of purified arachidonate esters or ω3 fatty acids or by in vitro lipoxygenase oxidation of epoxides, and neither their generation by cells nor any bioactivities have been described (14, 18, 19, 24, 25, 27, 28). Our study greatly extends these old in vitro observations by demonstrating that DX lipids are not only generated by live primary cells under physiological conditions, but they possess biological activity of relevance to innate immunity. This study places this eicosanoid in a new family of products likely relevant as a lipid mediator as are the prostaglandins, leukotrienes, and P450-derived eicosanoids. Extending these cell biology studies to in vivo measurements of leukocyte function and inflammation will be undertaken as soon as the synthetic standard becomes available.</p><p>Eicosanoids are essential lipid signaling mediators involved in diverse biological processes (29–32). Identification of new bioactive eicosanoids from this pathway could pave the way for additional and more selective therapeutic approaches. Thus, the proposed structure for DXA3 represents a new member of this family, characterized by a unique five-membered endoperoxide ring, and generated by a COX isoform known to play important roles in vascular disease and, more recently, in cancer.</p><p>Mac-1 (CD11b/CD18) is the predominant β2 integrin on neutrophils that mediates adhesion-dependent processes, such as binding to the endothelium or phagocytosis, recruitment, and transendothelial migration (33, 34). Herein, we show that DXA3 enhances Mac-1 on the cell surface (Fig. 3). The only other known Mac-1-inducing eicosanoids are leukotriene B4 and 5-oxo-ETE, both neutrophil-derived lipids (35, 36). Thus, neutrophil integrin activation by platelet-derived DXs could be of relevance during acute inflammation and infection. DXA3 was generated by platelets utilizing endogenous substrate in nanogram amounts that are ∼10-fold higher than platelet PGE2 (Fig. 4E). Its formation does not require supply of exogenous substrates and can be triggered directly by pathophysiological agonists in healthy primary cells, both important criteria in establishing that a new lipid mediator is endogenously relevant.</p><p>As DXA3 was generated via COX-1 in platelets, we reasoned that it could form through two potential mechanisms, either (i) rearrangement of 11-LOO•, known to be released by the enzyme during turnover, or (ii) that the dioxolane ring could form before the lipid exits from the active site (20, 21, 23, 37). In both cases, attack at C9 by the peroxyl radical would form the 9,11-dioxolane, which would be followed by oxygen addition at C8, and finally peroxidase reduction of the resulting LOOH by COX-1 peroxidase or GSH peroxidase in platelets. Our data using mutant COX-2 enzymes that generate less DXA3 but more 11-HETE suggest that the DX ring forms before lipid release by the enzyme. Thus, dioxolane ring formation occurs first and before prostanoid ring closure between C8 and C12 (Scheme 2). Finally, given that COX-1 generates 11R-HETE, we postulate that the dioxolane ring will likely be 9S,11R. Our observation of a single DX isomer in platelets but several in purified enzyme reactions indicates that platelets exert additional control over its biosynthesis. This may be at the stage of oxygen insertion into the chiral center at C8.</p><p>DXA3 was generated by platelets via a highly coordinated sequence of signaling events, including PAR-1 and -4, src tyrosine kinases, intracellular Ca2+, cPLA2, PLC, p38, and MAPK. This indicates tight control of its formation, similar to generation of other COX metabolites, such as TXA2. The signaling pathway is distinct from generation of free and esterified HETE and hydroxydocosahexadienoic acids, which form via 12-LOX, and require extracellular calcium, independent of PLC and MAPK (6, 26).</p><p>DXA3 was also generated by RAW cells as a single isomer, similar to platelets. Our preliminary data suggest that it originates primarily from COX-1 in these cells. In contrast, we found that either isoform could generate the lipid in vitro. In line with our observation that cellular DXA3 is a single isomer in platelets and RAW cells, although three isomers form via COXs in vitro, this collectively suggests that cellular DXA3 generation is under enzymatic control downstream of its synthesis by COX-1. Future studies will examine the ability of cellular COX-2 to generate the isoform and under which activation conditions. COX-1 is important not only in acute innate immunity but also in gastric function and development, and thus its generation by this isoform may have wider implications for eicosanoid biology in other organs.</p><p>Murine platelets also generated DXA3, and levels of this were enhanced in cells deficient in 12-lipoxygenase. This may be related to greater availability of substrate, although this has not been explored herein.</p><p>Eicosanoids include a large number of related structures formed via oxidation of arachidonate, following its release from intracellular membranes by phospholipases. A rapid burst of eicosanoid generation is a key event during cell activation and is stimulated during innate immunity by bacterial products, growth factors, cytokines, thrombin, and collagen. Most known eicosanoids from COXs were identified and structurally characterized in the 1980–1990s and include platelet-derived lipids, TXA2 and 12-HETE, as well as the PGs, exemplified by PGE2, and D2, well known as mediators of pain, fever, cell proliferation, and innate and adaptive immune responses. Our observation of a cellularly generated DX eicosanoid defines a new class of these lipids formed endogenously by mammalian cells. More members of this class are possible, given recent observations of purified LOXs being able to generate DX isomers via oxidation of epoxides in vitro in acellular experiments (14).</p><!><p>C. H., M. A., C. U., S. A., D. A. S., S. N. L., K. A. R., and C. P. T. conducted the experiments. C. H., P. W. C., M. A., V. O. D., R. C. M., and C. P. T. designed the experiments. L. J. M., H,. J. L., and T. D. W. provided reagents or patient samples. C. H. and V. O. D. wrote the paper. All authors edited the paper.</p><!><p>This work was supported in part by Wellcome Trust 94143/Z/10/Z, British Heart Foundation Grant RG/12/11/29815 (to V. B. O. and P. W. C.), and National Institutes of Health Grant U54HL117798 (to R. C. M.). The authors declare that they have no conflicts of interest with the contents of this article. The content is solely the responsibility of the authors and does not necessarily represent the official views of the National Institutes of Health.</p><p>This article contains Results and supplemental Figs. S1–S3.</p><p>thromboxane A2</p><p>dioxolane</p><p>formylmethionylleucylphenylalanine</p><p>dioxolane A3</p><p>protease-activated receptor</p><p>cytosolic phospholipase A2</p><p>hydroxyeicosatetraenoic acid</p><p>prostaglandin</p><p>N-methyl benzohydroxamic acid</p><p>analysis of variance</p><p>hydroperoxyeicosatetraenoic acid</p><p>Fourier MS</p><p>collision-induced dissociation</p><p>arachidonate</p><p>phospholipase C</p><p>lipoxygenase.</p>

PubMed Open Access

A molecular pathway for the egress of ammonia produced by nitrogenase

Nitrogenase converts N 2 to NH 3 , at one face of an Fe-Mo-S cluster (FeMo-co) buried in the protein. Through exploration of cavities in the structures of nitrogenase proteins, a pathway for the egress of ammonia from its generation site to the external medium is proposed. This pathway is conserved in the three species Azotobacter vinelandii, Klebsiella pneumoniae and Clostridium pasteurianum. A molecular mechanism for the translocation of NH 3 by skipping through a sequence of hydrogen bonds involving eleven water molecules and surrounding aminoacids has been developed. The putative mechanism requires movement aside of some water molecules by up to , 1A ˚. Consistent with this, the surrounding protein is comprised of different chains and has little secondary structure: protein fluctuations are part of the mechanism. This NH 3 pathway is well separated from the water chain and embedded proton wire that have been proposed for serial supply of protons to FeMo-co. Verification procedures are suggested.

a_molecular_pathway_for_the_egress_of_ammonia_produced_by_nitrogenase

<!>Results<!>Strategy.<!>Discussion<!>Methods

<p>T he enzyme nitrogenase converts dinitrogen to ammonia N 2 to NH 3 , with some reduction of protons to dihydrogen: N 2 1 6H 1 1 6e 2 R 2NH 3 ; 2H 1 1 2e 2 R H 2 . There is considerable understanding of the biochemical processes of this enzyme. The two proteins that comprise nitrogenase are the MoFe protein in which the chemical catalysis occurs at the iron-molybdenum cofactor (FeMo-co), and the Fe protein where the reducing potential is generated. The catalytic site, FeMo-co, is an unprecedented CFe 7 MoS 9 cluster (Fig. 1(a)) containing obligatory homocitrate (HCA, 2-hydroxybutane-1,2,4-tricarboxylate). Also in the MoFe protein is a redox-variable Fe 8 S 7 cluster called the P-cluster, approximately 12A ˚from FeMo-co, and understood to mediate electron transfer to FeMo-co. The Fe protein contains an Fe 4 S 4 cluster and the nucleotide (MgATP or MgADP) binding sites. The biochemical cycle involves docking of the Fe protein with the MoFe protein, hydrolysis of 2MgATP, and transfer of one electron via the Fe 4 S 4 cluster at the docking interface to the P-cluster and then to FeMo-co [1][2][3][4][5][6][7][8][9] . This biochemical cycle of association, hydrolysis, electron transfer, dissociation is repeated eight times during reduction of N 2 to 2NH 3 with concomitant formation of H 2 .</p><p>The understanding of mechanism at the chemical level is poorer, in part due to experimental difficulties in trapping and characterising intermediates. However, investigations of the reactivities of proteins with modified residues have revealed the locus of chemical catalysis, the Fe2,Fe3,Fe6,Fe7, face of FeMo-co, and particularly the Fe2-S2B-Fe6 region directly underneath the side chain of a-70 . The surrounding residues a-195 His , a-191 Gln , a-69 Gly and a-96 Arg when modified also influence the reactivity of nitrogenase 6,8,[13][14][15][16][17][18][19][20][21][22] .</p><p>A series of density functional investigations [23][24][25][26][27] has led to a detailed 21-step mechanism for the conversion of N 2 to NH 3 28,29 . A key aspect of this mechanism is the serial formation, migration, and accumulation of multiple H atoms on FeMo-co, and their sequential transfer to bound substrates and intermediates to generate hydrogenated products 25,30 . Coupled electron transfer and proton transfer to FeMo-co are proposed to generate the H atoms at S3B. A requirement of this general intramolecular hydrogenation mechanism for nitrogenase is a controlled pathway for serial provision of external protons to S3B, and in a recent paper I described the chain of water molecules that leads from the surface of the protein to S3B, Fig. 1(b). A Grotthuss mechanism for serial translocation of protons along the fully conserved single chain of hydrogen bonded water molecules W8 to W1 (see Fig. 1(b)) to S3B was developed 31 . This inner section is a tight proton wire: the outer section is variable and is postulated to function as a proton bay.</p><p>In this paper I report explorations for the pathway used for diffusion of product NH 3 /NH 4 1 from the reaction site to the surface of the protein. This product egress aspect of the activity of nitrogenase has been largely ignored. Durrant 32 recognized the water chain that transports protons (Fig. 1(b)) and suggested that it could be used for transport of ammonia, but ammonia and protons cannot travel in opposite directions along the same water channel. The eight water molecules comprising the inner section of the proton wire are ideally hydrogen bonded to each other and to the enclosing homocitrate and key residues, in order to facilitate and control the Grotthuss proton transfer. Intrusion or substitution of NH 3 here, and creation of NH 4</p><p>1 , would block proton transfer to FeMo-co (by the formation of too many hydrogen bonds to allow the Grotthuss proton hopping) and in the same manner would freeze translocation of ammonia. It is clear that separate pathways are needed for the proton supply to FeMo-co and the efflux of NH 3 . In the structures of MoFe proteins there is a collection of water molecules around homocitrate 2 , and it has been suggested that this water pool could absorb product ammonia 31 . However this does not account for transport of product ammonia to the protein surface, which is the question addressed here. The translocation of ammonia through other proteins has been investigated in detail, principally for the ammonium transporter (Amt) proteins which control movement of NH 3 /NH 4 1 across cell membranes in bacteria, archaea, fungi, plants and animals [33][34][35][36] , and for a collection of glutamine amidotransferase (GAT) enzymes. These latter enzymes utilise NH 3 , derived from the hydrolysis of glutamine, in a synthase reaction. They transport the NH 3 from the generation site to a distant synthase domain through intramolecular tunnels protected from the environment [37][38][39][40][41][42] . The consensus interpretation, based on structural information, targeted mutations, molecular dynamics simulations, and calculation of intra-protein pK a , is that NH 3 rather than NH 4</p><p>1 is the moving entity in these glutamine amidotransferases 37,[43][44][45][46][47] . However, the Amt proteins (and related human Rhesus proteins 48,49 ) have different functions, including transport of NH 4</p><p>1 against a concentration gradient and discrimination against K 1 . Three mechanisms are under consideration: transport of NH 3 , transport of NH , and co-transport of NH 3 and H 1 33-35,50-59 .</p><!><p>This investigation is based on detailed exploration of the structures of crystals containing the MoFe protein, from different species and in various trapped states, and including the independent occurrences of FeMo-co and surrounds in each protein. Unless stated otherwise, amino-acid numbering is that of the Azotobacter vinelandii (Av) MoFe protein, and water molecules are labeled as in the highest resolution structure, PDB 3U7Q 60 . Residue labels contain letters A, B. to denote the chain in the PDB file. In the following the term 'water path' is reserved for the collection shown in Fig. 1 (b), and the term 'water pool' refers to the different collection of water molecules around homocitrate (Fig. 1(a)).</p><p>Generation of NH 3 . Product NH 3 molecules are expected to be generated close to the reaction face of FeMo-co. In the complete mechanism proposed for conversion of N 2 to NH 3 28 , the NH 3 molecules are formed (late in the catalytic cycle) at atoms Fe6 and Fe2 of FeMo-co. The first NH 3 is generated at the exo position of Fe6, from which it dissociates in the exo direction. The second NH 3 is generated at the endo position of Fe2, and dissociates in an endo direction, looping around the endo side of Fe6. Residue Val 70A and its neighbours in the a-helix that covers the face of FeMo-co constrain both product NH 3 molecules to leave in the exo direction from Fe6. This is illustrated in Fig. 1(a) that shows the first NH 3 leaving Fe6, and the nascent second NH 3 as an NH 2 group bridging Fe2 and Fe6. The formation of the second NH 3 is completed by H atom transfer from S3B, and after dissociation it is proposed to follow the path of the first NH 3 . Fig. 1(a) also locates the hydrogen bonded water molecules that comprise the water pool, clustering around the uncoordinated carboxylate atoms O1, O2 and O4 of homocitrate. The separate water chain that supports the proton wire to S3B of FeMo-co uses different homocitrate atoms O5, O6 and O3 (Fig. 1(b)). Residue Gln 191A , hydrogen bonded to O1 of homocitrate, lies approximately in the path of the departing NH 3 molecules, whereas His 195A is out of the way.</p><!><p>The search for possible channels and pathways was undertaken first by a search for cavities in the MoFe protein in the vicinity of FeMo-co and the water pool. The species generality of the cavities so located was checked, together with the generality of their occurrence in the best available crystal structures. The most promising cavities near the NH 3 generation position were examined for size and extent, water content, and immediate surroundings. Then, in the one feasible cavity, possible locations for a moving NH 3 molecule were modeled, with hydrogen bonding to water and protein surrounds. Finally, the most probable pathway was assessed in terms of protein tertiary structure and likely dynamic fluctuations during NH 3 egress.</p><p>Calculations of pockets in the protein. The initial search for cavities involved calculations of pockets in the protein, using the program fpocket 61 . Pockets are clusters of intersecting alpha spheres. An alpha sphere contacts four atoms on its boundary and contains no internal atom: the four atoms are equally distant from the sphere centre, defining the radius of the sphere. The results reported used alpha molecule is dissociating from Fe6, the second is to be formed from NH 2 bridging Fe2 and Fe6 (in this and some following figures Fe2 and Fe6 are coloured differently). (b) The chain of hydrogen bonded water molecules from the protein surface to S3B of FeMo-co. The proton wire section W8 to W1 is fully conserved, while the branched proton bay section and the path to the surface are variable.</p><p>spheres with a minimum radius of 2.8A ˚, and required at least 35 intersecting alpha spheres to constitute a pocket. Water molecules are ignored in the calculation of pockets, but were included in the results, in order to examine the presence of water in or near the pockets. Thus the pocket calculations yielded domains in the protein sufficiently large to contain NH 3 or NH 4</p><p>1 , and possibly containing water.</p><p>Calculations were made for all of the high-resolution crystal structures (PDB 3U7Q 60 , 1M1N 62 , 2MIN 63 , 3MIN 63 , 2AFH 64 , 2AFI 64 , 1MIO 65 , 1QGU 66 , 1QH1 66 , 1QH8 66 , 1H1L 67 ) that include the MoFe protein and a high proportion of resolved water molecules, and encompassed the species Azotobacter vinelandii (Av), Klebsiella pneumoniae (Kp) and Clostridium pasteurianum (Cp).</p><p>Figure 2 shows the significant pockets (drawn as aggregates of alpha spheres) found in the vicinity of FeMo-co for the Av protein (PBD 3U7Q). The view direction in this figure is directly along the pseudo-threefold axis of FeMo-co, from the top. Two protein chains that envelop the pockets and cofactors are coloured green and red. In this and the following descriptions the pockets are labeled by their colours. The silver pocket contains the water molecules of the water pool, while the water path that supports the proton wire appears as the orange pocket extending upwards from FeMo-co. The key pocket is that coloured cyan, which extends from an edge of the silver pocket, twisting towards the surface of the protein. The inner section of the cyan pocket is very similar for the Av and Kp proteins, but variation occurs as the pocket reaches the protein surface. The gap between the inside edge of the cyan pocket and the silver pocket is devoid of protein main chain, and the point of approach of the cyan and silver pockets is essentially unchanged for all of the protein crystals investigated. The lime pocket extends from the P-cluster, and has variable shape and size. In some protein structures it is larger than shown in Fig. 2, extending to the protein surface near where the cyan pocket meets the surface. The lime pocket is not near the water pool, and is separated from it by a section of polypeptide chain, and therefore the lime pocket is not believed to play any role in the movement of NH 3 , and it will not be considered further. The only other pocket near the silver pocket is that coloured mauve: this pocket has variable size and shape in different proteins, and is separated from the silver pocket by a gap of 6.4A ˚that is blocked by the sidechain of Lys426 (Av, Lys424 in Kp).</p><p>These pocket calculations focus attention on the cyan pocket as the most likely domain for NH 3 efflux. Before proceeding with more detailed investigation of this silver-plus-cyan domain of the protein, two further checks were made. The first involved the location of the region where the Fe protein docks with the MoFe protein during the electron transfer cycle, in relation to the postulated exit point for NH 3 on the surface of the MoFe protein. This was to determine whether diffusion of NH 3 from the surface of the MoFe protein might be blocked when the Fe protein is docked. Crystal structures 2AFH and 2AFI (Av) contain the Fe protein docked with the MoFe protein, using two different but adjacent docking domains, dependent on the state of the nucleotide in the Fe protein 64 . In both structures, the surface terminus of the cyan pocket is not part of the docking domain, and does not encounter the polypeptide chains of the Fe protein (supplementary Fig. S1).</p><p>The second check involved the relationship between the water pool postulated for NH 3 movement away from FeMo-co and the chain of water molecules that constitutes the proton wire. Figure 3 shows the relevant part of the protein, around and 'underneath' homocitrate. The conserved chain of hydrogen-bonded water molecules comprising the proton wire (terminating with a hydrogen bond to S3B) is depicted with orange spheres, while water molecules extending from the other side of homocitrate towards the alpha spheres of the cyan pocket are coloured brown. There is an additional cluster of water molecules, coloured magenta, directly under homocitrate. The composition and structure of this water cluster varies in the different crystal structures, and the geometry is such that not all of the hydrogen bond distances within this cluster could be hydrogen bonds. This cluster is hydrogen bonded to one 'brown' water molecule (786B) and is not hydrogen bonded directly to the proton wire. The side-chain of arginine 105B is a barrier between the magenta cluster and the proton wire, and, as previously described, is a crucial part of the proton wire 31 .</p><p>It is clear from Fig. 3 that the putative ammonia path into the cyan pocket and the proton path occur in opposite directions through the protein, as required chemically. The water pool near homocitrate is comprised of the 'brown' and 'magenta' water molecules, not directly linked to the proton wire. The magenta water cluster does not appear to have a functional role other than as a small compact reservoir for water molecules.</p><p>Contents and surrounds of the postulated NH 3 exit pathway. The next step was to examine closely the contents and immediate surroundings of the indicated ammonia pathway towards and through the cyan pocket. The blue arrow in Fig. 4(a) shows the proposed general direction of movement from the vicinity of Fe6 (cf Fig. 1(a)), over homocitrate and associated water molecules in the water pool, and towards the water molecules at the entrance of the cyan pocket. The relevant water molecules close to O1, O2 and O4 of homocitrate are 573A, 512A, 585A, 604B and 674B, with hydrogen bond connections to amino acids, Gln 191A , Glu 380A , Lys 426A , Ile 59A and Gly 94B (all labels here and following are those of the Av protein, PDB 3U7Q). The gap between these water molecules and the entrance to the cyan pocket (the gap between silver and cyan pockets in Fig. 2) is surrounded by but not blocked by Lys 426A , Ile 59A and Gly 94B .</p><p>The cyan pocket (of the Av protein) contains a sequence of eleven hydrogen bonded water molecules, extending roughly in an arc. Around these there are interactions with Met 57A , Thr 58A , Arg 60A , Gln 53A , Asp 403A , Ser 115B and Ser 117B , marked in Fig. 4(b).</p><p>It is postulated that NH 3 egress starts from Fe2/Fe6 on FeMo-co, following the direction of the blue arrow (Fig. 4(a)) and probably engaging with some of the hydrogen bonding components O1, O2, O4 of homocitrate, the five associated water molecules, and sidechains of Gln 191A , Glu 380A , Lys 426A . At this point the NH 3 molecule would be positioned at the entrance of the cyan cavity, ready to move to the end of this cavity, engaging with the contained water and surrounding residues. If this hypothesis is correct, these features should be conserved in the various protein structures, and therefore details of these components and surroundings have been compared for the highest resolution crystal structures for Av (3U7Q) and Kp (1QH1). Fig. 5, a comparison of the hydrogen bonding connectivities, demonstrates that the two proteins are very similar. The domains in this depiction, left to right, are first the water pool and its homocitrate components, then a gap to the sequence of water molecules and protein surrounds in the vicinity of the cyan cavity. Functions from the A and B chains are differentiated by colour.</p><p>The similarity of the two proteins is evident. The principal difference is the absence in Kp of a water molecule in the same position as water 640A in Av, breaking the chain of hydrogen bonded water molecules, and changing some hydrogen bonding interactions with surrounding protein. This water is halfway through the cyan pocket (and more than 15A ˚from FeMo-co), and is at a point where the pathway possibilities for NH 3 are diverging (see next section). In the Kp protein the hydroxyl side chain of Ser 51A is near the position of water 640A in Av, and is disordered, which is indicative of movements that could be part of NH 3 passage here. These differences in the later section of the cyan pocket are consistent with the notion that as NH 3 approaches the protein surface its mechanism for molecular translocation is less controlled.</p><p>Figure 5 also draws attention to the differentiation of main chain atoms (boxed) and side-chain groups in the hydrogen-bonding network, and the prevalent involvement of main chain carbonyl functions, as potential hydrogen bond acceptors. Residue Ile 59A in Av is Val 58A in Kp, but since it is the main chain carbonyl O that participates in the hydrogen bonding this difference of hydrophobic side chains is inconsequential. There are differences in the region of waters 674B (3U7Q) and 25 (1QH1), and their hydrogen bonding with Lys 426 /Lys 424 , but this is not significant because these functions are distant from the detailed pathway described in the next section.</p><p>Atomic level mechanism for NH 3 movement. The preceding analysis has revealed a possible pathway for movement of NH 3 away from FeMo-co towards the protein surface. The questions now are (a) whether there is space for this movement of NH 3 , (b) whether the environment is consistent with the hydrogen bonding expected of NH 3 , (c) what impediments might exist, and (d) what fluctuations in the tertiary structure of the protein are likely to be involved. Possible pathways for molecular NH 3 to move from FeMoco to and through the cyan cavity towards the surface of the protein were investigated by manual molecular modelling.</p><p>Since water and NH 3 are similar in size and hydrogen bonding attributes, one question is whether NH 3 might occupy the locations that are observed to contain water in the resting protein structures.</p><p>[It is very unlikely that the crystallographically observed small 1 ) because the crystallisation medium for structure 3U7Q did not contain NH 3 /NH 4 1 buffers 60 .] Alternatively, is there space for NH 3 to move around the water molecules in the resting protein? Would it be necessary for water molecules to move aside in order to allow passage of NH 3 ?</p><p>The moving entity was modeled as NH 3 , because NH 3 is generated at the surface of FeMo-co and is released into a hydrophobic anhydrous space. On diffusion into the water pool and beyond it is possible that NH 4</p><p>1 is formed. In the early stages of the path there is no evident source of a replenishable proton, and so NH 4</p><p>1 formation would be just a temporary relocation of one hydrogen atom across a hydrogen bond, and the following modeling of hydrogen bonds would be unchanged. Near the protein surface the more aqueous medium would generate NH 4</p><p>1 .</p><p>Standard geometries were used for the three donor and one acceptor hydrogen bonds of NH 3 . The goal was to find a sequence of locations that engaged some or all of these possible hydrogen bonds, such that NH 3 could swing from one location to the next, breaking and forming hydrogen bonds. The starting point had NH 3 just dissociated from Fe6 (Fe6-NH 3 5 2.7A ˚), and NH 3 hydrogen bonded to both S2B and O1 of HCA. It soon became evident that NH 3 can only move away underneath the side chain of Gln 191A into the water pool. Through this first stage of NH 3 movement the important surroundings are O1, O2 and O4 of HCA, the side-chain CH 3 of Ala 65A , and both the terminal amide group and the CH 2 functions of the side chain of Gln 191A . In the resting state the terminal NH 2 of Gln 191A is well positioned to donate hydrogen bonds to O1 of HCA and to the side chain carboxylate of Glu 380A . During this stage it is probable that the side chain amide group of Gln 191A relocates and reconfigures in order the improve the hydrogen bonding with passing NH 3 , at the expense of the hydrogen bond to O1 and OCO of Glu 380A .</p><p>Figure 6 shows the first six possible positions for NH 3 . Position 1 has N-H hydrogen bonds to S2B and O1, then at position 2 the N-H hydrogen bonds are bent to O1 and O2 of HCA, then in position 3 breaking with O1 and in position 4 forming a new N-H hydrogen bond to water 585A. In position 5 the N-H hydrogen bond to water 585A is retained and the other N-H hydrogen bond bifurcates to O1 and O2, with a new hydrogen bond to water 512A. Finally, in position 6 the NH 3 swings around its N-H hydrogen bond to O1, and forms a new hydrogen bond to water 573A. Several comments are warranted. First, the NH 3 trajectory is a downwardly spiraling arc maintained by N-H hydrogen bonds to O1 and/or O2, and this carboxylate arm of HCA is regarded as a key controller of mechanism. Second, the side chain of Gln 191A is closely involved, and with reconfiguration could form Gln 191 -NH 2 R NH 3 hydrogen bonds at positions 3, 5 and 6: these potential hydrogen bonds are marked on Fig. 6. Third, water molecules are used as hydrogen bond acceptors at positions 4, 5 and 6, with water molecules 512A and 573A moving by ca 0.7A ˚away from the approaching NH 3 molecule. Sidechain b-CH 3 of Ala 65A and c-CH 2 of Gln 191A provide steric boundaries.</p><p>This modeling was continued, stepping the hydrogen bonds of NH 3 , to identify a total of 18 possible locations for NH 3 towards the protein surface. These positions are shown in Fig. 7, together with the water molecules and significant amino-acid functions. Detail is provided in Supplementary Figs. S2 and S3. In the journey from the water pool to the cyan cavity NH 3 passes between the sidechain of Lys 426A and the main-chain oxygen atom of Ile 59A . The terminal NH 2 function of Lys 426A is a hydrogen bond accepting pivot point, and because the distances from it are slightly short it is proposed that this side-chain function moves slightly away during NH 3 passage. The main-chain oxygen atoms of neighbouring residues Thr 58A , Ile 59A and Arg 60A function as hydrogen bond acceptors in positions 8 through 12. The side chain CH 3 function of Thr 58A is significant, because the lone pairs of NH 3 in positions 12 through 15 are distant 2.6-2.9A ˚from it, with the potential for formation of weak C-H R NH 3 hydrogen bonds. On the other side of this arced sequence of NH 3 positions the side chains of Ser 115 and Ser 117 in chain B are involved. Beyond the last modeled position, near Gln 53A , there is a 'river-delta' of possibilities for NH 3 to escape to the protein surface. Examination of the atom thermal parameters for 3U7Q and 1GQU reveals that above average values occur in the sections of chains A and B extending from the end of the NH 3 pathway to the protein surface.</p><p>In positions through the cyan cavity domain, N-H R OH 2 hydrogen bonds are part of the model. The ability of these water molecules to accept one such hydrogen bond at each position of NH 3 is favourable in terms of hydrogen atom accounting, because each water can form two other donor O-H R O hydrogen bonds, one to the next water in the chain, and one to protein main chain carbonyl oxygen. Displacements of water molecules by 0.4-1.0A ˚are involved in the model, and it is clear that some local geometric reorganisation would be required as NH 3 passed along this pathway.</p><p>This leads to considerations of secondary and tertiary structure, and structural fluctuations during NH 3 passage. Figure 7 portrays this structure of the protein, and reveals the lack of secondary structure on the viewer side of the proposed path. The path winds between the A and B chains (see also Fig. 5), with the B chain occurring as a structured backdrop and the A chain as a movable front curtain. Figure 8 is a different view of key structural elements in the surrounding protein. Three different sections of the A chain are involved. The part of chain A that lies over the reaction face and contains Val 70A is a-helical only as far as residue 62, but then is largely unstructured along the ammonia path to residue 52 (green). The section containing influential residues His 195A and Gln 191A (grey) is a-helical only near FeMo-co. Glu 380A and Lys 426A , both near the ammonia path, are connected by a long loop (yellow). Chain B (red) is near the proposed ammonia path at residues 93 and 115-117. The chain from 93 to 108 is a-helical, directed well away from the ammonia pathway, then loops back to serines 115 and 117.</p><!><p>This exploration of the structures of nitrogenase proteins has uncovered a pathway for the movement of product NH 3 from its site of formation on FeMo-co to the surface of the protein, and a molecular mechanism for the translocation of NH 3 has been developed. This pathway is conserved in proteins from species Av, Kp and Cp. The integrity of the acid-base dimension of the physiological reaction of nitrogenase, consuming acid and generating base, is maintained, because the NH 3 egress pathway is separate from and physically opposite to the proposed proton influx pathway. As in the glutamine amidotransferase enzymes, the moving entity in the early stages of the efflux is probably NH 3 , becoming NH 4</p><p>1 near the protein surface. Water molecules occur along most of the NH 3 pathway, at least in the resting protein as crystallised, and provide acceptor and donor hydrogen bonding sites for NH 3 . Polypeptide main chain atoms also provide hydrogen bonding sites, and the suggested mechanism is essentially a sequence of interchanges of hydrogen bonds over short distances. The mechanism does not involve a flow of water molecules accompanying NH 3 , but it does involve movements of some water molecules away from the main path by distances of up to 1A ˚, to create sufficient space for NH 3 to skip from one hydrogen bond to the next.</p><p>At this point it is appropriate to make comparisons with the wellstudied ammonia channels in glutamine amidotransferase (GAT) enzymes. In all of these, ammonia generated by hydrolysis of glutamine (or asparagine) in the amidase domain is translocated through tunnels to a distant synthase site: the ammonia is shielded from the medium and maintained as NH 3 because NH 3 is needed as nucleophile in the synthase domain. The GAT enzymes are diverse in product, function and structure, and include carbamoyl phosphate synthetase 68 , glutamine phosphoribosylpyrophosphate amidotransferase (GPAT) 46,69 , glutamate synthase (GltS) 39 , tRNA-dependent amidotransferases (or GatCAB). The ammonia tunnels are structured more by the residues of the synthase domain than those of the amidase domain, accounting for the variety of tunnel architectures 39,40,70 . In nitrogenase the formation of NH 3 at FeMo-co is analogous to the formation of NH 3 at the amidase site, but there is no requirement to prevent formation of NH 4</p><p>1 near the end of the pathway at the surface of the MoFe protein. Nevertheless, there are results and interpretations from the amidotransferase enzymes that can inform further investigation and understanding of NH 3 efflux from nitrogenase.</p><p>In crystalline GatCAB amidotransferases a 30-35A ˚-long hydrophilic tunnel links the amidase and synthetase sites. The tunnel is ungated along its length, and contains 18 water molecules that are hydrogen bonded to each other or to surrounding polar residues. The published representations 71,72 show a structure quite similar to that described here for nitrogenase. Molecular dynamics calculations have revealed gates and valves modulating the direction of NH 3 transport 47 , and confirmed large barriers associated with NH 4 1 rather than NH 3 as the moving entity. In glutamate synthase the tunnel is composed mainly of backbone atoms of hydrophilic residues, and water molecules fill the tunnel space 39 . There are some structural similarities between the various sections of the long intramolecular ammonia translocation tunnel in carbamoyl phosphate synthetase (CPS) 68 and the path proposed here for nitrogenase. Molecular dynamics simulations of CPS reveal hydrogen bond stepping, movements of side-chains, and involvement of small water clusters. It is calculated that in the first stage of a long tunnel, NH 3 is transported over a distance of ,13A ˚with a maximum barrier of 6 kcal mol 21 44,45 . Further, it is suggested that water molecules have an important role in the transfer of NH 3 , because sites vacated by NH 3 are refilled with water molecules such that the NH 3 hydrogen bonds cannot reform, thereby forcing NH 3 forward.</p><p>It is very probable that fluctuations of protein structure are part of the mechanism for nitrogenase, because the proposed route passes between different polypeptide chains, and those chain sections have minimal secondary structure and above average temperature factors. Gas transport channels in a variety of enzymes are characterised by structural flexibility 73 . Conformational variations are a primary feature of the ammonia-channeling glutamine amidotransferases [41][42][43]46,47 .</p><p>Amino acid Gln 191A is a key component of the mechanism in nitrogenase, because it is normally hydrogen bonded through sidechain NH 2 to O1 of HCA (another key component), and NH 3 has to move around both of these atoms, and to avoid the side chain CH 2 of Gln 191A , before escaping from the water pool (Fig. 6). Therefore the consequences of modification of this residue provide an experimental approach to testing the validity of the mechanism. Mutation of Gln 191A to lysine with a longer side chain shuts down NH 3 production from N 2 74 , but NH 3 is produced with azide as substrate by this mutant 15 , indicating that the NH 3 efflux channel is not closed by Lys 191A . The interpretation of the diminished activity of the Lys 191A mutant, that is only high K m reduction of C 2 H 2 and no reduction of N 2 , is probably that given previously 25 , which is that Lys blocks hydrogenation steps at Fe6 but not hydrogenation steps at Fe2. I am not aware of other experimental data that bears on the validity of the NH 3 efflux mechanism proposed here.</p><p>In this paper I have presented a plausible mechanistic pathway for egress of NH 3 produced by nitrogenase. How can this proposal be tested? A standard experimental approach is modification of key amino acids and assessment of reactivity. The early stages of the NH 3 path involve the side-chain of Gln 191A , in two ways: the terminal amide NH 2 group engages some hydrogen bonds, while c-CH 2 provides a steric barrier. In view of the flexibility of the side-chain it is not surprising that lysine in this position does not fully shut down the escape of NH 3 , but large non-hydrogen bonding side-chains such as Phe or Trp at position 191A are predicted to impede the NH 3 pathway. A large increase in size of the side-chain of residue 65A could also interfere. Another significant side-chain is that of Thr 58A (Fig. 7), because the proposed pathway passes closely around its c-CH 3 . A more voluminous side-chain, without hydrogen bonding capability, could interfere with NH 3 passage. These are suggestions for mutations to clog the pathway, but the likely fluctuations in protein structure may open alternative paths. In this context, residue 191A in the tighter earlier part of the NH 3 route would be the prime target.</p><p>Might it be possible to incorporate a reagent for NH 3 into the protein near the putative pathway, and then crystallise the protein containing the captured escaping NH 3 ?</p><p>Computationally, a normal modes analysis 75,76 of the protein structure would be informative about its flexibility around the proposed pathway. Molecular dynamics simulations, enhanced with various techniques 44,46,77 , should provide further insight into ammonia transport in nitrogenase.</p><!><p>The initial search for cavities was undertaken with the program fpocket 61 , using a procedure which strips water and heteroatoms (but not FeMo-co, homocitrate, or the P-cluster) from the atom coordinates (pdb) file, and then searches for alpha spheres. An alpha sphere contacts four atoms on its boundary and contains no internal atom: the four atoms are equally distant from the sphere centre, defining the radius of the sphere. Pockets are clusters of intersecting spheres, and pocket searches are controlled by filtering sphere properties such as radius, separation, and chemical attributes of the defining atoms. The variables in fpocket were first evaluated: for the searches of nitrogenase FeMo proteins, minimum alpha sphere radii in the range 2.7-3.0A ˚were found to be most informative, with pockets required to contain at least 35 alpha spheres. After each search the water excluded from the fpocket calculation was added back into the output files, in order to examine the presence of water in or near the pockets. Calculations of pockets were made for all of the high-resolution crystal structures (PDB 3U7Q, 1M1N, 2MIN, 3MIN, 2AFH, 2AFI, 1MIO, 1QGU, 1QH1, 1QH8, 1H1L) that included the MoFe protein and a high proportion of resolved water molecules (the best structures contain ca 1300 water molecules per FeMo-co in the MoFe protein: structures 3K1A, 1M34 and 2AFK describe considerably less water).</p><p>A 3U7Q pdb file with the modeled NH 3 positions included is available from the author.</p>

Scientific Reports - Nature

Role of Mitochondrial RNA Polymerase in the Toxicity of Nucleotide Inhibitors of Hepatitis C Virus

Toxicity has emerged during the clinical development of many but not all nucleotide inhibitors (NI) of hepatitis C virus (HCV). To better understand the mechanism for adverse events, clinically relevant HCV NI were characterized in biochemical and cellular assays, including assays of decreased viability in multiple cell lines and primary cells, interaction with human DNA and RNA polymerases, and inhibition of mitochondrial protein synthesis and respiration. NI that were incorporated by the mitochondrial RNA polymerase (PolRMT) inhibited mitochondrial protein synthesis and showed a corresponding decrease in mitochondrial oxygen consumption in cells. The nucleoside released by the prodrug balapiravir (R1626), 4′-azido cytidine, was a highly selective inhibitor of mitochondrial RNA transcription. The nucleotide prodrug of 2′-C-methyl guanosine, BMS-986094, showed a primary effect on mitochondrial function at submicromolar concentrations, followed by general cytotoxicity. In contrast, NI containing multiple ribose modifications, including the active forms of mericitabine and sofosbuvir, were poor substrates for PolRMT and did not show mitochondrial toxicity in cells. In general, these studies identified the prostate cell line PC-3 as more than an order of magnitude more sensitive to mitochondrial toxicity than the commonly used HepG2 cells. In conclusion, analogous to the role of mitochondrial DNA polymerase gamma in toxicity caused by some 2′-deoxynucleotide analogs, there is an association between HCV NI that interact with PolRMT and the observation of adverse events. More broadly applied, the sensitive methods for detecting mitochondrial toxicity described here may help in the identification of mitochondrial toxicity prior to clinical testing.

role_of_mitochondrial_rna_polymerase_in_the_toxicity_of_nucleotide_inhibitors_of_hepatitis_c_virus

INTRODUCTION<!><!>INTRODUCTION<!>Reagents.<!>Cell culture and cytotoxicity studies.<!>Measurement of active triphosphate metabolite in PC-3 cells.<!>DNA and RNA templates and primers.<!>Enzymatic assays.<!>Mitochondrial protein synthesis assay.<!>Measurement of the oxygen consumption rate in PC-3 cells.<!>Detection of cellular DNA level.<!>Detection of RNA transcripts using RT-PCR.<!>Computer modeling.<!>Evaluation of cytotoxicity in cell lines and primary cells.<!><!>Measurement of active triphosphate metabolite in PC-3 cells.<!>Inhibition of human DNA and RNA polymerases and incorporation by PolRMT.<!><!>Inhibition of human DNA and RNA polymerases and incorporation by PolRMT.<!>Structural modeling of PolRMT.<!><!>Effects on mitochondrial protein synthesis.<!><!>Effects on mitochondrial protein synthesis.<!><!>Effects on mitochondrial protein synthesis.<!>Inhibition of mitochondrial respiration.<!><!>Inhibition of RNA polymerases by 2′CMeNTP analogs.<!><!>Detection of RNA transcripts in PC-3 cells.<!><!>DISCUSSION<!>

<p>Hepatitis C virus (HCV) is a major cause of liver disease worldwide and is the leading reason for liver transplantation in North America and Europe (1). In 2011, the treatment of chronic HCV infection was advanced with the regulatory approval of two protease inhibitors, telaprevir and boceprevir, which directly targeted the virus and increased sustained viral response rates (2). However, these agents were given in combination with pegylated interferon (IFN) and ribavirin (RBV), adding new side effects on top of the already challenging tolerability profile of the prior standard of care. These combinations also had complicated dosing regimens, were only effective in patients with genotype 1 infection, and were less efficacious in many of the populations most in need of therapy (3). These limitations led to the continued pursuit of agents to treat HCV infection that exhibit greater efficacy and improved tolerability and more broadly address the needs of those afflicted with HCV infection, including patients around the world infected with other HCV genotypes and those with advanced liver disease.</p><p>In particular, nucleotide inhibitors (NI) of HCV RNA synthesis that serve as alternate substrates and inhibitors of the viral RNA-dependent RNA polymerase (HCV nonstructural protein 5B [NS5B]) have garnered substantial attention. The binding of NI to the highly conserved NS5B active site results in activity that is maintained across genotypes and substantial loss of viral fitness upon the infrequent development of resistance mutations (4, 5). The promising attributes of NI proved difficult to translate into clinical success until the recent approval of sofosbuvir in 2013. Failure during the clinical development of NI candidates has been primarily due to toxicity. The first two nucleoside analogs to enter clinical development, valopicitabine (NM283; prodrug of 2′-C-methyl (2′CMe) cytidine [2′CMeC, NM107] and balapiravir (R1626; prodrug of 4′-azido cytidine [4′-azidoC, R1479] (Fig. 1) had their development programs halted during phase 2 studies due to the observation of gastrointestinal and hematologic toxicity, respectively (6). Development of PSI-938 (prodrug of 2′F,2′CMe guanosine) was discontinued due to the observation of hepatic toxicity in phase 2b (7). IDX184 and BMS-986094, prodrugs that deliver the same pharmacologically active triphosphate (TP) (2′CMeGTP), had their clinical programs halted following the observation of cardiac and kidney toxicity with BMS-986094 (8, 9). Most recently, VX-135, a uridine analog with an undisclosed structure, was placed on partial clinical hold based on the observation of elevated liver enzymes (10). Other NI have likely never reached clinical trials due to the observation of toxicity during preclinical studies. For example, MK608 showed promising antiviral activity in chimpanzees (11), but further clinical studies have not been reported to date.</p><!><p>Structures of HCV NI and their clinical progress. The names of some corresponding nucleosides for the prodrugs are provided in parentheses.</p><!><p>The mechanisms for toxicity of ribonucleotide analogs have not been well characterized but could theoretically arise from incorporation into cellular RNA by human RNA polymerases I, II, III (PolI, PolII, and PolIII) and the mitochondrial RNA polymerase (PolRMT). In comparison to PolII, PolRMT has been shown to be particularly vulnerable to inhibition by ribonucleotide analogs due to the lack of a functional proofreading activity (12). However, neither PolI nor PolIII has been studied as a source of cytotoxicity for this class of compounds. In this study, we assessed the potential for clinical HCV NI (Fig. 1) to cause toxicity in various functional cell-based assays, including assays for (i) cytotoxicity in multiple human cell lines and primary cells, (ii) reductions in mitochondrial- and nuclear-DNA-encoded proteins, and (iii) changes in mitochondrial oxygen consumption. The interactions of NI with purified human DNA and RNA polymerases in biochemical assays were also assessed. While general cytotoxicity or the inhibition of the human DNA polymerases or RNA PolI, PolII, and PolIII did not consistently identify NI associated with clinical toxicity, interactions with PolRMT in biochemical assays coupled with corresponding decreases in mitochondrial protein production and cellular respiration suggest this polymerase as an important determinant of toxicity. A screening paradigm is proposed that will aid in identifying the potential for mitochondrial toxicity of nucleotide analogs that can also be applied more broadly to other classes of drugs.</p><!><p>All NI were synthesized by Gilead Sciences, Inc. Puromycin, 5-flurouracil (5-FU), and alpha-amanitin were purchased from Sigma-Aldrich (St. Louis, MO). All radioactively labeled nucleoside triphosphates (NTPs) were purchased from PerkinElmer (Shelton, CT).</p><!><p>The following cell lines were obtained from the indicated sources: Huh-7 (differentiated hepatocellular carcinoma; JCRB Cell Bank, Japan), HepG2 (hepatoblastoma; ATCC), PC-3 (prostate metastatic carcinoma; ATCC), MRC5 (fibroblast from normal lung tissue; ATCC), and MT-4 (human T-cell leukemia virus 1 [HTLV-1]-transformed human T lymphoblastoid cells; NIH AIDS Research and Reference Reagent Program). Primary human hepatocytes were from Bioreclamation IVT (Westbury, NY) or Life Technologies (Carlsbad, CA). Human peripheral blood mononuclear cells (PBMCs) were isolated from human buffy coats obtained from healthy volunteers (Stanford Blood Bank, Palo Alto, CA) using standard Ficoll separation and stimulated as described elsewhere (13) and were tested at both quiescent and stimulated stages. Quiescent PBMCs were stimulated with 10 units per ml of recombinant human interleukin 2 (hIL-2) and 1 μg per ml phytohemagglutinin P (PHA-P) for 48 h prior to drug treatment. Normal human primary bone marrow (BM) light-density cells were from three different lots obtained from AllCells (Emeryville, CA) or Lonza (Walkersville, MD). Primary rat neonatal cardiomyocytes were isolated from 3-day-old Sprague Dawley rat pups as described previously (14).</p><p>All cells were cultured at 37°C in a 5% CO2 incubator with 90% humidity unless noted otherwise. Detailed culture conditions for each cell line and primary cell can be found in the supplemental material. All cells were treated with the various compounds for 5 days, except for human erythroid and myeloid progenitors, which were treated for 14 days. After the incubation period, cell viability was measured by the addition of CellTiter Glo viability reagents (Promega, Madison, WI). The luminescence signal was quantified on an Envision luminescence plate reader (PerkinElmer, Waltham, MA) after incubation of the reagents and cells for 10 min at room temperature. The compound concentration that caused a 50% decrease in the luminescence signal (CC50), a measure of toxicity, was calculated by nonlinear regression using a sigmoidal dose-response (variable slope) equation, as follows: 1Y=bottom+(top−bottom)/{1+10^[(LogCC50−X)×hillslope]} where X is the log of the concentration of the test compound, Y is the response, and the bottom and top values were fixed at 0 and 100, respectively, unless a significant variation was noted. CC50 values were calculated as the average of the results of three or four independent experiments.</p><p>The effects of the compounds on the proliferation of human erythroid and myeloid progenitors were tested in MethoCult84434, a methylcellulose-based colony assay conducted by Stemcell Technology (Vancouver, Canada) (15). After a 14-day culture, hematopoietic progenitor colonies (CFU-E, BFU-E, CFU-GM, and CFU-GEMM [E, erythroid; BFU, burst-forming unit; GM, granulocyte/macrophage; GEMM, multilineage progenitors]) were enumerated and the CC50 values were calculated using equation 1.</p><!><p>The amount of nucleoside triphosphate was determined from three independent studies by ion-pairing liquid chromatography-tandem mass spectrometry (LC-MS/MS) as previously described (16). All compounds were incubated with the cells at 10 μM, except for BMS-986094 and NM107, which were incubated at 0.1 and 1 μM, respectively, due to the cytotoxicity observed at higher concentrations. Triphosphate concentrations for BMS-986094 and 2′CMeC were applied at doses normalized to the concentrations anticipated after a 10 μM incubation for comparison with the other NI. The reported TP levels are the maximal concentrations observed after 48 h during a 5-day incubation.</p><!><p>Activated fish sperm DNA was purchased from USB/Affymetrix (Santa Clara, CA) and used as a template for the DNA polymerases alpha, beta, and gamma. The DNA template used in the RNA Pol II assay was a 1,188-bp restriction fragment containing the cytomegalovirus (CMV) immediate early promoter (Promega, Madison, WI). The RNA and DNA oligonucleotides used in the mitochondrial RNA polymerase assay (see Table S1 in the supplemental material) were synthesized and PAGE purified by Thermo Scientific/Dharmacon (Lafayette, CO). RNA primer R12 was 5′-32P-phosphorylated with [γ-32P]ATP (3,000 Ci/mmol) and T4 kinase (New England BioLabs, Ipswich, MA).</p><!><p>Human DNA polymerase alpha, isolated from HeLa cell extracts, was from CHIMERx (Madison, WI). Recombinant human DNA polymerase beta, expressed in E. coli, was a gift from Zucai Suo at The Ohio State University. Recombinant human DNA polymerase gamma (including both the large subunit and the small subunit) was cloned, expressed, and purified from insect cells by Gilead Sciences (Foster City, CA) (13). RNA PolII was purchased as part of the HeLaScribe nuclear extract in vitro transcription system kit from Promega (Madison, WI). The recombinant human PolRMT and the transcription factors mitochondrial transcription factor A (mtTFA) and B2 (mtTFB2) were purchased from Enzymax (Lexington, KY). FLAG epitope-tagged RNA PolI and PolIII complexes were isolated from rat NSN1 cells and HeLa cells, respectively (17, 18). The inhibition of DNA polymerases alpha, beta, and gamma, PolRMT, and PolII and the rate of single-nucleotide incorporation by PolRMT have been described previously in detail (12, 13, 19). Inhibition of RNA PolI and PolIII was studied using published methods (20, 21). All concentrations were final unless noted otherwise. For the RNA PolI inhibition assay, a reaction mixture containing 50 mM HEPES (pH 7.9), 20% glycerol, 100 mM KCl, 1 mM dithiothreitol (DTT), 50 μg/ml calf thymus DNA, 4 mM MgCl2, 5 or 80 μM UTP (0.25 μl of [5,6-3H]UTP, >25 Ci/mmol; PerkinElmer), 1 mM noncompeting NTP, competing NTP (5 μM for GTP, CTP, and UTP and 25 μM for ATP), and various concentrations (0 to 500 μM) of NTP analogs was preincubated at 30°C for 5 min. The reaction was initiated with the addition of 5 μl of FLAG affinity-purified RNA PolI. After incubation at 30°C for 20 min to 1 h, the reactions were terminated by pipetting the mixtures onto DEAE-cellulose discs (DE81; Whatman), and the products analyzed as described previously (20, 21).</p><p>The RNA PolIII inhibition assay was conducted in a similar manner, except that the concentrations of the competing NTP were modified to 100 μM for CTP and GTP and 500 μM for ATP to compensate for the low assay signal at lower NTP concentrations.</p><!><p>PC-3 cells were treated with the compounds for 5 days and analyzed with the MitoTox MitoBiogenesis in-cell enzyme-linked immunosorbent assay (ELISA) kit (MitoSciences/Abcam, Eugene, Oregon) as described previously (13). The assay uses quantitative immunocytochemistry to measure the protein levels of nuclear DNA-encoded succinate dehydrogenase (SDH-A; complex II [succinate dehydrogenase]) and mitochondrial DNA-encoded cytochrome c oxidase (COX-1; complex IV [cytochrome c oxidase]) in cultured cells.</p><!><p>Mitochondrial respiration was monitored by measuring the rate of oxygen consumption (OCR) of PC-3 cells after 3-day treatments with the compounds, using a Seahorse extracellular flux analyzer (XFe-96) based on published protocols (22–24). The optimal cell seeding was determined by measuring the basal OCR of PC-3 cells that were seeded at different densities of 2.5 × 103, 5 × 103, 7.5 × 103, and 10 × 103 cells/well in XF 96-well plates (Seahorse Bioscience, North Billerica, MA) and incubated with 0.5% dimethyl sulfoxide (DMSO) for 3 days. The seeding density of 5 × 103 cells/well yielded basal OCR values of 80 to 100 pmol/minute (<15% coefficient of variation) and was chosen for future studies. Each reagent in the Mito Stress test kit (Seahorse Biosciences), i.e., ATP synthase inhibitor oligomycin, mitochondrial uncoupler carbonyl cyanide-4-(trifluoromethoxy)phenylhydrazone (FCCP), and a fixed-ratio mixture of the mitochondrial complex I (NADH dehydrogenase) inhibitor rotenone and the complex III (CoQH2-cytochrome c reductase) inhibitor antimycin A, was titrated. The lowest concentration that reached the maximal effect was chosen for each reagent. The OCR signals were normalized by cell numbers using DNA content determined by Hoechst's stain as described below.</p><p>PC-3 cells were seeded at a density of 5 × 103 cells/well in XF 96-well plates and incubated with the compounds for 3 days. On the day of the assay, the cell medium was replaced with XF assay medium (pH 7.4) containing 10 mM glucose and 1 mM freshly prepared pyruvate. Mitochondrial respiration was monitored by measuring OCR on a Seahorse extracellular flux analyzer (XFe-96). All concentrations listed are the final concentration after mixing unless noted otherwise. Multiple parameters were measured after the sequential injection of oligomycin (2 μM), FCCP (0.25 μM), and the mixture of rotenone (0.5 μM) and antimycin A (0.5 μM). Spare respiratory capacity was obtained by subtracting the rate of basal respiration from the rate of maximal respiration. The data reported for each treatment are the average of the results from six replicates.</p><!><p>After mitochondrial respiration was measured, the XF medium was removed and replaced with 60 to 80 μl of 1× TNE buffer (100 mM Tris base, 10 mM EDTA, and 1.75 M NaCl). The cells were lysed by freezing at −80°C for >2 h. The amount of DNA in each well was determined by staining the cells with Hoechst 33258 stain (Thermo Scientific, Wilmington, DE) and measuring the resulting fluorescence signal with an excitation wavelength of 352 nm and a detection wavelength of 461 nm. In parallel, another set of cells was treated with the compounds for 3 days and the ATP levels were measured using CellTiter Glo viability reagents (Promega, Madison, WI).</p><!><p>PC-3 cells were seeded at a density of 20,000 cells/well in a 96-well black clear-bottom plate (Corning, Tewksbury, MA) in 1× F-12K nutrient mixture, Kaighn's modification, with l-Glu, 10% heat-inactivated fetal bovine serum, and penicillin-streptomycin from Life Technologies (Carlsbad, CA) and grown overnight. CX-5461, a known specific RNA PolI inhibitor, was used as the positive control (25). The growth medium was aspirated and replaced with 200 μl of 100 μM BMS-986094, 5 μM CX-5461, or DMSO control prepared in growth medium. At selected time points (0.5, 1, 4, 8, and 24 h), duplicate wells of the compounds and DMSO control were aspirated, the cells were washed 1× with PBS (pH 7.4), and 100 μl of RNA lysis buffer was added (Promega, Sunnyvale, CA). RNA was purified using the Wizard SV 96 total RNA isolation kit (Promega) and eluted in 100 μl of nuclease-free water. The TaqMan RNA-to-Ct 1-step kit (ABI/Life Technologies) was used for reverse transcription-quantitative PCR (RT-qPCR) amplification and quantification, using the 7900HT Fast real-time PCR system (ABI) under the following conditions: 15 min at 48°C for reverse transcription and 10 min at 95°C for activation, followed by 40 cycles of 15 s at 95°C for denaturation and 1 min at 60°C for annealing and extension. Primer/probe sets included the following: Hs.PT.58.26770695 MYC (IDT PrimeTime standard qPCR assay; GenBank RefSeq accession number NM_002467), Hs.PT.58.264008 MYB (IDT PrimeTime standard qPCR assay; GenBank RefSeq accession number NM_005375), human RNA18S 5 FAM-MGB (Applied Biosystems assay; TaqMan assay identifier [ID] Hs02596862_g1), human mitochondrially encoded ATP synthase 6 (MT-ATP6)–FAM-MGB (Applied Biosystems assay; TaqMan ID Hs02596862_g1), and custom primers for the 5′-external transcribed spacer (ETS) of pre-rRNA (GenBank accession number U13369) (probe, 5′-/56-FAM/TCC GGT ACC/ZEN/CCC AAG GCA C/3IABkFQ/-3′; primer 1, 5′-CAT AAC GGA GGC AGA GAC AG-3′; and primer 2, 5′-AAA AGC CTT CTC TAG CGATCT G-3′ [IDT]). Standard curves for each gene were generated using a pool of the DMSO-treated control RNA diluted serially 1/5 in RNase-free water. Transcripts were quantified relative to the standard curves, and the results expressed as the percentages of the average value of DMSO control wells.</p><!><p>A homology model of PolRMT was constructed based on a ternary structure of T7 RNA polymerase that includes an RNA primer, DNA template, two bound Mg++ ions, and an incoming nonhydrolyzable ATP analog (PDB ID 1S76) (26). Although a crystal structure of human POLRMT was recently solved (27), this apo structure is expected to undergo significant conformational changes during transcription, and thus, the T7 RNA polymerase structure was deemed a better starting point. The model was constructed using the Prime package (version 3.0, 2011; Schrodinger, LLC, New York, NY). The overall homology was 27% identity and 44% similarity over residues 336 to 1188 (PolRMT numbering), but the active site was exceptionally well conserved. Only two residues with direct interactions with the substrate differed (T7 RNA polymerase R632 → PolRMT Q992, positioned under the 3′OH of the substrate, and T7 RNA polymerase S541 → PolRMT N926, positioned over the substrate/template base pair). Models of the NI in the PolRMT active site were refined using Macromodel (version 9.9, 2011; Schrodinger, LLC, New York, NY).</p><!><p>The cytotoxicity of a panel of clinical HCV NI was tested in various human cell lines and primary cells for 5 to 14 days (Fig. 1; Tables 1 and 2). Compounds not associated with clinical toxicity generally did not show marked cytotoxicity. For example, sofosbuvir showed modest effects in only one cell line (HepG2), and the nucleoside released by mericatibine, PSI-6130, had no effects on any of the cells at the concentrations tested (28). However, GS-6620, which was well tolerated in animals and in phase 1 clinical studies (29), had cytostatic effects in MRC5, MT-4, and bone marrow cells. Furthermore, the nucleosides released by valopicitabine and balapiravir, the nucleoside prodrugs whose clinical development was stopped due to toxicity, did not show marked cytotoxicity in the different cell types tested. In particular, 4′-azidoC showed no cytotoxicity in hematologic cells to correlate with the clinical findings reported with balapiravir. 2′CMeC was toxic to the MT-4 lymphocyte-derived cell line, with a CC50 of 19 μM, but there was little or no cytotoxicity in other cell types. BMS-986094 showed marked cytotoxicity in all human cell lines and primary cells tested. However, an alternate prodrug of 2′CMeG, IDX184, showed minimal toxicity. The lack of effect of IDX184 was likely due to inefficient activation in the cells studied (Table 3). Based on the clinical observation of cardiomyopathy in patients treated with BMS-986094, its cytotoxicity was also assessed in primary rat cardiomyocytes (30). After 5-day treatments with BMS-986094, cardiomyocytes showed the most profound sensitivity to cytotoxicity among the primary cells tested, with a CC50 of 0.68 ± 0.32 μM (mean ± standard deviation), almost equivalent to that of puromycin (CC50 of 0.33 ± 0.05 μM).</p><!><p>Cytotoxicity in human cell lines after 5-day treatment</p><p>The cytotoxicity of GS-6620 in some cell types was reported previously (13).</p><p>Complex, multiphasic dose response was observed, with an initial drop in viability of 50% followed by a rebound to over 50% at higher concentrations. Flow cytometry studies suggested that the effect of the compound on these cells was cytostatic instead of cytotoxic.</p><p>Control treatment.</p><p>Cytotoxicity in human primary cellsa</p><p>Hepatocytes and PBMCs were treated for 5 days, while bone marrow-derived cells were treated for 14 days.</p><p>Data are from reference 28.</p><p>ND, not determined.</p><p>Data are from reference 13.</p><p>Control treatment.</p><p>HCV NI intracellular activation in PC-3 cells</p><p>The amount of nucleoside triphosphate was determined from three independent LC-MS/MS analyses. The reported TP levels are the maximal concentration observed after 48 h during a 5-day incubation. BLQ, below the limit of quantification.</p><!><p>Prodrug activation and nucleotide phosphorylation can be highly cell type dependent and can lead to different degrees of cytotoxicity. However, it is impractical to measure this process in multiple cell lines. In this study, we chose to measure the triphosphate concentration after 48-h incubations of compounds with PC-3 cells. As shown by the results in Table 3, sofosbuvir and PSI-6130 efficiently formed TP, illustrating that the lack of toxicity (CC50 <100 μM) (Table 1) was not due to poor intracellular activation. In contrast, the TP level of IDX184 was >37-fold less than that of BMS-986094, which correlated well with the >47-fold difference in the CC50 values in PC-3 cells.</p><!><p>Cell-based assays for general cytotoxicity failed to reliably identify NI associated with clinical toxicity. Assessing the interactions of the active triphosphate metabolites of the test compounds with the potential molecular targets of toxicity may be a more sensitive way of assessing the potential for off-target effects. As summarized in Table 4, none of the active triphosphates inhibited human DNA polymerases alpha, beta, and gamma at the highest concentration tested (100 μM), nor did they affect PolII-catalyzed RNA synthesis (50% inhibitory concentration [IC50] of >200 μM).</p><!><p>Inhibition of human DNA and RNA polymerases and substrate utilization of 5′-triphosphate active metabolites of HCV NI</p><p>Pol α, polymerase alpha; Pol β, polymerase beta; Pol γ, polymerase gamma; PolII, polymerase II; PolRMT, mitochondrial RNA polymerase.</p><p>Specific positive control.</p><!><p>In contrast to DNA polymerases and RNA PolII, PolRMT was shown to interact with the active forms of many of the NI. The triphosphate formed by 4′-azidoC was a potent inhibitor of PolRMT-catalyzed RNA synthesis, with an IC50 of 3.8 μM. The active triphosphates of BMS-986094 and IDX184 and 2′CMeC were also inhibitors, with IC50s of 52 and 230 μM, respectively. In addition, we directly measured the incorporation of the triphosphates in biochemical assays at a fixed saturating concentration (500 μM) and compared the relative rates of incorporation to those of the corresponding natural ribonucleoside triphosphates (rNTPs). As shown by the results in Table 4, triphosphates formed by BMS-986084 and IDX184, 4′-azidoC, and 2′CMeC served as excellent substrates and were incorporated by PolRMT at rates similar to those of their corresponding natural rNTPs. In contrast, the active forms of sofosbuvir, PSI-938, mericitabine, and GS-6620 were all exceedingly poor substrates for PolRMT (rates of <1% of their respective rNTPs).</p><!><p>We constructed a homology model of PolRMT based on a ternary structure of T7 RNA polymerase (26). An examination of various NI presented here provided some insight into their relative activities (Fig. 2). Monosubstituted ribose analogs could be accommodated by the PolRMT active site to various degrees. The 4′-azido substitution was the most easily accommodated by the enzyme, with essentially no perturbation of the active site required for the inhibitor to bind. The 2′CMe substitution has a slight van der Waals clash with Tyr999, leading to a small shift in this residue toward 1′. The movement of this residue was more pronounced with the 2′F,2′CMe substitutions, likely due to the loss of hydrogen bonding capacity with the fluorine. Similarly, a 1′CN substituent leads to a slight van der Waals clash with His1125, thus leading to a small shift in this residue toward 2′. Consistent with the biochemical result of exceedingly poor incorporation for the active metabolite of GS-6620, the opposing movements of Tyr999 and His1125 would not be tolerated in the presence of both 1′ and 2′ substitutions.</p><!><p>Homology model of ATP incorporation into human PolMRT based on a crystal structure of T7 RNA polymerase (PDB ID 1S76). The active site is well conserved, with only N926 (shown) and Q992 (not shown, but positioned above the substrate/template bases) being different. A 1′CN substitution shifts H1125 toward 2′. A 2′CMe substitution shifts Y999 toward 1′. When combined, these two substitutions are then in conflict and the analog is a poor substrate.</p><!><p>In order to determine whether their interaction with PolRMT had any consequences for protein expression, we measured the effects of HCV NI on the levels of mitochondrial (transcribed by PolRMT) and nuclear (transcribed by PolII) DNA-encoded proteins in cultured cells. Mitochondrial toxicity studies have typically been conducted in the liver cell line HepG2. However, based on the greater sensitivity of PC-3 cells to general cytotoxicity to HCV NI, studies using the mitochondrial toxin dideoxycytosine (ddC) were done in these cells to determine whether they would provide a more sensitive model for mitochondrial toxicity. PC-3 cells were >30-fold more sensitive than HepG2 cells to mitochondrial DNA depletion after a 10-day treatment (see Fig. S1A in the supplemental material), and they were >500-fold more sensitive to reductions in COX-1 protein expression after a 5-day treatment (see Fig. S1B and C). Based on these results, PC-3 cells were chosen for subsequent cellular studies.</p><p>The effects on cellular protein production of model inhibitors of different cellular processes, including mitochondrial replication (ddC), mitochondrial translation (chloramphenicol), and general cellular translation (puromycin), were characterized (Fig. 3A). Consistent with their targets in mitochondria, ddC and chloramphenicol caused selective depletion of mitochondrial-DNA-encoded COX-1 prior to any effect on nuclear-DNA-encoded SDH-A or ATP. In contrast, puromycin had no selective effect with COX-1, SDH-A, and ATP, as shown by superimposable dose-response curves. The nonnucleotide RNA PolII inhibitor alpha-amanitin had a profile similar to that of puromycin (data not shown).</p><!><p>Effects of compounds on mitochondrial protein synthesis in PC-3 cells after 5-day treatments. The levels of mitochondrial protein synthesis, represented by COX-1 protein (•), nuclear protein synthesis, represented by SDH-A protein (◆), and ATP (■) were curve fitted with solid lines, dashed lines, and dotted lines, respectively. (A) Puromycin showed nonselective inhibitive effects on the COX-1, SDH-A, and ATP levels, while chloramphenicol and ddC specifically inhibited COX-1 synthesis. (B) 4′-AzidoC and BMS-986094 showed selective inhibition of COX-1 synthesis. (C) RBV at a concentration not resulting in any observed toxicity (10 μM) potentiated the mitochondrial and general cellular toxicity associated with BMS-986094. In contrast, at 10 μM, RBV showed no appreciable effect on mitochondrial or nuclear protein synthesis or ATP level.</p><!><p>As summarized in Table 5, many of the HCV NI tested in this system showed little or no cytotoxicity and no evidence for a selective effect on mitochondrial protein synthesis. Similar to the control mitochondrial toxins ddC and chloramphenicol, 4′-azidoC (nucleoside of balapiravir) showed selective inhibition of mitochondrial protein synthesis, with a CC50 of 11 μM, while no change in nuclear protein translation was observed at up to 50 μM (Table 5; Fig. 3B). BMS-986094 showed a mixed effect, somewhere between those of the selective mitochondrial toxins ddC and chloramphenicol and the general cellular cytotoxicity of puromycin (Fig. 3B). The slightly enhanced sensitivity to BMS-986094 of COX-1 relative to that of SDH-A was reproducible in three independent studies done in duplicate (F test, P < 0.0001). This effect on mitochondrial protein production is consistent with the reported effect of another 2′CMe-monosubstituted analog, 2′CMe adenosine, which showed selective depletion of mitochondrial RNAs in a prior study (12).</p><!><p>Effects of HCV NI on mitochondrial protein production, respiration, cellular protein production, and ATP levels in PC-3 cells</p><p>Cells were treated with compounds for 5 days.</p><p>Cells were treated with compounds for 3 days.</p><p>ND, not determined.</p><p>Control treatment.</p><!><p>To further assess the mechanism of BMS-986094 toxicity, this compound was studied in combination with RBV. RBV is an inhibitor of IMP dehydrogenase (IMPDH) and reduces endogenous GTP pools that compete with the pharmacologically active triphosphate formed by BMS-986094. While RBV itself had low cytotoxicity and no selective effect on mitochondria, its coincubation with BMS-986094 potentiated the mitochondrial and cellular toxicity of BMS-986084 by 3- to 4-fold (Fig. 3C).</p><!><p>In order to assess whether there was any functional consequence of the inhibition of mitochondrial protein production, mitochondrial spare respiratory capacity was assessed after 3-day treatments with the NI (Fig. 4; Table 5). A 3-day instead of 5-day time point was chosen due to the high sensitivity of the assay. The control inhibitors ddC, chloramphenicol, and puromycin showed results consistent with their effects on mitochondrial protein synthesis discussed above. For example, ddC and chloramphenicol selectively reduced respiration at concentrations that did not affect total cellular DNA content or ATP (Fig. 4A). Correlating with the biochemical and protein expression data, 4′-azidoC showed selective effects on respiration similar to those observed with the positive controls (Fig. 4B). Consistent with the results from mitochondrial protein synthesis assays, BMS-986094 showed a primary effect on mitochondrial respiration followed by general cytotoxicity in both PC-3 cells (CC50 of 0.48 ± 0.07 μM) (Fig. 4B) and primary rat cardiomyocytes (CC50 of 0.43 ± 0.14 μM) (Fig. 4C). All other HCV NI, including those showing some cytotoxicity in PC-3 cells, did not have a selective effect on mitochondrial respiration.</p><!><p>Effects of compounds on mitochondrial spare respiratory capacity in PC-3 cells (A and B) and primary rat cardiomyocytes (C) after 3-day treatments. The relative levels of spare respiratory capacity (•), DNA (◆), and ATP (■) were curve fitted with solid lines, dashed lines, and dotted lines, respectively. The spare respiratory capacity was normalized by cell numbers. (A) Puromycin showed nonselective inhibition of the spare respiratory capacity, DNA, and ATP levels, while ddC and chloramphenicol specifically inhibited mitochondrial spare respiratory capacity but showed minimal effects on cellular DNA and ATP levels. (B) 4′-AzidoC and BMS-986094 showed selective inhibition of mitochondrial respiration in PC-3 cells. (C) BMS-986094 showed selective inhibition of mitochondrial respiration in primary rat cardiomyocytes.</p><!><p>In the above-described studies, BMS-986094 showed a profile that was intermediate between the profiles of selective mitochondrial toxins like ddC, chloramphenicol, and 4′-azidoC and those of generally cytotoxic agents, exemplified by puromycin. Furthermore, the effects of BMS-986094 were increased by depleting the levels of GTP with ribavirin, suggesting that the effects are caused by competition at the nucleotide level. Combined, these results suggest that BMS-986094 and other 2′CMe monosubstituted NI may have an off-target polymerase in addition to PolRMT. Therefore, 2′CMe monosubstituted NTPs of the four natural ribonucleotide bases (2′CMeGTP, 2′CMeATP, 2′CMeCTP, and 2′CMeUTP) were tested for their inhibition of RNA PolI, PolII, PolIII, and PolRMT. As summarized in Table 6, all four 2′CMeNTPs showed inhibition of PolI but not of PolII or PolIII.</p><!><p>Inhibition of RNA polymerases by 2′CMe-modified nucleotide analogs</p><!><p>The BMS-986094-induced inhibition of the above-mentioned RNA polymerases was further tested in cell culture, where representative RNA transcription products from PolI (preribosome), PolII (myc), and PolRMT (ATP6) in PC-3 cells were measured using RT-PCR during a 1-day treatment. As shown by the results in Fig. 5, 100 μM BMS-986094 decreased the relative levels of multiple RNA transcripts, including preribosomal, myc, and mitochondrial ATP6, and total RNA, while the different RNA levels at 24 h may reflect the different degrees of stability of these RNA products. In contrast, at 5 μM, the positive-control RNA PolI inhibitor CX-5461 selectively decreased pre-rRNA but had no effect on the levels of myc and mitochondrial ATP6 RNAs.</p><!><p>Effects of BMS-986094 on RNA transcript levels in PC-3 cells during a 1-day treatment. The relative levels of transcripts of preribosome (PolI), myc (PolII), mitochondrial ATP6 (PolRMT), and total RNA are shown in green (•), blue (▲), red (▼), and black (⭘). Cells were treated with BMS-986094 (100 μM) (left) or the positive control RNA Pol1 inhibitor CX-5461 (5 μM) (right), and transcripts in purified total nuclear RNA were quantified by RT-qPCR.</p><!><p>Nucleotide analogs have played key roles as antiviral agents for herpes simplex virus (HSV), human immunodeficiency virus (HIV), and hepatitis B virus (HBV) (31). However, the early generations of NI targeting these viruses were plagued by toxicity (32). The toxicity potential of early 2′-deoxyribonucleotide analogs was perhaps best exemplified by the delayed liver failure observed during a clinical trial of fialuridine (FIAU) for HBV (33). Preclinical studies had not predicated this toxicity. Elucidation of the mechanism of FIAU toxicity, mitochondrial dysfunction caused by incorporation by the mitochondrial DNA polymerase gamma, enabled the development of cell-based and biochemical assays that serve as effective counterscreens to allow the development of more selective NI (34–37). Different from the NI targeting DNA viruses, all HCV NI are ribonucleotide analogs and, therefore, are more likely to target host RNA polymerases than DNA polymerases. An analogous understanding of key modes of toxicity for ribonucleoside analogs would serve to improve the chances of successful clinical development for indications like HCV.</p><p>In contrast to our understanding of the inhibition of DNA polymerases by dNTP analogs, little is known about the inhibition of RNA polymerases by rNTP analogs. In the few published studies where inhibition of RNA polymerases has been assessed, only RNA PolII and RNA poly(A) polymerase were studied and no inhibition of any appreciable level was noted for select analogs (28, 38). It has recently been reported that PolRMT can incorporate many modified rNTP analogs, including triphosphates formed by 4′- and 2′-modified NI explored as anti-HCV agents (12). This effect on PolRMT may be a valuable clue to the mechanism for the clinical toxicity observed with some ribonucleotide analogs.</p><p>In an effort to identify the best predictive toxicity parameters in vitro, we conducted multiple cell- and enzyme-based studies on clinically tested HCV NI with publically disclosed chemical structures. Similar to the observation with 2′-deoxynucleotide analogs in the past, cytotoxicity in human cell lines and primary cells was found to be an unreliable and insensitive predictor of toxicity potential. Two HCV NI that have had their clinical development terminated, 2′CMeC and 4′-azidoC, only showed inconsistent cytotoxicity against the panel of cells studied here. Of note, 4′-azidoC, the nucleoside released by the prodrug balapiravir, whose clinical trial was stopped due to clinical hematologic toxicity, did not show effects on lymphoid cell lines or primary bone marrow-derived erythroid and myeloid cells. Conversely, GS-6620 showed cytostatic effects on fibroblast- and lymphocyte-derived cell lines and cultured primary bone marrow cells but did not show effects on PBMCs in vitro or any corresponding effects during chronic toxicology studies of up to 39 weeks in rats and dogs. In general, PC-3 cells were found to be the most sensitive to HCV NI toxicity among the cell types tested. For example, PC-3 was the only cell line to show any cytotoxicity from the selective mitochondrial toxins ddC, chloramphenicol, and 4′-azidoC. Liver-derived HepG2 cells have historically been used as a standard for cytotoxicity evaluation; however, our study shows that they are less sensitive to the known mitochondrial toxin ddC and certain HCV NI. Presently, we do not have an explanation for this observation, since the two cell lines are very similar in the following areas: (i) the glucose concentration in the culture medium is 5 to 7 mM, (ii) the two cell lines have similar doubling times, (iii) both cell lines showed similar oxygen consumption rates and rates of glycolysis, and (iv) both cell types have been found to efficiently activate nucleoside analogs (data not shown).</p><p>The inability of in vitro toxicity assays to reliably identify NI associated with clinical adverse events may be due to many reasons, including (i) insufficient TP formation caused by cell type-dependent prodrug activation and nucleotide phosphorylation (Table 3), (ii) altered signaling cascades and cell death regulation in immortal cell lines, (iii) cell-specific sensitivity to mitochondrial insult, and (iv) altered metabolic pathways under cell culture conditions, including the presence of glucose-rich cell culture media (39). By avoiding these limitations, biochemical assays assessing the molecular determinant(s) of toxicity would serve as the most reliable indicator of toxicity potential if target(s) could be identified. Given inhibition of the viral polymerase as the mechanism of action for NI, we studied a number of host polymerases for interactions with HCV NI. In these studies, none of the three human DNA polymerases (alpha, beta, and gamma) or PolII was inhibited by the TP formed by the HCV NI tested. In contrast, PolRMT incorporated all of the 2′CMeNTP analogs and 4′-azido CTP efficiently, making PolRMT a candidate for the molecular target resulting in the observation of adverse events with HCV NI.</p><p>Assessing direct effects on mitochondrial gene products and function allowed the confirmation of the relevance of the PolRMT biochemical results. In particular, the consequences for mitochondrial function from the interactions with PolRMT were established for the active metabolites of 4′-azidoC and BMS-986094. Illustrating the importance of intracellular activation for the mitochondrial toxicity observed in cell culture, IDX184 and BMS-986094 are both prodrugs of 2′CMeGTP, and yet there was a difference of 2 orders of magnitude in their effects on mitochondrial COX-1 expression, reflecting the different levels of activation noted in PC-3 cells (Tables 3 and 5). Consistent with the different toxicities of these prodrugs, BMS-986094 has been reported to be 30- to 60-fold more potent than IDX184 against the HCV replicon (40, 41).</p><p>In the clinic, some HCV NI have demonstrated adverse effects on distinct organ systems, including gastrointestinal, hepatic, cardiac, and hematological tissues. While these diverse toxic effects may be related to different underlying mechanisms of insult, it is also possible that one or a few off-target activities exist and that the sensitive organ is based on drug distribution. For example, although both chloramphenicol and ddC are mitochondrial toxins, they display distinct clinical toxicities, with chloramphenicol causing anemia and ddC causing peripheral neuropathy. Likewise, the 2′-deoxynucleoside analogs used for HBV and HIV have been associated with anemia, neuropathy, myopathy, cardiomyopathy, lipid abnormalities, liver toxicity, renal toxicity and peripheral neuropathy, all apparently caused by inhibition of a single target, the mitochondrial DNA polymerase gamma (42). Like all methods designed to assess a specific mechanism of toxicity, our studies have limitations. For example, despite having tested it in a comprehensive battery of biochemical and cell-based assays, we were unable to elucidate a mechanism for the clinically observed adverse hepatic effects observed with PSI-938.</p><p>When assessing the potential toxicity of NIs, comprehensive testing across multiple platforms and cell lines is critical. Based on a review of available data, Ahmad et al. recently concluded that direct mitochondrial toxicity is unlikely to be the mechanism of the cardiac effects of BMS-986094 (30). This conclusion is at odds with the findings in the current study and an earlier report showing BMS-986094-induced lactic acid increases in HepG2 cells (43). Our study showed efficient incorporation of the triphosphate of BMS-986094 by PolRMT, primary effects on mitochondrial protein production, and an initial decrease in respiration in PC-3 cells and rat cardiomyocytes prior to decreases in cell viability. Furthermore, the marked shift in the mitochondrial effects caused by ribavirin, an inhibitor of IMP dehydrogenase, and the subsequent reduction of competing endogenous GTP pools further support the role of triphosphate incorporation by a host polymerase. Given the high requirements for mitochondrion-derived energy in cardiac tissue, it is hard to dismiss these findings as a causative factor. Of note, other mitochondrial toxins, including ddC and chloramphenicol, have been reported to cause cardiomyopathy (44–46). While the primary effect of BMS-986094 was on mitochondrial function, the profile of this compound is intermediate between those of highly selective mitochondrial toxins (e.g., ddC, 4′-azidoC, and chloramphenicol) and generally cytotoxic agents (e.g., alpha-amanitin and puromycin), suggesting a secondary target for toxicity. Further experiments suggested that the general effects of BMS-986094 may be mediated by inhibition of other RNA polymerases, based on the inhibition of RNA PolI by 2′CMe monosubstituted analogs in biochemical assays (Table 6) and the depletion of RNA transcripts generated by a number of RNA polymerases in cellular studies (Fig. 5).</p><p>In conclusion, a comprehensive study of the in vitro toxicity potential of HCV NI has identified PolRMT as a potential off-target mediator of tissue-specific adverse events observed clinically. Analogs that have been associated with adverse events, including valopicitabine, balapiravir, and BMS-986094, were observed to interfere with PolRMT in cell-based and biochemical assays. In contrast, mericitabine and sofosbuvir did not interact with PolRMT or cause mitochondrial toxicity, consistent with their progression into advanced clinical development. We propose that a screening paradigm for HCV NI and, more broadly, other drug classes should include specific monitoring of mitochondrial transcription and respiration in PC-3 cells and biochemical assays studying the interactions with PolRMT. A more comprehensive assessment of mitochondrial toxicity would reduce the observation of previously unappreciated toxicity during clinical trials and, thus, aid in the successful development of new therapies.</p><!><p>Supplemental material for this article may be found at http://dx.doi.org/10.1128/AAC.01922-15.</p>

PubMed Open Access

Nanoscale Metal-Organic Layers for Deeply Penetrating X-ray Induced Photodynamic Therapy

We report the rational design of metal-organic layers (MOLs) that are built from [Hf6O4(OH)4(HCO2)6] secondary building units (SBUs) and Ir[bpy(ppy)2]+- or Ru(bpy)32+-derived tricarboxylate ligands (Hf-BPY-Ir or Hf-BPY-Ru; bpy = 2,2\xe2\x80\xb2-bipyridine, ppy = 2-phenylpyridine) and their applications in X-ray induced photodynamic therapy (X-PDT) of colon cancer. Heavy Hf atoms in the SBUs efficiently absorb X-rays and transfer energy to Ir[bpy(ppy)2]+ or Ru(bpy)32+ moieties to induce PDT by generating reactive oxygen species (ROS). The ability of X-rays to penetrate deeply into tissue and efficient ROS diffusion through ultrathin 2-D MOLs (\xe2\x88\xbc1.2 nm) enable highly effective X-PDT to afford superb anticancer efficacy.

nanoscale_metal-organic_layers_for_deeply_penetrating_x-ray_induced_photodynamic_therapy

<p>By combining three intrinsically nontoxic components--a photosensitizer (PS), light, and tissue oxygen--to generate cytotoxic reactive oxygen species (ROS), particularly singlet oxygen (1O2), photodynamic therapy (PDT) provides a highly effective phototherapy against cancer.[1] Because ROS indiscriminately kill both diseased and normal cells, it is critical to selectively localize PSs in tumors in order to enhance PDT efficiency and minimize collateral damage to normal tissues.[2-13] We and others reported a series of porphyrin-and chlorin-based nanoscale metal-organic frameworks (nMOFs) as PSs for effective PDT.[14-19] Because the lifetime of ROS is short, it is not feasible for all the species generated to diffuse out of the 3-D structure of nMOFs to exert cytotoxicity on cellular organelles, thus limiting the overall efficacy of PDT in vivo.</p><p>We hypothesized that the in vivo PDT efficacy of nMOFs could be further improved by reducing the dimensionality to afford 2-D metal-organic layers (MOLs) and exciting the MOLs with more tissue-penetrating X-rays. We recently reported a new class of tunable and functionalizable MOLs that are composed of [Hf6O4(OH)4(HCO2)6] secondary building units (SBUs) and benzene-1,3,5-tribenzoate (BTB) bridging ligands.[20] The 2-D structure of MOLs allow s ROS to diffuse freely, thus presenting an ideal platform for designing nanoscale PSs for efficient PDT.</p><p>Ir[bpy(ppy)2]+ (bpy = 2,2′-bipyridine, ppy = 2-phenylpyridine) and Ru(bpy)32+ are two efficient PSs with the very high 1O2 quantum yields (ΦΔ) of 0.97 and 0.73, respectively.[21-23] However, due to large Stoke shifts, they can only be excited with photons at short wavelengths, ∼355 nm for Ir[bpy(ppy)2]+ and ∼450 nm for Ru(bpy)32+. Such UV-Vis photons cannot penetrate human tissue (penetration depth <0.1 mm),[24] which severely limits their application in PDT. Our previous work demonstrated that a Hf-based nMOF can absorb X-rays and transfer energy to coordinated anthracene-based ligands to luminesce in the visible spectrum.[25] We believe that coordination between Ir[bpy(ppy)2]+- or Ru(bpy)32+-derived tricarboxylate ligands and heavy Hf-based SBUs would enable direct excitation of the PSs by X-rays to achieve X-ray induced photodynamic therapy (X-PDT).[26] Here we report the rational design of two MOLs, composed of [Hf6O4(OH)4(HCO2)6] SBUs and Ir[bpy(ppy)2]+- or Ru(bpy)32+-derived tricarboxylate ligands, as potent PSs. The Hf-MOLs achieve greatly enhanced PDT efficacy both in vitro and in vivo upon X-ray irradiation.</p><p>Hf-BPY-Ir and Hf-BPY-Ru MOLs were synthesized by a postsynthetic metalation method. 4′,6′-dibenzoato-[2,2′-bipyridine]-4-carboxylic acid (H3BPY) was synthesized as show n in Figure S1 (SI) and treated with HfCl4 in N,N-dimethylformamide (DMF), formic acid, and water to afford Hf-BPY MOL as a white precipitate, which was then washed twice with DMF and once with ethanol. By optimizing the amounts of formic acid and H2O, the size of Hf-BPY could be controlled to a diameter of ∼500 nm, as verified by transmission electron microscopy (TEM) (Figure 1a). Hf-BPY was treated with [Ir(ppy)2Cl]2/Ru(bpy)2Cl2 to afford Hf-BPY-Ir/Hf-BPY-Ru MOL as an orange/brow n participate. Due to the 2-D structure of Hf-BPY, the bpy coordination sites are highly accessible, resulting in efficient metalation. The Ir and Ru loadings were determined to be 67% and 59% for Hf-BPY-Ir and Hf-BPY-Ru, respectively, as determined by Inductively Coupled Plasma-Mass Spectrometry (ICP-MS).</p><p>In Hf-BPY, each Hf6 cluster of 12-connectivity was capped by 6 formate groups (three at the top and three at the bottom), leaving the remaining six sites coordinated to 3-connected BPY ligands to form a 3,6-connected 2-D network of Hf6(μ3-O)4 (μ3-OH)4(HCO2)6(BPY)2 of kagome dual (kgd) topology (Scheme 1). High-resolution TEM (HRTEM) images of Hf-BPY, where Hf6 clusters appear as dark spots, and fast Fourier transform (FFT) patterns (Figure 1b) of Hf-BPY were consistent with the kgd topology. The distance between two adjacent dark spots in the HRTEM was 2.0 nm, which matched the distance between two adjacent SBUs. The powder X-ray diffraction (PXRD) pattern of Hf-BPY was identical to that of the Hf-BTB MOL (Figure 2a),[20] which further confirmed the kgd structure of Hf-BPY. Atomic force microscopy (AFM) images (Figure 1e, 1f and S13) of Hf-BPY showed a 1.2 nm thickness, which is very close to the van der Waals size of the Hf6 cluster capped by formate groups, indicating the monolayer structure of Hf-BPY. A nitrogen sorption study of Hf-BPY gave a BET surface are of 346 m2/g (Figure S15), indicating its porous structure. The ultrathin monolayer structure facilitates the diffusion of 1O2 as its diffusion length was estimated to be 20-220 nm in cells.[27]</p><p>TEM imaging showed that Hf-BPY-Ir (Figure 1c) and Hf-BPY-Ru (Figure 1d) have similar morphologies and sizes as Hf-BPY. The retention of the MOL structure after metalation was supported by the similarity among the PXRD patterns of Hf-BPY-Ir, Hf-BPY-Ru, and Hf-BPY (Figure 2a). In addition, the HRTEM images and FFT patterns of Hf-BPY-Ir and Hf-BPY-Ru (Figure S10) were identical to those of Hf-BPY. TEM images and PXRD patterns of the MOLs remained unchanged after incubation in DMEM media for 12 h (Figures S16 and 17), suggesting that the MOLs are stable for biomedical applications.</p><p>To further confirm the metalation of Hf-BPY and to understand the coordination environments of Ir and Ru centers in Hf-MOLs, we synthesized [(H3BPY)Ir(ppy)2] Cl (H3BPY-Ir) and [(H3BPY)Ru(bpy)2] Cl2 (H3BPY-Ru) as ligand controls (Figure S6-S9). The UV-visible absorption spectra of Hf-based MOLs exhibit similar MLCT bands as their corresponding ligands (Figure 2b). Importantly, X-ray absorption spectroscopy (XAS) indicated that Zr-BPY-Ir and Hf-BPY-Ru have the same Ir and Ru coordination environments as H3BPY-Ir and H3BPY-Ru, respectively (Figure S20-S23). Due to similar energy between Ir L3-edge (11215 eV) and Hf L1-dege (11271 eV), XAS data was collected for Zr-BPY-Ir instead of Hf-BPY-Ir.</p><p>We next examined singlet oxygen generation efficiencies of MOLs using 4-nitroso-N,N-dimethylanaline (RNO) assay. We also synthesized Zr-MOLs (Zr-BPY-Ir and Zr-BPY-Ru) using similar processes (Figure S11 and S12) and used them for comparison. Upon irradiation with a Xe lamp using a 400 nm long-pass filter or X-rays (225 KVp, 13 mA), the 1O2 generated by MOLs reacted with RNO in the presence of histidine, leading to a decrease of absorbance at 440 nm in the UV-visible spectra (Figure S24 and S25). By linearly fitting difference in RNO peak absorbance [Δ(OD)] against irradiation doses (which scale linearly with exposure times upon visible light or X-ray dose, Y = Ax + B), the RNO assay provides a quantitative measure of 1O2 generation efficiencies, with a more positive slope indicating more efficient 1O2 generation. Upon visible light irradiation, the linear fitting results showed that Ir-based Zr- and Hf-MOLs generated 1O2 more efficiently than Ru-based Zr- and Hf-MOLs (Figures 2c, Table S3), consistent with the difference in 1O2 generation quantum yields between [Ir(bpy)(ppy)2]+ (φΔ = 0.97) and [Ru(bpy)3]2+ (φΔ = 0.73). Furthermore, only very slight differences were observed between two Ir-based MOLs (A = 1.09×10-2 for Hf-BPY-Ir and A = 0.88×10-2 for Zr-BPY-Ir) or two Ru-based MOLs (A = 4.1×10-3 for Hf-BPY-Ru and A = 2.4×10-3 for Zr-BPY-Ru), suggesting minor effects of the SBUs in the 1O2 generation efficiency through spin-orbit coupling. [14,28] The efficient singlet oxygen generation by Hf-BPY-Ir upon light irradiation was also demonstrated by Singlet Oxygen Sensor Green (SOSG) assay (Figure S26). Comparisons with DBP-Hf, a porphyrin-based nMOF reported by us previously, indicated that Hf-BPY-Ir is more effective in generating 1O2 than DBP-Hf upon light irradiation (Figure S27), suggesting the facile diffusion of ROS through the MOL monolayer.</p><p>However, upon X-ray irradiation, there was a drastic difference in 1O2 generation efficiencies in Zr- and Hf-MOLs (Figure 2d, Table S4). Both Hf-MOLs (A = 1.22×10-2 for Hf-BPY-Ir and A = 1.0×10-2 for Hf-BPY-Ru) possessed much higher 1O2 generation efficiency than their corresponding Zr-MOLs (A = 0.39×10-2 for Hf-BPY-Ir and A = 0.19×10-2 for Zr-BPY-Ir), supporting our hypothesis that the X-ray energy was first absorbed by SBUs and then transferred to the PSs in the bridging ligands to lead to the X-PDT effect. Because the heavier Hf atoms absorb X-rays more efficiently than the Zr atoms, the Hf-MOLs are expected to be more effective at X-PDT Additionally, Ir-based MOLs showed only slightly better X-PDT efficiency than Ru-based MOLs, suggesting different energy transfer processes involved in X-PDT and PDT</p><p>In the clinic, PDT is typically applied to superficial malignant tumors such as skin lesions and esophageal cancer due to the limited penetration of light (∼3 mm at 800 nm). We sought to examine the potential of MOL-mediated X-PDT in the treatment of deeply seated tumors. Two types of murine colon adenocarcinoma cells, CT26 and MC38, were used for in vitro and in vivo studies. The cellular uptake was evaluated on CT26 cells incubated with Hf-BPY-Ir, Hf-BPY-Ru, or Hf-BPY at a Hf concentration of 50 μM for 1, 4, 8, and 24 h. At each time point, cells were digested and the Hf contents were determined by ICP-MS. Hf-BPY-Ru showed higher uptake (6580 ± 1770 ng/105 cells) than Hf-BPY-Ir (3317 ± 665 ng/105 cells) and Hf-BPY (1930 ± 716 ng/105 cells), presumably because of the higher positive charge of Hf-BPY-Ru, which favors interacting with the negatively charged cell membrane to facilitate endocytosis (Figure S28).</p><p>We next investigated the in vitro anticancer efficacy of three different Hf-based MOLs against CT26 (Figure 3a) and MC38 (Figure 3b) cells. To elucidate the key role of Hf in efficient absorption of X-rays, three corresponding Zr-MOLs were used as controls. MOLs were incubated with cells at various concentrations for 8 h, followed by irradiation with an X-ray irradiator at a dose of 2 Gy. Hf-BPY-Ir and Hf-BPY-Ru outperformed Hf-BPY and three Zr-MOLs. The IC50 values for Hf-BPY-Ir, Hf-BPY-Ru, and Hf-BPY against CT26 cells were calculated to be 3.82 ± 1.80, 3.63 ± 2.75, and 24.90 ± 7.87 μM, respectively. Against MC38 cells, the IC50 values were 11.66 ± 1.84, 10.72 ± 2.92, and 37.80 ± 6.57 μM, respectively. IC50 values exceeded 100 μM for Zr-BPY-Ir, Zr-BPY-Ru, and Zr-BPY in both CT26 and MC38 cell lines. No cytotoxicity was observed in dark control groups (Figure S29). We also tested cell viability with fixed Hf-MOL concentrations based on Ir, Ru, or BPY of 20μM, respectively, and various X-ray doses (Figure S30). All of the results showed greatly enhanced X-PDT potency of Ir[bpy(ppy)2]+ and [Ru(bpy)3]2+ in Hf-MOLs. Importantly, X-ray induced cytotoxicity of Hf-BPY-Ir and Hf-BPY-Ru remained essentially unchanged when the cells were covered with a beef block of 1 cm in thickness during X-ray irradiation (Figure S41). In contrast, light induced cytotoxicity of Hf-BPY-Ir and Hf-BPY-Ru was completely lost when the cells were covered with the same beef block during light irradiation (Figure S40). These results support our hypothesis that MOL-mediated X-PDT can be used to treat deep-seated tumors. Interestingly, control experiments with Hf-QPDC-Ir nMOF, a UiO nMOF built from Hf6(μ3-O)4(μ3-OH)4 SBUs and QPDC-Ir bridging ligands, further support the enhanced X-PDT efficacy of the MOLs due to facile ROS diffusion; upon X-ray irradiation, Hf-QPDC-Ir nMOF exhibited much higher IC50 values of 32.85 ± 3.02μM for CT26 cells and 26.08 ± 2.38 μM for MC38 cells, respectively (Figure S39).</p><p>We then explored the mechanism of X-ray induced cytotoxicity on CT26 cells. 1O2 generation in live cells was examined by SOSG and detected by confocal laser scanning microscopy (CLSM) (Figure S31). After preloading cells with SOSG and incubating them with PBS, Hf-MOLs, or H3BPY ligand for 8 h at a concentration of 20 μM based on Ir, Ru, or BPY, respectively, they were irradiated with X-rays at a dose of 2 Gy, immediately followed by CLSM imaging. Both Hf-BPY-Ir-and Hf-BPY-Ru-treated cells showed strong green SOSG fluorescence, indicating the efficient generation of 1O2 in the MOLs upon X-ray irradiation. In contrast, PBS, Hf-BPY and H3BPY ligand-treated groups showed no SOSG signal after X-ray induced 1O2 generation, which supported our proposed X-PDT process using Hf-BPY-Ir and Hf-BPY-Ru MOLs. We also performed γ-H2AFX assay (Life technology, USA) on CT26 cells to determine DNA double-strand breaks (DSBs) caused by MOLs upon X-ray irradiation. As shown in Figure S37, CT26 cells treated with three Hf-based MOLs showed significant red fluorescence, indicating DSBs induced by hydroxyl radical from X-ray irradiation. This result suggests that Hf6 SBUs are capable of radiosensitization to further enhance cytotoxicity of MOL-mediated X-PDT.</p><p>Encouraged by in vitro results, we carried out in vivo anticancer efficacy experiments on subcutaneous flank tumor-bearing mouse models of CT26 and MC38. When tumors reached 100-150 mm3 in volume, Hf-BPY-Ir, Hf-BPY-Ru, or Hf-BPY with amount of 0.5 nmol based on Ir, Ru or BPY, respectively, or PBS was intratumorally injected followed by daily X-ray irradiation at a dose of 1 Gy/fraction (120 kVp, 20 mA, 2 mm-Cu filter) for a total of 5 fractions on the CT26 model (Figure 3c) or 10 fractions on the MC38 model (Figure 3d) on consecutive days. Tumor sizes and body weights were measured every day. All mice were sacrificed 18 days after tumor inoculation, and the excised tumors were photogaphed and weighed (Figure S33-S34). To rule out any radiotherapy effects from the low dose X-ray, we used PBS-treated mice without X-ray irradiation as a dark control. The PBS groups with or without irradiation did not show any difference in tumor growth curves, indicating that low dose X-rays alone had no radiotherapeutic effects. The Hf-BPY goups appeared to show slight inhibition of tumor growth (P=0.047 or 0.048 for CT26 or MC38, respectively), consistent with the radiosensitization effects of the Hf6 SBUs. In stark contrast, Hf-BPY-Ir and Hf-BPY-Ru treatments led to effective tumor regression in CT26 with 5 fractions of X-ray irradiation (5 Gy total; total volume reduction of 83.6% or 77.3%, respectively) and in MC38 with 10 fractions of X-ray irradiation (10 Gy; total total volume reduction of 82.3% or 90.1%, respectively). The weights and sizes of tumors treated with Hf-BPY-Ir and Hf-BPY-Ru at the end point were significantly smaller than the other groups (Table S5 and S6). Histology of frozen tumor slices confirmed MOL-assisted X-PDT caused apoptosis/necrosis in tumors (Figure S35). No abnormalities were observed on histological images of frozen organ slices, which indicated that X-PDT with intratumoral injection of MOLs was not systemically toxic (Figure S36). The lack of systemic toxicity was further supported by steady body weights and similar weight gain patterns in all groups (Figure S32).</p><p>In summary, we rationally designed and synthesized two Hf-MOLs as powerful PSs for effective X-PDT of colon cancer models. Upon X-ray irradiation, Hf atoms in the SBUs absorb X-rays and transfer energy to Ir[bpy(ppy)2]+ or [Ru(bpy)3]2+ in the ligands to generate 1O2, demonstrated by both RNO assay and in vitro 1O2 detection as well as cytotoxicity studies. As a result of deep tissue penetration of X-rays, high 1O2 quantum yields of Ir[bpy(ppy)2]+ or [Ru(bpy)3]2+, and efficient ROS diffusion through ultrathin MOLs, X-PDT treatment led to an impressive 90% reduction in tumor volumes. MOLs thus represent a novel class of 2-D materials with great potential for cancer treatment and other biomedical applications.</p>

PubMed Author Manuscript

A highly sensitive fluorimetric method for determination of lenalidomide in its bulk form and capsules via derivatization with fluorescamine

BackgroundLenalidomide (LND) is a potent novel thalidomide analog which demonstrated remarkable clinical activity in treatment of multiple myeloma disease via a multiple-pathways mechanism. The strong evidences-based clinical success of LND in patients has led to its recent approval by US-FDA under the trade name of Revlimid® capsules by Celgene Corporation. Fluorimetry is a convenient technique for pharmaceutical quality control, however there was a fluorimetric method for determination of LND in its bulk and capsules.ResultsA novel highly sensitive and simple fluorimetric method has been developed and validated for the determination of lenalidmide (LND) in its bulk and dosage forms (capsules). The method was based on nucleophilic substitution reaction of LND with fluorescamine (FLC) in aqueous medium to form a highly fluorescent derivative that was measured at 494 nm after excitation at 381 nm. The factors affecting the reaction were carefully studied and optimized. The kinetics of the reaction was investigated, and the reaction mechanism was postulated. Under the optimized conditions, linear relationship with good correlation coefficient (0.9999) was found between the fluorescence intensity and LND concentration in the range of 25–300 ng/mL. The limits of detection and quantitation for the method were 2.9 and 8.7 ng/mL, respectively. The precision of the method was satisfactory; the values of relative standard deviations did not exceed 1.4%. The proposed method was successfully applied to the determination of LND in its bulk form and pharmaceutical capsules with good accuracy; the recovery values were 97.8–101.4 ± 1.08–2.75%.ConclusionsThe proposed method is selective and involved simple procedures. In conclusion, the method is practical and valuable for routine application in quality control laboratories for determination of LND.

a_highly_sensitive_fluorimetric_method_for_determination_of_lenalidomide_in_its_bulk_form_and_capsul

Background<!>Apparatus<!>Reagents and materials<!>Lenalidomide standard solution<!>Capsules sample solution<!>General recommended procedure<!>Determination of stoichiometric ratio<!>Excitation and emission spectra<!><!>Effect of FLC concentration<!><!>Effect of pH<!><!>Effect of time<!><!>Effect of diluting solvent<!><!>Stability of the fluorescent derivative<!><!>Stoichiometry, kinetics and mechanism of the reaction<!><!>Stoichiometry, kinetics and mechanism of the reaction<!><!>Calibration and sensitivity<!><!>Accuracy and precision<!><!>Robustness and ruggedness<!>Practical applications of the proposed method<!>Advantages of the proposed method over the reported methods<!>Conclusions<!>Abbreviations<!>Competing interests<!>Authors’ contributions

<p>Lenalidomide (LND) is a potent novel thalidomide analog which demonstrated remarkable clinical activity in treatment of multiple myeloma disease [1-5] via a multiple-pathways mechanism [6-9]. The strong evidences-based clinical success of LND in patients has led to its recent approval by US-FDA under the trade name of Revlimid® capsules by Celgene Corporation [10]. LND has an improved side effects profile than its parent compound thalidomide [11]. These side effects can be managed by combination therapy and/or careful dose adjustment [12]. The therapeutic benefits profile of LND is anticipated to encourage the development of new pharmaceutical preparations for LND. As a consequence, there is an increasing demand for proper analytical technologies for quality assurance of LND formulations.</p><p>Few methods have been reported for the determination LND in bulk material and in capsules. These methods included two spectrophotometric methods [13]. The first method was based on diazo-coupling reaction with N-(1-napthyl) ethylenediamine dihydrochloride and the second method was based on the formation of a colored condensation product with p-dimethylaminocinnamaldehyde. In addition, two HPLC methods have reported for analysis of bulk material of LND and its related impurities [14] and capsules [15]. These methods were associated with some major drawbacks such as lack of selectivity, time-consumption and/or use of expensive instruments.</p><p>Fluorimetry is considered one of the most convenient analytical techniques, because of its inherent simplicity, high sensitivity, low cost, and wide availability in most quality control laboratories. No attempt has yet been made for the fluorimetric determination of LND. The present study describes, for the first time, the development of a novel highly sensitive and simple fluorimetric method for the determination of LND in its bulk form and capsules. The method was based on the derivatization of LND with fluorescamine (FLC) in aqueous medium to produce a highly fluorescent product that was measured fluorimetrically at 494 nm after excitation at 381 nm.</p><!><p>Fluorescence measurements were carried out on a RF-5301 PC spectrofluorimeter (Shimadzu Corporation Kyoto, Japan) equipped with a 150 W xenon lamp and 1 cm quartz cells. The slit widths of both the excitation and emission monochromators were set at 1.5 nm. The calibration and linearity of the instrument were frequently checked with standard quinine sulphate. pH meter Model 211 a product of HANNA Instruments Inc. (Smithfield, RI, USA).</p><!><p>Lenalidomide (LND), free base (3-(4`-aminoisoindoline-1`-one)-1-piperidine-2,6-dione) (LC Laboratories®, Woburn, MA, USA ) was obtained and used as received; its purity was 100.2 ± 1.25%. Fluorescamine (FLC; Sigma Chemical Co., St. Louis, USA) was prepared in acetonitrile to contain 0.025% (w/v); the solution could be used for seven days when kept in the refrigerator. Revlimid® capsules (Celgene Corporation, New Jersy, USA) labeled to contain 5 mg LND per capsule was obtained from the local market. Double distilled water was obtained through WSC-85 water purification system (Hamilton Laboratory Glass Ltd., Kent, USA), and used throughout the work. All other solvents and materials used throughout this study were of analytical grade.</p><!><p>An accurately weighed amount (25 mg) of LND was quantitatively transferred into a 25-mL calibrated flask, dissolved in 20 mL methanol, completed to volume with the same solvent to obtain a stock solution of 1 mg/mL. This stock solution was further diluted with water to obtain a working stock solution containing 0.5 μg/mL.</p><!><p>The contents of 20 Revlimid® capsules (Celgene Corporation, NJ, USA), labeled to contain 5 mg of LEN per capsule were evacuated and weighed. An accurately weighed portion equivalent to 50 mg of LND was transferred into a 50-mL calibrated flask containing ~ 40 mL of methanol. The contents of the flask were swirled, sonicated for 5 min, and then completed to volume with methanol. The contents were mixed well and filtered rejecting the first portion of the filtrate. The prepared solution was diluted quantitatively with distilled water to obtain a suitable concentration for the analysis.</p><!><p>Accurately measured aliquots of LND working stock solution (0.5 μg/mL) were transferred into separate 10-mL calibrated flasks to obtain a series of LND standard solutions covering the working range of 25–300 ng/mL in the final solution. One milliliter of FLC solution (0.025% w/v) was added to each flask. The reaction was allowed to proceed at room temperature (25 ± 2°C) for 10 min, and then completed to volume with water. The fluorescence intensity of the resulting solutions were measured at 494 nm after excitation at 381 nm against a reagent blank prepared in the same manner but using water instead of the LND working stock solution.</p><!><p>The Job's method of continuous variation [16] was employed. Master equimolar (1.5×10−5 M) aqueous solutions of LND and FLC were prepared. Series of 10-mL portions of the master solutions of LND and FLC were made up comprising different complementary proportions (0:10, 1:9, . . ., 9:1, 10:0, inclusive) in 10-mL calibrated flasks and the reactions were allowed to proceed for 10 min. The solutions were further manipulated and the fluorescence signals were measured as described under the general recommended procedure.</p><!><p>Because of the absence of native fluorescence of LND, its derivatization with fluorogenic reagent was necessary for its fluorimetric determination. FLC was chosen as a derivatizing reagent because it forms highly fluorescent derivatives with primary amines under relatively mild reaction conditions [17]. It was found that LND reacts with FLC and forms a highly fluorescent derivative that exhibited maximum fluorescence intensity (λem) at 494 nm after excitation at wavelength (λex) of 381 nm. The excitation and emission spectra for the reaction product of LND with FLC are given in Figure 1.</p><!><p>Excitation (1) and emission (2) spectra of the reaction product of LND (275 ng/mL) with FLC (0.025%, w/v).</p><!><p>The study of the reaction between LND and FLC revealed that the reaction was dependent on the FLC concentration as the relative fluorescence intensity (RFI) of the reaction mixture increased steadily as the FLC concentration increased up to a final concentration of 0.002%, w/v (Figure 2). Beyond this concentration, the slope of the curve significantly decreased. For more precise readings, a concentration of 0.025% (w/v) of FLC reagent solution was used throughout the further experiments.</p><!><p>Effect of FLC concentration on its reaction with LND (75 ng/mL).</p><!><p>The effect of the pH on the reaction was studied by carrying out the reaction in borate buffer solution in the pH range of 6.5–9.5. The results indicated that the RFI increased initially as the pH increased and maximum readings were attained at pH 7.0 ± 0.2 (Figure 3). In previous studies involving FLC as a fluorofore, the maximum readings were obtained at pH around 8.0 [18]. This was possibly due to the predominance of the free amino group of the investigated substance rather than its salt form in acidic pH. Consequently, this facilitates the nucleophilic substitution reaction. In the present study, such alkaline pH was not necessary because the LND is already in the form of free base. Furthermore, at higher pH values, sharp decrease in the readings occurred (Figure 3). This was probably attributed to the hydrolysis of the reaction product between LND and FLC in alkaline medium. Neutral pH was found to be optimum for the reaction between LND and FLC. Distilled water was compared with borate buffer of pH 7, and similar results were obtained. Therefore, the reactions in all the subsequent experiments were carried out in distilled water. This was in favor of the simplicity of the proposed procedure, and environmental and health safety.</p><!><p>Effect of pH on the reaction of FLC (0.025%, w/v) with LND (130 ng/mL).</p><!><p>In order to determine the optimum time required for completion of the reaction, the derivatization reaction was carried out at room temperature (25 ± 2°C) and the induced fluorescence signals were measured immediately after the addition of FLC and monitored for 30 min. The optimum reaction time was considered as the time at which the highest fluorescence signals with reproducible results are obtained in a comfortable measurement region on the FI-time curve (wide plateau). The results indicated that the reaction was very fast and almost completed within 5 min (Figure 4). Beyond this time, the RFI values did not change by time. For comfortable readings with high precise results, all the subsequent reactions were carried out for 10 min.</p><!><p>Effect of time on the reaction of FLC (0.025%, w/v) with LND (130 ng/mL).</p><!><p>In order to select the most suitable diluting solvents for the formation and stability of the reaction product, different solvent were investigated. These solvents were: water, methanol, ethanol, and acetonitrile. The highest fluorescence intensities were obtained when water was used as a diluting solvent (Figure 5).</p><!><p>Effect of diluting solvent on the reaction between LND (130 ng/mL) and FLC (0.025%, w/v). Solvents were: water, methanol (MeOH), ethanol (EtOH), and acetonitrile (MeCN).</p><!><p>The effect of time on the stability of the LND-FLC fluorescent derivative was studied by monitoring the fluorescence intensities of the reaction solution (after dilution) at different time intervals. It was found that the RFI values remain constant for at least 1 hour. This allowed the processing of large batches of samples, and their comfortable measurements with convenience. This increased the convenience of the method as well as made it applicable for large number of samples in quality control laboratories.</p><p>A summary for the optimization of the variables affecting the reaction of LND with FLC is given in Table 1.</p><!><p>Optimization of variables affecting the reaction of LND with FLC</p><p>a Solvents tested: Water, methanol, ethanol, and acetonitrile.</p><p>b The stability of the LND-FLC was studied after dilution of the reaction solution.</p><!><p>Under the optimum conditions (Table 1), the stoichiometry of the reaction between LND and FLC was investigated by Job's method [16]. The symmetrical bell-shape of Job's plot (Figure 6) indicated that the LND:FLC ratio was 1:1. Based on this ratio, the reaction pathway was postulated to be proceeded as shown in Figure 7.</p><!><p>Job's plot for the reaction between LND and FLC.</p><p>Scheme for the reaction pathway between LND and FLC.</p><!><p>Under the optimum conditions (Table 1), the RFI-time curves for the reaction at varying LND concentrations (1.93×10–7 – 1.16×10–6 M)] with a fixed concentration of FLC [9×10–5 M)] were generated (Figure 8). The initial reaction rates (K) were determined from the slopes of these curves. The logarithms of the reaction rates (Log K) were plotted as a function of logarithms of LND concentrations (log C); Figure 9. The regression analysis for the values was performed by fitting the data to the following equation:</p><p>(1)Log K=log K′+nlogC</p><p>where K is reaction rate, K′ is the rate constant, C is the molar concentration of LND, and n (slope of the regression line) is the order of the reaction. As seen in Figure 9, a straight line with slope values of 0.9946 (≈ 1) confirming that the reaction was first order. However under the optimized reaction conditions, the concentration of FLC was in much more excess than that of LND in the reaction solution. Therefore, the reaction was regarded as a pseudo-first order reaction.</p><!><p>Relative fluorescence intensity (RFI)-time curves for the reaction of a fixed concentration of FLC with varying concentrations of LND. LND concentrations were 45 (▽), 95 (♦), 130 (○), 185 (▲), and 260 (□) ng/mL.</p><p>Linear plot for Log C versus Log K for the kinetic reaction of LND and FLC.</p><!><p>Under the optimum conditions (Table 1), calibration curve for the determination of LND by its reaction with FLC was constructed by plotting the RFI as a function of the corresponding LND concentration. The regression equation for the results was: RFI = a + bC, where RFI is the relative fluorescence intensity, C is the concentration of LND in ng/mL. Linear relationship with small intercept and excellent correlation coefficient (r = 0.9999) was obtained in the range of 25–300 ng/mL. The limit of detection (LOD) and limit of quantification (LOQ) were determined according to ICH guidelines for validation of analytical procedures [19]. The LOD and LOQ values were found to be 2.9 and 8.7 ng/mL, respectively. The parameters for the analytical performance of the proposed fluorimetric method are summarized in Table 2.</p><!><p>Statistical parameters for the determination of LND by the proposed fluorimetric method based on its reaction with FLC</p><!><p>The accuracy and precision of the proposed fluorimetric method was determined by replicate analysis of five different concentrations of the working standard. The recovery values were 97.8-101.4 ± 1.08 - 2.75% (Table 3), indicating the accuracy of the proposed method. The average recovery from all the concentrations was found to be 99.5% with SD of 1.40% indicating the good accuracy and reproducibility of the results. Furthermore intra- and inter-day precisions for determination of LND in bulk powder were assessed at three varying LND concentrations (low, medium, and high). The average recovery values were 101.40 and 102.27% with RSD values of 1.13 and 2.29% for intra- and inter-assay precision, respectively (Table 4). These good recovery values and low RSD values revealed the high accuracy and precisions, respectively.</p><!><p>Recovery studies for determination of LND by the proposed fluorimetric method based on its reaction with FLC</p><p>a Values are mean of three determinations.</p><p>Intra-assay and inter-assay precision and accuracy for determination of LND by the proposed fluorimetric method</p><p>a Values are mean of five determinations.</p><p>b Values are mean of four determinations.</p><!><p>Robustness was examined by evaluating the influence of small variation in the method variables on its analytical performance. In these experiments, one parameter was changed whereas the others were kept unchanged, and the recovery percentage was calculated each time. It was found that variation in the FLC concentrations (0.02–0.03%, w/v), temperature (optimum ± 2°C), pH (6.8 - 7.2) and time (optimum ± 5 min) did not significantly affect the recovery values. The most critical factor affecting the results was the FLC concentration and therefore it had to be added accurately to attain a fixed concentration in all the solutions. Ruggedness was also tested by applying the method to the assay of LND using fixed operational conditions within ± 10% changes and on different days. The results were reproducible and the RSD% did not exceed 2.29%.</p><!><p>It is evident from the above-mentioned results that the proposed method gave satisfactory results with LND in bulk form. Also the pharmaceutical dosage forms (Revlimid® capsules) were analyzed for their LND content by the proposed method. The percentage found from the label claim was 100.11 ± 1.61% of the label claim, indicating the successful applicability of the proposed method in quality control laboratories for determination of LND. The results obtained by the proposed method was compared with those obtained by a reported method [15] with respect to the accuracy (by t-test), and precision (by F-test). It was found that the calculated t- and F-values were lower than the tabulated ones. This indicated that there were no significant differences between the means and variance between the two methods in terms of the accuracy and precision.</p><!><p>This study represents the first report describing the successful evaluation of FLC as an analytical reagent in the development of a highly sensitive and simple fluorimetric method for the quantitative determination of LND. The proposed method is superior to the previously reported spectrophotometric methods in terms of the sensitivity and simplicity of the derivatization procedures. As well, the proposed procedure used water as a green, inexpensive, and safe solvent, rather than the costive and toxic organic solvents that have been employed in the previously reported HPLC methods. In addition, the method employed a simple inexpensive fluorimeter that is available in most quality control laboratories, rather than the expensive HPLC systems.</p><!><p>A novel simple and sensitive fluorimetric method for the determination of LND in bulk form and capsules has been successfully developed and validated. The method involved simple derivatization of LND with FLC reagent, and subsequent measurement of the fluorescence intensity of the fluorescent reaction product. The proposed method is specific, accurate, reproducible, and highly sensitive to be applied on the analysis of bulk form and capsules. Furthermore, the analysis requires a simple apparatus, thus the proposed method is suitable for routine analysis of LND in quality control laboratories.</p><!><p>LND: Lenalidomide; FLC: Fluorescamine; λex: Excitation wavelength; λem: Emission wavelength; RFI: Relative fluorescence intensity; ICH: The international Conference on Harmonization; LOD: Limit of detection; LOQ: Limit of quantification; SD: Standard deviation; RSD: Relative standard deviation.</p><!><p>The authors declare that they have no conflict of interests.</p><!><p>IAD proposed the subject, designed the study, participated in the results discussion and revised the manuscript. NYK participated in the assay design, conducted the validation of the assay, and participated in preparing the manuscript. AHB conducted the optimization of the assay conditions and prepared the draft version of the manuscript. NZA participated in study design, literature review, assay validation and preparing the manuscript. All authors read and approved the final manuscript.</p>

PubMed Open Access

Metabolic Regulation of Histone Acetyltransferases by Endogenous Acyl-CoA Cofactors

SUMMARY The finding that chromatin modifications are sensitive to changes in cellular cofactor levels potentially links altered tumor cell metabolism and gene expression. However, the specific enzymes and metabolites that connect these two processes remain obscure. Characterizing these metabolic-epigenetic axes is critical to understanding how metabolism supports signaling in cancer, and developing therapeutic strategies to disrupt this process. Here, we describe a chemical approach to define the metabolic regulation of lysine acetyltransferase (KAT) enzymes. Using a novel chemoproteomic probe, we identify a previously unreported interaction between fatty acyl-CoAs and KAT enzymes. Further analysis reveals that palmitoyl-CoA is a potent inhibitor of KAT activity and that fatty acyl-CoA precursors reduce cellular acetylation levels. These studies implicate fatty acyl-CoAs as endogenous regulators of histone acetylation, and suggest novel strategies for the investigation and metabolic modulation of epigenetic signaling.

metabolic_regulation_of_histone_acetyltransferases_by_endogenous_acyl-coa_cofactors

Introduction<!>A Sensitive Chemical Proteomic Method to Assess KAT Active Site Occupancy<!>Competitive Chemical Proteomic Profiling of Acyl-CoA/KAT Interactions<!>Biochemical Analysis of Fatty Acyl-CoA/KAT Interactions<!>Modulating Fatty Acyl-CoA Biosynthesis Can Affect Histone Acetylation in Cells<!>Discussion<!>Compounds, enzymes, and materials<!>Chemical proteomic analysis of KAT/acyl-CoA interactions<!>Continuous assay for lysine peptide acetylation (coupled-enzyme assay)<!>Separation-based assay for lysine peptide acetylation (microfluidic mobility shift)<!>Docking studies of palmitoyl-CoA/Gcn5 complex<!>Cellular analyses of histone acetylation

<p>Lysine acetylation plays a critical role in regulating chromatin structure. By neutralizing the positive charge of histone tails, acetylation serves to relax histone-DNA interactions and allows trans-acting factors to access genomic chromatin (Roth et al., 2001). Lysine acetylation also provides binding sites for effector proteins known as bromodomains, which can directly stimulate transcription by recruiting coactivators (Dhalluin et al., 1999). Global reductions in histone acetylation are correlated with aggressive disease and poor clinical outcome in many cancers (Seligson et al., 2009; Seligson et al., 2005), and small molecules that counteract this profile and restore acetylation are validated therapeutic agents (Marks and Breslow, 2007). Defining the cellular mechanisms that regulate acetylation is thus of critical importance to understanding the biology of cancer and developing novel strategies to combat disease.</p><p>Protein acetylation is established by the opposing functions of lysine acetyltransferase (KAT) and lysine deacetylase (KDAC) enzymes. In addition to their role in transcription, the activity of these enzymes appears to be intimately linked to the metabolic state of the cell (Meier, 2013). For example, KDAC activity can be modulated by endogenous inhibitors such as diet-derived short-chain fatty acids and ketone bodies (Donohoe et al., 2012; Shimazu et al., 2013). By comparison, less is known about the metabolic mechanisms that influence KAT activity. Disrupting production of the KAT cofactor acetyl-CoA has been shown to inhibit histone acetylation (Comerford et al., 2014; Wellen et al., 2009). However, since all characterized human KATs exhibit Michaelis constants for acetyl-CoA far below its estimated cellular concentration (Tanner et al., 2000a; Thompson et al., 2001), it has been proposed that rather than becoming inherently rate-limiting, low acetyl-CoA levels make KATs more susceptible to inhibition by CoA, an endogenous feedback inhibitor (Albaugh et al., 2011; Lee et al., 2014). Evidence from biochemical analyses and studies in yeast suggest the GCN5 family of KATs may be particularly susceptible to this mechanism of regulation, and thus serve as critical integrators of metabolic and epigenetic signals (Cai et al., 2011; Langer et al., 2002).</p><p>The proposed metabolic regulation of KAT activity by CoA led us to consider whether other endogenous inhibitors of these enzymes may exist. Cells contain a diverse repertoire of acyl-CoAs (i.e. malonyl-, succinyl-, butyryl-, propionyl-, crotonyl-, palmitoyl-CoA). Due to their function as key intermediates in different bioenergetic pathways, the concentration of these molecules directly reflects the metabolic state of the cell. Notably, while diverse lysine acylations have been characterized (Lin et al., 2012), no KAT enzyme has yet been discovered that can utilize these "alternative" acyl-CoA cofactors at rates comparable to acetyl-CoA. In contrast, extended CoA analogues capable of making high-affinity bisubstrate interactions with KATs are well known as inhibitors of acetylation (Lau et al., 2000). Therefore, we hypothesized that metabolic acyl-CoAs may serve as endogenous bisubstrate inhibitors of KAT enzymes (Figure 1). Metabolic modulation of KAT activity by acyl-CoAs may provide cells with a mechanism to integrate changes in metabolic state and histone acetylation, and potentially fine-tune gene expression under conditions of nutrient stress.</p><p>To investigate this hypothesis, here we describe a chemical proteomic approach to define acyl-CoA/KAT interactions in complex proteomes. This approach enables the rapid, direct, and quantitative study of acyl-CoA/KAT interactions in their native contexts, is applicable to multiple KAT family members, and does not require reconstitution. Applying this approach, we identified a previously unreported palmitoyl-CoA/KAT interaction. Biochemical profiling reveals palmitoyl-CoA and other fatty acyl-CoAs can inhibit KAT activity, with different KAT subfamilies displaying distinct in vitro sensitivities to this inhibition. Adding a palmitoyl-CoA precursor to cells or overexpressing fatty acyl-CoA biosynthetic enzymes reduces cellular histone acetylation. These studies suggest fatty acyl-CoAs may constitute a novel class of endogenous KAT inhibitors, and propose novel strategies for the metabolic modulation of KAT activity.</p><!><p>A major challenge to studying the metabolic regulation of KAT activity is that their biologically relevant forms are multiprotein complexes that are difficult to reconstitute in vitro (Roth et al., 2001). Testing whether a metabolite interacts strongly with one or more of these KAT complexes is thus a non-trivial task. To address this challenge, we developed a chemoproteomic approach to study the active site occupancy of intact, multiprotein KAT complexes. This strategy uses an enrichable active site probe to detect catalytically competent KATs directly from cell proteomes (Figure 2). In order to enable the study of the metabolically sensitive KAT enzymes Gcn5 and pCAF, the probe "core" incorporates H3K14-CoA, a known high affinity inhibitor of these two enzymes (Lau et al., 2000). For our affinity purification element we introduced a straightforward N-terminal biotin, designed to enable co-purification and identification of KAT complex members, and increase the sensitivity of KAT detection by circumventing potentially low-yielding photocrosslinking steps (Montgomery et al., 2014).</p><p>In pilot LC-MS/MS analyses, HeLa cell proteomes incubated with H3K14-CoA-biotin 1 (10 μM), and subjected to streptavidin-enrichment and a mild wash step observed enrichment of Gcn5, as well as several members of its associated multiprotein STAGA complex (Figure 2; Table S1). We also detected enrichment of Mof, a member of the MYST KAT family, including several proteins associated with its NSL complex. Interestingly, co-purification of members of other Mof-containing complexes (MSL and MLL) was not observed. For both KAT complexes, enrichment was competed by addition of an active site directed ligand (Figure 2). This indicates probe 1 is capable of profiling the active site occupancy of KATs within their endogenous multiprotein complexes, and represents a major leap in sensitivity over our previous efforts (Montgomery et al., 2014). In addition to Gcn5, the human GCN5 KAT family contains another member, pCAF, that is 97.8% identical in the KAT catalytic domain but is present at comparably lower levels in HeLa cell extracts. Theoretically pCAF should be amenable to active site directed enrichment by probe 1, but was not observed in our LC-MS/MS datasets. To test whether immunoblotting could enable the active site mediated analysis of low abundance KATs not observable by LC-MS/MS, we subjected unfractionated HeLa cell lysates to a single round of affinity enrichment by 1, followed by immunoblotting with an anti-pCAF antibody. Indeed, we found pCAF was enriched by probe 1, in a manner that was sensitive to active site competition. This likely reflects the signal amplification afforded by chemiluminescent detection strategies, and highlights immunoaffinity profiling as an ideal approach for the targeted quantification of KAT active site engagement.</p><!><p>We next assessed whether probe 1 could be applied in a competitive chemoproteomic approach to study the metabolic sensitivity of native, cellular KAT enzymes (Leung et al., 2003). In this experiment, blockade of KAT active sites by a competing acyl-CoA metabolite results in reduced enrichment of cellular KATs (Gcn5, pCAF, or Mof), impeding their subsequent detection (Figure 3). Thus, in parallel, HeLa cell lysates were incubated with a variety of CoA metabolites (CoA, acetyl-CoA, propionyl-CoA, butyryl-CoA, malonyl-CoA, succinyl-CoA, crotonyl-CoA, and palmitoyl-CoA; Figure S1a), followed by H3K14-CoA-biotin 1 enrichment and anti-KAT immunoblotting.</p><p>We observed an overall trend of metabolic acyl-CoAs inhibiting pulldown of Mof > Gcn5 > pCAF (Figure 3). Since 1 likely binds to these three KATs with differing affinities, it is logical that each KAT may display different intrinsic sensitivities to competition in this experiment. Therefore, we focused our analysis on the quantitative rank order trends of metabolite binding to a single KAT, rather than absolute magnitude of competition between different KAT family members. The positive control acetyl-CoA efficiently blocked enrichment of all three KATs, consistent with its known strong interaction with these enzymes (Figure 3). Competitive chemoproteomic analysis demonstrated the feedback inhibitor CoA also interacted strongly with all three active sites, and was one of the strongest metabolic inhibitors tested. Interestingly, Gcn5 and pCAF interacted more strongly with their feedback inhibitor (CoA) than their cofactor (acetyl-CoA), a finding that was distinct from Mof (Figure 3). This provides further evidence that GCN5 family members may be particularly sensitive to changes in the acetyl-CoA:CoA ratio, as has been previously proposed (Albaugh et al., 2011; Cai et al., 2011; Lee et al., 2014; Wellen et al., 2009).</p><p>Next we shifted our focus to investigating whether any metabolic acyl-CoAs are capable of interacting strongly with KATs, potentially indicating their ability to serve as alternative cofactors or endogenous inhibitors of these enzymes. Propionyl- and butyryl-CoA are known to be able to function as alternative cofactors for Gcn5 and pCAF (Leemhuis et al., 2008; Montgomery et al., 2014), and strongly competed their proteomic enrichment. Mof also interacted strongly with these C3/C4 acyl-CoAs, and propionyl-CoA antagonized Mof enrichment to a similar degree as its natural substrate acetyl-CoA. The ability of propionyl-CoA to function as a cofactor or inhibitor of Mof has not been reported (Chen et al., 2007). Malonyl-, crotonyl-, and succinyl-CoA did not interact strongly with Gcn5 or pCAF (<50% inhibition of pulldown), while Mof was observed to interact weakly with malonyl-CoA. However, we found that palmitoyl-CoA competitively inhibited enrichment of all three KAT enzymes (>50% inhibition of pulldown; Figure 3). Relative to acetyl-CoA, palmitoyl-CoA most strongly antagonized Gcn5 pulldown, followed by Mof and pCAF. In addition to these KAT catalytic domains, palmitoyl-CoA also antagonized enrichment of the STAGA/PCAF component TRRAP (Figure S1b). This indicates palmitoyl-CoA is able to block the interaction of H3K14-CoA-biotin 1 with a multiprotein KAT complex, presumably by directly competing for the enzyme active site. Control experiments showed palmitoyl-CoA was not significantly hydrolyzed during these experiments (Figure S2), suggesting competition was due to palmitoyl-CoA itself, and not a downstream effector or byproduct. Gcn5, pCAF, and Mof are all known to inefficiently utilize extended acyl-CoA analogues (Chen et al., 2007; Yang et al., 2013), and our own studies provided no evidence these KATs were capable of using palmitoyl-CoA as a substrate (vide infra). These observations, coupled with the potential significance of a link between transcriptional regulation and fatty acid metabolism, led us to further investigate palmitoyl-CoA as a potentially novel, endogenous KAT inhibitor.</p><!><p>While the inhibition of KAT activity by palmitoyl-CoA has not been previously reported, fatty acyl-CoAs and fatty acyl-CoA-binding proteins have been found in the nuclei of eukaryotic cells (Elholm et al., 2000; Huang et al., 2004; Ves-Losada and Brenner, 1996). Furthermore, these species have been proposed to regulate the activity of transcription factors (Hertz et al., 1998), kinases, and metabolic enzymes (Jenkins et al., 2011; Yang et al., 2005). To extend these observations to KATs, we first used a known crystal structure of the metabolically sensitive KAT Gcn5 to model the human palmitoyl-CoA/Gcn5 complex (Figure 4a) (Poux et al., 2002). The natural cofactor acetyl-CoA binds Gcn5 via hydrogen bonds between the oxygen atoms of the pyrophosphate moiety and the backbone amide nitrogens of Val587, Lys588, Gly589, Thr592 and Lys624, as well as between the phosphopantetheine arm and backbone amides of Cys579 and Val581. Docking suggests palmitoyl-CoA has the capacity to make identical interactions with the Gcn5 cofactor binding site, potentially anchoring the fatty acyl-chain and allowing it to make additional interactions with hydrophobic surfaces in the substrate-binding site similar to synthetic bisubstrate inhibitors (Figure 4a).</p><p>In order to biochemically characterize the effect of palmitoyl-CoA on KAT activity, we applied a coupled-enzyme assay to monitor histone peptide acetylation. To directly compare chemoproteomic and in vitro biochemical methods, we analyzed metabolic acyl-CoAs for inhibition of Gcn5 at a single concentration (20 μM). Consistent with our chemoproteomic analysis, significant differences (P < 0.05) were seen between palmitoyl-CoA and the weak interactors crotonyl-, malonyl-, and succinyl-CoA (Figure S3). No turnover was observed in the absence of acetyl-CoA, indicating palmitoyl-CoA is not used by Gcn5 as a substrate. Further analysis revealed that palmitoyl-CoA potently inhibits recombinant Gcn5 (Ki = 630 ± 80 nM) at levels far below its critical micellar concentration (~70 μM) (Constantinides and Steim, 1985). Palmitoyl-CoA inhibition best fit a model that was competitive with regards to acetyl-CoA (Figure 4b). Furthermore, free palmitic acid did not affect Gcn5 activity (Figure S4), demonstrating the C16 fatty acid must be covalently anchored to CoA to mediate its inhibitory effect. To evaluate specificity, we studied the effects of palmitoyl-CoA on the prototypical KAT p300. Palmitoyl-CoA inhibits p300-catalyzed acetylation, although less potently than Gcn5 (Ki = 8750 ± 1900 nM). Furthermore, we benchmarked palmitoyl-CoA's inhibitory potency against desulfo-CoA, a desulfurized CoA analogue that has previously been used to study feedback inhibition of Gcn5 and is compatible with KAT coupled-enzyme assays (Tanner et al., 2000b). Desulfo-CoA inhibited Gcn5 less potently than palmitoyl-CoA (Ki = 1220 ± 190 nM), highlighting palmitoyl-CoA as one of the most potent endogenous KAT-inhibiting metabolites identified to date.</p><p>To confirm these results and better understand the structural requirements for KAT inhibition, we next screened CoA, palmitoyl-CoA, and related fatty acyl-CoAs for their effect on Gcn5 activity using an orthogonal, separation-based assay (Fanslau et al., 2010). This assay monitors the ability of KATs to acetylate a fluorescent peptide substrate, which can be separated from non-acetylated peptides in a charged field and thus provide a direct and quantitative readout of KAT activity. While not amenable to the determination of biochemical inhibition constants, this assay does enable the direct comparison of the inhibitory potency of CoA and palmitoyl-CoA (CoA itself is technically incompatible with the coupled-enzyme method). Separation-based analyses of KAT activity confirmed that palmitoyl-CoA inhibited Gcn5 more potently than CoA (66% inhibition versus 29% inhibition at 10 μM, respectively). Expansion of this method to analyze a fatty acyl-CoA panel found Gcn5 was inhibited more potently by C18 fatty acyl-CoAs (linoleoyl- and oleoyl-CoA) than C14/C16 (Figure 4c). Beyond these differences in magnitude of inhibition, the larger general trend observed was that all fatty acyl-CoAs, regardless of chain length or unsaturation, had the net effect of inhibiting KAT activity (Figure 4c). The high number of rotatable bonds in palmitoyl-CoA limit rigorous computational treatment, and a definitive model of palmitoyl-CoA/Gcn5 binding will require structural analyses. However, docking studies support the finding that Gcn5 can bind a variety of fatty acyl-CoAs, as the majority of contacts of the fatty acyl chain are made generically with hydrophobic surfaces in the shallow substrate-binding groove of the enzyme, rather than in a deep, specific binding pocket (Figure 4a). The ability of Gcn5 to promiscuously interact with fatty acyl-CoAs is similar to the broad spectrum binding exhibited by fatty acid binding transcription factors (Lin et al., 1999). These data also suggest that overall cellular fatty acyl-CoA content, rather than any individual fatty acyl-CoA signal, may be the most important determinant for inhibition of cellular KAT activity by this mechanism.</p><!><p>Determining whether fatty acyl-CoAs can act as endogenous inhibitors of KATs in cells is challenging due to many factors, including the lack of cell-permeability of these molecules, the central role of fatty acyl-CoAs to many cell processes, and the difficulty of quantifying subcellular fatty acyl-CoA concentrations. Indeed, the overall nuclear/cytosolic concentration of fatty acyl-CoAs, which our biochemical analyses indicate may be a key regulator of KAT activity, have not been quantitatively determined for any cell line. Nevertheless, the inhibition constants measured here are lower than the Km values that many long chain acyl-CoA metabolizing enzymes exhibit for palmitoyl-CoA (~1 μM) (Powell et al., 1985), and fatty acyl-CoA concentrations in the micromolar range have been demonstrated to modulate the activity of many proteins including glucokinase (Tippett and Neet, 1982), protein kinase C (Majumdar et al., 1991), carnitine palmitoyltransferase (Agius et al., 1987), and the nuclear thyroid hormone receptor (Li et al., 1990). This suggests the observed inhibition of KATs by fatty acyl-CoAs may be biologically relevant, and led us to investigate the relationship between fatty acyl-CoA biosynthesis and histone acetylation in cells.</p><p>As mentioned above, fatty acyl-CoAs are not directly cell permeable. Therefore, we first evaluated the effect of palmitate, a palmitoyl-CoA precursor, on cellular histone acetylation. In HEK-293 cells, we found that supplementation of media with palmitate (0.1 mM) led to a slight, but measureable, decrease in several histone marks, including H3K9Ac, H3K14Ac, H3K27Ac, and H4K8Ac (Figure 5). However, this effect is small, and could be either due to stimulation of fatty acyl-CoA/KAT interactions or KAT-independent, off-target effects. Thus, to further implicate fatty acyl-CoAs directly in inhibition of histone acetylation, we tested how manipulation of palmitoyl-CoA biosynthesis affected this process. Palmitate is activated to palmitoyl-CoA by the activity of ATP-dependent, long-chain acyl-CoA synthetase (ACSL) enzymes (Figure 5). The human genome encodes 5 ACSL family members, and related medium-chain acyl-CoA synthetases are known to possess considerable overlap in their ability to activate palmitate (Watkins et al., 2007). Therefore, we employed a gain-of-function approach, testing whether overexpression of a fatty acyl-CoA biosynthetic enzyme further antagonized acetylation. As many ACSL enzymes are membrane proteins, we chose an ACSL family member (ACSL3) known to express partial nuclear membrane localization as a potential candidate for the production of nuclear fatty acyl-CoAs (Poppelreuther et al., 2012). ACSL3 was cloned with a C-terminal FLAG tag into a mammalian expression vector under control of a constitutive CMV promoter. Transient transfection and overexpression of FLAG-ACSL3 in HEK-293 cells prominently reduced histone acetylation. H3K9 and H3K14 are known targets of GCN5 family KATs (Jin et al., 2011), and acetylation of each of these residues was strongly reduced. H4K8 and H3K27, which are not Gcn5-regulated targets, were also antagonized. However, acetylation of the cytoskeletal protein tubulin was not affected (Figure 5). Interestingly, overexpression of ACSL3 alone (without palmitate supplementation) was also found to inhibit histone acetylation, suggesting endogenous fatty acids may be mobilized to stimulate deacetylation in an ACSL-dependent manner (Figure S5b).</p><p>We emphasize that these results must be interpreted with caution, as ACSL3 overexpression may have effects on metabolism and enzyme activity that facilitate KAT-independent mechanisms of deacetylation and will require further analysis to dissect. In particular, the KDAC enzyme SIRT6 has been shown to be biochemically activated by long-chain fatty acids (although this enzyme primarily acts on H3K9Ac/H3K56Ac substrates, and thus would not explain the observed inhibition of H4 acetylation) (Feldman et al., 2013). However, the strong stimulation of deacetylation by fatty acyl-CoA biosynthetic enzymes, together with our biochemical analyses, are consistent with the hypothesis that fatty acyl-CoAs may be capable of metabolically facilitating cellular deacetylation programs by functioning as endogenous inhibitors of KAT enzymes.</p><!><p>The discovery that chromatin-mediated signaling cascades are sensitive to the metabolic state of the cell has the potential to powerfully illuminate the etiology of cancer and also inspire new therapeutic strategies. However, a major challenge lies in identifying direct mechanistic links between metabolites and specific chromatin modifiers. Here we have applied a novel competitive chemoproteomic approach to identify a metabolic-epigenetic interaction that potentially links tumor hypoacetylation and altered lipid metabolism on a molecular level (Nomura et al., 2010; Seligson et al., 2005). Palmitoyl-CoA is capable of interacting with KATs in the context of their multiprotein complexes and can potently inhibit their catalytic activity at concentrations <1 μM, making it one of the most potent metabolic inhibitors of KAT activity characterized to date. Previous studies have identified fatty acyl-CoA and fatty acyl-CoA-binding proteins in the nucleus (Elholm et al., 2000; Ves-Losada and Brenner, 1996), and micromolar concentrations of palmitoyl-CoA have been implicated in the regulation of many cellular proteins, including transcription factors (Hertz et al., 1998). While the literature and data are compelling, a relevant question is whether fatty acyl-CoAs are able to accumulate in cells to concentrations sufficient to affect KAT activity. One possibility is that fatty acyl-CoAs may be supplied to KATs at high local concentrations by fatty acyl-CoA-binding proteins, as has been proposed for other fatty acyl-CoA regulated enzymes (Elholm et al., 2000). Alternatively, the finding that a wide range of fatty acyl-CoAs can inhibit KATs suggests the potential for overall fatty acyl-CoA concentrations to reach levels sufficient to impact KAT activity.</p><p>While the subcellular concentrations (i.e. nuclear/cytosolic v. mitochondrial) of fatty acyl-CoAs have never been quantitatively characterized, studies in rats have observed that several abundant fatty acyl-CoAs are upregulated under conditions of nutrient stress, such as fasting (Woldegiorgis et al., 1985). Wellen and coworkers have previously drawn a direct correlation between bioenergetic stress, reduced acetyl-CoA: CoA ratio, and inhibition of histone acetylation (Lee et al., 2014; Wellen et al., 2009). Our findings extend this paradigm, and suggest in some cases this phenomenon may be mediated by additional metabolic factors such as palmitoyl-CoA. Interestingly, a recent study observed that cells grown in galactose-palmitate exhibit both increased fatty acid oxidation, as well as greatly reduced histone acetylation, relative to cells grown on traditional high glucose media (Pougovkina et al., 2014). Since fatty acid oxidation requires transient synthesis of nuclear/cytosolic fatty acyl-CoA, it is intriguing to speculate that direct inhibition of KATs by endogenous fatty acyl-CoAs may play a role in this process, possibly facilitating gene expression programs that aid adaptation to conditions of bioenergetic stress.</p><p>Finally, beyond the potential biological ramifications of these findings, our results also suggest a solution to a long-standing challenge in acetylation biology: how to deliver potent KAT-inhibiting bisubstrate inhibitors to cells. Our studies imply that metabolic acyl-CoAs are capable of inhibiting KAT-catalyzed acetylation with some degree of selectivity. It is thereby intriguing to speculate that this mechanism may be exploited to manipulate the concentrations of metabolic acyl-CoAs, thereby producing endogenous KAT bisubstrate inhibitors in cells. Notably, a wide array of metabolites and xenobiotics are known to be activated to CoA analogues via the activity of acyl-CoA synthetase enzymes (Darnell and Weidolf, 2013). Defining the tissue-specific production and interaction of these metabolic acyl-CoAs with KATs may provide a new paradigm for the targeted regulation of KAT activity. Such studies will require improved tools for the rapid development of KAT-metabolite interactions, as the human genome encodes at least 30 enzymes that have demonstrated evidence of KAT activity, only three of which were analyzed with the chemical proteomic approach reported here. In the future, we envision improved chemical proteomic approaches will enable the generation of global KAT-cofactor affinity maps (Becher et al., 2013), providing novel insights into the metabolic regulation of KAT activity and strategic manipulation of acetylation-dependent signaling in disease.</p><!><p>Recombinant p300 (1195-1662) and Gcn5 (497-662) were obtained from Enzo. Acetyl-CoA, propionyl-CoA, butyryl-CoA, succinyl-CoA, crotonyl-CoA, malonyl-CoA, and palmitoyl-CoA were synthesized according to the literature procedure (Padmakumar et al., 1997). All synthesized CoA analogues were HPLC purified, with purity verified by LC-MS prior to use. Linoleoyl-CoA, myristoyl-CoA, oleoyl-CoA, and palmitoleoyl-CoA were purchased from Sigma with purity verified by LC-MS prior to use. H3K14-CoA and desulfo-CoA were synthesized according to previously reported procedures (Chase et al., 1966; Montgomery et al., 2014; Zheng et al., 2004). Qubit Protein Assay kit (Life Technologies) was used to determine cell lysate and histone extract concentrations. Pyruvate dehydrogenase, ketoglutarate dehydrogenase, and NAD+ were purchased from Sigma. Labchip EZ-Reader 12-sipper chip (#760404) and ProfilerPro Separation Buffer (#760367) were purchased from Perkin-Elmer. H3K9Ac (9649P), H3K14Ac (7627P), H3K27Ac (8173), H4K8Ac (2594P), acetylated tubulin (5225P), GAPDH (5174S), Gcn5 (3305S), and pCAF (3378S) antibodies were purchased from Cell Signaling Technologies, while Mof (A300-992A-T) antibody was purchased from Bethyl Laboratories.</p><!><p>For each analysis, 50 μL of streptavidin-agarose resin (prewashed with 3 × 1 mL of PBS) was incubated with 10 μM H3K14-CoA-biotin (1) in 500 μL PBS and rotated for 1 h at room temperature. Whole cell lysates were clarified via filtration (EMD-Millipore, SLGV033RS), adjusted to a final protein concentration of 1.5 mg/mL, and split into 500 μL (0.75 mg) aliquots. For competitive samples, the metabolic acyl-CoA or cognate competitor (H3K14-CoA) was preincubated with lysate for 30 min on ice. Following equilibration, beads were pelleted by centrifugation (1400g × 3 min) and the supernatant was removed, before addition of the corresponding lysate sample. Samples were rotated for 1 h at room temperature, and pelleted by centrifugation (1400g × 3 min). After removal of lysate solution, beads were washed with ice-cold buffer (50 mM Tris-HCl pH 7.5, 5% glycerol, 1.5mM MgCl2, 150mM NaCl, 3 × 500 μL) and collected in centrifugal filters (VWR, 82031-256). For western blot analysis, samples were eluted in 1X SDS sample buffer (95 °C, 2 × 10 min). For prote omic analysis, 400 μL trypsin buffer (50mM Tris-HCl pH 8.0, 1M urea) was added to each sample, followed by 0.4 μL 1M CaCl2 and 4 μL trypsin (0.25 mg/mL), and digests were allowed to proceed overnight at 37 °C. After extraction, tryptic peptide samples we re acidified to a final concentration of 5% formic acid and frozen at −80 °C for LC–MS/MS an alysis.</p><!><p>KAT activity was measured by a continuous coupled-enzyme assay (Kim et al., 2000). In this assay, CoA produced by KAT-catalyzed acetylation is used by pyruvate dehydrogenase (PDH) to produce NADH, which can be monitored spectrophotometrically at 340 nm. Assays were performed in 150 μL volumes containing 50 mM Bis-Tris, 50 mM Tris, 100 mM sodium acetate (TBA buffer, pH=7.5), 5 mM MgCl2, 1 mM DTT, 2.4 mM pyruvate, 200 μM thiamine pyrophosphate (TPP), 200 μM NAD and 0.035 units of PDH (as defined by supplier), unlabeled histone peptide (H3 5-20 for Gcn5, H4 3-14 for p300; 60 μM) and 150 nM Gcn5 or 100 nM p300. Reactions were plated in 96-well plates and allowed to equilibrate at room temperature for 10 min. Reactions were initiated by addition of acetyl-CoA and analyzed continuously for 5 min by measuring NADH production at 340 nm. Initial velocities were determined by linear regression and background corrected by subtracting the rate of spontaneous formation of CoA (determined from reactions lacking Gcn5/p300). Kinetic parameters (Km and Vmax) for acetyl-CoA and inhibition constants (Ki) for palmitoyl- and desulfo-CoA were determined by holding the concentration of substrate peptide constant (H3 5-20 for Gcn5, H4 3-14 for p300; 60 μM) and initiating reactions with acetyl-CoA (4.1 – 33.3 μM) in the presence or absence of inhibitor (0 – 20 μM). All calculations were performed using Graphpad Prism 6. Michaelis-Menten parameters for acetyl-CoA were calculated by nonlinear regression of initial velocities. Kinetic parameters for acetyl-CoA were used to appropriately constrain nonlinear regression analyses of inhibitor data. Data were fit to equations for competitive noncompetitive, or uncompetitive inhibition and assessed for global goodness of fit (R2-values) to determine the optimal mode of inhibition. The inhibition constant (Ki) for each inhibitor was computed by using equations Km(obs)=Km*(1+[I]/Ki) and Y=Vmax*X/(Km(obs)+X), where X is substrate concentration, Y is response, [I] is the concentration of inhibitor, Vmax is the maximum response in the absence of inhibitor (expressed in the same units as Y), and Km is the Michaelis-Menten constant (expressed in the same units as X). Reported values are based on competitive model of inhibition.</p><!><p>Gcn5 activity was also measured using an orthogonal separation-based assay (Fanslau et al., 2010). This method detects acetylation of a FITC-labeled Gcn5 substrate peptide (histone H3; FITC-Ahx-QTARKSTGGKAPRKQL) based on its altered electrophoretic mobility relative to non-acetylated peptide. Assays were performed in 30 μL of reaction buffer (50 mM HEPES, pH 7.5, 50 mM NaCl, 2 mM EDTA, 2 mM DTT, 0.05% Triton-X-100) with KAT (GCN5 [100 nM], p300 [50 nM]) and FITC-peptide (FITC-H3 for GCN5; FITC-H4 for P300). Reactions were plated in 384-well plates, allowed to equilibrate at room temperature for 10 min, and initiated by addition of acetyl-CoA (final concentration = 0.63 - 10 μM). Reactions were monitored in real-time following transfer to a Lab-Chip EZ-Reader at ambient temperature and analyzed by microfluidic electrophoresis. Optimized separation conditions were: downstream voltage of −500 V, upstream voltage of −2500 V, and a pressure of −1.5 psi for FITC-H3 5-20 and FITC-H4 3-14 Percent conversion is calculated by ratiometric measurement of substrate/product peak heights. Percent activity represents the relative acetylation in inhibitor treated reactions relative to untreated control, measured in triplicate, and corrected for nonenzymatic acetylation.</p><!><p>The crystal structures of the KAT catalytic domain of human Gcn5 (PDB code: 1Z4R) and H3K14-CoA ligand (PDB code: 1M1D) were retrieved from the Protein Data Bank for use in the docking calculations (Schuetz et al., 2007). The Protein Preparation Wizard of the Schrödinger suite was used to prepare the binding site of Gcn5. The protein was processed by assigning bond orders, adding hydrogens, removing co-crystallized water molecules and creating disulfide bonds. Finally, a restrained minimization with a root mean square deviation (RMSD) value of 0.30 was applied using the OPLS 2005 force field to optimize the hydrogen bond network. Docking studies were performed using Glide (Glide, version 5.8; Schrödinger, LLC). The prepared Gcn5 structure was employed to generate the receptor energy grid centered on the co-crystallized ligand. The extra-precision (XP) docking protocol was used. Palmitoyl-CoA has 38 rotatable bonds, which means the configuration space for docking poses is enormous. In order to facilitate the sampling of the ligand conformations, we therefore set up a core constraint for the flexible docking in Glide. This core constraint was defined using the heavy atoms of the cofactor moiety of the co-crystallized ligand from PDB structure 1Z4R. The docking poses obtained in this way were then redocked again, but this time without constraints, using the Glide's refine mode.</p><!><p>HEK293T cells were cultured in DMEM supplemented with 10% FBS and L-glutamine (2 mM). For histone acetylation analyses, HEK-293T cells were plated in 6-well dishes (6 × 105 cells/well), and allowed to adhere for 24 h. At this point, transfections were performed using Lipofectamine 3000 (7.5 μL/well; Life Technologies) and plasmid DNA (2500 ng) according to manufacturer's protocol. After 6 h, media was removed, cells were washed with PBS, and fresh media was added containing either palmitate-conjugated BSA (100 μM palmitate) or BSA control. Cells were incubated with/without palmitate for 24 h, and harvested for histone extraction. Following removal of media, cells were washed once with PBS and incubated with 300 μL of Triton Extraction Buffer (TEB: PBS with 0.5% Triton-X 100, 2 mM PMSF, and 0.02% NaN3), harvested by gentle lifting, and transferred to microcentrifuge tubes. Samples were incubated in TEB for 30 min on ice, pelleted at 6500 rcf for 10 min at 4 °C , and supernatant removed and saved for cytosolic protein analysis. Nuclei were resuspended, washed again with TEB (150 μL), and pelleted at 6500 rcf for 10 min at 4 °C. Fo r acid extraction of histones, each pellet was treated with 0.4N H2SO4 (75 μL) and rotated overnight at 4 °C. Samples were centrifuged at 11000 rcf for 10 min at 4 °C, and histones were precipitated from the supernatant by addition of 20% TCA (750 μL). After at least 1 h, samples were centrifuged at 16000 rcf for 10 min at 4 °C, and th e pellets were washed with acetone/0.1% HCl (750 μL) and neat acetone (750 μL), with centrifugations at 16000 rcf 10 min at 4 °C following each wash. Samples were air-dried at room temperature, dissolved in ddH2O, and quantified by Qubit protein assay kit (Life Technologies). Samples were prepared for western blot analysis using using Bis-Tris NuPAGE gels (12%) and MES running buffer in Xcell SureLock MiniCells (Invitrogen) according to the manufacturer's instructions. Following antibody incubation and washing, antibody-binding was detected using LumiGLO (Cell Signaling Technologies) and chemiluminescent signal visualized using an ImageQuant Las4010 Digitial Imaging System (GE Healthcare). Immunoblot signals were quantified using ImageQuant TL software (GE Healthcare).</p>

PubMed Author Manuscript

Amino-functionalized conjugated polymer electron transport layers enhance the UV-photostability of planar heterojunction perovskite solar cells

In this study, for the first time, we report a solution-processed amino-functionalized copolymer semiconductor (PFN-2TNDI) with a conjugated backbone composed of fluorine, naphthalene diimide, and thiophene spacers as the electron transporting layer (ETL) in n-i-p planar structured perovskite solar cells. Using this copolymer semiconductor in conjunction with a planar n-i-p heterojunction, we achieved an unprecedented efficiency of $16% under standard illumination test conditions. More importantly, the perovskite devices using this polymer ETL have shown good stability under constant ultra violet (UV) light soaking during 3000 h of accelerated tests. Various advanced spectroscopic characterizations, including ultra-fast spectroscopy, ultra-violet photoelectron spectroscopy and electronic impedance spectroscopy, elucidate that the interaction between the functional polymer ETL and the perovskite layer plays a critical role in trap passivation and thus, the device UV-photostability. We expect that these results will boost the development of low temperature solution-processed organic ETL materials, which is essential for the commercialization of high-performance and stable, flexible perovskite solar cells.

amino-functionalized_conjugated_polymer_electron_transport_layers_enhance_the_uv-photostability_of_p

Introduction<!>Results and discussion<!>Experimental

<p>As a new class of photovoltaic devices for solar energy utilization, perovskite solar cells have emerged as one of the promising alternatives to the conventional silicon-based photovoltaics over the past few years. [1][2][3][4] These organic-inorganic hybrid lead halide perovskite materials are advantageous because they can be fabricated with solution-processable, relatively low cost methods, as compared to their silicon-based counterparts. The organic-inorganic hybrid perovskite materials possess superior optoelectronic properties, including high absorption coefficient ($10 5 cm À1 ), 5,6 wide absorption (300-900 nm), 7,8 small exciton binding energy (19-50 eV), 9 long electron/ hole diffusion length (100-1000 nm), 10,11 suitable band gap ($1.5 eV), 12 and high bipolar conductivity (10 À2 -10 À3 S cm À1 ). 13,14 The efficiency of perovskite solar cells has increased from 3.8% in 2009 to 22.1% in 2016 through continuous efforts for the optimization of lm deposition as well as device fabrication processes. 15 Mesoporous-and planar-heterojunction (PHJ) structures are two main architectures adopted for efficient perovskite solar cells. In both device structures, the perovskite absorber layer is sandwiched between the electron transporting layer (ETL) and the hole transporting layer (HTL). 16,17 Therefore, charge transporting layers are the key components of the perovskite devices, and they play an important role in improving both the efficiency and stability of the devices. 18,19 The function of the charge transporting layer is to extract photon-generated charges from the perovskite layer and transport the charges to the corresponding current collecting electrodes. The dual crucial processes of fast charge transfer (forward) and slow recombination (backward) place challenging constraints on the choice of effective charge transporting layers. In general, a good charge transporting layer should have matched energy levels with the conduction band (CB) or valance band (VB) of the active layer, and should also possess high conductivity and charge mobility to ensure efficient charge transport, as well as good charge selectivity to increase the charge collection efficiency at the corresponding electrodes. Among the various materials, titanium dioxide (TiO 2 ) is the most widely used ETL for electron a Wuhan National Laboratory for Optoelectronics, Huazhong University of Science and Technology, Wuhan, Hubei 430074, China. E-mail: mingkui.wang@mail.hust.edu.cn b Institute of Polymer Optoelectronic Materials and Devices, State Key Laboratory of Luminescent Materials and Devices, South China University of Technology, Guangzhou, Guangdong 510640, China. E-mail: msangusyip@scut.edu.cn; msuang@scut.edu.cn transport and hole blocking, due to its good conductivity (1.1 Â 10 À5 S cm À1 ) and the suitable energy levels of the CB ($À3.9 eV) and VB ($À7.2 eV). [20][21][22] Impressive PCEs over 19% have been obtained for PHJ perovskite solar cells using TiO 2 as the ETL. 4,15 However, the long-term operational stability of this type of device has been considered to be a major, and concern has been raised about its suitability for practical application. Most stability studies have so far focused on the moisture or heat effect, but some recent experimental results have indicated that the performance of n-i-p structure perovskite solar cells with TiO 2 ETL suffer from rapid decay when exposed to illumination, even the devices that have been encapsulated in an inert atmosphere. [23][24][25] Thus, there is an urgent need to investigate the corresponding degradation mechanism for this type of perovskite solar cells under light illumination and develop appropriate strategies to improve the long-term operational stability of the solar cells.</p><p>The application of these inorganic ETLs, particularly the crystalline TiO 2 , usually requires a high-temperature process to improve crystallinity and charge carrier mobility. 26,27 High temperature sintering/annealing not only results in an increased cost and slow production, but also limits the utilization of plastic lms as the exible substrates. Therefore, replacing high-temperature processed ETLs with lowtemperature processable materials can provide a better processing window and eventually simplify the manufacturing process of perovskite solar cells. Organic materials have several attractive features that can make them efficient ETLs, which can settle the challenges mentioned above. Recent studies have revealed that some organic ETLs could reduce the density of trap states on the surface and at the grain boundaries of perovskite crystals to improve the electron extraction efficiency and decrease photocurrent hysteresis. 28 Despite many efficient organic ETLs being employed to improve the performance of pi-n planar heterojunction perovskite solar cells, there are very few reports suggesting that organic ETLs could also be used in the n-i-p structure PHJ perovskite solar cells. 20,29,30 Among them, fullerene (C 60 ) and its derivative [6,6]-phenyl-C 61 -butyric acid methyl ester (PCBM) have been proved to be efficient ETLs in this type of perovskite solar cell with impressive PCEs ($16%). 29,31 Compared with these small molecular fullerenes, polymer semiconductors possess several appealing features that make them good candidates for ETL, such as good lm formation properties, adjustable energy levels and excellent optical and electrical properties. Despite these advantages offered by polymer semiconductors, so far, they have not been explored as ETLs in the n-i-p structure PHJ perovskite solar cells.</p><p>In this work, we for the rst time introduced an aminofunctionalized copolymer as an alternative ETL to replace the commonly used TiO 2 in the n-i-p PHJ perovskite solar cells with device conguration of ITO/ETL/MAPbI 3Àx Cl x /2,2 0 ,7,7 0 -tetrakis-(N,N-di-p-methoxyphenylamine)-9,9 0 -spirobiuorene (spiro-MeOTAD)/Au, and studied the device stability under UV light soaking. Fig. 1 shows the chemical structure of the ETL material. The copolymer (named as PFN-2TNDI) has a conjugated backbone composed of uorene, naphthalene diimide, and thiophene spacers and the synthetic routes are reported elsewhere. 32 The matched energy levels, high electron mobility, and good lm formation properties make it favorable as a good electron transporting material for perovskite solar cells. When it was used in a p-i-n planar perovskite solar cell with ITO/ PEDOT:PSS/perovskite/ETL/metal electrode structure, the device showed a PCE of $16.7%. 32 The introduction of PFN-2TNDI into n-i-p planar structure devices was inspired by a successful application of amino-substituted perylene diimide derivative (N-PDI) as a new generation ETL for this type of perovskite solar cell in our group. 20 We found that the terminal amino groups in the N-PDI molecules could improve the wetting properties of the perovskite lm, reduce the work function of the transparent conductive oxides substrate and also passivate the surface trap states of perovskite lms. In this study, since the PFN-2TNDI also has amino groups in the alkylamine side chains, it may provide similar good properties for serving as an effective ETL. In addition, this ETL can be readily deposited to form a high-quality lm due to the good lm formation properties of the polymer, which signicantly inuence the morphology of the upper perovskite layer. 7,33 Indeed, our study conrmed that the PFN-2TNDI could be used as an alternative ETL candidate for n-i-p perovskite solar cells with excellent performance. Devices based on the PFN-2TNDI ETL showed a competitive PCE of $16% under a simulated irradiation of AM 1.5G at 100 mW cm À2 , which was comparable to the devices based on high-temperature sintered TiO 2 . More importantly, the devices based on PFN-2TNDI signicantly inhibited UV induced performance degradation and exhibited excellent UVphotostability, which retained 75% of their original efficiency under constant UV exposure over 3000 hours; the control devices with TiO 2 ETL only showed less than 10% of the original PCE value within 300 hours. This result highlights the possibility of using the low-temperature processed copolymer to replace inorganic metal oxide as ETLs for highly efficient and stable perovskite solar cells.</p><!><p>In this report, the n-type conjugated polymer PFN-2TNDI was used as the ETL in highly efficient n-i-p PHJ perovskite solar cells. Fig. 1a shows the device structure, in which we chose MAPbI 3Àx Cl x as the absorber layer, PFN-2TNDI and spiro-OMeTAD as ETL and HTL, respectively. The MAPbI 3Àx Cl x lm was deposited on the substrate pre-coated with the PFN-2TNDI using a two-step inter-diffusion method. 34 The spiro-OMeTAD HTL was deposited on top of the MAPbI 3Àx Cl x layer by spin coating. The cells were nished by thermally evaporating an Au back electrode. Details of the fabrication processes are described in the Experimental section. As depicted in Fig. 1b, the energy levels of each layer are well-matched with the adjacent ones, which facilitate efficient electron/hole transport and extraction.</p><p>Various characterizations, including top-view scanning electron microscopy (SEM), absorption spectra, and X-ray diffraction (XRD), were performed to illustrate the inuence of the under-layered PFN-2TNDI on the structural properties of the MAPbI 3Àx Cl x lms. The SEM images in Fig. 2a and b show the surface morphology of the MAPbI 3Àx Cl x lms on ITO and PFN-2TNDI (5 nm) coated ITO substrate, respectively. Both perovskite lms show compact multicrystalline structures with large grains (in the range of $1-2 mm), which is crucial for highperformance perovskite devices. Optical absorption spectra of both perovskite lms on the ITO and ITO/PFN-2TNDI substrates (Fig. 2c) show broad absorption ranging from the visible to near-IR region. As we can see from the spectra, the absorption from the ultrathin ETL layers introduces ignorable changes in the absorption properties of the perovskites as a result of very intense light absorption properties of the thick perovskite lms. The crystal structures of the MAPbI 3Àx Cl x lms were studied using XRD spectroscopy. As shown in Fig. 2d, the diffraction peaks of 14.2 , 28.4 , and 31.8 correspond to the (110), ( 220) and (310) planes of the tetragonal perovskite phase, conrming the formation of crystalline MAPbI 3Àx Cl x in both lms. However, the XRD intensity of the MAPbI 3Àx Cl x lm deposited on the ITO/PFN-2TNDI substrate was much stronger than that deposited on the bare ITO substrate, indicating that the ETL promoted the formation of a perovskite with higher phase purity and preferential orientation, which are important factors for high performance devices.</p><p>Steady-state photoluminescence (PL) and time-resolved PL decay measurements were further conducted to investigate the charge transfer and recombination occurring at the perovskite/ ETL interface. According to earlier reports, 20,32 the amino groups of this copolymer's side chains could passivate the surface trap states, which were introduced by halide vacancies in MAPbI 3Àx Cl x lms, leading to a recovery of the bandgap. The MAPbI 3Àx Cl x lm deposited on the ITO/PFN-2TNDI substrate shows stronger PL quenching than that on the bare ITO substrate (as in Fig. 3a), suggesting that the PFN-2TNDI is a good quencher for the perovskite lms. An excitation wavelength of 400 nm was used to excite the perovskite lms on the PFN-2TNDI modied substrates from either the ITO side or the perovskite side to investigate the trap passivation effect at the ETL/perovskite interface. Moreover, the perovskite lm with PFN-2TNDI shows a blue-shied PL peak from 769 (excited from the perovskite side) to 764 nm (excited from the ITO side), indicating a signature of lling the trap states close to the bottom surface of the MAPbI 3Àx Cl x lm by the PFN-2TNDI. 28,32 In contrast, the perovskite lm coated on the bare ITO maintains its PL peak at 769 nm under excitation from both sides. One potential explanation for the surface trap passivation is that the electron rich nitrogen atoms on the alkylamine side chain of PFN-2TNDI may coordinate with the unsaturated Pb atoms at the MAPbI 3Àx Cl x surface, thus providing an effective trap passivation effect at the ETL/perovskite interface. Furthermore, the perovskite lm on the ITO/PFN-2TNDI substrate underwent a faster PL decay as described in Fig. 3b. The PL decay curves were tted with a single-exponential equation with PL lifetime values of 5.56 and 4.59 ns for the MAPbI 3Àx Cl x lms on ITO and ITO/PFN-2TNDI substrates, respectively. The shorter lifetime for the PFN-2TNDI device suggests that a faster carrier extraction can be achieved from the MAPbI 3Àx Cl x lm to the ETL, further demonstrating that PFN-2TNDI is an efficient ETL.</p><p>The lm formation properties of PFN-2TNDI were studied using atomic force microscopy (AFM). As illustrated in the AFM topography images (Fig. S1, ESI †), the root mean square (RMS) roughness value of the quartz substrate was reduced from 1.189 to 0.328 nm by coating with 5 nm-thick PFN-2TNDI, suggesting that this polymer has good lm formation properties. The smooth PFN-2TNDI surface facilitates the formation of perovskite lms with higher surface coverage and reduced pinhole size. 7,33 Ultraviolet photoelectron spectroscopy (UPS) measurement was conducted to study the change in the work function (WF) of the ITO surface when coated with the ETL. We found that the WF of the ITO was reduced to 3.81 eV from its original 4.4 eV aer the deposition of PFN-2TNDI (5 nm), as displayed in Fig. S2, † which was a little higher than that of the TiO 2 -coated ITO sample (3.76 eV). This value matches well with the CB of MAPbI 3Àx Cl x lm, which facilitates ohmic contact formation and efficient charge transfer at this interface. The reduction of the substrate's WF can be attributed to a large dipole moment at the ITO/PFN-2TNDI interface, induced by the amino groups on the PFN-2TNDI side chain, which has been widely discussed in the eld of polymer solar cells. 20 The chemical properties of PFN-2TNDI are different from those of TiO 2 , which may affect the formation of perovskite crystals on them. Therefore, we compared the wetting capability of PFN-2TNDI and TiO 2 substrate surfaces (Fig. S3 †). The average contact angles are 30 and 100 for the PFN-2TNDI and TiO 2 lms to water, respectively. Because of the similar polarity of N,N-dimethylformamide (DMF) to water, we expect that this is an indication of the wettability of these surfaces by the perovskite precursor. It was reported that when the two-step inter-diffusion method was employed, the hydrophobic surfaces were good for the formation of perovskite lms with large crystalline grains and less charge trap density by preventing the formation of too dense nuclei from heterogeneous nucleation. 35 We carried out systematic characterizations to nd out the optimum thickness of PFN-2TNDI as the ETL in the MAPbI 3Àx Cl x solar cells. Fig. 3c shows the transmission spectra of ITO/PFN-2TNDI samples obtained by spin-coating PFN-2TNDI from solutions of different concentrations. The transmittance of the samples gradually decreased as the lm thickness of PFN-2TNDI increased from 3 nm to 12 nm, especially in the range of 350-450 nm and 560-700 nm, which can be attributed to the absorption from the p-p* transition and intra-molecular charge transfer characteristics of the PFN-2TNDI lm. 32 For comparison, Fig. S4 † shows the transmission spectra of the TiO 2 compact layer. The PFN-2TNDI lm with a thickness of 5 nm showed slightly less transmittance than the TiO 2 lm in the short wavelength range. This ensures the absorption of the perovskite layer deposited above it (Fig. 2c). The photovoltaic performances of the perovskite solar cells with a conguration of ITO/PFN-2TNDI/MAPbI 3Àx Cl x /spiro-OMeTAD/Au were evaluated by varying the PFN-2TNDI lm thicknesses. A control device without the PFN-2TNDI was also presented. Table 1 lists the photovoltaic parameters for various perovskite devices under the illumination of AM 1.5G, 100 mW cm À2 . As can be seen, the device performances signicantly depend on the thickness of the PFN-2TNDI lms. The optimized thickness of the PFN-2TNDI is about 5 nm. Further increasing the thickness results in the deterioration of the device performance due to an increase in electrical resistance (Table 1). Perovskite solar cells with an optimized PFN-2TNDI lm exhibit a short-circuit current density (J SC ) of 22.01 mA cm À2 , a ll factor (FF) of 0.74, an open-circuit voltage (V OC ) of 0.98 V, yielding an impressive PCE value of 15.96% (Table 1). The control device without ETL achieves a PCE of 11.99% with V OC , J SC and FF of 0.91 V, 19.42 mA cm À2 and 0.68, respectively. The enhancement of J SC and FF of the device with PFN-2TNDI ETL could be partially attributed to the passivation effect of PFN-2TNDI on the defects in perovskite lms, which reduces the charge recombination at the interface. 32,36 Additionally, the rectication effect (i.e., efficient electron extraction and hole blocking) induced by the PFN-2TNDI interlayer could lead to improvement in J SC and FF. The enhanced V OC from 0.91 V of the control device to 0.98 V of the PFN-2TNDI-based device can be attributed to better energy alignment at the ITO/ETL and ETL/perovskite interfaces. 20,32 Therefore, the existence of the PFN-2TNDI layer minimizes the potential loss between the ITO and perovskite and enlarges the built-in potential across the device. 37,38 Consequently, the enlarged built-in eld could reduce carrier accumulation and further reduce carrier recombination inside the solar cell.</p><p>We further compared the PFN-2TNDI-based devices with the TiO 2 -based ones to evaluate the viability of PFN-2TNDI as ETL in perovskite solar cells. Fig. 3d compares the J-V curves of solar cells based on the PFN-2TNDI (5 nm) and TiO 2 . The corresponding photovoltaic parameters for the TiO 2 (30 nm)-based devices are also shown in Table 1. With an optimized TiO 2 thickness of $30 nm, the TiO 2 -based device exhibited a PCE of 17.2% with a V OC of 1.00 V, a J SC of 21.66 mA cm À2 and a FF of 0.78. Impressively, the optimized performances of devices based on PFN-2TNDI ETL (5 nm) are comparable to that of TiO 2based devices, clearly indicating that the copolymer PFN-2TNDI can be used as an outstanding solution-processed ETL for perovskite devices. The relatively higher J SC of PFN-2TNDIbased devices could be ascribed to the larger perovskite crystalline grains on PFN-2TNDI, which led to an increase in IPCE value at longer wavelength (Fig. S5 †). The delicate difference in V OC between PFN-2TNDI-based devices and TiO 2 -based devices may be caused by the different work functions of PFN-2TNDI and TiO 2 .</p><p>In order to further understand the electronic transport processes at the perovskite/ETLs interfaces, electronic impedance spectroscopy (IS) measurement was performed for the n-ip perovskite solar cells based on low-temperature processed PFN-2TNDI and high-temperature processed TiO 2 . Fig. 4a and b show the Nyquist plots and Bode plots for both devices under illumination at a bias of À0.7 V in the frequency range of 1 MHz-10 mHz, from which two semicircles could be easily identied. The results indicate that similar interfacial charge transfer processes occur in both devices based on PFN-2TNDI and TiO 2 . Herein, the semicircle in the range of high frequency provides useful information on the charge carrier recombination process, i.e., interfacial recombination of electrons from the ETL with holes from MAPbI 3Àx Cl x , or electrons from MAPbI 3Àx Cl x with holes from HTL 39,40 and the charge transfer process at the ITO and Au electrodes. 16,40 The latter is too fast to be separated from the former. The semicircle in the relatively low frequency represents ion migration processes in the perovskite active layer. 16,33 The Nyquist plots were tted using an equivalent circuit model of two-RC elements in series as shown in Fig. 4a. For a better tting, all capacitor elements were replaced by constant phase elements; in all the cases, the constant phase element (CPE) exponent p was kept quite closely to the perfect capacitor value, p z 1. Fig. 4c and d present the charge recombination resistance (R rec ) and geometrical capacitance (C) as a function of bias for devices based on PFN-2TNDI and TiO 2 ETL. It is interesting to nd that the recombination resistance (R rec , Fig. 4c) in the PFN-2TNDI-based device is almost the same as that of the TiO 2 -based device. These results suggest that the charge ux for recombination processes in both devices are within the same order. However, the utilization of PFN-2TNDI shows larger capacitance (C, Fig. 4d), indicating that more charge would take part in the interfacial recombination process. This conclusion is based on the density of states, DOS, which can be reected by capacitance through the approximation of DOS $ C. This explains why the device using PFN-2TNDI shows lower output photo-voltage than that of the TiO 2 -based device (Table 1), even though a longer electron lifetime is observed in the former case (Fig. 4e). The IS results further conrm that the PFN-2TNDI can serve as an efficient ETL, like the TiO 2 , and provide an alternative choice for fabrication of n-i-p PHJ perovskite solar cells.</p><p>Stability measurements were also performed for the two devices under constant UV illumination. Fig. 5 presents the evolution of solar cell performance parameters as a function of testing time under UV illumination of the PFN-2TNDI-and TiO 2 -based perovskite solar cells. Those devices were encapsulated with glass cover-slips glued with epoxy resin. The devices were exposed to constant 365 nm UV light through ITO sides in an argon-lled glove-box with a humidity of <1.0 ppm. We found that all the photovoltaic parameters for the PFN-2TNDIbased perovskite solar cells improved a little within the rst 20 h of illumination and then slowly degraded to about 75% over the next period of aging time (3000 hours) with small variation in V OC , J SC , and FF. In contrast, for the TiO 2 -based devices, the PCE dropped signicantly to less than 10% of the original value within 300 hours of illumination, with a particularly rapid degradation in V OC , J SC , and FF. These results suggest that the perovskite solar cells based on PFN-2TNDI ETL are much more stable than the TiO 2 analogues under UV irradiation. Since the difference in the devices is related to ETL, we further conclude that the light-activated degradation at the interface between TiO 2 and perovskite is the major reason accounting for the degradation of the devices. In addition to the essential characteristics of an n-type semiconductor, the excited TiO 2 has a strong ability to extract electrons from electron-rich materials. Therefore the TiO 2 has been used as a typical photocatalyst for environmental purication, such as CO 2 reduction, organic compounds decomposition and water splitting. [41][42][43][44] Hence, electron extraction from the iodide anion in MAPbI 3Àx Cl x by the TiO 2 may provide a driving force for the deconstruction of the photoactive layer, resulting in the deterioration of device performance. 24 Other reports suggest that the UV degradation of perovskite solar cells originates from a large amount of oxygen vacancies in the TiO 2 layer. 23 Upon excitation by UV light, these oxygen vacancies could act as deep electron trap states. The photo-generated electrons by the MAPbI 3Àx Cl x layer could be trapped by these defeat states and then recombine with holes, and thus reduce the charge collection efficiency and cause further deterioration of the device performance. However, the organic PFN-2TNDI has less UV-light-induced structural defects and traps. Consequently the devices based on PFN-2TNDI have indeed shown much improved resilience to UV irradiation. A little deterioration observed with this device would be caused by the de-doping of spiro-OMeTAD and the reaction between spiro-OMeTAD and the Au electrode.</p><p>To further understand the change in electronic processes in the PFN-2TNDI-and TiO 2 -based devices aer the UV irradiation, we further analyzed their Nyquist plots as shown in Fig. 6a and b, respectively. The fresh and aged devices using the PFN-2TNDI ETL show two similar semicircles in the Nyquist plots, in which only a small decrease can be found in the rst semi-arc in the high frequency range, suggesting that there is no signicant change in the interfacial electronic processes in this type of device aer exposure to UV-light. In contrast, an additional semi-arc appears for the aged devices with the TiO 2 ETL in the high frequency range. The newly emerged semicircle implies that an additional interfacial process is generated during the UV aging process, which might adversely affect the carrier transport in the devices. 31 A large reduction in the recombination resistance (R rec ) was observed in the TiO 2 -based device aer UVexposure, probably due to an increase in charge recombination at the MAPbI 3Àx Cl x /TiO 2 interface. This led to the dramatically fast degradation of the device performance, especially the V OC and FF (Fig. 5). Interestingly, the UV-light aging process showed a less signicant effect on the Au/HTM interface for both devices, according to the IS results shown in Fig. 6.</p><!><p>The etched ITO (indium tin oxide) substrates were ultrasonically cleaned in detergent, milli-Q water, acetone and ethanol for 15 min, respectively. Prior to depositing the ETLs, the clean and dry ITO substrate was treated with UV/O 3 for 30 min. PFN-2TNDI layers were coated on the ITO by spin coating at 3000 rpm for 30 s using a solution of PFN-2TNDI in chlorobenzene with various concentrations of 0.5 mg mL À1 , 1 mg mL À1 , 2 mg mL À1 and 4 mg mL À1 , then annealed at 100 C for 10 min. The MAPbI 3Àx Cl x lm was fabricated by a two-step inter-diffusion method. 34 The PbI 2 layer was spin coated from the 70 C pre-heated PbI 2 solution in DMF (462 mg mL À1 ), then 30 mL of the pre-heated solution was dropped onto the ETLcovered ITO substrate as soon as possible, and then the spinning was immediately started at 3000 rpm. The PbI 2 lm was annealed at 70 C for 30 min. Aer cooling to room temperature, 25 mL MAI : MACl (50 : 5 mg) in 1 mL 2-propanol was dropped on the as-fabricated PbI 2 lm and spin coating was taken at 3000 rpm for 30 s. Aer the deposition of stacked precursor layers, the obtained lms were annealed on the hotplate at 135 C for 15 min. The HTL was deposited on top of perovskite lm by spin coating at 4000 rpm for 30 s. The acceleration was 3000 rpm per second. The spin coating solution was composed of 72.3 mg spiro-OMeTAD in 1 mL chlorobenzene with the standard additives of lithium bis(triuoromethylsulphonyl)imide in acetonitrile of 520 mg mL À1 (17.5 mL) and 4-tert-butylpyridine (30 mL). Then, an 80 nmthick gold electrode was thermally evaporated under the vacuum pressure of 5.0 Â 10 À4 Pa to complete the device fabrication. All the spin-coating processes were performed in a dry air-lled glove box with the humidity of <1.0 ppm. The effective area of the solar cell was dened to be 0.125 cm 2 . For the control device based on the TiO 2 , all the fabrication processes were the same, except that the compact layer was produced by spin coating at 3000 rpm twice using nanocrystalline TiO 2 solution precursors, and then annealed on a hot plate for 30 PFN-2TNDI ETL showed good performance with PCE of $16%, which was comparable to that of inorganic TiO 2 ETL-based devices. More importantly, in addition to the low temperature processability offered by the polymer ETL, the devices based on the polymer ETL show greatly enhanced photostability against UV irradiation, compared to the TiO 2 -based devices. We believe this work not only can provide important insights on designing new organic materials as efficient interfacial layers for high performance perovskite solar cells, but also demonstrates a feasible strategy to overcome the photostability issue encountered in perovskite devices using inorganic semiconductors as the charge transport layer.</p>

Royal Society of Chemistry (RSC)

First Principles Insights into Amorphous Mg 2 Sn Alloy Anode for Mg-ion Batteries

Rechargeable Mg-ion batteries (MIBs) are an advantageous alternative solution to Li-ion batteries in many ways. Mg is safer and abundant in the Earth, and has a high electrochemical capacity owing to its divalent nature. It is yet relatively less studied largely due to primal success of Li-base batteries and challenges associated with the design of MIBs including high performance electrode materials. Herein, using first-principles calculation, we study the electrochemical and mechanical properties of the most viable alloy anode Mg 2 Sn with special attention to its amorphous phase-unavoidable phase forming during cyclic Sn magnesiation in MIBs due to volume changes. We create amorphous Mg 2 Sn via simulated annealing technique using ab initio molecular dynamics. We find while Mg 2 Sn undergoes a substantial atomic-level structural changes during the crystal-to-amorphous transformation, its polycrystalline properties degrade slightly and become softer by only 20% compared to the crystal phase. Moreover, we predict competitive electrochemical properties for the amorphous phase assuming it goes under similar reaction path as the average electronic charge on Mg ions almost remain unaffected. This work thus not only demonstrate that a-Mg 2 Sn 1 phase could be a bypass to combat the challenges associated with the crystal cracking during volume change, but also serves as first step to better understand the widely used Mg 2 Sn alloy anode in MIBs.

first_principles_insights_into_amorphous_mg_2_sn_alloy_anode_for_mg-ion_batteries

Introduction<!>Methodology<!>Results and Discussion<!>Fundamental Elastic Behavior<!>Conclusions

<p>Li-ion batteries (LIBs) are considered the most important and widespread energy storage solution in consumer electronics largely owing to their unrivaled combination of high volumetric and gravimetric energy capacities, and indeed thier early technological maturity in 1991. 1,2 They are also seen as an immediate viable candidate for the next generation of electric vehicles reducing our dependence on fossil fuels. 3 The ever-increasing reliance on limited Li reserves as well as inherent safety issues and environmental impacts associated with LIBs, however, have raised concerns about the sustainability of this technology. 4 Na-ion batteries (NIBs) have emerged an appealing alternative solution with multifaceted benefits such as less expensive sodium precursors, and being more abundant and less toxic. 5 The performance of NIBs have constantly being improved especially during the last few years. 5,6 Yet this relatively unexplored technology is far behind its Li-counterparts limiting its application due to its intrinsic differences, and debated cost advantage. 5,7 Still they are deemed as an ideal storage technology for large-scale applications i.e. power grids, where size is not a major design concern. 2,5,8 The situation has fueled an active research towards the development of alternative safe and cost-effective technologies beyond LIBs technology. [9][10][11][12] Mg-ion batteries (MIBs) are further recognized as attractive low-cost and high-capacity substitute for energy storage. Mg is naturally abundant in the Earth and is environmentally benign, and has a negative reduction potential of -2.37 V vs. SHE. The divalency of Mg 2+ cations gives an attractive volumetric energy density in excess of 3833 mAh ml −1 , approximately twofold(threefold) higher than Li(Na). In addition, MIBs offer wider operating temperatures and better safety features compared to LIBs due to higher melting temperature of Mg (648.8 • C). The formation of a passivation layer at the interface of electrode and elec-trolyte during Mg plating and stripping completely blocking Mg 2+ ions to shuttle, and the strong electrostatic interaction of Mg 2+ with the host largely hampering Mg 2+ ions mobility are major drawbacks hindering their practical applications. Addressing these issues have become the mainstream of MIBs' research and development. [13][14][15] There has been a promising progress on finding compatible electrolytes and suitable electrodes led to the introduction of promising electrolytes and cathode candidates. [16][17][18][19][20] However, identifying failure-free anode materials has received less attention and remains a big challenge.</p><p>The dendrite-free Mg metal initially gained popularity as an anode material for MIBs. It was later realized that its application is largely impeded due to the formation of the reduced layer. 13,21 Magnesium alloys, a family of insertion-type materials based on Mg alloying/dealloying, are then found the most appealing anode candidates owing to their superior volumetric and gravimetric capacities as well as their compatibility with conventional electrolytes (e.g. magnesium salts). [22][23][24][25] The magnesiation of group-14,15 elements during the electrochemical reaction was reported experimentally. [26][27][28][29][30][31][32][33][34][35] It was shown that Sn could achieve higher gravimetric capacities than other group-14,15 elements, and operates in a low voltage window. 28,[32][33][34][35][36][37] Moreover, some theoretical studies suggest that Sn (α/β-Sn) is the most competitive alloy anode for MIBs due to relatively low diffusion barriers of Mg ions. 27,35,38 Nevertheless, Sn magnesiation results in large structural changes in an electrochemical cell-in excess of 180 % volume change with reference to pure Sn during the formation of crystalline Mg 2 Sn (c-Mg 2 Sn). [27][28][29] It is well established that the electrochemical performance of alloys is largely hampered by structural failures as a result of drastic volume changes upon successive intercalation and de-intercalation of active ions during charge and discharge. 13,[39][40][41][42] Sn magnesiation during charge and discharge results in remarkable capacity fade. 36 The underlying reasons are not clear, it is hypothesized that, along with the known issue of the formation of the passivation layer, large structural changes accompanied by pulverization and amorphization could be responsible. 28,36 Very recent strategies like the formation of nanostructured Sn 34 or dual phase alloying of nanoporous Bi-Sn alloy 25 or developing eutectic alloys have demonstrated a very good progress; still remains an open challenge in the field. Sn magnesiation is somewhat similar but less severe to alloy anodes in LIBs and NIBs in terms of expansion; there is a crucial difference though. In Li/Na-based alloy anodes, various crystalline and amorphous phases coexist during charge and discharge complicates the phase transformations leading to the accumulation of internal stresses due to the coherent boundaries between phases. [43][44][45][46][47] For Mg-Sn system, Mg 2 Sn is the only binary equilibria phase, hence less susceptible to detrimental phase transformations. 48,49 Yet Sn magnesiation cuasing drastic volume changes lead to the inevitable magnesiation-induced amorphization affecting the electrode performance. 36 Understanding the mechanical behavior during charge and discharge is essentailly the first step towards designing a failure-free electrode. [50][51][52] While c-Mg 2 Sn is studied extensively in literature, [53][54][55][56] to the best of our knowledge, no information is available about the atomic structure as well as mechanical properties of electrochemically formed amorphous Mg 2 Sn phase (a-Mg 2 Sn). These properties are well documented for Li and Na binary alloys, whereas such knowledge is missing for the Mg-Sn system. [43][44][45][46][47] In this work, we study the fundamental electrochemical and elastic properties of Mg 2 Snthe only phase in the binary Mg-Sn system at equilibrium-in crystalline and amorphous form using first principles calculations. 48,49 We obtained the atomic level structure of a-Mg 2 Sn by creating an amorphous phase using simulated annealing technique at three temperatures (above the melting point of c-Mg 2 Sn phase) using ab initio molecular dynamics.</p><p>We provide a quantitative and qualitative atomic level insight into the a-Mg 2 Sn phase using radial distribution functions and rings statistics. Our results suggest that a-Mg 2 Sn is elastically softer than c-Mg 2 Sn, and it is alike c-Mg 2 Sn in terms of electrochemical properties.</p><!><p>The Vienna Ab Initio Simulation Package (VASP) was used to perform density functional theory (DFT) calculations within GGA-PBE approximation. 57 Projector-augmented wave (PAW) pseudopotentials were employed for describing core and valence electrons interactions. 58,59 The 2p 6 3s 2 and 4d 10 5s 2 5p 2 states were treated as valence electrons for Mg and Sn, respectively. All crystalline phases were obtained from the crystal structure database. 60 For geometry optimization we used a 7×7×7 Monkhorst-Pack k-point sampling for the unit cell of c-Mg 2 Sn and a plane-wave kinetic energy cutoff of 500 eV ensuring the total energy converge within 1 meV/atom. 61 The structures were then optimized by allowing lattice vectors and ionic positions to relax until the Hellmann-Feynman forces were less than 0.01 eV/Å. A denser k-point (15×15×15 grid) and higher energy cutoff (700 eV) were used for the same cell for the single-point energy calculations.</p><p>We used a 2×2×2 super cell of c-Mg 2 Sn to create the a-Mg 2 Sn phase. Using ab initio molecular dynamics (AIMD) as implemented in VASP, the structure was annealed up to three different temperatures (1600 K, 1800 K, 2000 K) above the melting point of c-Mg 2 Sn (1053 K). These tests allow to examine the impact of annealing temperature on amorphousity. The temperature was then maintained for 1000 MD time steps (time step = 2 fs). The system was then quenched rapidly down to room temperature at the rate of 20×10 14 K/s. Finally the system was allowed to equilibrate at this temperature for another 2500 MD time steps for (time step = 2 fs). We analyzed the produced a-Mg 2 Sn phases using radial distribution functions (RDFs) and rings statistics as implemented in Interactive Structure Analysis of Amorphous and Crystalline Systems (I.S.A.A.C.S) software package. 62 The a-Mg 2 Sn structure was relaxed with a similar plane wave kinetic energy cutoff and an atomic force tolerance of 0.02 eV/Å, and a 7×7×7 k-point Monkhorst-Pack grid. We then calculated the formation energy E f with respect to Mg/Mg +2 as E f = E(Mg x Sn − [xE(Mg + E(Sn)] where x is the number of Mg atoms per Sn atoms, E(Mg x Sn) is the total energy per Sn atoms of Mg x Sn phase, E(Mg) is the total energy per atom of Mg in hcp crystal lattice, and E(Sn) is the total energy per atom of Sn in tetragonal crystal lattice. The optimized lattice parameters, formation energy per atom of all phases are summarized in Table 1. Our results for the crystalline phases are in excellent agreement with reported theoretical and experimental measurements. [53][54][55][56] Obtaining intrinsic elastic properties of single crystal structures using DFT is well established in literature and is explained in length in our earlier works. 46,47 Briefly, when a uniform and infinitesimal strain is applied to a homogeneous system, the internal energy of the system can be expressed in terms of strain components. Using energy-strain curve, one can obtain the elastic constants for all phases. For instance in c-Mg 2 Sn with cubic symmetry, uniaxial distortion, volumetric distortion and pure shear were applied to obtain the three independent elastic constants i.e. C 11 , C 12 , and C 44 , respectively (Voigt notation is used for all elastic constants). 63 The isotropic nature of a-Mg 2 Sn phase only gives two independent elastic constants, C 11 and C 12 similarly obtainable through uniaxial and volumetric strains, respectively. For C 11 , uniaxial strain was applied along three orthogonal directions and average C 11 was determined. In a similar fashion, we obtained the five and six independent elastic constants of Mg and Sn crystal structures, respectively. Moreover, crucial electrochemical properties were calculated including electrode potential -E f /x with respect to Mg/Mg +2 , and specific capacity nF/ΣM , where n is the total number of Mg electrons in the reaction, F is Faraday's constant (26.802 Ah/mol) and ΣM is the total molecular weight of Mg 2 Sn. We also obtained volumetric energy density versus a hypothetical 3.75 V cathode by a method introduced by Obrovac et al. 64,65</p><!><p>Electrochemical Properties. The specific capacity of c/a-Mg 2 Sn phases as used in MIBs is 641 mAh g −1 since both have the same total molecular weight and assuming the same number of Mg electrons participate in the reaction (it is noted that specific capacity in excess of 903 mAh g −1 is also interchangeably reported in literature by taking into account only the molecular weight of Sn instead of Mg 2 Sn). 31,34,36,37 The volume expansion of the optimized a-Mg 2 Sn is about 160 % with reference to pure Sn, which is lower by 20 % when compared to c-Mg 2 Sn. We find the electrode potentials as low as 0.15 and 0.06 V for c, a-Mg 2 Sn, respectively. Comparative low potentials ≈ 0.04 V have been found for fully lithiated amorphous Si (a-Li 3.75 Si), 66 and fully lithiated amorphous C (a-Li 0.75 C 6 ) in LIBs, 67 and have been reported experimentally for electrode materials in rechargeable batteries. 66,68 In Mg-Sn system, as obtained from our calculations, this gives a-Mg 2 Sn phase a volumetric energy density in excess of 8.50 Wh cc −1 greater than 8.25 Wh cc −1 for c-Mg 2 Sn.</p><p>Examination of Amorphous Mg 2 Sn Phase. The optimized c-Mg 2 Sn and a-Mg 2 Sn phases are shown in Fig. 1a and b, respectively. 69 The former has a well-defined atomic positions, whereas the latter has a disordered atomic positions. RDFs results, as shown in Fig. 1c, further demonstrate that while c-Mg 2 Sn is characterized with well-defined peaks indicating the crystallinity of the phase, the peaks of a-Mg 2 Sn are flattened indicating the amorphousity of the phase, similar to amorphous Li-Si phase in Li-ion battery. 70 We also compare the amorphousity of a-Mg 2 Sn at the three chosen temperatures.Evidenced by the RDFs, we obtain qualitatively similar a-Mg 2 Sn phases regardless of the choice of annealing temperature.</p><p>The RDFs give us a qualitative understanding on the formation of a-Mg 2 Sn phase upon annealing. It fails to elaborate on the origin of amorphousity, and provide an atomic-level insight into interatomic bonds. We employ rings statistics technique as implemented in I.S.A.A.C.S. package to obtain such insight and compare it against c-Mg 2 Sn. The a-Mg 2 Sn phase-obtained by annealing at 2000 K-was chosen for further analyses as no significant 2a (in red). We have not seen any significant qualitative differences in ring statistic results when other a-Mg 2 Sn phases, obtained by different annealing temperatures, were used. The number of rings per atom falls within a narrow range of 0.4-0.8 with a relative dominance of 8, 9-membered rings.</p><p>This disordered system characterized by existence of various rings-opposed to c-Mg 2 Snfurther signifies the formation of a-Mg 2 Sn phase. In order to provide an atomic-level view of a-Mg 2 Sn phase, a few multi-membered rings are identified and highlighted in Fig. 2c. C ij s of Mg, Sn and c/a-Mg 2 Sn phases as obtained from our DFT calculations. Our C ij s are in good agreement with the experimental measurements and first principles calculations of elastic constants for crystalline Mg, Mg 2 Sn, Sn phases (see Table 2). [53][54][55][56] These values are elastic constants of single crystals as opposed to polycrystalline structures with randomly oriented single crystals and do not reflect the elastic properties of the microstructure of an alloys anode. 50,71,72 The correlation between the elastic properties of single crystals and an aggregate is made by employing the Voigt and Reuss continuum theories. 73,74 Moreover, Hill demonstrated that the arithmetic average of the Voigt and Reuss values gives a better approximations for the elastic properties of the polycrystalline microstructures. Table 3 presents the bulk moduli (B), Young's moduli (E), and shear moduli (G) of polycrystalline Mg, c-Mg 2 Sn, Sn, and a-Mg 2 Sn adopting Hill averaging method. The B/G ratio, and anisotropy factor of Young's and shear moduli (calculated as A X = X max /X min with X being either Young's or shear moduli) are also given in the table. 75 In addition, we plot Young's and bulk moduli of all phases including a-Mg 2 Sn as a function of the Mg fraction y = x/(x + 1) in Fig. 3, where y= 0 and 1 represents the pure polycrystalline Sn phase and Mg phase, respectively.</p><!><p>It is evident from Table 3 that Young's and shear modulus for c-Mg 2 Sn are greater than that of pure Mg and Sn phases. This can be understood from B/G ratio-an indication to the ductility of metal alloys suggested by Pugh. 76 The ratio less(greater) than 1.75 suggests a brittle(ductile) behavior. 76 Pure Mg and Sn phases are thus ductile materials, while c-Mg 2 Sn is more brittle. The transition from ductile (Sn) to brittle (c-Mg 2 Sn) upon Sn magnesiation could be a responsible factor for a sudden increase in moduli. This elastic behavior may lead to the accumulation of internal stresses causing particle cracking and fracture during discontinuous phase transformation upon magnesiation/de-magnesiation during the formation of c-Mg 2 Sn phase. 36 This phenomenon is distinct from the mechanical response of group-14,15</p><p>elements to sodiation and Li-Si alloy anodes but somewhat similar to Li-Sn system (in which Sn lithiation also experiences a sudden increase in Young's and shear moduli with respect to pure Sn phase). [43][44][45][46][47] There is one remarkable difference: Sn magnesiation is susceptible to electrochemically driven amorphization. Our results demonstrate that a-Mg 2 Sn has become elastically soft by ≈ 20 % with respect to c-Mg 2 Sn. This elastic softening makes the moduli's trend to follow an approximate linear dependence on the Mg fraction as opposed to c-Mg 2 Sn (see Fig. 3). This behavior may suggest a more gradual magnesiation-induced phase transformation in favor of easing the build-up of severe internal stresses. We further see that all crystalline phases are rather elastically isotropic polycrystalline materials due to low anisotropy values consistent with the small deviation of their bulk and shear moduli from the linear dependency (see</p><!><p>In summary, we obtained the fundamental electrochemical and elastic properties of c/a-Mg 2 Sn alloy anodes for Mg-ion batteries. We created the a-Mg 2 Sn phase using annealingquenching technique, and examined its amorphousity through RDFs and ring statistics. We found while the elastic moduli of c-Mg 2 Sn phase show an abrupt increase (as large as 27 %) with respect to pure Sn, a-Mg 2 Sn is elastically softened by 20 % with respect to c-Mg 2 Sn and follows a linear trend of moduli in binary Mg-Sn system. The elastic softening is attributed to the formation of various complex rings resulted in uneven charge distributions. We believe that the electrochemically-induced failures during Sn↔c-Mg 2 Sn phase transformation could be due to the development of incompatible strains and stresses and subsequent sudden change in elastic moduli. This study suggests that the formation of a-Mg 2 Sn phase could be the key to mitigate this problem. The Sn↔a-Mg 2 Sn undergoes a relatively more gradual phase transformation which increases the likelihood of alleviating the build-up of internal stresses and capacity fade during volume change. Our results serve as the first step towards the development of failure-free Mg-Sn alloy anode suitable for the next generation of Mg-ion batteries. They can essentially be used in constructing macroscopic models to better understand deformation behaviors, magnesiation-induced amorphization, and different modes of failures.</p>

ChemRxiv

Cysteine sulfenic acid as an intermediate in disulfide bond formation and non-enzymatic protein folding\xe2\x80\xa0

As a posttranslational protein modification, cysteine sulfenic acid (Cys-SOH) is well established as an oxidative stress-induced mediator of enzyme function and redox signaling. Data presented herein show that protein Cys-SOH forms spontaneously in air-exposed aqueous solutions of unfolded (disulfide-reduced) protein in the absence of added oxidizing reagents, mediating the oxidative disulfide bond formation process key to in vitro, non-enzymatic protein folding. Molecular oxygen (O2) and trace metals (e.g., copper (II)) are shown to be important reagents in the oxidative refolding process. Cys-SOH is also revealed to play a role in spontaneous disulfide-based dimerization of peptide molecules containing free cysteine residues. In total, the data presented expose a chemically ubiquitous role for Cys-SOH in solutions of free cysteine-containing protein exposed to air.

cysteine_sulfenic_acid_as_an_intermediate_in_disulfide_bond_formation_and_non-enzymatic_protein_fold

<!>Reagents<!>Protein Reduction and Refolding<!>Assessment of Refolding Progress<!>Cysteine-containing Peptide Incubation with Dimedone<!>ESI-LC/MS<!>Proteolytic Digestion of RNAse A<!>MALDI-MS<!>Strategy<!>Trapping of Cys-SOH During Protein Refolding<!>Verification of Covalent Dimedone-Protein Product Formation<!>Roles of Oxygen and Metals<!>Cys-SOH Mediates Spontaneous Formation of Peptide Dimers<!>DISCUSSION<!>Conclusions<!>

<p>In the early 1960s Anfinsen and colleagues carried out seminal studies in protein disulfide bond formation and folding, showing that unfolded (disulfide reduced) proteins will spontaneously and completely refold, regenerating full protein activity, over a period of about a day in the absence of denaturants (1-6). For decades it has been known that free thiol-containing molecules will, over time, spontaneously form intra- and/or intermolecular disulfide bonds when stored exposed to air in aqueous solution. But thiols do not spontaneously oxidize one another to generate disulfide bonds in the absence of an oxidizing reagent (7). Moreover, an intermediate oxidized form of sulfur has not been clearly identified as a ubiquitous intermediate in the spontaneous formation of disulfide bonds in vitro. With the discovery that disulfide bond formation and protein folding in vivo is an enzyme catalyzed process that takes place in a matter of seconds, interest in identifying an oxidized sulfur intermediate relevant to non-enzymatic protein folding seems to have waned. Even so, such an intermediate remains important in peptide and protein research / production systems that, intentionally or not, involve non-enzyme mediated disulfide bond formation and protein folding. Knowledge of the identity of such an intermediate may allow for a better degree of control in manipulating in vitro protein systems.</p><p>Biologically, cysteine sulfenic acid (Cys-SOH) formation within folded protein molecules can serve as a means of regulating protein activity—helping to absorb oxidative insults (8, 9), informationally "register" such insults and/or alter protein activity (10-20), mediate redox signaling (16, 17, 21), and generally deflect (14-16, 22-26) what otherwise might have been injurious oxidative damage. In most studies on protein Cys-SOH, exogenous oxidants such as hydrogen peroxide are required to build up significant quantities of Cys-SOH within fully folded proteins.</p><p>In general, Cys-SOH is a transient intermediate that, until the past decade has been difficult to study. It has been understood that Cys-SOH likely renders the sulfur atom of cysteine electrophilic to the point of being susceptible to nucleophile attack (i.e., by a proximal thiol / thiolate anion) (10, 27). Yet, though they now appear to play a key role in the biochemical regulation of many different proteins, Cys-SOH modifications have, historically, been difficult to study due to their inherent instability outside of their native "cocoon-like" protein environment (9, 10, 28-31). Recent developments in analytical technologies and molecular probes—particularly those based on the specific alkylation (or irreversible "covalent trapping") of Cys-SOH residues with 5,5-dimethyl-1,3-cyclohexanedione (dimedone) (10, 32-39) have produced a great deal of interest and a flurry of research activity centered on this unique protein modification.</p><p>The purpose of this study was to determine whether or not cysteine sulfenic acid serves as the oxidized sulfur intermediate that mediates air-induced disulfide bond formation and non-enzymatic protein folding. Using dimedone as a well established (10, 32-39) mass-shifting molecular probe of Cys-SOH (Scheme 1) and bovine ribonuclease A (RNAse A) as a model protein1, we report herein the observation that air and trace metal-generated Cys-SOH is a non-site-specific intermediate in the disulfide bond formation process that occurs as part of the in vitro, non-enzymatic protein folding process.</p><!><p>All chemicals including Ribonuclease A Type III-A from bovine pancreas were purchased from Sigma-Aldrich. Synthetic peptide "IGF 57-70" (H2N-ALLETYCATPAKSE-CO2H) was purchased from American Peptide. Amicon centrifugal concentration units were from Millipore. Endoproteinase Lys-C from Lysobacter enzymogenes was obtained from Roche. Spilfyter Hands-in-Bag atmospheric chambers were purchased from VWR.</p><!><p>RNAse A was dissolved at 10 mg mL-1 in freshly prepared 8 M urea, pH 8.6 (adjusted with 5% (w/v) methylamine (2)), containing 50 mM dithiothreitol and incubated in a rotary shaker at 800 rpm and 37 °C for 1 hour. One hundred microliters of the reduced protein solution was adjusted to pH 3.5 with glacial acetic acid then loaded into an Amicon Ultra-4 5 kDa MWCO centrifugal concentration unit containing 4 mL of 0.1 M acetic acid. The sample was then centrifuged in a swing-bucket rotor for 20 minutes at 4,000 × g. The retentate was then re-diluted with 4 mL of 0.1 M acetic acid and the cycle of concentration and re-dilution was carried out a total of 5 times, resulting in a greater than one million fold dilution of urea and dithiothreitol.</p><p>The protein concentration of the final ~250-μL retentate was generally found to be ~120 μM by absorption spectrophotometry at 276 nm (ε = 9390 M-1cm-1 for reduced RNAse A (4)). To verify complete reduction, a 5-μL aliquot was removed and alkylated with 5 μL of 50 mM maleimide dissolved in 50 mM ammonium acetate, pH 5, and incubated at 50 °C for 15 minutes. The alkylated RNAse A was then analyzed intact by mass spectrometry to verify complete reduction by means of an anticipated 8 × 97 Da mass increase (Supporting Information Figure S1).</p><p>For each set of refolding conditions examined, two aliquots of purified, unfolded RNAse A were diluted to 0.075 mg mL-1 (5.5 μM), adjusted to pH 8 with a saturated solution of Tris base, and left to spontaneously refold at room temperature in the presence or absence of 50 mM dimedone. An additional negative control of native (folded) RNAse A was also incubated in the presence of 50 mM dimedone. When employed, additives such as metals (50 μM CuSO4 or Fe2(SO4)3) and metal chelators (1 mM EDTA and 0.1 mM DTPA) were added from 100-200x stock solutions (in 0.1 M acetic acid or tris acetate buffer) just prior to adjustment of solution pH and addition of protein. Final volumes were typically 1 mL. For incubations under controlled atmospheric conditions samples were placed into a Spilfyter Hands-in-Bag atmospheric chamber which was filled and continually pressurized with high purity nitrogen or oxygen. Samples were withdrawn with a gel loading pipette tip through a resealable pinhole in the top of the chamber.</p><!><p>Refolding progress (in the absence of dimedone) was monitored by taking 5-μL aliquots of sample and mixing with 5 μL of 50 mM maleimide in 0.2 M acetic acid (giving a final pH ≤ 5). Samples were incubated at 50 °C for 15 minutes then diluted to 1 μM final concentration in starting LC solvent before immediately injecting 2 μL on trap. (Dimedone-containing samples were not alkylated, but simply diluted to 1 μM final concentration in starting LC solvent before 2 μL was injected on trap.)</p><p>Refolding progress was monitored by maleimide-alkylation of partially refolded protein and analysis of samples by ESI-MS. As demonstrated by the raw mass spectra (Supporting Information Figures S1 and S3), the procedure employed resulted in sulfhydryl-specific protein alkylation with essentially no alkylation of protein amino groups. To determine the precise degree to which proteins were refolded, deconvoluted ESI mass spectra were integrated. Peak areas were then summed and the fraction of protein in each folding state (i.e., 0, 1, 2, 3, or 4 disulfides—see Supporting Information Figure S3) determined by dividing the area of the appropriate representative peak by the total area. These fractions were then weighted by folding state according to the following equation: Rp=(FSS4∗1+FSS3∗0.75+FSS2∗0.5+FSS1∗0.25+FSS0∗0)∗100Where Rp represents the percentage of protein refolded and FSSn represents the fraction of protein containing n disulfide bonds. Thus, for example, if two equal-area peaks were observed in a deconvoluted ESI mass spectrum representing protein molecules with 3 disulfides and protein molecules with 4 disulfides, the percentage of protein refolded would be reported as 87.5%. Notably, this summary method of reporting protein refolding progress as a single numerical value is useful for comparison with similar non-mass spectrometric techniques, but it mutes data on folding-state molecular statistics (which are available in the raw data) that might be considered informative toward certain kinetic models of folding (which were not under consideration here). In unmuted form—i.e., without applying the above equation—the raw data that are acquired using this technique contain more detailed information on protein disulfide status than is generally available with other non-mass spectrometric techniques.</p><!><p>Three microliters of a 670 μM solution of a single-free-cysteine-containing peptide (H2N-ALLETYCATPAKSE-CO2H) (a.k.a. IGF 57-70) was added to 96 μL of freshly prepared 0.15 M ammonium bicarbonate, pH 7.1 and 1 μL of 1 M dimedone in ethanol, giving final peptide and dimedone concentrations of 20 μM and 10 mM, respectively. A dimedone-lacking control was prepared in parallel. The samples were then allowed to sit exposed to air for 20 hours in the dark at room temperature, at which point they were diluted 30-fold in MALDI matrix solution and analyzed by MALDI-MS and MALDI-MS/MS (TOF/TOF) as described below.</p><!><p>A trap-and-elute form of sample concentration / solvent exchange rather than traditional LC was employed. Samples were injected by a Spark Holland Endurance autosampler in microliter pick-up mode and loaded by an Eksigent nanoLC*1D at 10 μL min-1 (90/10 water/acetonitrile containing 0.1% (v/v) formic acid, Solvent A) onto a protein captrap (Michrom Bioresources, Auburn, CA) configured for unidirectional flow on a 6-port divert valve. After 2 minutes, the divert valve position was automatically toggled and flow over the cartridge changed to 1 μL min-1 Solvent A (running directly to the ESI inlet) which was immediately ramped over 5 minutes to 10/90 water/acetonitrile containing 0.1% (v/v) formic acid. By 7.2 minutes the run was completed and the flow back to 100% solvent A.</p><p>The bulk of the RNAse A eluted between 3.5 and 5 minutes into a Bruker MicrOTOF-Q (Q-TOF) mass spectrometer operating in positive ion, TOF-only mode, acquiring spectra in the m/z range of 50 to 3000. ESI settings for the Agilent G1385A capillary nebulizer ion source were as follows: End Plate Offset -500 V, Capillary -4500 V, Nebulizer nitrogen 2 Bar, Dry Gas nitrogen 3.0 L min-1 at 225 °C. Data were acquired in profile mode at a digitizer sampling rate of 2 GHz. Spectra rate control was by summation at 1 Hz.</p><p>One to two minutes of recorded spectra were averaged across the chromatographic peak apex of RNAse A elution. The ESI charge-state envelope was deconvoluted with Bruker DataAnalysis v3.4 software to a mass range of 1000 Da on either side of any identified peak.</p><!><p>Reduced RNAse A that had been incubated in air for more than 20 hours in the absence or presence of dimedone was digested overnight at 37 °C and pH 7.5 with Lys-C at a protein to protease mass ratio of 3:1. One microliter of the resulting digestion mixture was added to 4 μL of MALDI matrix solution (see below) and allowed to dry on the MALDI target.</p><!><p>Peptides were mixed with MALDI matrix solution (33% (v/v) acetonitrile containing 0.4% (v/v) TFA and saturated with α-cyano-4-hydroxycinnamic acid) and spotted onto a gold-surfaced MALDI target and allowed to dry. Single stage and LIFT-TOF/TOF mass spectra were acquired on a Bruker Ultraflex MALDI-TOF/TOF instrument. Externally calibrated mass spectra were acquired in positive ion mode with the reflector engaged. For LIFT-TOF/TOF experiments, precursor ion selection width was set on an individual peptide basis to ensure that no undesired parent ions would contaminate the MS/MS spectra. Instrument voltages and other parameters were optimized for peptide resolution and sensitivity; at least 7,500 laser shots were acquired per sample to ensure excellent ion counting statistics.</p><!><p>Scheme 1 depicts the hypothesized ambient oxygen and trace metal-induced formation of Cys-SOH in an unfolded protein molecule. Previous studies have demonstrated that in the absence of a trapping agent (which allows for detection of Cys-SOH) a facile reaction occurs between hydrogen peroxide-generated protein Cys-SOH and intramolecular thiols (40). (Intermolecular protein thiols are also reactive with peroxide-generated Cys-SOH (7, 41).) We therefore reasoned that if Cys-SOH were to be detected within a refolding protein, then it must be serving as an oxidized sulfur intermediate in the in vitro formation of intramolecular disulfide bonds.</p><!><p>To determine the potential role of air-generated Cys-SOH in disulfide bond formation and in vitro, non-enzymatic protein folding, disulfide-reduced RNAse A was incubated at room temperature with exposure to air (analogous to Anfinsen and co-workers' seminal experiments in protein folding (1-6)) in the presence of the Cys-SOH trapping reagent dimedone (32-39). Deconvoluted ESI-mass spectra of intact RNAse A incubated in the presence of dimedone were acquired (Figure 1). Mass shifts caused by covalent dimedone reaction products were present. Control experiments conducted in the absence of dimedone and, separately, in the presence of dimedone but with native (fully folded) RNAse A did not result in dimedone-modified RNAse A as demonstrated (Figure 1).</p><!><p>To verify that dimedone was indeed covalently bound to RNAse A (due to its reaction with Cys-SOH), the samples described in Figure 1, panels A and C were digested with Lys-C and analyzed by MALDI-MS and MALDI-MS/MS (TOF/TOF). Mass spectra of the resulting peptide mixtures were acquired (Figure 2). With one exception, observed monoisotopic m/z values aligned to within 20 ppm of the calculated m/z values (Table 1). Average mass accuracy was 13.7 ppm (including the noted outlier). MS/MS (TOF/TOF) spectra for three of these peptides containing non-redundant cysteine residues were acquired; one of these spectra is shown as Panel C in Figure 2 (the others are available online as Supporting Information Figure S2). In agreement with the single-stage (peptide mapping) MALDI mass spectra, each of these MS/MS spectra confirm the covalent attachment of dimedone to the suspected cysteine residue.</p><!><p>Given its elemental composition, it is clear that formation of Cys-SOH requires oxygen. We hypothesized that diatomic oxygen from air contributes to Cys-SOH formation through the mediation of trace metals in solution. To elucidate the roles of ambient oxygen (O2) and trace metals on the rate of protein refolding, additional experiments were conducted under an oxygen atmosphere, under a nitrogen atmosphere, or in the presence of added copper (II), iron (III) or metal chelators (Figure 3). Samples of RNAse A were allowed to refold until either the refolding process was >95% complete or four days had passed—whichever came first. In the cases of refolding under a nitrogen atmosphere or refolding in the presence of metal chelators (in air), four days passed before 80% refolding was reached. It took just over two days for RNAse A to refold under air without added metals. The addition of 50 μM iron (III) had only a marginal effect on this rate. Refolding under a pure oxygen atmosphere was >95% complete within 24 hours and refolding in the presence of 50 μM copper (II) was >95% complete in less than 2 hours. For each sample described in Figure 3, parallel samples were refolded in the presence of dimedone. In each case covalent dimedone-RNAse A reaction products were discovered as described above. (A time course showing the degree of covalent incorporation of dimedone into refolding RNAse A over time is available as Supporting Information Figure S4.) Notably, none of the experimental alterations to the refolding environment resulted in formation of cysteine sulfinic acid (-SO2H).</p><!><p>The non-site-specific nature of Cys-SOH formation during RNAse A refolding suggested a generalized chemical phenomenon. To verify the hypothesis that Cys-SOH forms spontaneously on free thiols that are dissolved in neutral-to-slightly-alkaline aqueous solution and exposed to air, a 20 μM solution of the single-free-cysteine-containing peptide "IGF 57-70" (H2N-ALLETYCATPAKSE-CO2H) was placed under such conditions in the presence of dimedone for approximately 20 hours. MALDI-TOF mass spectra were acquired (Figure 4). Besides showing the expected formation of ample quantities of disulfide-linked homodimer, Panels B and C confirm that dimedone was found covalently attached to the cysteine residue of the monomeric peptide—indicating that Cys-SOH had formed under the mild conditions of neutral-to-slightly-alkaline pH and exposure to air.</p><!><p>The data shown here demonstrate air and trace metal induced formation of Cys-SOH as part of the reduced RNAse A protein molecule. When allowed to proceed for a long enough period of time, the refolding process completes and returns full activity to the RNAse A molecule (2-4, 6). Notably, arsenite is considered a reducing agent specific for Cys-SOH that does not affect protein disulfides (13, 42). As such, if Cys-SOH is involved in the formation of protein disulfide bonds, arsenite should inhibit in vitro protein folding. Such inhibition was recently well documented by Ramadan et al (43). Thus, considering 1) the detection of covalent dimedone-protein reaction products on cysteine residues within the unfolded protein in the absence of added oxidants, 2) that arsenite inhibits oxidative protein folding in vitro (43), 3) that no detectable Cys-SOH forms within a folded RNAse A molecule (Figure 1b), and 4) the known reactivity of free sulfhydryl groups with Cys-SOH (7, 8, 10, 21, 40, 41, 44-46), it is logical to conclude that ambient oxygen-induced Cys-SOH serves as an intermediate in the spontaneous disulfide bond formation process that takes place as part of in vitro, non-enzymatic protein folding. In 2007, Johansson and Lundberg (45) presented evidence for the spontaneous formation of protein-glutathione mixed disulfides. Based on inhibition of the process with dimedone and (separately) arsenite, they suggested that the mixed disulfides were formed via a Cys-SOH intermediate. Here we extend their studies by physically trapping and directly analyzing a spontaneously formed Cys-SOH intermediate which mediates intramolecular disulfide bond formation and in vitro protein folding. Additionally, we show that such natural Cys-SOH species mediate disulfide bond formation between free cysteine-containing peptides, suggesting a ubiquitous role for Cys-SOH in the spontaneous, frequently undesirable formation of disulfide-linked intermolecular dimers.</p><p>The data presented in Figure 3 demonstrate that molecular oxygen and trace metals are important reagents in the spontaneous in vitro generation of Cys-SOH. Compared to a "natural" folding rate of about two days in air (grey diamonds), refolding under nitrogen or in the presence of metal chelators (EDTA and DTPA) in air takes several days—but refolding under an oxygen atmosphere takes about one day and refolding in the presence of 50 μM copper (II) takes just a couple of hours. Notably, Anfinsen observed complete refolding of RNAse A in air in about 20-24 hours (3, 5). The differences between Anfinsen's results and those reported here may be due to the fact that we were able to utilize highly pure deionized water and virgin polypropylene test tubes which are likely to minimize trace metal contamination compared to reagents and reaction vessels available several decades ago.</p><p>The apparent plateau phases reached by samples folding under nitrogen or in the presence of metal chelators (Figure 3) may indicate that the extremely low quantities of available oxygen and trace metals, respectively, were effectively depleted during the refolding process—and that because of this the refolding process may never fully complete under these conditions. The mechanisms and rate laws governing formation of Cys-SOH are the subject of future investigation.</p><p>Notably, the means by which the folding process was monitored in these studies using maleimide alkylation and monitoring by mass spectrometry provides a unique viewpoint into the protein folding process by allowing for direct determination of the relative number of protein molecules in each state of disulfide bond formation at any given point in time. Though it is not obvious by the way the data are plotted in Figure 3, as explained in the experimental section, each data point was gleaned via a mass spectrum in which information on the relative abundance of protein molecules with 0, 1, 2, 3, or 4 disulfide bonds was unambiguously provided (see Supporting Information Figure S3 for as series of illustrative mass spectra). By examining these relative populations of protein folding states over time, it is clear that formation of the last disulfide bond is the slowest step in the refolding of RNAse A. For example, the sample containing copper (II) was nearly 80% folded within one minute, but took almost 2 hours to reach >95%. These observations support the RNAse A disulfide folding mechanism asserted by Creighton over 30 years ago (47, 48).</p><!><p>The data presented here demonstrate that oxygen and trace metal-generated Cys-SOH is a ubiquitous modification of cysteine residues in solution and an intermediate that mediates in vitro, non-enzymatic disulfide bond formation and protein folding.</p><!><p>Charge deconvoluted "singly charged" ESI-mass spectra of A) Negative control in which reduced RNAse A was allowed to refold in the absence of dimedone at room temperature, pH 8, and exposure to air. "X" indicates the absence of a peak at the expected m/z value of dimedone-modified RNAse A; B) Negative control in which native (non-reduced / folded) RNAse A was exposed to refolding conditions in the presence of 50 mM dimedone; C) Reduced and purified RNAse A refolded in the presence of 50 mM dimedone. The calculated MH+ mass of fully reduced RNAse A is 13691.3 and that of fully folded RNAse A is 13683.2.</p><p>Reflector-mode MALDI mass spectra of Lys-C-digested RNAse A. A) RNAse A that was refolded in the absence of dimedone. "X" indicates the absence of peaks that are found when RNAse A is refolded in the presence of dimedone. B) RNAse A refolded in the presence of 50 mM dimedone. Circled m/z values indicate peaks representative of covalent dimedone-modified RNAse A peptides. These peptides are shifted up in mass by that of covalently attached dimedone (138.07 Da). (Non-covalently attached dimedone would shift the mass by 140.08 Da.) Table 1 provides a list of pertinent calculated and observed masses. Though not visible, isotopic clusters are baseline-resolved, hence m/z values indicated are monoisotopic. C) MALDI-MS/MS (TOF/TOF) spectrum of the peptide represented at m/z 2310 in Panel B. Assigned peaks are accurate to within 0.3 Da. The y-ion series indicates that the cysteine residue is shifted up in mass by 138 Da. MS/MS spectra for the dimedone-modified peptide ions represented by peaks at m/z 996 and 2878 in Panel B are available as Supporting Information.</p><p>RNAase A refolding progress over time under different atmospheric and solution conditions. Grey diamonds represent RNAse A refolding under air without special additives.</p><p>MALDI mass spectra of synthetic peptide "IGF 57-70" (H2N-ALLETYCATPAKSE-CO2H) incubated for 20 hours at room temperature, pH 7.1 in the A) absence of dimedone and B) presence of 10 mM dimedone. Isotopic clusters are baseline resolved and indicated m/z values are monoisotopic. The final rows in Table 1 provide a list of pertinent calculated and observed masses. C) MALDI-MS/MS (TOF/TOF) mass spectrum of the dimedone-modified peptide at m/z 1634. The b- and y-ion series indicate that the cysteine residue is shifted up in mass by 138 Da.</p><p>Proposed reaction pathway for in vitro generation of disulfide bonds and protein folding. Dimedone reacts covalently and irreversibly with Cys-SOH, making it an effective probe for the existence of Cys-SOH (10, 32-39). Detection of covalent dimedone-modified protein in the experiments described here indicates the production of protein Cys-SOH. By logical extension of previous studies on the reactivity of intra- and intermolecular sulfhydryl groups with Cys-SOH (7, 40), such detection suggests that Cys-SOH is an oxidized sulfur intermediate that mediates in vitro disulfide bond formation and protein folding. Early experiments by Anfinsen and colleagues (2-4, 6) showed that in vitro protein folding goes to completion and full restoration of protein activity.</p><p>Calculated and observed monoisotopic masses of cysteine-containing peptides and their corresponding covalent dimedone-modfied peptides that were observed upon MALDI-TOF analysis of Lys-C digested RNAse A that was refolded in the presence of 50 mM dimedone. IGF 57-70 is the synthetic peptide (H2N-ALLETYCATPAKSE-CO2H) incubated for 20 hours in the presence of dimedone (see Figure 4). Observed m/z values aligned to within 20 ppm of the calculated m/z values (with the single exception indicated).</p><p>Mass accuracy outlier at 35 ppm.</p>

PubMed Author Manuscript

Ubiquitination, intracellular trafficking, and degradation of connexins

Gap junction channels provide a conduit for communication between neighboring cells. The function of gap junction channels is regulated by posttranslational modifications of connexins, the proteins that comprise these channels. Ubiquitination of connexins has increasingly been viewed as one mechanism by which cells regulate the level of connexins present in cells, as well as the corresponding intercellular communication. Here we review the current knowledge of connexin ubiquitination and the effects this may have on gap junctional communication.

ubiquitination,_intracellular_trafficking,_and_degradation_of_connexins

INTRODUCTION<!>Cx43 Ubiquitination<!>Internalization, endocytic trafficking, and lysosomal degradation<!>Autophagy<!>Proteasomal degradation<!>Cx32 ERAD<!>Cx43 degradation and disease<!>Concluding Remarks<!>

<p>Gap junctions are comprised of connexin hexamers, or connexons, that dock across the intercellular space with connexons on adjacent cells [1]. These plasma membrane channels mediate direct cell to cell communication, allowing the transfer of molecules less than 1000 daltons, such as small metabolites and secondary messengers. Gap junctional regulation and therefore, intercellular communication is critical for the proper maintenance of normal cell and tissue function [2–4].</p><p>Connexins are four-pass transmembrane proteins, with two extracellular loops, an intracellular loop, and N- and C-terminal cytoplasmic regions. Particularly for transmembrane proteins, connexins have remarkably short half-lives, ranging from 1.5 to 5 hours [5–9]. Since the amount of connexin protein correlates directly to the level of gap junctional intercellular communication (GJIC) 1, an understanding of how connexin levels are regulated is critical. There has been a great deal of work contributing to the current understanding of the processes that regulate GJIC and connexin turnover.</p><p>Connexins are co-translationally inserted into the endoplasmic reticulum (ER), then trafficked through the Golgi. At some point during the secretory pathway, connexins oligomerize into hexameric connexons. Depending on the connexin isoform, oligomerization has been observed in different locations from the ER, to the ER-Golgi intermediate compartment, to the trans-Golgi network [10–15]. These connexons, that will comprise one-half of the complete gap junction channel, are transported to the plasma membrane to form gap junctions and large accretions of gap junctions, known as gap junction plaques. Recently, undocked connexons at the plasma membrane have been demonstrated to have channel activity as well. These hemichannels can be opened under conditions such as mechanical shear stress [16, 17], membrane depolarization [18], and changes in ionic concentrations [19, 20], and can regulate the passage of ions and metabolites [16, 17, 21, 22]. Hemichannels may also have important roles in tissue remodeling [17, 23] and in cell death [24–27]. From the plasma membrane, connexins are internalized as double membrane annular gap junctions, or connexosomes, or possibly, as undocked individual connexons, then degraded [28]. It is clear that connexins undergo degradation through both the lysosomal and proteasomal degradation pathways [9, 29–37], with an additional pathway of autophagy recently revealed [38, 39] (Figure 1). However, the precise mechanisms that regulate connexin trafficking to and from the plasma membrane and subsequently through to degradation are still being studied and debated. Posttranslational modifications of connexins are thought to contribute to the regulation of connexin function and trafficking. Connexin phosphorylation, which is known to regulate gap junction channel closure and connexon membrane stability (reviewed in [40–42]), has been studied intensely. More recently, other modifications have been identified such as hydroxylation, methylation, and acetylation [43–45], the last of which has been reported to affect connexin43 (Cx43) localization in the heart [43]. This review will focus on a different type of posttranslational modification that may also have a significant impact on the connexin life cycle. This modification, which has been increasingly studied by a number of groups, is the ubiquitination of connexins, and we will discuss what effect this modification has on connexin trafficking and turnover (relevant studies summarized in Table 1).</p><p>Ubiquitination (or ubiquitylation) is a process that had primarily been studied as having a role in marking substrates for degradation by the 26S proteasome, with additional functions subsequently revealed. Ubiquitin is a 76 amino acid polypeptide that is highly conserved and expressed in all eukaryotes, from yeast to humans. In ubiquitination, an ubiquitin moiety is covalently attached to target substrates through a series of enzymatic events (reviewed in [46]), which begins with the E1 ubiquitin-activating enzyme forming a high-energy thioester bond with ubiquitin, activating ubiquitin. The activated ubiquitin is then loaded onto an E2 ubiquitin-conjugating enzyme, which will associate with an E3 ubiquitin ligase. The E3 ligase facilitates the covalent linkage of ubiquitin with either the target substrate at lysine residues, or with another ubiquitin forming a polyubiquitin modification. The creation of the polyubiquitin chains has been observed in vivo to attach at the lysine29, lysine48, and lysine63 residues in ubiquitin [46]. While lysine48 linkages are believed to function as the proteasomal degradation tag, lysine63 linkages are proposed to act as internalization signals during endocytosis [47, 48]. This review will focus primarily on the studies of the ubiquitination of Cx43, the most studied connexin regarding connexin ubiquitination.</p><!><p>The earliest reports of Cx43 ubiquitination arose from studies of Cx43 degradation. The first report of Cx43 ubiquitination was from Laing and Beyer in 1995 [30]. A cell line expressing thermo-labile E1 enzyme was used to study Cx43 ubiquitination and degradation in the absence of E1 activity. Loss of the E1 ubiquitin activation at the restrictive temperature resulted in the stabilization of Cx43 protein, which was presumed to be caused by the loss of ubiquitination and thus, the targeting of Cx43 to the proteasome for degradation. Additionally, sequential immunoprecipitations (first of Cx43, followed by ubiquitin) detected higher molecular weight Cx43 protein that was not detected in the presence of competing free ubiquitin [30]. These data suggested that Cx43 exists in cells with covalently-attached ubiquitin moieties, and perhaps, that ubiquitination is important for regulating Cx43 protein levels.</p><p>Subsequent studies reported that Cx43 ubiquitination could result from the cellular exposure to epidermal growth factor (EGF) and the phorbol ester 12-O-tetradecanoylphorbol 13-acetate (TPA) [33, 34]. Use of ubiquitin antibodies that had different specificities to mono- and polyubiquitin indicated that Cx43 was monoubiquitinated at multiple lysine sites in response to TPA [33]. Cx43 multi-monoubiquitination was confirmed in untreated conditions [49], which is significant because monoubiquitination is believed to be involved in proteasome-independent processes, such as internalization and localization [48, 50]. A decrease in the amount of Cx43 at the plasma membrane was also observed in response to both EGF and TPA treatments [33, 34].</p><p>The presence of ubiquitinated Cx43 in actual gap junction plaques at areas of cell-cell contact was captured in an immunoelectron microscopy study using freeze-fracture followed by immunogold labeling [51]. Using the fungal metabolite brefeldin A (BFA) to block the transport of Cx43 from the Golgi to the plasma membrane, a high amount of plaques were observed to be ubiquitinated (50%), which dropped to only 14% after a one hour wash-out of BFA. A reduction in GJIC was also observed with the BFA block that was reversed upon BFA wash-out [51]. These data indicated that ubiquitination might occur as a natural part of the Cx43 life cycle, where it may have a role in the trafficking of older gap junctions from the plasma membrane.</p><p>One key, but unknown, component in the process of Cx43 ubiquitination was the identity of the E3 ligase that was responsible for the ubiquitination. Nedd4 was the first E3 ligase identified initially as a Cx43 binding partner that appeared to be required for Cx43 internalization, as loss of Nedd4 by siRNA resulted in the accumulation of Cx43 gap junction plaques at the plasma membrane [52]. Further studies on Nedd4 reported that the presence of Nedd4 was required for Cx43 ubiquitination [49]. Other groups have observed the interaction of Cx43 with additional E3 ligases, another member of the Nedd4 family WWP1, and the RING E3 ligase, TRIM21. TRIM21 and WWP1 appear to have active roles in Cx43 ubiquitination (V. Chen, personal communication and L. Matesic, personal communication, respectively).</p><!><p>Cx43 internalization from the plasma membrane and subsequent trafficking through the endocytic pathway to the lysosome for degradation has been extensively studied. TPA and EGF cause a reduction in Cx43 levels at the plasma membrane [33, 34, 53], and are known to induce Cx43 internalization through the MAPK signaling pathway [33, 34, 54, 55]. Cx43 turnover following TPA treatment appeared to be due to both lysosomal and proteasomal degradation [33]. Since Cx43 monoubiquitination, and not polyubiquitination, was detected after TPA treatment, the proteasomal degradation data was surprising as monoubiquitination is not normally observed in proteasomally-degraded substrates. However, it was thought that the proteasomal inhibitors might have an indirect effect on Cx43 levels in the TPA response [33, 56]. Further work found that the subset of Cx43 localized to plaques in the plasma membrane was ubiquitinated in response to TPA treatment [56]. EGF treatment also causes Cx43 internalization and ubiquitination [34]. The type of Cx43 ubiquitination that occurs in response to EGF has not been analyzed and, therefore, it is unclear whether Cx43 is polyubiquitinated (such as by lysine48 linkages) for proteasomal degradation after EGF treatment or monoubiquitinated, as in the case of TPA treatment. In addition, a recent study of ubiquitinated Cx43 proposed that Cx43 may undergo a non-canonical internalization that is mediated by an unidentified ubiquitin modification that is not dependent on the YXXØ tyrosine-sorting signal previously identified to be required for membrane Cx43 internalization [57]. These data suggest that Cx43 ubiquitination is a determinant for the specific cellular internalization mechanism or pathway.</p><p>The previous studies of EGF and TPA-induced Cx43 internalization and ubiquitination also provided evidence of proteasomal involvement in the mechanism of Cx43 turnover [33, 34]. However, TPA induced the monoubiquitination of Cx43, which is not typically the ubiquitin tag that leads to proteasomal degradation [33]. In addition, it appeared that EGF stimulated Cx43 proteasomal degradation, because treatment with proteasomal inhibitors, but not lysosomal inhibitors, alleviated the EGF-induced reduction of Cx43 [34]. One possible explanation for these results is that another protein was more directly responsible for Cx43 internalization and turnover, and that the activity of this protein was somehow regulated by proteasomal degradation. A recent study using a Cx43 mutant that cannot be ubiquitinated suggested that proteasomal degradation is an indirect stimulator of Cx43 membrane localization [58]. This Cx43 ubiquitination mutant contained a series of point mutations where all the lysine residues (ubiquitin acceptor sites) were mutated to arginine residues. In the absence of ubiquitination, Cx43 was still trafficked to the plasma membrane and formed functional channels, although initial studies suggested the possibility that the gap junctional communication might not be efficient as gap junctions formed with wild-type Cx43. Akt/protein kinase B (PKB) was identified as the link between ubiquitination and the effect of proteasomal inhibition on Cx43 membrane stabilization. Akt has previously been demonstrated to phosphorylate Cx43 [59], and Akt phosphorylation of Cx43 in the cell membrane stabilized Cx43 in the membrane. Akt ubiquitination and subsequent degradation is one mechanism to regulate its activities, so disruption of Akt turnover by blocking its proteasomal degradation increases the amount of active kinase available to stabilize Cx43. Thus, Akt ubiquitination and degradation by the proteasome indirectly affects Cx43 levels at the plasma membrane [58]. However, the effects of complete loss of Cx43 ubiquitination on internalization and subsequent intracellular trafficking is unclear.</p><p>Biochemical fractionation identified the presence of ubiquitinated Cx43 in both the double membrane plaques (previously described as the TritonX-100 insoluble fraction [60]) and the soluble protein fraction, suggesting that the intracellular endocytosed Cx43 was also or still ubiquitinated [33]. Ubiquitination has been proposed to modulate the endocytic trafficking of Cx43 from gap junction plaques. Interestingly, presumably ubiquitinated Cx43 was observed to colocalize with two known ubiquitin-binding proteins, hepatocyte growth factor-regulated tyrosine kinase substrate (Hrs) and tumor susceptibility gene 101 (Tsg101) [56]. Both proteins are involved in the ESCRT (endosomal sorting complex required for transport) machinery for endocytic trafficking and reportedly mediate the trafficking of ubiquitinated growth factor receptors from the endosome to lysosomes [61]. Upon first glance, depletion of Hrs and/or Tsg101 by siRNA appeared to have little effect on Cx43, as membrane localization and GJIC are not significantly affected [56]. However, prolonged induction of internalization with TPA in concert with a block of protein synthesis in the siRNA-treated cells resulted in increased Cx43 levels at the plasma membrane and enhanced GJIC. This study proposes that Hrs and Tsg101 are responsible for directing Cx43 towards lysosomal degradation by sorting Cx43 at the early endosome, instead of possibly being recycled back to the plasma membrane. Cx43 recycling has previously been demonstrated during cellular stress situations [37] or mitosis [62]. Furthermore, prolonged (3 hours or more) induction of Cx43 internalization by TPA treatment in cells depleted for Hrs and Tsg101 by siRNA resulted in increased Cx43 localization at the plasma membrane, increased amounts of phosphorylated Cx43, and increased levels of GJIC compared to a shorter time point of 1.5 hours [56]. This strongly suggested that the Hrs and Tsg101 ubiquitin-binding proteins were necessary to dictate the movement of Cx43 towards lysosomal degradation, instead of being recycled back to the plasma membrane to again contribute to GJIC. Additionally, the loss of both Hrs and Tsg101 resulted in the increased amounts of ubiquitinated Cx43 and the ubiquitination of Cx43 at the membrane was observed to be dependent on the proteasome. This effect was thought to result from the depletion of intracellular pools of ubiquitin when proteasomal degradation is blocked, which would result in a reduction in the amount of ubiquitin available for conjugation [56]. The Hrs/Tsg101 study was followed shortly by the report of an interaction between ubiquitinated Cx43 and epidermal growth factor receptor substrate 15 (Eps15), another ubiquitin-binding domain-containing protein that is a part of the ESCRT machinery [49]. These studies suggested that a highly-regulated mechanism mediates Cx43 internalization from the plasma membrane, trafficking through the endo-lysosomal pathway, and ending with Cx43 degradation. The internalization and trafficking of the previously discussed lysine mutant form of Cx43 from the plasma membrane has not yet been extensively studied so it is not known how the loss of Cx43 ubiquitination affects the trafficking of Cx43 through the endo-lysosomal pathways as the interactions with the ubiquitin-binding proteins Hrs, Tsg101, and Eps15 are likely to be affected.</p><!><p>Besides direct lysosomal degradation, autophagy is a degradation pathway that utilizes lysosomes for the ultimate step of proteolysis, but is distinct from the endo-lysosomal degradation pathway (reviewed in [63]). Ubiquitination has not been demonstrated to be a requirement for autophagy; however, an ubiquitin-binding protein, p62/SQSTM1, is a substrate that is degraded via autophagy. p62 contains the ubiquitin-associated (UBA) ubiquitin-binding domain which has been demonstrated to regulate the turnover of ubiquitinated proteins (reviewed in [64]). A growing body of work suggests that p62 is a receptor for ubiquitinated proteins that can contribute to the autophagic degradation of its substrates (reviewed in [65]).</p><p>The involvement of autophagy in regulating Cx43 was not recognized until recently, when two studies identified Cx43-containing autophagosome structures [38, 39]. In the heart, Cx43 is normally localized to the intercalated disc regions of the plasma membrane in cardiomyocytes where functional gap junctions are responsible for propagating the electrical signal required for the normal cardiac rhythm. In heart disease, Cx43 is relocalized away from the intercalated discs to the lateral membranes of the cardiomyocyte cells [66–69]. Electron microscopy studies of failing canine heart myocardium showed Cx43-containing multilamellar membrane structures near the lateral membranes that morphologically resembled autophagosomes [39]. To further test the theory that internalized Cx43 gap junctions were in, or associated with, autophagosomes, the colocalization of Cx43 with a marker for autophagosomes, GFP-LC3 (microtubule-associated protein light chain 3), was examined in transfected HeLa cells. Colocalization of Cx43 and LC-3 was observed. This study also analyzed the targeting of Cx43 to lipid rafts, which has been previously reported [70–72]. It was suggested that the observed multilamellar structures might contain Cx43 targeted to lipid rafts [39]. Using TritonX-100 solubility fractions, two fractions of Cx43 were identified: one fraction that also contained lipid raft markers such as the sphingolipid GM1 and caveolin-3, and a denser, less buoyant fraction. There was an increase in the amount of Cx43 in the lipid raft fractions in the failing heart tissues, as well as an increase in the amount of the autophagy marker LC3-II, which is a cleaved form of LC3-I that results with the onset of autophagy [39]. One suggestion is that in diseased heart tissue, Cx43 is removed from the intercalated disc area, processed through lipid raft areas of the plasma membrane as a part of the degradation machinery that would involve autophagosomes [39]. A relationship between lipid rafts and autophagosomes has not yet been established and thus, will require more evidence to confirm. In a subsequent study, using cultured cells, the involvement of autophagy in Cx43 degradation was observed in cells subjected to starvation conditions, which induces autophagy [38]. Cx43 was found to localize to cytoplasmic structures surrounded by LC3, and blocking autophagic degradation with a lysosomal inhibitor or siRNA of the autophagy-related Atg proteins (Atg5) prevented the starvation-induced loss of Cx43 protein as well as increased the colocalization of Cx43 with GFP-LC3. Electron microscopy detected the double membrane structures that are characteristic of autophagosomes. p62, which has been implicated in the autophagy of ubiquitinated proteins, also colocalized with Cx43 [38]. However, it is still unclear what role the ubiquitination of Cx43 has on Cx43 autophagy. Since p62 has the UBA ubiquitin-binding domain, this may serve as the mechanism that targets ubiquitinated Cx43 to the autophagosome.</p><p>The involvement of autophagy in the turnover of gap junction plaques from the plasma membrane has also been recently reported where internalized intact annular gap junctions colocalized with LC3, p62, and other autophagy-related proteins. In addition, inhibition of autophagy by siRNA knock-down of the Atg proteins, Atg6 (beclin-1), or Atg8 (LC3) reduced Cx43 turnover, resulting in the accumulation of cytoplasmic annular gap junctions (M. Falk, personal communication). In light of the recent data suggesting that ubiquitination of Cx43 has an effect on the post-internalization trafficking of Cx43 from the plasma membrane, it would be interesting to determine whether the p62 ubiquitin-binding protein is part of the mechanism that determines the fate of internalized and ubiquitinated Cx43 gap junctions.</p><!><p>Like lysosomal degradation, there is extensive evidence of a role for proteasomal degradation in the Cx43 life cycle. The proteasomal degradation pathway is regulated by a multitude of proteins that result in the trafficking of target substrates to the 26S proteasome holoenzyme complex. This complex is comprised of two subunits, the 20S core particle (CP) and the 19S regulatory particle (RP) (reviewed in [73]). The 20S CP consists of four rings containing seven subunits that are stacked on one another. The β-type subunits comprise the inner two rings and are proteolytically active, while the α-type subunits comprise the outermost rings. The 19S RP, or cap complex, consists of a base and lid, and flank the core, with one RP on each end of the 20S CP. Several proteins associated with the 19S RP, such as Rpn1 and Rpn10, are able to bind ubiquitinated proteins [74–78].</p><p>The study that initially identified Cx43 ubiquitination pointed to a role for ubiquitin-mediated proteasomal degradation as a major process of Cx43 turnover [30]. Additional studies using proteasomal inhibitors further implicated the proteasome in regulating Cx43 levels at the plasma membrane and the corresponding effect on GJIC [9, 32, 36]. Two reasons for this effect have been proposed: first, by blocking proteasomal degradation, Cx43 in the ER has more time to fold properly and to traffic to the membrane, resulting in more Cx43 gap junctions. Or, alternatively, the proteasome has an immediate effect on membrane-localized Cx43, either by facilitating the degradation of that subset of Cx43 directly, or indirectly, by facilitating the degradation of another protein involved in Cx43 localization or trafficking.</p><p>The degradation of Cx43 through the process of ER-associated degradation (ERAD) has been demonstrated through a series of studies. Increasing the amount of Cx43 in the ER by blocking trafficking to the cell membrane using BFA, plus a concurrent block of proteasomal degradation, elevated Cx43 protein levels [32], which was suggestive of ERAD. Further studies using ER stress inducers, such as DTT, which prevents the disulfide bond formation in mature Cx43 protein, revealed enhanced dislocation from the ER into the cytoplasm and elevated proteasomal degradation in response to the ER stress [9, 36], again supporting a role for ERAD in Cx43 degradation. While proteasomal degradation and ERAD are typically associated with ubiquitinated substrates, the ubiquitination state of Cx43 undergoing ERAD has not been studied in depth.</p><p>Our laboratory identified a novel Cx43-interacting protein through a yeast two-hybrid screen that we called CIP75 for Cx43-interacting protein of 75 kDa [79]. CIP75 is a member of the ubiquitin-like (UbL)-UBA domain family of ubiquitin-binding proteins. With an N-terminal UbL domain, and one or more UBA domains at the C-terminus, this family of proteins has been demonstrated to bind to ubiquitin and ubiquitinated proteins via the UBA domain, as well as subunits of the proteasome via the UbL domain (reviewed in [64]). Our initial studies suggested that CIP75 was involved in facilitating Cx43 proteasomal degradation, as overexpression of CIP75 reduced the half-life of Cx43 and the knock-down of CIP75 by siRNA had the opposite effect of significantly increasing the Cx43 half-life [79]. Use of a proteasomal inhibitor indicated that this effect was via the proteasomal degradation pathway. Proteins that are targeted for degradation by the proteasome are typically marked for degradation by the covalently-attached ubiquitin tag so the ubiquitination state of the subset of Cx43 that specifically interacted with CIP75 was examined. Prior reports of Cx43 ubiquitination and ubiquitination of proteasome substrates suggested that the Cx43 interacting with CIP75 would be ubiquitinated [30, 33, 34, 46]. Surprisingly, extensive work demonstrated that while CIP75 is indeed an ubiquitin-binding protein, which was capable of binding both monoubiquitin and lysine48-linked tetraubiquitin, as well as ubiquitinated proteins, the subset of Cx43 that interacted with CIP75 did not appear to be ubiquitinated [80]. Our initial experiments included biochemical immunoprecipitations that failed to detect interacting ubiquitinated Cx43. We subsequently generated a series of three lysine to arginine Cx43 point mutants in an attempt to eliminate any covalent ubiquitin modification. The specific lysines that are ubiquitinated in Cx43 have not been identified so we mutated the lysines in or near the CIP75 binding domain in mutant 1, all the lysines in the C-terminal tail, which contains the CIP75 binding region, in mutant 2, and, all 27 lysines in Cx43 in mutant 3. In all cases, CIP75 was able to interact with Cx43, indicating that Cx43 ubiquitination is not a requirement for interaction with CIP75 [80]. While this result was unexpected, structural NMR data supported this conclusion [81]. An analysis of the binding region of the CIP75 UBA domain for the C-terminal tail of Cx43, as well as for ubiquitin, determined that both Cx43 and ubiquitin interact with overlapping regions of CIP75. This suggested a mechanism involving the competition between ubiquitin or ubiquitinated proteins and the non-ubiquitinated Cx43 C-terminus, which may be indicative of the regulation of CIP75 function and activity. Significantly, these data point to an uncommon situation of a non-ubiquitinated substrate that undergoes proteasomal degradation by the 26S proteasome. While many substrates have been identified as undergoing ubiquitin-independent proteasomal degradation [82], it should be noted that the bulk of these substrates have been demonstrated to undergo degradation by the 20S core proteasomal subunit, not the 26S proteasome holoenzyme complex. For example, extensive work has conclusively demonstrated that oxidized proteins do not require ubiquitination and are degraded specifically by the 20S CP [83–87]. This is thought to occur because oxidation may cause some conformational changes in the proteins that allow them to directly enter the 20S CP, instead of requiring the ATP-dependent activity of the 19S regulatory particle that typically is given credit for unfolding proteins to be able to enter the 20S CP barrel. Only a limited number of substrates have been identified that also have ubiquitin-independent degradation through the 26S proteasome [82, 88], the best known and most conclusively studied substrate being the enzyme ornithine decarboxylase [89–91].</p><!><p>Because Cx32 is linked to the causation of the X-linked Charcot-Marie Tooth disease (CMTX), a human peripheral neuropathy [92–94], there has been much interest in elucidating the regulation of Cx32, including its ubiquitination, as it may provide insight into the basis of CMTX. As for Cx43, Cx32 degradation is also mediated by both lysosomal and proteasomal degradation, although the bulk of the studies have focused on the degradation of Cx32 by the proteasome. One of the mutants that has been identified in CMTX patients is the Cx32 E208K point mutant [95], which exhibits an intracellular trafficking defect such that it remains in the ER. Use of this mutant has been particularly useful in studying Cx32 ubiquitination and ERAD. The Cx32 E208K mutant does not oligomerize into connexons and is not transported to the plasma membrane to form functional gap junction channels [35]. ER stress, induced by DTT treatment, triggers the dislocation of wild-type Cx32 from the ER into the cytoplasm where it undergoes proteasomal degradation, as was seen with Cx43 [29, 36]. The ER-localized Cx32 E208K mutant was observed in both non-ubiquitinated and polyubiquitinated states [29]. There was significantly more polyubiquitinated Cx32 E208K protein versus non-ubiquitinated in the cytoplasm than in the ER membrane, which could only be observed upon inhibition of the proteasome. The extent of polyubiquitination was decreased in response to cellular stresses such as a heat shock or oxidative stress. After cellular stress, there was a decrease in Cx32 polyubiquitination, particularly in the cytoplasm [29]. This indicated that, unlike Cx43, Cx32, which is dislocated from the ER into the cytoplasm for ERAD, is polyubiquitinated, as is the case for almost all proteins that undergo 26S proteasomal degradation.</p><!><p>The regulation of connexins and their function is known to be crucial for normal cell homeostasis. An increasing body of work argues for the importance of connexins in human diseases as well. As previously mentioned, Cx32 has been linked to the human CMTX disease. Connexin defects have also been discovered in other diseases including oculodentaldigital dysplasia (ODDD), deafness, cataracts, and skin disorders (reviewed in [28, 93]). A role for connexins in cancer has also been pursued (reviewed in [96]). A recent study analyzed the mechanisms by which proteasomal inhibition effects cell death in cancer cells [97], which is the reason for the use of the proteasomal inhibitor, bortezomib (commercially marketed as Velcade), as a cancer therapeutic in multiple myeloma and more. Using the proteasomal inhibitor, MG132, the study demonstrated that a long-term block of proteasomal degradation triggered cell apoptosis after induction of ER stress with tunicamycin or thapsigargin. Interestingly, overexpression of Cx43 can increase the apoptotic response to MG132, as well as sensitizing cells to apoptosis brought on by ER stress. While inhibition of proteasomal degradation has consistently been proven to increase GJIC, this study reported that the increased apoptotic sensitivity due to Cx43 overexpression did not appear to be dependent on the ability of Cx43 to create functional gap junctions [97]. Other studies have also implicated a GJIC-independent function for connexins [98, 99]. Thus, the regulation of connexins, either dependent or independent of their communication abilities, may have important consequences for a vast array of cellular and tissue processes and pathways.</p><!><p>It is evident from a growing body of work that connexins have critical functions not only in normal cell and tissue function and homeostasis, but also in human diseases under aberrant conditions. Altered gene and protein expression, and protein trafficking/localization have been observed in a number of disease states. These observations clearly demonstrate the necessity for a better understanding of the regulation of the connexin life cycle. One component that has commanded attention is the mechanisms that regulate connexin turnover. With such a short half-life of 1.5 to 5 hours, the cell must tightly regulate the processes that mediate connexin degradation. Lysosomal, and perhaps, autophagic actions appear to have an important role in regulating the turnover of ubiquitinated connexins that are internalized from gap junctions at the cell membrane. Proteasomal degradation, mediated by ERAD, acts in the turnover of both ubiquitinated and non-ubiquitinated connexins from the ER and may function as part of the cellular stress response which determines cell survival. With new tools and innovative techniques available to examine the role of ubiquitination in mediating connexin degradation, the processes that contribute to the regulation of connexin protein levels and GJIC will be elucidated, which will hopefully provide better insight into the mechanisms underlying disease pathologies.</p><!><p>Connexin43 ubiquitin modification affects intracellular trafficking and degradation</p><p>Monoubiquitination affects connexin43 internalization and endosomal trafficking</p><p>Ubiquitination is involved in connexin32 proteasomal degradation</p><p>Ubiquitination is not required for CIP75-mediated connexin43 proteasomal degradation</p><p>This is a PDF file of an unedited manuscript that has been accepted for publication. As a service to our customers we are providing this early version of the manuscript. The manuscript will undergo copyediting, typesetting, and review of the resulting proof before it is published in its final citable form. Please note that during the production process errors may be discovered which could affect the content, and all legal disclaimers that apply to the journal pertain.</p><p>Abbreviations used: GJIC, gap junctional intercellular communication; ER, endoplasmic reticulum; Cx43, connexin43; EGF, epidermal growth factor; TPA, 12-O-tetradecanoylphorbol 13-acetate; BFA, brefeldin A; UBA, ubiquitin-associated domain; LC3, light chain 3; ERAD, ER-associated degradation; Cx32, connexin32; CMTX, X-linked Charcot-Marie Tooth disease; ESCRT, endosomal sorting complex require for transport; Hrs, hepatocyte growth-factor regulated tyrosine kinase substrate; Tsg101, tumor susceptibility gene 101; Eps15, epidermal growth factor substrate 15.</p>

PubMed Author Manuscript

Near-infrared luminescent CaTiO3:Nd3+ nanofibers with tunable and trackable drug release kinetics\xe2\x80\xa0

750\xe2\x80\x93850 nm (NIR I) and 1000\xe2\x80\x931400 nm (NIR II) in the near infrared (NIR) spectra are two windows of optical transparency for biological tissues with the latter capable of penetrating tissue deeper. Monitoring drug release from the drug carrier is still a daunting challenge in the field of nanomedicine. To overcome such a challenge, we propose to use porous Nd3+-doped CaTiO3 nanofibers, which can be excited by NIR I to emit NIR II light, to carry drugs to test the concept of monitoring drug release from the nanofibers by detecting the NIR II emission intensity. Towards this end, we first used electrospinning to prepare porous Nd3+-doped CaTiO3 nanofibers by adding micelle-forming surfactant Pluronic F127, followed by annealing to remove the organic component. After a model drug, ibuprofen, was loaded into the porous nanofibers, the drug release from the nanofibers into the phosphate buffered saline (PBS) solution was monitored by detecting the NIR II emission from the nanofibers. We found that the release of the drug molecules from the nanofibers into the PBS solution triggers the quenching of NIR II emission by the hydroxyl groups in the surrounding media. Consequently, more drug release corresponded to more reduction in the intensity of the NIR II emission, allowing us to monitor the drug release by simply detecting the intensity of NIR II from the nanofibers. In addition, we demonstrated that tuning the amount of micelle-forming surfactant Pluronic F127 enabled us to tune the porosity of the nanofibers and thus the drug release kinetics. This study suggests that Nd3+ doped CaTiO3 nanostructures can serve as a promising drug delivery platform with the potential to monitor drug release kinetics by detecting the tissue-penetrating NIR emission.

near-infrared_luminescent_catio3:nd3+_nanofibers_with_tunable_and_trackable_drug_release_kinetics\xe

1. Introduction<!>2.1 Structure, morphology and the formation mechanism of CaTiO3:1%Nd3+ nanofibers<!>2.2 Cytotoxicity<!>2.3 Drug loading<!>2.4 Drug release kinetics and photoluminescence (PL) intensities<!>3. Conclusions<!>4.1 Reagents and materials<!>4.2 Synthesis of luminescence functionalized porous CaTiO3:Nd3+ nanofibers<!>4.3 In vitro cytotoxicity assay<!>4.4 Loading and release of drug<!>4.5 Characterization

<p>Over the past few decades, sustained drug delivery systems (SDDSs) have received increasing attention for biomedical and pharmaceutical applications in healthcare engineering. Conventional dosage administrations have shown considerable gastrointestinal or renal side effects.1 In contrast, sustained drug delivery systems can deliver the therapeutic drugs to the targeted cells or tissues at the expected concentration over a desirable period of time,2 thus improving the drug effectiveness and reducing the toxicity. Polymer-based pharmaceutical carrier systems have been utilized as such delivery systems for improving drug loading efficiency,3 but some of the synthetic polymers are non-biodegradable and potentially toxic.4 Recently, biocompatible porous inorganic materials, such as bioactive glass,5 silica,6 hydroxyapatite,7 calcium phosphate8 and calcium silicate,9 have been extensively investigated as efficient drug delivery systems owing to their sustained drug-release profiles, nontoxic nature and capability to deliver trace elements (e.g. Ca2+, Fe3+) essential for human health.</p><p>Calcium titanate (CaTiO3, CTO) is an established bioceramic, which has been widely used as a coating material for orthopedic implant.10 It has been reported that coating of titanium (Ti) and its alloys with hydroxyapatite (HA) results in the formation of a CaTiO3 interface layer.11 Such an interface layer was found to increase the bonding strength between the Ti-based substrate and HA, and also suppress the progression of HA dissolution in an acidic environment induced by osteoclastic bone resorption in the body.10c,12 While CaTiO3 is routinely used for bone implant devices, it has potential to serve as a drug carrier for building SDDSs. Apart from nontoxicity and biocompatibility, an ideal drug-delivery system is expected to transport the desired drug molecules to the targeted tissues and release them in a controlled manner. To achieve this, not only the drug release efficiency and the selection of disease therapy are of particular interest, but the monitoring and tracking of drug releasing kinetics in an external and in situ manner during the drug-release process is also equally critical.</p><p>Fluorescence materials have been established for bioimaging. The organic dyes and semiconductor quantum dots (QDs) have been synthesized and investigated extensively over the past few decades to monitor biological species and processes in living systems.13 Drug-carrying vehicles functionalized with fluorescence have been used to monitor and assess the drug release and disease therapy efficiency. In order to minimize high absorption and scattering as well as autofluorescence of light that occurs in biological tissues, near-infrared (NIR) light has played an eye-catching role for such a purpose in the biomedical field.14 The NIR light has shown the ability to penetrate ~10 cm of the breast tissue, ~4 cm of skull/brain tissue and deep muscle tissue under a microwatt spectral excitation. The NIR range of 750–850 nm, also called as the first near-infrared window (NIR I), has been considered as the 'window of optical transparency' for biological tissues,15 which allows for deep excitation light penetration, reduced photo damage effects, low autofluorescence and light scattering through tissue. Many organic dyes, including commonly used cyanine dyes, lie within this NIR I region and have been extensively investigated for drug delivery applications.16 Unfortunately, the rapid photobleaching rate of organic dyes limits the available detection time, and the short fluorescent lifetimes as well as broad emission are not beneficial for reducing the background interference to increase the signal to noise ratio.17 These shortcomings have been successfully overcome by semiconductor QDs, which possess high quantum yields, good photostability and narrow emission.14,18 However, NIR-emitting QDs can only be efficiently excited using visible or UV light due to the low intensity of the first excitonic absorption band.19 In addition, there have been wide concerns on the inherent toxicity and chemical instability of QDs.20</p><p>Recently, rare earth (RE) ion (typically trivalent) doped upconversion (UC) fluorescent nanomaterials, which can convert longer wavelength radiation (980 nm) to a shorter wavelength fluorescence (e.g. visible light) via a two-photon or multiphoton mechanism, have been proposed as an alternative candidate to quantum dots and dyes.21 The UC nanomaterials hold a range of advantages including narrow emission peaks, large Stokes shifts, good chemical stability, and low toxicity.22 However, a recent study has shown a temperature rise of 10 °C for samples (5 mg UCNPs per mL of deionized water) under the 980 nm excitation for 10 min at a power density of 35 W cm−2, and observed the death of all cells (HEK 293T) at an excitation power density of 400 mW cm−2 under 980 nm excitation.23 In addition, the UC fluorescence (e.g. Er/Yb doped UC nanomaterials) lies in the visible region (red and green),21d,22 where soft tissues show strong absorption or scattering. In contrast, Nd3+ ions are considered as good candidates to achieve high down-conversion (DC) quantum efficiency and can also lower the thermal effects associated with the UC process under excitation with the 980 nm spectrum.23 Besides, under excitation at 800 nm, Nd3+ ions induce a NIR DC emission within 850–1200 nm, which lies in the second near-infrared window (NIR II at 1000–1400 nm).24 Compared to NIR I, the NIR II spectral window is much more appealing due to its deeper tissue penetration, minimal autofluorescence and negligible light scattering, which facilitates the monitoring of excited emission.24,25 The Nd3+ ion DC emission may be a facile and effective approach to tracking the drug delivery phenomenon in deep natural tissue. Furthermore, Nd atoms can be feasibly doped and 'caged' within the lattice of CaTiO3 to avoid possible lanthanide element releasing due to its perovskite crystal structure and the dimensional matching of Nd and Ca atoms.26 Meanwhile, CaTiO3 is a transparent material with a high refractive index in visible and NIR wavelengths,27 which makes it an efficient host material for Nd3+ doping. Therefore, Nd3+ doped CaTiO3 nanomaterials are considered as a class of 'smart' carrier candidates for drug delivery applications.</p><p>In this study, a variety of Nd3+ doped CaTiO3 nanofibers with controlled structural characteristics were successfully synthesized via electrospinning. Ibuprofen (IBU), an anti-inflammatory drug, was used initially as a test model. The drug loading quantity and releasing kinetics were systematically studied. More importantly, the NIR II emission of the nanofibers was found to be directly related to the amount of drug released from the nanofibers. The main mechanism for such a phenomenon was proposed. A promising local drug delivery platform with monitorable releasing kinetics via an external optical field has therefore been suggested to the community.</p><!><p>Fig. 1a shows the typical photoluminescence (PL) emission spectra of CaTiO3:Nd3+ nanofibers excited at 800 nm. Pure CaTiO3 nanofibers are not luminescent, as expected. The luminescence of CaTiO3 nanofibers was observed when doped with different concentrations of Nd3+ ions. The characteristic peaks at 1072 nm correspond to 4F3/2–4I11/2 transitions of Nd3+, and the luminescence was quenched when the doping concentration was below or exceeds 1%. The maximum intensity at ~1072 nm was achieved on the nanofibers with the Nd3+ doping concentration of 1%, and thus this doping concentration was selected for further investigations.</p><p>The thermogravimetry-differential scanning calorimetry (TG-DSC) curves of the as-spun nanofibers (Fig. S1, ESI†) show ~14.82%, ~47.8% and ~10.88% weight losses with the increased temperature to ~600 °C, which are attributed to the evaporation of residual solvent, decomposition of nitrates and the degradation of PVP and surfactant Pluronic F127, and carbon oxidation released by the decomposition of tetrabutyl titanate and PVP, respectively.28 No further significant weight loss is observed beyond 650 °C, indicating that the perovskite CaTiO3 phase is formed. Therefore, the ultimate sintering temperature is set at 700 °C in the following experiments. Further, the crystallinity of 1%Nd3+ doped CaTiO3 nanofibers annealed with this temperature was determined by the X-ray diffraction (XRD). As shown in Fig. 1b, well-defined diffraction peaks at (110), (111), (112), (210), (103), (022), (220), (204), (224) and (110) match with those of the orthorhombic CaTiO3. When compared with orthorhombic CaTiO3 standard (JCPDS 82-0228), no impurity or secondary phases are present. The investigation using energy dispersive spectroscopy (EDS) shows the coexistence of the Ca, Ti, Nd and O elements with homogeneous distribution within CaTiO3:1%Nd3+ nanofibers (ESI,† Fig. S2 and Table S1). The results above suggest that Nd3+ ions were efficiently doped into the CaTiO3 crystal lattice.</p><p>The morphology of the CaTiO3:1%Nd3+ nanofibers prepared with various F127/CTO molar ratios (denoted as χ) was examined (Fig. 2). The as-spun CaTiO3:1%Nd3+ precursor fibers range from ~150 to ~180 nm in diameter with a smooth surface morphology (ESI,† Fig. S3). After annealing at 700 °C for 2 h, the nanofibers present a rough surface with a decreased diameter of ~80 to ~150 nm due to the decomposition of organic additives, inorganic salts and the formation of a perovskite CaTiO3 phase. The porous structure of nanofibers after thermal treatment at 700 °C is mainly attributed to the removal of surfactant Pluronic F127 self-assembly micelles and PVP. The internal microstructural features of nanofibers were examined further using transmission electron microscopy (TEM), where the highly porous nature of all fibers can be clearly distinguished due to the degree of electron penetrability on different parts of the samples. The high-magnification TEM images confirm the porous microstructural characteristics, and show that the pore size tends to increase from ~5 to ~20 nm with an increased χ value (the insets of Fig. 2d–f). All nanofibers present a large scale of neatly arranged lattice fringes with no obvious defects, indicating high crystallinity of CaTiO3 nanofibers. The spacing d-values are 0.382 nm, 0.270 nm and 0.342 nm, corresponding to the (110), (112) and (111) crystal facets of the orthorhombic CaTiO3 phase, respectively (ESI,† Fig. S4). These results are in agreement with the findings from XRD.</p><p>The N2 adsorption/desorption analysis was used to determine the effect of χ ratio on the porosity of CaTiO3:1%Nd3+ nanofibers prepared, as illustrated in Fig. 3. All samples show representative IV-type isotherms, a typical sign for porous materials. The textural parameters of the corresponding samples are summarized in Table 1. CaTiO3:1%Nd3+ nanofibers with a χ ratio of 5 m have a BET surface area of ~16 m2 g−1, a pore volume of ~0.06 cm3 g−1 and a narrow pore size distribution (insets of Fig. 3). The decrease of the χ ratio induces a significant increase of BET surface area and pore volume. The nanofibers synthesized without surfactant Pluronic F127 (χ ratio = 0) are found to have ~34 m2 g−1 surface area, ~0.14 cm3 g−1 pore volume and a relative broad pore size distribution (insets of Fig. 3). Both BET specific surface area and the BJH desorption cumulative pore volume have doubled and the pore dimension tends to be lager in comparison to the sample prepared with the highest χ ratio. The variation of surface area, pore volume and pore size is induced by the micelle aggregation of surfactant Pluronic F127, which was reported in previous studies.6a,29</p><p>The formation process of porous structures in CaTiO3:1%Nd3+ nanofibers is shown schematically in Fig. 4. Initially, PVP was added into the precursor of calcium titanate to prepare a suitable sol for a stable electrospinning process. When the surfactant Pluronic F127 was used, the F127 polymer chain self-assembled into micelles with hydrophobic groups as a core and hydrophilic groups as a shell in the fluid. When a large number of micelles aggregate, irregular round shaped pores formed and existed in the as-spun nanofibers. As χ increased, the micelles aggregate further, which promotes the formation of larger pores. When the F127/CTO molar ratio reaches 5m, the size of pores with irregular circular shape increases significantly. Therefore, CaTiO3:1%Nd3+ nano-fibers with rather controllable porous structures were obtained by tuning the F127 concentrations after the annealing process.</p><!><p>Cytocompatibility of a material is a crucial factor for its potential drug delivery applications. Hence, the in vitro cytotoxicity of CaTiO3:1%Nd3+ nanofibers synthesized with different F127/CTO molar ratios and cultured with bone marrow-derived mesenchymal stem cells (BMSCs), and 3-(4,5-dimethylthiazol-2-yl)-2,5-diphenyltetrazolium bromide (MTT) assay was used for the cytotoxicity assessment. As shown in Fig. 5, the viability of BMCSs with different concentrations of CaTiO3:1%Nd3+ nanofibers varying from 100 to 4000 μg mL−1 for 1, 3 and 5 days was investigated. It shows that the viability of BMSCs is not suppressed from the lowest concentration of 100 μg mL−1 to the highest of 4000 μg mL−1 for all time points (up to 5 days), indicating that the CaTiO3:1%Nd3+ nanofibers are biocompatible and non-cytotoxic to live cells, which allows further investigation of drug loading and releasing behaviors.</p><!><p>The Fourier transform infrared (FTIR) spectra of CaTiO3:1%Nd3+ nanofibers, pure IBU drug and IBU loaded CaTiO3:1%Nd3+ nano-fibers are shown in Fig. 6. When CaTiO3:1%Nd3+ nanofibers were loaded with IBU drugs, typical characteristic peaks such as Ti–O bands at ~445 and ~556 cm−1, C–C bands at ~1328, ~1460 and ~1512 cm−1, COOH groups at ~1720 cm−1 and C–Hx bonds at ~2862 and ~2953 cm−1 were present. The characteristic peaks of curve b well matched the curve combination of pure CaTiO3:1%Nd3+ nanofibers (curve a) and IBU drugs (curve c), confirming that IBU molecules were successfully loaded on the CaTiO3:1%Nd3+ nanofibers. In addition, the thermal gravity (TG) analysis was pursued to investigate the drug loading capacity of different nanofibers. The most dramatic weight loss of IBU occurs at 220 °C (ESI,† Fig. S5). As shown in Fig. 7, no distinguishable weight loss is observed during the heating process of pure CaTiO3:1%Nd3+ nanofibers. In contrast, all IBU loaded CaTiO3:1%Nd3+ nanofibers displayed substantial weight loss at ~220 °C, as expected. When the F127/CTO molar ratio is increased from 0 to 5 m, the weight loss of IBU loaded CaTiO3:1%Nd3+ nanofibers varies from ~50 wt% to ~30 wt%. Such a weight loss is mainly attributed to the decomposition of IBU molecules loaded on the nanofibers. Hence, the loading capacity of CaTiO3:1%Nd3+ nanofibers with F127/CTO molar ratios of 0, 2.5 and 5m is ~50 wt%, ~40 wt% and ~30 wt%, respectively. The decreased loading capacity is attributed to the enlarged pore dimensions, the decreased surface area and pore volume of the nanofibers with increased surfactant Pluronic F127 contents.</p><!><p>During the drug releasing process in phosphate buffered saline (PBS) aqueous solution, accompanied by fluid diffusion into the pores of nanofibers, IBU molecules are liberated and diffuse into the fluid by a diffusion-controlled mechanism.2b,5a,30 The cumulative drug release profiles of all three IBU loaded CaTiO3:1%Nd3+ systems in PBS are shown in Fig. 8a. IBU loaded CaTiO3:1%Nd3+ nanofibers with different χ exhibit dramatically different releasing phenomena. When the χ ratio is 5m, ~65% of total IBU loaded is released within the initial 5 h, and ~90% is released in ~20 h. With the decreased χ ratio, the drug releasing behavior shows a more sustained manner. The IBU loaded CaTiO3:1%Nd3+ nanofibers, synthesized without surfactant Pluronic F127, present the slowest drug releasing behavior. Only ~25% of the IBU drug is liberated from the nanofibers within the initial 5 h, and as low as ~50% of the total drug load is released after 45 h. Therefore, it is confirmed that IBU loaded CaTiO3:1%Nd3+ nanofibers synthesized without surfactant Pluronic F127 present the highest drug loading capacity and the most sustained release kinetics. This may be attributed to the strong interaction between IBU molecules and the fiber microstructures, such as the reduced pore dimensions and the increased surface area. Therefore, the drug loading and releasing behaviors have been successfully manipulated via the synthesis control of CaTiO3:1%Nd3+ nanofibers.</p><p>More importantly, the photoluminescence phenomenon of CaTiO3:1%Nd3+ nanofibers was found to respond effectively in tune with the drug releasing process. The PL intensity at ~1072 nm is dramatically decreased during the drug releasing for all types of nanofibers (ESI,† Fig. S6). The relationship between the relative PL intensity and the PBS immersion time is summarized in Fig. 8b. CaTiO3:1%Nd3+ nanofibers with fast drug release behavior quench the PL most rapidly. ~25% of PL intensity is quenched within the initial 5 h, and ~50% is reduced after 10 h drug releasing. With the reduced χ ratio, the PL quenching effect is weakened. In contrast, the nanofibers with most sustained drug releasing kinetics (χ ratio = 0) present a rather delayed PL intensity quenching effect; only ~10% of PL intensity is quenched within the initial 5 h, and as low as ~20% relative PL intensity is quenched after ~45 h immersion in PBS. The PL quenching effect is directly corresponding to the drug releasing process. In addition, one known fact is that the physiological environment is much more complex in comparison to the PBS solution. To verify the findings above, simulated body fluid (SBF), a well-known method for the bioactivity assessment of biomaterials,31 has been used to investigate the drug release behavior and PL variation in a simulated physiological environment. As shown in Fig. S7a (ESI†), the IBU release behavior in SBF has shown a quite similar phenomenon to that in PBS. More importantly, the relative PL emission intensity has been quenched in a similar manner to that for the samples immersed in PBS (Fig. S7b and c, ESI†). The relative PL intensity variation has thus been confirmed to effectively reflect the drug releasing phenomenon.</p><p>It is well-known that the PL emission of rare earth ions can be hindered by its environments that have a high phonon frequency.32 An FTIR study was carried out on the IBU loaded CaTiO3:1%Nd3+ nanofibers with a χ value of 5m to reveal the PL quenching mechanism associated with the drug release. As shown in Fig. 9, during the drug releasing process, the peaks at ~2862 cm−1 and ~2953 cm−1, assigned to C–Hx bonds of IBU molecules, gradually decrease with the drug liberation over time, as expected. In contrast, the intensity of the peak at ~3430 cm−1, which belongs to hydroxyl groups, increases significantly. It has been reported that the hydroxyl group with a high frequency vibration at 3430 cm−1 is an effective quencher of fluorescence at 1072 nm for rare earth Nd3+.32,33 Therefore, the IBU molecules play a shielding role in 'protecting' CaTiO3:1%Nd3+ nanofibers from the hydroxyl groups in the PBS aqueous solution during drug releasing. Owing to the liberation of IBU molecules, the fluid invades the fiber surface and diffuses into the pores. The quenching effect of hydroxyl groups is enhanced significantly. Therefore, the sustained drug release kinetics induces a postponed PL quenching phenomenon. Such a matching correspondence between the PL emission intensity within the NIR spectral range and the drug release represents an effective approach to tracking and monitoring the drug release from CaTiO3:Nd3+ nanofibers.</p><!><p>In this study, a series of fine CaTiO3:1%Nd3+ nanofibers, with the excitation in the first near-infrared window and emission spectra in the second near-infrared window, were synthesized for the first time via electrospinning. The in vitro study revealed that such photoluminescent nanofibers are cyto-compatible. The micro-structural characteristics of CaTiO3:1%Nd3+ nanofibers can be successfully tuned via the control of surfactant Pluronic F127 concentration during the synthesis. Due to the high surface area, large pore volume and low pore dimensions, the nanofibers synthesized without surfactant Pluronic F127 present a highest drug loading capacity and most sustained release kinetics. More importantly, the relative intensity of PL emission with the NIR II range from the nanofibers responds effectively to the released quantity of IBU molecules. Furthermore, the PL quenching rate has shown a good correspondence with the drug release kinetics due to the quenching effect of hydroxyl groups, confirming that the drug release kinetics can be feasibly identified and monitored via NIR spectra. Therefore, we anticipate that this study may provide another promising drug delivery platform with NIR-monitored releasing kinetics for biomedical applications.</p><!><p>All of the chemical reagents used as received without further purification, including acetic acid (C2H4O2, A.R.), ethanol (C2H6O, A.R.), N,N-dimethylformamide (C3H7NO, A.R.), cyclohexane (C6H12, A.R.) and phosphate buffered saline (PBS, pH = 7.4) (Sinopharm Chemical Reagent Co., Ltd), calcium nitrate tetrahydrate (Ca(NO3)2·4H2O, 99%), neodymium nitrate hexahydrate (Nd(NO3)3·6H2O), >99.9%, tetrabutyl titanate (Ti(OC4H9)4), >98.0% and polyvinylpyrrolidone (PVP, MW = 1 300 000) (Aladdin), Pluronic F127 (EO106PO70EO106, MW = 12 600, Sigma-Aldrich), and ibuprofen (IBU, 99 wt%, Nanjing Chemical Regent Co., Ltd).</p><!><p>In a typical process, Ti(OC4H9)4 (~0.82 g) and Ca(NO3)2·4H2O (~0.56 g) were dissolved in a mixture of acetic acid (2 mL) and ethanol (7 mL) to prepare solution A under constant stirring. Solution B was prepared by dissolving Nd(NO3)3· 6H2 O (~0.21 g) in 20 mL N,N-dimethylformamide. Subsequently, a series of mixture sols containing solution A and B were prepared with different Nd3+ ion concentrations (0, 0.5, 1, 2, and 3 mol%), and electrospun to synthesize CaTiO3:Nd3+ nanofibers with different doping concentrations. The sol, from which the nanofibers with highest photoluminescence properties were achieved, was added with surfactant Pluronic F127 with various F127/CTOmolar ratios (denoted as χ) of 0, 2.5 and 5m (m = 10−3). Finally, a N,N-dimethylformamide solution of polyvinyl pyrrolidone (PVP, 3.25 wt%) was mixed with the above solution to prepare spinnable precursor sols. The electrospinning sol was fed into the conducting nozzle (2 mm ID) using an infusion pump (KDS-100, KD Scientific, USA) at a constant flow rate of 0.5 mL h−1. The distance and voltage applied between the needle tip and the collector were set to be 15 cm and 10 kV (PS/FC30P04.0-22, Glassman High voltage Inc., USA), respectively. As-spun fibers were dried at 80 °C for 12 h and calcined in a furnace at 700 °C, with a heating rate of 2 °C min−1 for2h in air.</p><!><p>The in vitro cytotoxicity of Nd doped CaTiO3 was determined using bone marrow-derived mesenchymal stem cells (BMSCs) via 3-(4,5-dimethylthiazol-2-yl)-2,5-diphenyltetrazolium bromide (MTT) assay. Briefly, BMSCs were seeded in 96-well plates at a density of 6000–8000 cells per well and cultured in 5% CO2 at 37 °C for 24 h to allow the cells to attach to the bottom of the plates. The CaTiO3:1%Nd3+ nanofibers were sterilized using alcohol for 48 h, and then serial dilutions of the nanofibers at concentrations of 0, 100, 200, 500, 1000 and 4000 μg mL−1 were incubated with cells for 1, 3 and 5 days in 5% CO2 at 37 °C. After each incubation time point, MTT (20 μL, 10 mg mL−1) solution was added to each well and incubated for another 4 h at 37 °C. Subsequently, the media containing the CaTiO3:1%Nd3+nanofibers were removed, Dimethyl sulfoxide (DMSO, 150 μL) was added into each well, and the absorbance was measured at a wavelength of 490 nm using a microplate reader (Thermo ScientificInc.,USA). Tests were performed in six replicates.</p><!><p>Ibuprofen (IBU, an anti-inflammatory drug) was selected as a model drug in the study.30a Porous CaTiO3:1%Nd3+nanofibers (0.2 g) were added into IBU hexane solution (40 mg mL−1, 50 mL) at ambient temperature. The mixture was subsequently stirred for 24 h to induce the drug diffusion into the pores. The IBU loaded CaTiO3:1%Nd3+ sample was collected by centrifugation and washed with hexane to remove the adsorbed IBU on the outer surface. After the samples were dried at 60 °C for 12 h, the IBU release behavior was investigated in vitro. IBU loaded CaTiO3:1%Nd3+ nanofibers (~0.15 g) were equally divided into 12 portions, and immersed in PBS (10 mL) under gentle stirring at 37 °C. At each time interval, a 1 mL media sample was collected and diluted to measure the amount of IBU drug released by measuring the optical density at 222 nm using a UV/Vis spectrophotometer. Meanwhile, the nanofibers were filtered and air-dried for the examination of PL emission at each release time point (0 h, 4 h, 8 h, 12 h and 48 h). The nanofibers were pressed into a quartz plate with a cylindrical groove to ensure uniform thickness and distribution of the samples before PL measurement.</p><!><p>An X-ray diffractometer (X'PertPRO MPD, Netherlands) with Cu Kα radiation was used to investigate the crystal structure of nanofibers prepared. The morphological and microstructural characteristics were examined using field emission scanning electron microscopy (FESEM, Hitachi SU-70, Japan) and transmission electron microscopy (TEM, Philips TecnaiF20 S-TWIN, Netherlands), respectively. The specific surface area and pore size distribution were determined via N2 adsorption/desorption analysis at liquid nitrogen temperature (77 K) using a coulter OMNISORP-100 apparatus. The FTIR spectra were recorded on a Perkin-Elmer580B infrared spectrophotometer on KBr pellets (Tensor 27, Bruker, Germany). Thermogravimetry-differential scanning calorimeter measurement (TG-DSC, DSCQ1000, AT, USA) was used to measure the quantity of IBU drugs loaded onto the CaTiO3:1%Nd3+ nanofibers. UV/Vis adsorption spectra were measured on a TU-1810 spectrophotometer. The photoluminescence (PL) measurements were performed on a FLS920 spectrophotometer.</p>

PubMed Author Manuscript

The interplay between microRNAs and histone deacetylases in neurological diseases

Neurological conditions, such as Alzheimer\xe2\x80\x99s disease and stroke, represent a prevalent group of devastating illnesses with few treatments. Each of these diseases or conditions is in part characterized by the dysregulation of many genes, including those that code for microRNAs (miRNAs) and histone deacetylases (HDACs). Recently, a complex relationship has been uncovered linking miRNAs and HDACs and their ability to regulate one another. This provides a new avenue for potential therapeutics as the ability to reinstate a careful balance between miRNA and HDACs has lead to improved outcomes in a number of in vitro and in vivo models of neurological conditions. In this review, we will discuss recent findings on the interplay between miRNAs and HDACs and its implications for pathogenesis and treatment of neurological conditions, including amyotrophic lateral sclerosis, Alzheimer\xe2\x80\x99s disease, Huntington\xe2\x80\x99s disease and stroke.

the_interplay_between_micrornas_and_histone_deacetylases_in_neurological_diseases

1. Introduction<!>1.1. MicroRNAs<!>1.2. Histone acetylation and histone deacetylases (HDACs)<!>1.3. miRNAs and HDACs<!>2. Amyotrophic lateral sclerosis<!>3. Huntington\xe2\x80\x99s disease<!>4. Alzheimer\xe2\x80\x99s disease<!>5. Stroke<!>6. Conclusions<!>

<p>Neurological conditions including stroke, Alzheimer's disease and Parkinson's disease, are characterized by neuronal cell loss and represent a devastating, prevalent group of diseases, which exact tremendous personal and financial cost on our society. In each condition a diverse set of pathways is engaged leading to neuronal demise. Despite significant efforts, the primary causes for neurodegeneration remain largely unknown likely due to the complexity of the problem. In recent years, miRNAs and histone acetylation have separately garnered significant attention as they are dramatically affected in neurodegenerative diseases. An intriguing relationship has emerged between miRNAs and histone deacetylases (HDACs), which modulate acetylation of histones and other proteins. Indeed, each regulates the other in a variety of contexts in health and disease. This review will focus on the interaction between miRNAs and HDACs and how they impact neurodegenerative diseases. Table 1 provides a summary of the miRNA and HDAC interactions discussed within this review with respect to the each neurological condition.</p><!><p>miRNAs are short (~22 nucleotides) non-coding RNAs, which regulate post-transcriptional gene expression by blocking messenger RNA (mRNA), although they have been shown also to bind directly to RNAs. A number of miRNAs are up or down-regulated in neurological conditions. Of note, some miRNAs have been identified as having protective mechanisms to promote cell viability, while others contribute positively to disease pathogenesis. As nearly half of all known miRNAs have been found in the brain, they have a significant impact on gene expression in the central nervous system (Tardito et al., 2013). The functional roles of many non-coding RNAs, like microRNAs (miRNAs), have only recently been elucidated, which has vastly expanded our understanding of gene regulation. There are currently over 2500 known human miRNAs (from miRBase.org); and each, in turn, have the potential to regulate hundreds of mRNA transcripts (Krek et al., 2005; Lewis et al., 2003; Lim et al., 2005). Thus, it has been estimated that 50% of mammalian protein-coding genes are regulated by miRNAs (Tardito et al., 2013), making them a major factor in gene regulation. As a result, understanding their mechanisms of action and how they are regulated in the context of diseases of the nervous system is of vital importance.</p><p>miRNAs are produced through an elaborate but well-documented mechanism, shown in Fig. 1. In the nucleus RNA polymerase II or III is responsible for transcribing primary miRNA (pri-miRNA), which consists of at least one hairpin loop. The pri-miRNA is then excised to form pre-miRNA by the endoribonuclease, Drosha, with the help of DGCR8 (also known as Pasha), a double stranded RNA-binding protein. The ~70 nucleotide pre-miRNA is then exported from the nucleus to the cytosol by Exportin-5 in conjunction with Ran-GTP. In the cytosol, the pre-miRNA undergoes a second endoribonucleic cleavage by Dicer. The resulting double-stranded miRNA combines with an Argonaute (Ago) protein, which determines the complementary strand used to the target mRNA, and the remaining strand is degraded (reviewed in Goodall et al. (2013) and Yates et al. (2013)). While the canonical biogenesis, described above, is the most common method, it should be noted that a non-canonical biogenesis pathway also occurs, which bypasses the Drosha/Dicer processing (Babiarz et al., 2008; Saraiya and Wang, 2008).</p><p>The single stranded, mature miRNA is then combined into a multi-protein unit, known as miRNA-induced silence complex (miRISC), where it becomes fully functional. The miRNA will then bind to the 3′ untranslated region (UTR) of the target mRNA to prevent further translation (Bartel, 2009). Interestingly, much of the miRNA sequence does not match that of the target mRNA 3′UTR sequence, but a 2–6 nucleotide "seeding" region appears to be critical for recognition and binding of the mRNA (Lewis et al., 2005, 2003). However, there is a direct correlation between the sequence complementarity and the level of gene silencing. A perfect match between miRNA and mRNA will likely lead to degradation of the mRNA; whereas, less homologous sequences will result in transcriptional repression (Goodall et al., 2013).</p><!><p>DNA is condensed into chromatin in order to compress an enormous amount of genetic material into a relatively small nucleus. The nucleosome, the basic building block of chromatin, consists of 147 base pairs of DNA that are wrapped around an octamer of histone proteins. The histones can be heavily modified with small molecule post-translational modifications, such as methyl or acetyl groups, which influence the activity and transcription of nearby genes. The pattern of these small molecule post-translational modifications on histones, which govern their interactions with DNA and local propensity for transcription, is commonly referred to as the "histone code."</p><p>Acetylated histones generally are associated with increased transcription. When a histone is acetylated it is thought that the electronegativity of the acetyl group repels the already negatively charged DNA backbone, causing a loosening of the nucleosome and providing space for transcription factors to bind, as shown in Fig. 2. A group of proteins called histone acetyltransferases (HATs), also known as lysine acetyltranserases (KATs), are responsible for the addition of acetyl groups on the lysine residues of the histone tails. This process is undone by HDACs, which remove the acetyl group resulting in transcriptional repression (reviewed in Sleiman et al. (2009)).</p><p>HDACs are divided into four distinct classes based on their structure and cofactors. Classes I (HDACs 1, 2, 3 and 8), II (HDACs 4, 5, 6, 7, 9 and 10) and IV (HDAC 11) are all zinc dependent enzymes, while Class III HDACs, also known as Sirtuins (SIRT 1–7), require NAD+. Available evidence suggests that maintaining histone acetylation in a more active transcriptional state is beneficial to cells, especially during learning, disease or stress situations that require significant plasticity. Accordingly, a tremendous amount of ongoing research investigates the use of HDAC inhibitors to promote the activation of cyto-protective genes and cell survival.</p><p>In a number of neurodegenerative diseases, where investigators have looked, histones are notably hypoacetylated. Thus HDAC inhibitors have been used to restore proper histone acetylation and to promote the transcription of neuroprotective and/or neurorestorative genes. Indeed, HDAC inhibitors have been used to ameliorate neurodegeneration. Many HDAC inhibitors initially employed a wide range of HDACs, known as pan inhibitors. However, with the advent of more specific HDAC inhibitors, it is anticipated that we will be able to determine precisely which HDACs should be inhibited to achieve maximum neuroprotection with the fewest side effects.</p><p>It also should be noted that while most HDACs primarily act on histone proteins, HDACs are not specific to histone proteins but are also known to deacetylate a number of non-histone proteins. The most common example of this is HDAC6, which resides largely in the cytoplasm where it deacetylates tubulin. However, this review will focus primarily on their histone targets.</p><!><p>miRNAs and HDACs have a complex relationship that is not yet fully understood but could be critically important. miRNAs are capable of regulating HDACs and influencing histone acetylation, while HDACs themselves can regulate miRNA expression. Thus a careful balance between the two is important to maintain appropriate levels of each in the cell. In the case of neurological conditions, the expression of miRNAs and histone acetylation is dramatically affected, which can distort the balance between the two and may contribute to disease pathogenesis. Interestingly, HDAC inhibitors can alter the expression profiles of miRNAs in cancer and in the brain (Scott et al., 2006; Zhou et al., 2009), and could become an added benefit of HDAC inhibitors to treat diseases. The ability to control miRNAs and histone acetylation will likely become an important component in managing and treating neurodegenerative diseases. Thus, understanding how they influence each other and affect biological pathways is of significant interest.</p><!><p>Amyotrophic lateral sclerosis (ALS) is a progressive neurodegenerative disease that affects upper and lower motor neurons, resulting in muscle atrophy and paralysis. ALS is typically fatal within 3–5 years of the onset of symptoms. The overwhelming majority of ALS cases are sporadic; but about 10 percent of ALS cases are linked to genetic causes, such as a mutation in the gene that codes for copper-zinc superoxide dismutase (SOD1) (Valentine et al., 2005). There is currently no cure for the disease and riluzole, the only approved drug for ALS in the United States, is only moderately effective. Early in the disease motor neuron death results in deinnervated neuromuscular junctions (NMJ), which are often reinnervated by the surviving motor neurons (Schaefer et al., 2005; Wohlfart, 1957). Eventually, as more motor neurons die, this compensatory mechanism is unable to sustain the NMJ, resulting in paralysis. This process may explain why most ALS patients remain asymptomatic until the majority of motor neurons have died (Williams et al., 2009).</p><p>Many miRNAs are dysregulated in ALS; but one that has received significant attention is miR-206, a skeletal muscle-specific miRNA and associated with the NMJ (Bruneteau et al., 2013; Rao et al., 2006; Williams et al., 2009). Williams et al. showed that miR-206 is significantly increased in a mouse model of ALS expressing an SOD1 mutation (G93A) and is upregulated after acute nerve injury in wild-type mice. While miR-206 knockout mice develop normally, their ability to recover from sciatic nerve injury is greatly diminished. However, in the ALS mouse model the knockout of miR-206 accelerated the atrophy of skeletal muscles, but it did not impact the age of onset or survival (Williams et al., 2009). Therefore, they demonstrated that miR-206 plays a clear role in promoting the regeneration of the neuro-muscular synapse after acute injury and in ALS.</p><p>miR-206 was computationally predicted to target HDAC4, and HDAC4 has been strongly implicated in controlling neuromuscular gene expression (Cohen et al., 2007; Tang et al., 2009). Consistent with these studies, Williams et al. found that miR-206 knockout mice had increased HDAC4 protein levels but not HDAC4 mRNA, suggesting the miR-206 regulates HDAC4 via transcriptional repression rather than mRNA destabilization. In contrast to miR-206 knockout mice, mice lacking HDAC4 showed faster muscle reinnervation after injury. Taken together, these results demonstrate the opposing roles of HDAC4 and miR-206 to inhibit and promote reinnervation, respectively in ALS. Furthermore, the same group identified fibroblast growth factor binding protein 1 (FGFBP1) as a common downstream target of both miR-206 and HDAC4. FGFBP1 was downregulated in miR-206 knockout mice but upregulated in HDAC4 knockout mice. As FGFBP1 is known to interact with several fibroblast growth factors that participate in reinnervation, the knockdown of FGFBP1 resembled the miR-206 knockout mice after denervation. The identification of a common downstream target of miRNAs and HDACs, such as FGFBP1, is critical to determining how miRNAs and HDACs influence neurodegenerative diseases and will provide significant insight into disease pathogenesis.</p><p>More recently, Brunetau et al. expanded on the elegant work of Williams and colleagues by testing many of these findings in human muscle specimens from ALS patients (Bruneteau et al., 2013). While no change was observed in HDAC4 mRNA between healthy muscles and those from ALS patients, there was an increase in HDAC4 transcript levels in ALS patients with more rapid disease progression than long-term ALS survivors (>5 years survival). This demonstrated a positive correlation between HDAC4 transcript levels and disease progression. They also found that MIR206 and FGFBP1 were significantly upregulated in ALS patients, but no correlation was observed between disease progression or the degree of muscle reinnervation. Thus they were able to validate in human ALS patients that miR-206 is upregulated, likely in an attempt to promote reinnervation, while the inhibitory HDAC4 was downregulated. The relationship between miR-206 and HDAC4 is likely a component of the motor neuron's coping mechanism to handle the death of other motor neurons. However, this process is ultimately unsuccessful in maintaining the NMJ in advanced stages of ALS.</p><p>Given the negative effect of HDAC4 on muscle reinnervation, it has been suggested that HDAC inhibitors could be good drug candidates in slowing the progression of ALS. Trichostatin A, a Class I and II HDAC inhibitor, was recently tested in an ALS mouse model with promising results. Not only was the disease progression delayed, but it also increased the survival of ALS mice. More specifically the number of fully innervated NMJ was increased and muscle atrophy was reduced (Yoo and Ko, 2011). While these results are very promising, the development of HDAC4 specific inhibitors may further improve the outcome of these studies and reduce the numerous side effects associated with non-selective or pan-HDAC inhibitors. Alternatively, one could use infusions of miR-206 as a strategy to reduce HDAC4 and enhance muscle reinnervation.</p><!><p>Huntington's disease (HD) is a fatal neurodegenerative disorder that results in neuronal death specifically in the cerebral cortex and the medium spiny neurons of the striatum. The symptoms of HD include jerky involuntary movements (chorea), dementia and emotional dysfunction. HD is caused by a CAG repeat expansion in the first exon of the gene that codes for the huntingtin protein (Htt) and results in an unusually long and toxic polyglutamine repeat in Htt (MacDonald et al., 1993). Although the cause of HD is attributable to pathological polyglutamine expansions, the mechanism of pathogenesis remains unclear.</p><p>Recently, several dysregulated miRNAs have been linked to the pathogenesis of HD. For example,miR-22 is a potentially neuroprotective miRNA that is reduced in HD (Jovicic et al., 2013). It was shown that the overexpression of miR-22 can inhibit neurodegeneration in an in vitro model of HD. miR-22 was computationally predicted to target HDAC4, REST corepressor1 (Rcor1) and G-protein signaling 2 (Rgs2); and, indeed, luciferase-binding assays demonstrated that miR-22 could bind to the mRNA 3′UTR for each of the three predicted targets. miR-22 also targeted MAPK14 and Trp53inp1, which are cell death regulators. Based on these data, it is thought that the anti-apoptotic result of miR-22 overexpression is a combinatorial effect of its downstream targets. As histones are hypoacetylated in HD, blocking HDAC4 translation by increasing miR-22 could promote neuronal survival. It is important to note that the use of HDAC inhibitors, such as SAHA, can ameliorate the symptoms of HD in animal models, making them promising therapeutics in HD (Ferrante et al., 2003; Hockly et al., 2003; Mielcarek et al., 2011; Steffan et al., 2001; Thomas et al., 2008).</p><p>Other dysregulated miRNAs in HD are targeted by REST, a master regulator that largely represses neuronal genes in neuronal and non-neuronal cells. REST is known to recruit other co-repressors, notably HDAC1 and HDAC2 to assist in gene silencing. In HD REST appears to play a very complex role between Htt and miRNA transcription. Normal Htt is known to associate with the predominately cytoplasmic REST and, in part, prevents it from entering the nucleus and silencing genes. However, mutant Htt is unable to associate with REST; and, as a result, REST translocates into the nucleus and represses the expression of a number of genes, such as BDNF, other neurotrophic factors, and a number of miRNAs. One such miRNA is miR-124, which is down regulated in HD (Conaco et al., 2006) and subsequently its downstream targets are upregulated (Johnson et al., 2008). miR-124 is a brain specific miRNA that is capable of decreasing non-neuronal transcripts in neurons (Conaco et al., 2006). Based on these results, REST becomes localized to the nucleus in HD, where it silences the production of miR-124, resulting in an increase of non-neuronal gene transcription in neurons. This leads to reduced neuronal-like behavior in neurons, and may subsequently contribute to their degeneration (Lim et al., 2005).</p><p>In addition to being regulated by REST, miRNAs, miR-9 and miR-9* can target REST and Co-REST, respectively. Both miR-9 and miR-9* are downregulated in the HD brain so this negative feedback loop becomes dysfunctional and may exacerbate the cascade of genes inappropriately silenced by REST (Packer et al., 2008).</p><!><p>Alzheimer's disease (AD) is the most common age-related neurodegenerative disease, affecting ~2% of the population in industrialized nations. The clinical symptoms of AD include dementia and diminished cognitive functioning, and it is pathologically characterized by the formation of amyloid-beta plaques and neurofibrillary tangles. As people live longer, the population suffering from AD is expected to grow dramatically in the next few decades (Abbott, 2011). Without effective therapies to treat AD the social and economic burden of the disease will be immense. As with other neurodegenerative diseases, a number of miRNAs are differentially regulated in AD patients compared with healthy controls.</p><p>In AD some of the most affected miRNAs target SIRT1, a class III HDAC, which is known to have a positive effect on learning, memory and longevity (Gao et al., 2010; Howitz et al., 2003; Michán et al., 2010; Wood et al., 2004). SIRT1 can regulate miR-134 through a repressor complex with the transcription factor YY1 (Gao et al., 2010). Loss of SIRT1 leads to an increase in miR-134 and causes a decrease in CREB and BDNF, resulting in impaired synaptic plasticity (Gao et al., 2010). This demonstrates that enhancing SIRT1 may be of therapeutic benefit to neurodegenerative diseases like AD. In a recent study, reservatrol, a natural compound found in grapes and peanuts, was used to enhance SIRT1 activity. In 8–9 month-old mice reservatrol improved long-term memory and improved long-term potentiation in hippocampus CA1 slices (Zhao et al., 2013). The enhancements observed with reservatrol treatment were absent in SIRT1 mutant mice, directly linking the effects of reservatrol to SIRT1 activity. Reservatrol treatment also resulted decreased expression of miR-134 and miR-124. The reduced expression of these miRNAs was then correlated with increased CREB and subsequently increased BDNF (Zhao et al., 2013). Therefore, SIRT1 plays a powerful role in suppressing miR-134 allowing an increase in CREB and BDNF, which could be beneficial in AD.</p><p>SIRT1 is also associated with miR-34c, which is upregulated in the hippocampus of AD patients and in the hippocampus of the APPPS1-21 mouse model of AD, as well as in 24 month-old mice. In contrast to miR-134, miR-34c appears to negatively regulate memory consolidation but is also known to target SIRT1 (Yamakuchi et al., 2008; Zovoilis et al., 2011). Consistent with the increase in miR-34c, a decrease in SIRT1 in the hippocampus was observed in 24 month-old mice and the APPPS1-21 mouse model. Mice injected with a miR-34c mimic showed significant memory impairment and decreased SIRT1. Similarly, when APPPS1-21 mice were treated with a miR-34c seed inhibitor, their memory function was rescued to a level similar to that of age-matched controls and was correlated with an increase in SIRT1 (Zovoilis et al., 2011). These data suggest that the dysregulation of miR-34c and the reduced SIRT1 in the hippocampus may contribute to the age-related cognitive decline seen in AD and other dementias.</p><p>In AD there is a surprising overlap between the down-regulated miRNAs associated with SIRT1 in AD patients and the presence of amyloidgenic amyloid-beta (Hébert et al., 2008; Schonrock et al., 2010; Wang et al., 2008). Amyloidgenic amyloid-beta is generated from the sequential cleavage of the amyloid precursor protein by beta and gamma secretase, which is thought to predominate in AD patients. However, the preferred non-amyloidgenic pathway utilizes alpha secretase, rather than beta, to produce a slightly shorter amyloid-beta peptide that is not prone to aggregation. miR-9 and miR-181c are both down-regulated in AD patients and in AD models. Both miRNAs were computationally predicted to target the 3′UTR of SIRT1. Indeed luciferase assays showed significant repression of SIRT1 with the addition of miR-9 or miR-181c. Interestingly, when both miR-9 and miR-181c were combined there was an additive effect on the luciferase repression (Schonrock and Götz, 2012). The activation of SIRT1 has previously been shown to delay aging and regulate the aggregation and removal of amyloid-beta (Cohen et al., 2004; Donmez et al., 2010; Qin et al., 2006). In addition, the overexpression SIRT1 is capable of preventing amyloidgenic amyloid-beta by promoting the use of alpha-secretase and the non-amyloidgenic pathway (Qin et al., 2006). Thus, it has been suggested that the repression of miR-9 and miR-181c may result in a neuroprotective increase in SIRT1 to promote non-amyloidgenic amyloid-beta and aid in the clearance of amyloid-beta (Schonrock and Götz, 2012).</p><!><p>Stroke is a major cause of death and disability, affecting approximately 795,000 people in the US annually. In most cases, a stroke is caused by a loss of blood flow in the brain (ischemic stroke), but it can also be caused by a ruptured blood vessel in the brain (hemorrhagic stroke). Treatment options for either form of stroke remain limited, so there is significant motivation to develop new therapeutic options for stroke treatment. HDAC inhibitors have become potential candidates for the treatment of stroke as they can reverse the hypoacetylation seen after a stroke and can activate the transcription of neuroprotective genes (Chuang et al., 2009; Langley et al., 2009).</p><p>A number of miRNAs are differentially regulated after middle cerebral artery occlusion (MCAO), an in vivo model of ischemic stroke (Hunsberger et al., 2012; Jeyaseelan et al., 2008). Knowing that HDAC inhibitors can improve the behavioral outcomes in animal models of stroke, and that miRNAs expression can be influenced by HDAC inhibition, Hunsberger et al. measured miRNA profiles after MCAO with and without valproic acid, a Class I HDAC inhibitor. They found that valproic acid treatment reduced the neurological and motor deficits in rats after MCAO and significantly affected the expression of miR-885-3p and miR-331. miR-885-3p was upregulated after MCAO in untreated animals, but this upregulation was attenuated by valproic acid treatment. While relatively little is known about miR-885-3p in the brain, miR-885-3p is upregulated after cisplatin treatment in cancer cells and appears to play a role in cell viability and apoptosis and/or autophagy (Huang et al., 2011). Thus, decreasing miR-885-3p through HDAC inhibition could aid in reducing the cell death after an ischemic stroke and may contribute to the improved behavioral outcomes. In contrast to miR-885-3p, miR-331 was only upregulated after MCAO with valproic acid treatment. It was also upregulated in primary cortical neurons exposed to valproic acid, both preand post-treatment with oxygen-glucose deprivation, an in vitro stroke model. This suggests that miR-331 is directly upregulated by the valproic acid. Based on Ingenuity Pathway Analysis, it is predicted that miR-331 plays a role in cellular movement, cell death, and the organization of the nervous system during development (Hunsberger et al., 2012). In this case, valproic acid had controlled the expression of miRNAs in a neuroprotective manner.</p><!><p>In this review we have presented key studies that demonstrate the complex interplay between miRNAs and HDACs in the context of neurodegenerative disease. Their ability to regulate each other provides an intriguing insight into their biological functions in health and disease specific to neurodegenerative processes. However, this field is still in its infancy, and there are a number of challenges that need to be overcome in order to solidify our understanding of how the two components regulate and interact with each other. Perhaps the most important of these is to experimentally confirm many of the computationally determined roles of individual miRNAs. The computational predictions provide excellent insights for experimentalists; and, given the vast number of miRNAs, it is not practical to delve into the mechanistic effects of each miRNA. However, as specific miRNAs are identified as important to a particular disease or biological mechanism, it is imperative to determine the downstream effects of each. Likewise, it is important to further probe the connection between HDACs and miRNAs. One possibility is by using HDAC inhibitors more frequently as a tool to understand how particular classes of HDACs or even individual HDACs affect miRNA expression. This could significantly expand our knowledge of how both miRNAs and HDACs function. While this review focused solely on the effects of HDACs, miRNA expression can influence other chromatin modifications such as DNA and histone methylation. Understanding the functional roles of miRNAs, the effects of HDACs, how they affect each other and how to manipulate their expression could pave the way for new treatment options for neurodegenerative diseases.</p><!><p>The canonical miRNA biogenesis pathway. Pri-miRNA is transcribed by RNA polymerase II and excised by Drosha and DGCR8 to from pre-miRNA. Exportin 5 and Ran-GTP then export the pre-miRNA from the nucleus. In the cytoplasm, pre-miRNA is cleaved by Dicer, giving rise to double stranded miRNA, which is then incorporated into RISC and one strand of the miRNA is degraded. Subsquently, the RISC complex binds to target mRNA that is either repressed or degraded.</p><p>Proposed mechanism of reinnervation. After an injury at the neuromuscular junction (NMJ) there is an increase in MyoD, which activates the expression of miR-206. miR-206 binds to the HDAC4 3′UTR, causing a decrease in HDAC4. This in turn leads to an increase in FGFBP1, a secreted growth factor that interacts with FGF and promotes reinnervation. As miR-206 is reduced in ALS, this process is disrupted and prevents reinnervation. Adapted from Williams et al., 2009.</p><p>Summary of known miRNA and HDAC interactions in neurological conditions.</p>

PubMed Author Manuscript

Determination of Cyclopropane Fatty Acids in Food of Animal Origin by 1H NMR

Cyclopropane fatty acids (CPFAs) are unusual fatty acids of microbial origin, recently detected in milk and dairy products. CPFAs have been demonstrated to be interesting molecular markers for authentication of dairy products obtained without ensiled feeds. Moreover, they can also be recognized as a new secondary component of human diet. Information is lacking on the presence of cyclic fatty acids in other food sources. Cyclopropane fatty acids have been detected by GC-MS analysis in cheese and other animal fats in concentration ranging from 200 to 1000 mg/kg fat, but in some cases, the complex fatty acid profile and the possible presence of interfering peaks make the separation not straightforward and the quantification uneasy. Therefore, a new reliable 1H NMR method was developed to detect and measure CPFA content in different foods of animal origin, based on the detection of the characteristic signals of cyclopropane ring. The 1H NMR (600 MHz) method showed detection limits comparable with those of full scan GC-MS, and it allowed the identification and quantitation of the cyclopropane fatty acids in different foods.

determination_of_cyclopropane_fatty_acids_in_food_of_animal_origin_by_1h_nmr

1. Introduction<!>2.1. Materials<!>2.2. Fat Extraction<!>2.3.1. Synthesis of Internal Standard Trimethylsilyl Decanol (TMSD)<!>2.3.2. Preparation of CPFA and TMSD Standard Solutions<!>2.3.3. Preparation of Spiked Samples<!>2.3.4. 1H NMR Acquisition<!>2.3.5. Quantitative Analysis<!>2.3.6. Linearity and Limit of Detection and Quantification<!>2.3.7. Accuracy, Precision, and Recovery of the Method<!>2.4. Gas Chromatographic Analysis<!>3.1. CPFA Signal Detection<!>3.2. Quality Parameters of 1H NMR Analysis<!>3.3.1. Comparison of 1H NMR-Based Method and GC-MS Method<!>4. Conclusion<!>Conflicts of Interest<!>

<p>Cyclopropane fatty acids are unusual fatty acids found in microorganisms, both Gram-negative and Gram-positive, and seed oils of some tropical plants and protozoa [1, 2]. The bacterial production of cyclopropane ring is related to changes in the membrane fatty acids composition and represents one of the most important adaptive microbial responses that favours the stress tolerance of several bacteria, such as Lactobacillus helveticus, L. bulgaricus, L. acidophilus, and L. sanfranciscensis [1].</p><p>In plants, CPFAs are usually minor components, where cyclopropene fatty acids are the most abundant. They are present in Malvaceae, Sterculiaceae, and Sapindaceae, representing a significant component of Litchi chinensis and Sterculia foetida seed oils, principally sterculic acid [2].</p><p>Recently, we identified by GC-MS the presence of CPFAs (dihydrosterculic and lactobacillic acids, Figure 1) in milk and dairy products [3, 4] and more recently in meat (unpublished results).</p><p>Due to the undoubtable importance of these foodstuffs in human diet, it appears clear that a deep investigation on the dietary intake of these fatty acids and their effects on humans gains importance.</p><p>CPFAs have been also recently identified in human serum and adipose tissue [5], suggesting that they are absorbed as the other fatty acids and can exert physiological effects. Moreover, CPFAs are minor fatty acids but their presence in milk fat is in the hundred-ppm order [4], so their dietary intake may be not negligible.</p><p>CPFAs (mainly dihydrosterculic acid) also play an important role in food authentication: in fact, they were discovered in milk and dairy products from cows fed with silages, and their determination has been demonstrated to be a powerful tool for the authentication of Protected Denomination of Origin (PDO) cheeses, such as Parmigiano Reggiano, where the use of silages in cow feeding is forbidden [6]. In this context, "Consorzio del Formaggio Parmigiano Reggiano" has proposed a modification on the Production Specification Rules, including the determination of CPFAs among the official controls (UNI 11650).</p><p>Therefore, CPFAs represent an almost completely new field of research in food lipids and it is important to develop different methods of detection and quantification, in view of an expected growing body of research, both in food characterization/authentication and in food safety aspects.</p><p>Gas chromatography methods currently dominate the literature for the determination of main and secondary fatty acids in foods [7–9], and we previously applied this technique for the qualitative and quantitative determination of CPFAs in milk and dairy products [3, 4, 6]. However, gas chromatography analysis requires time-consuming sample derivatization with the risk of interfering by-products and use of large amount of solvents [10]. Moreover, in the particular case of fat from animal origin, the extreme complexity of fatty acid profile makes the separation and quantification of minor fatty acids a challenging issue. For example, more than 400 different fatty acids were detected in milk [11]. In the case of cyclopropane fatty acids, we obtained its separation in cheese fat by using apolar capillary column [6]; however, this column is not suitable for the optimal separation of fatty acids, so it is possible that changing the food matrix interferences occur. It is also possible that other cyclopropane fatty acids were present but undetectable because they were overlapped by the most abundant fatty acid signals. So, it is important to have an alternative method to confirm the cyclopropane ring presence and possibly to correctly quantify CPFAs. Moreover, the development of a rapid method that provides the necessary analytical information with minimal sample preparation would be advantageous. NMR spectroscopy is one such analytical tool that avoids sample derivatization and offers the benefit of short data acquisition times.</p><p>Nuclear magnetic resonance spectroscopy has started to represent an interesting tool to analyse biofluids and food and beverages, and in the case of lipids, it represents a reliable and fast alternative to traditional methods such as gas chromatography. This was due to the advantages of this technique as the simplicity of the sample preparation (usually it only requires the fat dissolution in deuterated chloroform) and measurement procedures, the instrumental stability, the increase of sensitivity, and modern pulse sequences, with simultaneous suppression of big signals [12].</p><p>For these reasons, the use of NMR spectroscopy has established a significant role in the analysis of lipids [13]. Several studies consider the analysis by 1H NMR of triacylglycerol composition as a useful tool for both triglyceride quantitation and sample classification [14]. Minor fatty acids were also object of investigation by NMR, especially conjugated linoleic acids (CLAs) [15].</p><p>NMR could represent an ideal method to detect CPFAs due to the characteristic signals of the protons of the cyclopropane unit between −0.30 and −0.35 ppm [16], which permit their detection in a zone of 1H NMR spectrum practically free from other signals. This highly shielded position of cyclopropane resonance is conventionally explained by the anisotropy of the C–C bond, just opposite to CH2 group in a three-membered ring, or by an aromatic-like ring current involving the six electrons in the three C–C bonds (σ aromaticity) that shields cyclopropane protons [17].</p><p>Therefore, with the aim to investigate on the presence of CPFAs in foods, we developed a new fast and reliable quantitative 1H NMR method, to be used as alternative to gas chromatographic methods and to confirm the presence of CPFAs in foods.</p><!><p>Methanol, n-hexane, dichloromethane, trimethylchlorosilane, hexamethyldisilazane, 1-decanol, sodium sulphate anhydrous, sodium carbonate, deuterated chloroform, and tetracosane were from Sigma-Aldrich (Saint Louis, MO, USA), and hydrochloric acid and potassium hydroxide pellets were from Carlo Erba (Milan, Italy). Dihydrosterculic acid methyl ester was from Abcam (Cambridge).</p><p>All the solvents, standards, and reagents were of analytical grade.</p><p>Cheese, meat samples from several species animals, cured meat, and commercial fish were analysed for the content of cyclopropane fatty acids. Most of them were purchased from the market (Parma, Italy). Samples of cheese and meat produced without ensiled feeds were kindly provided from Parmigiano Reggiano Cheese Consortium and Prof. Riccardo Bozzi of the University of Florence, respectively.</p><!><p>Lipid extraction following the Folch method [18] was performed. 10 g of sample was homogenized with 75 mL of dichloromethane : methanol (2 : 1, v/v). The mixture was centrifuged (10 min, 3000 rpm) and filtered. This procedure was repeated three times. The three filtrates were transferred to a graduate cylinder, and a volume of about 50 mL KCl 0.88% in distilled water was added. The mixture was shaken vigorously. The final biphasic system was decanted, and the upper aqueous phase was eliminated. The lower organic phase was filtered through anhydrous sodium sulphate and collected. Lipid content was then recovered after solvent was evaporated with a rotary evaporator under vacuum.</p><!><p>0.2 mL of 1-decanol, 0.3 mL of trimethylchlorosilane, and 0.6 mL of hexamethyldisyiazane were mixed in a screw cap septum vial. Mixture reacted for 1 h at 60°C, neutralized with sodium carbonate, and then dried with anhydrous sodium sulphate. Reaction mixture was diluted with 1 mL of hexane, filtered, taken to dryness in a rotary evaporator, and the residue weighed. Purity of trimethylsilyl decanol (TMSD) was confirmed by 1H NMR and by GC-MS analysis in the conditions reported in Section 2.4.</p><!><p>Appropriate amounts of trimethylsilyl decanol (TMSD, internal standard) and CPFAs were weighed and added separately to CDCl3 (10 mL) to yield two final stock solutions of about 500 mg/L each.</p><p>Adequate amounts of CPFA and TMSD stock solutions were transferred in 5 mm NMR tubes and taken to the final volume of 1 mL with CDCl3 to obtain working solutions at 100, 50, 25, and 5 µg/mL of CPFAs, all containing 10 µg/mL of TMSD.</p><!><p>100 mg of meat fat (chicken) and cheese fat (Parmigiano Reggiano) both negative to CPFAs were spiked with the appropriate amount of CPFA and TMSD solutions and taken to the volume of 1 mL of CDCl3 to obtain the same final concentrations reported above for standard solutions.</p><!><p>100 mg of fat was dissolved in 1 mL of CDCl3 containing 0.01 mg of TMSD as internal standard. 1H NMR spectra were recorded on a Varian INOVA-600 MHz spectrometer (Varian, Palo Alto, CA, USA), equipped with a 5 mm triple resonance inverse probe. Data were collected at 298 K, with 32 K complex points, using a 90° pulse length. 1024 scans were acquired with an acquisition time of 1.707 s and a recycle delay of 2 s. Presaturation of the fatty acids –CH2– signal (1.25 ppm) was performed in order to assure a correct digitization of small signals as CPFAs. The NMR spectra were processed by MestReC software 6.0.2 (Santiago de Compostela, Spain, EU): spectra were Fourier transformed with FT size of 64k and 1 Hz line-broadening factor, manually phased and carefully baseline corrected, and referenced to the chloroform signal (7.26 ppm). Baseline correction was further manually optimized in the zone of interest (from −1 ppm to 0.7 ppm).</p><!><p>CPFA concentrations were obtained by integrating the peak area of the 1H NMR signal at −0.35 ppm and the methyl signal of the trimethylsilyl group of the internal standard (TMSD) at 0.1 ppm.</p><p>The CPFA integral was converted in mass value (mg) according to the following formula, as previously reported [19]:(1)ACPFA×EWCPFAmg  CPFA=ATMSD×EWTMSDmg  TMSD,where ACPFA = spectral area of CPFA, ATMSD = spectral area of internal standard, EWCPFA = equivalent weight of the analyte, EWTMSD = equivalent weight of internal standard, and EW = (molecular weight/number of hydrogens in the signal).</p><p>Absolute amount of CPFAs obtained was finally expressed as mg/kg of fat.</p><!><p>The limit of detection (LOD) and the limit of quantification (LOQ) were calculated utilizing the S/N ratio methods, based on the determination of the peak-to-peak noise [20]. LOD and LOQ were, therefore, calculated as the concentrations of CPFAs producing a recognizable peak with a signal-to-noise ratio of, respectively, 3.3 and 10. LOD and LOQ were determined both in pure standard solution and in a sample of meat fat negative to CPFAs spiked with different concentrations of CPFAs.</p><!><p>The accuracy of the CPFA recovery was determined by assaying samples with known concentrations of CPFAs, both as pure compounds and as spiked matrix. The precision was expressed as coefficient of variation (CV%). Recovery of analytes was determined by spiking sample of fat free from CPFAs with pure dihydrosterculic acid.</p><!><p>GC-MS quantitative analysis was performed as previously reported [6]. Briefly, 200 mg of fat was dissolved in hexane (5 mL) and mixed for 1 min with 0.2 mL of KOH 10% (Carlo Erba, Milan, Italy) in methanol. After phase separation, the superior organic phase was added to internal standard (tetracosane) and injected (1 µL, split mode) on an Agilent Technologies 6890N gas chromatograph (Agilent Technologies, Palo Alto, CA, USA) coupled to an Agilent Technologies 5973 mass spectrometer (Agilent Technologies, Palo Alto, CA, USA). A low-polarity capillary column (SLB-5ms, Supelco, Bellafonte, USA) was used. The chromatogram was recorded in the scan mode (40–500 m/z) with a programmed temperature from 60°C to 280°C.</p><!><p>Figure 2 depicts the characteristic upfield zone of 1H NMR spectrum of dihydrosterculic acid and TMCD. 1H NMR shows two individual peaks at −0.35 and 0.60 ppm for the methylene protons of the cyclopropane ring. Assignments were previously made by Knothe [16] with the aid of 2D correlations: the upfield signal is assigned to the cis-proton and the downfield signal to the trans-proton. The two methine protons of the cyclopropane ring are located at 0.68 ppm. The other CPFA proton signals are located at lower fields; for example, the four protons in alpha position with respect to the cyclopropane ring display a distinct shift at 1.17 ppm and the signal of the other two protons is observed at 1.40 ppm, within the broad methylene peak [16]. Among all these specific signals, the cis-methylene proton of the cyclopropane ring can be easily assigned and used for quantification because it does not overlap with any other signal of fatty acids that could be observed in a complex food lipid 1H NMR spectrum.</p><p>Because the integral of a given peak in a 1H NMR spectrum is directly proportional to a corresponding number of resonant nuclei, peak areas of CPFA cis-methylene proton can be compared with the peak area of TMSD trimethylsilyl group close to 0.1 ppm and used for the determination of total CPFA content. This internal standard was specifically synthetized starting from a medium chain linear alcohol as decanol because it is not volatile, soluble in apolar solvents, and its trimethylsilyl group gives a singlet close to the CPFA selected signal. Tetramethylsilane (TMS) was commercially available but it cannot be used as quantitative internal standard due to its volatility. The synthesis of TMSD was quick and simple with a high yield (about 80%). The purity was determined by GC-MS and 1H NMR (data not shown).</p><p>1H NMR spectra of food fats are characterized by a dominating fatty acid methylene group peak at 1.2 ppm that is many orders of magnitude greater than those of the component of interest. This causes a number of problems: first of all, it prevents a correct digitization of small signals, hampering their observation and quantification, but the tail of this large signal can also determine a distortion of the baseline in the zone of CPFA chemical shifts. Therefore, a suppression of this signal was performed during spectra acquisition.</p><!><p>The quantitative 1H NMR method was developed with the aim to determine CPFA concentration in a broad range of food fats, in particular in fats of animal origin (dairy products, meat, and fish). As a first step, the method was subjected to validation in terms of precision, accuracy, linearity, detection, and quantitation limits, following recommendations of the International Conference on Harmonization (ICH 2005: Validation of analytical procedures: text and methodology. Harmonized tripartite guideline, Q2, R1). The validation tests were performed on pure solutions of CPFA (dihydrosterculic acid) and on cheese and meat matrices naturally free from CPFAs, spiked with dihydrosterculic acid as reported in Experimental.</p><p>To determinate accuracy and precision, solutions containing weighed amount of CPFAs were analysed by 1H NMR in the experimental conditions previously reported. Measured results for the standard solutions were in agreement with the amounts weighed in the range of concentrations of 0.005–0.1 mg/mL of CPFA, a range that corresponds with the final intube concentration of the analytes in real samples of fat. The linearity was demonstrated in the same range (Figure 3). The limit of detection and the limit of quantification were calculated utilizing the S/N ratio method described above. LOD and LOQ were calculated in pure standard solutions. The instrumental quantification limit for CPFA standard (LOQ, signal to noise ratio higher than 10) in the experimental conditions reported was about 0.01 mg/mL while the limit of detection (LOD) was obtained at 0.0025 mg/mL (S/N ratio 4). The 1H NMR (600 MHz) method developed showed detection limits in pure standard solutions comparable with those generally achieved by full scan GC-MS. Coefficients of variation (CV%) for three replicate measurements of each standard concentration were lower than 3%, indicating a good precision of the method.</p><p>Linearity, LOD, and LOQ were also calculated in two different matrices, cheese and chicken fat. The samples chosen for spiking were previously analysed by GC-MS and 1H NMR and were found negative to CPFAs. Each matrix was spiked with four different amounts of dihydrosterculic acid as reported in Experimental. Regression curves obtained are shown in Figure 4.</p><p>The linearity is maintained, as in the case of pure standard; however, in both cases, the intercept of the regression curve indicates a matrix effect, most pronounced for chicken fat. The limit of quantification for cheese was 120 mg/kg fat (S/N ratio of 10), and the limit of quantification was found to be 50 mg/kg fat (S/N ratio of 10). In the case of chicken meat, LOQ and LOD were 180 mg/kg fat and 70 mg/kg fat, respectively. Comparing these values with those obtained for pure standard solutions, results demonstrate not negligible matrix effect both in cheese and in meat fat, suggesting that an external calibration in a fat matrix is needed for an accurate quantification. On the contrary, the method can be easily applied for a rapid semiquantitative and qualitative analyses in both matrices.</p><!><p>Different samples of meat, fish, and cheese were analysed both by 1H NMR and GC-MS. GC-MS quantitative analysis was performed based on the method previously applied for cheese, as described in Caligiani et al. [6]. Figure 5 shows the enlargement of the diagnostic region for CPFAs in the 1H NMR spectra of lipids extracted from cheese, meat, and fish (containing the TMSD internal standard), confirming that the signal does not overlap with any other resonances representative of fatty acids.</p><p>Table 1 reports the list of the samples analysed for each food category, the number of samples negative or positive to CPFAs, and the comparison between GC-MS and 1H NMR results. A reference Grana Padano cheese was specifically analysed both by GC-MS and 1H NMR method to demonstrate the accuracy of the new 1H NMR method. Then, five samples of Parmigiano Reggiano and five samples of Grana Padano were tested because in the case of cheese, we had collected previously many data confirming the association between the use of ensiled feeds and the presence of CPFA [4, 6]. 1H NMR analysis confirmed the positivity at CPFA for all samples of Grana Padano (ensiled feeds allowed) and the negativity of all Parmigiano Reggiano samples (ensiled feeds forbidden), indicating that 1H NMR could be an attractive alternative technique to GC-MS to assure the authenticity of Parmigiano Reggiano and other cheeses forbidding the use of ensiled feeds in their disciplinary of productions. In the case of cheese, quantitative results suggested very good agreement between data obtained by the quantitative 1H NMR analysis and previous GC analysis method.</p><p>Concerning meat, CPFAs were detected in the GC-MS profiles of most of the commercial bovine meat samples in concentrations varying from 100 to 400 mg/kg of the total fat. CPFAs were detected by 1H NMR analysis in all commercial bovine meat samples previously resulted positive by the GC-MS analysis, with good agreement of the quantitative results. CPFAs were absent in two samples of certified meat from cows not fed with fermented forages, and this was evidenced by both techniques. The GC-MS analysis of other meat samples (pork and chicken) was negative to CPFAs both in GC-MS and 1H NMR method. In the case of pork cured meat (salami and ham), the GC-MS analysis showed the presence of a signal at the retention time of CPFAs with concentrations of 60–100 mg/kg of the total fat. However, the corresponding analysis by 1H NMR did not show the presence of cyclopropane ring, indicating the presence of an interfering peak in the GC-MS conditions adopted. This interfering peak was also resistant to oxidation as a saturated fatty acid, but it has not been identified yet and it was not easy to obtain a better separation varying chromatographic conditions. Therefore, in the case of pork, cured meat seems to be important to have the NMR confirmation of CPFA presence.</p><p>And in cured meat, the GC-MS analysis of fish samples generally showed the presence of interfering signals at the same retention time of cyclopropane fatty acids with the same corresponding mass spectrum (278 m/z), probably due to the presence of different isomers of nonadecenoic acid. Moreover, lactobacillic acid coeluted with another interfering substance with the corresponding mass spectrum of 165 m/z. This interfering peak was not resistant to oxidation and it has been suggested to be a furan fatty acid as discussed elsewhere [21, 22].</p><p>Therefore, GC-MS analysis alone was not able to confirm the presence or absence of CPFAs in fish samples, but it always required 1H NMR analysis. Moreover, observing the preliminary results showed in Table 1 on three fish samples, it seems that GC-MS underestimates the content of cyclopropane fatty acids, suggesting that besides cyclopropane fatty acids with 19-carbon atom skeleton, such as dihydrosterculic and lactobacillic acids, it is possible that other CPFAs with different chain lengths occur in fish.</p><!><p>A new quantitative 1H NMR method was developed for the determination of CPFA content in different food matrices, including dairy products, meat, and fish.</p><p>The new method reported here provides absolute quantities of CPFA (mg/kg of total fat) and shows a limit of detection comparable with those of full scan GC-MS. A complete and reliable sample analysis can be performed quickly and requires little sample preparation, reagents, and solvents. This was possible because the CPFA signal was very well defined and did not overlap with others. The role of NMR seems to be most important in meat and fish characterization because the GC-MS analysis was not able to confirm the presence of CPFAs in all the analysed samples due to the presence of interfering peaks.</p><p>Results suggested that the NMR analysis approach has potential application as a screening for quantifying cyclopropane fatty acids in meat and fish, as markers of quality and the preliminary data on few meat and fish samples presented here suggest some possible developments. For example, in the context of food authentication, cyclopropane fatty acids might be proposed, as in the case of cheese, as markers of silage feedings are able to authenticate high-quality costly meat whose producers declare the absence of silages in the feeding. This will require the construction of a robust database of meat certificated for the feeding system. This approach could also be extended to fish, to eventually distinguish farmed from wild fish.</p><p>Moreover, in the case of fish, NMR method is able to detect a higher amount of CPFAs with respect to GC-MS, indicating an important role of NMR when dietary intake of cyclopropane fatty acid has to be assessed.</p><!><p>The authors declare that there are no conflicts of interest regarding the publication of this paper.</p><!><p>Main cyclopropane fatty acids detected in dairy products.</p><p>1H NMR spectrum (600 MHz, CDCl3) of dihydrosterculic acid standard and trimethylsilyl decanol (TMSD) in the very upfield region of the spectrum. The CPFA signal at −0.34 ppm and the TMSD signal at 0.07 ppm were selected for quantification.</p><p>Regression curve for CPFA standard solution measured with the 1H NMR method.</p><p>Calibration curves of CPFA spiked matrices: (a) cheese and (b) chicken meat.</p><p>Enlargement of 1H NMR 600 MHz spectra in the zone from −0.5 to 0.8 ppm, showing the signal of the cyclopropane ring (at −0.35 ppm) used for CPFA quantification in cured ham (negative to CPFAs) and cheese and fish fat (positive to CPFAs).</p><p>Comparison of 1H NMR and GC-MS results on the presence of CPFAs in some representative samples of fat of animal origin.</p>

PubMed Open Access

Dual role of inorganic polyphosphate in cardiac myocytes: the importance of polyP chain length for energy metabolism and mPTP activation

We have previously demonstrated that inorganic polyphosphate (polyP) is a potent activator of the mitochondrial permeability transition pore (mPTP) in cardiac myocytes. PolyP depletion protected against Ca2+-induced mPTP opening, however it did not prevent and even exacerbated cell death during ischemia-reperfusion (I/R). The central goal of this study was to investigate potential molecular mechanisms underlying these dichotomous effects of polyP on mitochondrial function. We utilized a Langendorff-perfused heart model of I/R to monitor changes in polyP size and chain length at baseline, 20 min no-flow ischemia, and 15 min reperfusion. Freshly isolated cardiac myocytes and mitochondria from C57BL/6J (WT) and cyclophilin D knock-out (CypD KO) mice were used to measure polyP uptake, mPTP activity, mitochondrial membrane potential, respiration and ATP generation. We found that I/R induced a significant decrease in polyP chain length. We, therefore, tested, the ability of synthetic polyPs with different chain length to accumulate in mitochondria and induce mPTP. Both short and long chain polyPs accumulated in mitochondria in oligomycin-sensitive manner implicating potential involvement of mitochondrial ATP synthase in polyP transport. Notably, only short-chain polyP activated mPTP in WT myocytes, and this effect was prevented by mPTP inhibitor cyclosprorin A and absent in CypD KO myocytes. To the contrary, long-chain polyP suppressed mPTP activation, and enhanced ADP-linked respiration and ATP production. Our data indicate that 1) effect of polyP on cardiac function strongly depends on polymer chain length; and 2) short-chain polyPs (as increased in ischemia-reperfusion) induce mPTP and mitochondrial uncoupling, while long-chain polyPs contribute to energy generation and cell metabolism.

dual_role_of_inorganic_polyphosphate_in_cardiac_myocytes:_the_importance_of_polyp_chain_length_for_e

Introduction<!>Animal models<!>Cell Isolation<!>Permeabilized ventricular myocytes<!>Ischemia-reperfusion protocol and mitochondrial isolation from mouse hearts<!>Mitochondrial oxygen consumption<!>polyP extraction, quantification, and size determination<!>Synthetic polyP synthesis and fractionation<!>Confocal microscopy<!>Mitochondrial polyP accumulation<!>ATP Measurements<!>mPTP activity<!>Mitochondrial Membrane Potential<!>Chemical in-vitro Ischemia/Repefusion Protocol<!>Superoxide Generation during in-vitro Chemical Ischemia and Repefusion<!>Statistical Analysis<!>Ischemia-reperfusion leads to the decrease in polyP chain-lenght<!>Calcium dependent uptake of polyP by the energized mitochondria<!>PolyP stimulates ATP production in a chain-length and CypD dependent manner<!>Long-chain polyPs enhance ADP-linked respiration and respiratory reserve capacity while short-chain polyPs induce proton leak and mitochondrial uncoupling<!>Short-chain polyP activates mitochondrial permeability transition pore<!>PolyPs with short chain-length but not with long-chain length induce a dissipation of the mitochondrial membrane potential<!>Elevated superoxide generation led to mPTP opening during ischemia in mouse cardiac myocytes which was further exacerbated in reperfusion<!>Discussion<!>Conclusions

<p>Inorganic polyphosphate (polyP) is a polymer of orthophosphates linked together by phosphoanhydride bonds similar to those found in ATP (Fig. 1A) [1–5]. In eukaryotic organisms, polyP has been implicated in a wide range of physiological and pathological functions [2, 5]. In mammalian cells polyP is present in the size ranging from 5 to 800 orthophosphate residues depending on species and cell type tested [6–8]. One of the first reports of polyP presence in mammalian mitochondria dates back to over 50 years [9], however very little is still known about its roles and metabolism. The amount and turnover of polyP is much higher in tissues with high metabolic rates and demands such as brain and heart [6, 8–10]. Recent studies demonstrated that in mitochondria polyP could be involved in energy metabolism [11, 12], calcium (Ca2+) signaling [8, 13] and direct formation and/or activation of the mitochondrial Permeability Transition Pore (mPTP) [8, 14, 15]. In striking contrast to inorganic phosphate, a known inducer of mPTP [16, 17], which is typically present in the mitochondrial matrix in millimolar concentrations [2], polyP was detected in mammalian mitochondria only in micromolar concentrations [1, 6, 8], limiting its ability to buffer matrix Ca2+, Mg2+, or pH significantly and suggesting a direct regulatory role. Specifically, it has been suggested that involvement of polyP might be required for Ca2+-induced assembly of the mPTP channel-forming complex [8, 12, 15, 18, 19]. This hypothesis was built on the original discovery by Reusch and Sadoff [20] who demonstrated back in 1983 a strong correlation between bacterial transformation efficiency and the formation of poly-β-hydroxybutyrate and calcium polyphosphate (PHB/Ca2+/polyP) complexes in the plasma membranes of Escherichia coli (E. Coli) upon Ca2+-mediated induction of genetic competence. Further studies confirmed that PHB/Ca2+/polyP complexes function as voltage-activated calcium channels [21] in plasma membrane of E. Coli, and they have also been postulated to participate in co-export of Ca2+, phosphate, and deoxyribonucleic acid (DNA) transfer across the membrane during bacterial transformation [22–24]. Interestingly, a similar PHB/Ca2+/polyP complex was isolated later by Pavlov et al. [18] from rat liver mitochondria which upon reconstitution into planar lipid bilayer demonstrated the properties similar to the mPTP suggesting the role of polyP in mPTP activation. Accordingly, we previously found that enzymatic depletion of polyP from mitochondria leads to inhibition of the Ca2+-induced mPTP [8, 15]. However, polyP depletion did not prevent, and even enhanced cell death under conditions favoring reactive oxygen species (ROS)-induced mechanism of mPTP activation such as ischemia-reperfusion (I/R) [25]. Furthermore, exogenous polyP has been reported to induce death in cultured cells [26]. This suggests that involvement of polyP in mitochondrial pathology may occur through different mechanisms. The central goal of the present study was to investigate potential molecular mechanisms underlying these differential effects of polyP on mitochondrial function.</p><p>We hypothesize that the variable effects of polyP are linked to differences in polyP chain length under different physiological and pathological conditions. We, therefore, examined the size distribution of polyP in normal and pathological conditions and investigated effects of the synthetic polyPs with different chain length (short, medium, and long) on their ability to participate in energy metabolism and on Ca2+-induced mPTP. We found that polyP effects were strongly size dependent. Short chain polyP (14 phosphate residues) led to the mPTP activation, induction of proton leak, mitochondrial uncoupling and metabolic failure. Long chain polyP (130 phosphate residues) demonstrated protection against mPTP, enhanced mitochondrial coupling and improved energy generation. Medium chain polyPs (60 phosphate residues) had an intermediate effect on ATP generation, and did not affect mPTP activity. These data further support the idea that involvement of polyP in mitochondrial function should be considered in connection with specific metabolic and functional state of the organelle.</p><!><p>All protocols were in accordance with the Guide for the Care and Use of Laboratory Animals published by the National Institutes of Health (NIH Publication NO. 85–23, revised 1996) and approved by the University of California Davis Institutional Animal Care and Use Committee and by the University of Würzburg Institutional Animal Care and Use Committee. Experiments were performed in isolated hearts, cardiac myocytes or mitochondria from 10 week old wild type C57BL/6J controls (WT) and age-matched cyclophilin D (CypD) knock-out (KO) mice (mice with genetically deleted Ppif gene (Ppif−/−) [27] as we described before [28].</p><!><p>Mouse ventricular myocytes were isolated using a standard enzymatic technique using a Langendorff perfusion system [28]. Briefly, 5 min after mice were heparinized (1000 USP units) they were anesthetized by inhalation of isofluorane. When reflexes were absent, hearts were excised, placed into the Langendorff system and perfused for 5 min with a nominally Ca2+-free washing solution of composition (mM, unless specified): 140 NaCl, 4 KCl, 1 MgCl2, 5 HEPES, 10 Glucose, 1% (v/v) heparin; pH 7.4 adjusted with NaOH, followed by an enzyme solution containing 0.12 mg/mL of Liberase TM (Research Grade, Roche) in the above described solution. Upon digestion, hearts were minced and gently agitated to obtain the cells. All solutions were saturated with oxygen. Once myocytes were isolated, [Ca] was gradually raised to 1 mM using heparin-free Tyrode solution containing (in mM): 140 NaCl, 4 KCl, 1 MgCl2, 5 HEPES, 10 Glucose, 1 CaCl2, pH 7.4 adjusted with NaOH. Freshly isolated ventricular myocytes were plated on laminin-coated glass coverslips and used within 8 hours of isolation.</p><!><p>The sarcolemma was permeabilized with digitonin (10 μM for 60 s) as described previously [8, 28]. Digitonin was added to intracellular solution containing (in mM): 135 KCl, 10 NaCl, 20 HEPES, 5 pyruvate, 2 glutamate, 2 malate, 0.5 KH2PO4, 0.5 MgCl2, 15 2,3-butanedione monoxime, 5 EGTA, and 1.86 CaCl2 to yield a free [Ca2+]i of 100 nM with pH 7.2. After permeabilization, the bath solution was changed to the same intracellular solution but without digitonin. Free Ca2+ concentrations were calculated using the MaxChelator program (http://www.stanford.edu/~cpatton/maxc.html).</p><!><p>Hearts were excised from mice and placed in ice-cold Krebs-Henseleit buffer (KHB) containing in mM: 120 NaCl, 4.7 KCl, 1.2 KH2PO4, 25 NaHCO3, 1.2 MgSO4, 11 D-glucose, 1 CaC2, 1% (v/v) heparin; pH 7.4 which was oxygenized by 95% O2/5% CO2. The aorta was cannulated and hearts were perfused in a retrograded fashion on a Langendorff apparatus with Krebs-Henseleit buffer for 5 min and exposed to global no flow ischemia for 20 min followed by restoration of perfusion flow for 15 min (reperfusion). Mitochondria from 2–4 mouse hearts were isolated using the protocol of Rehncrona et al. [29] with some modifications [8, 30]. The minced heart tissue was subjected to protease treatment: it was incubated with 5 mg proteinase (bacterial; type XXIV, formerly called Nagarse; Sigma-Aldrich) dissolved in 10 ml medium A for 8 min at room temperature while gently stirring. The protease reaction was stopped by adding 1 ml of 0.2 mg/ml of bovine serum albumin dissolved in medium A containing (in mmol/L): 225 mannitol, 70 sucrose, 1 EGTA and 10 HEPES, pH 7.2. The tissue was then homogenized with a Potter-Elvehjem homogenizer, and mitochondria were isolated by differential centrifugation as described by us before [8]. The final mitochondrial pellet was suspended in isolation medium B containing (in mmol/L): 225 mannitol, 70 sucrose, and 10 HEPES, pH 7.2. Mitochondrial protein concentration was determined by protein assay (Bradford, 1976) (Pierce BCA; Thermo Fisher Scientific). All isolation procedures were performed on wet ice.</p><!><p>Mitochondrial oxygen consumption (respiration) was measured polarographically at 37°C with a Clark-type oxygen electrode (Model 782; Strathkelvin Instruments; Glasgow, UK) as previously described [28]. Freshly isolated mitochondria placed in 300 μl of MiRO5 respiration buffer containing (in mM): 110 Sucrose, 60 K-lactobionate, 20 HEPES, 3 MgCl2, 20 taurine, 10 KH2PO4, 0.5 EGTA, and 1 g/l BSA, pH = 7.1, and basal respiration rate was recorded. State 2 of respiration was initiated by the addition of 5 mmol/L glutamate and 5 mmol/L malate as substrates in the presence of 2 µM free Ca2+. Maximal respiration rate (State 3) was activated by adding 1 mmol/L ADP in the presence of 2 µM free Ca2+. Respiration rates were expressed as nmol O2 min−1 mg mitochondrial protein−1, and normalized to the no polyP treatment conditions. The respiratory control ratio (RCR) was calculated as state 3 divided by state 2 respiration rate (RCR=State3/State2) to estimate mitochondrial integrity and coupling efficiency.</p><!><p>polyP was extracted using a modified phenol/chloroform extraction protocol [6, 8]. The mitochondrial pellet (see above) was resuspended in 250 μl TELS buffer (100 mM LiCl, 10 mM EDTA, and 10 mM Tris, pH 8.0, 0.5% SDS) and mixed with 250 μl of acid phenol/chloroform, pH 4.5 (with isoamyl alcohol [IAA]; acid phenol/chloroform/IAA [125:24:1]; Invitrogen). 425–600-μm glass beads (Sigma-Aldrich) were added to the level right below the phenol fraction for extraction of polyP associated with the membrane fraction. Samples were vortexed for 5 min at 4°C, followed by centrifugation at 1,500 g for 5 min at 4°C. The water phase was transferred to a new tube and subjected to chloroform extraction with the equal volume of chloroform to remove traces of organic solvents from the water phase. polyP was precipitated from the water phase by adding 2.5 volumes of ethanol, followed by overnight incubation at 20°C. The water-ethanol mixture was centrifuged for 10 min at 10,000 g. The resulting pellet containing polyP was resuspended in 50 μl of a buffer (0.1% SDS, 1 mM EDTA, and 10 mM Tris-HCl, pH 7.4) treated with RNAse and DNAse to remove nucleic acid contamination and loaded on a 30% polyacrylamide gel prepared in the following solution (in mM): 90 Tris-KOH, pH 8.3, 90 boric acid, and 2.7 EDTA in the presence of 7 M urea. A gel size of 70 cm was run for 30 min at 80 V, followed by 6 h at 40 V. polyP was visualized by DAPI staining [31].</p><!><p>Synthetic polyP with 14 phosphate residues linked together (short chain), 60 (medium chain), and 130 (long chain) were synthesized and purified as described before in [32]. To synthesize polyP, sodium phosphate was rapidly heated to the temperature higher than 600°C and the melted phosphate solution was rapidly cooled to the room temperature (RT). Depending on the heating temperature and time, polyP of various chain lengths could be synthesized. To extract polyP, sodium polyP was dissolved in purified water (10 w/v%), and then 96% ethanol was gradually added to the solution at final concentration of 14 w/v%. The solution was vigorously stirred and allowed to stand at RT for approximately 30 minutes. Then centrifugation (10,000 g, 20 minutes at RT) was performed to separate the precipitate from the aqueous solution. The aqueous solution fraction was discarded, and 70% ethanol was added to the collected precipitate for washing, and then the precipitate was vacuum-dried. The resultant polyP solution had high viscosity and normally contained more than 60% polyP. Gel permeation chromatography (HPLC) was performed to separate polyP by size using Shimadzu LC-2010C with refractive index detector (RID-10A) and Ohpak SB-803 HQ column (Shodex, column size 30 cm with 8 mm internal diameter) that had been equilibrated with 100 mM NaCl (pH 7.5). Samples were run at a flow rate of 1 ml per minute at a temperature of 25°C. Protein absorbance was used to determine protein elution. The fractions were tested for polyP by DAPI staining [31]. Combining of this heating method, subsequent ethanol extraction and size fractionation protocol, we obtained and purified short, medium and long-chain polyP with ~14, 60 and 130 phosphate residues, respectively. Stock solutions of polyP standards (sodium salt with polyP content of 60%) were prepared in distilled water.</p><!><p>Laser scanning confocal microscopy (Nikon A1 and a Zeiss LSM 780) was used to follow the changes in mitochondrial polyP accumulation, mitochondrial permeability transition pore (mPTP), mitochondrial membrane potential (ΔΨm), and mitochondrial ATP generation using specific fluorescent indicators. For measurement the myocytes were plated on laminin-covered coverslips and incubated in Tyrode solution containing 1 mM Ca2+ at room temperature. The fluorescence image was recorded every 5 sec. Fluorescence levels were corrected for background fluorescence and normalized as described below.</p><!><p>These were estimated in intact cells loaded with 5 μg/ml 4',6-diamidino-2-phenylindole, dihydrochloride (DAPI) for 30 min at 37°C [8, 25]. DAPI was excited with 408-nm laser light, and emitted fluorescence was measured at 552–617 nm. For DAPI emission spectrum recording, cells were excited at 408 nm, and the emission spectrum was collected between 500 and 675 nm. Data are presented as background subtracted fluorescence in arbitrary fluorescence units collected from the whole cell.</p><!><p>ATP measurements were performed indirectly via the free magnesium (Mg2+) concentration using the fluorescent dye mag-fluo-4 [33, 34]. Since free [Mg2+]i is kept constant within a rather narrow range, ATP hydrolysis leads to concomitant increase in free [Mg2+]i as measured with fluorescent Mg2+ indicators such as mag-fluo-4 [33]. Therefore, an increase in mag-fluo-4 fluorescence indicates a decrease in ATP concentration. For ATP measurements myocytes were loaded with 10 µM mag-fluo-4/AM (λex=488 nm, λem= 565–605 nm) for 30 min at 37°C. All data from these measurements are expressed as R=1-F/F0.</p><!><p>mPTP activity was monitored in permeabilized myocytes loaded with 5 μM calcein/AM (λex=488 nm, λem=510 nm) for 40 min at 37°C [34]. Opening of mPTP resulted in the loss of mitochondria-trapped calcein (620 Da) and a decrease of fluorescence. At the end of each recording 10 μg/ml of the pore-forming antibiotic alamethicin was applied to provide a control measure for maximum calcein release from mitochondria. Loss of mitochondrial calcein induced by elevating [Ca2+]em was quantified as the rate of decline of fluorescence (d(ΔF)/dt) calculated from the linear fit to the initial decrease of calcein fluorescence. The rate of decline was normalized (%) to the basal decline of calcein fluorescence addition (taken as 100%) before [Ca2+]em elevation. In experiments performed in intact cells, cardiomyocytes were additionally incubated with 1 mM CoCl2 to quench cytosolic calcein fluorescence as described in [25, 35].</p><!><p>Changes in mitochondrial membrane potential (ΔΨm) were followed using the potential-sensitive dye tetramethylrodamine methyl ester (TMRM; λex=514 nm, λem=590 nm) [25, 34]. Cells were exposed to 5 nM TMRM for 30 min at 37°C prior to experiments. All solutions contained 5 nM TMRM during recordings. In the end of each experiment 10 µM carbonyl cyanide p-(trifluromethoxy)phenylhydrazone (FCCP) was added to calibrate the signal. All data were background corrected.</p><!><p>Ischemia was simulated by acidosis (pH 6.4), inhibition of glycolysis (glucose replaced with 20 mM deoxyglucose), and inhibition of mitochondrial respiration (with complex IV inhibitor sodium cyanide, NaCN) by cell exposure to glucose-free modified Tyrode solution containing (in mM) 20 2-deoxyglucose, 2 NaCN, 135 NaCl, 4 KCl, 1 MgCl2, 2 CaCl2, and 10 Hepes, pH 6.4 for 20 min [25]. Reperfusion was simulated by 15 min superfusion with standard Tyrode solution consisting of (in mM) 135 NaCl, 4 KCl, 10 glucose, 10 Hepes, 1 MgCl2, and 2 CaCl2; pH 7.4.</p><!><p>Reactive oxygen species (ROS) measurements were performed in intact cells loaded with the fluorescent probe MitoSOX Red that preferentially detects superoxide, O2.−. Cells were loaded with 0.5 µM MitoSox Red for 30 min at 37°C [25]. Changes in [ROS] are expressed as rates of ROS production (d(ΔF/F0)/dt) estimated from the initial linear phase of the MitoSOX Red (λex=543 nm, λem=555–617 nm) fluorescence increase normalized to basal rate.</p><!><p>All data are expressed as mean ± standard error. Statistical significance of differences between experimental groups was determined using Student paired t test or two-way ANOVA, followed by Tukey's post-test when appropriate. A value of p<0.05 was considered statistically significant.</p><!><p>We have previously shown that inorganic polyP is a potent activator of Ca2+-induced mitochondrial permeability transition pore (mPTP) in rabbit cardiac myocytes and that polyP depletion by expression of yeast exopolyphosphatase (PPX) prevents mPTP activation [8]. However, when experiments were performed to simulate ischemia-reperfusion (I/R) injury, polyP depletion failed to protect from oxidative stress-induced mPTP opening and even exacerbated cell death following I/R [25]. We, therefore, aimed to determine the reason for this discrepancy and evaluate whether polyP has an additional, mPTP independent mechanism of action in the heart. To achieve this goal, we subjected Langendorff-perfused mouse hearts to I/R injury using the protocol shown in Fig. 1B.</p><p>Specifically, hearts were perfused with Krebs-Henseleit buffer for 5 min and then exposed to global no-flow ischemia for 20 min, followed by restoration of flow for 15 min (reperfusion). Mitochondria were isolated from 3 groups of hearts (control perfusion with no exposure to I/R, ischemia exposure only, and hearts exposed to sequential I/R) and polyP was detected as described before [8]. Fig. 1C illustrates a decrease in polyP chain length in mitochondria which where subjected to ischemia and I/R (left image), and the presence of polyP is evident by its characteristic ladder-like polyP distribution on high-resolution polyacrilamide gel (right image) [36]. The appearance of this ladder is caused by the fact that both polyP standard and polyP sample from mitochondria are composed of a heterogeneous mixture of polymers of various sizes. Each band seen on the gel represents a polyP polymer of a certain size, with every next band representing the polymer that differs by a single Pi group (Fig. 1 C, right). Furthermore, we detected that total polyP concentration was actually increased from 230 ± 30 pmol/mg of protein in control (untreated conditions) to 350 ± 50 pmol/mg polyP (Pi units released by PPX treatment), n=4, P<0.05 following I/R (Fig. 1D). The amount of polyP was quantified using a highly specific enzymatic assay that relies on measurements of Pi released from the sample upon treatment with recombinant PPX [37]. Overall, these ex-vivo experiments suggest that both amount and size of polyP change dramatically during I/R, and are in agreement with our previous in-vitro study where polyP levels were increased in conditions of simulated chemically-induced ischemia and reperfusion [25]. PolyP depletion achieved by expression of the mitochondrially-targeted PPX resulted in nearly ~80% decrease in total mitochondrial polyP levels. While polyP depletion with PPX led to the decrease in mPTP activity, it also increased cell death following I/R [25].</p><p>Since polyP depletion with PPX eventually leads to the decrease of both long and short polyPs [8], we hypothesized that the observed shortening in polyP chain length during I/R could be responsible for its different effects on mitochondrial function. To test this hypothesis, we proceeded with in-vitro experiments to examine the effect of synthetic polyPs with different chain length on mitochondria. In-vivo experiments with synthetic polyPs are not feasible because polyP is known to enhance blood coagulation and clotting [38, 39] which will make data interpretation difficult. Furthermore, at the moment there are no known plasma membrane transporters of polyP that also precludes us from performing experiments in intact hearts or cardiomyocytes with synthetic polyPs. Polyphosphates with different chain length: short (14 phosphate residues linked together), medium (60 phosphate residues), and long-chain (130 phosphate residues) were synthesized (Fig. 1E) and purified as described before [32] and the purity of preparation is shown in Figs. 1F and G.</p><!><p>First, we tested whether mitochondria are capable of polyP uptake under energized conditions in permeabilized cardiac myocytes. To do this, we used DAPI fluorescent probe and confocal settings optimized for polyP detection as we described before [8, 12, 25]. As shown in Figs. 2A and B, addition of 100 µM short (polyP-14) or long (polyP-130) did not lead to a significant polyP accumulation inside of mitochondria in the conditions of low extramitochondrial Ca2+ ([Ca2+]em). However, when [Ca2+]em was elevated from 0.1 to 2 µM, both short and long polyP accumulated inside mitochondria but at different rates and extents. As demonstrated in Fig. 2C, more short-length polyP was accumulated inside of mitochondria compared to long polyP. Interestingly, in both cases the polyP uptake was inhibited by F0/F1-ATP synthase inhibitor oligomycin, indicating a link between ATPase activity and polyP uptake. Furthermore, mPTP inhibition with 1 µM cyclosporin A (CsA) significantly elevated accumulation of long-chain polyP but did not affect accumulation of short chain polyP. These observations can be explained either by chain-length dependent differences in the rate of polyP membrane transport or in rates of polyP consumption inside mitochondria.</p><!><p>Next, we investigated the effect of added polyPs on mitochondrial energy metabolism by monitoring ATP generation with the Mg2+-sensitive fluorescent dye mag-fluo-4 (Fig. 3). This method takes advantage of the fact that ATP is the main intracellular Mg2+ buffer, such that reduced [ATP]i raises free [Mg2+]i,, and a decline in ATP levels leads to an increase in [Mg2+]i, resulting in corresponding changes in mag-fluo-4 fluorescence [33]. To reflect [ATP]i changes, we inverted mag-fluo-4 signals in Fig. 3. As shown in Fig. 3A, long-chain and to a lesser degree, medium polyPs increased ATP levels in permeabilized WT myocytes, while short chain polyP did not. Addition of the mitochondrial uncoupler 10 µM FCCP led to a significant decrease in ATP levels in myocytes treated with long and medium-chain polyPs, however did not have any effect on cells treated with short-chain polyP consistent with mitochondrial already being uncoupled. Importantly, short chain polyP was able to increase ATP levels in myocytes lacking cyclophilin D (CypD KO) that are resistant to mPTP (Fig. 3B, C). This potentially suggests that the lack of stimulation by short chain polyP in Fig. 3A might be due to mPTP activation (and uncoupling), rather than its inability to stimulate ATP production (which was preserved when mPTP was inhibited).</p><!><p>Next, using a Clark type electrode, we monitored oxygen consumption in isolated cardiac mitochondria from WT and CypKO animals. Mitochondria were placed in MiRO5 respiration buffer and basal respiration was recorded for 1 min. Then state 2 respiration was activated by addition of the mitochondrial complex I substrates 5 mmol/L malate and 5 mmol/L glutamate in the presence of 2 µM free Ca2+ and rate of oxygen consumption was measured. Addition of extramitochondrial Ca2+ was required since synthetic polyPs were entering mitochondria only in the presence of Ca2+ as shown in Fig. 2. Then ADP-induced respiration (State 3) linked to oxidative phosphorylation was measured by subsequent addition of 1 mmol/L ADP to the respiration chamber. Respiration rates (in nmol O2 min−1 mg mitochondrial protein−1) were normalized to no-polyP treatment conditions for Fig. 4. As shown in Fig. 4A, short polyP significantly increased basal respiration, while state 2 (Mal/Glut) and state 3 (ADP) respiration was decreased indicating a possible proton leak and uncoupling of the mitochondrial respiratory chain resulting in the decline in the respiratory control ratio (RCR; Fig. 4C). RCR is calculated as state 3 divided by state 2 respiration rate and represents a measure of the mitochondrial integrity and coupling efficiency. Interestingly, in agreement with data presented in Fig. 3, long chain polyPs significantly increased ADP-activated state 3 respiration in WT mitochondria leading to the enhancement in RCR and therefore mitochondrial coupling. The effect of long-chain polyP on the increase in state 3 and RCR was not different in WT and CypD KO mitochondria, however the uncoupling effect of short-chain polyP was prevented in mitochondria from CypD KO hearts, suggesting that uncoupling effect of short-chain polyP most probably was due to mPTP activation, and the effect of long polyP on energetics did not involve CypD.</p><!><p>Our data suggest that polyP can be taken up by mitochondria in a calcium dependent manner and stimulate mitochondrial ATP production. At the same time short chain polyP causes uncoupling effect on mitochondria that is prevented by KO of mPTP activator CypD. This is consistent with our previous reports that demonstrate that polyP can act as a potent activation of mPTP. To investigate whether the length of polyP polymer is important for previously observed effect of polyP on mPTP activity, we directly monitored mPTP activity in permeabilized cardiac myocytes with mitochondrially-entrapped fluorescent probe calcein. Mitochondrially-entrapped calcein green with a molecular weight of 620 Da can be released from mitochondria only when mPTP opens which allows ions and compounds with the molecular weight less than 1500 Da to pass through [28, 34]. In these experiments, mPTP activity was quantified as % change in the rate of calcein fluorescence decline compared to basal conditions (defined as 100% in [Ca2+]em = 0.1 μM). As shown in Figs. 5A and D, an elevation of [Ca2+]em from 0.1 to 2 μM caused calcein release from mitochondria (rate of fluorescence decline was 310 ± 55%; n = 20) that was significantly inhibited in both CsA-treated WT (102 ± 3%; n = 10, P <0.01 compared to untreated WT; Figs. 5B and D) and CypD-KO (123 ± 8%; n = 19, P < 0.01 compared to WT; Figs. 5C and D) myocytes confirming the fidelity of this assay for mPTP detection. As shown in Fig. 5A and summarized in Fig. 5D, short chain polyPs significantly enhanced the rate of calcein release (524 ± 52%; n =27, P<0.05 compared to control), and therefore mPTP activity in WT myocytes. Medium chain polyP did not have a significant effect (385 ± 58%; n =20, P=0.18), while long-chain polyP decreased mPTP opening (158 ± 13%; n =21, P<0.05 compared to control; Fig. 5A and D). The effect of short chain polyP was abrogated in WT myocytes in the presence of CsA (185 ± 26%; n =13, P<0.05 compared to control; Fig. 5B, D) or in CypD KO myocytes (157 ± 16%; n =14, P<0.01 compared to control; Fig. 5C, D) further confirming mPTP involvement.</p><!><p>Since the opening of non-specific mPTP channel dissipates the mitochondrial membrane potential (ΔΨm), we tested how polyPs with different chain length affect ΔΨm. To monitor ΔΨm, we measured Ca2+-induced changes in ΔΨm using the voltage-sensitive dye TMRM in permeabilized WT cardiac myocytes. As shown in Fig. 6A, elevation of [Ca2+]em from 0.1 to 2 µM is associated with small ΔΨm depolarization (degree of ΔΨm depolarization normalized to total depolarization achieved after addition of the mitochondrial uncoupler 10 µM FCCP was 22 ± 2%; n = 6). Addition of short polyP, however, dramatically increased ΔΨm depolarization (62 ± 5%; n =6, P<0.01 compared to control; Fig. 6A, B) which corresponds to the increase in mPTP opening during mitochondrial exposure to high Ca2+ presented in Fig. 5A. Medium-chain polyP did not affect ΔΨm significantly (18 ± 2%; n =21, P=0.08 compared to control) while long-chain polyP actually diminished the degree of ΔΨm depolarization induced by Ca2+ (15 ± 2%; n=6, P<0.01 compared to control; Figs. 6A, B). To conclude, these data indicate that only short chain polyPs enhance mPTP activation which leads to dissipation of ΔΨm across mitochondria membrane.</p><!><p>Since polyP is known to enhance blood coagulation and clotting [38, 39] we were not able to perfuse the whole mouse hearts with synthetic polyPs. To verify that conditions of ischemia and reperfusion (as shown in Fig. 1) are associated with excessive oxidative stress and mPTP opening, we exposed intact mouse cardiac myocytes to 20 min of simulated ischemia followed by 15 min of reperfusion with a similar protocol as we used before [25]. As shown in Figs. 7A, B, mPTP opening as monitored by calcein release from mitochondria was already observed under ischemic conditions, but was further exacerbated in reperfusion. Superoxide generation monitored by changes in MitoSOX Red fluorescence (Fig. 7C) was already elevated to a maximal degree during ischemia, and remained elevated during I/R. Despite the observed protection against mPTP opening in WT myocytes treated with CsA or CypD KO mice (Figs. 7A, B), the levels of superoxide were not changed under these conditions suggesting that oxidative stress was causing mPTP opening under ischemic conditions. Based on the fact that ischemia was associated with the increased polyP generation in the heart (see Figs. 1C,D), we suggest that similar to bacteria [40] mammalian cells produce polyP in response to the oxidative stress. Under conditions of the excessive mPTP opening in reperfusion (red line in Fig. 7A), long polyP could be used to support ATP generation leading to generation of short-chained polyPs capable of mPTP activation (see Fig. 5A). This, in fact, is indirectly supported by our data presented in Fig. 3A where addition of the mitochondrial uncoupler FCCP which dissipates ΔΨm (see Fig. 6A) similar to that observed in I/R, was associated with dramatic decrease in ATP levels when long polyP was used to support respiration. Addition of FCCP had no effect on ATP levels when short polyP was used, suggesting that mitochondrial respiratory chain was already uncoupled. These data, together with other results presented in this study suggest that polyP effects could be different depending on the metabolic status of cell.</p><!><p>In this study we investigated the differential effects of polyP on cardiac mitochondrial function. The main motivation was to find the solution to the prior dichotomy that mitochondrial polyP depletion protected hearts from Ca2+-overload induced mPTP opening [8, 12], but failed to protect against mPTP opening during I/R and even exacerbated cell death [25]. In this study, we found that differences in the functional effects of short vs. long polyP chains (14 vs. 130 phosphates) may resolve this dichotomy. Specifically, short polyP promotes mPTP and proton leak (bad), while long polyP suppresses mPTP and increases both ATP production and RCR (good). Moreover, ischemia-reperfusion caused a shift in endogenous polyP from longer to shorter forms (Fig. 1). This would worsen mPTP and mitochondrial dysfunction because of both the higher maladaptive effects of short and the loss of beneficial effects of long polyP. The depletion of long polyP by mitochondrially-targeted exopolyphosphatase (PPX) would remove the benefit of long polyP on mPTP suppression and may explain why polyP depleted hearts fared somewhat worse than controls during I/R.</p><p>To elucidate this duality of polyP effects, we employed synthetic polyPs with different chain length (Fig. 1), tested their ability to enter mitochondria (Fig. 2), influence energy metabolism (Figs. 3 and 4) and modulate Ca2+-induced mPTP and mitochondrial membrane potential (Figs. 5 and 6). We found that these synthetic polyPs could be taken up by mitochondria when added to permeabilized ventricular myocytes (Fig. 2), but that mPTP activation could limit polyP uptake and that the mitochondrial ATP synthase might be involved in polyP transfer across the inner mitochondrial membrane. The potential molecular mechanism for this ATP synthase involvement may merit further study. The use of polyP of different lengths allowed us to distinguish the aforementioned differential functional effects of short vs. long polyP. That is, short chain polyPs led to the activation of the mPTP (Fig. 5), induced proton leak in the oxidative phosphorylation pathway (Fig. 4), mitochondrial uncoupling (Figs. 3 and 6) and failed to increase ATP levels in cardiac myocytes (Fig. 3). In contrast, longer chain polyPs demonstrated a mainly protective effect against mPTP opening and also enhanced energy generation. These data further support the idea that involvement of polyP in mitochondrial function should be considered in connection with specific metabolic and functional state of the organelle. Indeed, a prior study performed in human leukemic cell line HL-60 demonstrated that apoptosis induction by cell treatment with actinomycin D resulted in degradation of long chain polyP and accumulation of short chain polyPs [10]. Furthermore, their study has also reported a significant decrease in long chain polyPs in rat brain and liver with aging, again emphasizing the importance of long polyPs for cell survival [10]. This might be particularly important due to the link between polyP and levels of ATP and ADP which are known modulators of mPTP [30]. Indeed, our work supports earlier studies regarding the close link between mitochondrial energetics and levels of polyP.</p><p>The possible contribution of polyP towards energy production in mammalian cells has been suggested by us before [11, 12]. In these experiments stimulation of energy metabolism led to increased mitochondrial polyP, while in cultured cells polyP depletion led to energetic failure. Interestingly, it was also demonstrated that mitochondrial depolarization leads to the decrease in polyP chain length – an effect similar to what we observed in this work during I/R. Studies performed in bone-forming osteoblast SaOS-2 cells demonstrated that addition of inorganic polyP stimulated bone maturation which was associated with an increase in ATP levels [41, 42]. Ca2+ ions were required for polyP to be effective for stimulation of bone mineralization and energy production in SaOS-2 cells. Notably, another study found that only medium and long-chain polyPs stimulated tissue regeneration and bone formation while short-chain polyPs did not have this positive effect [32].</p><p>While, as mentioned above polyP involvement in metabolism can be important contributor to the regulation of mPTP, our data do not indicate such a direct link. We hypothesize that in case of mPTP activation, polyP might be involved as a chaperone or a structural component of the mPTP channel. Previous studies have demonstrated that polyP forms complexes with poly-β-hydroxybutyrate (PHB) and Ca2+ on bacterial plasma membranes, and participate in Ca2+ transport inside bacteria which is responsible for initiation of bacterial competence [22, 43]. Pavlov et al. [44] were able to isolate similar polyP-Ca2+-PHB complexes from rat liver mitochondria which upon incorporation into lipid bilayers induced a current with characteristics similar to that observed in mPTP channels. Interestingly, the study of Baev et al. [45] did not find any effects of polyP on permeability of lipid bilayers when polyP was added alone to the artificial membranes but demonstrated modulatory effects of polyP on the membranes from the native de-energized mitochondria. This is a very important observation that suggests that polyP needs a co-partner to activate mPTP in mitochondria. We further confirmed the presence of polyP in cardiac mitochondria and their contribution toward mPTP activation upon mitochondrial Ca2+ overload [8]. PolyP which are present in mammalian cells are typically longer [3, 6] compared to polyPs found in bacteria and which are used for commercial purposes [46]. This is consistent with our data presented in Fig. 1 where under physiological conditions polyPs with medium to long chain length were present in cardiac mitochondria. However, a significant shift towards short polyPs was detected under pathological conditions such as I/R.</p><p>The mPTP activation process is complicated and the molecular details are not completely understood, with several possibilities currently been investigated [19, 47–49]. What appears to be an essential condition for mPTP activation is Ca2+ uptake by energized mitochondria in the presence of orthophosphate. Further, while molecular composition of the pore part of mPTP is not well established [49–51], mPTP activity strongly depends on the interaction with protein cyclophilin D (CypD) which belongs to the family of peptidyl-prolyl cis-trans isomerases whose enzymatic activity can be inhibited by cyclosporin A [52].</p><p>The mPTP effects of polyP here are not non-specific polyP effects on lipid bilayer membranes because it is strongly dependent on CypD and presence of mitochondrial Ca2+ uptake. Together with previous reports, the polyP involvement in mPTP seen here could occur by several, non-mutually excluding, mechanisms: 1) polyP might be directly involved in mPTP channel formation through Ca2+-mediated interaction with PHB and ATP synthase; 2) polyP effects could be linked to its ability to regulate mitochondrial Ca2+ buffering capacity thereby regulating the mPTP activation threshold; 3) polyP might be involved in the mPTP through its chaperone activity [40, 53] and it's ability to modulate other proteins involved in mPTP activation and/or complex assembly. Specifically, the study of Gray et al. demonstrated that bacteria exposed to the oxidative stress respond by a dramatic accumulation of long-chain polyPs which are required for bacterial surviving in proteotoxic stress conditions [40]. This protective effect of polyP was related to its chaperone's action and the ability of polyP to protect certain proteins against stress-induced unfolding and aggregation [40]. The concept of misfolded proteins as a key trigger of ROS-induced mPTP opening is well known [54]. Lemasters's group [54] introduced the concept that mPTP forms by aggregation of misfolded integral membrane proteins damaged by oxidative stress. According to their model [54], mitochondrial damage after the oxidative stress causes misfolding and clustering of native mitochondrial membrane proteins. This misfolding can potentially expose hydrophilic residues to the bilayer phase to enclose aqueous channels conducting low molecular weight solutes. Initially, chaperone-like proteins are able to block conductance through these misfolded protein clusters [54]. However, when protein clusters exceed the ability of chaperons to block conductance [55] during increased oxidative stress observed in reperfusion, mPTP opens in high-conductance mode [54] leading to Ca2+, other ions and metabolites release from mitochondria. In our 2015 study [25], we demonstrated that total polyP levels in control rabbit ventricular myocytes were increased under conditions of simulated I/R suggesting that polyP is formed under the stress conditions. PolyP depletion by expression of mitochondria-targeted exopolyphoshatase (PPX) that specifically hydrolyzes polyP into inorganic phosphate resulted in nearly 80% decrease in mitochondrial polyP levels. While polyP depletion by PPX overexpression led to the decrease in mPTP activity, it also increased cell death following I/R [25]. Since polyP depletion with PPX eventually leads to the decrease of both long and short polyPs [8], we hypothesized that the observed shortening in polyP chain length (see Fig. 1 in the current manuscript) during I/R could be responsible for its different effects on mitochondrial function. Similar to the results obtained in rabbits, mPTP opening was observed in both ischemia (in a smaller degree) and reperfusion (more pronounced) conditions in isolated mouse cardiomyocytes (Figs. 7A, B). The decrease in calcein fluorescence (used as a measure for mPTP) was attenuated by cell treatment with CsA and was not observed in CypD KO mice confirming that the observed decrease in calcein fluorescence was indeed due to mPTP activity. Even though conditions of ischemia (in particular low pH) do not favor mPTP opening [56], other studies beside us showed mPTP opening during ischemia (see [57] for the review). However, in agreement with our previous study [25] and data from the Schumacker's group [58, 59], mitochondrial oxidative stress was already observed during ischemia (see also Fig. 7C) which could contribute to polyP formation, mPTP opening and prime cardiomyocytes for cell death during reperfusion.</p><p>We think the third scenario is most likely. While we could not exclude the direct involvement of polyP in channel formation, earlier reports suggested that medium length polyPs were forming complexes with PHB in bacteria [22, 43, 60], but our strongest effects were with short-chain polyP chains. It is possible that the effect of polyPs with different chain length could be tissue, species, and concentration-dependent. Indeed, the study of Baev et al. [45] revealed that polyPs with longest chain length had the most prominent effect on mitochondrial swelling in de-energized rat liver mitochondria. These experiments, however, were performed in highly isosmotic conditions in the presence of 40 mM Ca(NO3)2 and 1 µM rotenone, a known inhibitor of the mitochondrial complex I. These experimental conditions would prevent possible beneficial effects of long polyPs to enhance mitochondrial bioenergetics since rotenone inhibits mitochondrial respiratory chain, and therefore ATP generation. While it is possible that such a difference is due to the experimental conditions it is tantalizing to propose that effects of polyP could be also tissue specific. Earlier studies from Kornberg's lab [6] reported that heart and brain tissues contained the highest amount of polyP compared to other tissues. For examples, approximately 95 and 114 µM were extracted from brain and heart tissue, respectively, while only 38 µM polyP was extracted from liver tissues with similar protein content [6]. Furthermore, electrophoretic analysis revealed that all tissues samples examined in this study except the liver contained a mix of short, medium and long polyPs. Strikingly, only long chain polyPs were detected in liver samples [6, 9]. The physiological importance of this observation needs further evaluation but possibly can explain the difference between our study and the study of Baev et al. [45] which was performed using isolated liver mitochondria.</p><p>Ca2+ buffering effects are non-specific and thus are not expected to be chain dependent. Furthermore, our prior study did not find any significant changes in mitochondrial Ca2+ levels in cardiac myocytes treated with exopolyphosphatase PPX [8]. In contrast, polyP binding to molecular protein targets can be chain-length dependent [40, 61]. This interpretation is consistent with our observation that short chain polyP stimulates ATP production only in CypD KO cells (but not in WT cells) suggesting that in the absence of a mPTP inhibitor short chain polyP is bound to its putative molecular target. Whether the chain-length dependent target is CypD or some other protein remains to be established. Intriguingly, a recent study discovered that several human proteins are being post-translationally modified by polyP in the process called polyphosphorylation [62]. However, further studies are required to determine how this polyphosphorylation affects protein function and/or conformation.</p><p>We should also emphasize that long chain polyPs had no stimulating effect on mPTP activity, but rather inhibited mPTP opening (Fig. 5), and the effect of long polyP on ATP generation was not dependent on CypD. This may indicate that long chain polyPs lack the ability to bind to CypD and are thus potentially cardioprotective during I/R. While the enzyme responsible for polyP formation in mammalian cells is still unknown, it was demonstrated that human gastrointestinal tract bacteria (probiotics) produce polyP, and that polyP is responsible for probiotic actions that protect the intestinal epithelia from oxidant stress and epithelial injury [63]. Intriguingly, several studies [64, 65] demonstrated that probiotic administration attenuates myocardial infarction following I/R injury and prevents myocardial hypertrophy and heart failure development following myocardial infarction in the rat. Our data revealed that similar to bacteria, cardiac myocytes exposed to the oxidative stress during ischemia respond with compensatory increase in polyP formation (as shown in Fig. 1C, D). During reperfusion, however, polyP is used by cardiac myocytes to support cellular bioenergetics leading to the depletion of long polyP and accumulation of short chain polyPs which promote mPTP opening. From this point of view, promoting long polyP generation in reperfusion could be cardioprotective.</p><!><p>Taken together, these experiments have shown that inorganic polyP is actively generated and metabolized under conditions of stress such as ischemia and reperfusion. During I/R a significant reduction in polyP chain length is observed (from long to short). Using synthetic polyPs, we found that polyP has a dual effect on mitochondrial function in cardiac myocytes depending on polyP chain length. Short chain polyP (14 phosphate residues) strongly activated Ca2+-induced mPTP opening in cyclosporin A and cyclophilin D-dependent manner, indicating possible interactions of polyP with cyclophilin D. Long-chain polyP (130 phosphate residues) actually decreased mPTP opening and enhanced ADP-linked respiration and ATP levels in mitochondria. This could explain the fact that polyP depletion did not prevent mPTP opening and even exacerbated cell death following I/R in our previous study [25]. Therefore, promoting long chain polyP formation during I/R could be beneficial for cardiac function. Overall, these findings support the idea that the involvement of polyP in mitochondrial function should be considered in connection with the specific metabolic and functional state of the organelle.</p>

PubMed Author Manuscript

Accurate Zygote-Specific SNP Discrimination Using Microfluidic Electrochemical DNA Melt Curves**

We report the first electrochemical SNP detection system that can accurately discriminate homozygous and heterozygous genotypes using microfluidics technology. To achieve this, our system performs real-time melt-curve analysis of surface immobilized hybridization probe. As an example, we used our sensor to genotype two SNPs in the apolipoprotein E (ApoE) gene, where homozygous and heterozygous mutations greatly affect the risk of late-onset Alzheimer\xe2\x80\x99s disease. Using probes specific for each SNP, we simultaneously acquired melt curves for probe-target duplexes at two different loci and thereby accurately distinguish all six possible ApoE allele combinations. Since the design of our device and probes can be readily adapted for targeting other loci, we believe that our method offers a modular platform for SNP-based disease diagnosis and personalized medicine.

accurate_zygote-specific_snp_discrimination_using_microfluidic_electrochemical_dna_melt_curves**

<p>Single nucleotide polymorphisms (SNP) are important diagnostic indicators for inheritable diseases[1, 2] and predicting drug responses.[3] Genetically-informed treatment and dosing decisions will require effective platforms for SNP detection that are rapid, portable, and cost effective.[4] Such platforms must also support multiplexed detection and accurately differentiate between homozygous and heterozygous mismatches – a critical factor for many diagnoses.[5, 6]</p><p>Electrochemical DNA sensors offer robustness, portability, and compatibility with microfluidics and microelectronics,[7, 8] and many varieties of electrochemical DNA sensors (as reviewed by Palecek and Bartosik)[9] have been developed for SNP detection, including various enzymatic[10–12] and hybridization-based assays.[13–15] Surface-immobilized, redox-reporter labeled probes offer many advantages in this arena as they require no exogenous reagents, exhibit low background, and can operate directly in complex mixtures.[16] Innovative probe designs based on triple-stem probes,[17] polarity-switching probes,[18] and base-pair stacking probes[19] have all proven successful at SNP detection. Unfortunately, none of these sensors have shown the capacity to accurately discriminate heterozygous mismatches, as such samples contain a mixture of matched and mismatched DNA, and these probes cannot accurately resolve the minute differences in hybridization energy. Fluorescence-based, solution-phase methods such as dynamic allele-specific hybridization (DASH) can, in contrast, distinguish homozygous and heterozygous SNPs via melting curve analysis in solution phase.[20] This has spurred interest in integrating melt-curve analysis with electrochemical detection, but these efforts have been confounded by difficulties in obtaining accurate electrochemical and temperature measurements with sufficient speed and resolution.[21–23]</p><p>We present a novel microfluidic device that performs electrochemical melt-curve measurements with unprecedented temporal resolution at speeds that enable "real-time" analysis, and can accurately discriminate between homozygous and heterozygous SNP genotypes. Our 'microE-DASH' chip employs surface-bound, redox-tagged DNA probes[24] complementary to a SNP-containing target sequence. As we heat the chip, we can obtain a melt-curve by continuously measuring changes in the electrochemical signal. Furthermore, microE-DASH can achieve multiplexed detection of homozygous and heterozygous genotypes at multiple loci in a single step by integrating multiple probes and electrodes within the chip. As a model, we used microE-DASH to genotype two distinct SNP loci associated with the ApoE gene, where specific alleles strongly affect individual risk of late-onset Alzheimer's disease.[25]</p><p>MicroE-DASH achieves accurate SNP discrimination by monitoring thermal melting of duplexes formed by surface-bound probes and their DNA targets in real time (Figure 1A). The probes are linear, single-stranded DNA, modified with a redox reporter (methylene blue; MB) at their 3' end and self-assembled onto gold working electrodes via a tri-thiol modification at their 5' end. In the absence of target, these probes are unstructured and flexible, allowing the MB reporter to readily approach the gold electrode, generating a redox current that can be measured using alternating-current voltammetry (ACV). However, probe-target hybridization decreases the current because the stiffer double-stranded duplex greatly reduces reporter access to the electrode.[26]</p><p>To distinguish perfectly-matched (PM) targets from those containing single-base mismatches (MM), we ramp the temperature until the duplex completely melts. Due to its higher hybridization energy, the PM target (Figure 1A, top) melts at a higher temperature than the MM target (Figure 1A, bottom), and MicroE-DASH can readily distinguish the two by continuously measuring redox current as a function of temperature (Figure 1B, left). Furthermore, we can accurately determine the melting temperature of the duplex (Tm) by taking the derivative of the current as a function of temperature (dI/dT) as described below (Figure 1B, right). Heaton et al.[27] and Mahajan et al.[28] have shown that strong DC electric fields can influence DNA duplex formation and dyhybridization. However, given the lower magnitude and shorter duration of our AC measurements, we suspect that electric fields have negligible effect on the measured melt temperature of our duplexes.</p><p>The MicroE-DASH chip contains a single reaction chamber that incorporates platinum reference and counter electrodes and two probe-conjugated gold working electrodes to form a multiplexed electrochemical measurement cell (Figure 1C; see Supporting Information for fabrication details). The chip is mounted on a Peltier heater, which controls the temperature in a pre-programmed sequence while a potentiostat continuously records the peak redox current from the two working electrodes. Importantly, the chip is designed for minimal thermal resistance, so that the temperature of the electrodes and reaction chamber can equilibrate in seconds (Figure S1). This enables near-real-time acquisition of melt-curves, such that the entire assay can be completed within 30 minutes without the need to return to room temperature for measurement.[22, 29] To ensure probe stability, we immobilized them onto the electrodes via 5-tri-thiol modification.[30] This dramatically improves their thermal stability; mono-thiol probes detach from the electrodes at temperatures as low as 75 °C, whereas tri-thiol probes are stable at temperatures up to 85 °C, maintaining >99% of the maximum signal (Figure S2).</p><p>As proof of concept, we designed probes for SNPs within the ApoE gene that serve as important clinical diagnostic indicators for Alzheimer's disease.[25] In addition to the normal isoform (ε3), this gene has two variants (ε4 and ε2) arising from SNPs rs429358 (T:C) and rs7412 (C:T), referred to as T1 and T2 respectively in this work. Carriers of the ε4 allele – and ε4-ε4 homozygotes in particular – exhibit greater risk of developing Alzheimer's.[31] Conversely, the ε2 allele is associated with a reduced likelihood for Alzheimer's.[32] Accurate identification of the six possible allele combinations (Table 1) is thus important for identifying high-risk individuals.</p><p>We designed probes complementary to the T1 and T2 sequences associated with the ε3 isoform. This allowed us to readily identify homozygotes for ε3, ε2, ε4, or any heterozygous combination of alleles, based on whether T1 and T2 form PM or MM duplexes with their respective probes (Table 1). We considered three key criteria in designing the probes. First, we ensured that Tm occurs between 45–65 °C, which is sufficiently high for distinguishing secondary melt transitions but within the thermal stability range of the tri-thiolated probes. Second, we designed probes with minimal secondary structure because self- hybridization can result in complex melting behavior and compete with target hybridization. Finally, we designed the two probes with minimal sequence overlap to minimize potential cross-reactivity. We performed modeling with mfold[33] software to ensure our probe design satisfied the first two criteria (see Supporting Information for sequences), and verified their target specificity by immobilizing the probes onto gold electrodes and incubating them with non-matching targets (i.e., T1 target with T2 probe, and vice versa), resulting in minimal signal change compared to matched targets (Figure S3).</p><p>ACV measurements taken as a function of temperature can precisely track the melting characteristics of DNA targets. To demonstrate this, we hybridized the PM target to the T2 probe and measured the electrochemical current from 25 to 85 °C in 1 °C intervals. We observed a dramatic increase in redox current between 45 and 65 °C (Figure 2A), consistent with the sudden transition expected during melting of the T2-PM duplex. In the absence of target, instead of the sudden current increase expected from a duplex melting transition, we saw a current increase resulting from increased thermal motion of probe molecules at higher temperatures,[23] which in turn results in more frequent interaction between redox reporter molecules and the electrode (Figure S4).</p><p>Differences in melting behavior can be directly observed in real-time by plotting the peak redox current as a function of temperature. We demonstrated this after challenging T2 probes with PM, MM or a heterozygous (HET) mixture of PM and MM targets (1:1 ratio) (Figure 2B). We normalized peak-current readings to a 0 to 1 scale using the initial and maximum current values in each run, with five-point averaging to smooth the melt curve (see Methods). As a general trend, we observed clear differences among the three melt curves. The PM target melted at a higher temperature (Tm = 61 °C) than the MM target (Tm = 56 °C), while the curve for the HET mixture fell in between, with a Tm of 58 °C.</p><p>To determine accurate Tm we calculated the derivative of the current with respect to temperature (dI/dT) and defined Tm as the point at which this derivative is largest, analogus to established fluorescence melting approaches.[34] PM and MM targets produced single, distinct peaks with Tm of 63 °C and 56 °C, respectively (Figure 2C, red and black). This large difference enabled accurate and reproducible detection of SNPs in homozygous samples. Analysis of HET samples is challenging because the resulting dI/dT plot consists of two peaks from the PM and MM targets that can potentially overlap (Figure 2C, blue). To de-convolve the contribution of each probe-target duplex, we adopted an analytical strategy used for extracting peaks from multiple melt-curves (see Methods). We applied a curve-fitting algorithm to our single-peak PM and MM melt-curves, and determined that a Lorentzian function yields the best fit to our experimental data (Figure S5).[35, 36] When we applied this fit to the T2 heterozygous melt-curve (Figure 2C), we identified two curves with peaks at 65.4 °C and 56.6 °C. These correspond well to the measured Tm of PM and MM, respectively, confirming the validity of this approach.</p><p>MicroE-DASH's capacity to simultaneously differentiate multiple homozygous and heterozygous SNP samples enabled us to accurately genotype all six ApoE SNP combinations in a single-step assay. We assembled a duplex microE-DASH chip with a different probe immobilized on each of the two sensors, allowing us to simultaneously determine melt temperatures for both T1 and T2. We first obtained melt curves for each of the three homozygous ApoE genotypes (Figure 3A). As expected, these samples generated single peaks for both T1 and T2, since SNP mismatches occur in both alleles for any given homozygous genotype. The single-peak Tm measured in these duplex devices agree with repeated measurements from single-target experiments of T1 (PM: 61 ± 1.5 °C, MM: 50 ± 1 °C) and T2 (PM: 62 ± 0.6 °C, MM: 56 ± 0.6 °C). Furthermore, the shifts in the melt curve observed for the ε2-ε2 and ε4-ε4 samples corresponded with our predictions based on the target-probe mismatches identified in Table 1.</p><p>We subsequently used microE-DASH to accurately identify all three possible heterozygous genotypes. We obtained T1 and T2 melt curves as described above, and determined that each heterozygous combination yielded the predicted melt-curve shifts (Figure 3B). The ε3-ε2 and ε3-ε4 genotypes resulted in heterozygous melt curves for T2 and T1, respectively, while the ε2–ε4 genotype yielded heterozygous melt curves for both probes. We used deconvolution analysis on the heterozygous melt curves to extract individual Tm for each target-probe combination, which were in agreement with the PM and MM Tm obtained from homozygous samples. For example, the T2 melt curve for the ε3-ε2 target yielded peaks of 58°C and 64°C, which correspond well with our previous MM and PM Tm measurements of 57°C and 63°C respectively. The ΔTm for these heterozygous samples matches well with the individually measured single-peak Tm. This confirms that microE-DASH can accurately report both homozygous and heterozygous genotypes independent of the specific sample composition.</p><p>The microE-DASH microfluidic electrochemical SNP biosensor is thus capable of discriminating homozygous and heterozygous samples within ~30 minutes. Demonstrating the potential diagnostic utility of this, we have accurately resolved the six different SNP genotypes commonly associated with ApoE, an important diagnostic biomarker for Alzheimer's disease. microE-DASH can be readily expanded to incorporate microfluidic PCR and single-strand generation in an integrated device.[37] Since the design of our probes is relatively straightforward, we believe that our microE-DASH may offer a modular platform for SNP-based disease diagnostics and personalized medicine.</p><p>We are grateful for financial support from the Otis Williams Fellowship, National Institutes of Health, and the Institute of Collaborative Biotechnologies through the Army Research Office. Microfabrication was carried out in the Nanofabrication Facility at UCSB.</p><p>Supporting information for this article is available on the WWW under http://dx.doi.org/10.1002/anie.201xxxxxx.</p><p>Micro-Electrochemical Dynamic Allele-Specific Hybridization (microE-DASH). A) MicroE-DASH obtains real-time electrochemical melt curves from linear, single-stranded DNA probes complementary to the SNP target. The probes are linked to electrodes via a 5' tri-thiol anchor with a methylene blue (MB) redox reporter at the 3' terminus. The flexible unbound probe allows MB to readily approach the working electrode, generating high current. Rigid target-probe duplexes restrict MB-electrode interaction, resulting in a current decrease. Temperature-dependent changes in redox current reveal differences in melting temperature (Tm) between perfectly-matched (PM) and mismatched (MM) targets. B) The Tm corresponds to the temperature at which the rate of current change is greatest. We can thus determine the Tm by the peak of the first derivative of the redox current as a function of temperature (dI/dT). C) The microE-DASH chip consists of two glass pieces separated by a PDMS gasket, mounted on a programmable Peltier heater. The lower glass piece features two working electrodes, coupled with counter and reference electrodes, which form an electrochemical cell.</p><p>DNA sample genotyping using microE-DASH. A) Peak current traces taken in 10°C increments during temperature ramp of T2 probe incubated with a PM target reveal melting-dependent increases in current. B) Plotting the peak currents measured for T2 probe incubated with PM, MM and a heterozygous 1:1 mixture of PM and MM at 1°C increments reveals distinct melt-curves for each genotype. Based on the half-maximal current, the PM (black) target forms the most stable duplex with the highest Tm. The MM (red) target forms a less stable duplex with a lower Tm, while the heterozygous sample (blue) exhibits an intermediate Tm. C) By plotting the derivative of the current as a function of temperature (dI/dT) versus temperature, we can determine Tm with a precision of approximately ±1 °C. The target trace from each genotype has a unique peak.</p><p>ApoE genotyping via multiplexed real-time melt-curve analysis. Plots of independent melt-curve derivatives from samples containing PCR-length ApoE targets simulating different A) homozygous (PM and MM) and B) heterozygous (Het) genotypes, with the corresponding Tm reported in black. Each row represents a different allele combination, associated with a different response in the T1 or T2 melting curves. Black plots indicate perfect match between probe and target, whereas red plots indicate SNP mismatch. Blue plots represent profiles with dual peaks arising from heterozygous samples.</p><p>ApoE genotypes and their associated MicroE-DASH readout</p>

PubMed Author Manuscript