Datasets:

griffin

/

ChemSum

like 11

Stereoselective, Dual-Mode Ruthenium-Catalyzed Ring-Expansion of Alkynylcyclopropanols

A novel, dual-pathway ring-expansion of alkynylcyclopropanols is described. On treatment with a ruthenium catalyst, these compounds undergo highly selective enlargement to either (Z)-alkylidene cyclobutanones or \xce\xb2-substituted cyclopentenones. The unique ability to access the least selective double bond isomers of alkylidene cyclobutanones and the dramatic shift of reactivity observed further illustrate the particular intricacies of ruthenium catalysis when compared to other alkynophilic transition metals.

stereoselective,_dual-mode_ruthenium-catalyzed_ring-expansion_of_alkynylcyclopropanols

<!>Supporting Information Available<!>

<p>The fascinating chemistry of small-ring compounds stems almost invariably from the unique reactivity modes allowed by the intrinsic ring strain in these systems.1 In particular, ring-expansion reactions have been abundantly used in organic synthesis to fashion functionalized molecules in an efficient and expeditious manner, and the appearance of various transition metal-catalyzed ring expansion processes has only enriched this landscape. 2-3</p><p>There is a considerable body of work on the transition metal-catalyzed ring expansion of vinyl and allenyl cycloalkanols,4 which provide useful tools for the construction of various cyclic ketones. This contrasts with the scarcity of reports of transition metal-promoted skeletal rearrangements of alkynylcycloalkanols.5</p><p>Our recent interest in tapping the vast potential of alkynes as selective mediators in metal-catalyzed bond-forming reactions led us to speculate whether ruthenium catalysis would provide an interesting addition to the current arsenal of ring-expansion processes.6 The remote analogy between the isomerization of a propargyl alcohol 1 to an unsaturated carbonyl 1 (termed the redox isomerization reaction7, Scheme 1) and the skeletal rearrangement of a tertiary, cyclopropyl carbinol 4 further spurred our interest. Herein we report that ruthenium catalysis is unique in the activation of alkynyl cyclopropanols 4 as it mediates a highly selective, dual ring-expansion to either four- or five-membered cyclic ketones.</p><p>Gratifyingly, our initial forays were successful. Treatment of the TMS-substituted alkynylcyclopropanol 4a with catalytic amounts of ruthenium complex 2 smoothly triggered ring-expansion to alkylidene cyclobutanone 6a in essentially quantitative yield. Interestingly, the least stable (Z)-isomer was formed with nearly 6:1 stereoselectivity (Table 1, entry 1). With our curiosity piqued by these observations, the little precedent found for the expansion of silyl-substituted alkynyl cyclopropanols5c prompted us to examine more in detail this class of substrates. Our results are collected in Table 1.</p><p>As can be seen, the trend for the preferential formation of (Z)-silylalkylidene cyclobutanone products upon exposure to our conditions appears to be quite general. Strikingly enough, as the steric bulk of the silyl substituent increases, so does the Z:E ratio. The corolary of this premise is that the highly congested TIPS-substituted alkynylcyclopropanol 4f (Table 1, entry 6) leads exclusively (as far as NMR-detection is concerned) to the (Z)-cyclobutanone 5f, a most counter-intuitive result!</p><p>Realizing that the electronic properties of silyl moieties might be playing a prominent role in this outcome, we then decided to examine electron-withdrawing substituents. The results of these experiments are compiled in Table 2.</p><p>In contrast to the silyl-substituted substrates, in this case the conversion was slower, which could be ascribed to the lower electron-density at the alkyne (vide infra). Nonetheless, good yields of alkylidene cyclobutanones 8 were obtained and this regardless of the electron-withdrawing substituent being a ketone (entry 1) or ester (entries 2-4) group. It should be noted that the nature of the ester group (aliphatic, benzylic or nitroaromatic) also does not affect the outcome of the reaction. Importantly, and in analogy with the case of silyl-substituted alkynylcyclopropanols (cf. Table 1), a single isomer was obtained in all cases, which was assigned the (Z)-configuration. It is important to note that the stereochemical outcome for these reactions is the precise opposite of what was reported using gold-catalysis, suggesting that different mechanistic pathways may be operative in each case.5</p><p>Having witnessed the ability of our catalytic system to efficiently convert silyl- and acceptor-substituted alkynylcyclopropanols to stereodefined alkylidene cyclobutanones, we were eager to probe the stereoselectivity of the analogous process employing electron-"neutral" alkyl substituents at the alkyne.</p><p>To our surprise, when we exposed the hexyl-substituted alkynylcyclopropanol 10a to our reaction conditions (equation 1), a new product was formed which was not the anticipated cyclobutanone 11a. We quickly realized that the unexpected β-substituted cyclopentenone 12a had been generated instead!</p><p>Impressed by this complete shift in reactivity, we set out to examine the generality of this observation and briefly examined the alkyl-substituted substrates depicted in Table 3.</p><p>Interestingly enough, substrates comprising benzyl (entry 2), cycloalkyl (entry 3), or remote alkoxy (entries 4-5) and halide (entry 6) substituents all underwent completely selective ring-enlargement to the corresponding cyclopentenones. In all of these cases cyclopentenones 12 were obtained exclusively, with only trace amounts of the analogous cyclobutanones detectable by 1H-NMR analysis of the crude mixtures. To the best of our knowledge, only one example of a metal-catalyzed direct cyclopropanol-cyclopentenone rearrangement was reported prior to our findings.5a,b</p><p>Our working mechanistic hypothesis to accomodate these results is presented in Scheme 2.7 We believe that, in the case of both silyl and electron-withdrawing substituents, the electronic properties of the system are presumably exacerbated upon coordination to the metal catalyst. Thus, the ability of silicon to stabilize a developing β-positive charge (Scheme 2, R = SiR3) and the propensity of ynones and propiolate derivatives to undergo Michael addition (Scheme 2, R = COR) probably favor a rapid, substrate-controlled 1,2-alkyl shift. It is worthy of note that the observed (Z)-selectivity in these cyclopropanol/cyclobutanone rearrangements, suggests that internal chelation of the putative vinylmetal intermediate by the cyclobutanone carbonyl is not operative.</p><p>On the other hand, the electron-"neutral" substrates studied (Table 3) should be more prone to metal insertion into a carbon-carbon bond of the cyclopropane moiety (Scheme 2, R = alkyl). Such a process would provide ruthenacyclohexenone 13, from which reductive elimination accounts for the observed products. The fact that only trace amounts of the analogous cyclobutanones are obtained implies that a net 1,2-alkyl shift is much less favoured in these systems.</p><p>In summary, we have developed a novel ruthenium-catalyzed ring-expansion of alkynylcyclopropanols. This atom-economical8 reaction appears to proceed by two different pathways. The unique ability of ruthenium to selectively mediate either of the two pathways depending on the electronic properties of the substrate bears testament to the versatile nature of this metal in catalysis. In particular, the ability to access functionalized β-substituted cyclopentenones through a direct two-carbon homologation is very appealing. Moreover, the exclusive obtention of the (Z)-alkylidene cyclobutanone isomers through the cyclopropanol/cyclobutanone expansion manifold is unprecedented and serves to further distinguish ruthenium from other, alkynophilic transition metals.</p><!><p>Experimental procedures and characterization data for all new compounds. This material is available free of charge via the Internet at http://pubs.acs.org.</p><!><p>Redox isomerization and proposal for a ruthenium-catalyzed ring-expansion of alkynylcyclopropanols</p><p>Mechanistic proposal for the dual ring-expansions</p><p>Ruthenium-catalyzed ring-expansion of silyl-substituted alkynylcyclopropanols</p><p>Geometry was assigned by analogy to the Z and E isomers 5a/6a: see Supporting Information for details.</p><p>Total yield of two isomers determined by 1H-NMR with mesitylene as internal standard.</p><p>Isolated yield. BDMS = benzyl(dimethyl)silyl.</p><p>Ruthenium-catalyzed ring-expansion of electron-deficient alkynylcyclopropanols</p><p>Olefin geometry was assigned based on 1H-NMR chemical shift (see Supporting Information for details).</p><p>Yields refer to pure, isolated products.</p><p>Ruthenium-catalyzed ring-expansion of alkyl-substituted alkynylcyclopropanols to cyclopentenones</p><p>Yields refer to pure, isolated products.</p>

PubMed Author Manuscript

Cancer Therapeutics Following Newton’s Third Law

Cancer is a wound that never heals. This is suggested by the data produced after several years of cancer research and therapeutic interventions done worldwide. There is a strong similarity between Newton’s third law and therapeutic behavior of tumor. According to Newton’s third law “for every action, there is an equal and opposite reaction”. In cancer therapeutics, tumor exerts strong pro-tumor response against applied treatment and imposes therapeutic resistance, one of the major problems seen in preclinical and clinical studies. There is an urgent need to understand the tumor biology of therapy resistant tumors following the therapy. Here, we have discussed the problem and provided possible path for future studies to treat cancer.

cancer_therapeutics_following_newton’s_third_law

Editorial

<p>There is no doubt that cancer is a smart entity, which is evident by several treatment failures in preclinical and clinical trials. In part, host genetic mutations are known to be responsible for the limited success [1]. However, in most of the cases, following therapy, tumors itself acquires resistant properties. In this case, tumor is initially sensitive to the applied treatment and evolves itself to counteract the anti-tumor effects of drug. On the other hand, de novo resistance is the part of primary refractoriness to a therapy that should have been effective based on the underlying biology or genetics. According to Newton's third law "for every action, there is an equal and opposite reaction". In cancer therapeutics, tumor exerts strong pro-tumor response against applied treatment. Most of the therapies with short and transient benefit (measured in weeks or months), have witnessed relapse of malignant tumor growth [2,3]. Resistant tumors are characterized by hyper-vascularity and hyper-invasiveness, for example; breast cancer and glioblastoma [4,5]. Tumor cells become quiescent by reprograming into mesenchymal phenotypes. In epithelial tumors, this phenomenon is well evident and known as epithelial to mesenchymal transition (EMT) [6,7]. In addition, resistant tumor secretes several immunomodulatory signals in the form of secreted factors such as chemokines and growth factors [8]. By doing so, tumor modulates immune cells and governs to impose pro-tumor properties. Heterogeneous population of tumor associated macrophages (TAMs) and regulatory cells are the most abundant pro-tumor and immune-suppressive immune cells known, which contribute to tumor recurrence following therapy [9]. Recently, much attention has been given to the bone marrow derived cells (BMDCs), which is the host component. Ample amount of data suggests that immune cells are derived from bone marrow compartment and resistant tumor recruits these immune cells on regular basis either at the time of tumor initiation, progression or metastasis at the distant organs [5]. Studies, including reports from our lab support that recruitment of bone marrow-derived myeloid cells, especially; myeloid derived suppressor cells (MDSCs) are critical in therapeutic resistance [10-14]. We believe that MDSCs acquire vasculogenic properties to provide vasculature support to the transiently shrink tumor. Our previous study illustrates that depleting bone marrow-derived myeloid cells through CSF1R blockade, significantly decreased recruitment of BMDCs from bone marrow to the tumor site. In addition, CSF1R blockade decreased tumor associated MDSCs and reduced tumor growth. Thus, targeting TAMs is crucial in avoiding therapeutic resistance [13,14]. At this point, we need more in-depth understanding of resistant tumors and their microenvironments through detailed mechanistic studies. We have to evolve our approaches to monitor such resistance at the earlier period of therapy and during the therapy.</p><p>In summary, most of the cancer therapies are limited by the development of drug resistance and to bypass this circumventing resistance is our priority in the era of personalized medicine. [15]. Our understanding of the cellular and molecular mechanism(s) of drug resistance has been increasing day-by-day. Discovering biomarkers of therapeutic resistance could be a good tool to find resistant patients, during the therapy. In addition, using new experimental approaches coupled with the systematic genomic and proteomic technologies would identify novel targets [12].</p>

PubMed Open Access

Identification of the Electronic and Structural Dynamics of Catalytic Centers in Single-Fe-Atom Material

The lack of model single-atom catalysts (SACs) and atomic-resolution operando spectroscopic techniques greatly limits our comprehension of the nature of catalysis. Herein, based on the designed model single-Fe-atom catalysts with well-controlled microenvironments, we have explored the exact structure of catalytic centers and provided insights into a spin-crossover-involved mechanism for oxygen reduction reaction (ORR) using operando Raman, X-ray absorption spectroscopies, and the developed operando 57 Fe Mo ¨ssbauer spectroscopy. In combination with theoretical studies, the N-FeN 4 C 10 moiety is evidenced as a more active site for ORR. Moreover, the potential-relevant dynamic cycles of both geometric structure and electronic configuration of reactive single-Fe-atom moieties are evidenced via capturing the peroxido (*O 2 À ) and hydroxyl (*OH À ) intermediates under in situ ORR conditions. We anticipate that the integration of operando techniques and SACs in this work shall shed some light on the electronic-level insight into the catalytic centers and underlying reaction mechanism.

identification_of_the_electronic_and_structural_dynamics_of_catalytic_centers_in_single-fe-atom_mate

INTRODUCTION<!>The Bigger Picture<!>RESULTS AND DISCUSSION<!>Electronic States and Coordination Environment of Fe in Single-Fe-Atom Catalysts<!>Electrocatalytic Activities of ORR<!>Article<!>Quantum Chemical Studies<!>Conclusions<!>EXPERIMENTAL PROCEDURES<!>Materials Availability<!>Data and Code Availability<!>Synthesis of Fe-ZIF and Fe-ZIF-S<!>Preparation of Graphene-Based Single-Fe-Atom Catalysts<!>Characterization<!>Electrochemical Measurements<!>SUPPLEMENTAL INFORMATION<!>DECLARATION OF INTERESTS

<p>The oxygen reduction reaction (ORR) is essential for clean and sustainable energy conversion and storage technologies, such as those used in fuel cells and metal-air batteries. [1][2][3][4] Consequently, the development of effective nonprecious-metalgroup catalysts shall offer promising solutions to fixing the twin challenges of the world's growing energy demand and climate change. [5][6][7][8] Recently, M-N-C materials (M = Fe, Co) have shown a promising catalytic activity and stability, approaching those of the state-of-the-art Pt/C catalysts. [9][10][11][12][13][14] However, confusion over the origin of ORR activity retards further developments of high-performance M-N-C catalysts. [15][16][17][18][19][20] Over the past few years, significant efforts have been made to elucidate the active sites of M-N-C catalysts, and the MN x C y moieties are generally accepted as the main contributors to the ORR activity. [21][22][23][24][25][26][27][28][29] However, the exact local structure of the reactive MN x C y moieties has yet to be determined, and no electronic-level insights into the evolution of spin states of the catalytic sites have been gained. To this end, developing a model catalyst as the studying platform as well as novel operando spectroscopic techniques with high atomic resolution are highly demanded to unambiguously reveal the nature of active sites and the underlying ORR mechanism.</p><!><p>Single-atom catalysts (SACs) build a conceptual bridge between homo-and heterogeneous catalysis. However, the lack of model SACs and atomicresolution operando spectroscopic techniques greatly limits our comprehension of the nature of catalysis. Herein, based on the newly designed model single-Fe-atom catalysts, we explored the exact structure of catalytic centers and provided a spin-crossover-involved mechanism for oxygen reduction reaction (ORR) using operando Raman, X-ray absorption spectroscopies, and the newly developed operando 57 Fe Mo ¨ssbauer spectroscopy. The potential-relevant electronic and structural dynamic cycles of active single-Fe-atom moieties were evidenced via capturing the *O 2 À and *OH À intermediates and further supported by theoretical calculations. These results provide a proof of concept for the integration of operando techniques and SACs, which may direct the way toward the electronic-level insight into the catalytic centers and reaction mechanism.</p><p>Heterogeneous single-atom catalysts (SACs) with a number of desirable features are emerging as the most attractive research frontier in various chemical reactions. [30][31][32][33][34][35] M-N-C SACs, composed of isolated active metal centers, can be viewed as the conceptual bridge to link homogeneous and heterogeneous catalysts and thus offer great opportunities for precisely revealing the stepwise elementary reaction mechanism. 10,[36][37][38][39][40][41][42] However, both synthesis of M-N-C SACs with definite N-coordination environment and identification of the exact structure of the active center are challenging. Moreover, due to the lack of practical operando spectroscopic techniques with atomic resolution, tracking the dynamic evolution of SACs during catalysis is still a long-term goal toward elucidating the catalytic mechanism.</p><p>Herein, we have synthesized single-Fe-atom catalysts with well-controlled site density and definite N-coordination environment via a solvent-assisted linker exchange (SALE) method as model systems for electronic-level understanding of the ORR mechanism. Operando 57 Fe Mo ¨ssbauer spectroscopy is developed for the first time to identify the exact structures and spin state of active atomically dispersed Fe moieties during ORR. By combining with operando Raman and X-ray absorption spectroscopy (XAS) measurements, as well as density functional theory (DFT) studies, the potential-relevant structural and electronic evolutions of the active single-Fe-atom moieties are evidenced via directly capturing the *O 2 À and *OH À intermediates under ORR conditions.</p><!><p>Structural Characterization of the Single-Fe-Atom Catalysts As schematically illustrated in Figure S1, single-Fe-atom catalysts (Fe-NC, Fe-NC-S0.2, Fe-NC-S0.4, and Fe-NC-S) were prepared by pyrolyzing mixtures of four N-coordinated Fe (Fe-ZIF) and six N-coordinated Fe precursors (Fe-ZIF-S) were obtained from the SALE process with different ratios. The X-ray diffraction (XRD) patterns of Fe-ZIF-S as compared with Fe-ZIF (Figure S2) provide a clear evidence of linker substitution of 2-methylimidazole by 1,2,3-triazole during SALE. Correspondingly, the coordination number of Fe to N in the precursor increased from four to six, as shown in the Rietveld refinement results (Figure S3). After pyrolysis, no observable metallic Fe diffraction peaks could be detected for the single-Featom catalysts (Figure S4), indicating that Fe species are highly dispersed as tiny clusters or single atoms, which are undetectable by XRD, scanning electron microscopy (SEM), scanning transmission electron microscopy (STEM), and high-resolution transmission electron microscopy (HRTEM) (Figure S5). The negligible change in the intensity ratio of D/G-band (I D /I G ) in Raman spectra (Figure S6) reveals the presence of similar disorientated graphene in the as-prepared catalysts. The dispersion of single Fe atoms on graphene was confirmed by aberration-corrected high-angle annular dark-field scanning transmission electron microscopy (HAADF-STEM). As shown in Figures 1A-1D and S5, the bright spots corresponding to single Fe atoms were observed well dispersed across the entire graphene framework in all single-Fe-atom catalysts. Moreover, the increasing apparentness of graphene, accompanied with a gradual increase of surface Fe-N content, as evidenced in high-resolution X-ray photoelectron spectroscopy (XPS) N 1s spectra (Figures 1E-1H, S7, and S8), suggested an increase in the single-Fe-atom site density with an increasing proportion of FeN 6 in the precursor. 43,44 As determined by inductively coupled plasma optical emission spectroscopy (ICP-OES) (Table S1), the atomic concentration of Fe in Fe-NC, Fe-NC-S0.2, Fe-NC-S0.4, and Fe-NC-S was 0.51, 0.93, 1.79, and 1.93 wt %, respectively.</p><!><p>The 57 Fe Mo ¨ssbauer spectroscopy, which is highly sensitive for probing the oxidation state, electron spin configuration, and coordination environment, [45][46][47] was carried out to determine the electronic states and coordination environment of Fe in single-Fe-atom catalysts. As shown in Figures 1I-1L, the Mo ¨ssbauer spectra could be well fitted by three doublets. Based on the values of isomer shift (IS) and quadrupole splitting (QS) (see Tables S2 and S3), 19,28 the green, blue, and red doublets could be S9) and spin-quantitated 2 K electron paramagnetic resonance (EPR) results (Figure S10) suggested that the SACs mainly contain ferrous species. Additionally, in situ XAS measurements further confirmed the assignment for D1. As shown in Figure S11, the Fe K-edge X-ray absorption near-edge spectroscopy (XANES) spectrum remained unchanged below the Fe 3+/2+ redox potential (0.5 V versus reversible hydrogen electrode [RHE]) in Ar-saturated 0.1 M KOH at room temperature, which further confirmed the absence of trivalent iron and the result is well consistent with the Mo ¨ssbauer assignment. No signals for iron carbides, iron oxides, or metallic iron were detectable, demonstrating the high purity of the as-prepared single-Fe-atom catalysts that contain only isolated Fe moieties with definite N-coordination environments. The much larger IS for D3 as compared with D1 and D2 suggested a larger 3d-electron density of HS Fe 2+ located out of N 4 -plane resulting from a fifth coordinating N atom with a longer Fe-N bond length. 28 The corresponding structures of the Fe moieties in the single-Fe-atom catalysts are presented in Figure 1M (optimized by the DFT method). 48 D3 contents were observed to increase gradually with an increase in the proportion of FeN 6 in the precursor (see Table S2), and as high as 24.9 % of D3 content could be obtained in Fe-NC-S.</p><p>X-ray absorption fine structure (XAFS) spectroscopy measurements were performed to further probe the possible bonding between the Fe atoms and the light elements (N/C). Figure 1N shows the normalized Fe K-edge XANES spectra. The Fe K-edge of Fe-NC-S was observed to be shifted by approximately 1.5 eV toward the lower energy when compared with that of Fe-NC. Considering that the single-Fe-atom catalysts contained only Fe 2+ (d 6 configuration), as evidenced by 57 Fe Mo ¨ssbauer results, the shift of K-edge to lower energy can be ascribed to an increase in the Fe-N bond length and delocalization of unpaired electrons in the high-lying 3d x 2 Ày 2 orbital of HS Fe 2+ with an increase in the content of N-Fe II N 4 C 10 moiety. Figure 1O displays the k 3 -weighted Fourier transform spectra of the Fe K-edge extended X-ray absorption fine structure (EXAFS). The Fe-Fe peak at 2.18 A ˚was not detected in either Fe-NC or Fe-NC-S, further confirming the HAADF-STEM and Mo ¨ssbauer results and suggesting the presence of only isolated Fe atoms on graphene. The main peak attributed to the Fe-N scattering path 38 was observed shifted from 1.47 to 1.53 A ˚for Fe-NC and Fe-NC-S, which is similar to the peak shift from 1.53 to 1.60 A ˚for FeN 4 and FeN 6 in the precursor. In addition, the relative peak intensity for Fe-NC and Fe-NC-S at 2.64 A ˚,24 which can be attributed to the Fe-C scattering path, was observed to reduce with increasing proportions of FeN 6 in the precursor. The increase of Fe-N bond length and the possible existence of N-Fe II N 4 C 10 moiety in the as-synthesized SACs were supported by the fitting results of the EXAFS spectra of Fe-NC and Fe-NC-S (see Figure S12; Table S4). The simulation results of Fe L 3 edge XANES spectra further evidenced the higher content of HS Fe 2+ in Fe-NC-S as compared with Fe-NC (Figure S13). Furthermore, the high-resolution XPS O 1s spectra (Figure S14), comparison of the Fe-N x contents (Figure S15), and nuclear resonance vibrational spectroscopy (NRVS) spectra (Figure S16) of the series single-Fe-atom catalysts suggest the axial ligand as nitrogen atom in D3 moiety 49 and these results are well consistent with the Mo ¨ssbauer and XAFS results.</p><!><p>The ORR activities of the single-Fe-atom catalysts were evaluated on rotating ringdisk electrode (RRDE) in O 2 -saturated acidic and alkaline electrolytes. Compared with NC and NC-S, Fe-NC and Fe-NC-S exhibited much higher ORR activity (Figure 2A), significantly more positive shift of the cyclic voltammogram (CV) curve ll Chem 6, 3440-3454, December 3, 2020 3443 Article (Figure 2B), and a smaller Tafel slope (Figure 2C), indicating that the activity of single-Fe-atom catalysts should most probably originate from the Fe moieties rather than from CN sites. In addition, the much lower H 2 O 2 yield for Fe-NC and Fe-NC-S indicated a four-electron ORR pathway over single-Fe-atom sites, which could be further confirmed by the Koutecky-Levich (K-L) plots obtained from the rotating disk electrode (RDE) polarization curves at different rotating speeds (Figure S17).</p><p>The half-wave potential (0.88 V versus RHE) of Fe-NC-S was much higher than that of Fe-NC and even higher than that of the commercial Pt/C catalyst (see Figures 2A and S18), which can be ascribed to the higher density of isolated Fe moieties. In addition, the mass activity per mass of metal for Fe-NC-S ($515 mA mg À1 metal ) was about 20 times higher than that for commercial Pt/C catalyst ($25 mA mg À1 metal ), suggesting the higher intrinsic activity of the single-atom Fe moieties (Figure S19). Moreover, the half-wave potential and current density at 0.9 V (versus RHE) for Fe-NC, Fe-NC-S0.2, Fe-NC-S0.4, and Fe-NC-S were observed to gradually increase with increasing D3 site content, while they showed random correlation with D1 and D2 site contents, Brunauer-Emmett-Teller (BET) surface area, and electrochemically active surface area (ECSA) (see Figures S20-S23; Table S5). The linear correlation between mass activity at 0.9 V (versus RHE) and D3 moiety in the single-Fe-atom catalysts indicated the critical role of N-FeN 4 C 10 site for highly active single-Fe-atom catalysts (Figure 2D). A similar variation trend of the ORR activity observed in the acidic medium (Figure S24) further supports this hypothesis. Strikingly, despite a gradual decrease in the turnover frequency (TOF) resulting from the increase in D3 content (Figure S21), the TOF for Fe-NC-S at 0.9 V versus RHE (1.20 e À site À1 s À1 ) was still 24 times higher than that for commercial Pt/C catalyst (0.05 e À site À1 s À1 ) (see Table S1).</p><!><p>Structural and Dynamic Cycles of Single-Fe-Atom Moieties in ORR To clarify the nature of active sites and the underlying ORR mechanism, operando 57 Fe Mo ¨ssbauer spectroscopy measurements were designed and performed to track the dynamic evolutions of single-Fe-atom moieties under ORR conditions in O 2saturated acidic and alkaline electrolytes (Figure S25). Since the catalytic reaction (ORR in this case) is a dynamic cycle process, 50 all intermediates at the initial, final, and rate-determining steps could be captured, while the reactive intermediate of the rate-determining step had a larger proportion. The final compounds obtained from 56 Fe and 57 Fe, named as 56 Fe-NC-S and 57 Fe-NC-S, almost had the same ORR performance as well as structure (Figure S26). As shown in Figures 3A-3C, when polarized at 0.9 V (versus RHE), the D3 content was observed to decrease with an increase in the relative D1 content, reflecting adsorption of O 2 on the D3 site with the generation of O -Fe II N 4 intermediates at 0.9 and 0.5 V (versus RHE), respectively.</p><p>Besides, the variation of contents of different Fe moieties and reactive intermediates, the variation of QS value, which reflects the electric field gradient tensor at the Fe nucleus, has been proven relevant to the adsorption of OH À on the axial position of the active sites during ORR. 51 As shown in Figure 4B, QS of D3 were observed to decrease ($0.2 mm s À1 ) in alkaline electrolyte without polarization, and further reduced with increasing bias, indicating the adsorption of OH À on the N-Fe II N 4 C 10 moiety before ORR and the accumulation of *OH À intermediates under operando ORR conditions. The decreased QS could be well recovered in 0.5 M H 2 SO 4 (see Figures 4B and S30). As compared with the D1 moiety (Figure S31), the full recovery of QS for the D3 moiety, reflecting the higher flexibility, further proves its higher ORR catalytic activity, which is also well consistent with the operando Mo ¨ssbauer results. Moreover, the same values of IS for Fe moieties and *OH À intermediates suggest recycling of both the geometric structure and electronic configuration of the single-Fe-atom during electrochemical oxygen reduction.</p><p>For the operando 57 Fe Mo ¨ssbauer measurements performed in acidic medium, negligible changes associated with the single-Fe-atom moieties were observed (Figure S32; Table S7). The undetectable *O 2 À intermediate is probably due to the adsorption of O 2 onto the single-Fe-atom sites as the RDS in acidic medium (Figure S33). Moreover, the comparison of actual data of Mo ¨ssbauer spectra excluding the fits further supports our results and discussions (Figure S34).</p><p>The dynamic evolutions of the geometric structure and electronic state of the single-Fe-atom catalyst can be further corroborated by operando XAS measurements (Figure S35). As shown in Figure S36, the first derivative of the spectra clearly shows that there is no change of the chemical state of Fe during a reaction, which is reasonable since the catalysts contain mainly Fe 2+ . In addition, no redox peak could be observed in CV curves collected for Fe-NC-S in both Ar-saturated 0.1 M KOH and 0.1 M HClO 4 electrolytes (Figure S37). The slight shift of the Fe K-edge XANES spectra to the lower energy in O 2 atmosphere compared with that in Ar-saturated 0.1 M KOH (Figure S11) is probably due to the change of spin state for Fe 2+ in D1 moiety from LS to HS. 52 Figure 4C displays the Fourier transforms of the EXAFS spectra of Fe-NC-S at various biases. The main peak at approximately 1.5 A ˚corresponding to the Fe-N bond length shifts to shorter lengths when polarized at 0.9 V (versus RHE), while it shifts to longer lengths when polarized at 0.7 and 0.5 V (versus RHE). The variation trend of the Fe-N bond length further confirms the adsorption of O 2 on D3 and D1 moieties at different biases, which is well consistent with the operando 57 Fe Mo ¨ssbauer results.</p><!><p>Quantum chemical studies were performed to provide theoretical insights into the structural and dynamic evolutions of atomically dispersed Fe moieties during ORR. 53 The Fe d-orbital splitting and spin crossover in different scenarios can be understood with the ligand field theory for pseudo-D 4h , -C 4v , and -O h symmetry. The complete structural and dynamic cycles of D3 moiety in ORR are shown in Figures 5A and S38; Table S8 5C-5D). On the contrary, the orbital interaction between O 2 and D1 moiety decreases the -orbital splitting, leading to the conversion of the Fe 2+ electronic configuration from LS to HS upon the formation of O 2 À -FeN 4 intermediate (Figure S40). For the orbital interactions between O 2 and D2 moiety, a slight decrease of Fe 3d orbital splitting will not result in the conversion of the electronic configuration of Fe 2+ (Figure S41). The dynamic evolution energy barriers for D1 and D2 moieties are much higher (Figure S42), which provides a theoretical evidence of D3 as the more active site. Moreover, the geometric structure and electronic configuration of single-Fe-atom in both D3 and D1 moieties can be well recovered with the final formation of adsorbed OH À (see Figure S43; Table S9). All these theoretical insights well explain the spectroscopic and reactivity results obtained from operando measurements.</p><!><p>In summary, single-Fe-atom catalysts with a well-controlled site density and definite N-coordination environment have been designed as model systems to probe the ORR mechanism at the electronic-level. Taking the advantages of operando 57 Fe Mo ¨ssbauer, Raman, and X-ray absorption spectroscopies, the potential-relevant, structural, and dynamic cycles of the active single-Fe-atom moieties are evidenced</p><!><p>Resource Availability Lead Contact Further information and requests for resources and reagents should be directed to and will be fulfilled by the Lead Contact, Bin Liu (liubin@ntu.edu.sg)</p><!><p>This study did not generate new unique reagents.</p><!><p>The published article includes all datasets generated or analyzed during this study.</p><p>Chemicals 2-methylimidazole (98 %), 1H-1,2,3-triazole (97 %), zinc nitrate hexahydrate (98 %), ferrous chloride (98 %), and Nafion 117 solution ($5 % in a mixture of low aliphatic alcohols and water) were purchased from Sigma-Aldrich and used without further purification.</p><!><p>To synthesize Fe-ZIF, 13.14 g of 2-methylimidazole was dissolved in 400 mL of methanol under vigorous stirring, followed by slowly adding 400 mL of methanol solution containing 5.80 g of Zn(NO 3 ) 2 $6H 2 O and 0.09 g of FeCl 2 $4H 2 O. Subsequently, the obtained colloidal solution was further stirred for another 60 min. The resulting precipitates were then centrifuged and washed at least three times with methanol and dried in an electric oven at 60 C for 20 h. ZIF-8 was prepared in the same process, without adding FeCl 2 $4H 2 O. Fe-ZIF-S0.2, Fe-ZIF-S0.4, and Fe-ZIF-S were prepared via a SALE process by adding 0.2, 0.4, and 1.5 mL of 1H-1,2,3-triazole into 80 mL of Fe-ZIF (1.0 g) suspended methanol, respectively. The resulting mixtures were further stirred for another 72 h. The nanocrystals were then centrifuged and washed with methanol repeatedly to remove unreacted linkers and dried in an electric oven at 60 C for 20 h.</p><!><p>Graphene-based single-Fe-atom catalysts were prepared by pyrolyzing mixtures of four N-coordinated Fe precursor (Fe-ZIF) and six N-coordinated Fe precursor (Fe-ZIF-S) obtained from the SALE process with different ratios. NC, NC-S, Fe-NC, Fe-NC-S0.2, Fe-NC-S0.4, and Fe-NC-S were obtained via a pyrolysis and carbonization process (N 2 atmosphere, 950 C, 2 h) of ZIF-8, ZIF-S, Fe-ZIF, Fe-ZIF-S0.2, Fe-ZIF-S0.4, and Fe-ZIF-S, respectively, followed by acid leaching in 2 M HCl at 80 C for 4 h, and then heated again at 800 C for 1 h in N 2 to recover the crystallinity of carbon. 57 Fe ll Chem 6, 3440-3454, December 3, 2020 3449</p><p>Article enriched Fe-NC-S was prepared via the same process with the replacement of FeCl 2 $4H 2 O by 57 Fe 2 O 3 dissolved in hydrochloric acid solution.</p><!><p>The crystal structure and morphology of the catalysts were examined by powder XRD, Bruker AXS D8 Advance) equipped with Cu Ka radiation (l = 0.15406 nm), HRTEM (JEM-2100), and field-emission scanning electron microscopy (FESEM, JSM 7800F). Atomic HAADF-STEM characterization was conducted on a probe corrected JEOL JEM-ARM200F STEM/TEM. The elemental compositions were analyzed by inductively coupled plasma atomic emission spectroscopy (ICP-OES).</p><p>Raman spectroscopy was performed in backscattering mode on a Renishaw Raman microscope using a 514.5 nm laser. The XPS spectra were recorded on the ESCALAB 250 X-ray photoelectron spectroscope and fitted by the XPSPEAK41 software using Shirley-type background. All spectra were fitted by four peaks with FWHM in the range from 1.5-2.0 eV and binding energies fixed at 398.2, 399.5, 400.5, and 401.3 eV. The surface area was determined by the BET method on a Micromeritics ASAP 2010 instrument at 77 K. XAS data, including XANES and EXAFS at Fe K-edge, were collected in total-fluorescence-yield mode at ambient air in BL-17C at National Synchrotron Radiation Research Center (NSRRC, Hsinchu, Taiwan), in which the electron storage ring was operated at 1.5 GeV with a beam current of 300 mA. The XANES measurements at the Fe L 3 -edge were recorded on the NSRRC BL20A high-energy spherical grating monochromator (HSGM) beamline in total electron-yield (TEY) mode. In the TEY mode, the sample drain current was measured. The photon energies were calibrated with an accuracy of 0.1 eV. The scan range was kept in an energy range of 7,000-7,700 eV for Fe K-edge. Subtracting the baseline of pre-edge and normalizing that of post-edge obtained the spectra. EXAFS analysis was conducted using Fourier transform on k 3 -weighted EXAFS oscillations to evaluate the contribution of each bond pair to Fourier transform peak. The room temperature 57 Fe Mo ¨ssbauer spectra and operando 57 Fe Mo ¨ssbauer measurements were carried out with a proportional counter and a Topologic 500A spectrometer with 57 Co (Rh) as a g-ray radioactive source. X-band EPR was performed on a Brucker spectrometer at 1 mW (power), 9.41 GHz (frequency), 2 G (modulation amplitude), and 100 kHz (modulation frequency). The spectra were acquired at 2 K and spin integrated using a 0.59 mg CuSO 4 $5H 2 O standard under identical instrumentation parameters. NRVS measurements were carried out at beamline 3-ID of the Advanced Photon Source at Argonne National Laboratory. NRVS scans were performed from À50 meV to +90 meV for E-E 0 with sample temperature in the range of 40-50 K.</p><!><p>The ORR performances were measured on a RRDE in O 2 -saturated KOH (0.1 M) or HClO 4 (0.1 M) electrolyte. A three-electrode cell configuration was employed with the working electrode of a glassy carbon RDE of 5 mm diameter and a saturated calomel reference electrode (SCE). Pt wire or graphite rod was employed as the counter electrode. Catalyst ink was prepared by dispersing 5.0 mg of catalyst in a water-ethanol solution (1:1, v/v) with 1 vol % of 5% Nafion 117 solution. 8 and 24 mL of the ink were pipetted on the glassy carbon disk to reach a catalyst loading of 0.2 and 0.6 mg cm À2 for the measurement in alkaline and acidic environment, respectively. RRDE measurements were conducted by LSV from 1.2 to 0 V (5 mV s where I disk and I ring are the voltammetric current at the disk and ring electrode, respectively. N is the RRDE collection efficiency, which was determined to be 0.26.</p><p>The TOF was estimated by 24,25 :</p><p>where J k represents the kinetic current density (A cm À2 ), F is the Faraday's constant 9.65 3 10 4 C mol À1 , is the electron number per coulomb 6.24 3 10 18 , u Fe is the active Fe content in the catalyst, C cat is the catalyst loading, N A is the Avogadro's constant 6.022 3 10 23 , M Fe and is the mass per mole of Fe (55.845 g mol À1 ).</p><p>Operando 57 Fe Mo ¨ssbauer Measurements Operando 57 Fe Mo ¨ssbauer measurements were carried out with a proportional counter and a Topologic 500A spectrometer with 57 Co (Rh) as a g-ray radioactive source. The detail of the designed operando Mo ¨ssbauer-electrochemical cell is shown in Figure S25. The operando measurements were carried out in O 2 -saturated alkaline (1 M KOH) or acidic (0.5 M H 2 SO 4 ) electrolyte. Six carbon-fiber electrodes with each active area of $1.0 cm 2 and an overall loading of $20 mg of 57 Fe-enriched catalysts parallelly connected was used as the working electrode. At each stage, the spectrum was recorded for more than 12 h under operando ORR condition at various biases.</p><p>Operando Raman Measurements Operando Raman spectra were collected under controlled electrochemical potentials in O 2 -saturated alkaline (1 M KOH) electrolyte using a three-electrode epoxy pool cell with a counter electrode of Pt wire and a SCE. A controlled active area of 0.384 cm 2 by an insulation layer on fluorine-doped tin dioxide (FTO) drop-casted with 50 mL catalyst ink was used as the working electrode (Figure S27). Raman spectra were collected using a Raman spectrometer (Horiba-JY HR800) by a 532 nm excitation laser (25%) having a power of 2.1 mW measured at the objective. A 50x long working distance objective (Olympus, 0.5 NA) was used, focusing on the sample surface and avoiding the contact to the electrolyte. Before the backscattered light entered CCD, a 532 nm notch filter was added to eliminate the laser beam. Acquisition time was set as 10 s for the spectral Raman shift ranging from 800 to 1,800 cm À1 window using the 1,800 g/mm grating. Operando X-ray Absorption Measurements Operando XAS measurements (Figure S35) were collected in total-fluorescenceyield mode at room temperature using BL-17C at National Synchrotron Radiation Research Center (NSRRC), Taiwan. A homemade cell with a window sealed by Kepton tape was designed for the operando experiments, which were operated under identical conditions as the electrochemical measurements. XAS data were acquired with energy-calibrated by simultaneously recorded transmission spectra of the Fe foil, where the energy of the first inflection point for the reference sample absorption edge was 7,112 eV. For the operando measurements, the cell was filled with electrolyte (0.1 M KOH). SCE and Pt were used as the reference and counter electrode, respectively. A carbon-fiber electrode with an active area of $1.0 cm 2 and an overall loading with about $5 mg of catalyst was used as the working electrode. Operando ll Chem 6, 3440-3454, December 3, 2020 3451 Article XANES and EXAFS spectra were recorded at an open-circuit voltage (OCV), 0.9, 0.7, 0.5 V (versus RHE), and after ORR (AFT) conditions.</p><p>Theoretical and Computational Framework Details Geometry optimizations were performed with the Perdew-Burke-Ernzerhof (PBE) exchange-correlation functional of generalized gradient approximation, 54 as implemented in the Amsterdam Density Functional Program (ADF 2017.113). The Slatertype orbital (STO) basis sets of triple-zeta quality plus two polarization functions (TZ2P) were used for computing the final energies, while for C the double-zeta quality (DZP) basis set was adopted in optimizing the geometries for the sake of computational cost. The 1s-2p shells of Fe and 1s shells of C, N, and O were treated with frozen core approximation. 55 The scalar relativistic (SR) effects were taken into consideration by the zero th -order regular approximation (ZORA). 56 The natural bond orbital analysis was carried out with NBO6.0 in ADF. [57][58][59] A pyridine molecule was utilized to model the out-of-plane coordinating nitrogen, as suggested in a previous study. 29 In the optimization of D3 configuration with adsorbed *OH À , the pyridine group was constrained to constrain the distances between nitrogen and iron as those obtained in the bare cluster with intermediate spin polarization, based on the assumption that the adsorption of OH À in solution only slightly perturbs the catalytic center, while the geometries of all the other D3 species with adsorbates were optimized fully optimized.</p><p>For the electrochemical steps, we computed the reaction free energies by referencing to the electrochemical reaction, 60 following the computational hydrogen electrode (CHE) model in a manner proposed by Nørskov. 61 H 2 O + 2e À /2OH À + H 2</p><p>The zero-point energy (ZPE), entropy, and enthalpy corrections were taken from previous work. 62 While those for the step from *OH À to *O 2 were obtained from our DFT vibrational frequency calculations.</p><p>A detailed description of the methodology for the calculation of Mo ¨ssbauer parameters can be found in the Supplemental Experimental Procedures.</p><!><p>Supplemental Information can be found online at https://doi.org/10.1016/j.chempr. 2020.10.027.</p><!><p>The authors declare no competing financial interests.</p>

Chem Cell

Advances in the application of comprehensive two-dimensional gas chromatography in metabolomics

Due to excellent separation capacity for complex mixtures of chemicals, comprehensive two-dimensional gas chromatography (GC \xc3\x97 GC) is being utilized with increasing frequency for metabolomics analyses. This review describes recent advances in GC \xc3\x97 GC method development for metabolomics, organismal sampling techniques compatible with GC \xc3\x97 GC, metabolomic discoveries made using GC \xc3\x97 GC, and recommendations and best practices for collecting and reporting GC \xc3\x97 GC metabolomics data.

advances_in_the_application_of_comprehensive_two-dimensional_gas_chromatography_in_metabolomics

Introduction<!><!>Introduction<!>Method development for GC \xc3\x97 GC metabolomics<!>Sampling and sample preparation<!>Instrumentation<!>Data processing, analysis, and visualization<!>Applications of GC \xc3\x97 GC in metabolomics<!>In vitro analyses<!>Animal models of human disease<!>Human biospecimens<!>Non-human animals<!>Plants<!>Moving toward metabolic mechanisms<!>Recommendations and best practices<!>Clinical design and confounders<!>Chromatography<!>Sample handling and batch effects<!>Statistical analyses and overfitting<!>Concluding remarks

<p>Metabolism, which is the sum of chemical reactions of an organism, can be investigated at multiple scales, from a singular biochemical reaction, to metabolic pathways, to cellular, multicellular, tissue, organism, and population-scale analyses. As a part of the functional genome [1], metabolic analyses shed light on the translation of genes, transcriptomes, and proteomes to phenotypes, and the influence of the environment on this process. Characterizations of changes in the metabolome as a function of an external or internal perturbation are important for understanding how development, disease, diet, toxins, medications, stress, the microbiome, etc. govern living systems, and metabolome studies are therefore relevant to a broad range of the basic biological sciences (Fig. 1). Metabolome data are also useful in the applied sciences and industry, and consequently are of high economic importance; for instance, metabolome data play a central role in the discovery of new pharmaceutical targets and diagnostic biomarkers, in the production of fermented foods and beverages, and in the development of novel biosynthetic pathways and bioremediation strategies (Fig. 1).</p><!><p>Target analysis to measure the substrate or product of an enzyme or group of enzymes,</p><p>Metabolite profiling to identify and/or quantify a class of metabolites (e.g., fatty acid methyl esters (FAMEs)),</p><p>Metabolic fingerprinting to rapidly classify samples, where individual metabolites are not identified (e.g., via direct-injection mass spectral methods [3]), and</p><p>Metabolomics to comprehensively analyze the metabolome (or large fractions thereof), including identifying and quantitating individual metabolites.</p><!><p>Metabolomics aims to universally detect, characterize, and quantify all metabolites in a biological system [4], but of all of the 'omics approaches (i.e., genomics, transcriptomics, proteomics), metabolomics is the most analytically challenging. Like mRNA transcripts and proteins, metabolites can be present in hugely disparate concentrations, from single molecules to mole fractions, and the absolute and relative concentrations are context specific. However, unlike nucleic acids and proteins, made up of combinations of 4 and 22 chemical moieties, respectively, the metabolome contains thousands to hundreds of thousands of unique chemical species [5]. No single analytical platform can separate and detect all metabolites in a specimen, and to-date, even in the extensively studied human metabolome that is predicted to contain over 114,000 total metabolites, more than 80% are yet to be detected [5]. The key to advancing the field of metabolomics is developing the analytical tools to detect, identify, and quantify unknown metabolites, the software tools to manage and process large quantities of raw metabolomics data, and the chemometric tools to extract information from the data [2, 5].</p><p>Due to excellent separation capacity for complex mixtures of chemicals, comprehensive two-dimensional gas chromatography (GC × GC) is being utilized with increasing frequency for metabolite profiling and metabolomics analyses [4]. Typically, when compared to one dimensional gas chromatography (GC), three to ten-fold more peaks are detectable using GC × GC [4], and therefore, GC × GC metabolomics is rapidly increasing the metabolic catalogs for microbes [6, 7], plants [8], animals [9], and humans [10]. Here, we review recent advances in GC × GC method development for metabolomics, organismal sampling techniques compatible with GC × GC, and a selection of GC × GC metabolomic applications and discoveries that, in our opinion, will push the boundaries of their fields. Additionally, we provide recommendations and best practices for collecting and reporting GC × GC metabolomics data and perspectives on the future directions of GC × GC in metabolomics. An excellent review of GC × GC and metabolomics was published by Almstetter, et al. in 2012 [4], so we have focused our efforts on reviewing studies that have been published since.</p><!><p>There have been significant efforts and advancements in creating robust methods for generalized GC × GC analyses. However, there is still great need for the development of methods specialized for metabolomics, particularly validated protocols for specimen preparation, sampling, data collection, and data processing. Other reviews in this Special Issue summarize advancements in modulators, stationary phases, mass spectrometry, and novel instrumentation, but a few studies that are specifically focused on methods for metabolomics are reviewed here.</p><!><p>Analytical robustness in metabolomics is significantly impacted by uneven extraction efficiencies across chemical families as well as sample inhomogeneity in solid and semi-solid specimens. Therefore optimized methods for sampling and sample preparation are critically important. Several recent studies explored the impact of sample preparation on the GC × GC metabolomes of tissues and viscous biofluids, and provide recommendations for obtaining more comprehensive and reproducible metabolomes.</p><p>The metabolic profiling of organs and tissues has been explored as a strategy to obtain a variety of information about human health, however, the amount of blood in the samples can distort the information gained from these approaches. To explore this, Ly-Verdú, et al. used GC × GC with time-of-flight mass spectrometry (TOFMS) to study the effects that phosphate buffered saline (PBS) perfusion may have on the metabolite composition of the liver, and whether or not perfusion may constitute an essential experimental step in liver profiling [11]. Livers were collected from healthy male mice, and were either perfused with PBS or unperfused prior to harvesting, homogenization, extraction, derivatization, and GC × GC analysis. Results following multivariate analysis revealed more than 35 metabolites significantly differed between the profiles of unperfused and perfused livers. The authors observed that the GC × GC metabolomes of perfused livers were slightly less variable and concluded that the presence of blood metabolites can interfere with interpreting liver-specific metabolism in some cases. As a result, the authors suggest that the choice to perfuse organ and tissue samples must be carefully considered in the context of each study hypothesis since the metabolome will be influenced by the presence or absence of blood.</p><p>Sputum is an oft-used specimen for lung metabolomics analyses, but its high viscosity and inhomogeneity can impact the recovery and reproducibility of metabolites. To determine the best pretreatment method for sputum prior to chloroform/methanol/water extraction, derivatization, and GC × GC analyses, Schoeman, du Preez, and Loots compared four protocols using sputum spiked with Mycobacterium tuberculosis, the causative agent of tuberculosis (TB) [12]. The four methods evaluated included incubations with 1) a 1:1 v/v ratio of sputum and Sputolysin (a concentrate of dithiothreitol in phosphate buffer), 2) a 1:1 v/v ratio of sputum and 0.5 N NaOH with 20% w/v N-acetyl-L-cysteine, 3) a 1:2 v/v ratio of sputum and 1 N NaOH, and 4) a simple homogenization of sputum with 45% ethanol in a 1:2 v/v ratio. In the first three methods the pretreated mixtures were centrifuged and the cell pellets harvested for further extraction, and in the fourth method the entire homogenate was retained and dried prior to CHCl3/CH3OH/H2O (1:3:1) extraction and silylation. By analyzing the extraction efficiency (i.e., number and intensity of compounds extracted), repeatability, limits of detection (LOD), and the predictive accuracy of biomarkers selected from the GC × GC metabolomes, they determined ethanol homogenization is the superior pretreatment method, which allowed them to identify 19 metabolic biomarkers of TB using only 250 μL sputum. While it is not surprising that the ethanol extraction method, which retains the entire sputum sample, produces the greatest number and concentration of metabolites (roughly 80% more than Sputolysin, in second place), it is interesting to note that it also generated metabolite profiles that were highly discriminatory between the M. tuberculosis spiked sputum vs. unspiked controls and these metabolites correctly classified TB-positive vs. TB-negative patient samples. These data suggest that the secreted M. tuberculosis metabolites serve as useful biomarkers, not just the intracellular metabolites, which may be the key to diagnosing the presence of TB disease using sputum specimens with typically low bacterial cell densities (< 105 cells/mL).</p><p>The success of untargeted metabolomics studies that utilize relative quantitative data (e.g., multi-marker studies, biomarker panel discoveries) relies upon the ability to reproducibly and quantitatively extract a wide variety of metabolites while mitigating matrix effects. Pérez Vasquez, Crosnier de bellaistre-Bonose, et al. developed a novel double extraction method to capture additional urine metabolites, and analyzed the derivatized compounds by GC × GC-qMS [13]. The first urine extraction was modified from a commonly-used procedure wherein urea is removed via urease incubation, then the organic acids are extracted via a liquid-liquid extraction with ethyl acetate. They performed the second extraction on the remaining aqueous phase, which was first incubated with triethylamine at pH 9, then extracted recursively with tetrahydrofuran. The organic phases from both steps were pooled and silylated for analysis. This time-intensive sample preparation protocol combined with greater peak resolution achieved by GC × GC-qMS facilitated the detection of 92 additional compounds in urine compared to a commonly-used sample preparation method and GC-MS analysis. The validated GC × GC method is used in their hospital to analyze urine samples of children with neurological disorders of unknown origin, and the authors posit that their approach may be adaptable for metabolic profiling of other body fluids, such as cerebrospinal fluid, saliva, or breath condensates.</p><p>Marney, Synovec, and colleagues explored how the ratio of sample mass to solvent volume impacts the extraction efficiency of soluble metabolites from mouse heart tissues by measuring the GC × GC-TOFMS signal intensity of eight representative metabolites: fumarate, malate, glutamate, citrate, succinyl-CoA, myo-inosotol, glycerol-3-phosphate, and glycerol [14]. By recursive extraction experiments on 40 mg tissue in 1 mL solvent (3:1:1 v/v/v CHCl3/CH3OH/H2O), they determined that five of the metabolites were quantitatively extracted on the first round, while fumarate, glycerol, and citrate required four to five extractions to achieve quantitative results. To determine a ratio of tissue mass to solvent volume that would yield more universally quantitative extraction in a single step, they measured the linearity and reproducibility of each metabolite when using 2 mL solvent to extract tissue at four masses ranging from 5 – 40 mg. They found that a 20 mg sample provided an average relative standard deviation (RSD) of 20-30% in their metabolomics analyses, which was sufficiently low to detect relevant metabolic changes in their experiments. These data show that efforts to optimize extraction efficiency and enhance reproducibility prior to specimen analysis will yield higher quality relative quantitation data when extraction is a significant source of variation in an experiment.</p><p>Uniform extraction efficiency is also a concern for volatile metabolomics analyses in which a sorbent is used for sampling. Solid phase microextraction (SPME) has become an essential gas phase sampling technique, and several sorbents are commercially available to optimize sampling for each investigation. Purcaro, et al. conducted an analysis of five different SPME fibers consisting of combinations of three sorbents – divinylbenzene (DVB), carboxen (CAR), and polydimethylsiloxane (PDMS) – to determine the best fiber and sampling conditions for analyzing the volatile metabolites of cell cultures infected with human rhinovirus [15]. Based on the normalized peak areas of 12 volatile and semi-volatile standards extracted from the cell culture media at 43°C for 30 min, they identified the DVB/CAR/PDMS triphase fiber as the best option for their analyses as it yielded the highest chromatographic peak areas. They further optimized their sampling method by using a central composite design and response surface modeling to identify the best time (15 – 45 min) and temperature (37 – 50°C) combination to yield the highest peak intensities for each individual standard. While six of their standards were modeled to produce the highest peak areas when sampled at 43°C for 30 min, the other six standards did not produce peak area maxima, and therefore quantitative sampling may not be achievable under the experimental parameters that were tested. Because SPME sampling is often performed in parallel to GC – GC analysis, there are practical limits to the length of the sampling period, which are usually limited by the duration of the GC – GC runtime. Therefore, unequal extraction rates are a significant concern for relative quantitation by SPME. Another factor that can impact quantitation is interanalyte displacement, which has been of persistent concern for DVB/CAR/PDMS triphase fibers, and could have played a role in the differences in optimal sampling conditions for the 12 standards used by Purcaro and colleagues. To determine the extent of this problem among SPME fibers, Risticevic and Pawliszyn analyzed the performance of seven commercial phases, measuring analyte extraction efficiency and sensitivity, desorption carryover, linear dynamic range, and interanalyte displacement by performing headspace (HS)-SPME on apple homogenates and analyzing the volatile profiles using GC – GC-TOFMS [16]. The DVB/CAR/PDMS triphase coating outperformed other phases on extraction efficiency, as reported by Purcaro and colleagues [15], but were also more prone to carryover and interanalyte displacement of a subset of metabolites in the apples. However, decreasing the extraction time significantly improved both issues, with the tradeoff of increasing the LOD for some analytes.</p><p>The physical properties of samples (e.g., ionic strength) also influence the number and concentration of volatile metabolites that are detected, and salt is routinely used to increase the partitioning of semi-volatile compounds from the liquid phase to the headspace. In a comprehensive analysis of urine volatile metabolomes by HS-SPME and GC × GC-TOFMS, Rocha, et al. considered the influence of pH on metabolite detection [17]. When comparing chromatograms of aliquots of the same urine sample at pH 5.8 (physiological), pH 2.0, and pH 12.0, more than 40% of the approximately 700 GC × GC peaks could be tentatively identified in the pH 2.0 and pH 12.0 samples, whereas only 163 compounds were identified in the physiological sample (pH 5.8). The highest chromatographic area and compound numbers were obtained under acidic conditions; therefore Rocha, et al. concluded untargeted urine volatile metabolomics should be performed at pH 2.0. However, they also noted that targeted analysis or metabolite profiling might be more appropriate at a higher pH, depending on the metabolites of interest.</p><!><p>The majority of GC×GC analyses – metabolomics analyses included – are conducted using cryomodulation, which generates peak widths on the order of 100 ms. The narrowness of the peaks necessitates using MS detectors that can collect full scan spectra at a rate of 100 Hz to facilitate accurate peak quantitation and deconvolution. TOFMS is the most common method of ion separation used with GC × GC, capable of full spectrum collection rates up to 500 Hz, but these instruments are expensive, which limits accessibility. Compared to TOFMS, quadrupole mass spectrometers (qMS) are comparatively inexpensive, generally have a smaller footprint, and provide lower LODs via selected ion monitoring (SIM). However, typical "fast" qMS instruments have maximum acquisition rates of 20,000 amu/s, and therefore the mass spectral scan range will be restricted to 200 amu to meet the minimum scan rate for cryomodulation, which is often too narrow for metabolomics analyses, but can be sufficient for metabolic profiling.</p><p>GC × GC flow modulation is gaining in popularity and market share due to the significant advantage that it reduces consumable costs by forgoing the need for cryogens. Flow modulation produces broader peaks than cryomodulation, which reduces peak capacity and increases LODs, but the wider peaks are more compatible with qMS detectors. Tranchida, Mondello, and colleagues utilized qMS in a study to optimize a flow-modulated GC × GC method for the metabolic profiling of FAMEs [18]. After optimizing column diameters, gas flows, temperature programming, and modulation periods, they identified FAMEs in fish oil and human serum with limits of identification in the range of 100-200 pg on column, and limits of quantification (LOQ) as low as 3.4 pg in SIM mode. These results demonstrate that GC × GC-qMS is well suited for metabolite profiling, using only a few microliters of biofluid or micrograms of cells for analysis. Weinert, et al. set out to optimize GC × GC equipped with a fast-scanning qMS detector for large-scale untargeted metabolomics, and compared their results to TOFMS [19]. Their GC × GC-qMS method provided good separation in under an hour for 90% of the urine analytes detected by TOFMS. Scanning the range of m/z 60 – 500 at the maximum rate on the qMS (20,000 amu/s), they typically obtained 7 - 9 data points per peak above 10% peak height, providing good peak area and height precision (2.7 % and 2.4 % mean RSD, respectively). A potential concern with qMS is mass spectral skewing, which negatively impacts mass spectral library matching and peak alignment across chromatograms. Weinert, et al. quantitated skewing by a variety of metrics, observing 15 % mean RSD for apex spectra relative intensities (range 6.0 – 29.8 %) when they included trace-level peaks, and 10 % RSD (range 5.9 – 21.6 %) when trace peaks were excluded. While skewing was not insignificant, the quality and reproducibility of the apex spectra was sufficient for aligning the majority of peaks across their samples.</p><p>The application of high resolution (HR) mass spectrometry to GC × GC metabolomics is in its infancy, representing only three percent of the metabolomics studies published since 2012 (Appendix Table) [20-22]. The impact of using HRMS is greatest for untargeted metabolomics, where the accurate mass data provides molecular formulae for unknown compounds. However, their use remains niche because GC × GC-HRMS instruments are expensive, which precludes them from being purchased by most independent investigators. Further, the high resolution analyses generate very large data files, which makes them less amenable than nominal mass detectors for large scale comparative metabolomics studies for biomarker identification. As GC × GC metabolomics studies mature to the point of confirming the chemical identities of metabolites that were putatively identified in nominal mass analyses, the logical next step will be to obtain accurate mass data, and with that the proportion of publications that include GC × GC-HRMS data will increase.</p><p>Compared to MS detectors, vacuum ultraviolet absorption spectroscopy (VUVAS) has two significant hardware advantages: a small footprint and a lack of intensive vacuum requirements. Gruber, Groeger, et al. used a cryomodulated GC × GC-VUVAS to analyze four breath samples from an individual before and during a glucose challenge [23]. Results showed that detection with VUVAS, with selective monitoring for aromatics, provided similar performance to GC × GC-TOFMS, and gave good detection for small-oxygenated volatile metabolites (e.g. alcohols and ketones).</p><!><p>The ultimate goal of any GC × GC metabolomics analysis is to turn the data collected into chemical and biological information, which is strongly dependent upon reliable methods for processing, analyzing, and visualizing the data. The development of methods for GC × GC data processing and analysis is a rapidly-growing area, including novel approaches designed for metabolomics, or validated using metabolomics data [24-31]. Because this important topic is outside the authors' area of expertise, we refer readers to the GC × GC chemometrics review in this Special Issue and other reviews [32] for details on recent advancements and recommendations in data processing.</p><!><p>Due to the complexity of the metabolome and the heterogeneity that exists within and between organisms, many metabolomics studies are begun using reductionist models (e.g., cell culture), and then may graduate to more complex model systems (e.g., animal models), biospecimens (e.g., urine, blood, tissue), and ultimately, living organisms in natural and artificial environments. While the in vitro experiments may lack direct translation to organismal-level metabolism in native environments, they do provide important information on fundamentals of metabolism, with broad accessibility and low costs (Fig. 2). In this section we highlight applications of GC × GC to in vitro cultures, analysis of biospecimens, and organisms, and we review studies that used interesting biological and analytical designs to investigate the underlying mechanisms of metabolism and the roles metabolites play in multitrophic interactions. The handful of studies we review in this section were selected to demonstrate how GC × GC metabolomics studies can facilitate discoveries and push the boundaries of their fields. A more comprehensive list of GC × GC metabolomics studies published between the end of 2011 and June 2018 is available in Appendix Table.</p><!><p>The recent implementation of GC × GC for untargeted metabolomics of bacterial cultures has vastly expanded the volatile metabolome (or "volatilome") catalog for human pathogens, which have been studied for decades using GC-MS. Bean, Dimandja, and Hill pioneered the use of GC × GC for untargeted bacterial volatile metabolomics with a characterization of the volatilome of Pseudomonas aeruginosa strain PA14, detecting 56 chromatographic peaks associated with the bacterium, which nearly doubled the published volatilome of this well-studied organism [33]. The ability to detect more chemical diversity in in vitro samples via GC × GC has facilitated the exploration of the biological diversity within species, and underscores the degree to which study design impacts the volatilome. In order to investigate strain-to-strain diversity, Bean, Rees, and Hill used GC × GC to compare the volatilomes of 24 clinical isolates of P. aeruginosa [7]. They were able to detect 391 chromatographic peaks associated with P. aeruginosa, of which only 70 volatiles were detected in all 24 isolates, termed the core volatilome. Using accumulation and rarefaction curves of the pan-volatilome and core volatilome, respectively, they showed that they analyzed a sufficient number of samples to capture the volatilome diversity of P. aeruginosa clinical isolates under the studied conditions. Their curves also show that to approximate the core metabolome (with a 50% inflation in its size), a median of 12 and minimum of three isolates were required, and the pan-volatilome – or the collection of all volatile metabolites produced – required a median of 14 and minimum of six isolates to cover 95% of the metabolome. These data demonstrate that defining the metabolome of a species based on a single specimen (or a small collection of specimens) is likely to be misleading.</p><p>Growth conditions can also significantly influence the microbial metabolome in in vitro analyses. In a study of nine clinical Klebsiella pneumonia isolates grown in four rich media (lysogeny broth, brain heart infusion, Mueller-Hinton broth, and tryptic soy broth), a total of 365 K.pneumoniae-associated volatiles were detected by GC × GC-TOFMS, of which only 10% were conserved across all media [34]. Using principle components analysis (PCA) of the volatilomes, Rees, Hill, and colleagues showed that the bacterial samples clustered based on their growth medium and not bacterial strain. This finding was true even when only the 36 volatiles that were conserved across all four media were used as variables in the PCA. Therefore, the volatilome of K. pneumoniae is strongly dependent on the growth medium used, and the authors conclude that the choice of medium should be carefully considered in microbial metabolomics studies. Together Rees's [34] and Bean's [7] findings underscore the challenge in capturing the essence of an organism's metabolome with a single set of experiments, much less identifying in vitro growth conditions that can robustly mimic the in vivo infection environment. However, these experiments are still useful; the more variations in in vitro growth conditions we explore, the more we can understand the broad metabolic capabilities of individual organisms.</p><!><p>Primates, pigs, mice, and rats are used extensively in biomedical research to model human diseases and treatments, and the use of GC × GC to measure metabolic changes in these model systems is rapidly expanding. Juul and colleagues have used a primate model and GC × GC-TOFMS to investigate metabolic changes of the fetal-to-neonatal transition in healthy [35] and diseased animals [36]. To establish the healthy metabolome, six late-preterm Macaca nemestrina were delivered via hysterotomy, with plasma drawn from cord blood and eight additional post-birth time points through 72 h of age. A total of 100 metabolites were identified, of which 23 exhibited significant changes in concentration over the 72 h sampling period and were categorized by their association with signaling pathways, glucose metabolism, carbohydrates, and amino acids [35]. Beckstrom, et al. proposed that these metabolites could be used as baseline markers of normal birth transition in future perinatal metabolomics research. Chun, Juul, and colleagues built upon that hypothesis by utilizing the M. nemestrina primate model to investigate the plasma metabolome of hypoxic ischemic encephalopathy (HIE), a common complication of birth that can lead to early and/or long-term neurodevelopmental consequences, including cerebral palsy or death [36]. They used GC × GC-TOFMS to analyze blood samples from 33 macaques that were exposed to 0, 15, or 18 minutes of in utero umbilical cord occlusion to induce HIE. They treated a subset of the animals by two methods, hypothermia or hypothermia + erythropoietin, and obtained serial blood samples at baseline, 0.1,24, 48, and 72 h after hysterotomy. They identified twelve potential biomarkers of HIE that showed statistically-significant differences between the diseased and control animal groups. By collecting neurodevelopmental data of the macaques up to nine months of age, they identified eight metabolites that were correlated to early and/or long-term outcomes, and four metabolites (citric acid, fumaric acid, lactic acid, and propanoic acid) that predicted death or cerebral palsy.</p><p>Mellors, Hill, and colleagues posited that macaques would also be excellent models for identifying breath biomarkers of TB for novel diagnostics in humans [37]. In a feasibility study, they used GC × GC-TOFMS to analyze breath from three cynomolgus macaques (M. fascicularis) and two rhesus macaques (M. mulatta) before and one to two months after M. tuberculosis infection. Using random forest (RF) analysis, they identified 49 compounds – represented strongly (65%) by hydrocarbons – that significantly changed during the course of infection. They demonstrated that breath sampling and analysis is feasible in animal models, and that breath metabolites can serve as useful markers of infection. The fact that the same animal models are being used in diverse GC × GC metabolomics studies (e.g., the three macaque studies described here [35-37]), and that metabolomes are being compared across model systems (e.g., primate, murine [38], and cell culture [39] models of TB) and with human specimens [12, 40-42], a more comprehensive view of the animal models' applicability to human diseases can be built.</p><!><p>Because blood, serum, and plasma carry metabolites from all parts of the body and are routinely collected in a clinical setting, they are excellent biofluids for metabolomic analyses and the identification of biomarkers of disease. Winnike, Zhang, et al. compared the utility of GC-TOFMS and GC × GC-TOFMS in metabolic biomarker quantitation using pooled serum samples from 109 individuals, 54 of whom have a chronic neurodegenerative disorder [43]. When comparing metabolomic profiles between the healthy and unwell subject groups, 23 compounds detected by GC had statistically significant differences, compared to 34 detected using GC × GC. Similar advantages for metabolite detection were observed by Menéndez-Carreño et al., who developed and validated a method using GC × GC-TOFMS for phytosterol oxidation products (POPs) in human plasma [44]. Eleven POPs were spiked into human plasma samples to validate the detection method. The LODs and LOQs of GC × GC-TOFMS were found to be approximately 10-fold lower compared with GC-MS. In addition to the 11 known POPs, GC × GC facilitated the identification and quantitation of unsaturated brassicasterol and stigmasterol, reported in human plasma for the first time.</p><p>Like blood, urine is a rich source of metabolites from the entire body, and bears some significant advantages for biomarker research since it is plentiful and able to be collected non-invasively. Zhang, Brenna, and co-workers published a pair of studies in which they used GC × GC-qMS with positive chemical ionization (PCI) to detect complex steroid mixtures in urine of subjects on therapeutic steroid treatment [45] and of human athletes [46]. The steroids were extracted from urine, derivatized, and analyzed using GC × GC-qMS using either electron impact ionization (EI), CH4 PCI, or NH3 PCI.</p><p>Ionization with NH3 preserved structure-specific ions and the combination with GC × GC facilitated the identification of endogenous target steroids at physiological concentrations. Additionally, their results indicate that chromatographic structure provided by GC × GC may facilitate the detection of novel designer steroids in urine by anti-doping agencies. Luies and Loots measured urine metabolites to investigate host-pathogen interactions during active TB disease [41]. They extracted and derivatized the organic acid fraction of urine metabolites from 76 subjects: 30 TB-negative, and 46 with active TB. Using a multi-statistical approach on the 507 compounds they detected by GC × GC, they identified 12 metabolite markers in urine that could be used to distinguish the presence or absence of TB. The majority of the metabolic markers they discovered could be explained by changes in the host metabolome due to infection from M. tuberculosis. In particular, host fatty acid and aromatic amino acid metabolism is perturbed by infection, providing insights into symptom management and treatment.</p><p>Breath can be considered the headspace of the blood, and like urine, is plentiful and relatively easy to collect. Therefore, breath metabolomics is attractive for the development of sensitive, non-invasive diagnostics for a plethora of human ailments. To expand the catalog of the human breath volatilome, Phillips, et al. used GC × GC-TOFMS to analyze breath samples from 34 healthy individuals [10]. They detected approximately 2000 volatile metabolites, including numerous compounds that had not previously been described. Acetone, isoprene, benzene derivatives and alkane derivatives comprised the most abundant chemical species in the breath samples. Importantly, only 95 of these metabolites (out of the 2000) were shared among at least 90% of subjects, highlighting the inherent variation between individuals. This degree of variability indicates that human biomarker studies are likely to require large training cohorts and supervised machine learning methods to identify putative biomarkers, and independent testing cohorts to determine the accuracy of the biomarkers for predicting disease.</p><!><p>A few GC × GC metabolomics studies have been published on animals that are not considered human analogs, but these experiments have interesting study designs and observations that are relevant to human investigations. Rocha, et al. used GC × GC-TOFMS to examine differences in the volatile metabolomes from homogenized and salted tissues of wild Venerupis decussata and V. philippinarum, which are clams that can be found in the same geographical locations, but in different positions in the water column [47]. An average of 229 compounds were detected per species. Using multivariate analyses to reduce their data set and identify the most discriminatory compounds, they found 63 metabolites that significantly differed between the two species. The authors posited that the differences they observed between the two species could be due to the differences in clams' environments rather than their biology: dissimilarities in alkanes, alkenes, and terpenes were attributed to heterogeneous distributions of organic matter in the water column and marine sediments, while distinctions in aldehydes and alcohols were attributed to peroxidation of lipids from different dietary sources for the two species.</p><p>Rainbow trout (Oncorhynchus mykiss) naturally experience periods of starvation in their life cycle, which Baumgarner and Cooper hypothesized would cause different metabolic changes in different tissues [48]. They used GC × GC-TOFMS to compare the global metabolomes of serum, liver, and muscle tissues of 12 fish reared in captivity for two weeks, then split into two groups: six that were fed and six that were starved for four wks. They observed evidence that starved fish catabolize cellular protein in the liver for energy, but not in peripheral tissues. Additionally, they detected elevated xenobiotics (specifically n-alkanes) in fed fish that they posited were accumulated from fish food. However, in contrast to other xenobiotics, heptacosane was increased in starved fish, which they hypothesized was being liberated as specific tissues' energy reserves were being mobilized during starvation.</p><p>These two studies highlight a vexing complication of metabolomics analyses: parsing out metabolites versus xenobiotics, which we narrowly define here as exogenous substances that are accumulated from the environment and stored without chemical modification. Identifying xenobiotics is particularly difficult in cross-sectional studies of wild organisms, where past environmental conditions and exposures are not recorded. However, even experiments in captive organisms, like Baumgarner and Cooper's trout study [48], cannot definitively separate metabolites from xenobiotics without chemical characterization of the environment (i.e., food and water) and longitudinal sampling of tissues and biofluids to document bioaccumulation and/or release of xenobiotics. Alternatively, catabolism and anabolism of organic compounds can be traced using stable isotopes, examples of which are described in Section 3.6.</p><!><p>GC×GC is being used to characterize the complex metabolomes of plants to understand wide-ranging aspects of their physiology, ecology, and qualities as feedstocks and foods. For this review, we have chosen to exclude plant metabolomics studies that focus on food plant quality (spoilage, ripening, flavor, or aroma) or authenticity, and instead refer the reader to other reviews published in this Special Issue, as well as additional recent reviews on the topic of GC χ GC foodomics [49, 50].</p><p>To characterize the volatile and semi-volatile metabolomes of maturing 'Honeycrisp' apples (Malus × domestics Borkh.), Risticevic, Pawliszyn, and colleagues developed and optimized direct immersion (DI)-SPME for in vivo sampling [51]. In an earlier comparison of HS-SPME versus DI-SPME sampling of apple homogenates using a triphase fiber (DVB/CAR/PDMS), they obtained a 63% increase in metabolites using the latter method (555 vs. 906 compounds), and observed less bias against high molecular weight and polar metabolites [8]. For in vivo sampling Risticevic, et al. used DVB/CAR/PDMS fibers overcoated with PDMS, which were exposed to the apple tissue at a depth of 3 cm for 60 min at ambient temperatures. To remove interfering tissue material prior to analysis, they cleaned the fibers with lint-free wipes, washed the fibers in nanopure water for 10 s, then wiped the fibers again. By sampling in triplicate five apples of early maturity and five apples of late-harvest maturity, they found that inter-specimen variance was high, but they were able to clearly define the two groups of apples based on PCA of 225 peaks that were manually curated based upon high chromatographic quality.</p><p>As part of their natural defense systems, plants produce metabolites that deter attacks by other organisms. Wojciechowska, Geisen, and co-workers investigated the metabolic differences between two strains of tomatoes (Schmucktomate (ST) and Resi), which show differential resistance to the common fungal pathogen, Alternaria altemata [52]. They performed untargeted GC × GC metabolomics on the polar metabolites of ST and Resi, reproducibly detecting 267 metabolites from the tomatoes. Using volcano plot analysis, they identified 21 metabolites that were significantly elevated in ST, with chlorogenic acid (CGA) being the most discriminatory. Wojciechowska, et al. experimentally confirmed that CGA protects tomatoes from A. alternata colonization in a dose-dependent manner. Hantao, Augusto, and colleagues also used HS-SPME/GC × GC-qMS coupled with multivariate data analyses to identify biomarkers of Eucalyptus fungal infections [53], as well as disease-resistant clones [54]. Comparing biogenic volatile organic compounds produced by Eucalyptus globulus with and without infection by Teratosphaeria nubilosa fungus, they identified more than 40 volatiles that are putative biomarkers of infection [53]. Hantao, et al. also aimed to speed the selection of disease-resistant Eucalyptus hybrids by identifying volatile biomarkers of resistance to Eucalyptus rust [54]. They compared the volatilomes of E. grandis × E. urophylla hybrids resistant and susceptible to Puccinia psidii fungal infection, identifying two resistance biomarkers: eucalyptol and α-terpinyl acetate.</p><!><p>Most of the published GC χ GC metabolomics studies are descriptive, where the primary aim was to discover previously unidentified metabolites in a specimen, organism, or organismal interaction. While there is boundless need for these kinds of analyses, the data will obtain their greatest meaning when we are able to tie the metabolites to cellular mechanisms that produce or regulate their production, phenotypes they create, or interactions they facilitate. With these types of investigations, the data become information. Several groups are venturing in that direction by tying GC × GC metabolomics data to other chemical, biological, behavioral, and statistical analyses that contextualize the metabolome.</p><p>In vitro digestion models combined with GC × GC metabolomics have been used in two studies to characterize biotransformation of polyphenols in vivo. Aura, et al. measured the metabolic fate of polyphenols from Syrah red grapes, Syrah red wine, and extracted proanthocyanidins (PA) using a colonic model with fecal microbiota [55]. They observed a higher degree of C1-C3 phenolic acid formation from red wine than the fruit or PA in the colonic model. Vetrani and colleagues investigated the roles of the liver and of gut microbiota on transforming ingested polyphenols [56]. They analyzed the urine metabolomes of subjects consuming polyphenol-rich foods and beverages, and compared these data to the metabolomes generated by fecal microbiota fed the same diets in an in vitro colon model, and to hepatocytes cultured with green tea extracts. They found several associations between the urinary and colonic model metabolomes that partially explained polyphenol biotransformation in vivo, but complementing the colon model with hepatic metabolism significantly increased the correlation between in vivo and in vitro metabolism.</p><p>A technique that facilitates metabolic mapping is stable isotope labeling, which was employed by Žáček, Válterova, and coworkers to investigate the metabolic fate of dietary fatty acids (FAs) in the biosynthesis of bumblebee male marking pheromones [9]. The investigators fed or injected three species of bumble bees, Bombus lucorum, B. lapidaries, and B. terrestris, with 2H- or 13C-labeled C12, C14, C16, and C18 saturated FAs, then later harvested and analyzed the fat bodies and labial glands to characterized how the FAs had been stored and modified. They determined that FAs were stored as triacylglycerols in the fat body, and then modified and used for biosynthesis of pheromone precursors and pheromonal components. Importantly, Žáček et al. included an analysis on the effect of deuterium and carbon isotopes on the first and second dimension retention times (1tR and 2tR, respectively) in GC × GC. They observed larger shifts in 1tR with perdeuterated FAs (versus 13C-labeled FAs), and therefore used 2H labels for most analyses so they could more easily detect and quantify trace amounts of the labeled compounds in their untargeted analyses, even when a thousand times more unlabeled compound may be present. However, for confirmations of bioactivity, they relied upon the 13C-labeled FAs since the deuterated compounds can affect metabolism.</p><p>Cordero and colleagues used HS-SPME/GC × GC-qMS to investigate the metabolome of multitrophic interactions between mint plants (Mentha spp.) and their insect predator, the mint beetle (Chrysolina herbacea) [57. 58]. The first study explored how the mint beetle is able to tolerate the mono-terpenoids produced by the plant as a defense against insect herbivory [57]. They measured the volatile metabolomes of three different species of mint – two susceptible and one resistant to the pest – identifying four characteristic mint terpenoids in addition to 76 additional volatiles emanating from the leaves. By comparing the beetle frass (excrement) volatiles to the mint species on which they were reared, they were able to determine that the insects are biotransforming the toxic terpenoids during digestion, primarily by oxidation and acetylation. In a follow-up study, Pizzolante, et al. investigated the role of the gut microbiota in metabolizing Mentha aquatica terpenoids into sex-specific volatiles in beetle frass [58]. They identified 60 volatiles in the mint leaves and in beetle frass, including 9 terpenoid compounds that were nearly absent in the leaves but abundant in the frass, indicative of biotransformation of plant compounds during digestion. Additionally, they observed significant differences between the volatile metabolites in male and female frass that corresponded to differences in the cultivable species of bacteria that are found in their guts. To establish that the gut microbes were capable of metabolizing and biotransforming M. aquatica volatiles, they grew 16 C. herbacea gut bacterial isolates (10 from females, 6 from males) on mint extracts and measured the volatiles that were produced. They confirmed that the female and male gut microbiota biotransform the mint metabolites in unique ways, which they hypothesized may contribute to C. herbacea sex pheromone production.</p><p>As illustrated by these examples, deriving meaning from metabolic data requires a multidisciplinary approach. The most successful studies arise from deep collaborations between experts in disparate fields – chemistry, biology, medicine, mathematics – where all parties are involved in every stage of the project, from development to data analysis and publication.</p><!><p>The growth of the field of metabolomics is reflected by the recent investments by the National Institutes of Health (NIH) into the Common Fund Metabolomics program, which supports technology development, infrastructure, training, and an international repository for metabolomics data, the Metabolomics Workbench (metabolomicsworkbench.org) [59]. The goal is to generate publicly accessible metabolomics data, following the models of data sharing for genomics and transcriptomics via the GenBank and Gene Expression Omnibus (GEO) databases, respectively. Currently, depositing metabolomics data into public repositories is only encouraged under the NIH Data Sharing policies; however, it is possible that in the near future journal publishers may require data sharing as a condition of publication, as in genomics and transcriptomics. Under ideal circumstances, metabolomics databases will facilitate meta-analyses of multiple metabolomics experiments to generate new hypotheses and enhance translation of the data to practical applications in industry, medicine, and policy. In practice, achieving these ideals will require meticulous reporting of experimental metadata and very well designed experiments.</p><p>The Metabolomics Workbench and the Chemical Analysis Working Group (CAWG) Metabolomics Standards Initiative (MSI) [60] provide excellent guidelines on the biological and analytical metadata that should be included with metabolite data, and these parameters should be considered during experimental planning. When obtaining samples, thoughtful choices should be made regarding the source and selection of specimens, and data should be provided on the taxonomy of species, organs and tissues, cell types/lines, and strains. More detailed information such as animal husbandry, diet, growth media, and human data such as age, sex, body mass index, etc., are encouraged, as they can all impact metabolism. The CAWG provide detailed lists of analytical metadata to report for sample preparation, injection, separation, detection, method validation, and data preprocessing. Additionally, they provide recommendations on reporting the certainty to which chemical identity of metabolites have been determined using retention times/indices and mass spectral data [60]. Here, we review some important findings from GC × GC metabolomics studies that support the recommendations by the CAWG and inform clinical and analytical study design, and we make some recommendations for GC × GC metabolomics experiments to improve robustness and data sharing potential.</p><!><p>The Karlsruhe Metabolomics and Nutrition (KarMeN) Study is a large German cohort study (n = 301) of blood and urine metabolomes from healthy non-smoking adults [61]. Targeted and untargeted metabolomics analyses were performed using LC-MS/MS, 1H NMR, GC-MS, and GC × GC-MS, and metabolic differences due to body composition, age, sex, physical activity and diet are being characterized in the KarMeN Study. Due to the size of the cohort, the data being collected about each subject, and the variety of analytical techniques being applied, the metabolomics data from this study are providing a wealth of information on human health and metabolomics study design. A striking finding from their analysis is the degree to which the fasting plasma and urine metabolomes correlate with subject gender and age [62]. Using machine learning algorithms on the combined metabolomics data from all four analytical platforms, Rist and colleagues found that they could correctly predict the sex of the study participants with > 90 % accuracy from the urine metabolome, and > 95% accuracy from plasma. The correlation of the plasma and urine metabolomes to age were weaker, but still significant. A subset of the plasma metabolites in men correlated with age with R2 = 0.77, and a combination of plasma and urine metabolites in women predicted menopausal status with 90% accuracy. This study of healthy adults demonstrates that both sex and age are confounders of the human metabolome, and the authors recommend that sex, age, and sex-age interactions are included in statistical analyses of metabolomics data and reported with published results.</p><!><p>To obtain high-quality GC × GC metabolomics data, it is imperative to use the best chromatography available, as the data processing methods, chemometrics, and statistical analyses are only as good as the separations. There have been exciting developments in stationary phases and column configurations to optimize separations (reviewed in this Special Issue), however, phase thickness and phase ratio [63] are also important parameters to consider, which are often overlooked. While deconvolution procedures can enable identification and quantification of partially co-eluting peaks, they are impaired when peak shapes are compromised by column overloading. The problems created by poor peak shapes are compounded when multiple samples need to be aligned prior to downstream analyses [64]. Optimally, to obtain excellent data for mixtures of metabolites at concentrations that differ by many orders of magnitude, it is advisable to perform split and splitless injections and combine the data [64, 65]. However, this approach requires more supervised preprocessing and therefore is not amenable for large studies. An effective compromise can be reached by increasing stationary phase film volume by using thicker films and/or larger diameter columns, which increases loadability and therefore improves peak shapes for high concentration metabolites, but which requires longer separation times [66-68].</p><!><p>Sample handling and batch effects can play a significant role in metabolomics analyses, as described by Nizio, Forbes, and colleagues, [69] and should be closely controlled and accurately reported. Nizio, et al. set out to compare the volatilomes of six bacterial species, preparing four biological replicates of each bacterium and growth condition. Two biological replicates for each sample type were stored at −18°C for 2-5 d prior to GC × GC analysis, and the other two replicates were stored at −18°C for 48-50 d. They observed that the samples stored long-term produced more complex volatile profiles, as 200 more volatile metabolites were detected per sample, on average, compared to short-term storage. As a result, they posited that sample contamination, degradation, and/or biological activity may have contributed volatiles during storage [69]. However, a study by Wandro, et al., designed to test the stability of metabolites in cold storage [70] suggests that other factors may have been at play in Nizio's study. Wandro and colleagues measured the change in metabolites of sputum stored at 4°C, −20°C, and −80°C for up to eight weeks before extraction, derivatization, and GC-MS analysis was performed in a single batch. Samples stored at 4°C showed appreciable changes in their metabolome after one day of storage, whereas samples stored at −20°C for 1-56 days were statistically indistinguishable from samples stored at −80°C.</p><p>While Wandro's study does not specifically address the stability of volatile metabolites, which Nizio and colleagues were reporting, an important confounder to consider in Nizio's study is batch effects caused by instrumental drift, which can impact peak areas, and consequently metabolic profiles. In a five week long GC × GC-qMS run of 300 urine samples, Weinert et al., quantified the intra-day, inter-day, and inter-week RSD in peak height of 15 internal standards [19]. The intra-day and inter-day reproducibility was good (< 10 %), but despite weekly system maintenance to minimize matrix impacts (e.g., liner exchange, MS tuning), they observed a steady increase in peak heights of their internal standards (which was not explained by baseline drift), creating a mean inter-week RSD of 18 %. Weinert and colleagues did not use the internal standards to correct for the drift, as this method of normalization increased inter-week RSD of their sample analytes. However, the use of internal standards allowed them to track the performance of their system and quantify and report time-dependent variation in their data.</p><p>It is important to be aware of the natural structure of each study's GC × GC metabolomics data, especially in biomarker studies or other comparative analyses, so that biases can be accounted for and potentially corrected. An easy method to identify data structures that are not attributed to biological differences (e.g., batch effects) is to use PCA [71]. Figure 3 is an illustration of how we used PCA to uncover batch effects in a bacterial volatile metabolomics study of 81 P. aeruginosa clinical isolates analyzed in biological triplicates along with 15 media blanks. To briefly describe this study, isolates from 16 subjects, who provided 3-35 bacterial isolates over a span of at least five years of their chronic lung infections, were obtained from a biorepository. The 81 isolates were divided into two groups for HS-SPME/GC × GC-TOFMS volatile metabolomics, with 32 isolates placed into a "priority" group for the generation of preliminary data for a grant proposal, and the other 49 isolates placed into a "non-priority" group. Isolates from 11 subjects were divided between the priority and non-priority groups, while the isolates from five subjects were all in one group or the other. To minimize batch effects within each group, the order in which the isolates and their individual biological replicates were cultured and analyzed were randomized. Between culturing and analysis, samples were stored at −20°C, with a maximum storage time of 13 d. Table 1 summarizes the time frames for the preparation and analysis of the study samples.</p><p>All 258 chromatograms from both groups were processed, aligned, normalized, and statistically analyzed together. Without any prior feature selection, we performed PCA on all metabolite variables and sample observations, which shows that while the most significant variance in the metabolomes is accounted for by the metabolic differences between isolates and the media blanks (Fig. 3; PC1, 19.7%), approximately 10% of the sample variance is explained by priority vs. non-priority grouping (i.e., preparation batch), with samples in the priority group clustering in PC2 > 0, and samples in the non-priority group in PC2 < 0. As indicated by the marker colors in Fig. 3b, there is bias in the storage times between the priority and non-priority groups, but short and long storage times do not sufficiently explain the PC2 variance. Instrumental drift (visualized by analysis day, Fig. 3c) and growth media batches (not shown) also do not fully explain the separations in PC2. While we do not know all of the factors contributing to our data's structure, this unsupervised analysis of the entire data set revealed inherent correlations between a subset of the metabolites and the sample preparation batch. With this knowledge we can either exclude the batch-correlated metabolites from subsequent multivariate analyses, and/or conduct post-hoc testing for priority vs. non-priority group biases in the outcomes of supervised machine learning analyses of these data.</p><p>To minimize batch effects the unattainable ideal is to prepare or collect samples in a single small batch, store them for the same amount of time under the same (frozen) conditions, and analyze them in a single small batch. However, in practice it is not possible to meet these criteria when analyzing large numbers of samples or performing longitudinal studies. At a minimum, it is recommended that biological or technical replicates are obtained when possible, samples and replicates are prepared or collected in random order, and analyzed in a different random order. Most importantly, recognize that batch effects are impossible to avoid entirely and can arise due to unknown variables in sample collection, preparation, and analysis; therefore biases in the metabolomics data should be quantified, potentially corrected, and always reported.</p><!><p>The large number of metabolites that are revealed by GC × GC analyses causes a statistical conundrum, wherein the number of variables is often one or two orders of magnitude larger than the number of observations made. Therefore, statistical models that are employed to identify differences in group comparisons (e.g., putative biomarkers of disease) are at risk of overfitting. To address this problem, it is becoming more common for investigators to use multiple statistical models to identify the most discriminatory metabolites, and report the consensus set as the putative biomarkers of disease. For example, Phillips, et al. identified breath biomarkers of therapeutic radiation exposure by analyzing the breath of 31 individuals who received varying doses of radiation (180-1200 cGy/d) over five days [72]. Multiple Monte Carlo simulations were used to identify approximately 50 metabolites that significantly changed pre and post radiation. The 15 breath volatiles that were observed in 7 of 8 simulations were 99% accurate in identifying subjects who received at least 1.8 Gy. The approach of using multiple statistical analyses to identify putative biomarkers has been used in several other GC × GC metabolomics studies of lung specimens. To identify biomarkers of M. tuberculosis infection, du Preez and Loots analyzed methanol-extracted metabolites from the sputum of 61 TB-negative and 34 TB-positive subjects, detecting a total of 498 metabolites [40]. They applied a combination of supervised and unsupervised univariate and multivariate statistical analyses to select 22 metabolite markers of infection, which were identified as putative biomarkers by all three statistical methods (PLS-DA, fold-change, and t-tests). Beccaria et al. used GC × GC-TOFMS to identify M. tuberculosis infection biomarkers in breath [42]. By using three cutoffs for the frequency of observation of 1513 volatiles in 34 breath samples (20 %, 50 %, and 80 %), they generated three pools of variables for RF analysis. Twenty two breath volatiles were in common across all three RF analyses, which the authors considered the most promising leads for discriminating TB-positive and TB-negative breath samples.</p><!><p>Comprehensive two-dimensional gas chromatography is gaining popularity for metabolomics analyses, and has significantly expanded the metabolic catalogs of microbes, plants, animals, and humans. The field is progressing into studies of causality by incorporating longitudinal analyses, isotopic tracers, flux analysis, model systems, and multitrophic interactions into GC × GC metabolomics analyses. However, despite the clear analytical advantages that GC × GC offers for characterizing complex mixtures, especially for untargeted metabolomics studies, its adoption by new users is relatively slow. A significant barrier to growth is the availability of user-friendly software that can handle the entire data processing and analysis pipeline. While there have been brisk advancements in investigator-developed packages for peak picking, peak alignment, deconvolution, etc. [32, 73], utilizing them requires fluency with Matlab, R, Python, and/or other programming languages. The ideal software platform will incorporate a user-friendly interface and several different approaches for each step in the data analysis pipeline, allowing users to optimize the workflow for their particular samples. Obtaining the GC × GC metabolomics peak list is just the first step of linking the chemical data to biology, and as future studies aim to integrate metabolomics with transcriptomics, proteomics, and genomics information, easily-manipulated graphical displays of chromatographic and statistical data will be required.</p><p>For GC × GC metabolomics to be adopted by new investigators, and for the field of metabolomics, writ large, to continue to receive new investments, we in the field must demonstrate that the data we are generating are of high quality and can be independently validated. This requires excellence in study design, sample collection and storage, chromatographic analysis, data processing, and statistical analyses, and in order to objectively judge quality in all of these domains we must record and report ample metadata for all steps. It is not only incumbent upon investigators to be diligent in these pursuits, but also upon manuscript peer reviewers and journal editors to insist that methods are fully described and statistical evaluations of the data are appropriate and complete. Only with this level of rigor will GC × GC metabolomics become an essential component of any multi-omics study design.</p>

PubMed Author Manuscript

Isolation of tumorigenic circulating melanoma cells

Circulating tumor cells (CTC) have been identified in several human malignancies, including malignant melanoma. However, whether melanoma CTC are tumorigenic and cause metastatic progression is currently unknown. Here we isolate for the first time viable tumorigenic melanoma CTC and demonstrate that this cell population is capable of metastasis formation in human-to-mouse xenotransplantation experiments. The presence of CTC among peripheral blood mononuclear cells (PBMC) of murine recipients of subcutaneous (s.c.) human melanoma xenografts could be detected based on mRNA expression for human GAPDH and/or ATP-binding cassette subfamily B member 5 (ABCB5), a marker of malignant melanoma-initiating cells previously shown to be associated with metastatic disease progression in human patients. ABCB5 expression could also be detected in PBMC preparations from human stage IV melanoma patients but not healthy controls. The detection of melanoma CTC in human-to-mouse s.c. tumor xenotransplantation models correlated significantly with pulmonary metastasis formation. Moreover, prospectively isolated CTC from murine recipients of s.c. melanoma xenografts were capable of primary tumor initiation and caused metastasis formation upon xenotransplantation to secondary murine NOD-scid IL2R\xce\xb3null recipients. Our results provide initial evidence that melanoma CTC are tumorigenic and demonstrate that CTC are capable of causing metastatic tumor progression. These findings suggest a need for CTC eradication to inhibit metastatic progression and provide a rationale for assessment of therapeutic responses of this tumorigenic cell population to promising emerging melanoma treatment modalities.

isolation_of_tumorigenic_circulating_melanoma_cells

1. Introduction<!>Melanoma specimens<!>Animals<!>Human-to-mouse xenotransplantation studies<!>Antibodies<!>Isolation of circulating melanoma cells<!>Flow cyotometry<!>RNA extraction, reverse transcription, and real-time quantitative PCR<!>Isolation of tumorigenic circulating melanoma cells<!>Melanoma CTC contain ABCB5-positive subpopulations associated with metastatic disease progression in human patients<!>Melanoma CTC detection correlates with metastatic progression<!>4. Discussion<!>Isolation of tumorigenic circulating melanoma cells<!>Circulating melanoma cells contain ABCB5-positive tumor subpopulations<!>Melanoma CTC detection correlates with metastatic progression<!>

<p>Human melanoma is the most malignant skin cancer, with increasing incident rates worldwide. Although melanoma can be cured by surgical resection before the cancer has spread, metastatic melanoma is one of the most aggressive and drug-resistant cancers with very poor overall survival [1; 2; 3]. A more detailed understanding of the cellular mechanisms that drive metastatic melanoma progression would improve the development and assessment of more effective new or emerging treatment modalities that target disseminated disease.</p><p>In metastatic dissemination, primary tumor cells invade basement membranes, intravasate into blood or lymphatic vessels, reach secondary sites through the circulation, and extravasate and colonize again [4]. In recent years, circulating tumor cells (CTC) have been actively investigated as potential prognostic and therapeutic biomarkers [5]. In melanoma patients, CTC were first detected based on expression of tyrosinase RNA in the peripheral circulation [6]. Melanoma CTC have been found to correlate significantly with tumor stage and patient survival [7], and can serve as a biomarker for predicting antitumor responses in patients undergoing systemic therapies [8].</p><p>Different strategies have been employed to detect and characterize melanoma CTC, and several genes have been investigated as melanoma CTC biomarkers, including genes involved in melanin biosynthesis or genes encoding melanoma-associated antigens [9]. However, isolation of viable CTC, a prerequisite for functional studies in tumor initiation and metastasis of this cell population, has not been achieved to date. While observations of prognostic relevance of melanoma CTC have suggested that this malignant subpopulation might indeed be tumorigenic and responsible for the development of distant metastases, this hypothesis has not been experimentally proven to date.</p><p>Melanoma-initiating cells [10; 11] are minority subpopulations in which clinical virulence resides as a consequence of unlimited self-renewal capacity, resulting in inexorable tumor progression [1; 10; 11]. Our laboratory recently showed human malignant melanoma initiating cells (MMIC) to express the targetable biomarker [10; 12; 13; 14] and multidrug resistance mediator [14; 15], ATP-binding cassette subfamily B member 5 (ABCB5) [10]. In this study, proof of principle of immune-mediated MMIC destruction and consequent inhibition of tumor growth was demonstrated [10]. More recently, we have shown that MMIC employ mechanisms to thwart endogenous anti-tumor immunity [16], which could account for the observed relative enrichment of MMIC in patient lymph node (LN) metastases compared to visceral metastases [10]. Whether ABCB5 might also be expressed by melanoma CTC hypothesized to exert important roles in metastatic progression, has not been evaluated to date.</p><p>Here we provide initial evidence for the existence of circulating melanoma cells that are capable of causing tumor initiation and metastatic disease progression, considerably strengthening the rationale for ongoing clinical evaluations of melanoma CTC as a biomarker for disease progression, prognosis and outcome. Furthermore, we find that CTC can be comprised of heterogeneous cancer populations, including ABCB5-positive subsets previously associated with clinical metastatic melanoma progression. This additional result provides a rationale to examine in future studies whether specific CTC subsets might contribute differentially to tumor initiation and metastatic disease progression, of relevance to the selection of appropriate CTC markers for diagnostic, prognostic or therapeutic purposes.</p><!><p>Clinical melanoma cells were derived from surgical specimens according to an Institutional Review Board–approved research protocol (University of Würzburg, Germany). Metastatic human C8161 melanoma cells [17] were provided by Dr. Mary Hendrix (Children's Memorial Research Center, Chicago, IL) and metastatic human LOX and FEMX1 melanoma cells [18] were provided by Dr. Udo Schumacher (University Hospital Hamburg-Eppendorf, Hamburg, Germany). Melanoma cell lines were cultured as described [15]. Fluorescent protein-expressing cells were generated by stable transfection of C8161, LOX, or clinical melanoma cells with expression vectors encoding red fluorescent protein (RFP) or green fluorescent protein (GFP) genes (BD Bioscience, Franklin Lakes, NJ) as described [10]. Individual clones were generated for RFP-labeled LOX or clinical melanoma cells. Fluorescence-transfected C8161 cells were purified for high RFP or GFP expression using a Cytomation MoFlo sorter (Dako,Carpinteria, CA).</p><!><p>Athymic nude mice were purchased from Harlan Laboratories (Indianapolis, IN). Nonobese diabetic-severe combined immune deficiency (NOD-scid) mice and NOD-scid IL2Rγnull mice were purchased from The Jackson Laboratory (Bar Harbor, ME). Mice were maintained under defined conditions in accordance with the institutional guidelines of Children's Hospital Boston and Harvard Medical School. Experiments were performed according to experimental protocols approved by Children's Hospital Boston.</p><!><p>Human melanoma cells (C8161, LOX, FEMX1 or clinical specimen-derived cells) were injected subcutaneously (s.c.) into the right flank of recipient NOD-scid or nude mice (1×106 in 100 μL of PBS per inoculum). Tumor formation was assayed weekly up to 11 weeks and the development of pulmonary metastases was determined by examining hematoxylin-eosin (H&E)-stained paraffin-embedded tissue sections prepared by serial sectioning of the entire lungs of all primary tumor-bearing mice.</p><!><p>The murine IgG1κ anti-ABCB5 monoclonal antibody (mAb) 3C2-1D12 [10; 15; 16; 19] was used in the herein reported studies. Isotype control mAb and allophycocyanin (APC)-conjugated murine anti-human HLA-ABC IgG1 Ab were purchased from BD Bioscience. APC-conjugated isotype murine IgG1 mAb was purchased from Miltenyi Biotec (Auburn, CA). APC-conjugated donkey anti-mouse IgG was purchased from eBioscience (San Diego, CA).</p><!><p>NOD-scid IL2Rγnull mice were inoculated s.c. with RFP-transfected human melanoma cells (2×103 to 2×104 cells/inoculum). Mice were euthanized 5 to 7 weeks upon xenotransplantation and whole blood specimens were collected through heart acupuncture. PBMC were isolated as described [20] and stained with APC-conjugated anti-human HLA-ABC Ab or APC-conjugated isotype control Ab. To isolate CTC, RFP-positive and/or human HLA-ABC antigen-positive cells were purified from cell suspensions using a MoFlo high-speed sorter and collected in growth factor-reduced matrigel (BD Biosciences). Cultured RFP-transfected human melanoma cells or PBMC from tumor-free NOD-scid IL2Rγnull mice were used as positive or negative controls, respectively. Matrigel preparations containing sorted melanoma CTC were injected s.c. into the right flank of secondary NOD-scid IL2Rγnull mice (first-generation CTC xenotransplantation). Tumor formation was monitored weekly for 7 weeks and subsequently, second-generation CTC were isolated as above and xenografted s.c. to tertiary NOD-scid IL2Rγnull mice. Primary tumors and lungs from xenografted mice were collected and the expression of RFP was determined following counterstaining with DAPI by fluorescence microscopy as described [10].</p><!><p>ABCB5 protein expression in human melanoma cultures, primary tumors, distant metastases and CTC was determined by flow cytometry as described [10; 15; 16; 19]. First, NOD-scid IL2Rγnull mice were inoculated s.c. with C8161/RFP or C8161/GFP cells (2×104 cells/inoculum). Subsequently, primary tumors, lungs, and enlarged axillary lymph nodes were harvested and dissociated as described [10]. Cell suspensions were incubated with anti-ABCB5 mAb or isotype control mAb followed by counterstaining with APC-conjugated secondary antibody, and dual- or triple-color flow cytometry was subsequently performed at the FL1 (GFP) or FL2 (RFP) spectra, and the FL4 (APC) spectrum as described [10]. Single-cell suspensions from tumor-free mice were used as negative controls to exclude autofluorescent cells. Melanoma cells were identified as cells positive for RFP or GFP. Statistical differences in the percentages of ABCB5-positive melanoma cells were determined using ANOVA and Bonferroni's Multiple Comparison Test, with P<0.05 considered statistically significant.</p><!><p>Murine blood specimens were collected by cardiac puncture and erythrocytes were lysed using ACK lysis buffer (Invitrogen). Human blood specimens were obtained from metastatic melanoma patients or healthy controls in compliance with institutional IRB regulations and PBMC were isolated as described [21]. Total RNA was extracted and transcribed into complementary DNA (cDNA), and real-time quantitative PCR was performed as described [10]. Human GAPDH expression in PBMC samples of murine melanoma xenograft recipients was determined using a TaqMan assay (Hs99999905_ml, Applied Biosystems) and human ABCB5 expression was assayed using SYBR Green chemistry, with the use of 5'-ATGATGTGACTGATGAAGAGATGGAGAGA-3' and 5'-CTCTGTTTCTGCCCTCCACTCATTTGAGC-3' as forward and reverse primers, respectively. A Ct value of 33.5 was set as the threshold for ABCB5 positivity, as determined by using PBMC from tumor-free NOD-scid mice as negative controls. For ABCB5 detection in human metastatic melanoma patient blood specimens, a TaqMan assay (Hs00698751_m1, Applied Biosystems) was performed, with use of a ΔCt value<15 (compared to b-actin expression) as a threshold for significant ABCB5 mRNA detection, as determined by using PBMC from healthy donors, previously found negative for ABCB5 expression [19], as negative controls. Statistically significant associations between GAPDH and/or ABCB5 detection and the presence of melanoma metastases were determined using the Fisher's Exact Test and/or Relative Risk analyses, with P<0.05 considered significant.</p><!><p>In order to demonstrate tumorigenic potential of human melanoma CTC, we developed a novel metastatic melanoma xenotransplantation model whereby fluorescent transgene-expressing metastatic human melanoma cells are xenografted s.c. to immunodeficient mice and, upon primary tumor formation and systemic spreading, human melanoma cells are isolated from the murine circulation for use in further tumorigenicity experiments (Fig.1A). In order to maximize tumor take resulting from xenotransplantation of relatively low CTC numbers in this model, we grafted human melanoma CTC isolated from murine PBMC into severely immune-compromised NOD-scid IL2Rγnull mice, using matrigel [11; 22]. To maximize the efficiency of melanoma CTC detection for flow cytometric cell sorting, melanoma cells were transfected with the fluorescent transgene RFP, and CTC that expressed the RFP marker, or the human MHC Class-I antigen HLA-ABC, or both markers, were flow cytometrically sorted from the murine circulation (Fig.1B). These detection criteria were posited to maximize human CTC detection, because of the possibility of gradual loss of RFP transgene expression during prolonged in vivo tumorigenic growth, and because of the possibility that human melanoma cells might not invariably express human MHC Class I antigens [16]. Subcutaneous xenotransplantation of human melanoma cells (2×103-2×104 cells/inoculum) resulted in consistent primary tumor formation in NOD-scid IL2Rγnull primary recipient mice (n=18/18; Table 1). CTC isolated from the circulation of n=11 primary tumor-bearing mice (first-generation CTC) resulted in tumor formation in 7 of 11 secondary recipients upon s.c xenotransplantation of purified CTC ranging in numbers from 81 to 4.2×103, including 4 of 4 murine recipients of clinical specimen-derived melanoma CTC (Fig.1C, Table 1). Further re-isolated CTC from secondary xenograft recipient (second-generation CTC) performed for C8161 melanomas were likewise capable of tumor initiation in 2 of 2 tertiary xenograft recipients (Fig.1C, Table 1). Purified CTC were not only capable of primary tumor initiation (Fig.1C,D), but also of metastasis formation upon s.c. xenotransplantation (Fig.1D). These results demonstrate a capacity of circulating human melanoma cells for in vivo initiation of tumorigenic growth and metastatic progression.</p><!><p>We and others have demonstrated that the chemoresistance mediator ABCB5 [14; 15] marks immunoevasive [16] subpopulations of MMIC that correlate with metastatic malignant melanoma progression in human patients [10; 12; 13] and experimental xenotransplantation models [22]. We therefore examined whether ABCB5 might also be expressed by melanoma CTC found capable of tumor initiation and metastasis formation in this study. Similar to results of previous expression analyses in cultured melanoma cells and patient-derived primary tumor and metastatic specimens, ABCB5 was expressed on melanoma subpopulations ranging from 1.0% to 1.8% in clinical specimen-derived melanoma cells and established C8161, FEMX1 and LOX melanoma cultures (Fig.2A). Further investigation of ABCB5-positive MMIC frequencies at each step of tumorigenesis and metastatic progression revealed similarly low percentages of ABCB5-positive cells among melanoma cells derived from primary tumors and axillary lymph node (LN) metastases in the C8161 xenotransplantation model. Interestingly, ABCB5-positive melanoma cell frequency was significantly increased among melanoma CTC (33.4±7.4%) compared to the frequencies detected among melanoma cells in xenograft inocula, resultant primary tumors, LN metastases or pulmonary metastases (P<0.001, P<0.001, P<0.05 and P<0.001, respectively) (Fig.2B). Thus, circulating melanoma cells represent heterogeneous cell populations that include ABCB5-positive subsets previously associated with clinical metastatic progression [10].</p><!><p>Based on our demonstration that melanoma CTC can cause tumor formation and metastatic progression in human-to-mouse xenotransplantation experiments, we examined whether detection of melanoma CTC might correlate significantly with metastatic tumor progression. Human melanoma cells (C8161, LOX, FEMX1 or clinical specimen-derived cells) were injected s.c. into the right flank of recipient NOD-scid or nude mice (1×106 in 100 μL of PBS per inoculum), and the presence or absence of CTC and pulmonary metastases was assayed in n=32 xenograft recipients with established primary tumors. The presence of CTC among PBMC of murine recipients of s.c. human melanoma xenografts was established based on mRNA detection for human GAPDH and/or human ABCB5, a marker of drug resistant and immunoevasive MMIC [1; 2; 10; 15; 16]. Absence of CTC among PBMC preparations was defined as dual negativity for human GAPDH and ABCB5 expression. The presence or absence of pulmonary macro- or micrometastases was determined by macroscopic tissue examination and microscopic examination of serial tissue sections of the entire lungs of all primary tumor-bearing mice.</p><p>Among 27 CTC-positive primary tumor-bearing mice, 25 (92.6%) had pulmonary metastases, whereas only one of five (20%) of CTC-negative primary tumor-bearing mice showed evidence for pulmonary metastases (Fig.3A, B), demonstrating a significant association between CTC detection and the presence of pulmonary metastases (P<0.01). Mice with metastatic disease hereby exhibited human GAPDH positivity in 92% of cases, whereas mice without metastases were human GAPDH-positive in only 33% of cases (P<0.05) (Fig.3C). Regarding human ABCB5 detection, only mice with metastatic disease showed ABCB5 positivity (19% of metastatic cases), whereas none of the mice without metastases exhibited ABCB5 positivity (P<0.05) (Fig.3D), indicating that ABCB5 might represent a novel relatively specific, albeit relatively insensitive molecular marker for the detection of CTC that correlate with metastasis formation. In support of this possibility, only PBMC specimens derived from n=9 stage IV metastatic melanoma patients showed ABCB5 positivity in 3 of 9 cases, whereas none of the PBMC specimens derived from n=5 healthy human controls were found to express ABCB5 (P<0.05), consistent with previously determined ABCB5 mRNA negativity of human PBMC [19].</p><!><p>CTC are thought to be important mediators of tumor dissemination and metastatic disease progression in human patients. They represent unique malignant subpopulations that emigrate from the cellular microenvironment and extracellular matrix support of the primary tumor site, survive in a fluid dynamic environment in the context of immunocompetent PBMC, and possess the capacity to eventually home to and establish tumorigenic growth in secondary sites of metastasis. Not all CTC might necessarily be alike; some might readily adapt to a new microenvironment and reinitiate cellular proliferation causing metastasis, while others might remain dormant for prolonged time periods or ultimately die [23; 24]. Moreover, while some CTC might drive metastatic dissemination of the primary tumor, others have been shown to contribute to primary tumor growth through self-seeding [25].</p><p>CTC have previously been detected and phenotypically characterized in human malignant melanoma, using melanoma-specific or melanoma-associated molecular markers [9], and CTC detection has been shown to correlate with clinical tumor progression [26]. However, the existence of tumorigenic melanoma CTC capable of driving metastatic progression, a concept that underlies the posited prognostic value of CTC detection in this malignancy, has not been directly demonstrated prior to this study. Therefore, our finding that melanoma CTC are capable of tumor initiation and metastatic progression significantly strengthens the rationale for further evaluating CTC as a biomarker for melanoma progression, prognosis and outcome, and as potential therapeutic targets in this malignancy. Additionally, the prospective isolation of viable melanoma CTC promises to open new avenues for the functional study and further biological characterization of these cell populations, and to potentially lead to improved methods for evaluating therapeutic responses of melanoma CTC.</p><p>A second significant insight of our study is the finding that melanoma CTC represent phenotypically heterogeneous cell populations. This is indicated by our result of heterogeneous expression of the MMIC marker ABCB5 and MHC class I antigens on specific subsets of CTC, similar to previous findings in human primary tumor- or metastasis-derived melanoma cells [10; 16]. Intriguingly, ABCB5-positive tumor cells were >20-fold enriched among melanoma CTC compared to either primary tumors or pulmonary metastases. These results provide a possible explanation for the previously established role of ABCB5 as a marker of metastatic progression in human melanoma patients [10; 12]. Moreover, because ABCB5-positive MMIC and related melanoma initiating cell populations [11] often do not express melanoma-associated antigens such as MART-1 or tyrosinase [11; 16], our results suggest that caution should be applied when selecting melanoma CTC markers for diagnostic, prognostic or therapeutic purposes, as not all disease-relevant CTC populations might be measured or targeted when only select tumor antigens are considered. In addition, our results provide a rationale to examine in future studies whether distinct subpopulations of melanoma CTC might differ in their capacity for tumorigenic growth and metastatic progression.</p><p>Lastly, our results suggest that ABCB5 might represent a useful novel molecular marker for the detection and monitoring of tumorigenic melanoma CTC populations associated with metastatic disease, because ABCB5 detection in CTC was a highly specific predictor for the presence of metastatic disease in the here-examined experimental metastatic melanoma model. Moreover, our results revealed selective ABCB5 expression in PBMC of a subset of metastatic human melanoma patients but not in PBMC derived from healthy human controls as previously described [19], underlining the melanoma-specific expression pattern of this MMIC-associated gene. These findings resemble recent observations by others of melanoma-specific ABCB5 expression in a subset of patient sentinel lymph node biopsy specimens, but absent expression in all of a series of non-melanoma case-derived control lymphatic tissues [13].</p><p>In summary, our results show that melanoma CTC contain tumorigenic cancer cells and demonstrate that CTC are capable of causing metastatic tumor progression. These findings suggest a need for CTC eradication to inhibit metastatic progression and provide a rationale for assessment of therapeutic responses of this cell population to promising emerging melanoma treatment modalities.</p><!><p>A. Diagram depicting the procedure for CTC isolation and serial xenotransplantation. B. Representative flow cytometry results for RFP and human HLA-ABC antigen detection in blood specimens from tumor-free control mice (left) and mice xenografted with RFP-transfected human melanoma cells (right). The gate used for CTC sorting is shown in red. C. Tumor initiation capacity of CTC in NOD-scid IL2Rγnull mice. D. Top: Representative fluorescent and light microscopy of a tumor derived from RFP-transfected melanoma cells. Bottom: Representative fluorescent and light microscopy of a tumor derived from second-generation CTC. Scale bar, 100 μm. E. Representative image of a pulmonary metastasis derived from second-generation melanoma CTC. Scale bar, 100 μm.</p><!><p>A. Flow cytometry plots depicting ABCB5 expression in melanoma cell inocula for a clinical specimen and C8161, FEMX1 and LOX melanoma cells. The lower panels show isotype control-staining. B. Flow cytometrically-determined percentages of ABCB5-positive melanoma cells in cell inocula and resultant primary tumors, lymph node metastases, melanoma CTC, and pulmonary metastases. Means±SE of n=3-11 samples/group are illustrated.</p><!><p>A. Correlation between pulmonary metastases in primary tumor-bearing mice and CTC detection. B. Representative morphology of a primary tumor (left) and pulmonary metastases (right) following xenotransplantation of patient-derived melanoma cells. Scale bar, 100 μm. C. Detection of human GAPDH, and D. human ABCB5 expression in PBMC derived from primary tumor-bearing mice with or without the presence of pulmonary metastases. E. ABCB5 gene expression in PBMC derived from metastatic melanoma patients or healthy humans.</p><!><p>Tumorigenicity of melanoma CTC</p><p>Numbers in parentheses indicate the number of cells inoculated per mouse</p>

PubMed Author Manuscript

Synthetic reactions driven by electron-donor–acceptor (EDA) complexes

The reversible, weak ground-state aggregate formed by dipole–dipole interactions between an electron donor and an electron acceptor is referred to as an electron-donor–acceptor (EDA) complex. Generally, upon light irradiation, the EDA complex turns into the excited state, causing an electron transfer to give radicals and to initiate subsequent reactions. Besides light as an external energy source, reactions involving the participation of EDA complexes are mild, obviating transition metal catalysts or photosensitizers in the majority of cases and are in line with the theme of green chemistry. This review discusses the synthetic reactions concerned with EDA complexes as well as the mechanisms that have been shown over the past five years.

synthetic_reactions_driven_by_electron-donor–acceptor_(eda)_complexes

Introduction<!><!>Cyclization reactions<!><!>Cyclization reactions<!><!>Cyclization reactions<!><!>Cyclization reactions<!><!>Cyclization reactions<!><!>Cyclization reactions<!><!>Cyclization reactions<!><!>Cyclization reactions<!><!>Cyclization reactions<!><!>Cyclization reactions<!><!>Cyclization reactions<!><!>Cyclization reactions<!><!>Cyclization reactions<!><!>Cyclization reactions<!><!>Cyclization reactions<!><!>Cyclization reactions<!><!>Cyclization reactions<!><!>The construction of C–C bonds<!><!>The construction of C–C bonds<!><!>The construction of C–C bonds<!><!>The construction of C–C bonds<!><!>The construction of C–C bonds<!><!>The construction of C–C bonds<!><!>The construction of C–C bonds<!><!>The construction of C–C bonds<!><!>The construction of C–C bonds<!><!>The construction of C–C bonds<!><!>The construction of C–C bonds<!><!>The construction of C–C bonds<!><!>The construction of C–C bonds<!><!>The construction of C–C bonds<!><!>The construction of C–C bonds<!><!>The construction of C–C bonds<!><!>The construction of C–C bonds<!><!>The construction of C–C bonds<!><!>The construction of C–C bonds<!><!>The construction of C–C bonds<!><!>The construction of C–C bonds<!><!>The construction of C–S bonds<!><!>The construction of C–S bonds<!><!>The construction of C–S bonds<!><!>The construction of C–S bonds<!><!>The construction of C–B bonds<!><!>The construction of C–B bonds<!><!>The construction of C–B bonds<!><!>The construction of C–B bonds<!><!>The construction of C–N bonds<!><!>The construction of C–N bonds<!><!>The construction of C–N bonds<!><!>The construction of C–P bonds<!><!>The construction of C–O bonds and C–H bonds<!><!>The construction of C–O bonds and C–H bonds<!><!>The construction of C–O bonds and C–H bonds<!><!>Conclusion

<p>Electron transfer (ET) is a very common occurrence in the field of natural science, including photochemical, electrochemical, and enzymatic reactions and even major organic synthesis. From 1950 to 1952, Mulliken suggested an electron transfer hypothesis that could more precisely explain electron transfer phenomena based on the electron-donor–acceptor (EDA) complex [1–3]. Significantly, a broad absorption peak unrelated to the structure, called charge-transfer band, is typically located in the visible region of the UV–vis spectrum [4], which manifests the color variability of the mixed solution of the electron donor (D) and electron acceptor (A). The two components A and D may not absorb visible light, but the resulting EDA complex does [5]. If the EDA complex is irradiated with a particular wavelength (or heated to a corresponding temperature), the complex could be excited to the state [D, A]*, causing electron transfer and forming a pair of radical ions trapped in the solvent cage. The pair of radical ions escapes the solvent cage by diffusion to give radical ions, which could initiate chemical reactions or reverse electron transfer (Scheme 1) [6]. The continuously increasing demand for sustainable synthesis has encouraged chemists to pursue more efficient methods to manufacture fine and usable chemicals [7]. The reactions that EDA complexes participate in have been shown to be an enormous success, mainly due to the fact that they obviate photoredox catalysts or transition metal catalysts in the vast majority cases. Moreover, in line with the theme of green chemistry, light is the sole external energy source in EDA complex pathways. Except for the pioneering research on EDA complexes in the 20th century, there was not much progress in the follow-up. Until the past few years, EDA-complex photochemistry has attracted more and more chemists and provided new opportunities for synthetic chemistry [8]. Moreover, diverse photocatalyst-free photochemical reactions have been employed to construct carbon–carbon and carbon–heteroatom bonds [9]. Among these methods, the product formations by aid of EDA complexes are the simplest and most rapid way. In this review, the focus lies on cyclization reactions, C–C-, C–S-, C–B-, C–N-, C–P-, C–O-, and C–H bond formation, primarily summarizing the chemical reaction step involving the EDA complex (Table 1) as well as the underlying mechanisms that have appeared over the last five years.</p><!><p>The electron transfer process in EDA complexes.</p><p>The EDA complexes discussed in this review.</p><!><p>In the pharmaceutical industry, retail drugs with a heterocyclic composition have exceeded 60% of the market volume. Hence, cyclization reaction innovation seems to be a requisite for pharmaceutical industry and human health. As an outstanding way, free-radical cascade reactions could efficiently construct various carbocycles and heterocycles with multifarious structures and complexity [59–61]. Centered on this context, we give a clear overview on a variety of novel cyclization reactions initiated by EDA complexes from the recent years.</p><p>In 2016, Lakhdar and colleagues [10] obtained the target product 3 with LED (5 W) irradiation of a solution containing arylphosphine oxide 2, alkynes 1, eosin Y (EY, 4 mol %), N-ethoxy-2-methylpyridinium (4), and sodium bicarbonate in DMF (Scheme 2). As a distinct example of EDA complexes, the process efficiency depends on the association of eosin Y and oxidant 4 to a donor–acceptor EY–4 ground-system complex (high reactivity). Due to the ability of aryl groups to stabilize the formed alkenyl radical, this protocol could control regioselectivity efficiently with unsymmetrical alkynes. In addition, EPR spectroscopy shows that phosphono radicals could proceed throughout the reaction.</p><!><p>Synthesis of benzo[b]phosphorus oxide 3 initiated by an EDA complex.</p><!><p>A halogen bond (XB) is a noncovalent interaction formed between a halogen atom and a neutral or negatively charged Lewis base. It is a kind of weak intermolecular interaction analogous to a hydrogen bond and basically can be considered as a specific EDA complex [62]. In 2016, Yu and colleagues [33] employed perfluoroalkyl iodide 6 as halogen-bond donor (electron acceptor) and the organic base dibenzylamine as the halogen-bond acceptor (electron donor) to form the XB complex 8, and then a fluoroalkyl radical was given via visible-light-induced single-electron-transfer process. 1,2-Diisocyanato-4,5-xylene (5) was able to capture the fluoroalkyl radical, eventually providing the quinoxaline derivative 7 (Scheme 3).</p><!><p>Mechanism of the synthesis of quinoxaline derivative 7.</p><!><p>In 2017, Fu and colleagues [20] selected 1-phenyl-2-(piperidin-1-yl)ethanone O-(2,4-dinitrophenyl)oxime (9) as substrates under 23 W CFL (compact fluorescent lamp) irradiation, affording the desired imidazole derivative 10 by utilizing DMSO as the solvent at room temperature (Scheme 4). It is worth noting that, unlike most reported intermolecular electron-transfer via an EDA complex pathway, this approach transfers electrons from the electron-rich tertiary amine nitrogen atom to the electron-deficient benzene ring, achieving intramolecular electron transfer. Selective C–H-functionalization also includes no catalysts, oxidants, additives, acids and bases, which is of great significance in the synthesis and application of N-heterocyclic compounds.</p><!><p>Synthesis of imidazole derivative 10 initiated by an EDA complex.</p><!><p>In 2017, Wu and colleagues [58] uncovered the use of the bis(sulfur dioxide) adduct of DABCO, 1,4-diazabicyclo[2.2.2]octane·(SO2)2, as sulfone source in the EDA complex formation by 4-methyl-1-(p-tolyl)pentan-1-one O-(2,4-dinitrophenyl)oxime (11) towards the aminosulfonylation, employing blue light as irradiation source (Scheme 5). It has to be said that nitrogen radicals played a coordinating role in the sulfonation step. Additionally, to verify the applicability of this approach, 1H-benzo[d][1,2]thiazine 2,2-dioxides have been prepared successfully.</p><!><p>Synthesis of sulfamoylation product 12 initiated by an EDA complex.</p><!><p>A possible reaction mechanism for this transformation is as follows (Scheme 6): Firstly, O-aryloxime 11 forms EDA complex 13 by action of DABCO·(SO2)2 and then undergoes light-promoted single-electron transfer, affording the 2,4-dinitrophenol anion, nitrogen radical 14, and radical 15, respectively. 1,5-HAT (hydrogen atom transfer) occurs in nitrogen radical 14 to give radical 16, which further transforms to radical 17 after the addition of sulfur dioxide. Finally, HAT happens between 15 and 17, yielding quaternary ammonium salt 18 and product 12, respectively.</p><!><p>Mechanism of the synthesis of sulfamoylation product 12.</p><!><p>In 2017, Chen and colleagues [47] accomplished the cyclization through the EDA complex formed by N-tosyl-2-vinylaniline (19) and the Umemoto reagent 20 (CF3 radical source) in CH2Cl2 under blue LED irradiation. In the presence of base, 21 was produced with 98% yield after degassing. Along with straightforward posttreatment, the corresponding reduction product 22 can be afforded easily (Scheme 7). This procedure offers a novel cyclization method with bifunctionalization, causing a multicomponent reaction of vinylaniline, halide, and sulfonylate to give corresponding indole derivatives. Furthermore, a wide variety of applicable substrates and good functional-group tolerance are provided by this approach, yielding multiple indole analogues with biological activity.</p><!><p>Synthesis of indole derivative 22 initiated by an EDA complex.</p><!><p>In 2017, Liang and Bi [56] reported a visible-light-induced three-component cyclization of ethyl acetoacetate (23), perfluoroalkyl iodides 24, and guanidine hydrochloride (25) via a halogen-bond adduct. The first light-promoted three-component reaction has been realized by a halogen-bond adduct, forming perfluoroalkylated pyrimidines 26 (Scheme 8). A variety of perfluorinated chains were assembled with methylene compounds and guanidines or amidines, giving the corresponding perfluoroalkylated pyrimidines in good to excellent yield.</p><!><p>Synthesis of perfluoroalkylated pyrimidines 26 initiated by an EDA complex.</p><!><p>In 2017, Chen and colleagues [34] prepared the phenanthridine derivative 29 with CFL (25 W) irradiation of a solution containing 27, perfluoroalkyl iodide 28, amine additive N,N,N´,N´-tetraethylethylenediamine (TEEDA) in THF (Scheme 9). TEEDA and perfluoroalkyl iodide form a halogen-bond adduct, and then light-induced electron transfer happens in order to give a perfluoroalkyl radical. The protocol can realize alkene- and alkyne iodide perfluoroalkylation and C–H perfluoroalkylation of electron-rich heteroaromatic hydrocarbons, providing a novel protocol for the synthesis of perfluoroalkyl-substituted phenanthridines.</p><!><p>Synthesis of phenanthridine derivative 29 initiated by an EDA complex.</p><!><p>In 2018, Sundén and Hsu [31] proposed a method of adding an α-aminoalkyl radical to maleimide via an EDA complex based on previous work (Scheme 10). The corresponding products can be given in good yield by modifying substituents on the N-alkyl moiety in 31 or N,N-dimethylaniline (30). This approach utilizes N,N-dimethylaniline (30) as electron donor and N-methylmaleimide (31) as electron acceptor to form an EDA complex, so that single-electron transfer occurs under ultraviolet-light irradiation. Subsequently, intermolecular proton transfer takes place, giving radicals 33 and 34. Radical 34 is quenched by oxygen, and radical 33 attacks 31 in order to form radical 35. Intermediate 36 is achieved by cyclization of radical 35, followed by hydrogen-atom removal, providing the cis-tetrahydroquinoline 32 (Scheme 11). It is worth noting that the EDA complex not only undergoes charge transfer but also proton transfer in this approach. The optimization experiment showed that the wavelength of the light source must overlap with the absorption spectrum of the EDA complex. Most importantly, given that the best yield was achieved when the molar concentration of 30 was 7 times that of 31, the concentration of the EDA complex was essential for a high reaction rate.</p><!><p>Synthesis of cis-tetrahydroquinoline derivative 32 initiated by an EDA complex.</p><p>Mechanism of the synthesis of cis-tetrahydroquinoline derivative 32.</p><!><p>In 2018, Yu and colleagues [22] discovered a method that employed O-aryloxime 37 and triethylamine as substrates at room temperature and blue-light irradiation to give phenanthridine 38 (also including quinoline products, Scheme 12). In this way, a nitrogen-centered radical was given via the EDA complex that was initiated by single-electron transfer, accomplishing the synthesis of a variety of highly functionalized nitrogen-containing aromatics with excellent yield.</p><!><p>Synthesis of phenanthridine derivative 38 initiated by an EDA complex.</p><!><p>In 2019, Fu and colleagues [35] received the target product 40 with 23 W CFL irradiation of a solution containing 1-(4′-hydroxy-[1,1′-biphenyl]-2-yl)ethanone O-(2,4-dinitrophenyl)oxime (39) and 1,8-diazabicyclo[5.4.0]undec-7-ene (DBU) in CH3CN (Scheme 13). Ready-made DBU serves two roles: base and electron donor. Furthermore, due to the commercially available material and wide range of substrates, this approach has major significance for drug scaffold methodologies, providing useful strategy for the synthesis of spiropyrrolines.</p><!><p>Synthesis of spiropyrroline derivative 40 initiated by an EDA complex.</p><!><p>In 2019, Yu and colleagues [23] reported the utilization of tetramethylethylenediamine (TMEDA) as additive in the EDA complex formation with perfluoroalkyl iodides 42 to afford 2-perfluoroalkylbenzothiazole products 43, employing blue LEDs (25 W) as irradiation source (Scheme 14). Notably, as (2-isocyanophenyl)(methyl)selane was exploited instead of substrate 41, new fluoroalkylbenzoselenazole derivatives with biological potential could also be given successfully. Furthermore, several perfluoroalkyl iodides ICnF2n+1 (n = 3–8,10) and four other fluoroalkyl iodides, including ICF(CF3)2, ICF2COOEt, ICF2CF2Cl, and ICF2CF2Br were reacted smoothly with 41, affording the corresponding products in good to excellent yield. The protocol conforms to the characteristics of green and environmental conservation, having a reaction time of only 1 hour, achieving the purpose of saving energy.</p><!><p>Synthesis of benzothiazole derivative 43 initiated by an EDA complex.</p><!><p>In 2019, Liang and colleagues [24] reported a method for preparing perfluoroalkyl-s-triazine via visible-light-promoted [5 + 1] cyclization initiated by an EDA complex. Perfluoroalkyl-s-triazine derivative 45 was synthesized by the reaction of biguanidine derivative 44 and perfluoroalkyl iodide 28 in the presence of sodium hydroxide (Scheme 15). Considering that oxygen played an indispensable role in the process, the authors supposed that it may facilitate single-electron transfer between biguanidine anion and perfluoroalkyl halide. By constructing two C–N bonds at the same time, the triazine heterocyclic structure that is commonly utilized in medical and material fields was accomplished by [5 + 1] cyclization.</p><!><p>Synthesis of perfluoroalkyl-s-triazine derivative 45 initiated by an EDA complex.</p><!><p>In 2020, Taylor and colleagues [26] proposed a reaction for the preparation of spirocyclic indoline derivative 47 from indolylynone 46 and thiophenol under blue-light irradiation (Scheme 16). An abundant range of products was given to test various indole-tethered ynones and thiols, confirming that the reaction is broad in scope. Remarkably, C–S bonds and spiro compounds have been constructed simultaneously in this approach, which are promising for drug synthesis. Substrate 46 comprises both indole and alkynone groups, leading to light-promoted intramolecular electron transfer in order to form radical intermediate 48. Then, 48 absorbs the hydrogen atom of thiophenol, yielding a thiophenol radical. The addition of the thiophenol radical to alkynone forms radical intermediate 49. Radical intermediate 50 is given due to the cyclization of radical intermediate 49, followed by abstracting hydrogen atoms from thiophenol to produce the final product 47 (Scheme 17).</p><!><p>Synthesis of indoline derivative 47 initiated by an EDA complex.</p><p>Mechanism of the synthesis of spirocyclic indoline derivative 47.</p><!><p>In 2020, Alemán and colleagues [53] proposed an approach in which ketene 48 and diene 49 condense with the help of diamine 51 to afford cyclobutane product 50 (Scheme 18). The reaction could be catalyzed by a simple diamine due to the fact that diamine can condense with the enol substrate, forming an imine-ion intermediate absorbing in the visible region. The direct excitation of the intermediate leads to a charge-transfer excited state, completing the stereocontrolled intermolecular cycloaddition reaction with a good ratio of enantiomer to diastereomer.</p><!><p>Synthesis of cyclobutane product 50 initiated by an EDA complex.</p><!><p>A plausible mechanism is shown in Scheme 19. In the presence of acid, EDA complex 52 is formed by ketene 48 and diamine 51. Then, the ground state 52 transforms into excited state 53 or into unproductive charge-transfer excited state 54 that can restore ground state 52 by BET; moreover, 53 and 54 can be mutually transformed. Excited state 53 reacts with diene 49, forming a double radical intermediate 55 that subsequently evolves to cyclobutyliminium ion 56, and then product 50 is provided after hydrolysis, along with the release of diamine 51.</p><!><p>Mechanism of the synthesis of spirocyclic indoline derivative 50.</p><!><p>In 2020, Zhang and colleagues [25] developed a visible-light-induced [3 + 2] cycloaddition reaction between glycine derivatives 57 and aryl epoxides 58, which can efficiently prepare a series of multisubstituted 1,3-oxazolidines 59 at room temperature (Scheme 20). The strategy can be applied smoothly to an extensive range of glycine derivatives, including electron-donating or electron-withdrawing substituent groups in the para- or meta positions at the benzene rings, giving corresponding products in moderate yield. This protocol is also suitable for the structural diversity of epoxides, providing a new activation approach for C(sp3)–H-functionalization of glycine derivatives.</p><!><p>Synthesis of 1,3-oxazolidine compound 59 initiated by an EDA complex.</p><!><p>As a crucial element in the construction of various organic scaffolds, the formation of C–C bonds remains a hot topic in the field of synthetic organic chemistry. The conventional approaches of C–C-bond construction typically employ transition-metal catalysts, such as in the Suzuki–Miyaura and Heck coupling reactions. Methodologies for forming different C–C bonds have recently been developed in the field of single-electron chemistry [63–65]. Considering that EDA-complex-initiated free-radical reactions are carried out under mild conditions, more attention has been paid to this efficient strategy for C–C-bond formation.</p><p>In 2015, Yu and colleagues [27] proposed a method for direct C–H trifluoromethylation of aromatic hydrocarbons through an EDA complex. Trifluoromethylated product 61 was synthesized by employing tryptamine derivative 60 and Umemoto reagent 20 as substrates as well as N-methylmorpholine (NMM) as organic base additive at room temperature (Scheme 21). The highly functionalized indole, pyrrole, benzofuran, and electron-rich benzene containing CF3 can be given in good yield. Given the redox potential of NMM and Umemoto reagent, directly conducting thermodynamic intermolecular SET is impossible. Thus, it is worth noting that the SET in this approach can be carried out only at room temperature without visible light.</p><!><p>Synthesis of trifluoromethylated product 61 initiated by an EDA complex.</p><!><p>In 2015, Melchiorre and colleagues [28] proposed a visible-light-induced reaction to synthesize indole alkylation product 64 by exploiting the EDA complex formed by 3-methylindole (62) and 2,4-dinitrobenzyl bromide (63), with 2,6-dimethylpyridine as additive at room temperature (Scheme 22). The substrates with different substituents at position C2 and C3 of indole have been synthesized smoothly, including cis-fused pyrrolo- and furanoindolines. The X-ray single-crystal analysis showed that the EDA complex is received in the form of a face-to-face π–π complex, and the ratio of donor to acceptor is 1:1.</p><!><p>Synthesis of indole alkylation product 64 initiated by an EDA complex.</p><!><p>In the same year, based on previous experimental phenomena and data, Melchiorre and colleagues [45] designed a reaction, with indanone derivatives 65 and perfluorohexyl iodide (66) as substrates and a phase-transfer catalyst (PTC) to give perfluoroalkylation product 67 under white-light irradiation (Scheme 23). A variety of electron-withdrawing substituents on the aromatic ring of 65 were well tolerated; however, the presence of electron-donating substituents lowered the reactivity due to a negative impact on the EDA complex formation and led to a low yield. It is worth noting that the phase-transfer catalyst employed in this experiment is a suitable donor for the photosensitive EDA complex while at the same time providing effective asymmetric induction in the capture of the resulting radical along with enantioselectivity of the product.</p><!><p>Synthesis of perfluoroalkylation product 67 initiated by an EDA complex.</p><!><p>In 2016, based on the experimental work in 2015 [27], Yu and colleagues [29] reported a type of EDA complex that could complete the hydrotrifluoromethylation of unactivated olefins and alkynes. This approach employed 68 and Togni reagent 69 (electron acceptor) as substrates, NMM as electron donor, and pyrrolidin-2-one as solvent to give hydrotrifluoromethylated product 70 at room temperature (Scheme 24). CF3 was added to a variety of terminal alkenes, leading to corresponding hydrotrifluoromethylation products with moderate to good yield.</p><!><p>Synthesis of hydrotrifluoromethylated product 70 initiated by an EDA complex.</p><!><p>In 2017, Yu and colleagues [30] proposed an EDA-complex-induced alkyne trifluoromethylation reaction. The EDA complex formed by a catalytic quantity of Togni reagent 69 and NMM initiated the chain propagation, causing the final alkyne trifluoromethylation (Scheme 25). A variety of olefins, such as ene carbamates, styrene, aliphatic olefins, vinyl ethers, and acrylates are compatible in this approach, affording corresponding β-(trifluoromethyl)alkynes with good to excellent yield. The bifunctionalization was achieved by an EDA-complex-initiated three-component reaction in the absence of light, which is of great significance for the later study of temperature-driven EDA complexes.</p><!><p>Synthesis of β-trifluoromethylated alkyne product 71 initiated by an EDA complex.</p><!><p>In 2017, König and colleagues [42] discovered an EDA complex 75 formed by bromothiophene 72, aniline (73), and N,N-diisopropylethylamine (DIPEA) as organic base additive to give corresponding thiophene radical 76 and aniline radical cation under irradiation with light. Then, 76 reacted with 73, giving rise to corresponding radical 77. Finally, product 74 was given via hydrogen atom transfer (Scheme 26). In contrast to (hetero)aryl halides with indispensable electron-withdrawing groups, the scope of the reaction comprises anilines including electron-withdrawing or electron-donating substituents in the arene, except N-acetylated or ortho-halogenated anilines.</p><!><p>Mechanism of the synthesis of 2-phenylthiophene derivative 74.</p><!><p>In 2017, Chen and colleagues [21] reported a method that promoted the formation of an alkoxy radical through an EDA complex under visible-light irradiation. The EDA complex formed by electron donor Hantzsch ester (HE) 79 and electron acceptor N-acyloxybenzamide 78 was produced by light-promoted SET, providing alkoxy radicals that could give carbon radicals by removing one molecule of acetaldehyde (Scheme 27). It is worth noting that EDA complex has been firstly employed for the generation of alkoxy radicals under visible-light irradiation, achieving selective C(sp3)–C(sp3)-bond cleavage and allylation or alkenylation.</p><!><p>Synthesis of allylated product 80 initiated by an EDA complex.</p><!><p>In 2017, Li and colleagues [57] reported a reaction for the synthesis of trifluoromethyl alkynylation product 84 with alkynyl sulfone 83, alkene 81, and Togni reagent 82 as substrates catalyzed by 2,4,6-trimethylpyridine (TMP). The EDA complex formed by electron donor TMP and electron acceptor Togni reagent 82 facilitated electron transfer, yielding trifluoromethyl radical to initiate the subsequent reaction (Scheme 28). On account of a wide range of substrates and functional group compatibility, this protocol can be exploited to assemble various β-trifluoromethylated alkynes by three-component reaction without transition-metal catalysis.</p><!><p>Synthesis of trifluoromethyl-substituted alkynyl product 84 initiated by an EDA complex.</p><!><p>In 2018, Xu and colleagues [48] proposed the dearomatization of β-naphthol promoted by visible light via intermolecular charge transfer (Scheme 29). In this method, β-naphthol anion 87 (β-naphthol 85 formed in the presence of base) is employed as electron donor to form EDA complex with electron acceptor perfluoroalkyl iodide 28. Single-electron transfer occurs under white-light irradiation, leading to an electron-deficient fluoroalkyl radical. Thereafter, fluoroalkyl radical is captured by 87, affording radical 88 that gives rise to radical intermediate 89 by uptake of iodine. Finally, the dealkylation product 86 is given by removing iodide anion (Scheme 30). Various β-naphthols with different substituents in the 1- and 3-positions were tolerated, providing the corresponding products with excellent yield.</p><!><p>Synthesis of dearomatized fluoroalkylation product 86 initiated by an EDA complex.</p><p>Mechanism of the synthesis of dearomatized fluoroalkylation product 86.</p><!><p>In 2018, Chen and colleagues [18] reported a method to realize C(sp3)–H allylation by generating aryl carboxyl radical from EDA complex based on previous work in 2017 [21]. The reaction is initiated by the formation of EDA complex between electron acceptor N-acyloxyphthalimide 90 and electron donor HE 79, completing regio- and chemoselective C(sp3)–H allylation or olefin bifunctionalization (Scheme 31).</p><!><p>Synthesis of C(sp3)–H allylation product 91 initiated by an EDA complex.</p><!><p>In 2018, Tang and Studer [32] found a bifunctional group reaction of perfluoroalkylation and β-alkenylation by a perfluoroalkyl radical. Perfluoroalkylation product 93 was synthesized by utilizing (E)-3-methyl-1-phenylhept-1,6-dien-3-ol (92) and perfluorobutyl iodide (28) as substrates and potassium phosphate and DABCO as additives at 50 °C and under irradiation with light (Scheme 32). The reaction is compatible with phenyl substituents with high steric hindrance, indicating that steric effects of the aryl moiety in the migrating styrenyl group do not play a major role.</p><!><p>Synthesis of perfluoroalkylation product 93 initiated by an EDA complex.</p><!><p>In 2018, You and colleagues [19] reported the discovery of an EDA complex formed by indole derivative 94 and Umemoto reagent 20, which provided the trifluoromethyl-substituted spirocyclic indolene 95 with stereoisomeric center in good yield (up to 90%) under blue-light irradiation (Scheme 33). A variety of groups have been tolerated at the C2 position of indole, including phenyl groups with electron-donating or electron-withdrawing substituents, as well as the alkyl moiety. To further investigate the practicability of this approach, 95 was synthesized smoothly on the gram scale with a yield of 70%.</p><!><p>Synthesis of spirocyclic indolene derivative 95 initiated by an EDA complex.</p><!><p>In 2019, Czekelius and colleagues [43] found that the perfluoroalkylation of unactivated olefins can be realized with phosphine catalyst and perfluorobutyl iodide (28) under visible-light irradiation. The EDA complex formed by perfluorobutyl iodide (28) and phosphine catalyst induced SET, affording a perfluoroalkyl radical, and then perfluoroalkylation product 97 was yielded by addition of perfluoroalkyl radical with olefin 96 (Scheme 34). Upon the termination of the reaction, the desired product can be given by removing the solvent and precipitating the catalyst. The comparison experiments and electronic absorption spectra showed that the efficiency of the catalyst was related to enhancement of selective absorption, considering the use of a visible-light source.</p><!><p>Synthesis of perfluoroalkylation product 97 initiated by an EDA complex.</p><!><p>In 2019, Glorius and colleagues [11] proposed a method of employing an EDA complex formed by indole derivative 98 and the Katritzky salt 99 as well as morpholine as organic base to obtain alkylated indole derivatives 100 under blue-light irradiation (Scheme 35). Given the UV–vis spectra of the reaction mixture and its components, there was no evidence of the formation of a ternary EDA complex between Katritzky salt 99, 98, and morpholine. Moreover, the TDDFT calculation showed that electron transfer took place in complex 101 under visible-light excitation. Hence, it can be inferred that radical-chain propagation was initiated by a small amount of radicals that emerged from excited complex 101. The C–N bond in dissociated radical 103 is irreversibly broken, along with the appearance of radical 104. Subsequently, alkyl radical 104 is captured by indole 98, giving benzyl radical 105. The alkylated indole derivative 100 and morpholine salts are provided via proton-coupled electron transfer (PCET) with EDA complex 102 formed by morpholine and 99 (Scheme 36). As a rare example of EDA photochemistry, two kinds of EDA complexes were involved in this approach, explaining the reason why the yield increased significantly when morpholine was employed as an organic base additive, which was exploited in the screening stage of the reaction conditions.</p><!><p>Synthesis of alkylated indole derivative 100 initiated by an EDA complex.</p><p>Mechanism of the synthesis of alkylated indole derivative 100.</p><!><p>In 2019, Xia and colleagues [46] reported that the EDA complex formed by aryl halide 106 and oxindole 107 under alkaline conditions allowed single-electron transfer under irradiation with light, eventually affording arylated oxidized indole product 108 (Scheme 37). This reaction provides an effective method to construct various 3-arylindoles with medicinal value at ambient temperature, which has a wide range of substrates, including various (hetero)aryl halides and substituted oxindoles.</p><!><p>Synthesis of arylated oxidized indole derivative 108 initiated by an EDA complex.</p><!><p>In 2019, Gilmour and colleagues [12] transformed the classical Stetter reaction into a radical approach, solving the long-standing problem of chemical selectivity to convert α,β-unsaturated aldehydes selectively into 4-ketoaldehydes (Scheme 38). The imine salt (electron acceptor) that forms EDA complex 112 with electron donor α-keto acid 109 is synthesized by secondary amine catalyst and α,β-unsaturated aldehyde 110. Compound 112 turns to excited state 112* under irradiation with light. Then, radical intermediate 113 is afforded via intermolecular electron transfer, followed by removing one molecule of carbon dioxide to give radical intermediate 114. Species 115 is formed through radical coupling in 114, providing the target product 111 with the release of the secondary amine catalyst (Scheme 39).</p><!><p>Synthesis of 4-ketoaldehyde derivative 111 initiated by an EDA complex.</p><p>Mechanism of the synthesis of 4-ketoaldehyde derivative 111.</p><!><p>In 2019, Kappe and colleagues [36] reported a method to complete perfluoroalkylation of olefins under visible light in flow. Perfluoroalkylated olefin 118 was prepared by employing olefin 116 and perfluoroalkyl iodides 117 as substrates as well as triethylamine as additive at 20 °C and under 405 nm irradiation (Scheme 40). A standard residence time of 5 min was required for the full conversion via the EDA complex that formed by alkene and perfluoroalkyl iodide in flow, while longer residence times were requisite for less reactive alkenes. Moreover, the yield of this reaction can reach 7.6 g ⋅ h−1 on a gram scale, indicating that the flow step is promising in photochemistry.</p><!><p>Synthesis of perfluoroalkylated olefin 118 initiated by an EDA complex.</p><!><p>In 2019, Aggarwal and colleagues [38] employed Katritzky salt 119 as electron acceptor and HE 79 as electron donor to form an EDA complex, providing the corresponding alkyl radical that could react with olefin 120 with an electron-withdrawing group to give alkylation product 121 under irradiation with light (Scheme 41). The reaction is compatible with various substrates, including alkenes, secondary alkylpyridinium ions, benzylic pyridinium ions, and primary alkylpyridinium ions, which can be considered an effective method for the generation of alkyl radicals without catalyst.</p><!><p>Synthesis of alkylation product 121 initiated by an EDA complex.</p><!><p>In 2019, Yu and Zhang [15] reported a radical acylation reaction initiated by an EDA complex promoted by visible light. Imine 122 was employed as electron acceptor with α-keto acid 109 as electron donor to form the EDA complex, affording acylation product 123 under blue-light irradiation (Scheme 42). The quantum yield of the reaction was determined to be 0.08, suggesting that the reaction proceeded via radical coupling rather than a radical propagation. Moreover, the reaction was compatible with amides, cyanides, esters, ethers, halides, and heterocycles, and various α-aminoketones (32 examples) can be yielded in 90% isolated yield. According to the author, the EDA complex had a six-membered-ring transient state, and the imine also acted as an organic base (abstracting proton from α-keto acid), proving that electron transfer is accompanied by proton transfer in the process (Scheme 43).</p><!><p>Synthesis of acylation product 123 initiated by an EDA complex.</p><p>Mechanism of the synthesis of acylation product 123.</p><!><p>In 2020, Stephenson and colleagues [14] employed 2-methoxynaphthalene (124) and acylated ethyl isonicotinate N-oxide obtained from 125 and trifluoroacetic acid anhydride to form EDA complexes for the preparation of trifluoromethylated product 126 (Scheme 44). As a rare example of EDA photochemistry in the catalytic system, only a catalytic equivalent of the electron donor was employed in this approach. Further experiments showed that the addition of inorganic salts, calcium chloride and lithium chloride, could increase the absorption of EDA complex in the visible-light region.</p><!><p>Synthesis of trifluoromethylation product 126 initiated by an EDA complex.</p><!><p>In 2020, Xu and colleagues [39] proposed a visible-light-promoted alkylation reaction using Katritzky salts such as 128 and glycine derivative 127 (or glycine segments in peptides) initiated by an EDA complex. This successfully realized the simple synthesis of unnatural α-amino acids 129 and precise alkylation modification of peptides in the later stage (Scheme 45). Even in the presence of other amino acid residues, this protocol has excellent regio- and chemoselectivity, providing a sequence of novel corresponding dipeptides with good yield.</p><!><p>Synthesis of unnatural α-amino acid 129 initiated by an EDA complex.</p><!><p>C–S bonds are commonly present in amino acids, proteins, glycosides, nucleic acids, and other biological macromolecules. In recent years, photocatalyst- and transition-metal strategies have been employed to construct C–S bonds [66–69]. The C–S bond synthesis via EDA-complex pathways has the advantages of mild reaction conditions and a high tolerance to functional groups, which can be exploited for artificial syntheses of biological macromolecules.</p><p>In 2017, Miyake and colleagues [54] designed a type of C–S cross-coupling reaction initiated by an EDA complex promoted by visible light. In this approach, halogenated aromatic 130 was employed as electron acceptor with thiophenol (131) as electron donor to form an EDA complex. Light-promoted intermolecular electron transfer took place to give corresponding radicals, respectively, in the presence of base, and then cross-coupling between radicals was carried out, affording thioether derivative 132 (Scheme 46). It has been proved by UV–vis spectroscopy and TDDFT calculations that the EDA complex was formed between an electron-rich mercaptan anion and electron-deficient aryl halides. Most importantly, this approach can be successfully applied to the gram scale, providing a step towards assorted aryl sulfide structural units with medicinal value.</p><!><p>Synthesis of thioether derivative 132 initiated by an EDA complex.</p><!><p>In 2019, Yang and colleagues [52] developed a method for preparing S-aryl dithiocarbamates 135 by a multicomponent reaction of an EDA complex under visible-light irradiation (Scheme 47). A number of aryl halides reacted smoothly, providing moderate to good yields for analogous S-aryl dithiocarbamates. To further demonstrate the synthetic application of this protocol, a gram scale of 135 has been tested, giving a yield of 72%. By constructing C–N- and C–S bonds simultaneously in one step without any transition-metal catalyst, ligand, or photocatalyst, this method possesses a splendid application prospect.</p><!><p>Synthesis of S-aryl dithiocarbamate product 135 initiated by an EDA complex.</p><!><p>The reaction mechanism is as follows (Scheme 48): Firstly, carbon disulfide combines with N-methylaniline (134) in the presence of Cs2CO3 to form thiolate 136. Thiolate 136 is employed as electron donor to generate EDA complex 137 with halogenated electron-acceptor aromatics 133, and then electron transfer occurs, affording intermediate 138. Finally, radical coupling gives rise to S-aryl dithiocarbamate product 135.</p><!><p>Mechanism of the synthesis of S-aryl dithiocarbamate product 135.</p><!><p>In 2019, Liao and colleagues [13] utilized Katritzky salt 139 and thiobenzoic acid (140) to form an EDA complex, providing thioether derivative 141 with DIPEA as an organic base additive (Scheme 49). This reaction offers a novel and simple approach for the synthesis of α-mercapto acid derivatives under mild reaction conditions and demonstrates strong compatibility to the functional group. In addition, a gram-scale reaction also gives the desired thioether product in a yield of 99%.</p><!><p>Synthesis of thioether product 141 initiated by an EDA complex.</p><!><p>The C–B bond can be converted into a wide range of other functional groups by the conversion of alkyl borane [70–72]. Hence, it has become imperative to pursue more efficient syntheses for constructing C–B bonds. In recent years, the construction of C–B bonds via EDA complexes has attracted more chemists' attention.</p><p>In 2017, Glorius and colleagues [51] discovered a visible-light-induced decarboxylated borate of arylcarboxylic acid initiated by an EDA complex. First, the N-hydroxyphthalimide (NHPI) ester 142 is excited to electron acceptor 142* through visible-light intersystem crossing (ISC); diborate 143 combining with pyridine results in electron donor 145. Upon the formation of the EDA complex between 145 and 142*, electron transfer occurs, giving radical 146 and radical cation 147, respectively. Finally, radical 146 undergoes decarboxylation to afford an aryl radical and then combines with radical cation 147, yielding product 144 (Scheme 50). It should be noted that only when NHPI is firstly activated can it turn into an electron acceptor, and thus further combines with the electron donor to form an EDA complex, mainly due to the fact that the electron acceptor in an excited state allows for stronger oxidation, to integrate with the electron donor.</p><!><p>Mechanism of the synthesis of borate product 144.</p><!><p>In 2018, Glorius and colleagues [16] reported a method for the preparation of boron-substituted product 148 by employing Katritzky salt 119 as electron acceptor as well as the complex formed by bis(catecholato)diboron (B2cat2) and solvent DMA as electron donor to afford an EDA complex (Scheme 51). This approach can effectively convert primary-, benzyl-, and secondary amines into corresponding borated products, with only a coordinating solvent, DMA. Furthermore, functionalization of natural products and drug molecules has been accomplished smoothly with excellent yields.</p><!><p>Synthesis of boronation product 148 initiated by an EDA complex.</p><!><p>In 2018, Aggarwal and colleagues [17] proposed a method for preparing alkyl borate derivative 150 by employing Katritzky alkylpyridinium salt 149 and B2cat2 as substrates as well as DMA as coordination solvent under blue-light irradiation, followed by subsequent reaction with pinacol to afford boration product 151 (Scheme 52). A number of secondary alkylamines, even those that have carbamate- or phthalimide-protected amines, have been efficiently transformed to suitable pinacol boronic esters. This simple operation without transition-metal catalysis will be widely promoted in the synthesis of important boron-containing molecules in medicine and biology.</p><!><p>Synthesis of boration product 151 initiated by an EDA complex.</p><!><p>In 2019, Aggarwal and colleagues [55] proposed that the EDA complex was formed by 2-iodophenyl thiocarbonate 152, bis(catecholato)diboron, and triethylamine, which afforded boronic acid ester derivative 153 under blue-light irradiation. Simultaneously, pinacol boronic acid ester derivative 154 can be yielded by subsequent processing (Scheme 53). The protocol reveals a high functional-group tolerance that permits the transformation into boronic esters of natural alcohol products with high stereocontrol.</p><!><p>Synthesis of boronic acid ester derivative 154 initiated by an EDA complex.</p><!><p>The development of efficient methods to construct C–N bonds is an essential scheme in organic synthesis due to its widespread presence in pharmaceutical-, agrochemical-, and materials sciences [73–75]. At present, most of the C–N-bonding reactions require transition-metal catalysis, and the reaction conditions are more stringent; however, the EDA-complex pathway proceeds under mild, catalyst-free conditions, promoted by irradiation with visible light.</p><p>In 2017, Shirke and Ramaastry [40] proposed an organic catalyzed β-azide reaction of ketene 155 initiated by the EDA complex formed by DABCO and Zhdankin reagent 156 (Scheme 54). A variety of β-azidoketones was conveniently obtained with good to excellent yield with electron-rich as well as electron-poor arenes and heteroarenes. Subsequently, in order to prove the practicability of this approach, 1,2,3-triazoles were assembled by reaction of 157 with alkynes.</p><!><p>Synthesis of β-azide product 157 initiated by an EDA complex.</p><!><p>In 2019, Bosque and Bach [41] reported that 3-acetoxyquinuclidine (q-OAc) could be utilized as an electron-donor catalyst to form an EDA complex with electron acceptor 158, and then a molecule of carbon dioxide was removed under 455 nm light irradiation, giving decarboxylation product 159 (Scheme 55). It was found that many ester groups can be activated by the structural motif of tetrachlorophthalimide in 158. Significantly, in contrast to most traditional EDA complex approaches that consume the DA pair, the electron-donor catalyst q-OAc in this method could be regenerated.</p><!><p>Decarboxylation reaction initiated by an EDA complex.</p><!><p>In 2019, Frontera and colleagues [49] obtained target product 162 with blue LEDs irradiation of a solution containing electron-poor N-aryloxyamides 160, indole derivative 161, and carbonate or other multicharge anions in CH3CN (Scheme 56). The corresponding products can be given in good yield by modifying substituents on the amide moiety in 160 or N-substituted indoles. Inorganic-base electron donors formed transient complexes with N-aryloxyamides, driven by noncovalent anion–π interactions, which has been described for the first time in a light-promoted process.</p><!><p>Synthesis of amidated product 162 initiated by an EDA complex.</p><!><p>Many compounds contain phosphorus, which has gained a high degree of interest in materials, agriculture, medical science, and biology [76]. Cases of C–P bond construction employing photoredox [77–78] or photoredox/nickel dual catalysis [79] have been identified in the field of photochemistry. However, here we introduce the methods initiated by EDA complexes for C–P bond construction.</p><p>In 2018, Lakhdar and colleagues [44] reported a visible-light-mediated synthesis approach of arylphosphonates initiated by an EDA complex. Diethyl phenylphosphonate (165) was given by exploiting diphenyliodonium trifluoromethanesulfonate (163) as electron acceptor, triethylphosphite (164) as electron donor, potassium carbonate as base, and CH3CN as solvent (Scheme 57). The complex is bound together by weak halogen bonds, in which phosphorus lone-pair electrons interact with σ* orbitals of C–I bonds. A variety of arylphosphonates can be directly afforded by the simple combination of diaryliodonium salts and phosphite esters. In addition, calculations including EPR, NMR, and DFT have been carried out to prove that the reaction mechanism is consistent with inference (Scheme 58).</p><!><p>Synthesis of diethyl phenylphosphonate 165 initiated by an EDA complex.</p><p>Mechanism of the synthesis of diethyl phenylphosphonate derivative 165.</p><!><p>Although there have been few cases of constructing C–O bonds and C–H bonds via EDA-complex pathways in recent years, we also summarized them in view of their great significance in organic synthesis.</p><p>In 2018, Miyake and colleagues [50] found a protocol for the preparation of (Z)-2-iodovinyl phenyl ether 168 by utilizing ethynylbenziodoxol(on)e (EBX) 167 and phenol derivative 166 (Scheme 59). Due to the lack of significant electronic effects of phenol, a variety of phenols, including electron-donor and electron-withdrawing groups, were been converted into corresponding 2-iodovinyl phenyl ethers in moderate to excellent yield with high regio- and stereoselectivities.</p><!><p>Synthesis of (Z)-2-iodovinyl phenyl ether 168 initiated by an EDA complex.</p><!><p>According to the analysis of the mechanism (Scheme 60), a molecule of phenol anion is first added to the alkyne group of an EBX, forming electron acceptor 169, which causes the destabilization of the C–I bond. Then, electron acceptor 169 forms an EDA complex with phenol anion, along with light-promoted electron transfer occurring. Thereby, the C–I bond and the I–O bond break to afford the final product (Z)-2-iodovinyl phenyl ether 168. The electron acceptor can only be provided by the addition of phenol to the EBXs since an EDA complex cannot be directly formed from the original substrates, which means that the effect of the two-component ratio of the EDA complex must be taken into account.</p><!><p>Mechanism of the synthesis of (Z)-2-iodovinyl phenyl ether derivative 168.</p><!><p>In 2019, Rathnayake and Weaver, III [37] designed a method of visible-light-promoted EDA-complex-mediated dehalogenation of haloalkanes 170. The dehalogenation product 171 was afforded based on the presence of the EDA complex formed by DIPEA and haloalkanes 170 under blue-light irradiation (Scheme 61). It was worth mentioning that longer reaction times and increased DIPEA loading were required owing to the inactivity of α-bromoketones, esters, and nonactivated sulfones; however, corresponding products could be given in high yield. When the DIPEA molarity was double that of haloalkanes, the highest yield was given. Since a marked yellow color appeared immediately upon mixing substrates, the existence of an EDA complex could be confirmed by UV–vis spectroscopy.</p><!><p>Dehalogenation reaction initiated by an EDA complex.</p><!><p>In this review, reactions and mechanisms of EDA complexes were discussed from the aspects of cyclization reactions, C–C-, C–S-, C–B-, C–N-, C–P-, C–O-, and C–H bond formations. The absence of transitional-metal catalysts and photosensitizers is the most profound feature of EDA-complex- mediated reactions in most cases. On the other hand, the reaction conditions are mild, and light is utilized as the only external energy source, which is consistent with the theme of green chemistry. However, the comprehension of EDA complexes was established relatively late, mainly owing to the fact that the formation of EDA complexes was regarded as a unique chemical reaction rather than a branch of photochemistry; in addition, for the sake of avoiding BET processes, reactions involving EDA complexes require substrates with corresponding leaving groups, which also significantly limits the development of EDA complexes. In conclusion, although the research on EDA complexes is still in the initial stage, with many challenges to be solved in response, there is no doubt that the future of green chemical synthesis will surely have a very wide prospect for this strategy.</p>

PubMed Open Access

First DMAP-mediated direct conversion of Morita–Baylis–Hillman alcohols into γ-ketoallylphosphonates: Synthesis of γ-aminoallylphosphonates

An efficient synthesis of a series of γ-ketoallylphosphonates through a direct conversion of both primary and secondary Morita-Baylis-Hillman (MBH) alcohols by trialkyl phosphites with or without DMAP, used as additive, and under solvent-free conditions, is described herein for the first time. Subsequently, a highly regioselective Luche reduction of the primary phosphonate 2a (R = H) gave the corresponding γ-hydroxyallylphosphonate 5 that further reacted with tosylamines in the presence of diiodine (15 mol %) as a catalyst, affording the corresponding S N 2-type products 6a-d in 63 to 70% isolated yields. Alternatively, the alcohol 5 produced the corresponding acetate 7 which, mediated by Ce(III), was successfully converted into the corresponding γ-aminoallylphosphonates 8a-d.

first_dmap-mediated_direct_conversion_of_morita–baylis–hillman_alcohols_into_γ-ketoallylphosphonates

Introduction<!>Results and Discussion<!>Conclusion

<p>Phosphonates and their derivatives are an important class of substances that have a wide range of applications in numerous areas such as medicinal [1][2][3] and agricultural chemistry [4,5]. Among them, multifunctional derivatives have been exploited as valuable building blocks in natural product syntheses, e.g., calyculins A and B as potent serine-threonine protein phosphatase inhibitors [6], as well as ligands in enantioselective reactions [7]. Furthermore, allylphosphonates are important bioactive compounds that exhibit interesting antimicrobial and antimalarial properties [8,9], as well as useful substrates for the synthesis of valuable organic compounds [10][11][12].</p><p>Historically, the Michaelis−Arbuzov rearrangement [13] is the most widely and generally high yielding strategy for phosphonate synthesis. The current three step protocol involves firstly the mesylation of corresponding alcohols and then the conversion of the intermediate mesylates into their halides. Further an Arbuzov reaction of alkyl phosphites with such halides affords the phosphonates.</p><p>Interestingly, a direct conversion of common allyl alcohols into the corresponding phosphonates, with or without catalyst, was previously reported. Accordingly, Bodalski and co-workers [14] and then Swamy's research group [15] described a direct route to allylphosphonates by treatment of Morita-Baylis-Hillman alcohols with chlorophosphites without any additive, followed by thermal Arbuzov rearrangement of the intermediate allyl phosphites (Scheme 1, reaction 1).</p><p>Recently, an efficient protocol for the conversion of common allyl and benzyl alcohols into the corresponding phosphonates through their treatment with triethyl phosphite and ZnI 2 , was described [16]. Similarly, Das and co-workers [17] have directly converted acyclic Morita-Baylis-Hillman (MBH) alcohols into the corresponding allylphosphonates upon their treat-ment with trialkyl phosphite in the presence of FeCl 3 (Scheme 1, reaction 2).</p><p>On the other hand, treatment of MBH acetates with triethyl phosphite, without any additive, provided, after thermal Arbuzov rearrangement, a variety of diethyl allylphosphonates (Scheme 1, reaction 3) [18]. Moreover, the asymmetric allylic substitution of MBH carbonates with diphenyl phosphonate using chiral thiourea phosphite as catalyst, afforded the related allylphosphonates (Scheme 1, reaction 4) [19].</p><p>We have previously reported a direct nucleophilic allylic substitution of cyclic MBH alcohols by β-dicarbonyl compounds [20]. In addition, in our recent work [21], we have described an efficient protocol for the synthesis of a new series of allylphosphonates in high yields and selectivity, using cyclic MBH acetates as starting materials, in the presence of DMAP or imidazole as additives (Scheme 2, reaction 6).</p><p>Moreover, it has been demonstrated that the presence of a nitrogen atom and a phosphonate moiety in multifunctional compounds may improve their synthetic and biological potentials. Aminophosphonates and their derivatives [22][23][24][25][26] are thus recognized as promising compounds and a new class of drugs that are widely used in a variety of commercial applications. Indeed, they are known to influence various biochemical processes in plants, modifying or inhibiting them, or to have biological activities as for example antibiotics, antibacterial, anticancer or antithrombotic agents [22][23][24][25][26].</p><p>In this context, α-aminophosphonates have particularly attracted considerable attention owing to their biological activities [27][28][29][30][31][32][33] since they are considered as important surrogates for α-amino carboxylic acids, peptide mimics as well as versatile intermediates for the design of potential anticancer agents.</p><p>β-Aminophosphonates present also an important place among the various compounds containing both a P-C bond and an amino group. Indeed, they are known for their important role in medicinal chemistry [34][35][36] as anti-HIV agents, enzyme inhibitors and antibacterial agents. In addition, α-and β-aminophosphonates have been widely described in literature. However, γ-aminophosphonates have received much less attention. These derivatives were originally isolated from microorganisms [37,38], e.g., the fosmidomycin [39] is as an antimalarial drug that was isolated from broths of bacteria of the genus Streptomyces [40]. The γ-aminophosphonates were prepared through numerous synthetic methods [41][42][43][44][45]. For instance, they were prepared by reductive amination of γ-aminophosphonyl ketones using sodium borohydride [41], or by conjugate addition of diethyl methylphosphonite to 2-cyclohexenone followed by Bucherer-Bergs amino acid synthesis [42]. Another synthetic approach for a series of α-fluorinated-γ-aminophosphonates has been reported through a palladium-catalyzed hydrogenation of α-fluorovinylphosphonates [43].</p><p>In continuation of our interest [21] in the construction of carbon-phosphorus bonds using both cyclic and acyclic MBH adducts, we report herein, for the first time, a direct and facile access to γ-ketoallylphosphonates from primary and secondary MBH alcohols, with or without DMAP used as additive, and under solvent-free conditions (Scheme 1, reaction 5 and Scheme 2, reaction 7). Taking into consideration the importance of aminophosphonates (vide infra), we developed in the second part of this study an efficient conversion of these γ-ketoallylphosphonates into related γ-aminoallylphosphonates.</p><!><p>In our first attempt, a mixture of cyclic MBH alcohol 1a and triethyl phosphite was reacted in toluene without any additive.</p><p>After stirring the reaction mixture at 110 °C for 72 h, the starting materials were completely recovered (Table 1, entry 1). Under the previous conditions and in the presence of 0.5 equiv of DMAP, a sluggish reaction occurred, affording the desired phosphonate 2a within 72 h in a 30% yield (Table 1, entry 2). Using 1 equiv of DMAP in refluxing toluene required a longer reaction time of 120 h to completely convert the allyl alcohol 1a into the phosphonate 2a in 45% yield (Table 1, entry 3). The nucleophilic substitution of alcohol 1a with triethyl phosphite in THF at reflux was significantly less effective (Table 1, entry 4).</p><p>The TLC and 1 H NMR analyses of the crude reaction mixture conducted without any additive and under solvent-free conditions, at room temperature or at 80 °C, indicated that no reaction occurred and the starting materials were completely recovered (Table 1, entry 5).</p><p>However, when 0.5 equiv of DMAP was employed under the previous conditions, the phosphonate 2a was isolated within 4 h in 35% yield (Table 1, entry 6). After screening several amounts of DMAP, under solvent-free conditions at 80 °C, we observed that the best yield (75%) was obtained using 1 equiv of DMAP (Table 1, entry 7). Under these reaction conditions, the allylic phosphonate 2a was produced as the sole product within 1 h and its structure was confirmed by 1 H, 13 C and 31 P NMR analysis [21].</p><p>A series of γ-ketoallylphosphonates 2a-f was prepared from different cyclic MBH alcohols and triethyl phosphite using the optimized reaction conditions (Table 2). These adducts were obtained from primary (1a and 1f) or secondary (1b-e) MBH alcohols (R = linear/branched alkyl or aryl). The total conversion of alcohols 1a-f into phosphonates 2a-f was complete within 1-4 h in 58-84% yields. Our results suggested that this reaction worked well with six-membered cyclic MBH alcohols 1a-e (Table 2, entries 1-5), as well as five-membered cyclic MBH alcohol 1f (Table 2, entry 6).</p><p>A putative reaction pathway could start from a first β-conjugate addition of DMAP onto the allylic alcohol 1a (i), followed by the elimination of the hydroxide ion that would afford the intermediate I (ii). Similarly, a second β'-conjugate addition of triethyl phosphite onto I (iii), followed by the release of DMAP would provide the phosphonium intermediate II (iv). Finally, the hydroxide ion is expected to react with II via an Arbuzov rearrangement to provide the desired γ-ketoallylphosphonate 2a (v) (Scheme 3).</p><p>We next investigated the scope and the limitations of this synthetic method. We have thus examined, the behavior of the acyclic MBH alcohol 3a [46,47] towards triethyl phosphite under the above conditions (1 equiv of DMAP, solvent-free conditions, 80 °C, Scheme 4).</p><p>We were pleased to note that the nucleophilic substitution reaction worked also well and gave the allylphosphonate 4a in 90% yield within 1 h (Table 3, entry 1). More interestingly, repeating this reaction without DMAP and under solvent-free conditions at 80 °C gave, in only 30 min, the phosphonate 4a in 92% yield (Table 3, entry 2). Encouraged by these results, the behavior of a variety of acyclic MBH alcohols 3a-c towards trialkyl/triaryl phosphites was examined to determine the substrate scope of the present procedure (Table 3).</p><p>As shown in entries 3 and 4 (Table 3), the substitution of the primary alcohol 3a with a trialkyl/triaryl phosphites gave the corresponding phosphonates 4b and 4c in 60-62% yields.</p><p>The reaction of secondary alcohols 3b and 3c with trialkyl phosphites was also surveyed. Under the previous reaction conditions (solvent-free, 80 °C, without any additive) an S N 2'type reaction followed by an Arbuzov rearrangement gave the corresponding primary allylphosphonates 4d-f with a high Z-stereoselectivity (Z/E = 87-100/13-0) and in 56-92% yields (Table 2, entries 5-7). It seems that the Z/E ratios depend on the nature of the R allylic group. Indeed, when it is an alkyl group, i.e., R 1 = Me, the phosphonates 4d and 4e were obtained in high Z-diastereoselectivities (de = 74-88%), whereas the phosphonate 4f was provided with an excellent Z-diastereoselectivity (de = 100%) when R 1 = Ph. In this work, the observed high Z-diastereoselectities are in good agreement with a previous report by Basavaiah and co-workers on the nucleophilic addition of triethyl phosphite onto Morita-Baylis-Hillman acetates .</p><p>We envisaged in the second part of this study a further functionalization of these allylphosphonates as the literature survey revealed that the aminophosphonates are useful molecules in organic synthesis and in biology (vide supra).</p><p>The synthesis of γ-tosylaminophosphonates was achieved in a two-step sequence. First, the γ-hydroxyphosphonate 5 was prepared and isolated in a good yield (88%) via a highly selective Luche reduction [49,50] of the γ-ketophosphonate 2a using NaBH 4 in the presence of CeCl 3 •6H 2 O in methanol at 0 °C. The TLC of the reaction mixture and analysis of 1 H NMR and 31 P NMR spectra showed the exclusive formation of the 1,2adduct 5 (Scheme 5). The literature survey revealed that γ-hydroxyphosphonates may display interesting biological activities [51,52].</p><p>The next step was to realize the nucleophilic substitution of the γ-hydroxyphosphonate 5a with a variety of differently substituted tosylamines catalyzed by 15 mol % of iodine [53] in refluxing methylene chloride. Under these optimized conditions, the γ-tosylaminophosphonates 6a-d were obtained in good yields (63-70%) (Table 4, entries 1-4).</p><p>Scheme 6 illustrates a possible mechanism for the I 2 -catalyzed substitution reaction of γ-hydroxyallylphosphonate 5 with tosy-lamines. This could first involve an activation of the hydroxy moiety of adduct 5 through a coordination of the iodine catalyst with the hydroxy group to give an iodine-coordinated intermediate I that would subsequently undergo a nucleophilic attack by the tosylamine. The release of diiodine and elimination of a molecule of water is then expected to afford the desired γ-tosylaminoallylphosphonates 6.</p><p>However, this reaction failed using primary, secondary amines or aniline as nucleophiles. Therefore, an alternative synthesis of γ-aminophosphonates 8 has been developed. Following our previous report [54], we have first converted the MBH alcohol 5 into its acetate 7 in 90% yields using a mixture of Ac 2 O and DMAP as reagents (Scheme 7). The MBH acetate 7 was further converted into the γ-aminophosphonates 8a-c within short reactions times of 1-2 h and in 60-81% isolated yields using anilines, and 1 equiv of cerium(III) chloride hexahydrate in refluxing toluene (Table 5, entries 1-3). The nucleophilic allylic substitution of the acetate 7 worked also with β-naphthylamine and gave the corresponding γ-aminophosphonate 8d in only a modest yield (Table 5, entry 4). a Yields of isolated pure compounds after column chromatography.</p><!><p>We have developed a mild, direct and convenient procedure for the conversion of both cyclic and acyclic MBH alcohols, with trialkyl and triaryl phosphites, into γ-ketoallylphospho-nates under solvent-free conditions, in the presence or absence of the promoter DMAP. The corresponding products have been further involved in two alternative efficient synthetic routes for γ-amino-and γ-tosylaminophos-</p>

Beilstein

An In-Depth Analysis Approach Enabling Precision Single Chain Nanoparticle Design

The synthesis of single chain nanoparticles (SCNPs) is a vibrant field in macromolecular science, enabled by a rich variety of synthetic strategies to induce macromolecular chain folding. Due to the decrease of the hydrodynamic volume upon folding, SCNP formation is typically characterized by a shift towards higher elution volumes in size exclusion chromatography (SEC). However, a step-change in the methodologies for SCNP analysis is required for the in-depth understanding of the nature of intramolecular polymer folding and internal SCNP structure, which is critical to enable their application as catalytic nanoreactors.Herein, we exploit a unique combination of small-angle neutron scattering (SANS), 19 F NMR spectroscopy, and quadruple detection SEC to generate an encompassing and systematic view of the folded morphology of poly(tert-butyl acrylate) based-SCNPs as a function of their reactive group density (5, 15, and 30 mol%) and absolute molar mass (20, 50, 100 kDa). In addition to detailed morphological insights, we establish that the primary factor dictating the compaction of SCNPs is their reactive group density, which is ineffective below 5 mol%, reaching maximum compaction close to 30 mol%. The molar mass of the precursor polymers has a minor impact on how an SCNP compacts for molar masses above 20 kDa.

an_in-depth_analysis_approach_enabling_precision_single_chain_nanoparticle_design

Introduction<!>Results and Discussion<!>Linear Precursors<!>Single Chain Nanoparticles (SCNPs)<!>Figure 3. Synthetic strategy of the PFTR-based folding reaction (right box). Composition determination (left two boxes):<!>Composition of the SCNPs<!>𝐿𝐷 = (𝜒 𝑃𝐹𝑇𝑅 * %PFB)<!>Figure 4. Possible products of the folding reaction and their validation using SEC-, and 1 H NMR spectroscopy. Right box: The distinct ratio of the backbone-attached PFB methylene protons (orange) to the aliphatic protons of the crosslinker (green) allow to distinguish between loop-building or scavenged crosslinker. The aromatic resonances of the crosslinker (grey) are not suitable for this quantification due to overlap with the solvent resonances. The table shows the ratio of the peak areas of PFB signal (normalized to 1) to the aliphatic signals of the crosslinker. The ratio of ligation characteristic signals is approx. 0.5 (for 5%PFB with PFTR conversion 50%) and 1.0 (15% and 30%PFB with PFTR conversion 100%) confirming the PFTR conversion obtained from 19 F NMR spectroscopy. This is additionally confirmed by the Mn-increase (left box) after incorporation of the corresponding number of crosslinker molecules (determined by NMR spectroscopy) from precursor (red dots) to SCNP (colour coded dots) obtained by SEC.<!>Molar mass, size and conformation via advanced size exclusion chromatography<!>Figure 8. (a) KMH plots of the precursors (open dots) and their corresponding SCNPs (full dots) of 50 kDa series. (b) KMH exponents of the whole copolymer library depending on the crosslinks per chain. Data for precursors (open dots) and SCNPs (full dots)are presented. Limits for statistical coil and hard sphere are indicated in the plot and in the table within the figure. (c) Correlation between loop-length (stars) calculated from the LD and polymerisation degree vs SCNP conformation according to KMH slope (dots).<!>In-depth size and conformation validation via SANS<!>(d) Scaling exponent  (from SANS, dots) and KMH exponent  (from SEC-D4, trinagle) vs the ligation density (precursorsopen; SCNPs -full symbols). (e) Apparent density calculated according to Eq. 6 (precursorsopen; SCNPs -full dots) and the change of the density after folding vs. LD (crosses). (f) Contraction factors g (full dots<!>(g) k-parameter, a generalised ratio between Rg (from SANS) and R (from SEC-D4), gives an insight into the segmental distribution and shape of the particles (see scheme) depending on the ligation density (precursorsopen; SCNPs -full symbols).<!>Conclusions

<p>The precisely folded structure of biomacromolecules, such as proteins and enzymes, dictates their biochemical function. 1 Taking inspiration from nature, synthetic linear copolymers can be folded intramolecularly, forming dense single-chain nanoparticles (SCNPs). 2,3 The intramolecular crosslinking can be of static or dynamic, non-covalent or covalent nature. 4 The versatile molecular design of SCNPs allows for precise control over their architecture, 5,6 conformation and size, 7,8 rendering them ideal candidates for enzyme mimicry, 9 advanced catalysis, 10 drug delivery, 11 or targeted imaging. 12 Deducing generalized structure-property relationships for the size and conformation of SCNPs, regardless of the employed synthetic strategy, represents a fundamental step towards the advanced design of polymer nanoparticles, which is still lacking today. Yet, the essential requirement for the rational design of SCNP with properties close to the perfection of their natural counterparts is the fine control of the key parameters affecting SCNP formation. Thus, the precise analysis of the macromolecules before and after the folding process is critically required to establish complete understanding of these key parameters.</p><p>Analysis of the primary chemical structure of copolymers is conventionally carried out via 1 H and 13 C nuclear magnetic resonance (NMR) spectroscopy, while versatile NMR active nuclei enable monitoring the chemical transformations occurring during SCNP synthesis. A noteworthy illustration is a study that monitored the collapse of Pt II -SCNPs by phosphine ligand coordination using 31 P{ 1 H} NMR spectroscopy and corroborated via 195 Pt NMR measurements of the metal nuclei. 13 1 H and DOSY NMR measurements was used by our team to monitor step-wise unfolding of SCNPs by changes in chemical composition, and hydrodynamic size. 14 Furthermore, exploiting the high sensitivity of the 19 F nucleus, Perez-Baena et al.</p><p>studied successfully B(C6F5)3 catalyzed folding reactions. 15 Currently, size exclusion chromatography (SEC) is the most widely employed technique for mapping compaction during SCNP synthesis, as it separates molecules according to the hydrodynamic volume. The folding of SCNPs leads to a shift of the SEC trace towards higher elution volumes, relative to the starting material of the linear polymer (Figure 1). 2 The assumption of relative size change is only valid under ideal SEC conditions, i.e. the separation occurs purely entropically and no enthalpic interaction with the columns occurs. 2 Meeting these requirements is non-trivial, as the folding reaction usually entails chemical transformation. Thus, if the precursor did not experience enthalpic interactions with the stationary phase, the SCNP might behave differently, and vice versa. If the delay of the elution volume relative to a linear precursor is caused by both column interaction and hydrodynamic compaction of the analyte, the results will be significantly skewed. The current limitation in monitoring the folding process of SCNPs relies on using SEC with simple differential refractometry (dRI) detection and calculation of sizes based on SEC standards. Even detection of absolute molar masses using multi-angle static light scattering (MALS) in SEC-dRI-MALS can be strongly limited due to the small sizes of SCNP structures. 16 In order to gain deeper insight into the nature of SCNP folding, advanced hyphenated techniques such as hyphenating SEC-dRI-MALS systems to viscometry, fluorescence, or UV/Vis detectors are critically required. SEC-dRI-MALS enables the study of the molecular conformation of polymers and their apparent density, as well as revealing undesired multichain aggregation. 17,18 Evaluation of the conformation via MALS requires a certain molar mass range (spanning at least two orders of magnitude), making narrowly distributed polymer samples difficult to analyze. 2 In the case of isotropic scattering (SCNPs below 10 nm in diameter), or samples with a low optical contrast, the fundamental evaluation of the radius of gyration (Rg) and the absolute weight average molar mass Mw becomes impossible via MALS. Quasi-elastic light scattering (QELS, known also as dynamic light scattering) can determine the hydrodynamic radius (Rh) of particles down to only a few nm, providing information on the segmental density and shape of the separated particles via the ratio = Rg/Rh. 2 As aggregation or broad distributions lead to a drastic overestimation of the Rh, QELS ideally should be coupled to a size separation technique (such as SEC). Unfortunately, the downstream dilution after SEC is an obstacle that SEC-QELS characterization of SCNPs has not overcome to date. By measuring rheological properties in solution, online viscometry (VS) is truly orthogonal to the optical detectors (QELS, MALS, dRI), and therefore capable of analysing samples that isotropically scatter light, as well as samples with low optical contrast. The spatial compaction of SCNPs is indicated by a decrease in the intrinsic viscosity [], or the viscometric radius (Rη), respectively. [19][20][21][22][23][24][25] Viscosity data from SEC-dRI-MALS-VS experiments allows for the analysis of the solvent specific conformation of the polymer, providing valuable information about the macromolecular scaling or the solvent quality. Although limited techniques have been reported beyond SEC, complementary techniques to SEC separation such as mass spectrometry, 26 atomic force microscopy (AFM) 7,[27][28][29][30][31][32][33] or cryogenic transmission electron microscopy (Cryo-TEM) 21,34 enable detailed insights into the properties of SCNPs for appropriate materials.</p><p>Small-angle neutron scattering (SANS) provides unrivalled insights into soft matter materials. Due to the use of neutrons as the scattering source, SANS allows for non-destructive determination of the absolute radius of gyration (Rg) for particles between 1 nm -300 nm in size. 35 A lot of information can be obtained from SANS data by a model-independent analysis, where standard plotting approaches such a Holtzer, Kratky, and Porod plots are the most commonly used. Furthermore, a model-dependent analysis can be performed by fitting the SANS data to various form factor models, which aids in deducing critical information regarding the microstructure (e.g. shape, mass fractals or segment density, compactness, stiffness of the polymer backbone, etc.), as long as a sufficient contrast is manipulated, e.g. by using either a (partially) deuterated solvent or by deuteration of parts of the solute. 35 The disadvantages of SANS include high costs, the need for high sample quantity and purity, and extensive data-processing. However, its ability to provide characterisation in the sub 10 nm realm, which includes the majority of SCNPs, makes this an extremely powerful analytical tool. Initial neutron scattering characterization by Pomposo et al. provided insights into the shape and the compactness of SCNPs. 36 Other studies have applied Kratky plots to the scattering data of SCNPs to confirm the increase of intramolecular density. [36][37][38][39][40] Recently, a good prediction of experimentally observed relative chain collapses was set by a relation from a pool of literature values by invoking Flory-Fox theory. 41 Theoretical calculation and simulation frequently assist the interpretation of experimental characterization. 6,13,15,24,36,37,[42][43][44][45][46][47][48][49] Chain-dynamics and collapse processes have been assessed, mostly by semi-empirical and statistical approaches. 42 Not surprisingly, the complexity of macromolecular dynamics makes ab initio quantum chemical calculations less applicable for SCNPs. Limiting the range of temperature and polymer lengths, Danilov et al. presented an extensive thermodynamic characterization of a reversible selective point folding, spanning the opening and closing transition with atomistic models. 47 Importantly, Monte Carlo simulations enable the simultaneous tracing of the folding process. 50 Very recently, Sommer and co-workers revealed the underestimated impact of solvent conditions during the folding on the SCNP structure. 51,52 Fundamental progress towards understanding the structure-property relationships of SCNPs has thrived in the last 10 years. Nevertheless, an in-depth understanding of the correlation between the key parameters molar mass and folding group density to the resulting SCNP morphology is critically missing. Thus, advanced characterisation techniques need to be developed to facilitate the purposeful design of SCNP structures. Herein, we fill this critical gap based on a unique combination of analytical techniques. The investigated SCNPs are constructed from poly(pentafluorobenzyl-stat-tert-butyl acrylate) precursors spanning a defined range of reactive group density (5 to 30 mol% pentafluorobenzyl acrylate) and molar masses (20 to 100 kDa), and the resulting morphologies are investigated by a fusion of small-angle neutron scattering (SANS), 19 F NMR spectroscopy and quadruple size-exclusion chromatography (SEC-RI-MALS-VS-QELS), coined SEC-D4. This combination of sophisticated techniques provides an unprecedented picture of both the outer as well as inner structure of compactly folded single chains to deduce general trends of structure property relationships for any SCNP material.</p><!><p>The SCNP folding strategy in the current work is based on the regioselective para-fluoro thiol reaction (PFTR), which allows for the folding of the polymer precursors using an external dithiol crosslinker 1,4benzenedimethanethiol (BDMT). Inspired by the work of Roth et al., we employed pentafluorobenzyl acrylate (PFBA) as a functional monomer to introduce pentafluorobenzyl (PFB) moieties for the selective nucleophilic aromatic substitution at the para fluorine atom along the precursor backbone. 53 Thiols exhibit increased acidity and nucleophilicity compared to amines or alcohols, allowing for less harsh reaction conditions for the regioselective PFTR reaction. 54 Importantly, 19 F NMR in combination with 1 H NMR spectroscopy facilitates the precise determination of the absolute SCNP composition, which we purposefully manipulated by molar mass and number of crosslinks of our defined precursor library (Figure 2). Thus, we tune the morphology of the SCNPs by both the systematic variation of the primary structure of precursor chains (repeat-unit folding approach), and of versatile length by a folding strategy in a poor solvent. 55</p><!><p>The functional monomer pentafluorobenzyl acrylate (PFBA) was synthesized in a one-pot procedure described in the Supporting Information (SI, section 3.3). 53 The chain transfer agent cyanomethyl dodecyl trithiocarbonate (CMDT) facilitated the reversible addition-fragmentation chain transfer (RAFT) polymerization of the functional monomer pentafluorobenzyl acrylate (PFBA) and tert-butyl acrylate ( t BuA). 56 Herein, three chain lengths of linear precursors with targeted molar masses of 20, 50, and 100 kDa were obtained. As depicted in Figure 2, each molar mass was synthesized for three different PFBA monomer contents ranging from low (approximately 5%, colored green, A samples) to mid (approximately 15%, colored blue, B samples) to high (approximately 30%, colored black, C samples) feed ratios, randomly distributed per chain. As the amount of PFBA incorporation into the precursor backbone is in-line with the targeted values, the precursor library is suited for the fine-tuning of the ligation-density in the subsequent folding reaction. The number of repeating units, n and m, are calculated from Mn, molar mass of repeating units and copolymer composition determined from 1 H NMR spectra. Yellow background in the table indicates samples that were additionally characterized via SANS.</p><p>Initial SCNP folding of the RAFT-based precursors proved difficult due to the sulfur end groups, therefore, we employed a radical-induced oxidation in a facile one-pot procedure as described by Barner-Kowollik and coworkers in order to remove the terminal trithiocarbonate moieties. [57][58][59][60] The optimization of the conditions of end group removal was performed on representative copolymers as described in the Supporting Information (section 3.4). The success of RAFT group removal was confirmed by 1 H NMR spectroscopy on smaller copolymers as evidenced by the disappearance of the C(S)SCH2 signal at 3.25 ppm (section 3.4.2, SI-Figure 12). 60,61 The quantification of the polymer composition was performed for both the parent RAFT copolymer, and for the precursor after trithiocarbonate group removal via 1 H NMR spectroscopy with only negligible deviation observed.</p><p>The composition of the polymer (comonomer ratio and crosslinker content for the SCNPs) per chain directly impacts the optical properties of the copolymers (refractive index, dn/dc, the extinction coefficient, etc.) and all concentration dependent measures, respectively. The determination of the optical contrast dn/dc is critical for accurate molar mass determination using light scattering techniques and was determined for all precursors and the SCNPs. We employed a calculation procedure allowing to deduce reliable refractive index increments (Table 1) for any composition of the precursors as well as of the SCNPs (refer to section 3.2 of the SI).</p><!><p>The primary structure of the precursor was designed to enable a systematic variation in the final size and conformation of the SCNPs. The mol% of PFB in the precursor polymers defines the reactive group density and therefore fine-tunes the number of crosslinks in the folding reaction. All crosslinkable moieties of the precursor as well as the procedure for the evaluation of the absolute SCNP composition, are depicted in Figure 3.</p><!><p>The relative composition of the polymers is associated with an absolute number of building blocks for a given Mn of the precursor and the molar mass of the building blocks. 19 F NMR spectroscopy gives access to the amount of substituted para-fluorine atoms and the incorporated crosslinker under assumptions based on the distinct amount of PFB units in the precursor chains via 1 H NMR spectroscopy.</p><p>Poor solvent conditions during the folding reaction, enabling the densest possible conformation of the linear precursor, are critical to generate compact SCNPs, as was demonstrated by Sommer et al. 51 Optimal solubility of the precursor library was observed in polar solvents (i.e. poor solubility, below thetaconditions), facilitating a more densely packed random coil of the precursor polymer, thereby leading to a more tightly intramolecularly folded SCNP. Thus, the polar aprotic solvent acetonitrile (ACN) was ideal for the intramolecular folding process.</p><p>The external crosslinker 1,4-benzenedimethanethiol (BDMT) was added equimolarly with respect to the functional PFB groups of the precursor (refer to the SI, section 3.5). The use of an external crosslinker is analogous to substrate binding for enzyme activitya diffusion governed process -thus a higher than stoichiometric amount of external crosslinker is required for successful intramolecular folding. 62 Polar solvents not only facilitate the PFTR, but also the disulfide formation -a parasitic side reaction in this instance. 63 To overcome this issue, a catalytic amount of dimethylphenylphoshine (DMPP) is utilized as a reducing agent to effectively avoid disulfide-linkage of the crosslinker. 64,65 This parasitic side reaction would otherwise lead to decreased PFTR turnover and intermolecular ligation.</p><p>The successful generation of a structural hierarchy towards SCNPs, based on PFTR using an external crosslinker, required comprehensive optimization until a facile one-pot procedure was established (SI, section 3.5.1). Small additions of a good solvent (tetrahydrofuran, THF) was found to assist the PFTR reaction, whilst maintaining poor solvent conditions for the macromolecules. THF was added dropwise, under stirring, as the reaction rate of PFTR is rather slow in THF, 63 until the solution became transparent.</p><p>The required amount of good solvent was found to depend on the molar mass and on the amount of PFB incorporation. A clear dependence of the solvent quality with increasing PFB concentration (5% PFB (A) > 15 % PFB (B) > 30% PFB (C)) in the polymer backbone of the precursor was observed. Highly PFB decorated precursors are less soluble because of the highly polar C-F bonds, which can alter the overall polarizability and solubility of highly fluorinated polymers. [66][67][68] Excess of the base triethylamine (TEA) served not only to further solubilize the precursor, but also mitigated competitive behavior between the hydroxy endgroup of the precursor and the thiol-groups of the external crosslinker. 53,[69][70][71] Attempts to increase the reaction rate by a stronger base were not successful. 54,[70][71][72][73] To compensate for the use of TEA and the dilution required for SCNP synthesis, long reaction times are required. 63 The long reaction time is also intended to compensate for the increasing rigidity of the polymer as intramolecular ligation increases, and the decreasing of available PFB moieties. 29 The folding reaction was quenched by scavenging the unreacted thiols via a Michael reaction with methylacrylate (MA), successfully avoiding intermolecular side reactions or column interaction during SEC characterization. 63 The reaction times allowed the crosslinker molecules to reach even sparsely functionalized precursors in order to transform them into SCNPs at room temperature. Near quantitative conversion of the PFTR reaction was observed for the PFB groups of the B and C precursors, whereas the side-groups of sparcely decorated A-samples (approximately 5% PFB) were transformed with approximately 50-60 % PFTR conversion only (Figure 4). The lower reaction rate is in good correlation with the lower amount of reactive groups but additional reason is the more open precursor conformation, when compared to higher decorated samples of similar molar mass. 46 Thus, the composition and the conformation of the precursor during the folding reaction are influencing the probability of folding events and the intra-chain distances, respectively. 51</p><!><p>The turnover of the intramolecular PFTR can be accurately quantified (refer to the SI, section 3.5.4) via highly sensitive 19 F NMR spectroscopy. The pentafluorobenzyl moiety shows three distinct signals of the ortho-, meta-, and para-fluorine atoms as depicted in Figure 3 (refer to the SI, section 3.5.4). 53 Whereas the integration of all signals (Fo, Fm, Fp) enables the accurate quantification for small molecule experiments, only the signal integrals of the meta-flourine atoms in the precursor (Fm) and in the reacted PFB moiety (Fm') can be used to quantify the average conversion of the PFTR for polymers by</p><p>To quantify the ligation events, we use the term of ligation density (LD). The LD is the number of crosslinking events of the functional side groups with respect to the degree of the polymerization of the precursor chain. When comparing different chain lengths, the relative value of the LD is a useful measure.</p><p>Figure 3 shows that the amount of substituted para-fluoro atoms is a product of the PFTR-conversion χPFTR (determined via 19 F NMR spectroscopy) and the molar amount of PFB units in the precursor´s backbone (determined by 1 H NMR spectroscopy).</p><!><p>We obtained a SCNP-library with approximately 3% LD (A-samples), 14% LD (B-samples) and 30% LD (C-samples) of PFTR-converted side-groups (Figure 2 and Table 1).</p><p>Strictly speaking, crosslinking events cannot be derived from 19 F NMR analysis of the PFTR reaction, since this only describes the reaction with the PFB moiety. The bifunctional crosslinker could potentially only be attached with one thiol (Figure 4). A crosslinking reaction has occurred only when both sides of the crosslinker have demonstrably intramolecularly reacted. According to this folding strategy, potentially oneside PFTR-reacted crosslinkers will be scavenged at the other thiol-group by methylacrylate (MA), ensuring that any residual free thiols will not form undesired intermolecular disulfide linkages upon concentration of the solution. Thus, the double-sided attachment of the crosslinker to the macromolecules must be confirmed. Due to signal overlay in the 1 H NMR spectra, direct quantification of the characteristic scavenger resonances is not necessarily indicative of the number of potential scavenging events. Instead, the ratio of the integral of the methylene group of the PFB-unit to the integral of the methylene groups of the crosslinker presented in Figure 4, can serve as a validation of the nature of the attachment of the crosslinker. As confirmed via the comparison of these 1 H NMR spectroscopic integrals to the χPFTR (Table in Figure 4), mainly two-side attached crosslinker molecules are present. Additionally, the scavenging scenario is negligible and the amount of incorporated bifunctional crosslinker molecules is equal to half of the ligation points, respectively.</p><p>As the relative composition of the polymer ultimately dictates absolute number of PFB moietiesgiven the knowledge of the absolute Mn of the precursors and the molar mass of the PFBA monomer units -the average number of crosslinker per chain can be estimated for every sample. The average number of crosslinker per chain, equal to half of the number of the total number of ligation events per chain, increases with the total chain length for comparable LD. Comparison of the experimental SCNP molar masses (Figure 4, left box, color coded) shows good agreement with the expected molar mass increase compared to the linear precursor (Figure 4, left box, red dots). Using SEC, we indicate an average molar mass increase of approximately 1-, 6-, and 12 % for the A-, B-, and C precursor samples, respectively. These values are an important further confirmation of the primary structure of the SCNPs, i.e. an increase in molar mass but a decrease in size (see discussion below).</p><p>Thus, the obtained library of precursor-SCNP pairs serves well to deduce structure-property relationships with regard to the chain lengths and the amount of crosslinker units per chain (Table 1). The library is thus key for elucidation the size and conformation of the precursor and their corresponding SCNPs by SEC-D4 and SANS characterization.</p><!><p>Table 1. Key parameters of the polymer library determined via 1 H-, and 19 F NMR spectroscopy, SEC-D4 and SANS (in THF, THF-d8). The full list of parameters and calculations is available in the SI Tables 8-10.</p><!><p>State-of-the-art SEC-D4 characterization enables a comprehensive analysis of the precursor library and their folded counterparts. The successful interpretation of the SEC-D4 experiments depends on the precise determination of dn/dc (Eqs. SI-2 and SI-3), which is challenging in case of copolymers with composition variation. Very low optical contrast was observed for the precursor library in THF (dn/dc approx. 0.05 mL/g, see Table 1). Such an observation is somewhat expected due to the highly polar C-F bonds decreasing the overall polarizability of multi-fluorinated organic molecules, thus, influencing the refractive index increment dn/dc. [66][67][68] In contrast, sulphur belongs to the group of highly polarizable main group elements and the C-S bond has low polarity. 74 Therefore, the refractive index of the SCNPs is significantly increased due to incorporation of the dithiol-crosslinker. A comprehensive strategy for reliable evaluation of dn/dc dependent on the varied copolymer composition for both precursor and SCNPs is provided in the SI section 3.2. The exact knowledge of dn/dc is critically needed for samples of non-quantitative mass-recovery during SEC separation, which by definition limits the precise online determination of the dn/dc. Quantitative massrecovery in SEC separation for all precursors and most SCNPs (SI Table 8) allowed for reliable SEC interpretation and comprehensive online analysis of the dn/dc values of the complete library.</p><p>Figure 5a shows the dRI traces from the SEC-D4 separation for the 50 kDa samples (complete data in SI, section 4.2), including the molar mass and the intrinsic viscosities of the precursors and their folded analogues. Initial observation of the dRI traces shows monomodal distribution for the starting material, as well as for the SCNPs. The small high molar mass shoulder in some chromatograms is most probably caused by backbiting as a side reaction. 75,76 This shoulder disappears after folding due to further precipitation procedures. Table 1 shows that the overall polydispersity Đ is not significantly altered after the folding, indicating that the purification after the folding compensates for the generally expected increase of polydispersity during the folding process. 77 The shift of the peak from the precursor to the corresponding SCNP reveals the apparent decrease of the hydrodynamic size, which is more pronounced for higher LDs of the SCNPs. This observation is directly supported by the change of the intrinsic viscosity from the precursor to the corresponding SCNP (Figure 5a). Figure 5b indicates the direct correlation between the extent of apparent hydrodynamic collapse by the peak apex shift along the elution volume axis (Ve) to the number of crosslinks per chain for all samples. Comparing the samples with similar LD (Figure 5c), it is obvious that the apparent hydrodynamic collapse is more significant for longer chains, because they possess larger number of segments, which increases the freedom for stronger conformational changes as discussed below. Hyphenation of SEC to dRI and MALS detectors allows for evaluation of the absolute molar mass of the macromolecules (theoretical background is given in the SI section 1). Additional interpretation of the angular dependence of the scattering intensity enables the determination of the radius of gyration (Rg). The dependence between Rg and the molar mass delivers the scaling exponent  according to Eq. (3).</p><p>Unfortunately, the determination of Rg is limited for particle sizes smaller than /20 nm (with  the wavelength of the incident beam) because of the lack of angular dependence of their scattering intensity (isotropic scattering). With size ranges of below 12 nm, the SCNPs fall within the isotropic scattering regime. Even though, the reproducibility of the triple determination of the Rg values for the 50 kDa-, and the 100 kDa samples is high (SI determination of the scaling parameter  from MALS requires a spanning of at least two orders of magnitude of molar mass range, making narrow distributions challenging. 2 Nevertheless, absolute molar mass determination was successfully performed and a steady decrease of the molar mass with increasing elution volume was observed indicating no significant influence of enthalpic interactions on the SEC results. The increase of molar mass from precursors to corresponding SCNP (Figure 5a and Table 1) clearly correlates to the theoretical amount of incorporated crosslinker (C > B > A), confirming the success of the folding reaction.</p><p>As discussed in the introduction, QELS is a complementary approach to MALS for determination of small polymers down to several nm. Therefore, part of the SEC-D4 equipment was a QELS-detector for the determination of the hydrodynamic radius Rh. According to the Stokes-Einstein equation (Eq. SI-6) Rh corresponds to the radius of a hypothetical sphere which possesses the same diffusion coefficient as the measured macromolecule. Rh was frequently used for investigation of SCNPs using QELS in batch. 15,[78][79][80] Yet, the scattering intensity is proportional to the radius to the 6 th power, leading to overestimation of the radii whenever impurities (dust or broad distributions) are present in the sample. The limited precision of QELS-batch characterization was analysed by our team, noting that it can lead to strong deviation from reality. 80 Consequently, QELS detection is most effective when coupled to a size separation, as each elution fraction can be considered monodisperse. 18 Unfortunately, poor signal-to-noise ratio is a well-known obstacle due to dilution in the SEC column additionally to the low optical contrast dn/dc. Therefore, only the Rh from samples with sufficiently high dn/dc (SCNP with high crosslinker incorporation) were successfully characterized via online-QELS technique and are listed in SI Table 9. In conclusion, due to the limits of the light scattering, the reliable analysis of Rg and Rh was not possible over the complete range of samples' library via SEC-D4.</p><p>Online viscometry is orthogonal to all previously described detectors, as it measures rheological properties in solution and is independent of their optical features. In contrast to the results from MALS and QELS detectors, the signal of the viscometer shows excellent signal-to-noise ratio -even for particle sizes below the size-limit of light scattering. The spatial compaction of SCNPs should lead to a decrease in intrinsic viscosity [η] which in turn is correlating with a decrease in the viscometric radius, Rη, according to Eq. ( 4).</p><p>[𝜂] = 10 𝜋 3</p><p>Although the methods for determination of Rh and Rηare based on different physical principles, both radii are derived under the assumption of a sphere equivalent and as a fact, it was frequently found that their values are in a good agreement. 81 Viscometric radii between 3 and 15 nm are obtained for the linear precursors (Table 1) with a trend of increasing average Rηwith the molar mass as expected (Figure 6a). The SCNPs of all molar mass ranges clearly follow the expected trend that increasing number of crosslinks per chain lead to increasing chain collapse of the SCNPs and confirm the volume reduction found by the shift of Ve (Figure 5b,c).</p><p>In contrast to the 50 kDa and 100 kDa samples, we obtain atypical results for the smallest (20 kDa) samples (Figure 6b). Sample 20A shows a slight increase of R, in contrast to the expected decrease as found for 20C. This unexpected result is unlikely associated with experimental error as the triple determination of R for all 18 samples shows not only a low uncertainty for each experiment, but is also reproducible for each sample (SI Table 9). Nevertheless, it is not possible to suggest a lower physical limit for small particle size of this technique, as it strongly depends on the pressure, the temperature and the set-up.</p><p>We hypothesize that the short chain length is to be the reason for the atypical size development of the 20 kDa samples resulting in smaller volume reduction compared to larger precursor chains. Only short distance interactions (i.e. more local crosslinking as opposed to end-to-end crosslinking) are possible in shorter chains, while longer chains have greater potential for long-distance interactions between the macromolecular segments. This is justified with the small number of segments at a fixed persistence length.</p><p>Figure 7 shows the effect of the polymerisation degree on the degree of compaction, which is even more pronounced for poor solvent conditions. While characterisation is carried out in good solvent conditions in the stages before and after the folding reaction (Figure 7, green modes), poor solvent conditions are chosen for the folding process (Figure 7, red mode). The reason for this difference is that high dilutions and the used binary solvent required during folding are not ideal for the clear interpretation of analytical experiments. Furthermore, the characterization of the starting material in a good solvent represents the state of statistical linear chains. However, it should be noted that while the larger polymerisation degrees lead to conformations close to ideal coil statistics with pronounced fractal behaviour, the lower polymerisation degrees are excluded from this behaviour and have conformations closer to worm-like chains. Thus, the change to poor solvent conditions (Figure 7, red arrows) close to the theta-state leads to a more compact precursor conformation when the polymerisation degree is sufficiently high, increasing the probability of self-interaction to foster a tight intramolecular ligation (Figure 7, red mode). The base-induced intramolecular ligation (Figure 7, black arrows) can be referred to as a 'covalent freeze' of the chain conformation under these conditions, giving the quality of the solvent a paramount meaning for the final SCNP conformation. 11 The subsequent characterization of the SCNPs in a good solvent mirrors the dominant morphology of the precursors under the folding conditions.</p><p>Of course, such behaviour depends on the number of crosslinks per segment which is reflected in the relative numbers of crosslinks per precursor chain (or LD in %). The trend of increasing compaction is similar for all molar mass series (see Figure 5c). However, the effect of the absolute amount of crosslinks per chain seems to be the driving parameter of the folding behavoiur and is depending on the contour length of the chains. Furthermore, having a particular composition and thus, particular persistence length of the macromolecule implies an increasing number of the segments (increasing molar mass) will lead to increasing number of long-distance interactions between the segments for long chains. In contrast, predominantly short distance interactions occur for shorter chains, which may rather occur along the chain (2D contraction) than over intra-chain space (3D contraction, according to the classical understanding of folding). Further experimental confirmation of other repeat unit folding SCNP-systems is beneficial to validate this hypothesis of the interplay between chain-length and folding conditions.</p><p>Information on the compactness of polymer chains is obtained by the molar mass dependence of the intrinsic viscosity according to the Kuhn-Mark-Houwink (KMH) plot, which by theory correlates with the scaling law (Eq. 3). The KMH plot is shown for the 50 kDa sample-set (50A, 50B, 50C) in Figure 8a.</p><p>Significant change of the slope in the log [η] vs. log M plot indicates the change of the macromolecular conformation from precursor to the folded analogue according to Eq. ( 5).</p><p>[𝜂] = 𝐾𝑀 𝛼 (5)</p><p>The average KMH-slopes of all samples are given in Table 1 and in Figure 8. All precursors show KMHslopes above 0.5, corresponding to a polymer coil in good solvent conditions. 35 Lower values indicate a more compact structure for the corresponding SCNPs. The SCNPs of the lowest LD (A-samples) show no significant change of compactness, as the KMH-slope only changes marginally. In contrast, the compactness of the B-samples is significantly higher due to intramolecular ligation. Accordingly, the tightly crosslinked C-samples adopt a mostly compact spherical conformation (close to  = 0 for solid sphere). In contrast to the clear impact of the LD on the KMH slope, the molar mass does not show significant influence on the conformation (Figure 8b).</p><!><p>The knowledge about the number of crosslinks per chain provides insight into the average number of monomer units per loop, a suitable value to estimate the compactness of SCNPs. For simplicity, we can assume that the number of incorporated crosslinker units is equal to the average number of loops per chain. Relating the average number of loops per chain to the total number of building blocks per chain (degree of polymerisation, DP), one can estimate the number of building blocks per loop for the SCNPs, coined the loop length. Sommer and colleagues recently showed that the distances between ligation-events (considered as average loop length) follow the distribution for Gaussian chains after folding in poor solvent conditions. 51 Thus, the average values of LD and polymerisation degree of the SCNPs can deliver an average of the loop length (see also SI, section 3.5.4). 55 The results of the average loop length show the same trend as the results of the KMH-slopes, indicating a direct correlation of the loop length to the conventional measure of polymer conformation (Figure 8c). Interestingly, the shortest loop lengths found in the C-series correspond with approx. 4 blocks per loop to approx. 2 nm length (assuming C-C bond length of 0.125 nm). The shortest found block length, thus, corresponds to the average Kuhn length of 4 nm calculated from SANS experiments (Table 1) related to a segment length of approx. 2 nm. This fact shows that not only the ligation density but also the segment length and flexibility of the precursor chain is a limiting factor during SCNP formation.</p><p>SEC-D4 analysis with state-of-the art detectors allows for valuable insight into the molecular conformation of the precursor library of nine precursors and their folded counterparts. While the size determination using online scattering techniques (MALS, QELS) were pushed to their limits -due to both, small sizes and low optical contrast -absolute molar mass determination and viscosity detection reliably validate the success of the folding strategy. Analysis of the viscosity and molar mass show substantial potential for size and conformation characterization, keeping in mind the applied assumptions. In order to evidence the reliability of the observed SEC-D4 data and to obtain fully comprehensive image of an SCNP architecture, a scattering technique not limited to the approximate 10 nm size regime or optical contrast issues is required. We have therefore employed the powerful SANS technique to sharpen our understanding of these nanoobjects.</p><!><p>SANS is unrivalled in analytical power for soft matter materials, as it provides the non-destructive determination of the absolute Rg for particles down to 1 nm in size and additionally provides the information about the microstructure (shape and segment density) of the analyte. 82 Contrary to light-scattering techniques, the contrast of SANS is given by the difference of the scattering length between a dissolved polymer and the solvent. Deuterium labelling provides decent contrast to investigate organic macromolecules and makes polymers, which have poor optical contrast, "visible". Due to the employment of THF-d8 as a solvent for SANS experiments, we have good solvent conditions for the polymer-library and excellent contrast as already demonstrated in previous studies on acrylate-based polymers. 83 For our SANS experiments, we selected a representative sample set of our original nine precursors and the corresponding SCNPs (samples 50A, 50B and 50C as well as 20C, 100A and 100C). Concentration dependent measurements enable the complete Zimm analysis of each sample. 84 The evaluation of the scattering intensity I(q) over large range of scattering vector q enables information about: (i) the global scaling of the macromolecules at low q values and (ii) the macromolecular segments down to the size scale of the monomers (details on the setup and theoretical background are given in SI, section 6). Figure 9a shows an example of a scattering curve of the 50C-pair of precursor and SCNP and the ranges of the scattering curve delivering information about size (Rg), conformation (scaling exponent  in Eq. 3) and flexibility (segment length or Kuhn length lK) of the polymer chain (all scattering curves and their interpretation are given in SI, section 6). As the concentrations of the samples are held sufficiently below chain overlap concentration c* (SI, section 1), interparticle interaction is ruled out, 85 as confirmed by the absence of a structure factor peak. Therefore, we can focus on the form factor P(q) (SI, section 6.1) to investigate the scattering function of a single isolated polymer particle, revealing information about size and shape of the precursors and the corresponding SCNPs.</p><p>Using the Zimm analysis, the absolute Rg and the second virial coefficients A2 in THF-d8 were determined and are summarized in Table 1. Absolute particle sizes between 3 and 17 nm radius of gyration are obtained.</p><p>Good agreement with the tendencies found for R was found (Table 1, Figure 6). Furthermore, good correlation of the Rg values with the calculated molar masses from SEC-D4 is observed: already small changes of chain length for the 50 kDa precursor demonstrate an impact on the Rg (samples 50A, 50B, 50C</p><p>in Table 1). The extent of collapse from precursor to SCNP is in direct relation to the extent of LD and correlates with the development of the size reduction calculated from the SEC-D4. Figure 9b Having reliable information about the size reduction of the SCNPs after folding, which is confirmed by two independent analytical tools, SEC-D4 and SANS, the next logical step is to validate the reliability of the peak shift of the SEC chromatograms Ve (Figure 5). A linear fit of the Ve increase with increasing crosslinks per chain (Figure 9c) allows an assessment of the deviation of the normalized Vg and V values from this fit. Although the general tendency is similar, the plot makes immediately clear that in contrast to the well overlapping Vg and V, their values are strongly deviating from Ve. The reason for this deviation is most likely the non-ideal entropic separation of the macromolecules, which can lead to a mixed size-exclusion / interaction mode of separation, thus, influencing the position of the peak apex of the chromatogram. Although most of the samples show linear dependence of the absolute molar mass (SEC mode) of the elution volume (see SI, section 5), some deviations from linearity can be indicated, showing that co-elution of different sizes especially in case of the SCNPs is taking place. The origin of this coelution should be investigated in depth using instruments of separation theory and advanced fractionation tools. It should be noted that for the purpose of the current study, the mixed mode of separation does not affect the calculation of the main parameters from SEC-D4 listed in Table 1 and in Tables 5-7 of the SI.</p><p>However, this phenomenon of non-ideal entropic separation can be misleading in regards to the extent of folding and compaction of the SCNPs based only on the single shift of the chromatogram Ve.</p><!><p>) and g' (half full dots) calculated according to Eqs. 7 and 8 vs crosslinks per chain. Values of g found for 3-arm (0.78) and 18-arm (0.23) polystyrene stars under theta-conditions are marked in grey. 35</p><!><p>Further information from the Zimm analysis of the SANS results at low q-values delivers the second osmotic virial coefficient A2, which is a quantitative measure of the strength of interaction between the macromolecular segments. Good solvent conditions will lead to positive A2 values, whereas A2 = 0 corresponds to unperturbed dimensions of the macromolecules under theta-conditions. The obtained second virial coefficients in THF (Table 1) are similar for most precursors indicating good solvent conditions and decreasing to theta-conditions with increasing molar mass as expected. 35 Clear dependency of the A2 from the copolymer composition is observed for the 50 kDa samples (see Table 1). In accordance with the proposed hypothesis (Figure 7), the SCNPs show a negative A2 close to theta-state corresponding to the adjusted and "frozen" compact conformation by folding in a poor solvent (ACN/THF).</p><p>The compact conformation of the folded SCNPs and conformation of their precursors is impressively reflected in their different scattering behaviour. Figure 9a depicts the representative chain collapse of sample 50C visible in the significantly faster decay of the scattering intensity of the SCNPs compared to the precursor. The evaluation of the negative, reciprocal slope of the scattering curves delivers information on the fractal dimension, e.g. the scaling exponent , which describes the relation between the Rg and the molar mass of the macromolecules (Eq. 3). Additional fit to model form factors (polymer excluded volume) of the scattering curves P(q) was used to validate the evaluated values of the Rg from Zimm analysis and  from the slope of P(q). 86 The results from these fits show very good agreement with the manually evaluated data (see SI Table 10 and section 6.3 in the SI).</p><p>Figure 9d shows that the scaling exponents of the precursor correspond to linear Gaussian chain conformation, which is defined by ν = 0.6 in a good solvent. 87 Colmenero et al. recently emphasized the similarity of Gaussian chain conformation to the scaling behaviour of intrinsically disordered proteins (IDPs) defined by their biological task. 44 Comparing the precursor to their corresponding SCNPs, a clear trend of increasing compaction with increasing LD is confirmed. Maximal compaction approaching spherelike conformation ( = 0.33) is obtained for the SCNPs of the C-series of the strongly crosslinked 50 and 100 kDa samples, while the 20 kDa sample show still pronounced but smaller compaction. This strong contraction is in full agreement with the slightly negative A2 values found in a good solvent (THF-d8), which can only be explained by the folding strategy under carefully chosen poor solvent conditions (see Figure 7). Furthermore, the analysis of the C-series shows convincingly that the conformation changes during folding are not only depending on the frequency of the crosslinking events (LD). Having maximum reduction of the SCNP volume (at 30% LD, Figure 9b) there is still room to control the final conformation from slightly compact chain conformation (20C) to sphere like (50C) to hard sphere (100C) by adjusting the absolute number of crosslinks, e.g. chain length.</p><p>The orthogonal analytical tools in this study allow for a unique comparison between the scaling exponent  and the KMH exponent both obtained in the same solvent conditions (THF for SEC-D4 and THF-d8 for SANS). Figure 9d further shows the values relation to the  values, which has its origin in the Flory mean-field theory as = (1 + ) / 3. 11,87,88 Strong agreement between both values confirms transition of the statistical precursor coils into compact, poorly drained by the solvent SCNP morphologies. The changes in  from SEC-D4 are, in general, less pronounced than the changes calculated by transformation of to deduced by SANS. However, these deviations are within an allowable range, considering the different principle of analysis: SANS is performed in batch while SEC-D4 delivers molar mass dependent values after a separation according to size.</p><p>Increasing compaction is related with increasing density, thus, a complementary calculation of the macromolecular apparent density can validate the observed compaction. The apparent density dapp is derived from the volume of gyration (evaluated via SANS characterization) and the corresponding molar mass Mw (evaluated via SEC-D4) as follows 87 𝑑 𝑎𝑝𝑝 = 𝑀 𝑤 𝑉 𝑅 𝑔 = 3 4𝜋𝑁 𝐴 𝑀 𝑤 𝑅 𝑔 3 (6) Figure 9e shows the apparent density values as well as the change of the density after folding depending on the LD. The precursors show an average density of dapp = 0.03 g cm -3 and negligible change of this density is observed for the slightly crosslinked SCNPs (A-samples). In contrast, the medium crosslinked SCNPs show already significant increase (B-samples), and the strongly crosslinked SCNPs show the highest dapp (C-samples). Our team earlier suggested that the minimum SCNP density is 0.1 g cm -3 , despite observing a large number of SCNP systems below this density threshold. The highest found dapp = 0.25 g cm -3 for our SCNPs is far below the 1 g cm -3 expected for bulk materials, which surprisingly, was reported for some SCNPs. 80 It should be noted that, the swelling behaviour in a good solvent as well as bulky functional groups within the SCNP structure could strongly influence the apparent density.</p><p>The change of density upon folding increases steadily with LD (Figure 9e). The highest density is observed for the smallest SCNP sample f20C, but peculiarly, the change of density due to the folding is only changed to an extent comparable to the medium crosslinked SCNPs. In order to understand this effect more precisely, the density distribution within the macromolecule should be analysed. For this purpose, the generalised ratios kR/Rg) andRg/Rhcan be used asquantitative indicators of the molecular topology of polymers in solution. 18,89 kindicates the mass distribution around the centre of gravity and the hydrodynamic draining of the solvent, thus giving information about the segmental density in complex macromolecules. Figure 9g presents the kvalues of the samples investigated by SANS and SEC-D4. The topology of the precursors corresponds to the density-distribution of a linear coil (k = 0.6). The increase of kis minimal for the weakly crosslinked SCNPs (A-samples), and significant for the compact SCNPs (Csamples). As a reference, the theoretical maximum compaction of solid spheres has the ratio of k= 1.3. 90 The SCNP sample f20C has the highest mass-centred compactness. These results are confirmed by the complementary ρ-parameter based on the calculation of the Rh from SEC-D4 in combination with Rg form SANS (SI, section 4.3.2). The observation of the 20C sample behaviour coincides with the phenomena observed for the viscometric radii and the apparent density of the smallest polymers. We explain this phenomena by two main reasons: 80 (i) as discussed previously, shorter chains undergo predominantly short distance interactions under the poor solvent conditions due to limited number of segments able to form a loop over longer distances. Taking into account that the stiffness of the macromolecular chain is defined by the segment length, low number of segments does not allow a significant contraction in 3D sense, and the volume of the polymer coils is not significantly changed due to the folding reaction; (ii) at the same time, the mass-increase due to the incorporation of the external crosslinker contributes proportionally to the apparent density dapp (Eq. 6). Visualisation of the impact of mass increase on the density upon folding for small molar mass precursors is given in the SI, section 7.4.</p><p>The reduction of macromolecular size due to conformational changes is a well-known key issue in the branching theory, where the concept of contraction factors g and g´ is applied. 35,91,92 The contraction factors indicate the relative change of the main structure parameters used in this study, [] (viscosity model) and the Rg (radius model). We define the contraction factor as the ratio of the intrinsic viscosities of the precursor and the corresponding SCNP for a given DP (Eqs. 7 and 8). 35,88 The original introduction of contraction factors within the branching theory takes into account changes in [] or Rg at equal molar mass of a linear polymer an the branched analogue. In our study, we translate this relationship for SCNPs contraction at similar polymerisation degrees because of the increase of the molar mass of the precursor after folding.</p><p>g and g' possess values between 0 and 1, where smaller values indicate a higher contraction.</p><p>Figure 9f shows that both, g and g' calculated from SEC-D4 and SANS show similar tendencies. Since this is the first study in which the concept of contraction factor is applied to SCNPs, we can compare the obtained values only with reported branched polymers. As a reference, a 3-arm polystyrene star under thetaconditions shows weak contraction with g = 0.78 while an 18-arms polystyrene strong contraction with g = 0.23 is reported. 35 It should be noted that the nature of the contraction factor is an independent measure and neither the architecture, nor chemical structure and solvent conditions need to be considered. The experimental data in Figure 9f demonstrate the direct correlation of the contraction factors to the crosslinks per chain revealing the minor impact of the molar mass, but strong impact of the crosslinking events. The samples with comparable LD (for example the C-samples, black) show only slightly different contraction factors, indicating that the extent of contraction increases slightly with the length of the precursor chain. As seen by several previous parameters, the smallest polymer (sample 20C) shows smaller contraction in comparison to larger polymers of the same LD, which supports our hypothesis of limited conformational freedom during folding for such small polymers. In conclusion, for quantification of the relative compaction of the SCNPs, contraction factors show better suitability than the apparent density dapp. Further details on the relation 87,88 between the g and g' are given in the SI, section 7.3.</p><p>The SANS scattering curves of our library contain much more information than only the structural parameters discussed above. The extraction of this information can be performed by analysis of the scattering intensity P(q) according to Kratky or Casassa. 84,93,94 Kratky and Casassa plots can qualitatively assess the degree of folding and are suitable for determination of the flexibility of linear chains using the high q-regime (see Figure 9a). This analysis is usually applied to concentration dependent scattering curves, which leads to simplified data analysis. Although, Hawker et al. in early SCNP research, that no remarkable differences of Kratky plots are given for different concentrations for slightly branched polystyrene-based SCNPs, 40 we plot the form factor P(q) corrected for angle and concentration dependency for sake of accuracy as obtained by our Zimm analysis.</p><p>Kratky plots of the dimension q 2 P(q) vs. q facilitate to decipher the segment density, as a plateau appears for polymers of a Gaussian coil conformation (q -2 -slope). In contrast, globular structures show a bell-shaped peak (Figure 10c). 95,96 Kratky plots have been frequently applied to reveal the folding state of proteins, by referring the area of the bell above the plateau the degree of folding. 95 This calculation approach can therefore, semi-qualitatively assess the degree of intramolecular ligation, as the position of the local maximum refers to the linear mass fractions of the particles. 93,96 Kratky plots have already been applied to experimental SANS data of SCNPs to evaluate the degree of compaction, for example by comparison to Kratky plots of form factors for globular particles. [36][37][38][39][40] Nevertheless, we exercise caution when directly comparing Kratky plots of completely different structures, as performed for the scattering form factors of SCNPs and intrinsically disordered proteins to that of compact globular proteins. 36 Instead, we compare the Kratky plots within a chemically similar system (e.g. precursor and the correlated SCNPs, or SCNPs of the same building blocks) to elucidate the density and sphericity of the polymers of our polymer library (Figure 10). For all precursors, a plateau in a Kratky plot is obtained, resembling a Gaussian coil conformation (see SI, section 6). Increasing plateau height is, amongst other factors, caused by the different particle sizes (molar mass), which can be shown in a dimensionless Kratky plot (qRg) 2 P(q) vs. qRg, see SI, section 6.1.2).</p><p>Dimensionless Kratky plots have been, for example, applied to synthetic SCNPs, intrinsically disordered proteins and globular proteins by Pomposo et al. 36 In Figure 10a the 50 kDa series of the precursors are shown. Figure 10b shows the Kratky plots of the 50kDa SCNPs (all plots available in SI, section 6). In contrast to the less crosslinked SCNPs (A-samples), the strongly crosslinked SCNPs (B-and C-samples)</p><p>show a pronounced Gaussian distribution in the Kratky plot indicating globular particles as also reported for globular proteins. 95 These local maxima of the Kratky plot indicate an increased segment density. The position of the apex of the peak (in q-dimension) is related to the size of the mass-fractals and resembles the trends observed for Rg of the SCNPs (Table 1). Contrary to the classical Kratky plots of spheres, a plateau is still present for the compact B-and C-series of the SCNPs at high q-values. This observation indicates the coexistence of multiple domains composed of dense-fractions and linear chain conformations also observed for proteins, 97 leading to the assumption that the macromolecule does not possess homogenous segment distribution, but linear segments and dense segments jointly exist. 93,96 The denser segmented regions in the SCNP structures can be described as cross-linking crowded regions however, this assumption needs further elucidation using theoretical consideration in future studies. Closer look at the 20C SCNPs (Figure 10b) shows that such a plateau is less pronounced for this small molar mass sample. This observation, analysed in the context of the calculated high apparent density and high segment density found for this particular sample (Figure 9e and 9g), confirms low amounts of linear segments but rather dense structure with small dimensions. These findings support the above discussed hypothesis of short distance interactions responsible for crosslinking at low degrees of polymerisation, which cannot lead to joined linear and dense, globular segments in one SCNP as in the case of the 50 and 100 kDa samples.</p><p>Similar to the Kratky plots, the Casassa plots of the SCNPs show a significantly increased maximum at low q-ranges (SI, section 6), which is typical for dense coil-structures. We apply this plot to deduce semiquantitative information about the coil behaviour itself, as the height of the maximum increases with decreasing deformability. The transition (denoted with q* in SI, section 6.1.3) from coil-like (low q-range)</p><p>to rod-like behaviour (high q-range) enables the calculation of the segment length. Another way to calculate the segment length is using a fit with model form factors as described in the SI, section 6.3.1. leading to precise analysis of the particle's features. The form factor model flexible cylinder provides the ability to determine the Kuhn length lK, based on simulations of a discrete representation of a worm-like chain model of Kratky and Porod and normalized by the volume of a flexible cylinder. 98 For this purpose the SasView 86 application was used, as described in the SI, section 6. Very good fit quality was achieved for all precursor samples and assures high reliability of the obtained data. The obtained Kuhn lengths of the precursors between 3.2 and 4.2 nm are slightly lower than the approx. 4.6 nm found for poly( t Bu acrylate). 83 The shortest lK, thus, the most flexible chain, is found for the precursor 20C, whereas sample 50C shows the longest lK -although both precursors exhibit a comparable backbone composition. Generally, the A-samples show better flexibility, thus, the low crosslinking turnover of the low functionalized precursor is not caused by steric hindrance of potentially stiff backbone features, but is solely caused by statistically reduced intramolecular ligation events. Focusing on the 50kDa samples, a clear increase of lK is indicated, which correlates well with the decreasing solubility in the order A > B > C. From these reasons, we assume maximal flexibility of the polymer chain of A-precursors. The solvent quality and the chain stiffness, both in means of the degree of rotational freedom, are of paramount importance for the efficacy of the folding reaction. 99,100 As the bending stiffness of the linear precursor determines the probability of intramolecular interaction in space, and therefore influences the morphology of the SCNPs, we conclude negligible influence of changing backbone composition or stiffness on the SCNP morphology.</p><!><p>Regioselective para-fluoro thiol cross-linking allows for the precise design a polymer library and its systematic folding into SCNPs defined by predetermined numbers of linkage points and molar masses.</p><p>Thus, two general properties of the precursor structure are investigated regarding their impacts on the SCNPs properties, making findings applicable to SCNPs regardless of the employed chemical strategy of the folding reaction.</p><p>Extensive 1 H NMR spectroscopy, 19 F NMR spectroscopy, SEC-D4 and SANS provide unique insight into the parameters limiting the formation of SCNPs. We combine these orthogonal instruments of advanced polymer analysis to obtain a multidimensional picture of the resulting architectures -from the chemical structure to the segment distribution to the global conformation.</p><p>The turnover of the crosslinking reaction and resulting size reduction of the formed SCNPs relative to the linear precursors directly correlates with two main parameters: the ligation density and the conformation of the linear precursor under the conditions of the folding reaction.</p><p>Based on molar masses from 20 to 100 kDa and ligation densities from 3 to 30% of the precursors, we estimate (i) a lower limit of 5% ligation density for effective crosslinking with minimum size reduction during SCNP formation and (ii) a higher limit of 30% ligation density leading to maximum achievable size reduction of 80%. These limits are additionally influenced by the molar mass of the precursor. For molar masses of 20 kDa, 80% volume contraction is not achievable due to lack of long-distance interactions as a result of low number of segments, which enable limited variation of the conformation even under poor solvent conditions.</p><p>In the molar mass range between 20 and 100 kDa, an effective control of the size reduction can be observed mainly depending on the absolute number of crosslinks per chain in the range between 5 and 40. These ranges are entirely independent on the chain length for molar masses above 20kDa. For more than 40 crosslinks at sufficiently high molar masses, no further increase of the compaction can be indicated.</p><p>These results are validated by the systematic combination of complementary scaling approaches using key parameters including molar mass, second virial coefficient, segment length, intrinsic viscosity, gyration, hydrodynamic and viscosity radii and apparent density. As an alternative to the apparent density, we introduce the contraction ratios g and g´ as simple yet meaningful tools to compare SCNPs properties regardless of the synthetic strategy. Calculations of the scaling exponent as as the complementary Kuhn-Mark-Houwink exponent unlock the highest achievable density for the 50 and 100kDa polymers leading to a hard sphere shape. However, the segment density based on generalized ratios of different radii and detailed Kratky and Casassa analysis of the SANS scattering behaviour unravel joint dense globular and linear segments within these SCNP structures.</p><p>These fundamental insights into the parameters ruling the formation of SCNPs enable the precise assessment of the commonly used shift of the elution volume using SEC in comparison with multidetector SEC-D4 and profound validation of these results by SANS. While SEC-D4 has its limits in terms of the use of optical detectors due to very small SCNP sizes and low refractive index increment, it is a reliable technique for SCNP analysis, when viscosity detection is used. The accuracy of these experiments is, however, essentially dependent on precise absolute molar mass analysis using MALS detector and the careful determination of the refractive index increment.</p><p>The obtained results from SANS and SEC-D4 demonstrate that the shift of the elution volume, even if mainly entropic separation (according to particle size) is operational, cannot be used as a reliable detection of size reduction due to SCNP folding. The reasons of this limitation can be found in the changes of the chemical composition after folding as well as in the very complex conformation of the SCNP structures.</p><p>Further studies on the separation mechanisms in SEC should shed light on this question.</p><p>When exploring approaches for size tuned artificial nanocontainers, SCNPs represent an ideal avenue between synthetically demanding dendrimers and architectural delicate macro-complexes such as micelles or polymersomes. Exploiting the chemistry available for SCNP construction, we uncover general guidelines for the tailored design of SCNPs with regard to molar mass and number of linkable units for the precursor material. The introduced alternative parameters such as g, g', and k show better applicability than conventional ones such as the apparent density and size that are of generalized use for encapsulation, retention behaviour or for catalytic activities, and bring SCNPs applications within reach.</p>

ChemRxiv

Influence of Chemical Kinetics on Predictions of Performance of Syngas Production From Fuel-Rich Combustion of CO2/CH4 Mixture in a Two-Layer Burner

Numerical investigations on partial oxidation combustion of CO2/CH4 mixture were executed for a two-layer burner using a two-dimensional two-temperature model with different detailed chemical reaction mechanisms that are DRM 19, GRI-Mech 1. 2, and GRI-Mech 3.0. Attention was focused on the influence of these mechanisms on predictions of the temperature distributions in the burner, chemical structure as well as syngas production. The equivalence ratio was a fixed value of 1.5, while the volumetric ratio of CO2 to CH4 was changed from 0 to 1. The predicted results were compared with the available experimental data. It was revealed that the chemical reaction mechanisms have little effect on the temperature distribution in the burner except for the exothermic zone. It indicted that the smaller kinetic DRM 19 can precisely predict the temperature distributions in the burner, using DRM 19 was recommended to save computational time when the detailed components of the syngas was not taken into consideration. In addition, all the three mechanisms predicted the same trend of molar fraction of CO, H2, and CO2 with experimental results. Good agreement between the experiment and predictions of major species was obtained by GRI-Mech 1.2 and GRI-Mech 3.0, the two mechanisms had the same accuracy in predicting CO, H2, and CO2 production. However, computations with DRM 19 under-predicted the molar fraction of CO and H2. Furthermore, it was shown that the thermal conductivity of porous media has significant effect on the syngas production. In general, the temperature was increased as the thermal conductivity of the porous media was reduced and the H2 production was increased.

influence_of_chemical_kinetics_on_predictions_of_performance_of_syngas_production_from_fuel-rich_com

<!>Introduction<!>Physical Model<!><!>Physical Model<!>Boundary Conditions<!><!>Initial Conditions and Solution<!><!>Temperature Distributions<!><!>Temperature Distributions<!><!>Temperature Distributions<!><!>Temperature Distributions<!>Chemical Structure<!><!>Chemical Structure<!>Major Species in the Exhaust Gases and Conversion Efficiency<!><!>Major Species in the Exhaust Gases and Conversion Efficiency<!><!>Conclusions<!><!>Data Availability Statement<!>Author Contributions<!>Conflict of Interest<!>

<p>- The highest combustion temperature in the reaction zone is predicted by DRM 19</p><p>- Computations by DRM 19 under-predict the molar fraction of CO and H2</p><p>- GRI-Mech 3.0 and 1.2 have the same accuracy in predicting the syngas component;</p><p>- Increase in thermal conductivity of solid leads to increase in temperature in the two-layer burner;</p><p>- Increase in thermal conductivity of solid leads to increase in H2 production.</p><!><p>The interest in syngas production from fuel-rich partial oxidation within inert porous media has increased significantly during the last two decades. This is because this technology combines several positive features, such as quick start-up, fast dynamic response, no need for external heat, steam, and catalyst. Kennedy et al. (1999) and Muhammad (2016) presented detailed reviews of this subject.</p><p>In order to improve combustion process and increase syngas production, a large amount of researches have been conducted in recent years and some significant developments have been made (Mujeebu, 2016).</p><p>There are many experimental and numerical studies on fuel-rich partial oxidation in porous media burner. In general, there are two design approaches commonly employed for syngas production in porous burner: the transient combustions systems and stationary systems.</p><p>For the transient combustion system, the porous media burner was filled with homogeneous porous media and flame propagations were observed either in downstream or upstream direction in most cases. Hydrogen production by transient filtration combustion was extensively studied by Kennedy group (Drayton et al., 1998), they proposed reciprocal flow burner (RFB) for syngas production and the combustion wave was restricted in the burner by periodically changing the direction of the flow. An extremely high flammability limit was extended to an equivalence ratio of eight for CH4/Air mixtures. In the experimental and numerical studies of Kennedy et al. (2000), upstream, downstream, or standing wave was observed experimentally, mainly depending on the CH4/Air equivalence ratio (φ) for the wide range of 0.2 ≤ φ ≤ 2.5 and the fixed gas velocity of is 0.25 m/s. The experimental results showed that 60% of the methane was converted to CO and H2 for φ > 2.</p><p>Toledo et al. (2009) studied experimentally the transient combustion system for syngas production in a packed bed filled with 5.6 mm Al2O3 spheres. It was shown that the maximum hydrogen yield was close to 50% and CO yield was close to 80%. To increase the syngas production, two methods were conducted, namely, adding stream during filtration combustion (Araya et al., 2014), or providing external heat source to the burner system (Zheng et al., 2012a). Experimental and numerical results by Araya et al. (2014) showed that hydrogen yield increases with increasing steam content in methane-air mixture. Several fuels conversion to syngas were experimentally studied by Toledo et al. (2016) and Gonzalez et al. (2018), they (Toledo et al., 2016) studied conversion efficiency of partial oxidation combustion in pellet packed bed for liquefied Petroleum Gas, butane, propane, Diesel fuel, and Heavy Fuel Oil. It was shown that conversion efficiency of Heavy Fuel Oil was highest than other fuels and reached up to 45%.</p><p>For the transient combustion system, combustion wave always propagates upstream or downstream and this leads to flame extinction at the end. To stabilize the flame within porous burner, two types of burner, namely, RFB (Drayton et al., 1998) and two-layer burner (Zeng et al., 2017; Wang et al., 2018) filled with different material or structures of the porous media were developed.</p><p>For the stationary systems, the burner was filled with different material or structures of porous media, the flame was restricted near two sections of the burners under a certain range of equivalence ratio and gas velocity. Zeng et al. (2017) studied experimentally and numerically the syngas production in a two-layer burner for fuel-rich combustion with a range of CO2 content in the CH4 fuel. The experimental results showed that the reforming efficiency increased from 39.1 to 45.3% when the CO2 was injected into the system. In their subsequent study (Wang et al., 2018), the performance of methane partial oxidation in a two-layer burner filled with alumina pellets of different diameters in the downstream was conducted. According to the highest reforming efficiency, an optimized burner was determined, which was composed of 7.5 mm pellets in the downstream section and 2–3 mm pellets in the upstream section.</p><p>For simulation of syngas production from fuel-rich partial oxidation in porous media, it is essential to use detailed or reduced chemistry for investigation of detailed composition of syngas and intermediary components, GRI-Mech combustion mechanism was widely used, which includes detailed Kinetics GRI-Mech 1.2, GRI-Mech 2.11, and GRI-Mech 3.0 (Bowman et al., 2009). For saving computational cost, smaller chemical kinetics like Peters (Mauss and Peters, 1993) or overall mechanism was used by the researchers. Based on the volume-averaged method, one-dimensional or two-dimensional model with these chemical kinetics has been applied to predict the temperature profiles, syngas production, and conversion efficiency.</p><p>Kinetic simulations with GRI-Mech 1.2 were conducted and CHEMKIN software was used to solve the chemical reactions (Drayton et al., 1998). Their results indicated that the partial oxidation of methane in porous media occurs ignition and steam reformation processes. Kennedy et al. (2000) analyzed chemical structures of CH4/Air mixture in packed bed using one-dimensional model with GRI-Mech 1.2. Their analysis of the reaction pathway showed significant changes of the combustion mechanism from ultra-lean to ultra-rich conditions.</p><p>Dhamrat and Ellzey (2006) modeled transient filtration combustion of fuel-rich combustion in the range of equivalence ratio from 1.5 to 5 using a two-dimensional two-temperature model with GRI-Mech 3.0. Special attention was focused on the transient behavior of fuel-rich combustion. It was shown that high solid temperature zone enlarged as combustion wave propagated downstream, which was preferred to steam-reforming reaction and methane conversion efficiency was increased. In addition, they presented the effect of solid properties on the syngas production.</p><p>Toledo et al. (2009) modeled syngas production for multiple fuels in porous burner using a one-dimensional two-temperature model with GRI-Mech 3.0. Their predictions showed that fuel-rich partial oxidation in porous media could be used to reform C1-C3 gaseous fuels into hydrogen and syngas. Partial oxidation of methane in a porous reactor was investigated numerically by Zheng et al. (2012a) based on a two-dimensional two-temperature transient model with GRI-Mech 1.2. Results showed that both the gas and solid temperatures increased in the first 400 s and then the variation of maximum combustion temperature was negligible. To increase the CO and H2 yields, adding external heat energy for the combustion system was proposed by Zheng et al. (2012b), they modeled partial oxidation of methane in a RFB applying a one-dimensional two-temperature model with GRI-Mech 1.2. Results showed that CO and H2 yields increased significantly with the external heater power. An industrial reformer was numerically studied using two-dimensional model with GRI-Mech 3.0 and turbulent effect was taken into account (Xu et al., 2014). Their results showed that the methane conversion efficiency increased as the operating pressure was increased and 3.0 MPa was recommended for industrial operation.</p><p>Kostenko et al. (2014) modeled methane partial oxidation in a porous reactor using a one-dimensional two-temperature model with detailed kinetic mechanism including soot formation. Their predictions showed that the steam-soot reaction reduces soot formation and increases hydrogen production. An overall three-step six-component chemical kinetic model was developed for methane partial oxidation in porous media (Dobrego et al., 2008). The parameters of kinetic model were derived by adjusting to the experimental conditions and they suggested that the developed model could be combined with detailed chemistry.</p><p>To increase conversion efficiency, Dorofeenko and Polianczyk (2016) proposed a new version of RFB for syngas production, in which the methane-steam and air were supplied separately to the RFB. The air was directly feed into the RFB without preheating, whereas the methane-steam was preheated to high temperature prior to entering into the reaction zone by periodically changing the direction of its flow. Zeng et al. (2017) modeled fuel-rich partial oxidation of CH4/CO2 mixture in a two-layer burner using a two-dimensional two-temperature model with the detailed chemical kinetic (Peters). It was shown that the predictions of temperatures and major species matched well with the experimental results. Fuel-rich combustion in porous media was studied using skeleton diagrams and sensitivity analysis (Futko, 2003). They suggested that the combustion wave could be divided into three regions that are preheating zone, exothermic zone and endothermic zone, depending on the heat release.</p><p>To save computation time, for fuel-lean combustion in porous media, one-step kinetic still plays an important role in understanding the combustion and heat transfer processes using volume-averaged method (Liu et al., 2009; Fan et al., 2017, 2019; Wang et al., 2019). The influence of chemical reaction mechanisms on predictions of the filtration combustion has been conducted for fuel-lean condition (Hsu and Matthews, 1993; Mohammadi and Hossainpour, 2013). Hsu and Matthews (1993) numerically investigated the effect of chemical reaction mechanisms on the CH4/air premixed combustion characteristics in porous media under the condition of equivalence ratio <1. They concluded that it is essential to use multistep kinetics for accurate predictions of temperature profiles, species distribution, energy release rate and pollutant emissions. Mohammadi and Hossainpour (2013) simulated the experiment of Trimis and Durst (1996) applying a two-dimensional two-phase model with different four multistep kinetics that are GRI-Mech 3.0 mechanism, GRI-Mech 2.11 mechanism, the skeletal and 17 species mechanism (Peters). They studied the effects of these models on temperature, species profiles and pollutant emissions at the fixed equivalence ratio of 0.67 for CH4/Air mixture. Results showed that the four models have the same accuracy in predicting temperature distributions and the difference between these profiles was not more than 2%. In addition, the GRI-Mech 3.0 showed the best prediction of NO emission in comparison with the experimental data.</p><p>As reviewed above, a lot of numerical studies on fuel-rich combustion for syngas production in porous media burner have been conducted applying different detailed mechanisms, the influence of these kinetics on prediction of premixed combustion for fuel-lean combustion in porous burner has been revealed. However, it is still not certain the influence of the prediction of syngas production for fuel-rich combustion in porous media using different chemical reaction mechanisms. To save computation time, one may prefer to use a smaller detailed kinetic for modeling of syngas production in porous burner, but it was not known if this mechanism have the same accurate in predicting the combustion characteristics as the more detailed mechanisms. At the same time, one-dimensional model was widely used to save computational time and the two-dimensional study is scarce, which is more accurate to compute the heat loss to the surrounding through burner walls.</p><p>The aim of the present work is to investigate the influence of chemical reaction mechanisms on predictions of the temperature profiles, chemical structures and therefore the output of the syngas production in the exhaust gas, using a two-temperature two-dimensional model. In contrast to previous studies which emphasized on the effect of chemical kinetics on fuel-lean combustion in porous media, we explore the syngas production for fuel-rich combustion in porous media.</p><!><p>A two-layer porous burner reported by Zeng et al. (2017) is considered in the present work to study the sensitive of different kinetics to the predictions of syngas production in porous media. As shown in Figure 1, the burner was designed to study the effect of CO2 addition on the conversion efficiency of CH4 partial combustion within packed bed, which consists of two layers of alumina pellets with different diameters. In the upstream section the burner was filled with 2–3 mm alumina pellets that are 20 mm long, while the downstream layer was filled with 7.5 mm alumina pellets with length of 60 mm. The premixed mixture of methane with different amount of CO2 and air was feed into the burner and combustion waves were restricted near the interface of the upstream and downstream sections. For simplification, the pellet diameter in the upstream section is assumed to be a constant value of 2.5 mm. 2-D computations are considered in this work to save computational time. To simplify the problem, the following assumptions are made;</p><!><p>The alumina pellets are assumed to be inert homogeneous and optically thick media, the solid radiation is taken into account using the effective radiation thermal conductivity model.</p><p>Gas flow in the packed bed is assumed to be laminar and gas radiation is ignored.</p><p>The porosity variation near the tube wall is ignored and the thermal conductivity of the packed bed is the same for the two layers.</p><p>Schematic of the two-layer burner. All dimensions are in mm.</p><!><p>The chemistry is treated with two detailed kinetics and a reduced chemical mechanism in the computation. These mechanisms include two full mechanisms complied by the Gas Research Institute, namely GRI-Mech 1.2 (32 species, 177 reactions), GRI-Mech 3.0 (53 species, 325 reactions), and DRM 19 (Kazakov and Frenklach, 1994) (20 species, 58 reactions) which is a reduced mechanism based on GRI-Mech 1.2. The gas thermal and transport properties are obtained from the Chemkin and Tranfit packages (Kee et al., 1986).</p><p>Under the above assumptions, a set of differential equations can be obtained.</p><p>Continuity equation:</p><p>where ρg represents the gas density; vdenotes the velocity vector.</p><p>Vertical momentum equation</p><p>Where urepresents the vertical velocity; μ is dynamic viscosity, p is the pressure.</p><p>Horizontal momentum equation:</p><p>where v denotes the horizontal velocity. The pressure loss in the vertical direction is computed as (Ergun, 1952),</p><p>where ε is the porosity of the porous media, d is the pellet diameter, u′ is superficial velocity and u′ = εu.</p><p>Gas phase energy equation:</p><p>where Tg, λg, cg are the gas temperature, thermal conductivity and specific heat, respectively. Ts denotes the solid temperature; ωi, Wi are the chemical reaction rate and molecular weight of species i. hv is the volumetric convective heat-transfer coefficient between the gas and solid phases (Kaviany, 1994),</p><p>where Nuv, Pr, and Re is the Nusslet number, Prandtl number, and Reynolds number, respectively.</p><p>Solid phase energy equation:</p><p>Where λeff is the effective thermal conductivity of the porous media and can be expressed as λeff = λs + λrad, λs, and λrad are the solid thermal conductivity. We assume that the thermal conductivities for the two sections are same and the influence needs to be further studied. The radiative heat transfer coefficient of the alumina pellets, respectively. λrad is expressed as (Serguei et al., 1996).</p><p>Species conservation equation:</p><p>Where Yi is mass fraction of species i.</p><!><p>The following boundary conditions are specified in the model:</p><!><p>Inlet (10)Tg=Ts=300K,u=u0,v=0YCH4=YCH4,in,YO2=YO2,in,YCO2=YCO2,in,YN2=YN2,in</p><p>Outlet (11)∂Tg∂x=∂Ts∂x=∂(Yi)∂x=0.</p><p>Solid temperature at the inlet and outlet; (12)λeff∂Ts∂x=-εrσ(T4s,in/out-T04)</p><p>εr is the solid surface emissivity, σ is the Stefan-Boltzmann constant, T0 is ambient temperature.</p><p>At y = 15 mm, symmetry conditions are imposed; (13)∂Tg∂y=∂Ts∂y=∂Yi∂y=∂u∂y=v=0</p><p>Wall</p><p>At y = 0 mm, heat loss to the surroundings through the burner walls by convective heat transfer is considered and heat flux q˙ is defined as, (14)q˙=λδ(Twall-T0)</p><p>where λ is thermal conductivity of insulation, δ is thickness of insulation.</p><!><p>The governing equations presented above are numerically solved by a CFD software Fluent 15.0. To allow the gas and solid phases have different temperatures, user defined function and scalars provided by Fluent 15.0 are used to solve a separate energy equation for the solid phase. Radiative and conductive transport through the packed bed, convective heat transfer between the gas and solid phases is taken into account in the model, as shown in the above Equation (7).</p><p>The SIMPLE algorithm is used to handle the pressure and velocity coupling. At the downstream, the solid temperature with a thickness of 4 mm is set to be 1,800 K to model the ignition process. Mesh independence of the results are verified. The computational domain is discretized into 600 cells in the upstream section and 3,600 cells in the downstream section. When the solution is converged, the mesh of the reaction zone is densified.</p><p>A residual error of 10−6 for energy equations and 10−3 for all other equations are taken as convergence criteria.</p><p>The syngas energy conversion efficiency is defined as:</p><p>where LHVH2, LHVCO, LHVCH4 are the low heating value of H2, CO, and CH4, respectively. Table 1 presents the symbol used in this work.</p><!><p>Symbols used in this work.</p><!><p>In the experiment (Zeng et al., 2017) the equivalence ratio and air flow rate are fixed while the ratio (α) between the CO2 and CH4 is changed from 0 to 1. The solid thermal conductivity of alumina is 7.22W/m▪Kat T = 1,000 K, according to reference by Munro (1997). For volume average method used in this work, the thermal conductivity of packed bed is defined as 0.04 times of the solid thermal conductivity at T = 1,000 K as reference, this means that thermal conductivity of the packed bed is 0.2888 W/m·K unless otherwise stated. Table 2 shows the simulation cases carried out in this work. In the computation, for all computed cases the equivalence ratio is set to be a fixed value of 1.5 and the thermal conductivity of packed bed is varied due to uncertainty of its values. In the following, the temperature reported is along the centerline of the burner.</p><!><p>Simulation cases carried out in this work.</p><!><p>Figure 2 illustrates the predicted Tg and Ts at the centerline (y = 0 mm) for different α as well as experimental values for comparison. Figures 2A,B shows predicted Tg, Ts with GRI-Mech 3.0 at λs = 2.888 W/m·K. The calculated results show that the flame shifts from upstream of burner to the interface of two sections as α is increased from 0 to 1. One may expect that the maximum combustion temperature decreases with α due to injection of greater amount of CO2 to the burner, but the oppose trend is observed as shown in Figures 2A,B. One of the reasons for this phenomenon is due to the different flame stabilization positions where hv are different. hv is getting greater as the pellet diameter is decreased, as shown in Equation (6). At the reaction zone, the reaction heat is redistributed through convective heat transfer between the two phases. When the flame is stabilized near the interface, the convective heat transfer weakens and thus the temperature difference between the two phases is enlarged, which results to higher gas combustion temperature in the reaction zone for the smaller pellet diameter.</p><!><p>Temperature distributions along the centerline of the burner for different kinetics. (A) Gas, solid temperature for molar ratio of CO2/CH4 = 0, 0.25, and 10 times of thermal conductivity of packed bed with GRI-Mech 3.0; (B) Gas, solid temperature for molar ratio of CO2/CH4 = 0.5, 1, and 10 times of thermal conductivity of packed bed with GRI-Mech 3.0; (C) Gas temperature for molar ratio of CO2/CH4 = 0 with three different mechanisms; (D) Solid temperature for molar ratio of CO2/CH4 = 0 with three different mechanisms.</p><!><p>Figures 2C,D show the predicted Tg, Ts by three different kinetics for α = 0 at λs = 0.2888 W/m·K as well as experiment results (Zeng et al., 2017) for comparison. For α = 0, it can be seen that the predicted Tg, Ts are similar through the burner for the different three kinetics and the temperature differences predicted by the three kinetics are quiet small except for the reaction zone. In the pre-heat zone, the predicted Tg is almost independent of the mechanisms used. Small difference is observed downstream the reaction zone. The more detailed mechanism is used, the greater peak temperature is obtained in the computation as shown in Figure 2C. In the reaction zone the combustion temperature by the DRM 19 is highest and reaches up to 1,879 K, Tg by GRI-Mech 3.0 is lowest and the value is 1,786 K. In a word, the temperature distributions in the pre-heat and post reaction zone are almost independent of the mechanisms used, in the reaction zone the peak temperature predicted by different kinetics is rather small. Considering the computation cost and time, choosing a smaller mechanism for predicting temperature distribution in the burner may be a good choice and rather accuracy can be obtained, when the detailed components of the syngas is not taken into consideration. It is noted that the predicted Tg and Ts by the three kinetics are always greater than the corresponding experiment values, in the following this deviation will be discussed.</p><p>In the experimental study of Zeng et al. (2017), λs was not presented and the effect of λs on the syngas production was not clear. We test the sensitive of λs to Tg and Ts by increasing λs by 10 times with other parameters are fixed. The results with GRI-Mech 3.0 for α = 0 are shown in Figures 3A,B. One can see that λs has significant influence on the temperature distribution in the burner. For α = 0, the flame is stabilized near the burner inlet. The temperatures both for gas and solid phases in the entire burner decrease with λs. This is because an increase in λs leads to enhance heat condition in the solid phase, more heat is conducted through the solid phase, thus the temperature gradient is getting lower with λs. According to the Equation (12), the heat loss to the surrounding through burner inlet and outlet is increased as λs is increased, which leads to decrease in temperature in the burner. The predictions with 10 λs match well with experimental results and the predictions are greater than the experimental value as λs is decreased.</p><!><p>Effect of the packing bed thermal conductivity on gas and solid temperature with GRI-Mech 3.0 for molar ratio of CO2/CH4 = 0, 1. (A) Gas temperature for different solid thermal conductivities at CO2/CH4 = 0; (B) Solid temperature for different solid thermal conductivities at CO2/CH4 = 0; (C) Gas temperature for different solid thermal conductivities at CO2/CH4 = 1; (D) Solid temperature for different solid thermal conductivities at CO2/CH4 = 1.</p><!><p>For α = 1, as shown in Figures 3C,D, the flame stabilizes just downstream the interface, Ts distribution in the pre-heat zone is oppose to that for α = 0, Ts is increased with λs. This is attributed to the fact that the heat recirculation from the high solid temperature zone to the downstream direction is enlarged when λs is increased, which leads to increase in Ts in the first section of the burner, thus the gas mixture is effectively preheated and Tg is increased.</p><!><p>The major species in the exhaust gases includes H2, H2O, CO, CO2, and CH4. Consistent with the experiment (Zeng et al., 2017), the Figure 4 show the predicted species based on the wet base. Figure 4 presents the predictions of molar fraction of H2O, O2, H2, CO2, CO, CH4 by GRI-Mech 1.2 along the vertical direction for α = 0, 1 at 10 λs. For visible, the major species near the reaction zone is also presented in the Figure 4. According to Futko (2003), the combustion wave is composed of the preheating zone, an exothermic zone with partial combustion of methane in the reaction CH4+0.5O2 = CO+2H2(16), and an endothermic zone characterized by the reforming processes CO+H2O = CO2+H2(17),CH4+H2O = CO+3H2(18). For α = 0, it can be seen that extensive reaction occurs in a small region with length about 3 mm near the burner inlet. Methane begins to break down near the burner inlet, CH4 and O2 are quickly consumed in the exothermic zone. As shown in Figures 4A,B, all the syngas components almost peak after the exothermic zone in this case, only small change in major species is observed due to the reforming reaction. This indicates that the reaction (18) and (19) contribute little to the CO and H2 production in the endothermic zone in this case.</p><!><p>Distribution of major species (molar fraction) by GRI-Mech 1.2 along the centerline and near the reaction zone for molar ratio of CO2/CH4 = 0 and 1 at 10 times of thermal conductivity of packed bed. (A) Molar fraction of major species for CO2/CH4 = 0; (B) Molar fraction of major species at the reaction zone for CO2/CH4 = 0; (C) Molar fraction of major species for CO2/CH4 = 1; (D) Molar fraction of major species at the reaction zone for CO2/CH4 = 1.</p><!><p>For α = 1, the predicted distribution of major species is similar to that of α = 0, except for the distribution of CO2 in the exothermic zone. In this zone XH2 and XCO increase sharply and then XH2 increases slowly while XCO reduces slightly in the endothermic zone. As shown in Figures 4C,D, XCO2 decreases first in the reaction zone, this means that CO2 reacts and is consumed. Then XCO2 continually increases along the vertical direction. It can be seen from Figures 4A,B that, after the exothermic zone the variation rate of the major species for α = 1 is greater than that for α = 0. This is the effect of CO2 injection in the burner and also the effect of reforming reaction (17) and (18).</p><!><p>Figure 5 depicts the predicted major species of H2, CO, CO2, CH4 in the exhaust gases as a function of α for three different kinetics at λs = 0.2888 W/m·K. In Figure 5 the experimental and computational results with Peters by Zeng et al. (2017) are also shown for comparison. As shown in Figure 5A, both the experimental results and predictions show that XH2 decreases with increasing α from 0 to 1 and the predicted XH2 by three different kinetics are lower than the experimental values, whereas the computational results by Zeng et al. (2017) are greater than the experimental values. The predicted values by GRI-Mech 1.2 and GRI-Mech 3.0 match well with the experiment when the measurement error is taken into account. The prediction by DRM 19 also shows the same trend with experiment, but its values are significantly lower than the experimental values. For α = 1, the calculated results by GRI-Mech and Peters precisely predict the experimental results, this may be due to the dilution effect of the CO2 in the fuel.</p><!><p>Predicted molar fraction of H2, CO, CO2, and CH4 by three different kinetics. (A) Molar fraction of H2; (B) Molar fraction of CO; (C) Molar fraction of CO2; (D) Molar fraction of CH4.</p><!><p>Figure 5B shows the comparison of predicted YCO with experimental results and Zeng et al. (2017) for three different kinetics. The predictions by GRI-Mech 1.2 and GRI-Mech 3.0 are greater than the corresponding experimental results, as shown in Figure 5B, Qualitative agreement between the numerical and experimental results can be noted when the measurement error is taken into consideration. It is noted that the predictions by DRM19 deviates significantly from experimental results.</p><p>As demonstrated in Figures 5A–C, XCO increases while XH2 decreases with α. Meantime, XCO2 increases with α, although part of the CO2 reacts and is consumed in the exothermic zone as reactant when CO2 was injected into the burner. The reaction (17) is an important reformation reaction in the endothermic zone and the injection of CO2 into the burner may promote the inverse reaction (17), this may be responsible for the continue increase in XCO and decrease in XH2 when α is increased from 0 to 1.</p><p>Figures 5C,D presents predictions of XCO2, XCH4 by GRI-Mech, DRM19, and Peters (Zeng et al., 2017). All the combustion models precisely predict XCO2 for 0 ≤ α ≤ 1. Although all combustion models predict the trend of XCH4 with α, but the predictions are about two times of the corresponding experimental values.</p><p>To test the effect of λs on conversion efficiency, λs is varied from 0.2888 W/m·K to 2.888 W/m·K and the computation is conducted with GRI-Mech 3.0. These results are presented in Figure 6. As discussed above, a lower λs serves to reduce heat conduction through the solid phase and it is advantageous to form a high solid temperature zone around the reaction zone, thus results in a higher combustion temperature. The increased temperature drives the chemical kinetic to yield a higher percentage conversion. As shown in Figure 6A, XH2 always increases when λs is reduced for α ≤ 0.5. However, the effect of λs on the production of H2 diminishes for α = 1. The effect of λs on the production of CO is shown in Figure 6B, in which it can be seen that increasing in λs results in a small increase in XCO for α ≤ 0.25. Then the oppose trend is observed for α > 0.25, XCO decreases with α. A combined effect of on the syngas production is shown in Figure 6C, one can see that a decrease in λs results in a very small increase in the percentage conversion. For example, the conversion efficiency increases from 41.66 to 42.95% as λs is decreased from 2.888 to 0.2888 W/m·K for α = 0.5, thus the increase in conversion efficiency can be ignored.</p><!><p>Effect of packing bed thermal conductivity on molar fraction of CO, H2, and conversion efficiency with GRI-Mech 3.0. (A) Molar fraction of H2; (B) Molar fraction of CO; (C) Reforming efficiency.</p><!><p>The influence of chemical mechanisms on the syngas production for rich CO2/CH4 and air mixture combustion in a two-layer porous burner is investigated using a two-dimensional two-temperature model with GRI-Mech 1.2, GRI-Mech 3.0 and a reduced mechanism (DRM19) based on GRI 1.2. In the computation, the equivalence ratio is a fixed value of 1.5 while the ratio of CO2 to CH4 is changed from 0 to 1. The sensitive of predicted temperature distributions in the burner and major species in the exhaust gases to the mechanisms used in the model is conducted. The major conclusions from the present study are as follows;</p><!><p>Kinetic has no obvious influence on the temperature profiles in the burner expect for the narrow exothermic zone. The predicted temperature distributions by the three kinetics match well with experimental results. For predictions of temperature profiles without consideration of major species, using DRM 19 is recommended to save computational time.</p><p>The predicted major species (H2, CO, CO2) by GRI-Mech 1.2 and GRI-Mech 3.0 indicates that the two mechanisms have almost the same accuracy in predicting detailed components, little difference is observed for the whole investigated range. Thus, when the NOx emission is not focus of the study, GRI-Mech 1.2 is recommended to save computational time. However, the predicted molar fraction of H2 and CO by DRM 19 is under-predicted compared to experimental values. The three kinetics over-predicted the molar fraction of CH4 by a factor about two times of the experimental values.</p><p>Thermal conductivity of the porous media used in the burner has significant effect on predicting the syngas productions. Increase in the thermal conductivity leads to a decrease in the combustion temperature, and thus increases in H2 and decreases in CO, a very small increase in conversion efficiency is observed when the thermal conductivity is decreased by a factor of 10 times and this effect can be ignored.</p><!><p>All datasets generated for this study are included in the article/supplementary material.</p><!><p>JS, MM, and HL conceived and designed the study. YoL analyzed the data. YaL and YD wrote 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><!><p>Funding. The authors wish to acknowledge the support to this work by the National Natural Science Foundation of China (Nos. 51876107) and SDUT & Zibo City Integration Development Project of China (Grant No. 2018ZBXC084).</p>

PubMed Open Access

Attractive force-driven superhardening of graphene membranes as a pin-point breaking of continuum mechanics

Bending at the nanometre scale can substantially modify the mechanical, chemical and electronic properties of graphene membranes. The subsequent response of chemical bonds leads to deviations from plate idealisation in continuum mechanics. However, those phenomena have thus far been investigated exclusively by measuring the electronic properties of graphene deformed by compressing and stretching with local-probe techniques. Here, we report that the interatomic-attractive forces applied on the convexly-curved graphene by the probe tip give rise to a pin-point breaking of the plate idealisation in the continuum mechanics, facilitating atomically-localised enhancements in its chemical reactivity and mechanical strength. Thorough characterisations were conducted by atomic force microscopy and force field spectroscopy on hollow nanotubes, rolled-up graphene, with different diameters. Their topmost parts supplied well-defined curvatures of the convex graphene. We found that a significant enhancement in the out-of-plane Young's modulus from 13 to 163 GPa, "superhardening", was realised with the nonlinear transition of bond configurations. Our findings provide a fundamental understanding of the relationships between the structure of atomistic membranes and the dynamic behaviour of approaching exterior atoms or molecules and their subsequent interplay with chemical and mechanical properties. Thus, these results encourage the application of such membranes in functionally-controllable materials or devices.Graphene is a single atomic layer of sp 2 -bonded carbon atoms arranged in a two-dimensional (2D) honeycomb lattice and a basic building block for graphitic materials of all other dimensionalities, such as 0D fullerenes (spherical graphene), 1D nanotubes (rolled-up graphene), and 3D graphite (stacked graphene) [1][2][3][4][5] . Owing to the unique combination of an extremely small out-of-plane stiffness with a high in-plane modulus (~1000 GPa) and tensile strength (~100 GPa), the behaviour of curved graphene is of fundamental importance for studying graphene-based nanostructures ranging from 0D to 3D and for their application in a variety of devices 4,[6][7][8][9][10] . The bending properties not only control the morphology of graphene under external stimuli [11][12][13][14] but are also related to its electronic, magnetic, and chemical properties [1][2][3]6,[15][16][17][18][19][20] . The carbon atoms located within the plane of graphene are chemically inert due to π-conjugation, whereas the curved π-conjugation in the carbon networks of curved graphene has not only π-character but also substantial σ-character (i.e., π-σ re-hybridisation) 21,22 . According to the "π-orbital axis vector" (POAV) theory, carbon atoms that reside on highly curved surfaces exhibit increased chemical potential due to diminished electronic delocalisation [22][23][24][25][26][27] . When the local curvature is on the nanometre scale, the electronic structure is substantially modified by altering the π-orbital energy and modifying the nearest-neighbour hopping integrals, which can induce a local shift in the electrochemical potential 28 and give rise to large pseudomagnetic fields 29 .Regarding the characterisation of the mechanical properties of curved graphene (or hollow nanotubes), there are always concerns about the applicability of existing continuum mechanics theories 6,[30][31][32] . Although only atomically thick, graphene membranes under bending can be still described by these theories 33,34 . However, they usually require slowly varying, harmonic deformation conditions. These conditions are violated in realistic situations, such as sub-nanometre ripples or out-of-plane displacements of individual atoms in the carbon networks,

attractive_force-driven_superhardening_of_graphene_membranes_as_a_pin-point_breaking_of_continuum_me

<!>3D topographies and force fields on convexly curved graphene<!>Correlation of interatomic forces and potentials to the curvatures<!>Chemical activation and superhardening of curved graphene<!>Applicability of our findings<!>Method

<p>which may be beyond first-order continuum elasticity 30,31,35 . This calls for a fundamental study of the geometry of atomistic membranes and their subsequent coupling to electronic degrees of freedom, down to unavoidable atomic-scale fluctuations 11,12,35 . The discrete geometry is relevant for addressing spin diffusion in rippled graphene 35,36 as well as for understanding the chemical properties of nonplanar 2D crystals 37 , and it may even be important for the strain engineering of 2D crystals with topological defects 35 .</p><p>Under a pure bending distortion of single-layered graphene, likely caused by being rolled-up around an arbitrary axis into a hollow nanotube (Fig. 1a) from a plane (Fig. 1b), each carbon atom and its three nearest neighbours are no longer planar but are instead located in the corners of a pyramid. This pyramidalization is accounted for using the POAV construction, as indicated by arrows in the insets of Fig. 1a,b [23][24][25][26][27]30 . The geometrical tilting of σ i -bonds (i = 1, 2, 3) by an angle θ p (Fig. 1a) is accomplished in POAV by introducing a degree of p z atomic orbital mixing into the σ i framework. Note that to the first order in curvature (1/R), the three tilting angles as well as the bond lengths are common 38,39 . Remarkably, the pyramidalization angle θ p is sufficient for describing the curvature-induced shift in sp 2 hybridisation 30 and is useful for gauging the reactivity of the carbon atom sites of the curved graphene 21 .</p><!><p>The nanotubes used in our study were single-walled carbon nanotubes (SWNTs); their original radii R o ranged from 6.3 to 9.2Å 40 . The individual R o values were determined by comparing the overall heights in their topographies with those for the well-defined standard of R o = 6.9 ± 0.1 Å 41 . The validity is based on the finding that the 2 to sp 3 bonding configurations for the rolled-up form of graphene, i.e., nanotube, and silicon AFM tip apex (a), of the n-orbital in the planar graphene (b), and of dynamic-mode AFM imaging of the nanotube isolated on a planar substrate (c). The n-orbital axis vector (POAV) is indicated by an arrow for a conjugated carbon atom (• ) bonded with the nearest neighbours (○ ) by the σ i -bond (i = 1, 2 and 3). The pyramidalization angle θ p is defined by the angle between the POAV and σ i -bond minus 90°, as depicted in (a). (d-f) Atomic-resolution AFM topographies of the nanotubes with different radii of curvature in 3D views 58 . The scale bar is 1 nm. The nanotubes were sparsely deposited on an atomically flat substrate. Atomically resolved AFM topographies were obtained in ultrahigh vacuum (< 10 overall heights are linearly correlated to R o as long as < 10 Å. The R o values obtained in this way show good agreement with those obtained in advance using radial breathing modes in Raman spectroscopy 40 . The topographies, representing slender and convexly curved features, as shown in Fig. 1d-f, were measured over the individually isolated nanotubes on the same substrate by atomic force microscopy (AFM) 42 with the same silicon (Si) tip with an atomically sharp apex. Atomically resolved topographies enable us to determine the chiral indices (n, m) that are utilised to confirm the accuracy of the evaluated R o values 43 . The R o values of the nanotubes in Fig. 1d-f were found to be 8.1 Å, 7.5 Å, and 6.5 Å, respectively, with a standard deviation of 3.8%(± 0.25 Å) at the maximum.</p><p>The atomically resolved topographies obtained with the same tip represent characteristic features unique to the respective nanotubes with different radii ( ≠ ′ R R o o ), as three-dimensionally demonstrated in Fig. 1d-f. The upper part of the slender and convexly curved features in Fig. 1d, corresponding to the topmost area of the R o = 8.1 Å nanotubes 43 , exhibits corrugations with atomic-scale periodicities. On the other hand, those atomically corrugating features explicitly decline in Fig. 1e and become inconspicuous in Fig. 1f, corresponding to the R o = 7.5 Å and 6.5 Å nanotubes, respectively.</p><p>In 3D force fields F(x, y, z) 44,45 , the regions in which interatomic attractive forces attain their maximum values, i.e., the blue-coloured areas in Fig. 2a,d, correspond to the ridges of the convexly curved single-layered graphene (i.e., the hollow-tubes' upper halves), as illustrated in Fig. 1a. Thus, the blue-coloured areas in Fig. 2a,d represent the features unique to their different curvatures. The F(x, y, z) over the R o = 8.1 Å and 6.5 Å nanotubes are displayed, respectively, in Fig. 2a,d within a rectangular parallelepiped (10 × 10 × 6 Å 3 ). F(x, y, z) represents the spatial distributions of the interatomic forces acting exclusively on the foremost atom of the tip apex over the ridges where the 3D topographies in Fig. 1d,f were obtained. The interatomic forces were derived by subtracting long-range background forces acting comprehensively on the tip apex towards horizontally wide and perpendicularly intersecting sample areas, including steep sidewalls and plane substrates (Fig. 1a,c).</p><p>A comparison of the 3D force fields with the 3D damping fields 43,45 simultaneously measured revealed that the foremost atom of the tip apex exclusively contributes to the elastic interactions with the individual carbon (C) atoms of the central ridge. The 3D damping fields U dmp (x, y, z) in Fig. 2b,e three-dimensionally specify the locations in which inelastic interactions occurred within the same parallelepiped as shown in Fig. 2a,d, respectively. Indeed, they are almost completely absent (< 3 meV) in Fig. 2b and over the shown in Fig. 2e area, except lower peripheral areas, meaning that the interatomic interactions are elastic over the whole ridge and central ridge of the R o = 8.1 Å and 6.5 Å nanotubes, respectively.</p><p>The contrast in the radial force maps was found to be closely related to the corrugation amplitudes in the 3D topographies, where the atomic features were very prominent for the larger R o = 8.1 Å (Fig. 1d) but inconspicuous for the smaller R o = 6.5 Å (Fig. 1f). The convexly curved sections radially crossing the middle of the blue-coloured regions in F(x, y, z) (Fig. 2a,d) are presented as "radial force maps" F(x, θ) (Fig. 2c,f, respectively). They almost dependably trace the ridge of the convexly curved graphene. The F(x, θ) maps in Fig. 2c,f are rescaled by individual colour codes in which the least upper and greatest lower bounds are, respectively, set to the minimal and maximal forces. The contrast between the red-and blue-coloured spots in F(x, θ), i.e., the difference between the attractive-force minima and maxima, reaches approximately 40pN for R o = 8.1 Å (Fig. 2c), whereas it is nearly half (~20 pN) for R o = 6.5 Å (Fig. 2f).</p><!><p>The red-and blue-coloured spots in Fig. 2c,f, designating the locations of the "relative" minima and the maxima in F(x, θ), can be assigned to the C-atom and hollow sites, respectively, because the interatomic attractive force F(z) acting on the tip-apex atom over the red-coloured spots in F(x, θ) was found to be clearly dependent on the radius R o , whereas the F(z) curve over the blue-coloured spots showed no clear dependence on R o . The F(z) at the normal z position in the out-of-plane direction over the sites corresponding to the red-and blue-coloured spots in F(x, θ) is plotted in Fig. 3a,b, respectively, for the nanotubes of the four different original radii R o (8.1 to 6.5 Å). The z position is arranged to be the equilibrium z o (= 3.35 Å) position in the case F(z) = 0. Each plot in Fig. 3a,b is the mean of F(z) in F(x, y, z), respectively, corresponding to the red-and blue-coloured spots around the central ridge in F(x, θ), where the 3D topographies (Fig. 1d-f) exhibit the corrugating features, and U dmp (x, y, z) (Fig. 2b,e) represents the elasticity.</p><p>Figure 3a shows that the strength of F(z) over the red-coloured spots in F(x, θ) is negatively correlated with the original radius R o , that is, positively correlated with the original bending curvature 1/R o . The positive correlation between F(z) and 1/R o may conflict with the expected negative correlation of "nonbonding" interactions. Over the C atoms of graphene, the π-orbitals forming a reciprocal weak bond (i.e., π-bond) predominantly contribute to the attractive forces acting on the tip-apex atom, unless any electron-transfer reactions occur 5,46,47 . Since not only the closest atom but also the nearest neighbours additively contribute to such "nonbonding" interactions, following the inverse power law, the latter's contributions decrease as their distances d int to the tip-apex atom increase with 1/R o , as depicted in Fig. 1g. On the other hand, Fig. 3b shows that the F(z) curves over the blue-coloured spots in F(x, θ) were found to be more independent of 1/R o . The framework of the hexagonal ring is thought to preserve its original structure even in rather heavily curved graphene. The six individual C atoms contribute "nonbonding" interactions equally and are always hexagonally arranged around the hollow site.</p><p>The depths of the potential wells of the tip-apex atom over the locations assigned to the C atoms show a quadratic relationship to the original bending curvature 1/R o of the convexly curved graphene. For F(z) in Fig. 3a, the mean of the potential U(z), averaging over the locations corresponding to the red-coloured spots around the central ridge in F(x, θ), is plotted in Fig. 3c F(x, y, z) was derived from the short-range term of the frequency shift, i.e., Δ f sht (x, y, z), using Sader's formula 46,47 . The Δ f sht (x, y, z) was derived by subtracting the long-range background term from the frequency shift Δ f(x, y, z) originally obtained when retracting the tip during the measurement [45][46][47] . (b,e) The 3D damping fields U dmp (x, y, z) simultaneously obtained with F(x, y, z) in (a and d), respectively. The locations exhibiting slight amounts of inelastic interactions (10-20 meV) in e correspond to the sidewalls of the nanotube with the smaller radius (R o = 6.5 Å), as displayed in (d). The locations further apart from the central ridge consist of the steeper sidewalls, in which not only the foremost atom of the tip apex but also its nearest-neighbouring atom was thought to non-elastically interact with the sidewall. (c,f) The radial force maps F(x, θ) corresponding to the convexly curved sections radially crossing the middle of the bluecoloured regions (shaped similar to a "barrel roof ") in (a and d). F(x, θ) almost dependably traces the ridge of the convexly curved graphene, and thereby the curved surfaces in (c and f) directly represent the differences in their curvatures. The red-and blue-coloured spots in F(x, θ) correspond to the carbon atom and hollow sites of the hexagonal honeycomb lattice of the convexly curved graphene, respectively. It should be noted that in F(x, θ), the successive distributions of the blue-coloured spots in specific directions might be induced by superposition of the interactions successively acting on the "dimer row" of the tip-apex atoms arranging only in the specific direction, as depicted in Fig. 1a. Hence, by excluding the directions exhibiting those artefacts, the force maps showing atomic arrangements certainly enable us to quantitatively analyse the attractive interactions on the atomic scale.</p><p>indicates that |U o | (i.e., binding energy) is proportional to the square of 1/R o . Thus, the inset of Fig. 3d supports the validity of the continuum mechanics theories 6,30,46,47 .</p><p>To describe the interatomic potentials of curved graphene, Kostov et al. proposed a simple bond parameter of the "mixed" state, consisting of the linear combination of the sp 2 and sp 3 bond states by introducing a curvature parameter, g(1/R), and using the corresponding bond parameters, X sp 2 and X sp 3 , respectively 48 . This method is based on the interatomic potential functions developed for carbon atoms with sp 2 and sp 3 hybridisation and derives new parameters for carbon atoms with π-σ re-hybridisation explicitly dependent on the curvature 48 . We adopt this method to describe U(z) in our study. As such, |U o | can be described as a function of 1/R using the "mixed" state based on the linear combination of the corresponding sp 2 and sp 3</p><p>where the curvature parameter g(1/R) is defined as binding energy for single-layered graphene 49 . The reference constant 1/R t was based on the radius of curvature R t = 5.8 Å of the tip apex, estimated by analysing the 3D topographies 51 in our study.</p><p>Figure 3c shows approximate curves to experimental-data plots that were derived from the expansion of Eq. ( 1), where the Lennard-Jones and Morse potential functions were adopted to describe the bond parameters,</p><p>, respectively, corresponding to the sp 2 and sp 3 hybridisations, as follows:</p><p>where the decay length parameter λ in the Morse potential was individually estimated to find an excellent fit to the experimental-data plots. As illustrated in Fig. 3f, the normal position of the C atom was set to the origin such that the tip-apex atom is located at the equilibrium z o (= 3.55 Å) position. By adopting the "lift" displacement z lft of the C atom, corresponding to the relaxation originating from the interatomic attractive forces applied by the tip-apex atom, the interval z int between those two atoms is properly described as = − z z z int l ft . The approximate curves to the force plots in Fig. 3a were obtained by differentiating Eqs (3a) and (3b) in the interval z int , as follows:</p><p>In contrast, differentiating only Eq. (3a) yields the approximate curves to the force plots in Fig. 3b, over the hollow sites, exhibiting no clear dependence on the curvatures.</p><p>Since single-layered graphene has an extremely small out-of-plane stiffness 6,7 , the closest C atom of the convexly curved graphene is expected to be lifted towards the tip-apex atom due to the interatomic attractive force in close proximity, as depicted in Fig. 3f. Consequently, 1/R would locally increase further from 1/R o . The absolute values of the minima in F(x), i.e., the attractive-force maxima F a , are plotted as a function of 1/R in Fig. 3d, where the 1/R values were rearranged taking the "lift" displacement z lif into account. The solid curve . Assuming 1/R rigidly stays at 1/R o without any relaxation, then the relationship of F a versus 1/ R o is additionally given by the plots with open (small) markers and their approximate (dashed) line in Fig. 3d. The force curve estimated under this assumption is represented by the dotted line in Fig. 3e, showing large deviations in the normal z direction from the experimental-data plots for R o = 6.9 ± 0.1 Å. In contrast, the approximate curve (solid line), showing excellent agreement with all the experimental-data plots in Fig. 3e, was obtained by rearranging 1/R. Indeed, the approximate curves in Fig. 3a, showing excellent agreement with all experimental-data plots, were derived from Eq. ( 4) using the rearranged 1/R in Eq. ( 2).</p><p>An empirical analysis of the experimental finding of how much 1/R would locally increase from 1/R o revealed that within the first-order approximation, the local increment of the curvature, i.e., ∆ ≡</p><p>/ / lft o , would not be inversely but would be directly proportional to the "lift" displacement z lft :</p><p>The z lft and ∆ 1 R ( / ) lft values were estimated in the process of determining the approximate curves to the experimental-data plots in Fig. 3a,c. The resultant positive value β is a linear coefficient corresponding to an increasing rate and can be expressed as a linear function of 1/R o , as follows:</p><p>where the linear coefficient γ and the lowest limit of the strained curvature 1/R II were estimated to be 7.770 nm −1 and 1.185 nm −1 , respectively, from the values shown in Table 1.</p><!><p>Figure 4a explicitly shows that the out-of-plane elastic stiffness k S of the convexly curved graphene [Method] attains a much larger maximal value k s max under the maximal "lift" displacement z lft max than the original value k s o at the equilibrium z o position (z lft = 0). The upper two variations of the plots in Fig. 4a show that k s max (at z lft max ) represents conspicuous differences between the C-atom and hollow sites in the variations as a function of the maximal-strained curvatures In the case that the tip-apex atom is located at the equilibrium z o position (z lft = 0), the out-of-plane Young's modulus ≡ E E s s o , ranging from 7.1 to 13GPa [Method], was found to be independent of any atomically specific site and almost uniform over the whole ridge, following the relationship</p><p>, where n was estimated to be 2.67 and 2.87 at the C-atom and hollow sites, respectively. These results show excellent agreement with those of many previous reports on the nanoindentation and compression of carbon nanotubes by the AFM tip, for which the Hertzian model based on the plate idealisation of continuum mechanics is still applicable 52 . On the other hand, the maximal ≡ E E s s max under the maximal "lift" displacement z lft max explicitly demonstrates a conspicuous disparity or difference between the C-atom and hollow sites, resulting in tremendous atomic-site dependency. The maximal-strained curvature (1/R) max and E s max were found to follow the relationship</p><p>where n was estimated to be 2.80 at the hollow sites but 3.47 at the C-atom sites, as in the case of k s max . However, more interestingly, the C-atom sites exhibit a much more pronounced dependence of E s max on (1/R) max , attaining a significantly large value of</p><p>, which is almost comparable to that of silicon (i.e., the tip-apex material) with a so-called diamond structure holding sp 3 orbitals in a tetrahedral framework 53 .</p><p>To elucidate the reason why E s is dramatically higher at the C-atom sites under the maximal "lift" displacement z lft max , the pyramidalization angle θ p (see Fig. 1a) was evaluated based on the "lift" displacement z lft of the C atom and its relationship to the local curvature increment ∆ 1 R ( / ) lft . The individual θ p at the equilibrium z o position (z lft = 0) was first derived for the respective 1/R o of the four different nanotubes, taking their helical indices (n, m) into account. Figure 4c demonstrates the variations of θ p as a function of the normal z position of the tip-apex atom, in which the plots at the left ends and the maxima correspond to the original θ p o at the equilibrium z o position (z lft = 0) and the maximal θ p max under the maximal z lft max , respectively. Furthermore, the plots of E s o versus θ p o and E s max versus θ p max are shown in the bottom and middle groups of in Fig. 4d, respectively, together with a data point (open triangle) for the tetrahedral bond angle (θ p = 19.5°) in the sp 3 hybridisation of the diamond (E s ≅ 1TPa), as the upper limit. Interestingly, all the plots were almost on a parabolic line, indicating their quadratic correlation (i.e., E s ∝ (θ p ) 2 ).</p><p>Furthermore, to gain clear insight into the relationship of E s versus 1/R, as demonstrated in Fig. 4b, the variations of θ p were replotted as a function of 1/R in the inset of Fig. 4d. The individual data sets obtained for the four different nanotubes showed that the respective θ p values linearly increased with ( / ) lft was found to be directly proportional to z lft , as in Eqs ( 5) and ( 6), where the linear coefficient β also linearly increased with 1/R o . In contrast, the slope of the rows in the inset of Fig. 4d 4b. Consequently, the dramatic increase in E s max at the C-atom sites was found to be closely related to the nonlinear increment of θ p max , reaching up to 8.4° at the end. The increase in θ p would be accompanied by an increase in the out-of-plane attractive potential according to the POAV theory, in which the degree of the valence orbital hybridisation depends on θ p , as illustrated in Fig. 1 [23][24][25][26][27] : a slight increment in θ p leads to a continued weak π-state following sp 2 hybridisation, whereas its further increment towards the tetrahedral bond angle (θ p = 19.5°) yields a transition towards the chemically radical σ-state of a dangling bond following sp 3 hybridisation. Intermingling the σ-state of chemically radical dangling bonds with the nonbonding π-state in the transition from the sp 2 to sp 3 hybridisation (i.e., π-σ re-hybridisation) triggered by the increment in θ p (3.1° to 8.4°) certainly explains not only the increase in F(z) but also the significant enhancement of E s (13 covered by continuum mechanics. In contrast, the unexpectedly great variation in E s max for the maximal-strained curvatures (1/R) max , i.e., under the maximal "lift" displacement z lft max specifically at the C-atom sites, could be attributed to the result from the π-σ re-hybridisation, and thereby indicates an atomically pin-point breaking of the continuum mechanics.</p><!><p>Very recently, the functionalisation of graphene, especially hydrogenation, has attracted much attention for two main reasons: it can be used to tune the band gap for realising semi-conducting behaviour with a high carrier mobility, and it can also be harnessed as an energy-conversion/storage material 7 . For the case of hydrogenation, the ripples with large-curvature, likely narrow or highly curved nanotubes, usually with a diameter < 1 nm, have thus far been considered necessary 11,12 for binding hydrogen 54 , leading to the hybridisation of carbon atoms from sp 2 into sp 3 , and thereby removing the conducting π-bonds and opening an energy gap [54][55][56][57] . However, our findings suggest that the interatomic attractive forces applied by any inactive atom or molecule beyond the tip-apex atom could trigger the transition of its bond state from sp 2 to sp 3 hybridisation, although the original radius of curvature in convexly curved graphene is larger than 5 Å (i.e., > 1 nm in diameter). Furthermore, the significant strength enhancement of the out-of-plane elasticity (i.e., superhardening) discovered by our study suggests that the interatomic attractive forces acting between nanostructured graphene and other components would play an important role in enhancing the mechanical strength of composite materials.</p><!><p>Out-of-plane elastic stiffness. Using the "lift" displacement z lft of the C atom, the interatomic forces acting on the tip-apex atom can be expressed as</p><p>) because their interaction was found to be elastic, as demonstrated in U dmp (x, y, z) (Fig. 2b,e). The elastic stiffness k S of the convexly curved graphene was expected to vary with the local increment of the curvature ∆ 1 R ( / ) lft , with a linear correlation to z lft , as described in Eq. (5). Since the applicability of Hooke's law is guaranteed for the small displacement ∆z lft , the elastic stiffness k S can be derived from the infinitesimal force change dF for the infinitesimal displacement dz lft as k S = dF/dz lft . Indeed, the k S value was found to be almost constant k S o around z lft = 0 and attained the maximum k S max at the maximal "lift" displacement z lft max . Those k S o and k s max values were plotted as a function of 1/R for the C-atoms and hollow sites in Fig. 4a to examine the site dependency on the atomic scale (see illustrations in the inset).</p><p>The out-of-plane Young's modulus. E s was evaluated at the C-atom and hollow sites based on the simple model, in which the effective areas of interatomic attractive forces applied by the tip-apex atom were estimated taking into account their variations dependent on the "lift" displacement z lft of the closest C atoms. As illustrated in Fig. 4a,b, the constituent bond elements playing the leading roles are the following: (i) the three in-plane σ-bonds surrounding the closest C atom in the case where the tip-apex atom is located directly over it and pulling it up by the maximal "lift" displacement z lft max ; and (ii) the six in-plane σ-bonds surrounding the hexagonal ring in the case where the tip-apex atom is directly over the hollow site and = z z lft l ft max H ( ) . In the case where the tip-apex atom is located at the equilibrium z o position (z lft = 0), where the convexly curved graphene has a curvature of the original 1/R o and is free from any local strain, as illustrated in Fig. 3f ( ) ), the corresponding area estimated, and Eq. ( 5) to Δz, A, and 1/R, respectively.</p>

Scientific Reports - Nature

A Photoprotective Effect by Cation-\xcf\x80-Interaction? Quenching of Singlet Oxygen by an Indole-Cation-\xcf\x80 Model System

We investigated the effect of the cation-\xcf\x80 interaction on the susceptibility of a tryptophan model system towards interaction with singlet oxygen, i.e. Type II photooxidation. The model system consists of two indole units linked to a lariat crown ether to measure the total rate of removal of singlet oxygen by the indole units in the presence of sodium cations (i.e. indole units subject to a cation-\xcf\x80 interaction) and in the absence of this interaction. We found that the cation-\xcf\x80 interaction significantly decreases the total rate of removal of singlet oxygen (kT) for the model system, i.e. (kT = 2.4\xc2\xb10.2)\xc3\x97108 M\xe2\x88\x921sec\xe2\x88\x921 without sodium cation vs. (kT = 6.9\xc2\xb10.7)\xc3\x97107 M\xe2\x88\x921sec\xe2\x88\x921 upon complexation of sodium cation to the crown ether. Furthermore, we found that the indole moieties undergo Type I photooxidation processes with triplet excited Methylene Blue; this effect is also inhibited by the cation-\xcf\x80 interaction. The chemical rate of reaction of the indole groups with singlet oxygen is also slower upon complexation of sodium cation in our model system, although we were unable to obtain an exact ratio due to differences of the chemical reaction rates of the two indole moieties.

a_photoprotective_effect_by_cation-\xcf\x80-interaction?_quenching_of_singlet_oxygen_by_an_indole-ca

INTRODUCTION<!>MATERIALS AND METHODS<!>Synthesis of the Bis-indole-crown (1)<!>Singlet Oxygen Luminescence Quenching Experiments.<!>Steady-state irradiation of compounds 1 and 2.<!>RESULTS AND DISCUSSION<!>CONCLUSIONS<!>

<p>Singlet molecular oxygen, the lowest excited state of the dioxygen molecule, is well known to react with several amino acid residues. In cellular systems, proteins are one of the main targets of singlet oxygen, leading to oxidized amino acid residues which may affect the structure and function of the protein (1–7). Generally, histidine, tyrosine, and tryptophan as well as the sulfur-containing amino acids methionine and cysteine are most reactive with singlet oxygen. Rate constants of the reaction of these free amino acids with singlet oxygen in solution have been known for several decades to be generally on the order of 107 M−1sec−1 (8–10). However, in recent years it has become clear that within a protein matrix, these rate constants may vary over a wider range. In some cases, native proteins having been found to react with singlet oxygen with rate constants on the order of 108-109 M−1sec−1(10). Perhaps not surprisingly, steric accessibility of the amino acid residue may cause a large amount of variation in these reaction rate constants (11). Elegant work by the group of Ogilby has shown that electronic effects also cause a significant amount of variation in the susceptibility of a specific amino acid residue within a protein towards photooxidation (12).</p><p>The cation-π interaction, which involves an electron-rich π-system and a cation (commonly a simple alkali metal ion or a quaternary nitrogen atom), is one of the most important noncovalent interactions in biochemistry. This interaction plays an important role in protein structure involving the amino acid side chains of tryptophan, tyrosine, histidine, and, to a lesser extent phenylalanine (13–18). We noticed that the amino acids which are most reactive with singlet oxygen at their conjugated π-systems (histidine, tyrosine and tryptophan) are also the amino acids involved in the stabilizing electrostatic cation-π interaction observed in protein structures and in biological recognition. This led us to hypothesize that the cation-π interaction may possibly affect the susceptibility of these amino acids towards degradation by singlet oxygen – in other words, in addition to its key role for protein structure, does the cation-π interaction also possess a photoprotective aspect?</p><p>The most common residue involved in cation-π pairing is tryptophan (19). Free tryptophan in solution scavenges singlet oxygen with a rate constant of 3.2×107 M−1sec−1, leading to a very complex degradation process with a large number of oxidized products (9, 20–23). The initial attack of the singlet oxygen molecule occurs via a [2+2] cycloaddition on the π-bond of the indole system (as opposed to the benzene ring). Interestingly, the indole moiety is also the preferred π-donor in the cation-π interaction of the tryptophan molecule (24). Therefore, we chose a tryptophan-based model system to investigate the possible photoprotective effect of the cation-π interaction on tryptophan residues, namely the model system originally developed by Gokel and coworkers in which a tryptophan residue can be either free or completely bound to a cation by an intramolecular cation-π interaction (Fig. 1), with the equilibrium concentration of free tryptophan being negligible (24).</p><!><p>All reagents were purchased from Sigma Aldrich or TCI, were verified by 1H NMR, and used without further purification unless otherwise specified. The control lariat crown dibenzyl ether 3 was also obtained from Sigma Aldrich. All solvents were purchased from either Sigma Aldrich or Fischer Scientific and were HPLC grade. Deuterated solvents were obtained from Cambridge Scientific.</p><!><p>Compound 1 was prepared by a modified literature procedure (24): A 500 mL three neck flask was charged with 1,10-diaza-18-crown-6 (0.787 g, 3.00 mmol), sodium carbonate (1.590 g, 15.00 mmol), sodium iodide (0.0899 g, 0.600 mmol), and a magnetic stir bar. The flask was fitted with an addition funnel with glass frit (left), reflux condenser (center) and a tightly fitting rubber septum (right). All joints were grease-sealed. The top of the condenser was sealed with tightly fitting rubber septum. To the addition funnel, 3-(2-bromoethyl)indole (1.345 g, 6.000 mmol) was added, and the top was also sealed with a tightly fitting rubber septum. Using a Schlenk line, the entire apparatus was purged with N2. Using cannula techniques, 30.0 mL of dry and degassed acetonitrile was added to the addition funnel (a glass syringe degassed with N2 was required to disturb the solution in the addition funnel so that it became homogenous), while 50.0 mL was added to the three neck flask and stirred. Positive N2 pressure was established through the system. After ~15 min of reflux, the bromo-indole solution was added dropwise to the three neck flask. More rapid addition would lead to formation of intractable polymeric compounds. Addition of the bromo-indole was completed after three hours. The entire apparatus was covered in foil and left to reflux for 48 hours. Intermittently, the reaction was monitored by TLC (6:4 EtOAc:Hexanes and 2:8 MeOH:Acetone) and 1H NMR.</p><p>Upon reaction completion, the crude product was collected via filtration and concentrated under reduced pressure. The crude was then washed three times with DCM and water to remove any salt. The organic layer was collected and concentrated under reduced pressure. To this dark yellow solution, ~3 g of high grade (43–75 μm) silica gel was added, making a slurry. This slurry was fully mixed, then dried via rotary evaporator and vacuum oven (~35 °C) overnight. The following day, the solid was transferred to a flash column chromatography holder; a 40 g silica flash column was used for separation. The product was separated in stages (EtOAc was eluent for the first 30 minutes, then 20% methanol in acetone for the remaining time). The flash column was monitored using UV light (λ = 254 nm and 280 nm). Fractions with UV absorbance were collected and concentrated. The purest (determined by 1H NMR) fractions corresponded to the third column peak. The solid from these fractions were left to recrystallize out from cold acetone for three days. The desired bis-indole-crown was collected (0.170 mg, >99%) and verified by UV-vis (peaks: λ = 223 nm 282 nm), and NMR.</p><p>1H NMR (400 MHz, CD3COCD3) δ: 10.00 (s, 2H), 7.58 (d, J = 8, 2H), 7.37 (d, J = 8, 2H), 7.25 (s, 2H), 7.08 (t, J = 8, 2H), 7.00 (t, J = 8, 2H), 3.62 (t, J = 6, 8H), 3.57 (s, 8H), 2.94–2.84 (m, 16H). 1H NMR (400 MHz, CD3CN) δ: 9.13 (s, 2H), 7.57 (d, J = 8, 2H), 7.38 (d, J = 8, 2H), 7.17 (s, 2H), 7.12 (t, J = 8, 2H), 7.04 (t, J = 8, 2H), 3.59–3.55 (m, 16H), 2.91–2.79 (m, 16H). 13C NMR (100 MHz, CD3COCD3) δ: 136.70, 127.87, 122.59, 120.95, 118.37, 118.31, 113.52, 111.17, 70.59, 70.19, 56.71, 54.46, 23.39. ESI MS: Exact mass calculated for C32H44N4O4 [M + H]+: 549.34. Found: 549.42.</p><!><p>Stock solutions of the quencher to be analyzed (e.g. bis-indole-crown, dibenzyl-crown, etc.) were prepared in d3-acetonitrile. Zinc tetraphenylporphyrin (ZnTPP) was used as external photosensitizer for singlet oxygen production. For each quenching experiment, three cells (fluorescence grade quartz cuvette) were prepared with 2.00 mL of solvent containing a small amount of ZnTPP. Pulsed irradiation (2 ns) via a 400 V Nd:YAG laser (New Wave Research Mini-Lase II) was used at λex = 532 nm to generate 1O2. The band-pass filtered fluorescence emission signal of 1O2 was monitored by a Ge photodiode detector recording near IR emission in the range of 1200–1310 nm, with a maximum transmission at 1269 nm, the emission wavelength of 1O2. All measurements were taken of air saturated solutions. For each data point, 10.00 μL of a concentrated stock solution of the quencher was added to the cuvette cell such that the concentration of the quencher throughout the run was ~10−6-10−5 M. Each quenching experiment consisted of 6–10 data points of different quencher concentrations. The recorded luminescence signals from each run were processed on Origin Pro Software using equation I = Ioe−kt. The observed rate constant of the decay of 1O2 was recorded as a function of quencher concentration and plotted, yielding a linear graph. The slope of the curve gave kT, or the rate constant of total 1O2 quenching. All reported kT values are the average of three or more runs.</p><!><p>A 50 mL of photosensitizer stock solution was prepared using ZnTPP (0.0049 g, 7.23 μmol) in acetonitrile. The solution was diluted by a factor of 10. Bisindole-crown (1, 0.0042 g, 7.65 μmol) was added to 10 mL of the diluted ZnTPP stock solution. To 5 mL of the bisindole-crown solution NaBF4 (0.0029 g, 26.4 μmol) was added, and formation of 2 was confirmed via 1H NMR. Both solutions (1 and 2) were diluted so that their absorbance at 223 nm was ~2 (absorbance at 282 nm was ~0.6) and exposed to steady state irradiation with a 12-A Xenon lamp for 40 min. UV-vis scans were taken every 5 min using a Varian Cary 300 Bio Spectrophotometer to monitor the disappearance of the peaks at 223 and 282 nm.</p><!><p>The system developed by Gokel and coworkers consists of a two-armed lariat crown ether possessing identical two side-arms of a (3-indolyl)ethyl system attached to the nitrogen atoms of the crown ether (1). In the absence of a cation (K+ or Na+), the sidearm indole groups point away from the crown ether. However, addition of a cation leads to the cation-π complex 2 with occupation of the apical coordination sites by the indole groups (Fig. 2) (24).</p><p>This crown-tryptophan model system was synthesized by a modified literature procedure (for details see experimental section) (24). Proton NMR titrations showed that, as reported by Gokel and co-workers, upon addition of a slight molar excess of NaI in d3-acetonitrile relative to 1, the proton signals of the system shifted appreciably (Figure S1). Most notably, the proton on the second carbon of the pyrrole subunit shifted by −0.30 ppm, indicative of the binding of the cation, in agreement with the literature (Fig. S2). Further addition of the salt has no effect, indicating that all of the tryptophan is already bound to the cation. Since iodide anion is a strong physical singlet oxygen quencher (25), we decided to use NaBF4 instead of NaI as the cation source. The coordination of the sodium was not affected by the change of counterion, as was verified by following the 1H NMR shift of the protons on 1 during addition of NaBF4; the observed changes were identical for both NaI and NaBF4.</p><p>In preliminary qualitative experiments, we exposed an acetonitrile solution of crown-indole system 1 to steady-state irradiation (cut-off filter at 493 nm such that compound 1 is not irradiated) in the presence of the singlet oxygen sensitizer Methylene Blue. This let to rapid disappearance of the characteristic indole peaks in the UV spectrum (282 and 223 nm) concomitant with rapid photobleaching of the Methylene Blue (ca. 90 % after 10 min irradiation. The photobleaching is most likely due to a Type I photooxidation process, as discussed further below. Since the primary objective of this study is to determine as to whether or not there is a measurable effect of the cation-π interaction on Type II (singlet oxygen) photooxidation processes, we used zinc tetraphenylporphyrin, (ZnTPP) as a photosensitizer instead of Methylene Blue for further experiments.</p><p>We irradiated of a solution 1 ([1] = 3.0×10−5 M) in the presence of ([ZnTPP] = 1.45×105 M, solvent = CH3CN, cut-off filter at 493 nm) for 30 min, and subsequently irradiated under identical conditions a solution of 2 formed by addition of five equiv. of NaBF4 to the solution of 1. No photobleaching of the sensitizer was observed in either case. For the solution of the cation-complexed system 2, the characteristic peaks of the indole moiety in the UV/vis spectrum at 282 nm and 223 nm decreased considerably more slowly (about three-fold under these conditions) than without NaBF4 being present (Figure 3). This qualitative experiment indicates that the presence of the sodium cation does indeed have a notable effect on rate of photooxidation of the indole moiety.</p><p>Singlet oxygen can both be physically quenched by (kq) and/or chemically react (kr) with a substrate. Free tryptophan undergoes both of these processes with singlet oxygen. The sum of these two parameters is usually referred to as the kT value, which is a measure of the total singlet oxygen scavenging rate constant of a substrate. Since tertiary amines are known to be strong physical quenchers of singlet oxygen, one would expect both the tryptophan and the tertiary amine of the lariat crown to exhibit significant physical quenching, and the actual kT value would be the sum of these contributions. Indeed, in the absence of sodium cation, time-resolved singlet oxygen luminescence quenching measurements (in CD3CN with ZnTPP as photosensitizer) revealed a very large kT value of (2.4±0.2)×108 M−1sec−1 for the model system 1 (Fig. 4a). Upon addition of five equiv. of NaBF4 (formation of the cation-π interaction was again verified by 1H NMR), we obtained a much smaller kT value of (6.9±0.9)×107 M−1sec−1 (Fig. 4b) We did not observe any quenching of singlet oxygen by NaBF4 in a control experiment conducted under otherwise identical conditions. The value of kT could also be affected by changes in the ionic strength of the solution upon addition of NaBF4. However, a control experiment using the same flash-photolysis measurement of kT in the presence of 1 and five equiv. of N(CH3)4BF4 (the N(CH3)4+ cation is too large to complex to the crown ether) gave a value for kT identical to that of 1 without any cation present. Hence the change in the ionic strength of the solution has no effect on the value of kT for 1 and 2 under the conditions of our time-resolved measurements.</p><p>As mentioned above, the crown ether model system 1 contains two tertiary amines which are well known to quench singlet oxygen (26,27). To obtain an estimate of how much the kT value of 1 and its sodium complex 2 may be due to quenching by the tertiary amine of the lariat crown system, we investigated the singlet oxygen quenching by the commercially available dibenzyl crown ether, 7,16-dibenzyl-1,4,10,13-tetraoxa-7,16-diazacyclooctadecane, 3. Crown ether 3 has the same lariat crown system as our model system 1, but the two benzyl sidearms are too short to allow a cation-π interaction between the benzene rings and a sodium cation (28). In the absence of sodium cation, this system quenches singlet oxygen with a rate constant kT of (4.0±0.6)×107 M−1sec−1. In contrast, upon addition of five equiv. of sodium cation no singlet oxygen quenching was detected. The latter result is consistent with the charge-transfer mechanism by which tertiary amines quench singlet oxygen (26,27). The control experiments with crown ether 3 imply that the kT value for the sodium-complexed indole crown system 2 must be solely due to quenching of singlet oxygen by the indole moiety subjected to a cation-π interaction, and for the uncomplexed model system 1, more than 80 % of the quenching appears to be due to the indole system (cf. kT of 1 = (2.4±0.2)×108 M−1sec−1 while kT for 3 = (4.0±0.6)×107 M−1sec−1). The various kT values for compounds 1-3 are summarized in Table 1 below.</p><p>As mentioned earlier, the photosensitizer Methylene Blue is rapidly bleached upon irradiation in the presence of 1. For example, during a steady-state photooxidation of the model system 1 ([1] = 5.0×10−5 M, A(668 nm) = 0.4), we observed disappearance of 90 % of the peak at 668 nm after 10 min, concomitant with rapid disappearance of the UV peaks for the indole moiety of 1 (decrease of the indole peak at 282 nm in the UV spectrum by ca. 50 % within ten minutes which is much faster than upon irradiation with ZnTPP). However, no photobleaching of the Methylene Blue is observed over a period of two hours upon addition of five equiv. NaBF4. To further investigate this effect, we conducted singlet oxygen luminescence quenching experiments for compounds 1 and 2 in the presence of Methylene Blue. The kT values obtained for the cation-π complexed system 2 were indeed identical to those obtained with ZnTPP as a photosensitizer. In contrast, for the uncomplexed model system 1 nonlinear behavior was observed (Fig.4c). While there could be an aggregation between the methylene blue and the indole moiety (29), the rapid photobleaching of methylene blue during steady-state irradiation suggests that there is a Type I interaction between the triplet excited methylene blue and the indole-crown compound 1. We therefore hypothesize that the non-linear plot of kobsd vs. [1] (Figure 4c) is due to electron-transfer from 1 to the excited cationic Methylene Blue photosensitizer, i.e. a Type I photooxidation process which leads to generation of products that can act as additional singlet oxygen quenchers. It has previously been observed that for free base porphyrin sensitizers, Type I processes compete with singlet oxygen oxidation for tryptophan in micelles (30). A Type I process is also consistent with the rapid photobleaching of Methylene Blue upon irradiation in the presence of 1. Upon formation of the cation-indole complex 2, the Type I process is completely inhibited as we did not observe any photobleaching of Methylene Blue during the steady-state photooxidation of a sample of 2. The Type I photooxidation process of compound 1 could involve the tertiary amine from its lariat crown system or the indole moieties. However, a singlet oxygen luminescence control experiment with compound 3 (which has the same lariat crown system as 1) and methylene blue showed no curvature in the plot of kobsd vs. [3] (Fig. S3), and the kT value was identical within limits of error to that obtained when ZnTPP was used as a photosensitizer (Fig. S4). Hence the indole moiety is solely responsible for the rapid photobleaching of Methylene Blue, and the cation-π interaction is also able to inhibit this Type I photooxidation process of the indole moiety.</p><p>We also attempted to elucidate the rate of chemical reaction only (kr) for 1 and 2. This value can be determined by competition kinetics using singlet oxygen acceptors with known kr values. We used the well-known singlet oxygen co-acceptor, 1,5-dihydroxynaphthalene, as described in detail in a previous paper by our laboratory (31). This involves monitoring the disappearance of 1 at 277 nm (where the 1,5-dihydroxynaphthalene has an isosbestic point) and monitoring the increase of the oxidation product of 1,5-dihydroxynaphthalene at 450 nm where neither the indole nor its oxidation products absorb. However, despite many attempts we were unable to obtain sufficiently reproducible data to report exact values for kr. This appears to be due to the fact that for complex 1, the value of kr decreases with increasing oxidation of 1. Our model system 1 contains two indole moieties. It appears that the reactivity of the second indole moiety of our model system is affected (i.e. slowed down) by oxidation of the first indole group. Furthermore, for the complex 2, the value of kr appears to increase with increasing conversion of 2. There appears to be at least partial loss of the sodium cation (by 1H NMR) upon oxidation of one or both of the indole units. Nevertheless, during the competition experiments, the ratio of the kr values of 1 vs. 2 generally ranged from two to six at any given time, i.e. the chemical reactivity of the indole moiety is slowed down significantly upon complexation of a sodium cation. This range is also consistent with the qualitative experiments described earlier where loss of the indole moiety in the sodium-complexed system 2 was slower than for the uncomplexed system 1.</p><p>All of our singlet oxygen quenching measurements for compounds 1 – 3 were carried out in deuterated acetonitrile solution. Due to the partial charge-transfer character of the encounter complex between the singlet oxygen meolcule and the indole moiety in the quenching process, the kT values for methyl indoles and tryptophan are larger than in water as compared to aprotioc solvents (9,32). While it would be interesting to determine the kT values for 1 and 2 in water, this is unfortunately not possible for our system, as complex 1 is insoluble in water. It is probable that in solution the cation-π interaction is weaker than for a crystal structure. We therefore conducted density functional theory (DFT) optimizations on both the complexed and uncomplexed indole-crown structures and compared these optimized structures with their crystal structures (Figs. S5–S6). Stationary point calculations were conducted on the crystal structure geometries of 1 and 2 provided by Gokel and coworkers (24). DFT calculations were performed using the Orca program system (33). Three separate stationary point calculations were performed, all using the second generation triple-zeta valence polarization basis set, def2-TZVP which is recommended for main group elements (34). All calculations were solvent corrected using the conductor-like polarizable continuum model (CPCM) for acetonitrile (35). Grimme's meta hybrid M06 functional and was used in addition to the popular hybrid Becke 3-parameter functionals, Lee-Yang-Parr (B3LYP) and Perdew-Wang (B3PW91). Note that while B3PW91 has been reported to be more accurate for modeling noncovalently bounded complexes (36), all three functionals chosen (M06, B3PW91, and B3LYP) have been applied for modeling the cation-π interaction before by Ramos and coworkers, demonstrating errors within chemical accuracy (errors of 0.2 kcal, 0.1 kcal, and 0.8 kcal, respectively) for a benzene-Na+ system (37). Additionally, recent work by Mahanta et al. effectively illustrated the usage of B3LYP and M06 on a cofacial molecular dyad, 4,5-biphenyl acridine, which contains an aromatic heteroatom participating in cation-π interaction with secondary phenyl groups (38). Based on this literature precedent we felt confident choosing these methods for our computational analysis. All calculated values are available in the Supporting Information.</p><p>Since B3PW91 has shown the most accuracy of these for modeling the cation-π interaction (36,37), we proceeded to optimize the geometries of 1 and 2 using B3PW91/def2-TZVP—CPCM(Acetonitrile). While the optimized structure of 1 shows little significant deviation upon solvent correction, there is a noticeable increase in the distance between the sodium cation of 2 and the indole π-system in acetonitrile solution (Figure S6): Upon solvation, the cation-π interaction distance exhibits an average increase of 0.561 Å. An increase in the distance of the two moieties is likely the result of a weakened cation-π interaction in the model complex when in solution. Thus, one would expect that for more strongly held cation-π interactions the decrease in reactivity with 1O2 could be more significant than for the model system examined in this work. Furthermore, it may also be the case that in solution, multiple species exist in equilibrium, i.e. uncomplexed, mono- and bis-indole complexed. To the extent that this may be the case, this study has established a lower limit to the possible protective effects of cation-π interactions.</p><p>One interesting result from the kT values for the bis-indole crown model system 1 and its sodium complex 2 as well as the control benzyl crown ether 3 and its sodium complex appears to be that while the quenching of the indole moieties is considerably reduced in 2 relative to 1, the quenching of 1O2 by the tertiary amines is completely inhibited upon complexation of a sodium cation.</p><p>Our density functional theory (DFT) calculations of the frontier orbital energies and locations for 1 and 2 confirm this observation. We found that across all three functionals, the HOMO of the uncomplexed indole-crown system 1 was delocalized through the tertiary amine and side arms onto the indole moieties. However, upon complexation with sodium forming 2, the electron density of the HOMO was shifted from the tertiary amine (Fig. 5). Since singlet oxygen reacts as an electrophile, theory predicts that the tertiary amine is no longer able to interact with 1O2 from the HOMO of the complexed molecule. Thus, our computational results are in agreement with the observation that binding a cation electronically inhibits the reactivity of the tertiary amine towards singlet oxygen.</p><!><p>In summary, we have been able to show for the first time that the cation-π interaction can exhibit a photoprotective effect by using a crown-indole model system. There is a significant reduction of the total rate of singlet oxygen removal by the indole system upon complexation of a sodium cation. We would like to point out that the cation-π interaction is an electronic effect that leads to stereochemical consequences. As such, the measurements reported herein should be interpreted in the same way: Coordination of a sodium cation to 1 inevitably leads to electronic and stereochemical changes of the bis-indole crown system both of which may be involved in the reduced kT values. We have also shown that the cation-π interaction can suppress a Type I photooxidation involving electron transfer from an indole moiety to triplet excited methylene blue. Finally, we have found that there can be complete inhibition of physical quenching of singlet oxygen by a tertiary amine upon coordination of a cation in a crown ether. In biological systems, the most common type of cation-π interactions are those involving tryptophan residues. It seems likely that these residues are less reactive with singlet oxygen than other exposed tryptophan residues that are not subject to the cation-π interaction.</p><!><p>Figure S1. 1H NMR spectra in d3-acetonitrile of both the indole-crown model complex (1, top) and the complexed indole-crown-Na+ system (2, bottom) are shown. Note the shift of the peaks due to complexation. The peak at 3.1 ppm is the result of MeOH solvent</p><p>Figure S2. Shift of the protons in a CD3CN solution of bis-indole-crown (1) upon addition of NaBF4, leading to formation of 2. The concentration of the indole crown 1 was 4.6 × 10−6 M. Each NaBF4 addition corresponds to 1 equivalent of sodium cation. Based on the plot, full complexation is seen after ~2 equivalents of sodium cation have been added.</p><p>Figure S3. Singlet oxygen quenching by control compound 3 (dibenzyl crown ether, 7,16-dibenzyl-1,4,10,13-tetraoxa-7,16-diazacyclooctadecane) using photosensitizer methylene blue in d3-acetonitrile. The slope is the kT value for compound 3.</p><p>Figure S4. Singlet oxygen quenching by 3 using photosensitizer ZnTPP in d3-acetonitrile. The slope is the kT value for compound 3.</p><p>Figure S5. Overlapped structures showing the comparison between crystallographic coordinates reported by Gokel et al. (with grid lines) and the B3PW91/def2-TZVP—CPCM(Acetonitrile) optimized geometry (no grids). Top: Indole-crown, 1, and bottom: Indole-crown-Na+, 2. Optimized frontier orbital energies of 1: EHOMO/LUMO = 5.02 eV; 2: EHOMO/LUMO = 5.08 eV.</p><p>Figure S6. Left: The crystal structure of 2 is shown. Distance from the sodium cation to center of five-membered ring is 3.496 Å. Right: The B3PW91/def2-TZVP—CPCM(Acetonitrile) optimized geometry of 2 is shown. Average distance from the sodium cation to center of five-membered ring is 4.057 Å. Upon solvation, the cation-π interaction distance exhibits an average increase of 0.561 Å.</p>

PubMed Author Manuscript

Safety, pharmacokinetics, and pharmacodynamic properties of oral DEBIO1143 (AT-406) in patients with advanced cancer: results of a first-in-man study

PurposeTo assess safety/tolerability, pharmacokinetics (PK), pharmacodynamics (PD), and antitumor activity of DEBIO1143, an antagonist of inhibitor apoptosis proteins.MethodsThis first-in-man study in patients with advanced cancer used an accelerated dose titration design. DEBIO1143 was given orally once daily on days 1–5 every 2 or 3 weeks until disease progressed or patients dropped out. The starting dose of 5 mg was escalated by 100 % in single patients until related grade 2 toxicity occurred. This triggered expansion to cohorts of three and subsequently six patients and reduction in dose increments to 50 %. Maximum tolerated dose (MTD) was exceeded when any two patients within the same cohort experienced dose-limiting toxicity (DLT). On days 1 and 5, PK and PD samples were taken.ResultsThirty-one patients received doses from 5 to 900 mg. Only one DLT was reported at 180 mg. No MTD was found. Most common adverse drug reactions were fatigue (26 %), nausea (23 %), and vomiting (13 %). Average t max and T 1/2 was about 1 and 6 h, respectively. Exposure increased proportionally with doses from 80 to 900 mg, without accumulation over 5 days. Plasma CCL2 increased at 3–6 h postdose and epithelial apoptosis marker M30 on day 5; cIAP-1 levels in PBMCs decreased at all doses >80 mg. Five patients (17 %) had stable disease as the best treatment response.ConclusionDEBIO1143 was well tolerated at doses up to 900 mg and elicited PD effects at doses greater 80 mg. Limited antitumor activity may suggest development rather as adjunct treatment.Electronic supplementary materialThe online version of this article (doi:10.1007/s00280-015-2709-8) contains supplementary material, which is available to authorized users.

safety,_pharmacokinetics,_and_pharmacodynamic_properties_of_oral_debio1143_(at-406)_in_patients_with

Introduction<!>Design<!>Patient population<!>Treatment<!>Safety<!>Pharmacokinetics<!>Pharmacodynamics<!>Efficacy<!>PK and PD analyses<!>Data analysis<!><!>Results<!><!>Safety<!><!>cIAP1 levels in tissues and PBMCs<!><!>Antitumor activity<!>Discussion<!>

<p>Inhibitors of apoptosis proteins (IAPs) may play a role in the development of cancer [1, 2]. Their over-expression has been linked not only to tumor growth and poor prognosis, but also to low treatment response or resistance [2, 4]. Therefore, the IAP protein family is generally considered a promising target for cancer drug development [2, 5, 6]. So far, six IAP antagonists have entered clinical development [3]. One of these is DEBIO1143 (formerly AT-406, SM-406), a small molecule mimetic of second mitochondria-derived activator of caspase (SMAC) [7].</p><p>In vitro studies have demonstrated DEBIO1143 to inhibit cell growth in various human cancer cell lines [2, 4] through binding of X-chromosome-linked IAP (XIAP) and cellular IAPs 1 and 2 (cIAP-1 and -2). DEBIO1143 rapidly induced degradation of cIAP-1 in a cell-free functional assay [7] and apoptosis in xenograft tumors. Moreover, it was able to enhance the antitumoral effects of irradiation or various chemotherapeutic agents in multiple mouse cancer models [1, 4, 8]. Preclinical data further revealed good oral bioavailability in mice, rats, dogs, and non-human primates, enabling PK/PD modeling to predict tumor and plasma concentrations in humans [1].</p><p>Multiple high doses (40–120 mg/kg/day) induced hepatotoxicity in rats. In dogs, liver cell degeneration was seen at 3 and 10 mg/kg/day. In 4-week toxicology studies, the severely toxic dose (STD) in the rat was determined at 40 mg/kg and the highest non-severely toxic dose (HNSTD) in a non-rodent species at 1 mg/kg in dogs. Based on metabolism data and observed adverse events (AEs), the dog was considered the most relevant species, in line with reports on other IAP inhibitors [9]. The no observable adverse event level (NOAEL) of 1 mg/kg in dogs led to a calculated starting dose of 5 mg in humans. An intermittent dosing schedule was chosen to further mitigate the risk of unacceptable toxicity when entering clinical development.</p><p>The primary objective of this first-in-man study was to characterize the safety and determine the maximum tolerated dose (MTD) and schedule of DEBIO1143 when administered to patients with advanced solid tumors and lymphomas. Secondary objectives were to explore (a) PK of DEBIO1143, (b) any PD effects, (c) any observable antitumor activity during the trial, and (d) its correlation with PK.</p><!><p>This was a multicenter, uncontrolled, open-label, dose-escalation study on DEBIO1143 in patients with advanced cancer. It employed an accelerated titration design for dose escalation with 100 % dose increments in consecutively enrolled single patients until drug-related grade-2 toxicity was observed during the initial treatment cycle (until day 28 or day 21 as per protocol amendment). If this was the case, cohort size was expanded to three patients and dosing increment was reduced to 50 % of the last dose. Dose escalation was to be stopped at the MTD which was considered exceeded if at any dose level more than one patient experienced dose-limiting toxicity (DLT) during the first treatment cycle. DLT was defined as any of the following: (a) non-hematological toxicity of grade ≥3 (excluding nausea, vomiting, diarrhea unless not controlled by maximal antiemetic/diarrheal therapy for >24 h); (b) anemia or neutropenia of grade ≥3 or thrombocytopenia of grade 4 or any grade if associated with clinically significant bleeding; (c) any AE resulting in dose delay or reduction; (d) any toxicity considered dose-limiting by the investigator. If only one out of the three patients of a cohort experienced drug-related DLT, the cohort was expanded by another three patients to be treated at the same dose level. If none of these additional patients experienced DLT, the dose escalation by 50 %, rounded down to the nearest capsule strength combination, continued in the next cohort of three patients.</p><p>Pharmacokinetic samples were taken from all patients on day 1 (predose, 0.5, 1, 2, 3, 4, 6, 8, 12, 18 h postdose) and on day 5 (predose, 0.5, 1, 2, 3, 4, 6, 8 h postdose) of the first cycle. In addition, for exploratory PD analysis of IAP inhibition and activation of apoptosis, optional skin and tumor biopsies were taken from consenting patients on days 1 (predose) and 5 and blood samples on day 1 (predose, 1, 3, 6, 8, 12 h postdose), day 2 (predose), and day 5 (3 h postdose).</p><p>The study was compliant with all applicable legal obligations, the requirements of the Declaration of Helsinki and Good Clinical Practice. It was approved by the institutional review boards of the three participating sites and registered under Clinicaltrials.gov (identifier: NCT01078649).</p><!><p>Eligible were male and female adult outpatients with histologically confirmed advanced or metastatic solid tumors or lymphoma for which no life prolonging or appropriate standard therapy was available. Patients had to be ambulatory (Eastern Cooperative Oncology Group (ECOG) performance status ≤1) with adequate hematological (ANC ≥1,500/mm3; hemoglobin >9.0 g/dL; platelet count ≥100,000/mm3), renal (creatinine ≤1.0 × upper limit of normal (ULN) or creatinine clearance of >60 ml/min), hepatic (serum albumin ≥3.0 gm/dL; total bilirubin <1.0 × ULN; aminotransferases and alkaline phosphatase ≤2.5 × ULN, including negative hepatitis testing), and cardiac function without evidence of QTc prolongation.</p><p>As clinically significant bleeding formed part of the definition of DLT, further exclusion criteria were a history of gastrointestinal bleeding during the preceding year, of treatment-requiring diabetes mellitus, or any condition associated with chronic inflammation (e.g., rheumatoid arthritis, inflammatory bowel disease, chronic infections) or affecting copper accumulation or regulation (e.g., Wilson's disease).</p><p>Last radiation and intake of steroids had to date back at least 14 days from study entry (thoracic radiation 28 days); patients had to be clinically stable and to have recovered to toxicity grade ≤1 from any prior cancer therapy. Patients had to have never received IAP inhibitors before. All patients had to give written informed consent to be enrolled in the trial.</p><!><p>Oral treatment had to be taken daily on days 1–5, initially every 14 days, later every 21 days as per protocol amendment. The amendment was put in place to be more conducive to future combination with common chemotherapy regimens and to reduce the potential for adverse drug reactions (ADRs) through an additional week of recovery between doses. The starting dose of 5 mg was increased in subsequent cohorts, based on DLT observed by the end of cycle 1. Stable or responding patients who experienced DLT were allowed to continue therapy at the next lower dose, once those had resolved to grade ≤1 within 2 weeks. End of treatment was triggered by disease progression, unacceptable toxicity, or withdrawal from the study for any reason. Cancer therapy other than DEBIO1143 was not allowed, but supportive care measures were. Concomitant treatment with aspirin at doses >81 mg/day or with any anticoagulants was prohibited.</p><!><p>The incidence of AE, ADR, and DLT was recorded at all scheduled visits (on days 1, 15, and 28 of each cycle and additionally on days 5, 8, and 22 of cycle 1) and graded according to the Common Terminology Criteria for Adverse Events, version 4.0 of the National Cancer Institute. Moreover, safety laboratory, 12-lead ECG, vital sign measurements, and physical examinations were performed.</p><!><p>C max and t max were determined by direct assessment of the observed concentration versus time curves. The area under the curve until the last quantifiable concentration (AUC0−t) was estimated by a linear up/log down method if ≥3 values were available and extrapolated to infinity (AUCinf), if the extrapolated part was <30 %. The terminal elimination half-life (T 1/2) was calculated as the ratio of loge2 to the apparent terminal phase rate constant (λ z), determined through unweighted linear regression analysis on ≥3 log-transformed concentrations on the linear portion of the terminal slope, excluding the peak concentration. In general, points maximizing R 2 up to at least 0.9 were included for linear regression.</p><!><p>cIAP-1 levels were measured in tumor tissue and surrogate tissue as available. Peripheral blood mononuclear cells (PBMCs) were analyzed for DEBIO1143-induced cIAP-1 degradation. Plasma native cytokeratin-18 (M65) or caspase-3 generated cytokeratin-18 fragments (M30), interleukin 8 (IL8), chemokine ligand 2 (CCL2, MCP1), and tumor necrosis factor α (TNFα) were measured by ELISA as markers of epithelial cell death and inflammation on days 1 (predose, 1, 3, 6, and 12 h postdose), 2, and 5.</p><!><p>Tumor evaluations were scheduled before therapy and after every other cycle of therapy. Changes were determined by physical examination, tumor markers, or standard imaging techniques. Response to DEBIO1143 was assessed for solid tumors based on RECIST guidelines, version 1.1 [10] and for lymphoma as per the Revised Response Criteria for Malignant Lymphoma [11]. Read-outs were complete response (CR), partial response (PR), stable disease (SD), and disease progression (DP).</p><!><p>DEBIO1143 plasma concentrations were measured using a validated LC–MS/MS assay. For the detection of cIAP-1 in paraffin-embedded human tissue, a validated immunohistochemical (IHC) assay was used. DEBIO1143-induced cIAP-1 degradation was measured in PBMCs using Western blot. Plasma biomarkers were measured through commercial ELISAs. All laboratory analyses were performed by MPI Research, Inc. (PK) and Mosaic Laboratory, LLC (PD).</p><!><p>Descriptive statistics were presented as appropriate. Safety laboratory assessments, vital signs, body weight, ECOG, ECG, as well as PD and tumor response data were compared over time to assess change from baseline during treatment and follow-up. In case of sufficient sample size, Wilcoxon matched pairs test was used for inferential comparisons. Data were analyzed using the Statistical Analysis System (SAS), version 9.1.3. Individual PK parameters were derived using WinNonlin (version 5.3, Pharsight Corp., Mountain View, CA).</p><!><p>Patient flowchart according to CONSORT. DP disease progression (second line: duration of stable disease); UPR upon patient request</p><!><p>Patients were treated with DEBIO1143 for up to 117 days and all patients completed at least one cycle; 2 cycles: 27 (87.1 %) patients; 3 and 4 cycles: 5 (16.1 %) patients each; 5 and 6 cycles: 2 (6.5 %) patients each; 7 and 8 cycles: one patient each (3.2 %) (Fig. 1; median 2 cycles). A drug-related grade 2 fatigue in a patient treated with 80 mg prompted expansion to 3-patient cohorts. Subsequently, a grade 3 reversible ALT elevation in a patient receiving 180 mg was the only reported DLT which resulted in the expansion of this cohort to six patients. Dose was escalated to 900 mg daily before enrollment was halted due to the excessive number of pills to be taken. Thus, the MTD was not reached.</p><!><p>Number of patients with ADRs and ADR frequency by system organ class</p><!><p>No clinically meaningful trends were seen in measurements of safety laboratory, ECG, vital signs, body weight, or ECOG performance status. There were no dose reductions, delays, or modifications due to AEs or lack of tolerability.</p><!><p>Dose proportionality of C max and AUCinf</p><p>Pharmacokinetics of DEBIO1143, means (standard deviation)</p><p>ND not determined</p><p>** Median (minimum–maximum)</p><p>Expression of cIAP. a in skin biopsies of 12 patients (H-scores; on the top). b in PBMC (quantitative Western blot results as % from baseline) across doses (on the bottom; for results per dose see Suppl. 2)</p><!><p>The expression of cIAP1 was evaluable in PBMCs from 28 patients with doses above 80 mg using Western blot (Fig. 3b; Suppl. 2). In 20 patients, cIAP1 was readily detectable at baseline but undetectable or extremely low in eight patients. In all patients with detectable cIAP1, DEBIO1143 led to rapid and persistent cIAP1 degradation regardless of dose.</p><!><p>Pharmacodynamic measurements</p><!><p>No patient had a complete or partial response. One patient with metastatic melanoma with latero-cervical lymph node involvement showed an 11 % reduction in target lesion dimensions at 400 mg/day. Progression in the same nodes was noted after six cycles. Stable disease as best response was seen in five patients (16.1 %) for a median duration of 93 days (range 85–197 days; Fig. 1). All these patients had different cancer types (Hurthle cell, melanoma, breast, rectal, hemangiopericytoma).</p><!><p>Several SMAC mimetic IAP antagonists have entered clinical development, including DEBIO1143 (formerly AT-406), HGS1029 (formerly AEG-40826), GDC-0917 and -0152, LCL-161, and birinapant (TL-32711). The latter two compounds have entered phase II trials. Our results on DEBIO1143 add to the existing body of evidence from these clinical trials on SMAC mimetic IAP antagonists. In general, tolerability and safety of this class of drugs have been acceptable [12–14]. These trials have not revealed any consistent AE in humans [3, 12–14] except an increased incidence of Bell's Palsy syndrome at higher doses of birinapant which can be prevented by dose titration in the initial treatment cycle [15]. In this regard, it is noteworthy that the cerebrovascular accident and cranial nerve disorder in our study were both unlikely related to DEBIO1143. The former occurred 23 days after last intake, and signs of thrombocytopenia, coagulopathy, or hypertension were absent during the two cycles of treatment as well as at the time of the event. By contrast, an occult metastasis could not be ruled out. The cranial nerve disorder was not Bell's Palsy syndrome as it affected cranial nerve 5 rather than 7; it was most consistent with progression of leptomeningeal disease. In general, there was no grade 4–5 treatment-related AE at all, even during treatment for over 6 months. The only case of transient G3 hepatotoxicity is consistent with preclinical data and linked to the mechanism of action of SMAC mimetic IAP inhibitors [16, 17].</p><p>Our PK data in humans confirmed drug exposures at or above those needed for activity in preclinical models [7]. PK disposition of DEBIO1143 is suitable for once daily dosing over 5 consecutive days every 3 weeks. DEBIO1143 showed proportional plasma increases at doses >80 mg in our study (Fig. 2). However, beyond this threshold, neither PD nor antitumor activity showed any dose relationship, but all doses resulted in the degradation of cIAP-1 in PBMCs and at least in a trend for a decrease in cIAP levels in surrogate skin tissue. The on-target activity of DEBIO1143 was also supported by the significant increase in CCL2 plasma levels (Fig. 4), which might be a consequence of the cIAP-1 degradation through modulation of NF-κB. CCL2, actually a marker of inflammation, has also been associated with the stimulation of a host antitumor response [18] which would be in line with the observed M30 increase indicating drug-induced epithelial apoptosis [19].</p><p>However, interpretation of PD data remains limited due to the small samples sizes of dose groups and the lacking dose–response relationship. Our data also remain inconclusive regarding a recommendable dose, although 900 mg/day resulted in acceptable tolerability and in exposures with proven activity in preclinical experiments. It may thus serve at least as a starting dose for phase II. In addition, the modest clinical activity of IAP inhibitors shown in unselected refractory cancer patients so far suggests the need for combination approaches and screening for more sensitive subpopulations.</p><!><p>Supplementary material 1 (DOC 195 kb)</p><p>ClinicalTrials.gov identifier: NCT01078649.</p>

PubMed Open Access

Asymmetric light reflectance from metal nanoparticle arrays on dielectric surfaces

Asymmetric light reflectance associated with localized surface plasmons excited in metal nanoparticles on a quartz substrate is observed and analyzed. This phenomenon is explained by the superposition of two waves, the wave reflected by the air/quartz interface and that reflected by the metal nanoparticles, and the resulting interference effects. Far field behavior investigation suggests that zero reflection can be achieved by optimizing the density of metal nanoparticles. Near field behavior investigation suggests that the coupling efficiency of localized surface plasmon can be additionally enhanced by separating the metal NPs from substrates using a thin film with refractive index smaller than the substrate. The latter behavior is confirmed via surface-enhanced Raman spectroscopy studies using metal nanoparticles on Si/SiO2 substrates.Localized surface plasmons (LSPs), the charge density oscillations confined in metallic nanostructures, have attracted tremendous interest for a broad range of emerging applications, such as chemical and biomolecular sensing 1-4 , subwavelength optical imaging 5,6 , optoelectronic devices such as light emitting diodes (LEDs) 7-9 , solar cells 10-14 and photodetectors [15][16][17][18] . LSPs excited by an electric field at a particular incident wavelength where resonance occurs will lead to strong light scattering, intense absorption band and enhancement of the local electromagnetic fields 19 , The oscillation frequency and intensity i. e. coupling efficiency of LSPs can be modulated by the type of metals 20 . They are also highly sensitive to the size, size distribution, shape and the medium which surround/near the metallic nanostructure 21,22 .In this work, we characterize and analyse asymmetric light reflectance observed for metal nanoparticles (NPs) fabricated on quartz substrate. Far field reflection spectra show different behaviours at wavelengths close to the LSP resonance wavelength when light is normally incident from air compared to that when light is incident through the quartz substrate. This phenomenon can be explained using a modified Fresnel coefficients model [23][24][25] . Specifically, the far field reflected wave can be regarded as a superposition of the wave reflected by the air/quartz interface and that reflected by the LSPs. The reflection-phase shift of LSPs is π at the LSP resonance wavelength, so the superposition leads to either constructive or destructive interference when light is incident on the air/NPs/ quartz interface from air or quartz, respectively. Theoretical analysis combined with FDTD simulations show that the reflection intensity at the LSP resonance wavelength can be reduced close to zero by varying the density of Au NPs. This behaviour can be used to enhance the sensitivity of LSP based sensors.Theoretical analysis and FDTD simulation also indicate that when light is incident from quartz, the extinction peak intensity of the LSPs at the wavelength close to LSP resonance wavelength is larger than that when light is incident from air. This phenomenon can be attributed to the different local driving field intensities of LSPs when light is incident from different media. The ratio of the extinction peak intensities when light is incident from different media is equal to the ratio of the refractive indices of the two media. However, for many LSPs applications, light must be incident from air. Theoretical analysis and Surface-enhanced Raman scattering (SERS) measurements demonstrate that when metal NPs are separated from a substrate by a thin film with refractive index lower than the substrate, the local driving field intensity can be adjusted. The local driving field intensity can therefore be optimized

asymmetric_light_reflectance_from_metal_nanoparticle_arrays_on_dielectric_surfaces

<!>Results<!>(c). When<!>(d). However, regardless of the value of<!>Discussion<!>Methods<!>Measurements and Simulations.

<p>by including a thin film with optimized thickness. This behaviour provides a general method to enhance the LSPs coupling efficiency that may improve the performance of the LSPs based devices for a variety of applications.</p><!><p>Far field behaviour. Figure 1(a-d) present the far field transmittance and reflectance spectra, respectively, of Au and Ag NP arrays on a quartz substrate. One can see from Fig. 1(a) that when light is incident normally from air to the air/Au NPs/quartz interface (designated as front incident), the transmission spectrum shows a valley at approximately 525 nm corresponding to the LSP resonance mode of the Au NPs. When light is incident normally from the quartz substrate (designated as back incident), the transmission spectrum is almost the same as that of front incident, agreeing with the principle of reciprocity. Similarly, a dip in transmittance near the LSP resonance wavelength and nearly identical transmission spectra for front and back incidence are observed for Ag nanoparticles, as seen in Fig. 1(b). However, the reflectance spectra when light is incident from different directions differ substantially, as shown in Fig. 1c for the structure with Au nanoparticles on quartz. At the LSP resonance wavelength, the reflectance spectrum measured from the front side shows a peak while the spectrum measured from the back side shows a valley. For the sample consisting of Ag NPs on quartz substrate, peaks are seen in the reflectance spectra when light is incident from both sides. However, when light is incident from back side, the peak intensity is smaller than that when light is incident from front side (Fig. 1d).</p><p>The behavior seen in Fig. 1 can be explained using a model based on modified Fresnel coefficients 23 . This model take into account excess current and charge densities present due to the discrete subwavelength metal NPs at the interface. For a plane wave propagating through a medium with refractive index n i toward an interface consisting of metal NPs against a medium with refractive index n t , the reflection coefficient r for a normally incident wave can be written as</p><p>Here, α ω</p><p>2 , where V is the volume of the NPs, ω LSPR is the LSP resonance frequency, and L is a geometrical depolarization factor calculated by the image dipole theory 23,26 . For a hemisphere structure, one can obtain ≈ .</p><p>L 0 1. γ ≈ . × / rad s 2 5 10 0 13</p><p>are the width of resonance determined by the resistive Drude damping factor 25 and the factor</p><p>3 describes the radiative damping contribution that arises due to the finite size of the particle. Thus, one can see that when the frequency of the incident wave is equal to the LSP resonance frequency, i.e. ω ω = LSPR , the additional term arising from the localized surface plasmon,</p><p>, is a positive real number. Equation (1) can then be written as , where r back is the reflection coefficient when light is normally incident from the back side. One can see that the reflectance spectrum does not always show valley when light is incident from the medium with high refractive index. When</p><p>, we see that < r r back 0 , and the reflectance spectrum will exhibit a valley at wavelengths close to the localized surface plasmon resonance wavelength as shown in Fig. 1</p><!><p>back 0 , and the reflectance spectrum will contain a peak as shown in Fig. 1</p><!><p>Thus when light is incident from the back side, the reflectance at wavelengths close to the localized surface plasmon resonance wavelength is smaller than that when light is incident from the front side.</p><p>Figure 2 illustrates the mechanism leading to the asymmetric light reflectance phenomenon described above. For a plane wave propagating through a medium with refractive index n i toward an interface consisting of metal NPs against a medium with refractive index n t , the reflected wave can be regarded as superposition of two waves: the reflected wave from the interface without metal NPs on it and the reflected wave arising from the metal NPs. Thus the reflection coefficient arising from the metal NPs can be written as</p><p>From Equation (3) one can see that r LSPR is a negative real number which indicates that a-phase shift of π is introduced when light is reflected by metal NPs. When light is incident from the medium with lower refractive index, the superposition of the two reflected waves leads to constructive interference. Thus the reflectance spectrum always shows a peak at wavelengths close to the localized surface plasmon resonance wavelength. When light is incident from the medium with higher refractive index, however, the superposition of the two reflected waves leads to destructive interference. Thus, when A LSPR is not too large, the reflectance spectrum exhibits a valley rather than a peak at wavelengths close to the localized surface plasmon resonance wavelength (Fig. 2a). One can see that, to a certain degree, this phenomenon is similar to the asymmetric light reflectance effect we have reported previously in AAO on glass 27 . However, the phenomenon observed here still has some major differences from previous report, which are described as follows. (i) This phenomenon is wavelength selective and can be regulated by the resonance wavelength of the LSPs. (ii) The two superposing waves come from metal NPs and dielectric interface which attach to each other rather than two separated interfaces. (iii) Metal nanostructure is not necessarily to be embedded into an optical film. Thus it is more suitable for further near field applications.</p><p>One can see both the volume and density of metal NPs will affect the value of A LSPR and therefore the reflectance intensity. If the diameter of the metal NPs is relatively small, peaks will not be present in the reflectance spectra when light is incident from back side even with relatively high NPs density. Figure 2b shows the simulated reflectance spectra of Au NPs (10 nm in diameter) with density as high as 2.9 × 10 3 μ m −2 , the reflectance spectra show peak and valley when light is incident from front and back sides respectively. When the diameter of metal NPs is relatively large, the far field behavior can be tuned by varying the densities of the NPs. In particular, when = − A n n LSPR i t , the reflection coefficient is zero at the the LSP resonance wavelength. Since A LSPR is a positive real number, zero reflectance can happen only when light is incident from the medium with high refractive index. Figure 2c-d show the simulated reflection spectra with different Au NPs densities. One can see that when the density of Au NPs (60 nm in diameter) equals to 80.2 μ m −2 ; the reflection spectra show peaks regardless the incident directions (Fig. 2c). For the large Au NPs with relatively low densities, the reflectance spectra show valleys when light is incident from the back side (Fig. 2d). One can see that when the density of Au NPs increases, the valley intensity decreases close to zero first and then increases. When the density of Au NPs equals to 28.9 μ m −2 , the reflection intensity at the LSP resonance wavelength can be as low as 0.15%.</p><p>Near field behavior. Equation (3) also indicates that when light is incident from media with different refractive indices, the reflection coefficients associated with the metal NPs are not the same. The reflection by metal NPs can be regarded as arising from a portion of the light scattered by the metal NPs. Fig. 3b shows the simulated extinction spectra of a single hemisphere Au NP (60 nm in diameter) on quartz when light is incident from front and back sides. Extinction peaks at approximately 600 nm are present in both spectra regardless of the incident direction of the light. However, the peak intensities are different. When light is incident from the front side, the extinction peak intensity is smaller than that when light is incident from the back side. At the LSP resonance wavelength, the extinction peak intensity when light is incident from the back side is approximately 1.5 times that when light is incident from the front side.</p><p>This behavior can be explained by the different local driving field intensities E d at the position of the Au sphere when light is incident from different directions. When light is incident to the air/NPs/substrate, the local driving field intensities can be regarded as superposition of field intensities of incident wave and reflecting wave as shown in Fig. 3a. Thus we obtain</p><p>, where E i is the electric field intensity of incident light. Then we can obtain</p><p>, where E dF and E dB are the local driving field intensities of LSPs when light is incident from front and back side respectively, n 1 and n 2 are the refractive indices of the materials above and beneath the Au sphere respectively. Despite the different LSP resonance wavelengths caused by the different effective refractive indices beneath the NPs, one can see from Fig. 3b-d 28,29 . In this way, the efficiencies of the LSPs enhanced solar cells and the intensity of the surface-enhanced Raman scattering (SERS) signal have been significantly enhanced. However, if light is not generated from the devices, the energy loss by reflection at the interface is unavoidable, for example, photodetectors, solar cells, and most TERS, SERS setups. When light is incident from air and the metal NPs are located at the back side of a thick non-absorbing dielectric film, as shown in Fig. 4a, the local driving field intensity should be</p><p>, where n 1 and n 2 are the refractive indices of air and substrate respectively. Although this local driving field intensity of LSPs is larger than that when light is incident from front side, which equals to</p><p>, it is still not an optimized structure. Considering the structure shown in Fig. 4b, when metal NPs are separated from a dielectric medium with high refractive index n 3 by a thin film with low refractive index n 2 , the local driving field intensity of LSPs can be written as</p><p>, where h is the thickness of the thin film, and λ is the wavelength of the incident wave. One can see that when</p><p>1 , the local driving field intensity of LSPs shown in Fig. 4b will be higher than that shown in Fig. 4a. Figure 4c shows the local driving field intensity of LSPs as a function of h, where A LSPR , n 2 and n 3 are set as 0.6, 1.5 and 3.5 respectively, λ equals to 532 nm. One can see that when = ,( = , , , .....)</p><p>, the local driving field intensity of LSPs is maximized. Figure 4d presents SERS results of R6G molecules using Au NPs/SiO 2 /Si as substrates. One can see that when the thickness of SiO 2 equals to 90 nm and 270 nm, the SERS signals are much larger than when the thickness of SiO 2 equals to 0 nm and 180 nm.</p><!><p>In summary, we observed the asymmetric light reflectance phenomenon in metallic NPs fabricated on quartz substrate. The difference of the reflectivity when light is incident from different directions can be attributed to the superposition of waves reflected from metallic NPs and from the dielectric medium interface. A modified Fresnel coefficient model indicates that the phase shift of the wave reflected from metal NPs should be π . Thus the superposition between the reflected waves from metallic NPs and dielectric medium interface creates either constructive or destructive interference when light is incident from media with lower or higher refractive indices, respectively. Theoretical analysis and FDTD simulation suggest that this behavior can achieve zero reflectance via adjusting the density of metal NPs that can enhance the sensitivity of LSP sensors. Near field FDTD simulation shows that the ratio of the extinction peak intensities when light is incident from different directions equals the ratio of the refractive indices of two mediums beside the interface, implying that when light is incident from the medium with higher refractive index, metallic nanostructures would have higher coupling efficiency with the incident light. This behavior can be attributed to the different local driving field according to the Fresnel equation. Further investigating shows that the LSPs coupling efficiency when light is incident from air can be regulated by separating the metallic NPs from substrate using a low refractive index thin film. The highest LSPs coupling efficiency is achieved when the thickness of the thin film equals to , ( = , , , .....)</p><p>. This work provides a general method to opti-</p><!><p>Sample Fabrication. Au and Ag NPs were fabricated on quartz wafers of 0.5 mm thickness. The quartz wafers were ultrasonically degreased in acetone, ethanol and then double deionized water for 3min each. Au and Ag films with thickness of approximately 2 nm were sputtered by using SCD005 (Balzers Union, Balzers, Liechtenstein). The sample was then annealed in N 2 ambient by using the RTA device at 450 °C for 60 s to form Au and Ag NPs. The average sizes of fabricated Au and Ag NPs are 28 and 20 nm respectively. The densities of Au and Ag NPs are 8.5 × 10 10 and 4.5 × 10 10 cm -2 respectively. For SERS measurement, SiO 2 /Si wafers with different SiO 2 thickness were used as substrates to fabricate Au NPs. The thickness of dry-oxidized SiO 2 layers was 0 nm, 90 nm, 180 nm and 270 nm respectively. Au film with thickness of approximately 2 nm was then sputtered and followed by annealing in N 2 ambient by using the RTA device at 700 °C for 60 s to form Au NPs.</p><!><p>The optical characterizations of transmittance and reflectance spectra were performed using a UV-Vis-NIR spectrophotometer (Varian Cary 5000). All simulations in this work were performed with commercial Lumerical FDTD solutions (version 7.5) software. The incident plane wave propagated perpendicular to the interface of two media from z or -z directions with the same incident energy density. The polarization direction of incident wave is along the x-direction. The refractive index of quartz was set as 1.5.</p><p>SERS spectra were acquired using a confocal Raman system (Xplora, Horiba) using 532 nm laser excitation. The laser power was 5mW for the SERS measurements. The typical exposure time for our measurements was 20 s. All the spectra are presented after baseline correction by a polynomial fitting method. The SERS analysis probe R6G was dissolved in DI water to a concentration of 10 −4 mol/L. The samples were soaked in the R6G solution for 1h. Then the samples were taken out and rinsed using DI water followed by drying in N 2 gas.</p>

Scientific Reports - Nature

Characterization of tetracycline modifying enzymes using a sensitive in vivo reporter system

BackgroundIncreasing our understanding of antibiotic resistance mechanisms is critical. To enable progress in this area, methods to rapidly identify and characterize antibiotic resistance conferring enzymes are required.ResultsWe have constructed a sensitive reporter system in Escherichia coli that can be used to detect and characterize the activity of enzymes that act upon the antibiotic, tetracycline and its derivatives. In this system, expression of the lux operon is regulated by the tetracycline repressor, TetR, which is expressed from the same plasmid under the control of an arabinose-inducible promoter. Addition of very low concentrations of tetracycline derivatives, well below growth inhibitory concentrations, resulted in luminescence production as a result of expression of the lux genes carried by the reporter plasmid. Introduction of another plasmid into this system expressing TetX, a tetracycline-inactivating enzyme, caused a marked loss in luminescence due to enzyme-mediated reduction in the intracellular Tc concentration. Data generated for the TetX enzyme using the reporter system could be effectively fit with the known Km and kcat values, demonstrating the usefulness of this system for quantitative analyses.ConclusionSince members of the TetR family of repressors regulate enzymes and pumps acting upon almost every known antibiotic and a wide range of other small molecules, reporter systems with the same design as presented here, but employing heterologous TetR-related proteins, could be developed to measure enzymatic activities against a wide range of antibiotics and other compounds. Thus, the assay described here has far-reaching applicability and could be adapted for high-throughput applications.

characterization_of_tetracycline_modifying_enzymes_using_a_sensitive_in_vivo_reporter_system

Background<!><!>Background<!>Construction of a Sensitive Luminescence-Based System for the Detection of Inducing Ligands for Members of the TetR family of Repressors<!><!>Construction of a Sensitive Luminescence-Based System for the Detection of Inducing Ligands for Members of the TetR family of Repressors<!>The Detection of Tc-Modifying Enzymatic Activity In Vivo Using the pYRtetOR Luminescence System<!><!>Quantitative Analysis of the In Vivo TetX Assay<!><!>Conclusions<!>Bacterial strains, plasmid constructs and culture conditions<!>DNA manipulation procedures<!>Construction of the pYR plasmids and protein expression vectors<!>Luminescence assays<!>Enzyme Kinetics Data Fitting for Enzyme-Containing Cells<!>Estimation of Intracellular Enzyme Concentrations<!>Protein expression and purification of repressors and enzymes<!>In vitro enzymatic assays<!>Abbreviations<!>Authors' contributions<!>Additional file 1<!><!>Acknowledgements

<p>Over many decades, a wide variety of in vitro and in vivo screens have been used to identify small molecules with useful activities, such as antibiotics and enzyme inhibitors. However, there is still a need for simple and widely applicable assay systems for characterizing the activity of enzymes against specific small molecules that avoid the necessity of enzyme purification or high level expression. In the work described here, we have developed an in vivo luminescence-based reporter system that can be used to detect and characterize enzymatic activities against the antibiotic, Tetracycline (Tc). In the future, systems designed on the same principle could be used to investigate enzymes active against a variety of other small molecules.</p><p>Tc and its derivatives are highly effective broad specificity antibiotics that have been widely used for many decades [1]. The ubiquitous utilization of tetracyclines has resulted in the emergence of numerous resistance mechanisms mediated by a variety of proteins including efflux pumps, drug modifying enzymes, and ribosome protection factors [2]. The significant negative clinical impact of resistance to tetracyclines has led to intensive efforts to elucidate the mechanisms of this resistance and to develop new Tc derivatives that will overcome resistance mechanisms. To this end, much research has focused on enzymes capable of modifying tetracycline derivatives, either as a resistance mechanism or as a step in the tetracycline synthesis process [3]. It is hoped that characterization of these enzymes will lead to approaches for combating resistance and creating more potent tetracycline derivatives. Although in vitro spectroscopic methods are available to assess some enzymatic modifications of tetracyclines, in vivo assays of these enzymes are quite complicated if the enzyme activity does not confer resistance to the growth inhibitory effect of the antibiotic [4]. To aid in characterizing enzymes that modify tetracyclines and in identifying novel enzymes active against tetracyclines, a simple in vivo method to detect these activities would be very useful.</p><p>In the work presented here, we describe a system for characterizing tetracycline-modifying enzymes that takes advantage of the tetracycline repressor (TetR). Many of the genes conferring resistance to tetracyclines are regulated by TetR, which was first isolated and characterized more that 25 years ago [5]. TetR is homodimeric with each monomer composed of an N-terminal DNA-binding domain, and a C-terminal domain that mediates dimerization and binds to tetracyclines [6]. In the typical TetR-regulated regulon, TetR binds to two DNA operator sites, thereby repressing transcription of its own gene as well as the divergently transcribed tetA gene, which encodes an exporter of tetracyclines (Figure 1A). Binding of tetracyclines to the C-terminal domain of TetR leads to a conformational change in the DNA-binding domains, which causes them to lose affinity for DNA, relieving repression of the tetR and tetA genes. TetR is the founding member of a huge family of transcriptional regulators, which we refer to as the TetR family of transcriptional regulators (TFRs). TFRs constitute the third most frequently occurring transcriptional regulator family found in bacteria [7] with more than 10,000 proteins in the non-redundant protein database annotated as members of this group. TFRs have been identified that control the expression of genes conferring resistance to most known antibiotics including tetracycline, chloramphenicol, erythromycin, ampicillin, and streptomycin [6]. As well, TFRs are involved in the regulation of many aspects of bacterial physiology in response to various small molecule inducers, including quorum sensing, biofilm formation, morphological differentiation and antibiotic production. All characterized TFRs are homodimers with the same domain structure and general fold as TetR, and most of them function in a manner similar to TetR, mediating transcriptional repression that is relieved only in the presence of their specific small molecule ligand. TFR sequences in bacterial genomes can be reliably identified due to the high level of sequence conservation in their N-terminal DNA-binding domains [6]; however, their ligand-binding domains display tremendous diversity commensurate with the broad range of ligands recognized by TFRs.</p><!><p>The TetR regulon and pYR cell based repression-induction system. A) The tetracycline resistance conferring tetR-tetA regulon. The repressor gene, tetR, is depicted as a dark gray arrow and the resistance gene, tetA, as a light gray arrow. TetR regulates the transcription of tetA and its own gene by binding operators O1 and O2. The binding of TetR to the operator sequences is inhibited by interaction with Tc. B) Plasmid map of pYRtetOR and the scheme of the cell based repression-induction system. TetR expressed from the pBAD promoter on pYRtetOR represses trancription of the lux genes on the same plasmid. Upon addition of Tc, TetR is induced, the lux genes are transcribed, and luminescence is emitted from the cells. If active TetX enzyme is produced from pETTetX, then the effective concentration of Tc is reduced by the activity of the enzyme and the transcription level of the lux genes is also reduced, resulting in decreased luminescence. The box represents the E. coli cell membrane.</p><!><p>In a previous study, we designed a TetR-based biosensor that produced luminescence upon addition of tetracycline derivatives [8]. The goals of the work described here were to improve the sensitivity of this system and to then exploit it to detect the activities of tetracycline modifying enzymes. To this end, a TetR regulated transcriptional promoter was cloned upstream of the lux operon in such a way that luminescence was elicited at very low concentrations of Tc. We then demonstrated that introduction of the Tc-modifying enzyme, TetX [9], into this system led to a significant reduction in luminescence due to the activity of the enzyme, which degrades the inducer of TetR. This reporter system could be modified to both identify ligands for TFRs of unknown function, and to detect enzymes active against these ligands. Since there are currently at least 50 known ligands for TFRs [6], the assay principle described here is applicable to the characterization of a considerable number enzymes active against small molecules.</p><!><p>We previously constructed a lux-based biosensor system to investigate the binding of TetR to its DNA binding site (tetO), and to measure transcriptional induction elicited by tetracycline (Tc) and its derivatives [8,10]. For the work described here, we sought to create a more sensitive Tc-responsive system. We placed the luxCDABE gene cluster under the control of a TetR-repressible promoter on a low copy number pSC101-derived plasmid [11]. This plasmid, named pYRtetO, mediated production of a high level of luminescence in E. coli (Figure 2A, white bars), while a plasmid containing the lux genes with no promoter (pYR) produced no detectable luminescence (Figure 2A black bars). TetR was then cloned into the same plasmid under the control of the arabinose-inducible pBAD promoter to produce pYRtetOR (Figure 1B). Expression of TetR in pYRtetOR led to repression of luminescence (Figure 2A, gray bars) caused by the binding of TetR to the tetO site in the promoter of the lux genes. Repression occurred even in the absence of arabinose, indicating that the small amount of TetR expression from the pBAD promoter in the absence of arabinose was sufficient for repression of lux expression, even though TetR expression could not be detected by western blot under these conditions. Addition of arabinose only led to a small reduction in luminescence (Figure 2B).</p><!><p>Luminescence production by the pYR plasmids in the presence or absence of Tc derivatives. A) Luminescence produced by pYR, pYRtetO and pYRtetOR harboring cells. B) Degree of repression by TetR at various arabinose concentrations. TetR expression levels are indicated by an anti-TetR western blot on top of the plot. C) Induction of TetR in pYRtetOR bearing cells at different concentrations of Tc and Atc in the absence of arabinose. D) TetR induction in pYRtetOR bearing cells induced by different concentrations of Atc, Dox and ClTc in 0.02% arabinose.</p><!><p>To demonstrate the utility of our system for detecting Tc and its derivatives, luminescence production from pYRtetOR was measured in the presence of varying concentrations of these antibiotics. In these experiments the degree of induction is expressed as an induction ratio, the luminescence generated from pYRtetOR divided by that from pYRtetO (Figure 2A, white bars). Significant luminescence production was observed at Tc concentrations as low as 1 ng/mL, which is far below its minimum inhibitory concentration (MIC) in E. coli of 500 ng/mL (Figure 2C, white bars) [12]. Anhydrotetracycline (Atc), a stronger inducer of TetR [13], also displayed a greater ability to induce TetR in this assay, relieving repression of the lux operon at a concentration of only 0.1 ng/mL. The addition of 0.02% arabinose to the system, which greatly increases the intracellular concentration of TetR (Figure 2B), led to a requirement for much higher concentrations of Tc and Atc, as well as the Tc analogs doxycycline (Dox) and chlorotetracycline (ClTc), to relieve lux repression (Figure 2D). For example, full induction with Atc under these conditions required a concentration of 200 ng/mL, indicating that minimizing the cellular concentration of TetR increases the sensitivity of the reporter system. Concentrations of Tc high enough to induce luminescence under conditions of increased TetR expression caused significant inhibition of cell growth (data not shown). Together, these data demonstrate that the pYRtetOR system can detect very low concentrations of Tc and its derivatives. The reporter system is sensitive to the concentration and chemical properties of inducer molecules, and to the intracellular concentration of TetR.</p><!><p>To test the ability of the pYRtetOR system for in vivo detection of enzymatic activity against Tc, we investigated the TetX enzyme. TetX is an FAD-dependent monooxygenase from Bacteroides fragilis that has been shown to hydroxylate Tc creating an unstable compound that undergoes rapid decomposition [9]. To measure the effect of TetX when expressed in pYRtetOR-containing cells, we introduced a separate plasmid into these cells that expressed this enzyme (pETtetX, Figure 1B). For comparison, we also tested pYRtetOR-containing cells co-transformed with a plasmid expressing the D311A mutant of TetX (pETtetXD), which is substituted at a highly conserved residue in the FAD-binding site and was expected to possess no enzymatic activity (Additional file 1: Figure S1). We verified that the pETtetX construct produced active TetX and that the D311A mutant was reduced in activity using an in vitro fluorescence based assay for TetX activity (Additional file1: Figure S2). As shown in Figure 3A, we measured the luminescence generated by pETtetX- and pETtetXD-containing cells at varying concentrations of Atc, and found that considerably less light was emitted by cells containing the plasmid expressing the WT version of TetX. For example, pETtetX-bearing cells required a concentration of ~100 ng/mL Atc to generate a level of luminescence similar to that emitted from pETtetXD-containing cells at an Atc concentration of only 2 ng/mL. We surmised that the reduction in luminescence in pETtetX-containing cells was the result of TetX-mediated catalysis of Atc into an unstable product and/or a product that could no longer bind TetR. Thus, the concentration of Atc available within the cells to bind TetR was decreased, which led to a greater degree of transcriptional repression of the lux genes by TetR (Figure 1B). We found that a similarly large reduction of luminescence was elicited by pETtetX when the assay was done under conditions of high TetR expression (0.02% arabinose) even though induction did not occur until a much higher concentration of Atc was added (Figure 3B). Assays performed with Dox and ClTc indicated that, as expected [9], TetX was also active against these Tc derivatives since reduced luminescence was observed in pETtetX-containing cells treated with these compounds (Figures 3C, D). It should be noted that the expression levels of WT TetX and TetX-D311A were similar (Figure 3E), indicating that the luminescence differences observed above were due to differences in the activity of these enzymes. Similar reductions in luminescence were observed when cells were treated with Tc (data not shown).</p><!><p>Activities of TetX detected using the pYRtetOR system and fluorescence-based assays. A-D) Induction ratios in the presence of the active enzymes (open circles) or mutated inactive enzymes (solid circles). Best fit lines using the enzyme kinetics model (see text and Methods for details) are shown. A, C and D were performed in the presence of 0.0002% arabinose, which was the minimum arabinose concentration where consistent full repression could be obtained under these conditions. B was perfomed in 0.02% arabinose. E) A Western blot (anti-6xHis) showing the expression level of TetX at the point where drugs were added. Bands indicated by arrows correspond to the enzymes and bands indicated by the triangle correspond to purified TetR-6xHis protein. The enzyme bands show the total amount of enzyme from ~1 μL cell pellet, and the TetR bands show the amount of 6xHis tagged protein from 1 μL solution containing the indicated concentration protein.</p><!><p>Our success in detecting the enzymatic activity of TetX in a cell based assay prompted us to determine whether the behavior observed in our assays could be accounted for by the known kinetic parameters of the TetX enzyme. To this end we formulated a series of equations to describe the TetX enzymatic activity within the cell based system in terms that were as simple as possible (see Methods for details). The objective of our analysis was to account for the difference between the dose-response curves generated in the presence of TetX as compared to TetXD311A, which we showed above is an inactive enzyme. Our equations were predicated on the assumption that once tetracycline is added, the media becomes an infinite drug reservoir. In the absence of TetX, drug molecules enter cells from the media driven by diffusion and rapidly reach an effective steady state concentration, which is referred to as [Ieff]. With TetX present inside the cell, the intracellular concentration of drug is simultaneously increased by the process of diffusion and decreased by the enzymatic activity of TetX. The intracellular drug concentration reaches equilibrium only when the rate of inward diffusion is matched by the rate of enzymatic modification. Thus, the final effective concentration of drug within these cells ([Ieff]) is a function not only of the extracellular concentration of drug ([Iout]), but also the rate of drug diffusion, the enzymatic activity of TetX, and the time taken after drug addition for the intracellular drug concentration to reach equilibrium (t).</p><p>In our data fitting, the enzymatic activity of TetX on Dox and ClTc was modeled by entering the known in vitro Km and kcat values of TetX for these compounds as fixed parameters [9]. The only free parameter in the fitting process was an arbitrary diffusion constant, K that accounted for the diffusion properties of Tc derivatives. Although parameters for the diffusion of Tc into E. coli have been experimentally determined [14,15], the value of K for our fitting could not be determined a priori because it is not known how the diffusional properties of Tc derivatives would change upon modification by TetX or how quickly the modified Tc derivatives might degrade within the cell. In fitting the data from each experiment, the data generated for TetXD311A was used as a reference to predict how much luminescence would be generated at a given effective concentration of drug in the absence of enzyme. It can be seen in Table 1 that we were able to fit the Dox and ClTc data effectively using the known kinetic parameters of TetX (R2 values equaled 0.90 and 0.66 for Dox and ClTc, respectively). To fit the curves generated using Atc, for which Km and kcat values for TetX were not known, we allowed Km and kcat to also be free parameters. Notably, we were still able to obtain good fits to our data and the parameters returned were similar to those in the other fits. These data suggest that TetX acts on Atc with similar kinetic parameters as on Dox and ClTc. The ability to obtain fits to different experiments with consistent enzyme parameter values supports our conclusion that the behavior of this system is the result of the enzymatic activity of TetX against Tc derivatives. The use of the data generated from the pETtetXD-containing cells as the reference curve requires that the time taken after drug addition for the intracellular drug concentration to reach equilibrium in pETtetX-containing cells (t) is short enough to not affect luminescence accumulation. With the enzyme parameters from the above data fitting, we were also able to estimate t (see Methods for details). As shown in Table 1, t was less than 10 min in all experiments, which is much shorter than the time at which luminescence was measured (2-4 hours).</p><!><p>Fitting results of cell based TetX activity experimentsa</p><p>a Fixed parameters were shown in regular fonts and values returned from fitting were shown in bold italic fonts</p><!><p>In the work described here we have developed a sensitive luminescence-based in vivo assay to detect enzymatic activity against tetracyclines. Our assay system is based on the ability of these compounds to induce TetR regulated transcription and the diminution of this induction that occurs when a Tc modifying enzyme activity is present within the cell. An important aspect of this system is that it not only detects enzyme activity, but also allows quantitation of this activity. The only requirement is that the Tc derivative is modified in a way that lowers its affinity for TetR. Of course, our system will fail to detect an enzyme activity if the modification produced does not change the affinity of the Tc derivative for TetR. However, due to the sensitivity of our assay system and the use of titrations to detect activity, even a small change in affinity would be detected. In addition, there are a number of TetR mutants with varying reactivities towards different Tc derivatives [16,17] that could be utilized in our assay system to maximize the number of Tc modifications that could be detected. Finally, there are other TetR-like repressors that are induced by Tc derivatives and at least one of these, TtgR of Pseudomonas putida [18], binds to Tc by a completely different mechanism compared to TetR. By using these different repressors, a wide range of modifications of tetracyclines would likely be detectable. Since there is great interest in the identification of enzymes that may modify tetracyclines to produce more potent antibiotics [19], our assay could be useful as a rapid screen to determine the level of activity of a given enzyme against a range of tetracyclines. Because of the wide range of ligands bound by TFRs [6], the assay system described here could be adapted to test for enzymatic activities against a huge variety of antibiotics and other small molecules. Supporting the general utility of our system, we have constructed reporters analogous to pYRtetOR for 22 diverse TFRs, and all of these TFRs are able to repress transcription of the lux operon when expressed in E. coli (data not shown). A reporter similar to the one described here was used to discover new ligands for the ActR repressor of S. coelicolor [20,21]. Future studies will determine whether enzyme activities for a variety of small molecules will be detectable using these systems. It should be noted that the same principles used to design this TetR-based system could be used for any of the other families of transcriptional regulators that are induced by small molecules.</p><p>In general, our TetR-based system or other systems designed in a similar manner present many advantages for investigation of enzymes with activities against small molecules. First, we are able to detect enzyme activity using only nanogram quantities of compound. This feature could be critical in screening for activities against compounds that are available only in small quantities as in the case of compound libraries used for high-throughput screening. The ability to modulate the expression level of the repressor in our system by addition of varying concentrations of arabinose allows the intracellular repressor concentration to be adjusted to a level at which there is just enough present to repress transcription. Consequently, addition of only a small amount of inducer is required for derepression of the system. A second advantage of our system is that it provides the potential to investigate enzymatic activities without having to purify the enzymes or know their co-factor requirements. Our assay system would be equally capable of measuring the activity of difficult to purify membrane proteins, such as drug efflux pumps. TetR-based reporter systems have been shown to function in a wide range of cell types including mammalian cells [22-24], thus, enzymatic assays operating by the same principle as ours could be adapted to many cell types. A final advantage of our system is that it could be used to screen for enzymes with activity against a given compound of interest. For this purpose, a library of plasmids expressing candidate enzymes would be transformed into a strain containing a pYR-derivative that responded to the compound of interest. Individual colonies could then be screened in a 96-well format for reduced luminescence resulting from expression of an enzyme that modified the TFR-binding molecule. In this way, it will be possible to systematically identify novel enzymes with activities against many important small molecules.</p><!><p>Bacterial strains and plasmids used in this work are shown in Additional file1: Table S1. E. coli cells were grown at 37°C in Luria broth (LB) or LB agar medium containing the following antibiotics when necessary: kanamycin (50 μg/mL), ampicilin (100 μg/mL) and streptomycin (50 μg/mL). Protein expression for purification and E. coli drug susceptibility assays were carried out in E. coli BL21*(DE3). In vivo repression and induction assays were performed using E. coli Top10 (Invitrogen) because it is an arabinose metabolism deficient strain and produces more luminescence than BL21*(DE3). A T7 polymerase bearing E. coli Top10 (DE3) strain was constructed for the in vivo enzymatic activity assay using the Lambda DE3 Lysogenization Kit from Novagen.</p><!><p>Standard procedures were employed for all DNA manipulation and molecular cloning [25]. The oligonucleotides and primers used in this study were synthesized from the ACGT Corporation (Toronto) and listed in Additional file1: Table S2. PCR reactions were carried out using Vent DNA polymerase (New England Biolabs). The QuikChange site-directed mutagenesis protocol (Stratagene) was employed to create point mutations of TetX.</p><!><p>The pCS26-Pac plasmid [11], which contains the luxCDABE operon on a low copy number pSC101-derived vector [26], was modified by inserting a double stranded oligonucleotide (oligo YLuxupdateF annealed to oligo YLuxupdateR) upstream of luxCDABE operon in between XhoI and BamHI sites. The araBAD promoter and the araC gene from the pBAD vector (Invitrogen), were amplified using primers SR204 and SR205 and then ligated into the EcoRI and KpnI restriction sites introduced as described above, creating the pYR plasmid (~13 kbp). A synthetic promoter consisting of the TetR operator and promoter tetO from the Tn10 tetA promoter region was prepared by annealing YZ201 and YZ202 oligonucleotides. This promoter was then introduced into pYR between KpnI and PmlI upstream of the luxCDABE cluster to give pYRtetO. The tetR gene was amplified from pETtetR [27] using primer YZ203 and YZ204 and cloned into pYRtetO between XhoI and EcoRI, downstream of the araBAD promoter, to give pYRtetOR. Expression vector pETtetX was prepared by introducing the PCR amplified tetX gene using primer pair YZ241-2 YZ242 into pET21 between EcoRI and XhoI sites.</p><!><p>For the repression and induction assays, isolated E. coli colonies were used to inoculate 2 mL cultures, which were grown overnight. 10 μL of overnight culture was added into 2 mL of LB media in the presence of varying concentrations of arabinose and drugs. These cultures were grown for 12 to 16 h before measurement of luminescence using a BMG Fluostar OPTIMA luminometer. In the enzyme assays, we inoculated 4 μL overnight cultures into 200 μL fresh pH 7 buffered LB media containing 0.04% glycerol, then grew for 2-4 h until early log phase and added varying concentrations of drugs. Luminescence and optical density were measured every 15 min using a TECAN Infinite M200 luminometer. Luminescence from similar early stationary phase cells (around 2-4 h after drug induction) was used to calculate the induction ratio. Multiple replicates (N > = 2) were performed and the 95% confidence intervals which are 1.96 times standard errors were displayed as error bars.</p><!><p>In the presence of TetX, the effective cytoplasmic inducer concentration, [Ieff] was expected to reach steady state when the enzyme catalytic rate equalled the diffusion rate. The enzyme catalytic rate was expressed using the Michaelis-Menten equation:</p><p>(1)vt=kcat⋅[E]⋅[Ieff]Km+[Ieff]</p><p>where vt is the rate of reaction at time point t after inducer addition with an intracellular inducer concentration of [Ieff] at that time point, kcat is the catalytic rate constant, Km is the Michaelis constant and [E] is the enzyme concentration, which was assumed to be constant. The concentration of enzyme was estimated through analysis of Western Blots (see below). To determine [Ieff] as a function of the extracellular inducer concentration ([Iout]), the diffusion process across the cell membranes was modeled. Diffusion is normally described by the Fick's first law of diffusion:</p><p>(2)vd=P0⋅A0⋅[Iout−Ieff]</p><p>where vd is the diffusion rate, P0 is the permeability coefficient through diffusion barrier, and A0 is the barrier surface area. Although P0 and A0 are known for Tc [28], to simplify the model and generalize it to situations where P0 and A0 might not be known for a given compound, we replaced these parameters with an arbitrary diffusion constant K. The diffusion process was then expressed by:</p><p>(3)vd=K⋅[Iout−Ieff]</p><p>K accounts for not only P0 and A0, but also the diffusion-effective concentration difference across the cell membranes. Since enzymes may only make subtle changes on the inducer, it is impossible to predict how these changes would affect diffusion properties. In addition, some enzymatic modifications will lead to inducer degradation. In the steady state, vt equals to vd, therefore the inducer concentration [Ieff] could be determined by solving:</p><p>(4)kcat⋅[E]⋅[Ieff]Km+[Ieff]=K⋅[Iout−Ieff]⇒[Ieff]=−(kcat⋅[E]/K−[Iout]+Km)+(kcat⋅[E]/K−[Iout]+Km)2+4⋅Km⋅[Iout]2</p><p>Thus, the steady state [Ieff] can be described as a function of three parameters: K, Km and kcat. The relationship between [Ieff] and the luminescence induction ratio (L) was determined from curves generated for pETtetXD-containing cells, which were assumed to contain no active TetX. In this situation, the steady state [Ieff] and [Iout] were assumed to be equivalent. While this may not be strictly true in all conditions (in fact, the intracellular concentration of Tc has actually been found to be 4-fold higher than the extracellular concentration under some conditions [15]), the success of our fitting procedure requires only that [Ieff] and [Iout] are related by a constant value over the range of inducer concentrations used. Any deviation from equivalence of [Ieff] and [Iout] is accounted for in the K value described above, which is a free parameter in our fitting scheme. The curves of L versus [Iout] for the pETtetXD-containing cells were fit to a hyperbolic equation (in the presence of 0.0002% arabinose) or a sigmoidal model (in the presence of 0.02% arabinose) as shown:</p><p>(5)L=a⋅[Iout]b+[Iout]+c⋅[Iout]</p><p>(6)L=L0+a⋅[Iout]bcb+⋅[Iout]b</p><p>where a, b, c and L0, are arbitrary parameters giving the completed standard curves. Although these equations have no physical relevance to the functioning of the system, they accurately captured the relationship between L and [Iout] under these conditions (R2 > 0.9 in every fitting). Finally the L versus [Iout] curves for pETtetX-containing cells were fitted to the following equations using the information derived from the fits of the pETtetXD-containing cells.</p><p>Under 0.0002% arabinose:</p><p>(7)L={a⋅[Iout](b+[Iout])+c⋅[Iout],[Iout]≤[Is]a⋅[Ieff](b+[Ieff])+c⋅[Ieff],[Iout]>[Is]</p><p>Under 0.02% arabinose:</p><p>(8)L={L0+a⋅[Iout]b(cb+[Iout]b),[Iout]≤[Is]L0+a⋅[Ieff]b(cb+[Ieff]b),[Iout]>[Is]</p><p>The equations above also took into account the competition for drugs from TetR. Since TetR binds Tc derivatives much more tightly than TetX, TetX will only induce when TetR is saturated. The TetR saturating inducer concentration [Is] was set to be the inducer concentration corresponding to the onset of luminescence induction. For fitting curves to data derived from pETtetX-containing cells in response to Dox and ClTc, only the K value was left as free parameters and published in vitro Km and kcat values were used.</p><p>The steady state [Ieff] can also be expressed as a function of the equilibration time t:</p><p>(9)[Ieff]=∫0t(vd−vt)⋅tdt</p><p>where vd was determined from equation (3) and vt was determined from equation (1). With K, Km and kcat determined in previous steps, t became the only free parameter and can be returned by data fitting using equation (7) or (8).</p><!><p>To estimate in vivo enzyme concentrations, TetX-bearing cells were harvested at the time when drugs were added to induce luminescence. Approximately 1 μL cell pellet was lysed and loaded on a SDS-PAGE and subjected to a Western blot using anti-6xHis antibody. A series of purified 6xHis tagged TetR was loaded on the same gel as concentration standards (Figure 3E). TetX was estimated to be 15 μM.</p><!><p>Proteins were expressed as C-terminal hexahistidine fusions in E. coli strain BL21*(DE3) using plasmids derived from pET21d (Novagen). TetR was purified as previously described [27]. For TetX and mutants, cells bearing pETtetX were grown at 30°C to an OD600 of 0.8, induced with 1 mM IPTG, and the grown overnight at 20°C. Cells were harvested and lysed in 6 M GuHCl. Protein was purified using nickel affinity chromatography by the denatured protein procedure (Qiagen). TetX was refolded by dialyzing into 10 mM Tris-HCl (pH 8.0), 50 mM NaCl, 0.1 mM EDTA, 1 mM DTT and 2% glycerol. These proteins were further purified through a hi-load sephadex-75 gel filtration column (Pharmacia) using a Pharmacia LKB FPLC. All in vitro protein assays were performed in the dialysis buffer described above.</p><!><p>Fluorescence assays were performed using an Aviv ATF 105 spectrofluorometer in a 1 cm path length cuvette. Complex solutions with the indicated concentration of ingredients were excited at 340 nm and emission between 350 and 600 nm was recorded over time, averaging for 2 seconds at each wavelength.</p><!><p>TetR: tetracycline repressor; TFR: TetR family of transcriptional regulators; Tc: Tetracycline; Atc: Anhydrotetracycline; Dox: doxycycline; ClTc: chlorotetracycline; MIC: minimum inhibitory concentration.</p><!><p>ZY, SER and LC performed research; ZY analyzed data; ZY, SER, JRN and ARD designed research; and ZY and ARD wrote the manuscript. All authors read and approved the final document.</p><!><p>Additional figures and tables for Characterization of tetracycline modifying enzymes using a sensitive in vivo reporter system This file contains additional figures and tables of the main manuscript</p><!><p>Click here for file</p><!><p>Funding was provided by operating grants from the Canadian Institutes of Health Research awarded to A.R.D. (MOP-13609) and J.R.N. (MOP-97729). Z.Y. was supported by an Ontario Graduate Scholarship and L.C. was supported by a postdoctoral fellowship from the Natural Sciences and Engineering Research Council of Canada. S.E.R was supported by a Canadian Institutes of Health Research Training Grant in Protein Folding. The authors would like to thank William Navarre for allowing us to use his TECAN Infinite F200 plate reader.</p>

PubMed Open Access

Theoretical Study on the Sensing Mechanism of Novel Hydrazine Sensor TAPHP and Its ESIPT and ICT Processes

The photophysical and photochemical properties of the novel hydrazine sensor TAPHP and the TAPDP generated by the cyclization reaction of TAPHP with hydrazine are investigated using the density functional theory and time-dependent density functional theory. The results show that both the excited-state intramolecular proton transfer and intramolecular charge transfer can occur for TAPHP and TAPDP. Analysis of bond parameters and infrared vibrational spectra indicate that hydrogen bonds are enhanced in the first excited state, which is beneficial to excited-state intramolecular proton transfer. The strength of hydrogen bonds is also visualized by using the independent gradient model and topological analysis. The core-valence bifurcation index and bond critical point parameters are further employed to measure hydrogen bonds. The reaction path of proton transfer is obtained through the potential energy curves. The excitation of TAPHP and TAPDP is attributed to the charge transfer excitation, which is determined by the characteristics of the hole-electron distribution. The reaction site and product configuration are verified by atomic charge and 1H-NMR spectra. The negative free energy difference indicates that the reaction between TAPHP and hydrazine can proceed spontaneously. In addition, the absorption and fluorescence spectra agree well with the experimental results, confirming that TAPHP is an excellent sensor of hydrazine.

theoretical_study_on_the_sensing_mechanism_of_novel_hydrazine_sensor_taphp_and_its_esipt_and_ict_pro

Introduction<!><!>Theoretical Calculation Methods<!><!>Geometrics<!><!>Geometrics<!>IR Vibration Analysis<!><!>IR Vibration Analysis<!>Hole-Electron Analysis<!><!>Hole-Electron Analysis<!><!>Absorption and Emission Spectra<!><!>Topologocal Analysis<!><!>CVB Index<!><!>Independent Gradient Model<!><!>Potential Energy Curves<!><!>Analysis of Sensing Mechanisms<!><!>Analysis of Sensing Mechanisms<!><!>Analysis of Sensing Mechanisms<!><!>Conclusions<!>Data Availability Statement<!>Author Contributions<!>Conflict of Interest<!>

<p>Hydrogen bond (HB) observed in DNA, water, proteins, and other materials (Zhao and Han, 2011; Tanioku et al., 2013; Gole et al., 2014; Ling and Gutowski, 2016; Huang et al., 2018) formed between a hydrogen (H) atom of one molecular fragment D-H and another atom A (i.e., D-H…A) (Li et al., 2011; Wilcken et al., 2013; An et al., 2016, 2017; Ma et al., 2018; Zhao et al., 2019). Where D atom is connected to H atom by a covalent bond (or ionic bond), the A atom has a high electron density and is easy to attract hydrogen proton. HBs are ubiquitous in molecular structures and have been the research focus due to their distinct properties. Studies have shown that many sensing mechanisms can be well-explained by HB interactions, such as the excited-state intramolecular proton transfer (ESIPT), fluorescence quenching, and intramolecular charge transfer (ICT) (Wen and Jiang, 2004; Plasser et al., 2009; Rawat and Biswas, 2014; Zhu et al., 2017). The ESIPT was first found by Weller when the double fluorescence emission of methyl salicylate was observed (Weller, 1956; Zhou et al., 2014). Such a pioneering discovery soon led to a flurry of studies on the proton transfer mechanism. Recently, Han et al. discovered that HB interaction can be enhanced in excited states (Zhao and Han, 2007; Zhao et al., 2007; Chai et al., 2009; Liu et al., 2018; Song et al., 2019). According to Weller, the ESIPT process of methyl salicylate was induced by its intramolecular HB.</p><p>ESIPT is a photochemical process that produces a tautomer with a different electronic structure from the original excited form through a four-level photo cycle (Enol-Enol*-Keto*-Keto) (Mahanta et al., 2011; Demchenko et al., 2013; Ray et al., 2014; Chen et al., 2016; Kumpulainen et al., 2016; Liu et al., 2019b). Among them, the role of H proton in the ESIPT process is extremely important due to its active characteristics. The energy difference between the initial and the relaxed excited state will power the proton transfer. The ESIPT model can be described by a variety of reaction coordinates. One of the simplest ways is to use the D-H bond length as the reaction coordinate. As the D-H bond is stretched, the reaction will continue to drive (Lee et al., 2013). Molecules with ESIPT property have been used in various fields due to their unique photophysical properties (Liu et al., 2019a). One of the more valuable applications is the use of fluorescent sensors for ions, biomolecules, and chemicals (Padalkar and Seki, 2016). For example, Wang et al. synthesized a fluorescence sensor to monitor methanol based on ESIPT characteristics (Qin et al., 2018). Yang et al. proposed that Al3+ sensor can be prepared by ESIPT and PET mechanisms (Yue et al., 2017). Interestingly, a novel hydrazine sensor TAPHP was recently synthesized by Wang et al. (Luo et al., 2018).</p><p>Hydrazine is a colorless oily liquid, which is well-soluble in polar solutions such as water and alcohol. The benefit is that it can make fuel for rockets and jet engines, mirror silver plating and foaming agent. While, the drawback is that it can cause severe skin erosion and damage to the eyes and liver due to its extreme toxicity. Therefore, we are motivated to investigate the sensing mechanism of the newly designed hydrazine sensor TAPHP. It is commendable that the ESIPT process is considered in the probe skeleton, which produces long-wavelength emission to avoid autofluorescence of the probe. We found that the TAPDP molecule generated by the cyclization reaction can also occur ESIPT and ICT, which was not observed experimentally. Hydrazine can be specifically recognized by the probe since its amino group has stronger nucleophilicity than other amines. However, it is difficult to distinguish hydrazine and hydroxylamine composed of -OH group and -NH2 group by ordinary fluorescent probes. The probe TAPHP can recognize hydrazine and hydroxylamine because it can undergo cyclization reaction with hydrazine, and the -OH group on hydroxylamine has a lower nucleophilicity, which is not conducive to cyclization reaction. As shown in Figure 1A, the cyclization reaction of TAPHP with hydrazine is accomplished by condensation of an amine with ketone and the conjugation addition of another amine of hydrazine to α, β-unsaturated carbonyl group. In this work, we focused on the ESIPT and ICT processes of TAPHP and TAPDP through analyzing bond lengths and bond angles, infrared (IR) vibration spectra, potential energy curves (PECs), and hole-electron distribution, etc. The calculated atomic charge and proton nuclear magnetic resonance (1H-NMR) spectra verify the reaction site. The sensing mechanism of TAPHP is also well-confirmed by the large stokes shift of the fluorescence spectra.</p><!><p>(A) Chemical structures of probe TAPHP and TAPDP formed by cyclization reaction between TAPHP and hydrazine. (B) Optimized geometries of the TAPHP and TAPDP.</p><!><p>All ab initio calculations were carried out by using Gaussian 16 suite (Frisch et al., 2016). The density functional theory (DFT) (Lee et al., 1988) and time-dependent density functional theory (TDDFT) (Becke, 1993) methods are adopted to optimize the geometries of TAPHP and TAPDP in the ground (S0) and first excited (S1) states without any constraints, respectively. A series of functionals [B3LYP (Miehlich et al., 1989), CAM-B3LYP (Yanai et al., 2004), PBE0 (Adamo and Barone, 1999), M06-2X (Zhao and Truhlar, 2008), and PW6B95 (Grimme et al., 2011)] suitable for calculating weak interactions are tested. Table 1 shows that the absorption peaks of TAPHP and TAPDP calculated by functional PW6B95 are the closest to the experimental results. Meanwhile, the triple-zeta valence quality with one set of polarization functions (TZVP) (Eichkorn et al., 1997; Vargas et al., 2000) performs well in this work. Therefore, we employ PW6B95/TZVP as the most reliable method in the current work. In order to be as consistent as possible with the experimental environment, the solvent effect [dimethyl sulfoxide (DMSO)] (Bordwell et al., 1990) based on the polarizable continuum model (PCM) using the integral equation formalism variant (IEFPCM) is utilized in ab initio calculations (Cammi and Tomasi, 1995; Cances et al., 1997). To improve calculation accuracy, the implicit solvation model based on density (SMD) (Marenich et al., 2009) was employed to calculate free energy and 1H-NMR spectra. The IR vibration analysis after geometric optimization shows that all the molecular configurations are the true minimum. The independent gradient model (IGM), topological analysis, and core-valence bifurcation (CVB) index are employed to visualize the strength of intramolecular HBs as well as electron-hole distribution were obtained by using Multiwfn program (Lu and Chen, 2012a,b). Additionally, to quantitatively describe the reaction path of proton transfer, the PECs are obtained by scanning the bond length of O1-H1 with a step of 0.1 Å. In the present work, the enol and keto forms of TAPHP and TAPDP are represented as TAPHP-Enol, TAPHP-Keto, TAPHP-Enol*, TAPHP-Keto*, TAPDP-Enol, TAPDP-Keto, TAPDP-Enol*, and TAPDP-Keto* in the S0 and S1 states, respectively.</p><!><p>The absorption maxima (nm) of TAPHP and TAPDP in experimental and theoretical calculation.</p><p>Maximal absorption peak in experiment.</p><!><p>The two optimized geometric configurations (keto and enol) of TAPHP and TAPDP are shown in Figure 1B. To make the following description clearer, we labeled the atoms associated with HBs (O1, H1, O2, and N1). The relevant bond parameters are collected in Table 2. By comparing the HB parameters of TAPHP's enol form of S0 and S1 states, one can observe that the O1-H1 bond length is elongated by 0.019 Å (0.991 Å→1.01 Å), the distance of H1…O2 is shortened by 0.093 Å (1.607 Å→1.541 Å), and the bond angle of O1-H1…O2 is increased by 4.9° (149.6°→154.5°). Obviously, the strength of the O1-H1…O2 is strengthened in the S1 state, which is conducive to proton transfer. Optimization calculation shows that the keto form of TAPHP only forms in S1 state. In other words, TAPHP only performs ESIPT, rather than the ground state intramolecular proton transfer (GSIPT).</p><!><p>Bond parameters of TAPHP-Enol, TAPHP-Keto, TAPDP-Enol, and TAPDP-Keto forms in the S0 and S1 states.</p><!><p>Moreover, HB interaction is found in TAPDP. By comparing the HB parameters of TAPDP's enol form of S0 and S1 states, we can find that the O1-H1 bond length is elongated by 0.007 Å (0.983→0.990 Å), the distance of H1…N1 is shortened by 0.054 Å (1.747 →1.693 Å), while the bond angle of O1-H1…N1 changes little. These changes indicate that O1-H1…N1 is enhanced in the S1 state. The keto form of TAPDP has been obtained by optimizing calculation. By comparing the HB parameters of TAPDP's keto form of S0 and S1 states, the distance of O1…H1 is shown to be elongated by 0.160 Å (1.717→1.877 Å), the bond length of H1-N1 is reduced by 0.014 Å (1.034→1.020 Å), and the bond angle of O1…H1-N1 is decreased by 0.8° (132.0→129.2°). Such variation can clearly show that O1…H1-N1 is stronger in the S0 state, which means the reverse proton transfer is more likely to occur in the S0 state.</p><!><p>IR vibration spectra is a useful tool for analyzing hydrogen bonds. The IR spectra of O1-H1 and H1-N1 related to hydrogen bonds in TAPHP and TAPDP are displayed in a, b and c of Figure 2, respectively. It can be seen clearly from this figure that the vibration frequencies of O1-H1 stretching in TAPHP's enol form show the red-shift of 415 cm−1 (3,203→2,788 cm−1) from S0 to S1 states. The red shift of IR spectra indicates the enhancement of hydrogen bond, whereas the blue shift indicates the weakening of hydrogen bond. Obviously, the O1-H1…O2 of TAPHP's enol form is strengthened in S1 state.</p><!><p>IR spectra of TAPHP and TAPDP. (A) TAPHP, O1-H1; (B) TAPDP, O1-H1; (C) TAPDP, H1-N1.</p><!><p>Similarly, the stretching vibration frequencies of O1-H1 in TAPDP's enol form also show a red shift of 165 cm−1 (3,354→3,189 cm−1) from S0 to S1 states. This means that the O1-H1…N1 in TAPDP's enol form is enhanced in S1 state. However, the stretching vibration frequencies of H1-N1 in TAPDP's keto form reveal a blue shift of 208 cm−1 (3,181→3,389 cm−1) from S0 to S1 states. The conclusion that the O1…H1-N1 in TAPDP's keto form is stronger in S0 state is confirmed. Such conclusion is consistent with the above analysis of bond parameters.</p><!><p>Hole-electron analysis is an intuitive way of graphically examining electron excitation characteristics. The hole-electron distribution and the Chole-Cele diagrams drawn by the Multiwfn program (Gao et al., 2019) are shown in Figure 3. The orange isosurfaces represents the electron distribution and the blue isosurfaces represents the hole distribution.</p><!><p>Hole-electron distribution and the Chole-Cele diagrams of TAPHP and TAPDP.</p><!><p>The Chole-Cele diagram smoothes out the complex isosurfaces by erasing the details of hole and electron distribution. At the same time, Bahers et al. (2011). proposed a series of indexes including the DCT index, which are adapted by Lu (2018) in the Multiwfn. The relevant indexes (D, Sr, H, and t) are collected in Table 3. The four defined indexes and Chole-Cele can be expressed by the formula:</p><p>where X, Y, Z (formulas 1, 2, 3) refer to the centroid coordinates of holes and electrons, respectively. The D index represents the distance between the centroid of the hole and the electron, which are 4.65 and 4.57 Å for TAPHP and TAPDP, respectively. The distance between the centroid of the hole and the electron is very large, which is obviously the result of the charge transfer excitation. And it can be seen from the hole-electron distribution map that electrons are transferred from triphenylamine to ketone. That is to say, both TAPHP and TAPDP undergo ICT instead of the experimental literature description that only TAPHP can occur. Both holes and electrons can be defined as σ (formula 5), and its three components of x, y, and z correspond to the root mean square deviation (RMSD) of holes or electrons distributed in the x, y, and z directions, reflecting the breadth of the distribution of holes and electrons. Therefore, the H index can reflect the overall average distribution breadth of electrons and holes. The t index measures the separation of holes and electrons. The t index > 0 implies that the separation of holes and electrons is sufficient due to charge excitation. In addition, Sr index indicates the overlap degree of holes and electrons.</p><!><p>Indexes related to hole-electron distribution in TAPHP and TAPDP.</p><!><p>In order to more vividly illustrate the rationality of our calculation method, the electronic spectra of TAPHP and TAPDP are drawn in Figure 4. The calculated absorption peaks of TAPHP and TAPDP are 309, 475, and 344 nm, respectively, which coincide with the experimental values (295, 450, and 309 nm). The calculated results indicate that both the TAPHP and TAPDP are ESIPT fluorophores with double fluorescence emission, but only their single fluorescence peaks can be found in the experimental data. The reason that the fluorescence of TAPDP at 503 nm has not been experimentally observed can be considered to be that its intensity is too weak. We attribute the peak at 610 nm obtained in the experiment to the emission peak in the enol form of TAPHP, because the following PECs analysis indicates that TAPHP-Enol* is stable in the S1 state. The fluorescence peak movement of TAPDP relative to TAPHP affords available sensing mechanism for the specific recognition of hydrazine.</p><!><p>Electronic spectra of TAPHP and TAPDP in DMSO. (A) Absorption spectra. (B) Fluorescence spectra.</p><!><p>Atoms-In-Molecule (AIM) theory, which describes bonding in molecules, is a reliable tool for characterizing HB interactions. The topological properties of electron density [ρ(r)] can be intuitively expressed in the Lewis structure of molecules by using the AIM method. According to Bader's theory (Bader and Essén, 1984), critical points and bond paths in equilibrium are both the necessary and sufficient conditions to specify the interaction (HB interaction) between two atoms. Topological analysis diagrams of geometric configurations related to HBs in TAPHP and TAPDP are displayed in Figure 5. Fortunately, there are bond critical points (BCPs) and bond paths between the two atoms associated with HBs in all configurations. This is a visual representation of the presence of HBs in the molecule. We call the BCPs of each configurations at different states as BCPi (i = 1,….7). The indicators (shown in Table 4) of BCPs are the focus of our attention because they are the key to the strength of the interaction. For homogeneous interactions, the higher the ρ(r) and the more negative the density of potential energy [V(r)] at the BCPs, the stronger the interaction between the two atoms connected by the bond path. It can be found that absolute values of ρ(r) and V(r) at BCPs follow the following relation: BCP2 > BCP1, BCP5 > BCP4, and BCP6 > BCP7. Meanwhile, the absolute values of hydrogen bond energy (EHB), Laplacian of electron density [∇ 2ρ(r)], kinetic energy density [G(r)] and total electron energy density [H(r)] also fit the rule well. The strength of HB can then be concluded to have the following relations: TAPHP-Enol* > TAPHP-Enol, TAPDP-Enol* > TAPDP-Enol, and TAPDP-Keto> TAPDP-Keto*. In addition, the indicators such as ρ(r) at BCP3 are also relatively high, which means that a strong HB favoring reverse proton transfer is formed in TAPHP -Keto*.</p><!><p>Topological analysis diagrams of TAPHP and TAPDP including bond paths and bond critical points.</p><p>Obtained parameters of BCPs in the TAPHP and TAPDP.</p><!><p>The CVB index obtained by using the topological analysis of electronic localization function (ELF) is a method proposed by Silvi et al. to research the strength of hydrogen bond (Fuster and Silvi, 2000). In the present work, hydrogen bond is represented as D-H…A, where D is the donor atom, and A is the acceptor atom. In the ELF basin analysis, this area is made up of the following ELF basins: V(D, H): valence basin formed by D atom and its bonding H atom, C(D) and C(A): the core basins of D and A atoms, V(A): valence bath of A atom. The CVB index is defined as:</p><p>where ELF (C-V, D) is the core-valence bifurcation point value of D atom. ELF (DH-A) represents the bifurcation point value between V(D, H) and V(A). It has been verified that the more negative CVB index is, the stronger the HB is in general. The stronger the HB, the closer the distance between H atom and A atom, and the more covalent the interaction will be. Therefore, ELF (DH-A) is bound to become larger, resulting in a more negative CVB index. The CVB index of TAPHP and TAPDP in different states are displayed in Table 5. All the CVB indexes are negative, which indicate the existence of hydrogen bond interactions between the related atoms. Their relationship of absolute value of the CVB indexes are: TAPHP-Enol* > TAPHP-Enol, TAPDP-Enol* > TAPDP-Enol, and TAPDP-Keto> TAPDP- Keto*, which corresponds to the strength of hydrogen bond. It is worth to note that the CVB index of TAPHP-Keto in S1 state is the most negative, which demonstrates that O1…H1-O2 is a strong hydrogen bond.</p><!><p>The CVB index of TAPHP and TAPDP in S0 and S1 states.</p><!><p>The method of investigating weak interactions by reduced the density gradient (RDG) is well-known. Recently, a new model (IGM) has been proposed that can precisely extract the characteristics of interactions in RDG diagrams (Lefebvre et al., 2017). The IGM provides a method for identifying and quantifying the gradient attenuation of net electron density. Its core is a descriptor (δg) that uniquely defines the region of interaction. The general way to calculate the density gradient [g(r)] of the pro-molecular is to sum over the density gradient of each atom, while the IGM type density gradient [gIGM(r)] is to sum over the absolute value of the density gradient of each atom. The difference between the two density gradients is called δg function, which is a three-dimensional real space function, can be written as,</p><p>where i denotes the atom number, ∇ρ the gradient vector, and abs(∇ρ) the absolute value of each component of the vector ∇ρ. The advantage of IGM is that it can examine the strength of interaction between each pair of atoms and quantify how much each atom affects the interaction between fragments.</p><p>We define the D-H of the HB interaction D-H…A as one fragment, A as another fragment. The scatter plots and isosurface maps, which can intuitively show the interaction, are drawn in Figure 6. The dark blue isosurfaces indicate that strong intramolecular HBs exist in TAPHP, TAPDP, and their isomers. The strength of the HBs can be shown by the spikes in the scatter diagrams. The peak value of O1-H1…O2 in S1 state is more negative than that in S0 state, which provide a strong proof for the enhancement of O1-H1…O2 in S1 state and the existence of O1…H1-O2 in S1 state as a strong hydrogen bond. For TAPDP, the peak value of O1-H1…N1 is more negative in S1 state than in S0 state, while the peak value of O1…H1-N1 has an opposite trend to that of O1-H1…N1. Obviously, O1-H1…N1 is stronger in the S1 state, while O1…H1-N1 is stronger in the S0 state. This further implies that proton transfer occurs in S1 state and reverse proton transfer occurs in S0 state.</p><!><p>The scatter plots and gradient isosurfaces of TAPHP, TAPDP's enol form, and TAPDP's keto form.</p><!><p>To investigate the reaction path, the PECs of TAPHP and TAPDP are drawn in the Figure 7. As expected, the scans showed that the H proton moved along the reaction path near the O1 atom in TAPHP and the N1 atom in TAPDP. From this, we can speculate that proton transfer occurs in both TAPHP and TAPDP. According to the PECs of TAPHP, there is no local minimum corresponding to the stable keto form in S0 state but exists in S1 state, and there is only a small barrier (1.99 kcal/mol) from enol to keto form in S1 state. It is clear that the proton transfer of TAPHP can only occur in S1 state. In addition, the potential barrier (0.44 kcal/mol) needed for reverse proton transfer of TAPHP-Keto* to form TAPHP-Enol* is extremely small. Thus, TAPHP-Keto* and TAPHP-Enol* can coexist in S1 state. But the energy of TAPHP-Enol* is less than the energy of TAPHP-Keto*, which means that TAPHP-Enol* is most stable in S1 state. Thus, the dynamic process of TAPHP can be expressed as follows: TAPHP-Enol is firstly excited to S1 state (TAPHP-Enol*) by absorbing photons. Because of the small potential barrier, TAPHP-Enol* will form the TAPHP-Keto* by proton transfer. Then reverse proton transfer of TAPHP-Keto* can easily occur due to the small reverse barrier and form the TAPHP-Enol* isomer. Finally, the TAPHP-Enol* performs a radiation transition back to the S0 state (TAPHP-Enol). This loop process reasonably explains why the emission peak of TAPHP's keto form was not observed in the experiment. However, TAPDP is quite different. It can be seen that the PECs have four stable points both in S0 and S1 states, namely, TAPDP–Enol, TAPDP-Keto, TAPDP-Enol*, and TAPDP-Keto*. The barrier to be crossed in S0 state is 10.41 kcal/mol, while the barrier in S1 state is only 5.37 kcal/mol. Obviously, ESIPT of TAPDP is easier to occur than GSIPT. The barriers that TAPDP needs to cross for reverse proton transfer in S0 and S1 states are 0.66 and 4.98 kcal/mol, respectively. Small barriers also make it possible for TAPDP to perform reverse proton transfer. The circulation system of TAPDP undergoing proton transfer should be: TAPDP-Enol→TAPDP-Enol*→TAPDP-Keto*→TAPDP-Keto →TAPDP-Enol.</p><!><p>PECs and associated barriers of (A) TAPHP and (B) TAPDP.</p><!><p>It is known that hydrazine is more nucleophilic than hydroxylamine. The more positive the atomic charge, the easier it is to attract nucleophiles to react. It is therefore possible to verify the reaction site by calculating the atomic charge. The Hirshfeld atomic charge of the probe TAPHP is shown atomically colored in Figure 8. The size of the atomic charge increases in the order of blue, white and red. It can be seen that the atomic charge of C1 is the most positive, so C1 should be the site that most easily attracts the nucleophilic amino group. While the other amino group of hydrazine should react with C2 with an atomic charge close to 0. This is consistent with the reaction sites predicted experimentally. In order to verify the preference of the probe TAPHP for hydrazine, we calculated the binding energy of TAPHP for hydrazine and hydroxylamine, which are 4.59 and 3.56 kcal/mol, respectively. Therefore, the larger binding energy is one of the reasons why TAPHP shows excellent selectivity to hydrazine.</p><!><p>Hirshfeld atomic charge plotted in atomic coloring, the atomic charge corresponds to the blue-white-red color change from minimum to maximum.</p><!><p>The energy changes in biochemical reactions that occur during biological oxidation can be described by thermodynamic free energy changes. As shown in Figure 9, the free energy of the products is −15.87 kcal/mol relative to the free energy of the reactants. The negative value of ΔG indicates that the reaction between TAPHP and hydrazine is an exothermic reaction, which can proceed spontaneously.</p><!><p>Free energy of reactants (TAPHP+N2H4) and products (TAPDP+H2O).</p><!><p>The 1H NMR spectra of TAPHP and TAPDP were obtained with tetramethylsilane (TMS) as reference material. As shown in Figure 10, the magnetic shield values of the two proton signals (Ha1 and Hb1) on the TAPHP were 23.18 and 23.53 ppm, and the NMR chemical shifts are 8.32 and 7.98 ppm, taking TMS as the standard. However, the peaks of TAPDP's signal at 23.18 and 23.53 ppm completely disappeared, and three new proton signals appeared, 26.68(4.82), 28.30(3.21), and 27.97(3.53) ppm, respectively. The data we obtained is consistent with the 1H NMR spectra measured experimentally. As a result, the configuration of the product TAPDP is affirmed. TAPHP's detecting mechanism for hydrazine are confirmed again.</p><!><p>The 1H NMR spectra of TAPHP and TAPDP.</p><!><p>The ESIPT and ICT properties of the TAPHP and the TAPDP are researched using DFT and TDDFT methods. Analysis of the PECs can conclude that both TAPHP and TAPDP can have ESIPT instead of GSIPT. The bond parameters and IR vibrational spectra confirm the enhancement mechanism of hydrogen bond in the S1 state. The same conclusion is obtained by using the visualized isosurface maps equipped with IGM. Meanwhile, topological analysis based on AIM theory and CVB index are also applied to characterize the strength of hydrogen bonds. Hole-electron analysis suggests that both TAPHP and TAPDP undergo charge excitation, rather than only TAPHP as described in experimental literature, that is, ICT is inevitable. In addition, the calculated electronic spectra coincide with the experimental results. The fluorescence of TAPDP at 503 nm was not observed experimentally, which could be considered as the result of its weak intensity. Negative free energy difference implies a spontaneous exothermic reaction from TAPHP to TAPDP. Furthermore, TAPHP can indeed recognize hydrazine specifically by the movement of the fluorescence peaks. TAPHP's sensing mechanism for hydrazine is also characterized by atomic charge and 1H-NMR spectra.</p><!><p>The datasets analyzed in this manuscript are not publicly available. Requests to access the datasets should be directed to yzsong@sdnu.edu.cn.</p><!><p>SL, JL, and QL carried out the ab initio calculation. JF and LL analyzed the results. SL wrote the manuscript. CW and YS supervised this project.</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. This work was supported by National Natural Science Foundation of China (Grant Nos. 11874241 and 11874242), Shandong Province Higher Educational Science and Technology Program (Grant No. J15LJ03), Taishan scholar project of Shandong Province, China Post-doctoral Foundation (No. 2018M630796), and Natural Science Foundation of Shandong Province (No. ZR2018BA034).</p>

PubMed Open Access

Multiplexed Method to Calibrate and Quantitate Fluorescence Signal for Allergen-Specific IgE

Using a microarray platform for allergy diagnosis allows for testing of specific IgE sensitivity to a multitude of allergens, while requiring only small volumes of serum. However, variation of probe immobilization on microarrays hinders the ability to make quantitative, assertive, and statistically relevant conclusions necessary in immunodiagnostics. To address this problem, we have developed a calibrated, inexpensive, multiplexed, and rapid protein microarray method that directly correlates surface probe density to captured labeled secondary antibody in clinical samples. We have identified three major technological advantages of our calibrated fluorescence enhancement (CaFE) technique: (i) a significant increase in fluorescence emission over a broad range of fluorophores on a layered substrate optimized specifically for fluorescence; (ii) a method to perform label-free quantification of the probes in each spot while maintaining fluorescence enhancement for a particular fluorophore; and (iii) a calibrated, quantitative technique that combines fluorescence and label-free modalities to accurately measure probe density and bound target for a variety of antibody\xe2\x80\x93antigen pairs. In this paper, we establish the effectiveness of the CaFE method by presenting the strong linear dependence of the amount of bound protein to the resulting fluorescence signal of secondary antibody for IgG, \xce\xb2-lactoglobulin, and allergen-specific IgEs to Ara h 1 (peanut major allergen) and Phl p 1 (timothy grass major allergen) in human serum.

multiplexed_method_to_calibrate_and_quantitate_fluorescence_signal_for_allergen-specific_ige

<!>Platform Design Simulations<!>Reagents and Equipment<!>Si/SiO2 Coating by Copoly(DMA-NAS-MAPS)<!>Proof of Concept of CaFE Using IgG and \xce\xb2-lactoglobulin<!>IgG Calibration<!>\xce\xb2-Lactoglobulin Calibration<!>CaFE Implemented as an Allergy Testing Platform<!>Platform Design Simulations<!>Proof of Concept of Quantification and Calibration of CaFE using IgG and \xce\xb2-Lactoglobulin<!>CaFE Implemented as an Allergy Testing Platform<!>CONCLUSIONS

<p>Allergy, a disorder of the immune system characterized by a maladaptive immune response to otherwise harmless environmental antigens ("allergens"), affects nearly 50 million people in the U.S. with total estimated annual costs at nearly $7B.1,2 The recent development of protein microarrays and the availability of recombinant allergens over the past decade have led to microarray- based allergy immunoassays testing for specific IgE in patient serum. The main advantages of these microarray in vitro diagnostics resides in the inherent capability to quantify allergen-specific IgE using only 10–100 μL of serum per test and to better characterize allergen sensitization by measuring specific IgE to the component major allergens of a crude allergen extract (component-resolved diagnostics, CRD).3 However, two specific shortcomings exist with this technology. First, the typical microarray chips utilize assays with probes placed directly on a simple glass slide. In this configuration, the presence of a high index solid substrate (glass) in the immediate vicinity of the fluorophores reduces the fluorescence yield. Simple layered structures offer an inexpensive alternative to overcome the limitations imposed by glass slides and provide significant signal enhancement.4 Second, the more important limitation is related to the difficulty in obtaining quantitative results in conventional fluorescence based microarray tests. This difficulty arises from the variability in the amount of immobilized allergens that affects specific IgE capture and quantitation.3,5,6 As a result, fluorescence detection on typical glass (SiO2) slides, the "gold-standard" technique used in microarrays, has limited sensitivity and may yield inaccurate results. These inadequacies can cause under-estimation or failure of detection for captured targets7 and concurrently yield unreliable clinical results.8–10</p><p>Variability in microarray technology in general has become an essential concern in producing reliable data not only due to the technical variation, such as array printing, sample processing, analytes, plate, or person, but also due to the inherent nature of proteins themselves.11–13 Label-based procedures have been developed to account for this variation in probe deposition and binding to the surface in order to visualize the printed slides prior to experimentation.14,15 Although these techniques verify the presence of uniformly bound probe, they may negatively affect the activity of the probe, fail to quantify the amount of bound probe on surface, and alter physiochemical properties. Recently, an approach that utilizes a photonic crystal biosensor surface and a high resolution label-free imaging detection instrument to formulate prehybridization images of spotted nucleic acid array was recently reported as a sensitive method of quality control.16 Aside from being demonstrated only for DNA microarrays, this quality control method merely bins the spot as being suitable or unsuitable for analysis and does not offer the quantified amount of bound probe relative to secondary antibody (i.e., fluorescence). Although a variety of techniques have attempted to advance quality assurance of microarray technologies, a need for quantitative assessment providing calibrated microarray measurements still remains.</p><p>To address these issues, we have integrated our label-free technology, the interferometric reflectance imaging sensor (IRIS), a quantitative, high-throughput, simple, robust, and versatile technology used for multiplexed detection of DNA and proteins with high sensitivity comparable to surface plasmon resonance (SPR),17–20 with a new enhanced fluorescence technology to develop the calibrated fluorescence enhancement (CaFE) method.21 By combining the sensitivity of fluorescence with the quantitative nature of IRIS, the CaFE method addresses microarray reproducibility issues by (1) quantifying the probe amount with IRIS, (2) measuring the enhanced fluorescence signal generated by labeled secondary antibodies, and (3) calibrating the fluorescence signal utilizing the quantitative assessment of the spots by IRIS. While this technique is broadly applicable to a variety of ligand-analyte based microarray platforms, it is particularly effective for allergy chips. Detection of allergen-specific IgE molecules necessitates the use of secondary antibodies to distinguish them from the large amount of physiologic allergen-specific IgG molecules that bind to the probe but are not indicative of allergic sensitization. The additional quantification challenge imposed by the large variability of immobilized probe density makes in vitro allergy diagnostics a perfect candidate for the demonstration of CaFE technology proof-of-concept.</p><p>A variety of layered structure designs could be used to achieve fluorescence enhancement for DNA and protein microarray applications.22–26 In this paper, we choose to implement a simple oxide on Si structure due to its low-cost and well-established characteristics. Our primary design parameters include wavelength range for fluorescence enhancement and label-free detection accuracy. We have designed two types of sample platforms for enhanced fluorescence and label-free protein sensing. For design optimization we used the dipole emission model27,28 to simulate fluorophore emitters near a dielectric interface on a layered structure. Details of modeling for the label-free sensing by IRIS technology can be found in Daaboul et al.19 Platform-1 consists of two areas of SiO2, one optimized specifically for label-free and the other optimized for fluorescence sensing (500 nm for IRIS and 100 nm for enhanced fluorescence19,21,29), and has been designed to enhance a broad emission range of fluorophore wavelengths. Recently, this combined chip was shown to be of high practical use during the assay development process. Label-free sensing was utilized to image arrays prior to incubation with labeled antibodies to assess the robustness of the array and to quantify the amount of immobilized probes. Via this method, fluorescence measurements on the same chip yielded calibrated bioassay results in a single experiment.21 However, this platform relies on uniformity and repeatability within a chip since label-free and fluorescence measurements are performed on two separate parts of the same chip. Platform-2 consists of a silicon chip with a single oxide thickness optimized for single spot enhancement of a particular fluorophore(s) and label-free analysis. In this new configuration, each spot is measured with both fluorescence and label-free modalities, and hence calibrated, effectively accounting for spot-to-spot variation commonly reported in microarrays. For both platforms, regardless of having single or multiple regions, the top layer is oxide and thus chemical surface preparation is identical. This feature allows the use of specialized polymeric coatings (copoly(DMA-NAS-MAPS)30,31 to covalently link capture agents to the surface, while maintaining high functionality and preventing nonspecific binding. Therefore, the self-calibrated CaFE platform offers an opportunity for quantitative assessment of allergy chips which uses labeled secondary antibodies to detect captured IgE by integrating both label-free and fluorescence measurements.</p><!><p>Structure optimization is defined as the oxide thickness on Si that yields intensity enhancement across fluorescence emission wavelengths of interest and performs label-free sensing with high accuracy. These simulations address two goals regarding oxide thickness design and optimization: (1) platform-1 (called CaFE chip) should yield emission enhancement for maximum coverage of wavelengths while performing label-free sensing on a separate spot on the same chip and (2) platform-2 maintains emission enhancement for a limited wavelength range (covering 1 or 2 fluorophores) while allowing for accurate calibration of the same spot using IRIS. Platform design for Cy3 and Cy5 emission enhancement was chosen due to their extensive use in microarrays to study the ratio of expression of genes from two sources.32–36</p><p>To investigate the effect of oxide thickness on fluorescence intensity, the dipole emission model27,28 was used to simulate fluorophore emitters near a dielectric interface on a layered SiO2/Si structure. Simulations were carried out for oxide thicknesses ranging from 1 to 1000 nm over the entire visible wavelength range to select the optimized oxide thickness for each of the two platforms described above. In Figure 1a, we plot the wavelength dependence of fluorescence intensity for two specific oxide thicknesses in comparison to a glass substrate along with typical spectra of fluorophores that are commonly used as labels in bioassays (Alexa Fluor 359, Alexa Fluor 488, Cy3 dye, Cy5 dye, and Alexa Fluor 647). The initial optimization is performed assuming a collection angle of 0.7. For a practical fluorescence collection system, the numerical aperture (NA) is large (typical values around 0.5–0.7) (Herman, B., Fluorescence Microscopy, Second Edition, BIOS Scientific Publishers (1998)) and hence the theoretical enhancement calculations should consider the dependence on NA. In Figure 1b, the NA dependence of emission enhancement of Cy3 is plotted and the two optimized designs are compared at typical NA values.</p><p>The CaFE verification of fluorescence measurement is described in the Supporting Information, and details on label-free detection using IRIS have been described.19 As previously reported, 500 nm oxide on Si is optimal for IRIS sensing because the spectra of three of the LEDs (455, 525, and 632 nm) sample the linear region of the curve while the yellow LED (598 nm) helps determine the amplitude of the curve over one period. Because the linear region falls on the inflection point of the curve, any small change in height, or biomass accumulation, will be detected. To check for label-free sensing capability for the oxide thickness chosen for emission enhancement of Cy3/Cy5 fluorophores (goal-2), we generated an additional reflectivity curve at this oxide thickness as a function of illumination wavelengths and compared it to the curve produced from 500 nm oxide (Figure 1c).</p><p>Finally, to verify the results of the simulations with experimental data, platform-1 (CaFE) chips and platform-2 (320 nm) chips were fabricated as seen in Cretich et al.21 An experiment described in IgG Calibration was performed on each platform to validate the accuracy of label-free detection and enhancement of fluorescence (Figure 1d).</p><!><p>TRIS, BSA, Tween 20, PBS tablets, rabbit immunoglobulin G, carbonic anhydrase, and bovine serum albumin were purchased from Sigma (St. Louis, MO). Rabbit anti-β-lactoglobulin was purchased from Bethyl Laboratories (Montgomery, TX), goat anti-α-lactalbumin from GeneTex Inc. (Irvine, CA), and AbCam (Cambridge, U.K.). Secondary antibodies (Cy3-labeled goat antimouse IgG and mouse antigoat IgG) were purchased from Jackson ImmunoResearch (West Grove, PA), and anti-IgE was purchased from BD Bioscences. Allergens Bet v 1a (Bet v 1.0101), Phl p 1 (Phl p1.0101), Phl p 5 (Phl p 5.0101), and Alt a 1 (Alt a 1.0101) were recombinant allergens from Biomay, (Vienna, Austria) and allergens Bet v 2 (Bet v 2.0101), Phl p 7, nDer p 1, nDer p 2, and nFel d 1 were recombinant (or native, when prefix "n" is used) allergens from Indoor Biotechnologies Ltd. (Warminster, U.K.).</p><!><p>Si/SiO2 slides were immersed for 30 min in a solution of copoly(DMA-NAS-MAPS) at 1% w/v concentration in a solution of deionized water and 20% saturated ammonium sulfate. Slides were washed with water and dried at 80 °C for 15 min. This polymeric coating was chosen for our CaFE chips due the feasibility and reproducibility of its synthesis and coating process. This particular polymeric coating does not change the optical properties of the setup.</p><!><p>To demonstrate the universal application of the CaFE method, we have performed the quantification and correlation of captured IgG and β-lactoglobulin probe to fluorescence signal of Cy3-labeled-secondary antibody. For these experiments, we utilize the CaFE chip platform to optimize for all fluorophores and achieve maximum sensitivity on the IRIS device. IgG and β-lactoglobulin proteins were chosen due to their well-established spotting protocols and reliable spotting morphologies.</p><!><p>As proof of concept, 20 replicates IgG of varying concentrations (0.015, 0.03, 0.063, 0.125, 0.25, and 1 mg/mL) were spotted onto 2 CaFE chips. After overnight humid chamber incubation, the chips were washed with 50 mM ethanolamine in TRIS/HCl 1MpH 9 for 1 h, rinsed with water, dried with a stream of argon gas, and then measured using IRIS. They were then incubated with 100 μL of specific labeled antibody in incubation buffer (Tris/HCl 0.05 M pH 7.6, NaCl 0.15 M, Tween 20 0.02%) with 1% w/v BSA for 1 h at 1 μg/mL. Another IRIS measurement was taken after washing with PBS for 10 min, rinsing with water, and drying with argon.</p><p>Fluorescence evaluation was performed by a fluorescence scanner using 40% PMT and 33% laser power for maximal fluorescence value without saturation. Mean fluorescence intensity and standard error from all 20 spots is depicted.</p><!><p>To model a sandwich assay similar to the allergen immunoassay and to demonstrate the versatility of the CaFE method, 20 replicates of β-lactoglobulin of varying concentrations (0.015, 0.03, 0.063, 0.125, 0.25, and 1 mg/mL) were spotted onto three CaFE chips, and after overnight humid chamber incubation, the CaFE chips were washed with 50 mM ethanolamine in TRIS/HCl 1 M pH 9 for 1 h, washed with water, dried with a stream of argon gas, and then measured using IRIS. The chips were then incubated with 100 μL of specific antibody in incubation buffer with 1% w/v BSA, for 2 h at 10 ng/mL. Slides were then washed with washing buffer (Tris/HCl 0.05 M pH 9, NaCl 0.25 M, Tween20 0.05%) for 10 min, rinsed with water, and dried with argon gas. The chips were then incubated with 100 μL of the solution of the specific labeled secondary antibody 1 μg/mL in incubation buffer for 1 h. Slides were then washed with PBS for 10 min, rinsed with water, dried with argon, and measured with IRIS.</p><p>Fluorescence evaluation was performed using 90% PMT and 90% laser power for maximal fluorescence value just below saturation. Mean fluorescence intensity and standard error were plotted for the 20 replicate spots.</p><!><p>To evaluate CaFE as a clinical diagnostic platform, two major allergens (peanut (Ara h1) amd timothy grass (Phl p1) were spotted in replicates of three at four concentrations (0.25, 0.5, 0.75, and 1.0 mg/mL) on two CaFE chips. In addition, PBS and IgG were spotted as negative and positive control parameters. After overnight humid chamber incubation, the chips were washed with 50 mM ethanolamine in TRIS/HCl 1 M pH 9 for 1 h, washed with water, dried with a stream of argon gas, and then measured using IRIS. The chips were then incubated with 100 μL of patient serum with documented allergy to peanut (specific IgE 19.40 kU/L, Phadia ImmunoCAP) and timothy grass (positive allergen skin prick test) in incubation buffer with 1% w/v BSA for 2 h at 10 μg/mL. (Subject recruitment was approved by the Boston University Medical Campus Institutional Review Board, Protocol H-29428.) Slides were then washed with washing buffer for 10 min, rinsed with water, and dried with argon gas. After incubating with 1 ng/mL anti-IgE labeled with Cy3, the chips were washed with PBS (10 min), rinsed with water, dried with argon, and measured with IRIS.</p><p>Scanning for fluorescence evaluation was performed. CaFE slides were analyzed using 90% PMT and laser power to maximize the signal-to-noise ratio without pixel saturation. Mean fluorescence intensity is depicted.</p><!><p>After running the radiation model of emitters near a dielectric interface, a 100 nm oxide thickness yielded emission enhancement for maximum coverage of wavelengths and a 320 nm oxide thickness yielded enhancement for emission wavelengths of Cy3 and Cy5 fluorophores (Figure 1a). This enhancement effect is a result from the interference of the emission wavelength reflected at the oxide-silicon interface and the air-oxide interface. The oxide thickness serves as a spacer in which emitted light is reflected at the air-oxide interface and refracted through the oxide layer. The refracted light travels through the spacer layer and is then reflected by the silicon layer. Depending on the wavelength of the emitted light, the thickness of the spacer will determine which wavelengths undergo constructive or destructive interference. Using this method, we designed two layered structures: (1) a 100 nm SiO2 layer on Si that yields enhancement for maximum coverage of wavelengths and (2) a 320 nm SiO2 layer on Si that maintains fluorescence enhancement for Cy3 and Cy5 fluorophores.</p><p>In Figure 1b, the NA dependence of emission enhancement of Cy3 is graphed and the two optimized designs are compared. At low NA, a 100 nm oxide thickness yields a greater enhancement effect compared to a 320 nm oxide layer. As the collection angle increases, the enhancement effect decreases because higher angles do not undergo complete constructive interference due to the increase in optical path length through the spacer region. The enhancement values converge as a greater percentage of the reflected emission is collected. When comparing the platforms at the typical NA used in fluorescence scanners (NA = 0.7), both platforms yield a 2-fold emission enhancement of Cy3. Furthermore, the enhancement produced by 100 nm SiO2 is only 1.04 times greater than the enhancement yielded by 320 nm SiO2. On the basis of these results, designing a chip with 100 nm oxide on Si would enhance all wavelengths of commonly used fluorophores and constructing a chip with 320 nm oxide on Si would maintain enhancement for Cy3 and Cy5 fluorophores.</p><p>In Figure 1c, normalized reflectivity curves as a function of illumination wavelength are generated for both platforms and compared. Because the rate of change with wavelength is slightly slower in the 320 nm oxide compared to the 500 nm oxide, the illumination wavelengths intersect the curves at slightly different locations. As a result, the spectra of the LEDs now sample the 320 nm reflectivity curve close to the minima and maxima regions. Although none of the LEDs fall directly on the inflection point of the graph, the green and yellow LED sample above and below the linear region, helping to define the slope of the curve and determine the reflectivity profile of a 320 nm oxide on Si. These reflectivity comparisons suggest that IRIS measurements on a 320 nm oxide are feasible; however, a reduction in level of detection and sensitivity may occur.</p><p>Fluorescence simulations and IRIS reflectivity curves facilitated the design of two platforms: (1) a CaFE chip with one area of 100 nm oxide on Si for fluorescence enhancement coverage over all visible wavelengths and an area of 500 nm for sensitive label-free biosensing and (2) a 320 nm oxide on Si chip that maintains fluorescence enhancement of Cy3 and Cy5 fluorophores and operates with IRIS. These designs were tested and compared by conducting the IgG calibration experiments on each platform (Figure 1d). As expected, the CaFE chip yields a higher fluorescence signal and slightly better sensitivity compared to the 320 nm oxide chip. Standard deviation is also higher on the 320 nm oxide chip, and the measurement was not able to detect the lowest spotting concentration (0.015 mg/mL). On the basis of this data, the radiation model of emitters and IRIS reflectivity curves helped design two dual modality platforms that optimize for enhanced fluorescence of one or more wavelengths and quantify biomass accumulation, label-free.</p><!><p>Images of both label-free measurements (Figure 2a) and fluorescence measurements (Figure 2b) show a gradient that correlates with varying concentration of immobilized rabbit IgG and captured Cy3-anti-rabbit IgG. The fluorescence enhancement of the 100 nm oxide islands is noticeable when compared to the 500 nm oxide island. Similar effects are seen in the β-lactoglobulin array but are not shown.</p><p>Theoretically, each chip should bind the same probe density on the surface. However, IRIS measurements show that both IgG and β-lactoglobulin showed a large immobilization variation between chips resulting in varied fluorescence measurement of secondary antibody. Quadrant-to-quadrant variation on chip was also present in the IgG data. The CaFE method was then applied to fluorescence and IRIS data in order to quantify and calibrate specific secondary antibody for both IgG and β-lactoglobulin systems (Figure 2c,d). In these plots, all error bars correspond to spot-to-spot variability, while each subset shows chip-to-chip variability. In both cases, each of the proteins demonstrates a calibrated, strong linear response (R2 ≥ 90%) between fluorescence and probe density despite chip-to-chip variability. Because each protein has different finite binding capacities, the slopes of the CaFE curves differ. This capacity also may result in possible saturation point at higher concentrations. Overall, these results suggest that the CaFE method will be an effective and versatile platform to quantify and correlate bound probe to secondary labeled antibody and has potential application in immunodiagnostics. In addition, on-chip variation could be addressed by implementing platform-2, which would assess each spot individually.</p><!><p>The performance of the two chips, defined by fluorescence signal of secondary IgE, varies between the allergens because different levels of IgE in serum for each particular allergen are present. This is confirmed with the ImmunoCAP results (Figure 2-S in the Supporting Information). When analyzing the fluorescence signal of anti-IgE in the fashion of typical ELISA and microarray assays, measurement variation of allergen-specific IgE of the same allergen between chips is present, particularly in Phl p1 allergen (Figure 3a). The degree of chip-to-chip variation between allergens most likely occurs due to the physiochemical properties of the allergens themselves (i.e., affinity to immobilize to the surface) and technical variation (i.e., spotting). While only slight chip-to-chip fluorescence variation is seen for Ara h1 allergen (R2 of 0.88), a significant chip-to-chip fluorescence variation is seen for Phl p1 allergen (R2 of 0.24), despite the same conditions, reagents, and serum samples used in this single experiment. On the basis of the fluorescence data alone, it is unknown as to why Phl p1 allergen chips yield different fluorescence responses.</p><p>To account for this deviation, the CaFE method is applied to collected fluorescence and IRIS data. As a result, a calibrated, linear response between allergen-specific IgE and amount of allergen immobilized on surface emerges (Figure 3b). In both Ara h1 and Phl p1 examples, the R2 value increases to at least 90%, demonstrating the higher degree of correlation between fluorescence signal and immobilization density compared to spotting concentration. Although the CaFE method only slightly improves upon Ara h1 allergen data, the effects of including label-free IRIS measurement are dramatically seen in the Phl p1 data. Most importantly, use of the CaFE method clarifies that higher immobilization density of Phl p1 allergen on chip 2 results in higher fluorescence signal, indicating that any variation in immobilization density will affect the amount of IgE captured. In accordance with the literature, this data supports that high variation in allergen immobilization microarrays is a concerning issue. The strong linear correlation between fluorescence and immobilization density demonstrates the value of the CaFE method as an opportunity for calibrated quantitative assessment of serum allergen-specific IgE and should improve accuracy in predicting clinical reactivity in susceptible individuals.</p><p>Finally, in order to verify that the CaFE method differentiates between allergic and nonallergic samples, five allergens were arrayed in replicates of three onto a CaFE chip. This chip was incubated with characterized serum in which there was selective allergen-specific IgE. Results were compared to ImmunoCAP or skin test characterization (Figure 2-S in the Supporting Information).</p><!><p>We have designed two platforms: (1) a chip with multiple sections to optimize label-free measurement and fluorescence over a broad range of wavelengths and (2) a silicon chip that has been engineered specifically for the enhancement of a particular fluorophore and is operational in the label-free modality and allows for calibration of the same spot, to create a calibrated fluorescence enhancement, or CaFE, method that improves upon quality and quantity control in microarrays. In our experiments, we utilize one of the two platforms, the CaFE chip, to demonstrate that the correlation between the fluorescence signal and immobilized probe density is linear despite chip-to-chip variance. We also show that these experiments are repeatable on a single oxide thickness platform. Without the ability to quantify the amount of capture probes (major allergens), the typical assay quality control would be limited to controlling the spotting conditions, i.e., the intended concentration of the probes. Thus, the CaFE method is an effective, self-calibrated, multiplexed, and sensitive platform and offers technological advances that are not currently or readily available.</p><p>Additional applications of this technology are by no means limited to the described model of allergy diagnosis; this platform could be effectively applied where readout from a labeled secondary antibody must be calibrated against quantity of immobilized capture probe for accurate diagnosis.</p>

PubMed Author Manuscript

Exceptionally long-lived light-emitting electrochemical cells: multiple intra-cation π-stacking interactions in [Ir(C^N)₂(N^N)][PF₆] emitters

A series of cyclometalated iridium(III) complexes [Ir(C^N) 2 (N^N)][PF 6 ] (N^N ¼ 2,2 0 -bipyridine (1), 6-phenyl-2,2 0 -bipyridine (2), 4,4 0 -di-tert-butyl-2,2 0 -bipyridine (3), 4,4 0 -di-tert-butyl-6-phenyl-2,2 0 -bipyridine (4); HC^N ¼ 2-(3-phenyl)phenylpyridine (HPhppy) or 2-(3,5-diphenyl)phenylpyridine (HPh 2 ppy)) are reported. They have been synthesized using solvento precursors so as to avoid the use of chloridodimer intermediates, chloride ion contaminant being detrimental to the performance of [Ir(C^N) 2 (N^N)]-[PF 6 ] emitters in light-electrochemical cell (LEC) devices. Single crystal structure determinations and variable temperature solution 1 H NMR spectroscopic data confirm that the pendant phenyl domains engage in multiple face-to-face p-interactions within the coordination sphere of the iridium(III) centre. The series of [Ir(Phppy) 2 (N^N)] + and [Ir(Ph 2 ppy) 2 (N^N)] + complexes investigated include those with and without intra-cation face-to-face p-stacking. All the complexes display excellent luminescent properties, in particular when employed in thin solid films. The most important observation is that all the LECs using the [Ir(Phppy) 2 (N^N)] + and [Ir(Ph 2 ppy) 2 (N^N)] + emitters (i.e. with and without intra-cation p-stacking interactions) exhibit very stable luminance outputs over time, even when driven at elevated current densities. The most stable LEC had an extrapolated lifetime of more than 2500 hours under accelerated testing conditions.

exceptionally_long-lived_light-emitting_electrochemical_cells:_multiple_intra-cation_π-stacking_inte

Introduction<!>Experimental<!>Solvento-precursors [Ir(C^N) 2 (MeOH) 2 ][PF 6 ]<!>Electrochemical behaviour of [Ir(C^N) 2 (N^N)][PF 6 ] complexes<!>Photophysical properties of [Ir(C^N) 2 (N^N)][PF 6 ] complexes<!>LEC performances and electroluminescence<!>Conclusions

<p>Light-emitting electrochemical cells (LECs) are a class of lightemitting devices in which the active material is a charged species (an ionic transition-metal complex, iTMC-LECs, or a polymer, PLECs). 1 LECs operate in a unique fashion: aer application of a bias, the charged species in the active layer move towards the electrodes, accumulating at the interfaces and causing a sharp drop of potential near the electrode interfaces with the consequent formation of doped zones. In this situation emission of light takes place at the intrinsic region. 2 As a consequence of this behaviour, it is not necessary for LECs to incorporate a low work-function metal, because the injection barrier of the charges is reduced by the formation of an electric double layer. Air-stable electrodes such as Al can be used, negating the need for rigorous encapsulation of the device as is essential for organic light-emitting diodes (OLEDs). Compared to OLEDs, LECs possess a simpler architecture allowing them to be prepared by solution processes.</p><p>The rst report of an iTMC-LEC was by Maness et al. 3 and utilized a ruthenium(II)-containing complex as the single component in the active layer. The emission band of ruthenium(II) complexes such as those based on [Ru(bpy) 3 ] 2+ is centred in the orange-red region, and this limits the emission colours that can be achieved with this class of compound. Even more problematic is the low stability of these materials under device conditions. By changing from a second to a third row transition metal (e.g. iridium) in the iTMC, it is possible to improve the stability of the device and achieve higher ligand-eld splitting energies leading to higher colour tunability. [4][5][6][7] The use of iTMCs containing ligands with substituents that are capable of intra-cation face-to-face p-interactions can stabilize the complex in the excited state and consequently enhance the lifetime of the LEC device. This strategy has been used to produce LECs with lifetimes of thousands of hours. [8][9][10][11][12] The archetype member of the family is [Ir(ppy) 2 (bpy)] + (Hppy ¼ 2-phenylpyridine, bpy ¼ 2,2 0 -bipyridine). Within the octahedral sphere, a 6-phenyl substituent introduced into the bpy unit is perfectly positioned to stack over the phenyl ring of the cyclometalated ppy domain. The interaction is present in both the ground and excited states of the complex, stabilizing it with respect to attack at the metal centre by nucleophiles such as H 2 O. The phenomenon was initially established with phenyl/phenyl p-interactions, [8][9][10][11] but is also effective for other aryls, e.g. phenyl/pyrazolyl 12,13 and phenyl/pyridyl contacts. 14 Surprisingly, replacing 6-phenyl-2,2 0 -bipyridine (6-Phbpy) by 6,6 0 -diphenyl-2,2 0 -bipyridine (6,6 0 -Ph 2 bpy) does not result in additional enhancement of LEC device lifetimes on going from [Ir(ppy) 2 (6-Phbpy)] + to [Ir(ppy) 2 (6,6 0 -Ph 2 bpy)] + . 8 We now report a series of new [Ir(C^N) 2 (N^N)] + complexes in which both the C^N and N^N domains contain pendant phenyl substituents and demonstrate the effects of differing degrees of p-stacking interactions in the coordination sphere of the iridium(III) centre on the emission behaviours and LEC device characteristics. In addition, we use a solvento-iridium(III) precursor to circumvent the detrimental effects associated with chlorido-impurities. 15 When used as the primary active component, these complexes lead to LECs with exceptional stabilities.</p><!><p>All experimental details including crystallographic data and device preparation are given in the ESI. †</p><!><p>The conventional method for preparing [Ir(C^N) 2 (N^N)][PF 6 ] compounds is to treat the chlorido dimer [Ir 2 (C^N) 4 Cl 2 ] with two equivalents of an N^N chelating ligand, followed by anion exchange by addition of NH 4 PF 6 (Scheme 1, le). 16 However, even small amounts of residual chloride ion in the nal product result in signicant reductions in the performance of the iTMC in LECs. 15 ). The coordinated solvent must be sufficiently labile to allow displacement with an N^N ligand in the nal step. Use of AgPF 6 results in precipitation of chloride ion as AgCl, and we have previously shown that this is a reliable means of removing residual Cl À from [Ir(ppy) 2 (bpy)][PF 6 ] to give material with optimal LEC performance. 15 The C^N and N^N ligands used in this study are shown in Scheme 2, and the single crystal structure of HPh 2 ppy is described in the ESI (Fig. S1 † and 13 C NMR spectra of CD 3 OD solutions of the complexes were consistent with the formulations. Most importantly, for each compound, a singlet at d 3.35 ppm in the 1 H NMR spectrum correlating in the HMQC spectrum with a signal at d 49.9 ppm was assigned to the coordinated MeOH; the proton resonance was distinct from the multiplet arising from residual bulk CD 2 HOD (Fig. S2 †). Attempts to obtain electrospray mass spectrometric evidence for the [Ir(Phppy) 2 (MeOH) 2 ] + or [Ir(Ph 2 ppy) 2 (MeOH) 2 ] + ions were not successful, presumably because of the lability of the methanol molecules. In the spectrum of [Ir(Ph 2 ppy) 2 (MeOH) 2 ][PF 6 ], a peak envelope at m/z 805.5 corresponding to [Ir(Ph 2 ppy) 2 ] + was observed; the isotope pattern matched that calculated. The solvento-complexes were used in the subsequent steps as soon aer synthesis as possible. [Ir(Phppy) 2 ( 1)][PF 6 ] and [Ir(Ph 2 ppy) 2 ( 1)][PF 6 ]$EtOH crystallize in the monoclinic space group P2 1 /n and orthorhombic space group Pna2 1 , respectively, each with one cation in the asymmetric unit (Fig. 1a and 2). The octahedral iridium(III) trischelates are chiral and in both structures, the L and D-enantiomers are present in the lattice. The bpy unit in 1 is slightly twisted in [Ir(Phppy) 2 (1)] + (angle between the bpy ring planes ¼ 13.1 ) but is close to planar in [Ir(Ph 2 ppy) 2 (1)] + (angle ¼ 6.3 ). In [Ir(Phppy) 2 (1)] + , both ppy units are close to planar (angles between the planes of rings containing N4/C34 and N3/C17 ¼ 5. and 61.8 , respectively, and these large twist angles are associated with face-to-face p-stacking of these rings over the [Ph 2 ppy] À pyridine rings containing N3 and N4 (Fig. 3). The pinteraction between the rings containing N4 and C23 is 1)][PF 6 ]$EtOH (H atoms omitted, ellipsoids plotted at 40% probability). Selected bond parameters: characterized by an angle between ring planes of 9.9 , phenyl ring plane/centroid of pyridine ring distance of 3.27 Å, and centroid/centroid separation of 3.48 Å. The corresponding parameters for the p-stacking of rings with N3 and C46 are 18.6 , 3.37 Å and 3.51 Å. The cations in [Ir(Phppy) 2 ( 1)][PF 6 ] are closely associated through embraces of the arene domains (Fig. 1b) leading to assembly of anion-separated columns running along the b-axis.</p><p>[Ir(Ph 2 ppy) 2 ( 2)][PF 6 ]$2C 6 H 5 Me crystallizes in the triclinic space group P 1, and Fig. S3 † shows the L-enantiomer of [Ir(Ph 2 ppy) 2 (2)] + ; both enantiomers are present in the lattice. Although the three chelate angles in [Ir(Ph 2 ppy) 2 (2)] + are comparable with those in [Ir(Phppy) 2 (1)] + and [Ir(Ph 2 ppy) 2 (1)] + , the remaining angles in the coordination environment sphere of Ir1 vary greatly (Table 1). The widening of the cis-C-Ir-N angles in [Ir(Ph 2 ppy) 2 (2)] + is coupled to the three intra-cation pstacking interactions shown in Fig. 4. The face-to-face contacts are between pairs of phenyl and pyridine rings containing C20/ N44 and C49/N2 (see Fig. S3 †) and between the cyclometalated ring with C47 and pendant phenyl ring containing C38; the pinteractions are characterized by centroid/ring-plane and centroid/centroid distances and interplane angle of 3.37 Å, 3.61 Å and 14.9 between rings with C20/N44, 3.18 Å, 3.47 Å and 5.8 for rings with C49/N2, and 3.24 Å, 3.42 Å and 10.9 for rings with C47/C38. Packing interactions involve extensive CH/F contacts between cations and anions, and one of the toluene molecules engages in edge-to-face p-contacts with a pendant phenyl ring of the cation. 19,20 Upon cooling (Fig. S5 † and 5), the broad signals collapse and give rise at 218 K to two doublets (d 5.90 and 7.06 ppm, H G2 and H G6 ) and two multiplets (d 6.85 and 6.55 ppm, H G3 and H G5 ) (Fig. 5a). To understand the observations, we consider the modelled structure of [Ir(Phppy) 2 (2)] + (Fig. 5b). Apart from H G2/G3/G5/G6 , one other signal is signicantly affected by temperature, and shis from zd 7.4 ppm at 298 K to d 7.18 ppm at 218 K. In the HMQC spectrum at 218 K, this proton correlates to a 13 C NMR signal at d 149.4 ppm. A second highfrequency signal at d 150.0 ppm correlates to a 1 H NMR signal at d 7.76 ppm which is temperature independent (Fig. 5a). The high-frequency 13 C NMR signals are characteristic of pyridine C 6 nuclei and are identied as C D6 and C B6 ; the remaining pyridine C 6 (C E6 ) is observed at d 151.2 ppm at 218 K. Protons H B6 or H D6 (red and orange in Fig. 5b) were distinguished from NOESY spectra (at 298 and 218 K). Ring G (green in Fig. 5b) is spatially closer to phenyl ring J than the corresponding phenyl ring H, and NOESY cross peaks are observed between H G3 /H J2 and H G4 /H J2 at 298 K, and between H G3 /H J2 , H G4 /H J2 and H G5 /H J2 at 218 K. This allows the spin systems of the two [Phppy] À ligands to be discriminated. Although proton H B6 is closer to ring G, it is H D6 that is affected more as the hindered rotation of ring G is frozen out. We propose that as the p-interaction between rings C and G strengthens at low temperature, a concomitant deformation of the bpy domain (rings E and F) occurs leading to an enhanced C-H/p interaction between H D6 and ring E (and an associated shi to lower frequency for H D6 ). Twisting of the bpy unit is substantiated by the structural data (see above) and is also responsible for the dynamic behaviour of [Ir(ppy) 2 (Naphbpy)] + (Naphbpy ¼ 6-(2-naphthyl)-2,2 0 -bipyridine). 21 While phenyl ring H in the coordinated [Phppy] À ligand is free to rotate on the NMR timescale, spectroscopic data show that phenyl ring K in metal-bound [Ph 2 ppy] À is static at K. The data in Table S1, † and in particular the shi to lower frequency for all ring B protons on going from [Ir(Phppy) 2 ( 1 The effects of introducing a third phenyl group are seen by comparing the 1 H NMR spectra of [Ir(Ph 2 ppy) 2 ( 1)][PF 6 ] and [Ir(Ph 2 ppy) 2 ( 2)][PF 6 ] (Fig. 6). Pendant rings K and L are static in [Ir(Ph 2 ppy) 2 (2)] + ; each is p-stacked over an adjacent cyclometalated ligand (Fig. 6c), as indicated by the relatively low frequency shis for signals in the B, D, K and L rings. The exceptions are the signals for H B6 and H D6 which shi to higher frequency on going from [Ir(Ph 2 ppy) 2 (1)] + to [Ir(Ph 2 ppy) 2 (2)] + (Fig. 6a and b). The chemical shis for H B6 and H D6 in [Ir(Ph 2 ppy) 2 (2)] + are similar to those in [Ir(Phppy) 2 (2)] + , indicating that similar effects are operative in both complexes. The effect of cooling a CD 2 Cl 2 solution of [Ir(Ph 2 ppy) 2 ( 2)][PF 6 ] is shown in Fig. S7 and S8. †</p><!><p>The redox activity of the iridium(III) complexes was investigated by cyclic voltammetry; data are given in Table 2 and a representative CV is shown in Fig. 7. Each complex exhibits a reversible metal-centred oxidation. The trends in the iridiumcentred oxidation potential are consistent with the introduction of electron-releasing phenyl and/or tert-butyl groups. In [Ir(ppy) 2 ( 1)][PF 6 ], E ox 1/2 occurs at +0.84 V (versus Fc/Fc + , in DMF) 22 and the process occurs at increasingly lower potential on going Each complex shows a quasi-reversible reduction (Table 2) assigned to reduction of the bpy ligand (the LUMO is localized on the bpy domain). The value of E red 1/2 shis to more negative potential upon introducing t Bu substituents, consistent with previous observations. 23</p><!><p>The absorption spectra of CH 2 Cl 2 solutions of the complexes are shown in Fig. 8. The [Ir(Phppy) 2 (N^N)][PF 6 ] family shows an intense, broad absorption with l max in the range 276-278 nm arising from spin-allowed ligand-centred p*)p transitions. For the four [Ir(Ph 2 ppy) 2 (N^N)][PF 6 ] complexes, the corresponding bands are broader and exhibit two or three maxima in the approximate range 250-300 nm. The weaker absorptions around 400 and 420 nm are assigned to MLCT transitions. Excitation into the MLCT bands results in broad, unstructured emissions which are, in solution, centred at 600 nm for [Ir(Phppy) 2 ( 1)][PF 6 ] and 611 nm for [Ir(Ph 2 ppy) 2 ( 1)][PF 6 ] (Table 3). The red-shi in the emission is consistent with destabilization of the HOMO as the electron-releasing phenyl group is introduced into the C^N ligand. An analogous red-shi is observed on going from [Ir(Phppy) 2 ( 2 4)][PF 6 ] (Table 3). For both the [Ir(Phppy) 2 (N^N)][PF 6 ] and [Ir(Ph 2 ppy) 2 (N^N)][PF 6 ] series of complexes, introducing the 6-phenyl substituent into the bpy domain leads to a red-shi in the emission, while introducing the tert-butyl groups into the 4-and 4 0 -positions results in a blue shi (Table 3).</p><p>The emission spectra of powdered samples of the complexes were recorded and are presented in Fig. 9. In each case, a blue shi in the emission is observed compared to the solution spectrum (Table 3 and Fig. 10). As in solution, the emission undergoes a red-shi on introducing the additional phenyl which may be a consequence of packing effects in the solid state in the sterically crowded [Ir(Ph 2 ppy) 2 (2)] + . Both solution and solid-state emission data conrm that the introduction of the tert-butyl groups into the N^N ligand results in signicant blue-shis in l max em . The photoluminescence (PL) data for the complexes in the device conguration (thin lm), but without electrodes and PEDOT:PSS, are given in Table 3. The similarity between the emission maxima for a given complex in thin lm and solution is in contrast to the signicant blue shis observed for most complexes on going from solution to the solid state. This suggests that packing effects may be dominant in determining the latter, since the complex is present in the lms only in low concentration. The emission maxima for lms of the [Ir(Ph 2 ppy) 2 (N^N)] + complexes are slightly red-shied compared to those of lms [Ir(Phppy) 2 (N^N)] + . In each set of complexes, the presence of tert-butyl substituents causes a blue-shi in the emission maxima.</p><p>The photoluminescence quantum yields (PLQY) are generally enhanced on going from solution to the solid state (Table 3). The four complexes in which C^N ¼ Ph 2 ppy exhibit the highest PLQY values. Lifetimes of the emissions are given in Table 4. For each complex, the luminescence decay was tted using a biexponential function. Going from solution to the solid state generally results in an increase in the emission lifetime. This is especially noteworthy for the most sterically crowded cations [Ir(Ph 2 ppy) 2 (2)] + and [Ir(Ph 2 ppy) 2 (4)] + which exhibit values of s ave of 617 and 1148 ns in the solid state compared to 36 and 88 ns, respectively, in argon-degassed solution.</p><!><p>Simple LECs were prepared using all complexes, the devices were prepared on ITO-coated glass plates and consisted of a PEDOT:PSS hole injection/planarization layer (60 nm), the lightemitting layer, consisting of the iridium complex and 1-butyl-3methyl-imidazolium hexauoridophosphate ([BMIM][PF 6 ]) at a molar ratio of 4 : 1, and an aluminum top electrode. The voltage behaviour is typical for LEC devices, starting at a high value due to an initial high injection barrier and rapidly dropping as a result of the electric double-layer formation that reduces the injection barriers 24 (Fig. S9 †). Most LECs reported in the literature have been driven using a constant voltage mode. However, this leads to an increase of the width of the doped zone over time. As doped materials are efficient exciton quenchers, this leads to (partially reversible) reduction in the luminance. 7,18 To avoid this decrease in performance, we have driven the devices using a pulsed current mode, with a frequency of 1 kHz and a duty cycle of 50%. 25,26 Using pulsed current driving, iridium iTMC-based LECs are usually operated at an average current density of 50 or 100 A m À2 . The luminance and voltage versus time curves for the different devices (tested at 50 and 100 A m À2 ) are depicted in Fig. 11 and S9, † respectively. For a number of devices, the luminance does not appear to decay over time. This is obviously a good property, yet does not allow an analysis of the relationship between iridium complex composition and the device performance. Therefore, to distinguish between the different LECs, all devices were also driven using a much higher current density, of 300 A m À2 , which permits acceleration of the degradation of the device due to the higher stress that the materials are subjected to. The luminance increases with higher current density although not linearly. This is due to a reduction in device efficiency as a result of charge induced carrier quenching. 27,28 The devices containing the [Ir(Ph 2 ppy) 2 (N^N)] + iTMCs have a slightly lower luminance than those based on [Ir(Phppy) 2 (N^N)] + . Under these pulsed current conditions, the efficiency scales directly to the luminance, and it follows that the efficiencies are also lower for the [Ir(Ph 2 ppy) 2 (N^N)] + complexes. The effect of introducing the tert-butyl groups in the N^N domain does not lead to an increase in luminance or in the efficiency of the LECs as might be expected by comparison with previous results. 11 This is probably related to the fact that the [Phppy] À and [Ph 2 ppy] À ligands are sterically demanding which results in reduced close packing in the lm, thereby enhancing the radiative decay pathways. The efficiency of the LEC devices containing N^N ligands 2 or 4 with the 6-phenyl group is lower than those in which N^N ¼ bpy; this is consistent with previous results. 11,29 This is directly related to a lower PLQY of the complexes that exhibit the p-p stacking.</p><p>The trend in the lifetimes of the devices is both important and interesting. In general, the devices based on iTMCs that contain the phenyl group on the bpy ligand, and hence show intra-cation p-p stacking, show a faster decay of luminance than the devices using the iTMCs without the p-p stacking ability. However, this trend is not observed for the device containing [Ir(Ph 2 ppy) 2 (2)] + which, although exhibiting a rather low luminance of 200 cd m À2 , stays constant over a period of 350 hours (Fig. 11a). The series of iTMCs evaluated in this study all exhibit exceptional stabilities in LECs. The best performances are observed for devices containing [Ir(Phppy) 2 ( 1)][PF 6 ], with a maximum efficiency of 3.5 cd A À1 and luminance of 1024 cd m À2 (at an average current density of 300 A m À2 ) and an extrapolated lifetime in excess of 2800 hours (time to reach 50% of the maximum luminance). In [Ir(Phppy) 2 ( 1)][PF 6 ], the phenyl substituents on the C^N ligand reside on the periphery of the complex (Fig. 1) and are not involved in inter-ligand p-stacking within the iridium(III) coordination sphere. Although incorporation of an intra-cation p-stacking domain may be advantageous, 7,9,10 this is not necessarily a general design principle 8,19 and in the current study, the presence of intra-cation p-stacking does not improve the stability of the light emitting device.</p><p>The electroluminescence spectra (Fig. S10 †) are slightly blue shied with respect to the photoluminescence maxima as reported in Table 3.</p><!><p>We have designed a series of cyclometalated iridium(III) complexes [Ir(C^N) 2 (N^N)] + in order to examine the effects of having multiple p-stacking domains within the coordination sphere of the iridium(III) centre. The complexes have been synthesized via solvento precursors, thus avoiding the use of chlorido-dimer intermediates. Single crystal structure determinations and variable temperature solution 1 H NMR spectroscopic data conrm that the pendant phenyl domains engage in multiple face-to-face p-interactions. The [Ir(Phppy) 2 (N^N)] + and [Ir(Ph 2 ppy) 2 (N^N)] + iTMCs all show excellent luminescent properties, in particular when employed in thin solid lms. LECs using these complexes exhibit a very stable luminance output over time even when driven at elevated current densities. The most stable LEC had an extrapolated lifetime in excess of 2500 hours at a starting luminance above 1000 cd m À2 achieved under accelerated testing conditions. These remarkable lifetimes were obtained for devices using complexes both with and without the ability to form intra-molecular face-to-face p-stacking.</p>

Royal Society of Chemistry (RSC)

Stress Granules in Cancer

The capacity of cells to organize complex biochemical reactions in intracellular space is a fundamental organizational principle of life. Key to this organization is the compartmentalization of the cytoplasm into distinct organelles, which is frequently achieved through intracellular membranes. Recent evidence, however, has added a new layer of flexibility to cellular compartmentalization. As such, in response to specific stimuli, liquid-liquid phase separations can lead to the rapid rearrangements of the cytoplasm to form membraneless organelles. Stress granules (SGs) are one such type of organelle that form specifically when cells are faced with stress stimuli, to aid cells in coping with stress. Inherently, altered SG formation has been linked to the pathogenesis of diseases associated with stress and inflammatory conditions, including cancer. Exciting discoveries have indicated an intimate link between SGs and tumorigenesis. Several pro-tumorigenic signaling molecules including the RAS oncogene, mTOR, and histone deacetylase 6 (HDAC6) have been shown to upregulate SG formation. Based on these studies, SGs have emerged as structures that can integrate oncogenic signaling and tumor-associated stress stimuli to enhance cancer cell fitness. In addition, growing evidence over the past decade suggests that SGs function not only to regulate the switch between survival and cell death, but also contribute to cancer cell proliferation, invasion, metastasis, and drug resistance. Although much remains to be learned about the role of SGs in tumorigenesis, these studies highlight SGs as a key regulatory hub in cancer and a promising therapeutic target.

stress_granules_in_cancer

Introduction<!>Formation of Stress Granules<!>Stress Granule Structure<!>Dysregulated Cancer Signaling and Stress Granule Formation<!>RAS<!>mTORC1<!>Glycolysis and the Hexosamine Biosynthetic Pathway<!>HDAC6<!>Stress Granules and Cancer Hallmarks<!>Stress Granules and Proliferation<!>Stress Granules and Suppression of Cell Death<!>Stress Granule-Mediated Suppression of Stress-Induced Apoptosis<!>Stress Granules and Chemotherapy Resistance<!>Stress Granules and Tumor Metastasis<!>Concluding Remarks

<p>Stress granules (SGs) are non-membranous cytoplasmic organelles that assemble when cells are exposed to stress (Buchan and Parker 2009; Kedersha et al. 2013; Buchan 2014). They consist of a vast proteomic and transcriptomic network and range in size from tens of nanometers to several micrometers (Anderson and Kedersha 2008; Jain et al. 2016; Protter and Parker 2016; Khong et al. 2017; Namkoong et al. 2018). These structures are highly dynamic and can undergo fusion and fission. Furthermore, once stress subsides, SGs disassemble, and their components disperse back into the cytosol (Protter and Parker 2016; Wheeler et al. 2016). Earlier reports proposed that SGs function to store and target mRNA for degradation during stress (Anderson and Kedersha 2008). Studies over the past decade, however, have drastically expanded our understanding of their function. It is now well-established that SGs function as signaling hubs that regulate gene expression and signal transduction, and are critical to the cellular stress response and survival under adverse conditions.</p><p>In vivo, SGs have been associated with several pathologies, including cancer (Grabocka and Bar-Sagi 2016; Cruz et al. 2019; Herman et al. 2019; Wolozin and Ivanov 2019). Studies in cancer cells and animal models of tumorigenesis have established SGs as a stress-adaptive strategy hijacked by cancer cells to support tumorigenesis (Somasekharan et al. 2015; Grabocka and Bar-Sagi 2016; Protter and Parker 2016). Stress adaptation is emerging as an important property of cancer cells (Sharma et al. 2016; Truitt and Ruggero 2016; El-Naggar and Sorensen 2018). As oncogenic-driven hyperproliferation demands a high expenditure of cellular resources, cancer cells are often faced with stress conditions (Solimini et al. 2007; Stylianopoulos et al. 2012; Urra et al. 2016). As such, the increased demand for protein synthesis and flux in the endoplasmic reticulum results in proteotoxic stress and misfolded proteins, and hyper-replication of DNA leads to DNA damage and genotoxic stress (Joyce and Pollard 2009; Fiaschi and Chiarugi 2012; Yadav et al. 2014; Anastasiou 2017; Gouirand et al. 2018). The increased metabolic demand contributes to nutrient stress, reactive oxidant species (ROS), and pH imbalances (Vincent et al. 2015; Panieri and Santoro 2016). Furthermore, as tumors outgrow the local vascularization, the inadequate blood supply leads to reduced oxygen and nutrient levels (Fig. 1) (Wellen and Thompson 2010; Semenza 2012). Such levels of stress would normally lead to cell death, but cancer cells are able to quickly adapt and survive (Wellen and Thompson 2010; Gorrini et al. 2013; Wang and Kaufman 2014; Senft and Ronai 2016; Lee et al. 2020). Stress adaptation, therefore, can contribute to tumorigenesis by enhancing the cellular fitness and supporting the survival of cancer cells.</p><p>The stress adaptation of cancer cells is conferred, in large part, by the capacity of oncogenic molecules to elicit compensatory responses to tumor-associated stresses in order to promote tumor cell survival (Solimini et al. 2007; Commisso et al. 2013; Ruggero 2013; Easwaran et al. 2014; Eirew et al. 2015; Perera et al. 2015; Amaravadi et al. 2016; Lee et al. 2020). Such responses include alterations to the genetic, epigenetic, and transcriptomic landscape and, more recently, the hijacking of stress-coping cellular processes. Examples of the latter include oncogene-induced upregulation of macropinocytosis and autophagy, as well as modifications of lysosomes, which enable cancer cells to cope with nutritional stress (Commisso et al. 2013; Perera et al. 2015; Amaravadi et al. 2016). In addition, cancer cells upregulate the unfolded protein response (UPR) to cope with ER stress and misfolded proteins (Obacz et al. 2017). Whereas these processes are upregulated by cancer cells to cope with specific stress stimuli, SGs have been identified as a cancer cell stress-adaptive mechanism for a broad spectrum of tumor-associated stresses including oxidative-, proteotoxic-, osmotic- stress, as well as for nutrient deprivation (Fig. 1) (Somasekharan et al. 2015; Grabocka and Bar-Sagi 2016; Protter and Parker 2016).</p><p>Several studies have demonstrated that pro-tumorigenic signaling pathways that are hyperactivated in cancer stimulate the formation of SGs (Somasekharan et al. 2015; Grabocka and Bar-Sagi 2016; Protter and Parker 2016). This enhanced formation of SGs, in turn, may promote cancer development and progression. Evidence exists that SGs may support tumorigenesis not only through facilitating cancer cell survival but also through contributing to tumor cell proliferation and metastasis (Fig. 1). In addition, it has been shown that SGs may play an important role in the development of drug resistance (Fig. 1). These studies highlight that while SGs are important in the normal cellular stress response and may impact several diseases (reviewed in excellent detail elsewhere (Protter and Parker 2016; Mahboubi and Stochaj 2017; Cruz et al. 2019; Herman et al. 2019; Wolozin and Ivanov 2019)), the hijacking of this process by cancer cells may be critical for tumorigenesis and a promising therapeutic target. Here we review recent data illuminating the oncogenic signaling pathways that promote the formation of SGs in cancer cells and the mechanisms through which SGs may contribute to tumor progression and response to chemotherapy. In addition, we discuss how leveraging this knowledge may instruct the development of therapeutic strategies for the treatment of cancer and overcoming drug resistance.</p><!><p>Since the initial discovery of SGs in tomato cells exposed to heat shock, several studies have revealed SG formation as an evolutionary conserved response to stress produced by plants, protozoa, yeast, C. elegans, Drosophila, and mammalian cells (Nover et al. 1983; Arrigo et al. 1988; Collier et al. 1988; Buchan et al. 2008; Farny et al. 2009; Thomas et al. 2011; Gutierrez-Beltran et al. 2015). SG formation is induced by a variety of stress stimuli including oxidative stress, heat shock, ER stress, nutrient deprivation, UV irradiation, proteotoxic stress, and several chemotherapeutic agents (Kedersha et al. 1999; Kimball et al. 2003; Kwon et al. 2007; Mazroui et al. 2007; Fournier et al. 2010; Emara et al. 2012; Kaehler et al. 2014; Moutaoufik et al. 2014; Adjibade et al. 2015; Grabocka and Bar-Sagi 2016; Reineke et al. 2018; Lin et al. 2019).</p><p>The formation of SGs is closely linked to translation inhibition (Kedersha et al. 1999; Protter and Parker 2016). Cells respond to stress by blocking protein synthesis via the phosphorylation of the α subunit of the eukaryotic initiation factor 2 (eIF2). In mammalian cells, the phosphorylation of eIF2α is mediated by a family of four different serine/threonine kinases each of which is activated by specific forms of stress. These kinases include the general control non-derepressible 2 (GCN2) kinase which is activated by amino acid deprivation; the heme-regulated inhibitor (HRI) kinase which is activated by oxidative or osmotic stress; the double stranded RNA-dependent protein (PKR) kinase which is activated in response to viral infections; and the PKR-like endoplasmic reticulum kinase (PERK) which is activated by ER stress (Wek et al. 2006; Donnelly et al. 2013). eIF2α is one of the three subunits (α, β, γ) of eIF2 which mediates the binding of initiator methionyl-tRNA (Met-tRNAiMet) to the ribosome in a GTP-dependent manner (Donnelly et al. 2013). The eIF2-Met-tRNAiMet-GTP complex binds the 40S ribosomal subunit, as well as eIF1, eIF1A, eIF5, and eIF3, to form the 43S pre-initiation complex (PIC); PIC then associates with eIF4F on mRNA to form a new 48S complex which scans the mRNA for the start codon (AUG). Phosphorylation of eIF2α under stress prevents the formation of eIF2- Met-tRNAiMet-GTP, thus resulting in a translationally stalled, noncanonical 48S complex that is unable to recruit the 60S ribosomal subunit (Jackson et al. 2010). Consequently, ribosomes runoff the transcripts, causing a flux of messenger ribonucleoprotein complexes (mRNPs) and exposed RNA, which are critical for SG formation (Wheeler et al. 2016).</p><p>It is important to note that while translation inhibition is key for the formation of SGs, it can also occur independently of eIF2α phosphorylation (Jackson et al. 2010). For one, changes in the composition or activity of the eIF4F-cap binding complex (eIF4A, eIF4E, eIF4G) can inhibit translation and induce SG formation. As such, hydrogen peroxide initiates SG assembly by inhibiting translation initiation through disrupting the interaction of eIF4E with eIF4G (Emara et al. 2012; Fujimura et al. 2012). Also, chemicals such as hippuristanol and pateamine A interfere with translation initiation by blocking the eIF4A helicase, which is required for the ribosome recruitment phase of translation initiation. Lastly, the anti-inflammatory lipids 15-deoxy-Δ-12,14-prostaglandin J2 and prostaglandin A1, which are potent inducers of SG formation, inhibit translation by preventing the association of eIF4A with eIF4G (Bordeleau et al. 2006; Dang et al. 2006; Kim et al. 2007; Grabocka and Bar-Sagi 2016).</p><p>Regardless of the mode of translation inhibition, the resulting flux of mRNPs and exposed RNA are essential for the formation of SGs (Protter and Parker 2016; Wheeler et al. 2016; Ivanov et al. 2019). These molecules initiate the first steps of SG assembly by binding to RNA-binding proteins termed SG-nucleating proteins, which include the poly(A)-binding protein (PABP1) PABP1, the T-cell internal antigen 1 (TIA-1), TIA-1 related (TIAR), and Ras-GTPase-activating protein SH3-domain-binding protein1 (G3BP1). SG assembly is further aided by the ability of SG-nucleating proteins to phase-separate and coalesce in the cytoplasm, leading to the formation of nascent SGs. Further recruitment of proteins and transcripts via protein-protein, protein-RNA, and RNA-RNA interactions results in the formation of mature SGs. The phase-separation capacity of SG-nucleating proteins is mediated by intrinsically disordered domains (IDD), which are a prominent feature of SG-nucleator proteins (Protter and Parker 2016; Mahboubi and Stochaj 2017). The role of these domains in the coalescence of SG-nucleator proteins is supported by several studies showing that, at high concentration, IDD domains can induce spontaneous phase separation. In agreement with this notion, one study showed that overexpression of SG nucleators, which would presumably increase the concentration of IDD domains, is sufficient to induce SG formation in vitro even in the absence of stress (Gilks et al. 2004; Matsuki et al. 2013; Lin et al. 2015; Molliex et al. 2015). In addition to IDD domains, phase separation of SG nucleators is regulated by posttranslational modifications which can enhance or weaken the multivalent interactions between these molecules (Kwon et al. 2007; Tsai et al. 2008; Carpio et al. 2010; Xie and Denman 2011; Owen and Shewmaker 2019). Shedding further light onto the macromolecular interactions that contribute to SG assembly, a recent study indicated that RNA-RNA interactions and ability to self-coalesce wherever there is a high concentration of RNA may also contribute to SG assembly (Van Treeck et al. 2018). Taken together, these features support a model where SGs are formed by the concerted action of phase separation, protein-RNA, protein-protein, and RNA-RNA interactions.</p><!><p>SGs are non-membranous structures of the cytoplasm that contain stalled mRNA transcripts, poly(A) mRNAs, microRNAs, translation initiation factors, large and small ribosomal subunit protein components, and a vast network of proteins (Jain et al. 2016; Khong et al. 2017; Markmiller et al. 2018; Namkoong et al. 2018). Recent studies suggest that SGs have a biphasic architecture consisting of a stable core, which is surrounded by a dynamic shell (Fujimura et al. 2009; Souquere et al. 2009; Jain et al. 2016; Wheeler et al. 2016; Markmiller et al. 2018). This architecture is thought to provide multiple levels of functionality within SGs whereby the shell provides a platform for an active exchange of transcripts and protein with the cytoplasm, whereas compartmentalization to the stable core by definition allows for more stable retention (Jain et al. 2016; Protter and Parker 2016; Wheeler et al. 2016; Van Treeck et al. 2018).</p><p>The dynamic nature of SG shells has rendered their full isolation and characterization intractable to date. However, stable SG cores have been purified and reveal a vast network of 411 proteins (Jain et al. 2016). These include several RNA-binding proteins, an array of signaling proteins including protein kinases, phosphatases, GTPases, ATPases, adaptor proteins, endoribonucleases, helicases, glycosyltransferases, ubiquitin modifying enzymes, and components of the RNAi machinery (Jain et al. 2016). Building on the methodology of Jain et al., characterization of the SG-core transcriptome revealed that 10–12% of the total mRNA molecules accumulate in SGs (Khong et al. 2017; Namkoong et al. 2018; Matheny et al. 2019). This recruitment does not appear to be random. The ~185 gene mRNA transcripts that have been identified as most likely to find their way to SGs follow patterns of shared transcript length and translation efficiency and share a handful of specific RNA motifs (Khong et al. 2017; Namkoong et al. 2018; Matheny et al. 2019). Longer mRNAs and ncRNAs, transcripts with lower translation efficiency, and transcripts with RNA sequence motifs such as adenylate-uridylate (AU)-rich element, Pumilio-binding element, and guanylate-cytidylate (GC)-rich element are highly common in SGs (Lin et al. 2007; Khong et al. 2017; Namkoong et al. 2018; Van Treeck et al. 2018; Matheny et al. 2019; Moon et al. 2019). While SG cores induced by different stimuli shared several protein and transcript components, considerable differences were also observed, depending on the specific type of stress (Khong et al. 2017; Namkoong et al. 2018). Thus, the composition of SG cores is specific to the type of stress. Research has yet to illuminate exactly which proteins and transcripts associate with the SG shells, but it is likely that similar to SG cores, they will capture, modify, and exchange proteins and transcripts based on the specific kind of stress that the cell is experiencing.</p><!><p>In vivo, SGs are found in cancer cells of osteosarcomas and tumors of the pancreas and colon but are absent in normal cells from the same tissues (Somasekharan et al. 2015; Grabocka and Bar-Sagi 2016). As previously mentioned, tumors are frequently faced with stress conditions, and perhaps not surprisingly, SGs are often detected in tumor regions experiencing stress. Evidence suggests, however, that the presence of SGs in tumors is not a sole consequence of heightened stress stimuli, but that dysregulation of several signaling pathways also contributes to SG formation. Dysregulated cancer signaling appears to facilitate SG formation in response to stress through promoting translation inhibition and protein-protein interactions important for SG assembly. This section discusses how dysregulated RAS, mTOR, HDAC, glycolytic, and hexosamine biosynthetic pathways can promote SG formation in cancer cells in vitro and in vivo.</p><!><p>Mutations in the RAS genes (KRAS, NRAS, and HRAS) constitute one of the largest oncogenic alterations in cancer and are present in approximately 30% of all human cancers (Pylayeva-Gupta et al. 2011). Mutant RAS proteins drive several cell functions that support cancer development and progression including cancer cell proliferation, apoptosis, metastasis, metabolism, immune modulation, cancer-associated fibroblast modulation, and ECM composition and structure modification (Liu et al. 2011; Tao et al. 2014; Dias Carvalho et al. 2018; Yang et al. 2018). Furthermore, accumulating evidence indicates that mutant RAS proteins stimulate stress-adaptive responses, allowing the cell to resist tumor-associated stresses and chemotherapeutic agents (Commisso et al. 2013; Tao et al. 2014; Amaravadi et al. 2016; Yang et al. 2018).</p><p>SGs were observed in mutant KRAS pancreatic cancer cells but not in normal pancreas tissue, in both human samples and mouse models of pancreatic cancer (Grabocka and Bar-Sagi 2016). As SGs in this setting were detected in the absence of exogenous stress stimuli, but were present in hypoxic tumor regions, this study first linked SG formation with tumor-associated stresses in vivo. Moreover, the study showed that SGs were present in mutant KRAS pancreatic tumors but not in wild-type (WT) RAS tumors, despite similar levels of hypoxia. This observation indicates a cooperation between mutant KRAS signaling and tumor-associated stresses in SG formation. Of note, SGs were also detected in non-hypoxic regions of mutant KRAS pancreatic tumors, suggesting that mutant KRAS may cooperate with additional stresses in stimulating SG formation. Consistent with this model, mutant KRAS enhanced SG formation in cells exposed to various forms of stress in vitro. These included oxidative stress, proteotoxic stress, UV-C irradiation, and chemotherapeutic agent-induced stress. Mutant KRAS cells also showed a heightened dependence on SGs for survival under stress stimuli, when compared to KRAS-WT cancer cells (Grabocka and Bar-Sagi 2016). As such, inhibition of SG formation in KRAS mutant cells led to higher levels of cell death compared to KRAS-WT cells. Thus, higher cellular levels of SGs may indicate a heightened dependence on SGs for cancer cell survival. An earlier study reported that overexpression of mutant HRAS also stimulated SG formation thus suggesting that all mutant Ras isoforms may be able to stimulate SGs in vivo (Tourriere et al. 2003). With all of this in mind, SGs may be a unique vulnerability that can be exploited for the treatment of all RAS mutant tumors, the treatment options for which are currently limited.</p><p>Mechanistically, mutant KRAS was shown to stimulate SG formation by enhancing the levels of the prostaglandin 15-deoxy-delta 12,14-prostaglandin J2 (15-d-PGJ2) (Fig. 2) (Grabocka and Bar-Sagi 2016). 15-d-PGJ2 can induce SG formation by inhibiting translation through covalently binding to eIF4A to block its interaction with eIF4G, as well as stimulating eIF2α phosphorylation (Kim et al. 2007; Tauber and Parker 2019). Recently, the nuclear factor erythroid 2-related factor 2 (NRF2) was also implicated in 15-d-PGJ2-mediated SG formation in KRAS mutant pancreatic cancer cells (Mukhopadhyay et al. 2020). The mechanisms through which 15-d-PGJ2-stimulated NRF2 promotes SG formation remain unclear. However, NRF2 regulation of SGs was shown to rely on glutamine, thereby linking KRAS-mediated SG induction to glutamine metabolism.</p><p>Mutant KRAS was shown to stimulate 15-d-PGJ2 production through the downstream effector molecules RAF and RALGDS (Grabocka and Bar-Sagi 2016). Signaling from these KRAS effector molecules upregulates15-d-PGJ2 through two different paths. For one, mutant KRAS signaling upregulates cyclooxygenase 2 (COX2), which catalyzes 15-d-PGJ2 synthesis. Secondly, mutant KRAS downregulates the NAD+-dependent 15-hydroxyprostaglandin dehydrogenase (HPGD), which inhibits prostaglandin catabolism.</p><p>Interestingly, 15-d-PGJ2 is also secreted from mutant KRAS cancer cells to stimulate SG formation in a paracrine manner (Grabocka and Bar-Sagi 2016). In addition, paracrine stimulation of SG formation by mutant KRAS cancer cells promoted the survival of KRAS-WT cells when exposed to stress stimuli. The observation that mutant KRAS cells can promote survival of KRAS-WT cells through paracrine induction of SG formation is important; it raises the possibility that SG formation serves as a platform for mutant KRAS to promote the stress resistance and survival of the various cell types in the tumor stroma. Coupling this idea with the well-established role of tumor stroma in the development, growth, and drug resistance of mutant KRAS tumors, these findings also highlight the need for a better understanding of how SGs are integrated in KRAS-driven tumorigenesis.</p><!><p>The mammalian target of rapamycin (mTOR) is a crucial signaling node that regulates cell survival, proliferation, and metabolism (Saxton and Sabatini 2017; Mossmann et al. 2018). mTOR operates in two multi-protein complexes: mTOR complex 1 (mTORC1) and mTOR complex 2 (mTORC2) which have distinct compositions, functions, and substrate specificities. Both mTORC1 and mTORC2 are commonly hyperactivated in cancer; however, only mTORC1 activity has been linked to SG formation in cancer cells (Guertin and Sabatini 2007). mTORC1 is considered to be an essential factor for cancer metabolism reprogramming and adaptation to cellular stress (Chantranupong et al. 2016; Wolfson et al. 2016; Harachi et al. 2018; Lee et al. 2018). The mTORC1 catalytic complex consists of mTOR and co-factor molecules which include regulatory-associated protein of mTOR (RAPTOR), DEP domain containing mTOR interacting protein (DEPTOR) proline-rich AKT substrate 40 kDa (PRAS-40), FKBP38, and mammalian lethal with Sec13 protein 8 (mLST8). mTORC1 activity is both positively and negatively regulated by components of the catalytic complex.</p><p>Several studies have shown that mTORC1 activation in cancer cells can facilitate SG formation (Fig. 2). Inhibition of mTORC1, either genetically through shRNA-mediated downregulation or through pharmacological inhibition of its catalytic activity, impairs SG formation in cancer cells exposed to oxidative or proteotoxic stress (Fournier et al. 2013; Wippich et al. 2013). Similarly, inhibition of mTORC1 activity through the downregulation of RAPTOR also impaired SG formation (Fournier et al. 2013). Furthermore, mTORC1 inhibition impaired the activations of SG-mediated anti-apoptotic pathways under stress conditions (Fournier et al. 2013; Wippich et al. 2013).</p><p>While the mechanistic pathways through which mTORC1 facilitates SG formation have not been fully elucidated, two major downstream effector molecules, the ribosomal protein S6 kinase 1 and 2 (S6K1, S6K2) and the eukaryotic translation initiation factor 4E (eIF4E) binding protein 1 (4EBP1), have been implicated (Fournier et al. 2013; Sfakianos et al. 2018). Both S6K1 and S6K2 were shown to localize to SGs, and their kinase activity was required for SG formation under conditions of oxidative stress (Sfakianos et al. 2018). Interestingly, S6K1 and S6K2 appear to play distinct roles in SG formation. S6K1 was shown to promote the formation of SGs by regulating eIF2α phosphorylation, whereas S6K2 was required for SG maintenance after assembly (Sfakianos et al. 2018). mTORC1-stimulated 4EBP1 has also been implicated in SG formation under oxidative stress conditions (Fig. 2). In contrast to mTORC1-S6K1 mediated SG formation, the formation of SGs via mTORC1-4EBP1 appears to be peIF2α independent (Fournier et al. 2013). Instead, mTORC1-4EBP1 is thought to promote SGs by impinging on the eIF4E-eIF4GI interaction and translation initiation under stress.</p><p>It is somewhat paradoxical for a pathway that is best known for stimulating protein translation to be associated with translation inhibition. A potential explanation may be that under specific stress stimuli, mTORC1 promotes SG formation independent of its effect on protein synthesis. In this context, activation of stress kinases would counter mTORC1 signaling and inhibit translation; this would initiate SG formation which is then aided by the capacity of mTORC1 to promote protein interactions and modifications with roles in SG formation. Once components of the mTORC1 complex and effector molecules are recruited to SGs however, mTORC1 would be prevented from stimulating translation in the cytosolic compartment. In agreement with this model, SGs have been shown to inhibit mTORC1 activity (Thedieck et al. 2013; Wippich et al. 2013). This is consistent with studies showing that while mTORC1 activation is required for cancer cell survival, chronic hyper-active mTORC1 can lead to apoptosis (Wippich et al. 2013). Thus, by recruiting mTORC1 and inhibiting its cytosolic function, SGs would contribute to cell survival by blunting chronic hyperactivation of mTORC1 (Wippich et al. 2013).</p><p>Mechanistically, SGs restrict mTORC1 hyperactivation through sequestering components of the catalytic complex including mTOR and RAPTOR and all three subunits (α-catalytic subunit; β and γ- regulatory subunits) of the upstream activator AMP-activated protein serine/threonine protein kinase (AMPK) (Hofmann et al. 2012; Takahara and Maeda 2012; Thedieck et al. 2013; Wippich et al. 2013). Distinct from mTOR and RAPTOR however, inhibition of AMPK does not impair SG formation in cancer cells. These studies have suggested that AMPK recruitment via interaction with G3BP1 occurs in the later stages of SG formation as a potential mechanism to restrain mTORC1 hyperactivation and promote survival.</p><p>Notably, the reactivation of mTORC1 in the recovery phase after stress has subsided has also been linked to SGs. As such, SG disassembly was shown to contribute to the activation of mTORC1 by the dual specificity tyrosinephosphorylation-regulated kinase 3 (DYRK3) (Wippich et al. 2013). In its inactive state, DYRK3 promotes SG formation and, consequently, the recruitment of mTOR to SGs and inhibition of mTORC1. Stress recovery is associated with DYRK3 activation which stimulates SG dissolution to release mTORC1 components while simultaneously phosphorylating and inhibiting the mTORC1 inhibitor PRAS40. Altogether, these studies indicate that SG formation contributes to the inactivation of mTORC1 during oxidative stress, whereas SG dissolution contributes to the necessary reactivation during stress recovery.</p><!><p>It is well established that cancer cells alter their metabolism to derive energy from glycolysis instead of mitochondrial oxidative phosphorylation and utilize more glucose than normal cells (Ganapathy-Kanniappan and Geschwind 2013; Pavlova and Thompson 2016). Whereas the majority of glucose that enters the cell proceeds to glycolysis for ATP generation, the remaining glucose enters the hexosamine biosynthetic pathway (HBP) and along with glutamine, glucosamine, and acetylcoenzyme A is utilized to generate the amino sugar uridine diphosphate N-acetylglucosamine (UDP-GlcNAc) (Akella et al. 2019). Both glycolysis and the HBP pathway are known to promote SG formation, indicating that enhanced glycolytic and HBP flux may contribute to the formation of SGs observed in tumors (Fig. 2) (Jain et al. 2016).</p><p>A proteomic analysis of mammalian stress granule cores in cancer cells revealed a large number of proteins with ATPase activity as components of SGs (Jain et al. 2016). This study also showed that ATP is required for both SG assembly and dynamics. Whether specific ATPases are required for these processes remains to be elucidated (Jain et al. 2016). However, given that cancer cells have a preferential reliance on glycolysis for ATP production, inhibition of glycolysis might be expected to impair SG formation in cancer cells. This prediction is supported by evidence that blocking the glycolytic pathway impaired SG formation in sarcoma cells (Jain et al. 2016). Glycolytic inhibitors, therefore, may be a promising strategy to inhibit SGs in vivo.</p><p>The final product of the HBP pathways, UDP-GlcNAc, is critical for the generation of metabolic intermediates as well as for the glycosylation of proteins; UDP aids in the N-linked and O-linked glycosylation of proteins in the ER and Golgi and in the O-Linked N-Acetylglucosamine (O-GlcNAc) modification of nuclear and cytoplasmic proteins by OGT (O-GlcNAc transferase) (Akella et al. 2019). Several studies support a role for glycosylation in tumorigenesis, and changes in protein glycosylation have been observed in several cancers including those of the pancreas, colon, melanoma, lung, liver, and prostate (Sharma et al. 2016). One proposed mechanism through which altered protein glycosylation contributes to cancer progression is through enabling cancer cells to cope with stress. Stress stimuli have been shown to enhance protein O-GlcNAc modifications, and oncogenic pathways such as PI3K-mTOR-MYC and MEK/ERK signaling pathways can stimulate OGT activity and O-GlcNAc modifications to enhance cell survival under stress (Sohn et al. 2004; Zachara et al. 2004; Taylor et al. 2008; Ferrer et al. 2014; Sodi et al. 2015; Zhang et al. 2015; Katai et al. 2016).</p><p>An RNA-mediated interference-based screen in osteosarcoma cells identified the HBP pathway and O-GlcNAc protein modifications as critical to SG formation, thereby implicating SGs as a potential mechanism through which enhanced HBP flux promotes tumorigenesis (Ohn et al. 2008). Knockdown of either sortilin (trans-membrane protein that regulates the vesicular transport of the GLUT4 glucose transporter and glucose uptake), GFAT2 (glutamine: fructose 6 phosphate amidotransferase 2), or OGT inhibited SG formation in cancer cells but had no effect on eIF2α phosphorylation. Thus, the HBP pathway acts independently of peIF2α in inducing SGs in cancer cells. Instead, HBP-dependent SG formation is mediated by O-GlcNAc modification of receptor for activated C kinase 1 (RACK1), PROHIBITIN-2, and several ribosomal proteins, which promotes their aggregation into SGs (Ohn et al. 2008). In addition to components of the translational machinery, the HBP pathway regulates multiple proteins with established roles in SG formation (mTOR and AMPK) and can be manipulated by endogenous metabolites (glutamine) and oncogenic signaling pathways implicated in SG formation (PI3K and RAS/RAF). It remains to be elucidated how these signals integrate in vivo to promote SG upregulation. Nonetheless, given the role of the HBP pathway in SG formation, it is a promising therapeutic target for inhibiting SG formation in cancer.</p><!><p>Histone acetylation and deacetylation is an important posttranslational mechanism that regulates gene expression. Histone deacetylase (HDAC) proteins are key enzymes that regulate acetylation levels. Within the HDAC family, HDAC6 is unique as it determines the acetylation status of not only histones but also of several non-histone substrates such as dynein, α-tubulin, cortactin, and HSP90 (Li et al. 2018). HDAC6 is overexpressed in melanoma, lung, pancreas, breast, and bladder cancer and is thought to promote tumorigenesis by regulating cancer cell proliferation, metastasis, and immune regulators through both its histone and non-histone substrates (Lee et al. 2008; Wickstrom et al. 2010; Lafarga et al. 2012; Woan et al. 2015; Li et al. 2018). HDAC6 was shown to be a stable integral component of SGs and critical to SG formation in cancer cells under various stress stimuli including oxidative stress, UV irradiation, mitochondrial stress, or heat shock (Kwon et al. 2007). Mechanistically, HDAC6 is thought to promote SG formation through deacetylating G3BP1 which stimulates its RNA-dependent interaction with PABP1, a key component and regulator of SG assembly (Fig. 2) (Gal et al. 2019). In addition, the interaction of HDAC6 with dynein and microtubules has also been suggested to promote SG formation, potentially through mediating mRNP translocation to SGs (Kwon et al. 2007).</p><!><p>Significant amount of evidence supports a role for SGs in tumorigenesis. However, the cancer hallmarks impacted by SGs, and the mechanisms through which they do so, remain to be fully understood. This section discusses the role of SGs in the pathogenesis of cancer and the evidence linking SGs to cancer cell proliferation, metastasis, and survival and chemotherapy resistance.</p><!><p>Accumulating evidence indicates that SG formation is linked to the proliferative status of cells. One such example is cellular senescence, which has been shown to have a profound effect on the cellular capacity to form SGs. Cellular senescence is a cytostatic program that can be triggered by multiple mechanisms. These include (a) telomere attrition that occurs with replicative "aging" of cells – known as replicative senescence – and (b) exposure to exogenous agents that induce DNA damage (oxidative stress, chemotherapy, UV light), which is referred to as stress-induced premature senescence and leads to cell-cycle arrest via the activation of the DNA damage response (DDR). Lastly, acquisition of oncogenic mutations can lead to oncogene-induced senescence, which can also be mediated by DDR (Campisi and d'Adda di Fagagna 2007; Campisi 2013).</p><p>As a cell cycle-arrest program, senescence functions as a tumor-suppressive mechanism. However, senescent cells can also contribute to the generation of a tumor-promoting microenvironment via the secretion of several cytokines and chemokines (senescence-associated secretory phenotype; SASP) that promote cell cycle progression, de-differentiation, and metastasis. In in vitro models of stress-induced premature senescence, transition from the proliferative state to pre-senescence and senescence correlated with a progressive impairment of the cellular capacity to form SGs (Lian and Gallouzi 2009; Moujaber et al. 2017; Omer et al. 2018). These observations raised the possibility that SGs may play a role in preventing cells from exiting cell cycle and entering senescence. Mechanistically, senescence-dependent SG impairment was shown to be driven by the depletion of the transcription factor specific protein 1 (SP1) which regulates the expression of the SG-nucleating proteins G3BP1, TIA-1/TIAR, eIF4G, hnRNPK, and HuR (Moujaber et al. 2017). Consistent with a role for SGs in inhibiting senescence, a study showed SG inhibition via G3BP1 knockdown promoted stress-induced cellular senescence (Omer et al. 2018). SGs therefore appear to inhibit cellular senescence, while acquisition of the senescent phenotype on the other hand suppresses SGs. This is consistent with the tumor suppressive role of senescence and the function of SGs as a cytoprotective mechanism that can promote tumorigenesis. It should be noted, however, that PAI-1 – a marker of senescence as well as a downstream effector and key component of SASP – is recruited to SGs (Omer et al. 2018). As translocation to SGs would prevent the secretion of PAI-1, SGs in this setting could, in theory, function to suppress SASP and the associated tumor-promoting activity. This raises a note of caution regarding SG inhibition for the treatment of cancer; while inhibition of SGs could push cells toward senescence and, thus, halt tumor growth, the lack of SGs in senescent cells could also aid tumorigenesis by contributing to SASP.</p><p>Several proteins and mRNAs that carry out cell division localize to SGs in cancer cells; this observation supports the idea that SGs play a role in coordinating cell proliferation. RBFOX2 is a member of the RBFOX family of proteins that regulate alternative pre-mRNA splicing and mRNA stability (Jin et al. 2003; Lovci et al. 2013). Under stress conditions, RBFOX2 is recruited to SGs via its RNA-binding domain and preferentially binds cell cycle-related mRNAs including retinoblastoma 1 (RB1) mRNA (Park et al. 2017). RB1 is a negative cell cycle regulator, and excess RB1 arrests cells in G1. This study proposed that SGs promote cell cycle progression via RBFOX2-mediated recruitment and inhibition of RB1 mRNA translation (Choi et al. 2019). Recruitment of mRNA transcripts encoding for proteins involved in proliferation appears to be a general theme of SGs, as gene enrichment analysis of mRNAs in SG cores demonstrated that proto-oncogene transcripts (e.g., ABL2, PDGFRA, GSK3B, RUNX1, AKAP11) were highly enriched (Namkoong et al. 2018). Importantly, this study showed that while distinct stresses showed differences in the variety of mRNAs that were preferentially recruited to SGs, enrichment of proto-oncogene transcripts was shared across stress types. Given that proto-oncogene transcripts are rich in adenylate-uridylate (AU) sequences and consequently subject to mRNA processing and degradation, it has been suggested that recruitment to SGs may promote their stability, expression, and function to promote tumorigenesis (Namkoong et al. 2018). While several lines of evidence support a role for SGs in cancer cell proliferation under stress, studies also indicate that transcripts of both negative and positive regulators of proliferation are recruited to SGs. In addition, several of the protein components of SGs are involved in both negative and positive proliferation pathways. It is not clear how SGs would support cell proliferation by capturing proteins and mRNAs with seemingly opposite functions. It is possible that these components may be recruited at different levels relative to one another and that the sum of all parts ultimately favors proliferation.</p><p>In principle however, depending on cell intrinsic and extrinsic stimuli, the cellular levels of either positive or negative regulators of proliferation, as well as the levels at which they are recruited to SGs relative to one another, can shift. In addition, these transcripts could also be differentially modified through interactions with SG components with roles in mRNA processing and stability. As such, there is perhaps a context specific and dynamic balance between proliferative and anti-proliferative components of SGs. This would suggest that the impact of SGs on proliferation could also depend on context. With this in mind, it is interesting that primary osteosarcoma tumors where SGs were downregulated by shRNA-mediated knockdown of G3BP1 showed no difference in proliferation rate compared to control. Further studies are needed to understand whether this is specific to osteosarcoma or a shared phenotype of all tumors. However, this study may indicate that in the context of tumorigenesis, SGs may aid proliferation in later stages of tumor development or in specific cancer cell subclones, perhaps when a specific threshold of SG formation and SG signaling output is reached.</p><!><p>The relationship between stress granules and cell death is perhaps the earliest studied function of SGs. A number of in vitro studies demonstrate that SGs block the cellular apoptotic machinery that is triggered by stress stimuli in cancer cells, and several SG-mediated anti-apoptotic pathways have been defined. In addition, numerous studies have also shown that SGs are critical determinants of the sensitivity of cancer cells to chemotherapeutic agents.</p><!><p>SGs control live-or-die cell fate decisions along two broad paths. The first is through the sequestration of pro-apoptotic factors, limiting their activity at their target locations. Secondly, SGs curb the production of reactive oxygen species, limiting apoptosis-inducing stress stimuli and cell damage. As discussed above, recruitment of components of the mTORC1 complex allows SGs to prevent mTORC1-hyperactivation-induced apoptosis in cancer cells. In addition, Arimoto et al. reported that SGs inhibit apoptosis by preventing p38 and JNK activation (Arimoto et al. 2008). Specifically, under oxidative stress conditions, SGs recruit RACK1 thus preventing its interaction with the MAPK kinase MTK1, which is required for p38/pJNK mediated apoptosis of cervical cancer cells (Arimoto et al. 2008). The coiled coil containing protein kinase (ROCK1) is another activator of JNK that is recruited to SGs (Tsai and Wei 2010). Sequestration of ROCK1 to SGs in cancer cells prevents the ROCK1-mediated phosphorylation of the JNK-interacting protein 3 (JIP)-3, thus inhibiting JNK activation and the induction of JNK-mediated apoptosis. In addition, translocation of TRAF2 to SGs inhibits TNF-mediated activation of nuclear factor (NF)-kB and apoptosis (Kim et al. 2005). Studies in non-cancerous cells show that SGs can recruit arginylated calreticutin to prevent its translocation to the plasma membrane and apoptotic function during stress; whether this mechanism also occurs in cancer cells remains to be determined (Lopez Sambrooks et al. 2012). More recently, it was demonstrated that macrophages utilize the SG translocation of DEAD-box helicase 3 X-linked (DDX3) to inhibit NLRP3 inflammasome activation (Samir et al. 2019). Activation of the NLRP3 inflammasome induces the secretion of pro-inflammatory cytokines and pyroptosis – a form of inflammatory cell death (Samir et al. 2019). Given the role of macrophages in driving tumor progression, it is tempting to speculate that SGs may promote tumorigenesis by preventing macrophage pyroptosis.</p><p>SGs have been shown to reduce ROS levels and ROS-dependent apoptosis; however, the mechanisms behind the antioxidant activity of SGs are not fully elucidated (Takahashi et al. 2013). One study identified an antioxidant function of the ubiquitin-specific peptidase 10 (USP10) and proposed that SGs may reduce ROS levels by facilitating the activation of USP10. It is currently unknown, however, how SGs may promote the antioxidant function of USP10 and how USP10 functions as an antioxidant (Takahashi et al. 2013). As discussed above, the NRF2 antioxidant pathway has been shown to promote SG formation. Given the role of SGs in regulating ROS levels, it would be interesting to determine whether SGs also impact the antioxidant activity of NRF2 (Mukhopadhyay et al. 2020).</p><!><p>Studies in various in vitro models have explored the relationship between SGs and cancer cell resistance to chemotherapy. These studies have shown that chemotherapeutic agents including bortezomib, cisplatin, etoposide, oxaliplatin, paclitaxel, and sorafenib induce SG formation. A comprehensive review of these studies has been recently published (Gao et al. 2019). This section highlights the most salient aspects of SG-mediated drug resistance.</p><p>A shared feature of all chemotherapeutic agents that drive SG formation is that they do so by inducing the phosphorylation of eIF2α. The exact kinases responsible for eIF2α phosphorylation, however, differ across agents. Sorafenib is a Raf1/Mek/Erk kinase inhibitor that is FDA approved for the treatment of patients with advanced hepatocarcinoma (HCC), renal carcinoma, and metastatic, progressive, and differentiated thyroid carcinoma refractory to iodine treatment. Sorafenib has been shown to induce SGs (Lin et al. 2012; Adjibade et al. 2015). Further studies showed that sorafenib treatment lead to the activation of the unfolded protein response (UPR) and induced SG formation via PERK-mediated phosphorylation of eIF2α (Adjibade et al. 2015; Pakos-Zebrucka et al. 2016; Feng et al. 2017).</p><p>Bortezomib is a proteasome inhibitor that is FDA approved for the treatment of multiple myeloma and mantle cell myeloma. Bortezomib treatment induced SGs in cancer cells of the colon, lung, cervix, head, and neck via HRI-mediated phosphorylation of eIF2α (Fournier et al. 2010; Kaehler et al. 2014; Burwick and Aktas 2017). Furthermore, bortezomib-induced SGs were shown to recruit and promote the degradation of transcripts of the cyclin-dependent kinase inhibitor p21 (WAF1/CIP1). As p21 is a protein that promotes cell cycle arrest and apoptosis, it was proposed that bortezomib-induced SGs lead to apoptosis inhibition and treatment resistance through downregulating p21.</p><p>Chemotherapeutic agents such as 5-fluorouracil (5-FU) cisplatin, etoposide, or oxaliplatin – which are used for the treatment of several cancers including colorectal, pancreas, breast, and head and neck – have been shown to induce SGs. 5-FU induces SG assembly by stimulating PKR-mediated phosphorylation of eIF2α (Kaehler et al. 2014). In all reported instances, SG formation in response to chemotherapeutic agents functioned as a mechanism of resistance to chemotherapy-induced cell death. In addition, inhibition of SGs, or of the kinases responsible for peIF2-α-mediated SG formation, sensitized cancer cells to chemotherapeutic agents.</p><p>Taken together these studies suggest that the blockage of SG formation would enhance chemotherapy cancer treatment. In addition, tumors driven by oncogenic pathways that stimulate SGs such as mutant RAS are well documented as refractory to chemotherapy. The capacity of these pathways to stimulate SGs therefore may also provide mechanistic insight into chemotherapy resistance and identify patients that could most benefit from anti-stress granule therapy.</p><!><p>Invasion of local tissue and spread to distant sites to form metastases is a central feature of cancer and the primary cause of death for >90% of cancer patients (Hanahan and Weinberg 2011). Understanding the biological mechanisms of the metastatic process is crucial in finding successful therapeutic opportunities. The development of metastasis is a complex process that requires cancer cells to leave the local environment, circulate in the bloodstream, and acclimatize and survive the new environment of a secondary site. Consistent with the idea that highly metastatic cells utilize SGs for migration and survival, SGs have been observed in disseminated tumor cells isolated from the bone marrow specimens of breast cancer patients (Bartkowiak et al. 2015). In addition, Somasekharan et al. demonstrated that the metastatic potential of osteosarcoma cells is linked to SG formation (Somasekharan et al. 2015). SG inhibition by shRNA-mediated knockdown of G3BP1 led to an impairment of the invasive and metastatic potential of sarcoma cells in vivo. Formation of SGs in this study was linked to the upregulation of YB-1, which can directly bind to the 5' UTR of G3BP1 mRNA to upregulate its translation. In agreement with these observations, a class I HDAC inhibitor suppressed sarcoma metastasis by enhancing YB-1 acetylation, which blocked the interaction of YB-1 with its mRNA target G3BP1, and downregulated G3BP1 levels and SG formation (El-Naggar et al. 2019). While the mechanisms through which SGs promote invasion and metastasis were unexplained in this study, the authors raised the possibility that SGs might sequester mRNAs encoding for proteins that inhibit invasion and metastasis. Sequestration of these mRNAs to SGs would prevent synthesis of the proteins they encode, thus enhancing the cellular capacity to invade and metastasize. In addition, based on the observation that G3BP1 knockdown reverted the growth pattern of primary tumors to noninvasive borders, this study proposed that SGs facilitate invasive capacity by selectively releasing mRNAs that encode matrix-degrading enzymes for translation.</p><p>Other studies suggest that SGs promote metastasis via inhibiting the ribonuclease inhibitor 1 (RNH1) which promotes metastasis through stimulating the activity of angiogenin (Pizzo et al. 2013). RNH1 is a component of SGs, and downregulation of RNH1 promoted migration and metastasis (Pizzo et al. 2013; Yao et al. 2013). Recruitment of RBFOX2 to SGs has also been shown to promote metastasis of melanoma cells to the lung as inhibiting the localization of RBFOX2 to SGs diminished lung metastasis in a mouse model (Choi et al. 2019). It is currently unknown how RBFOX2 recruitment to SGs may promote metastasis, but selective recruitment or exclusion of mRNAs encoding proteins, which inhibit or promote metastasis respectively, have been proposed as potential mechanisms (Choi et al. 2019). Another study in pancreatic cancer cells proposed that SGs may be implicated in the degradation of mRNA transcripts encoding for Binder of Arl Two (BART), which impairs cell invasion and metastasis by inhibiting ARL2-mediated activation of the RHO small GTPase, which is a key mediator of cell migration and metastasis (Taniuchi et al. 2011a, b). Although direct evidence that SGs contribute to BART downregulation is lacking, given that BART can be degraded by G3BP1, it is possible that SG formation enhances the interaction of BART mRNA with G3BP1 as well as BART degradation to facilitate cell invasion. Studies in noncancerous cells also showed that RHO is both a component and a mediator of SG formation, suggesting that a potential mechanism through which RHO promotes metastasis may be through SG formation (Tsai and Wei 2010). Taken together these studies indicate that while multiple lines of evidence point to a role of SGs in metastasis, further work is needed to identify and characterize the molecular mediators through which SGs may support this process.</p><!><p>Stress adaptation, driven by dysregulated cancer signaling, is a fundamental property of cancer that has yet to be fully elucidated. As reviewed here, evidence from multiple in vitro and in vivo models indicates that oncogenic mutations and dysregulated signaling pathways in cancer modify the canonical molecules that regulate SG formation. By doing so, cancer cells take advantage of SG formation to enhance stress adaptation.</p><p>Oncogenic Ras mutations, hyperactivation of mTORC1 and HDAC, and dysregulation of glycolytic and hexosamine biosynthetic pathways have emerged as key pathways that stimulate SG formation in cancer cells. However, the full scope of oncogenic signaling pathways that may regulate SG formation remains to be established and may include a broader signaling network than is currently known. A recent study indicated that mutations in the E3 ubiquitin ligase binding adaptor SPOP1, which occur in ~15% of primary prostate cancers, led to enhanced SG formation in prostate cancer cells in vitro (Shi et al. 2019). As such, prostate cancers with SPOP1 mutations may be another example of tumors with enhanced SG formation and stress adaptation. Additional metabolic processes may also play an important role in SG formation in cancer. Glutamine deprivation was shown to impact SG formation in pancreatic cancer cells (Mukhopadhyay et al. 2020). However, cancer cells are depleted of several non-essential amino acids with roles in purine/pyrimidine synthesis, protein translation, and glutathione regulation which can impact translation inhibition and cellular stress and consequently SG formation. Lastly, protein levels of SG nucleators are upregulated in several tumors compared to normal tissue raising the possibility that higher levels of free SG-nucleator proteins in cancer may also amplify SG formation (French et al. 2002; Wang et al. 2018).</p><p>The initial view that SGs function solely to store RNA has been offset by a wealth of data that link SGs to several signal transduction and gene expression regulation pathways. In addition, it has been clearly demonstrated that the composition of SGs can vary significantly depending on the type of stress and tissue. The model that has emerged from these studies is that SG levels, composition, and dynamics determine their signaling output. As such, current studies aimed at understanding the role of SGs in cancer and their molecular mediators must address their context-dependent specificities and relevance.</p><p>Much remains to be learned about the cellular processes that SGs regulate in cancer and how they impact tumorigenesis. In addition, it still remains to be understood whether SGs are a feature of all tumors or only those of specific tissues (e.g., pancreatic cancer, osteosarcoma). As stress and the dysregulated signaling pathways described here are common in cancer, the expectation would be that SGs may also be a shared feature for most types of tumors. In the same vein, mTORC1, HDAC, and metabolic pathways are often dysregulated in the tumor stroma which is also exposed to stress stimuli. In addition, evidence that mutant KRAS can promote SG formation in a paracrine manner suggests that cancer cells may also instruct SG formation in the tumor stroma (Grabocka and Bar-Sagi 2016). The question that inevitably arises is whether SGs are present in the tumor stroma and does this impact tumorigenesis. It is also important to note that SGs are part of a larger group of stress-adaptive organelles that are hijacked by cancer cells under stress including macropinosomes, autophagosomes, and lysosomes (Commisso et al. 2013; Perera et al. 2015; Amaravadi et al. 2016). In yeast, SGs are cleared by autophagy, and evidence suggests that targeting of SGs to degradative organelles by autophagy may be conserved in mammalian cells (Buchan et al. 2013; Ryu et al. 2014; Marrone et al. 2018). This raises the possibility that SGs may interact and integrate with other stress-adaptive organelles in cancer. Such interactions could have important implications for the stress adaptation of cancer cells and tumor progression. Future studies aimed at answering these questions can provide important insight into the role of SGs in tumorigenesis.</p><p>Given the evidence that SGs may play an important role in tumorigenesis, it will be essential to develop animal models that assess their tumor-relevant functions. Development of tools for in vivo imaging of SGs in such models may allow visualization of the context-dependent specificities of their formation. Another current challenge is the lack of specific SG inhibitors. Current pharmacological agents that inhibit SGs have broad effects. Additionally, genetic inhibition of SG formation is usually achieved through targeting one or more SG nucleators which, generally, have functions beyond SGs. The understanding of specific interactions or modifications that determine the SG-nucleating capacity of these molecules is critical for the development of tools that allow for the distinction of their SG-specific function and roles in tumorigenesis. The exciting work that lies ahead to fully elucidate the function of SGs in tumorigenesis also has promising therapeutic prospects. Given the well-documented roles of SGs in the chemotherapeutic response, the development of anti-SG therapies has the potential to provide efficacious treatment modalities for cancer patients.</p>

PubMed Author Manuscript

High‐Pressure CO Electroreduction at Silver Produces Ethanol and Propanol

AbstractReducing CO2 to long‐chain carbon products is attractive considering such products are typically more valuable than shorter ones. However, the best electrocatalyst for making such products from CO2, copper, lacks selectivity. By studying alternate C2+ producing catalysts we can increase our mechanistic understanding, which is beneficial for improving catalyst performance. Therefore, we investigate CO reduction on silver, as density functional theory (DFT) results predict it to be good at forming ethanol. To address the current disagreement between DFT and experimental results (ethanol vs. no ethanol), we investigated CO reduction at higher surface coverage (by increasing pressure) to ascertain if desorption effects can explain the discrepancy. In terms of product trends, our results agree with the DFT‐proposed acetaldehyde‐like intermediate, yielding ethanol and propanol as C2+ products—making the CO2 electrochemistry of silver very similar to that of copper at sufficiently high coverage.

high‐pressure_co_electroreduction_at_silver_produces_ethanol_and_propanol

<!>Conflict of interest<!>

<p>S. J. Raaijman, M. P. Schellekens, P. J. Corbett, M. T. M. Koper, Angew. Chem. Int. Ed. 2021, 60, 21732.</p><p>Few electrocatalytic systems are known to be capable of generating carbon‐coupled products from the CO2 reduction reaction (CO2RR) and/or the CO reduction reaction (CORR). [1] Out of these, copper is by far the most capable electrocatalyst for making C2+ molecules, yielding ethylene, [2] ethanol, [3] and n‐propanol [4] as its primary multi‐carbon products.[ 1d , 5 ] Other catalysts (in aqueous media) include molybdenum disulfides, [6] enzymatic nitrogenases with a vanadium/molybdenum active center [7] (and its organometallic homologues [8] ), bimetallic palladium/gold nanparticles, [9] heteroatom (N, B)‐doped nanoparticles, [10] transition‐metal (Ni, Fe)‐doped carbon xerogels, [11] certain surfaces when coated with functionalized films, [12] nickel/gallium alloys, [13] nickel phosphides, [14] and metallic nickel and silver. [15] However, these non‐copper catalysts exhibit comparatively low (on the order of a few %) faradaic efficiencies (FEs) for C2+ products.</p><p>As for the currently existing theories on the C−C coupling mechanism, an in‐depth review concerning non‐copper systems has recently been published by Zhou and Yeo, [16] whilst comprehensive reviews regarding the mechanism on copper can be found, for example, here [17] and in a review by Fan et al. [18] who compare mechanisms on a per‐product basis. For comprehensibility, summaries of the main theories for making C2 and C3 products on metallic Cu in aqueous media are also provided in the Supporting Information (SI) in Schemes A‐C2 to I‐C2 (with, where applicable, reaction paths to C3 products in accompanying Schemes A‐C3 to J‐C3).</p><p>To increase molecular‐level understanding of the formation mechanism for C2+ products, Hanselman et al. carried out density functional theory (DFT) calculations on CO reduction to C2 products for various transition‐metal surfaces (including silver), suggesting two reaction pathways: one to ethylene and one to ethanol, bifurcating from a surface intermediate that is one hydrogen short of acetaldehyde. [19] This mechanism, where acetaldehyde is the precursor to ethanol, agrees with experiments on copper single‐crystal electrodes. [20] Their DFT calculations indicate that, among nine transition‐metal surfaces, only copper has a reasonably low onset potential for ethylene formation whilst ethanol has a slightly later onset. The former agrees well with literature as copper is reported to yield reasonable FE towards C2H4 at overpotentials of a few hundred mV,[ 2a , 21 ] although experimentally no large differences are observed between the formation onsets of ethylene and ethanol.[ 2b , 22 ] Importantly, their calculations also indicate silver should have a lower onset potential for ethanol formation than copper whilst being incapable of producing ethylene. In chemical terms, silver is seemingly too noble to break the last C−O bond.</p><p>This prediction is, however, in apparent disagreement with experimental studies as the maximum reported FE of CO2 to ethanol is ca. 0.1 % on silver vs. 40 % on copper.[ 3 , 15a , 22a ] Hanselman et al. hypothesized this disagreement may be a consequence of CO desorbing rather than reacting further on silver due to its unfavorable adsorption strength. [19] Hence, herein we probe the validity of the theory that silver can produce ethanol if the CO coverage on the surface is sufficiently high. To this end, we study CO reduction at elevated pressure as a means of increasing surface coverage which enhances the likelihood of (intermolecular) reactions involving COads. In line with DFT calculations we observe ethanol (whose formation is positively influenced by increasing the pressure) and no ethylene during CORR. Furthermore, ethylene glycol and n‐propanol are also observed and found to exhibit a similar pressure dependency as ethanol, providing us with additional insight into carbon−carbon bond formation and the mechanistic aspects of C3 production.</p><p>Experiments were carried out in a three‐compartment electrochemical cell inside an autoclave that could be pressurized up to 60 barg, with the gaseous products leaving the cell analyzed by gas chromatography, and liquid products analyzed by NMR. The working electrode was a silver gas diffusion electrode (GDE) with a 1 cm2 exposed geometrical area. Alkaline conditions were employed as these promote C2 formation from CO on copper. [23] A Ag|AgCl|KCl (3 m) reference was used as a reference electrode and potentials are reported on this scale unless denoted otherwise. Reported potentials are not IR‐corrected because of the inherent inhomogeneity of the interfacial potential on a GDE, rendering the nominal reported potentials unrepresentative of the "real" potential. As a figure of merit, the nominal IR‐corrected potential of the most negative potential employed in this work (−4.5 V) was calculated to be ca. −1 V vs. RHE (see SI). A comprehensive description of the experimental setup can be found in the SI, including control experiments conducted in the absence of CO and in the absence of applied potential in the presence of CO to prove that the products we report are indeed the result of electrochemical CO reduction.</p><p>Absolute formation rates of CORR‐related products obtained for CO reduction in 0.5 m KOH on a silver GDE at various potentials are depicted in Figure 1 a–c, for reactant (carbon monoxide) pressures ranging between 10 and 60 barg. Investigated reaction times were between 2.6 and 73 hours, with more positive potentials necessitating longer times to guarantee a minimum of charge had passed. The CORR products depicted in Figure 1 are minority species, with hydrogen and formate (Figure S3a and S3b, respectively) being the main products. As we study the carbon−carbon bond formation mechanism on silver, we will disregard H2 and HCOO− as neither is the result of CO reduction or contains a C−C bond. However, to briefly address the possible origin of formate (being in equal oxidation state as CO), we refer the reader to literature wherein formate is proposed to form through a solution phase reaction between CO and hydroxide, which may occur in this work given the high electrolyte alkalinity and elevated carbon monoxide pressures. [24]</p><p>Color‐coded formation rates for CORR products (methane: red, methanol: blue, acetic acid: green, ethanol: black, ethylene glycol: orange, n‐propanol: pink) plotted as a function of applied potential (non‐IR corrected) for three different reactant pressures; 10 barg (a, d, and g), 40 barg (b, e, and h) and 60 barg (c, f, and i) expressed in absolute rates (a, b, and c) and relative rates (d, g and e, h and f, i). All axes in a given row are of equal magnitude. Not detected products are marked by an "x" in the subfigures depicting relative rates.</p><p>Specifically, the CORR‐related products (Figure 1) comprise a product with carboxylic acid functionality (acetic acid, green), the simplest hydrocarbon (methane, red), and four compounds with alcohol functionality (methanol, ethylene glycol, ethanol, and n‐propanol; blue, orange, black, and purple, respectively). Notably, ethylene, which is very commonly observed on copper electrodes, [1d] was not observed. The predominance of oxygenates (excluding methane) agrees with the DFT predictions of Hanselman et al., who computed silver to be a poor catalyst for breaking C−O bonds. [19] Unconventionally, formation rates rather than partial current densities are depicted in Figure 1. This approach allows for directly comparing molar product ratios, which is valuable from a mechanistic point of view considering certain reaction pathways yielding C2 species (e.g., Cannizzaro disproportionation [25] ) would result in equimolar concentrations of particular types of products. Partial current densities are provided in Figures S4 (for CORR products) and S5 (for hydrogen), whilst the overall current response of the system is depicted in Figure S6. Faradaic efficiencies are given in Table S1.</p><p>Pressure and potential dependencies for these CORR products can be determined from Figure 1 a–c. Overall, formation rates increase when either the overpotential or CO pressure is increased, although formation rates at 10 barg/−3 V and methane formation at 60 barg/−3 V are exceptions. However, because the products' formation rates overlap to a considerable degree, these figures can only provide us with general trends. To better distinguish individual trends, each product has been normalized to its highest observed formation rate and is depicted on a per‐pressure basis in Figure 1 d–f (for methane, methanol, and acetic acid) and Figure 1 g–i (for ethanol, ethylene glycol, and n‐propanol) for 10, 40, and 60 barg from left to right, respectively. The first group (methane, methanol, and acetic acid) comprises products weakly correlating to pressure, potential, and one another whereas the second group (ethanol, ethylene glycol, and n‐propanol) is comprised of products that show fairly straightforward trends that are shared between them.</p><p>The behavior of these latter three higher alcohols yields important insights into the C−C formation mechanism since they all exhibit very similar trends: at the lowest applied pressure and potential (10 barg, −2 V) they are just barely detectable. Then, as the potential is decreased (−3 V) their formation rates go through a maximum and subsequently slightly decrease again for higher overpotentials (−4.5 V). Increasing the CO pressure from 10 to 40 barg results in this maximum disappearing, with observed relative formation rates increasing rapidly as higher overpotentials are applied. However, this potential dependency becomes weaker as the pressure is increased further, with more moderate increases of ca. 5–25 % observed between successively more negative potentials at CO pressures of 60 barg.</p><p>Exhibiting such strong similarities in their potential and pressure dependency indicates commonalities in their formation mechanism, separate from the pathway via which methanol and acetic acid form (to be discussed later). The absence of ethylene (which cannot be explained by insufficient hydrogen coverage, considering the still high rate of H2 formation) in concert with the comparable behavior of ethanol and n‐propanol is especially interesting. Namely, this observation makes it unlikely that the coupling of CO and ethylene ("hydroformylation") is responsible for the formation of C3 products on silver, as hypothesized to occur on copper by Ren et al. [26] Instead, acetaldehyde, being both reactive and difficult to detect via standard NMR techniques (especially in alkaline media), [27] is known to only reduce to ethanol and not ethylene (on copper). [28] Its high reactivity would facilitate further reduction rather than desorption. This possibility would agree with recent work by Xu et al. who showed that propanol is formed on copper via the coupling between CO and a surface‐bound methylcarbonyl, an intermediate which is one hydrogen short of acetaldehyde. [28b] This latter observation agrees well with DFT calculations conducted by Hanselman et al., who propose ethanol formation takes place via a surface‐bound acetaldehyde species. [19]</p><p>The fact that both ethylene glycol and ethanol are observed and exhibit similar behavior proves that silver is capable of breaking one of the C−O bonds in a molecule comprised of two carbon atoms containing two C−O bonds. However, the absence of ethylene shows that silver is indeed a poor catalyst for breaking the final C−O bond, as predicted by DFT calculations. From these observations, our results suggest that an oxygenated intermediate, probably surface‐bound methylcarbonyl (as proposed by Hanselman et al. and Xu et al.),[ 19 , 28b ] is involved in the formation of ethanol, as well as in the coupling with adsorbed CO to lead to the formation of n‐propanol (through propanal).</p><p>Additional insights regarding C−C coupling on silver can be derived from the behavior of the other "group" of products (methane, methanol, and acetic acid) whose trends with regards to potential, pressure, and one another are more inconsistent. Of these, the methane "trends" disagree with all other observed CORR products. The most notable observation that can reasonably be made is that it is more prevalent at increased CO pressures and more cathodic potentials. More important are methanol and acetic acid, as they exhibit some similarities although their correlation is much weaker than the previously discussed alcohols. Comparing these products, we find that methanol generally exhibits higher relative formation rates than acetic acid at lower overpotentials, and for all investigated potentials in the case of 10 barg of CO pressure. However, when the pressure is increased (from 10 to 40 or 60 barg), relative acetic acid formation rates start to become very similar to those of methanol formation for the most cathodic potentials investigated (−4.5 V). This results from the fact that methanol formation rates are relatively invariant with potential and pressure, whereas acetic acid is strongly influenced by both of these parameters. (This observation that acetic acid formation remains strongly potential dependent also at increased pressures is what makes its behavior different from the previously discussed "alcohol group" as they exhibit much weaker relative increases in formation rate with potential at 60 barg of CO.)</p><p>The strong pressure dependency of acetic acid suggests that CO is involved in its formation. Furthermore, the fact that this dependency persists even at elevated reactant pressures signifies that the C−C coupling step for its formation has a significant barrier. Additionally, the (weak) correlation observed between methanol and acetic acid can be interpreted as them sharing a common intermediate. Hence we speculate there may exist a pathway where CO couples with a methanol‐like moiety to form acetic acid. Some plausibility for this hypothesis can be derived from the existence of a rhodium‐catalyzed industrial process for acetic acid synthesis involving the carbonylation of methanol called the Monsanto process. [29] However, we emphasize that the most important observation from Figure 1 is that the pathway for the formation of acetic acid differs from the pathway via which ethanol, ethylene glycol, and n‐propanol are formed.</p><p>In summary, high‐pressure CO electroreduction experiments reveal that silver is capable of further reducing carbon monoxide if the CO surface coverage is sufficiently high, with the total production rates of C2+ CORR products (ethanol, ethylene glycol, and propanol) increasing as the pressure is increased. Contrary to one literature report, [15b] ethylene formation was not observed in this work. The fact that silver is capable of reducing CO to ethanol but not to ethylene is in agreement with DFT calculations. [19]</p><p>The comparable potential and pressure dependence of the formation of ethanol, n‐propanol, and ethylene glycol indicates a commonality in their formation pathways. An oxygenated surface species is likely to be the shared intermediate between ethanol and n‐propanol, and this species is likely to be one hydrogen short of acetaldehyde, as suggested by Hanselman et al. and Xu et al.[ 19 , 28b ] We propose it is the coupling of this species with adsorbed CO that is responsible for the formation of propanal, which is then further reduced to n‐propanol, as opposed to a reaction between a surface‐bound ethylene molecule and carbon monoxide (Figure 2).</p><p>Proposed mechanistic pathway based on literature and the products (and their trend similarities) observed in this study.</p><p>If the CO coverage is sufficiently high, as can be achieved by increasing CO pressure, the product spectrum of silver starts to resemble that of copper under CO2RR conditions. [5] However, the formation rates for CORR products on silver are orders of magnitude lower than what is observed on copper, making detecting minority products beyond the scope of this work. The main difference between the two systems seems twofold. Firstly, due to the rather unfavorable adsorption energy of CO, silver has the propensity for desorbing CO rather than reducing it further, even though thermodynamically speaking it is capable of doing so. Secondly, due to silver being a poor catalyst for breaking C−O bonds, [19] no ethylene (nor ethane) formation is observed although the rest of the products observed compare favorably with copper‐catalyzed CO(2) reduction.</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>Supporting Information</p><p>Click here for additional data file.</p>

PubMed Open Access

Noncovalent Interaction Analysis in Fluctuating Environments

Noncovalent interactions play a central role in many chemical and biological systems. In a previous study, Johnson et al developed a NonCovalent Interaction (NCI) index to characterize and visualize different types of weak interactions. To apply the NCI analysis to fluctuating environments as in solution phase, we here develop a new Averaged NonCovalent Interaction (i.e., aNCI) index along with a fluctuation index to characterize magnitude of interactions and fluctuations. We applied aNCI for various systems including solute-solvent and ligand-protein noncovalent interactions. For water and benzene molecules in aqueous solution, solvation structures and the specific hydrogen bond patterns were visualized clearly. For the Cl\xe2\x88\x92+CH3Cl SN2 reaction in aqueous solution, charge reorganization influences over solvation structure along SN2 reaction were revealed. For ligand-protein systems, aNCI can recover several key fluctuating hydrogen bond patterns that have potential applications for drug design. Therefore, aNCI, as a complementary approach to the original NCI method, can extract and visualize noncovalent interactions from thermal noise in fluctuating environments.

noncovalent_interaction_analysis_in_fluctuating_environments

1. Introduction<!><!>2.2 The aNCI analysis<!><!>2.3 aNCI combined with QM/MM MFEP simulations<!>3. Computational details<!>4.1.1 The aRDG definition: s(r)\xc2\xaf vs. \xe2\x8c\xa9s(r)\xe2\x8c\xaa<!>4.1.2 Electron density: ab initio vs. promolecular<!>4.2 Case I: solute-solvent systems<!>Water-water interactions<!>Benzene-water interaction<!>4.3 Case II: solvation structures during SN2 reaction<!>4.4 Case III: Ligand-protein binding interactions<!>4.4.1 BIIB021 interaction features<!>4.4.2 VHD interaction features<!>5. Conclusion

<p>Noncovalent interactions play a predominant role in chemistry and biochemistry1,2. For instance, such interactions are the driving force to fold3 and stabilize protein structures4, to coil the DNA into double helix and to self-assemble molecules5. In our previous studies6,7 a novel NonCovalent Interaction (NCI) index was proposed to characterize NCI and estimate their strength. This index is a valuable tool to study weak interaction within biological systems8,9. Using a promolecular10,11 density, NCI can characterize large systems with low computational cost and with insightful results. NCI was also applied to metal complexes12,13 and visualized binding modes of the cations. Influence of noncovalent interactions on the reaction mechanism was also investigated in several studies8,14 using the NCI analysis. For instance, Gillet et al.15 showed that an interpretative analysis crossing Electron Localization Function (ELF) and NCI results could allow following every step of a reaction mechanism to reveal reaction details. Finally Contreras et al.7 linked NCI index with the interaction energy for the specific case of hydrogen bonding. In this study they proposed a scheme to integrate the density on the NCI surfaces and showed that using this approach, the NCI integrated density could give accurate estimation of the interaction energy. This study made a first quantitative link between NCI analysis and interaction energy.</p><p>One limitation of the NCI analysis is that the noncovalent interactions are characterized based on one single structure. However, geometric fluctuations are constantly present in practical molecular systems. For example, in solutions, the positions of solvent molecules fluctuate and the solvent molecules change between solvation shells, which play important roles in solvation and chemical reactions. As such, the application of the NCI index for fluctuating systems is still unclear. In this work, we develop an averaged NonCovalent Interaction index (aNCI), which uses an ensemble of the structures, to overcome this problem.</p><p>This paper is organized as follows: In the next section, we briefly review the NCI analysis and introduced the aNCI analysis. We also explain how aNCI can be readily combined with classical or QM/MM simulations such as the QM/MM-MFEP (Quantum Mechanics/Molecular Mechanics Minimum Free-Energy Path16,17) method. In section 3, we provide the details of simulations and model systems. In section 4, we elucidate how to choose between different averaged reduced density gradient definitions. Then, using the aNCI index and fluctuation index, we scrutinize three cases, including solvation structures, an SN2 reaction, and ligand-protein binding environments. We demonstrate the strength and robustness of aNCI analysis to characterize noncovalent interactions in fluctuating systems. Finally, we conclude our works in section 5.</p><!><p>Blue for the highly attractive interactions (such as hydrogen bonds);</p><p>Green for the weak interactions (such as dispersive-like van der Waals);</p><p>Red for repulsive interactions (such as steric clashes).</p><p>Definition a): using averaged density ρ(r)¯ and averaged density gradient ∇ρ(r)¯, one can define: (2)s(r)¯=12(3π2)1/3|∇ρ(r)¯|ρ(r)¯4/3 </p><p>Definition b): using RDG of each single structure, one can define: (3)〈s(r)〉=∑i=1Nsnapshotssi(r)/Nsnapshots si(r)=12(3π2)1/3|∇ρi(r)|ρi(r)4/3 </p><!><p>To illustrate how thermal motions can affect the weak interactions, we define the thermal fluctuation index as (4)f(r)=std({ρi(r)})mean({ρi(r)}) where ρi(r) is the electron density from structure i.</p><!><p>Blue for highly stable interactions, which can be barely affected by thermal motions;</p><p>Red for flexible interactions, which can be easily distorted by thermal motions;</p><p>Green for fluctuations between blue and red types.</p><!><p>In principle, one can use Eq. (2) and (4) to carry out the aNCI analysis for any systems with thermal motions. However, some technical problems can make aNCI unfeasible. For instance, all the structures generated by molecular dynamics simulations are required to be aligned based on some criteria such as the minimization of root mean square deviations. This alignment process can cause artificial bias in the aNCI analysis. Hence, we partition the entire system into the subsystem (that is the targeting region analyzed by aNCI, such as solute in solution) and the environment (that is the surrounding regions of the subsystem, such as solvent). The subsystem structure is fixed at an optimized structure in the aNCI analysis while the environment fluctuates. Therefore, the aNCI analysis needs a representative subsystem structure and an ensemble of structures for fluctuating environment.</p><p>Since aNCI is an analysis technique based on given system conformations, it is possible to interface the aNCI analysis with any classical or QM/MM simulation methods. In this work, we incorporate the recently-developed quantum mechanics/molecular mechanics minimum free energy path (QM/MM-MFEP) optimization technique into the aNCI analysis. QM/MM-MFEP has been applied to solvation reactions and enzyme systems8,17,1920–22. In QM/MM-MFEP, the subsystem is described by QM while the environment is simulated by classical force fields. The QM/MM-MFEP optimized structure of the subsystem is ensemble-averaged since the subsystem region is optimized over the potential of mean force surface, which is defined by, A(rQM)=−1βln(∫drMMexp(−βE(rQM,rMM))) where E(rQM, rMM) is the total energy of the entire system expressed as a function of the Cartesian coordinates of the QM and MM subsystems. The QM/MM interaction energy in E(rQM, rMM) includes the electrostatic interactions from classical point charges and van der Waals (vdW) interactions between QM and MM subsystems. The integration of QM/MM MFEP into the aNCI analysis is efficient and intuitive. The aNCI analysis can also be combined with any type of molecular dynamics simulation. One just needs to fix the solute molecules at an optimized geometry and carry out the molecular dynamics simulation with the solvent molecules.</p><!><p>We carried out the aNCI analysis on systems that are summarized in Table I, including single molecule solvation, SN2 reaction in water, and ligand-protein binding systems. In QM/MM simulations, the subsystem was treated as quantum mechanics at a B3LYP/6-31+G* level23. CHARMM22 force field24 and TIP3P water model25 were employed for environments. The protein systems were prepared with MolProbity26. Each system was optimized by QM/MM-MFEP with a 640 ps simulation. The 320 ps simulations then were performed to generate the snapshots for the aNCI analysis. 1000 snapshots were generated for each system.</p><p>The aNCI analysis was carried out with modified version of NCIPLOT software12. To reduce the computational cost, the promolecular density6 from atoms within a 15 Å radius cutoff of subsystem were chosen to compute the indices using Eq. (2) and (4). The cube grids, with a 0.05 (if not further mentioned) Å step size along x, y, and z directions, were generated with 3.0 Å buffer regions from the subsystem. In the 3D visualization process with VMD, the threshold for RDG/aRDG was 0.25, and the color scale was [−2, 2], [0, 1.5] for aNCI analysis and fluctuation index, respectively.</p><!><p>To validate two possible definitions for aNCI, we used promolecular density to characterize the noncovalent interactions of one water molecule in water. As shown in Figure 1, s(r)¯, defined in Eq. (2) approaches zero at small density regions and the spikes can be clearly identified. In contrast, the 〈s(r)〉, defined in Eq. (3) does not have (close) zero RDG regions and only two spikes can be observed. According to the original NCI paper6, such spikes represent noncovalent interaction regions. These indicate the microscopic details of interactions are lost in 〈s(r)〉. This phenomenon can be further explained by a detailed analysis over the RDG distribution. Three 3D grids, located at three ICPs (with effective density as −0.04, −0.02, and 0.005, which will be observed later to correspond with hydrogen bond donor, hydrogen bond acceptor, and vdW regions, respectively) under s(r)¯ definition were selected. For each grid, 1000 RDG values were calculated from each snapshot and their distributions were shown in Figure 2. Some interactions were blurred by the broad RDG distribution, although these interactions do exist under averaged density and gradient sense. This suggests that large thermal fluctuations of unstable interactions can bury the useful information of aNCI in 〈s(r)〉. Therefore, we chose the definition s(r)¯ for our aNCI analysis.</p><!><p>To examine how promolecular and ab initio electron densities affect the aNCI analysis, we compared the computed RDGs using both electron densities. The ab initio density is constructed using density functional theory calculations with B3LYP/6-31G* basis set over a small rectangle water box (around 200 atoms) with 5.0 Å buffer zone to the QM water molecule for each snapshot. As shown in Fig. 3, two aRDG plots (black and red dots) against effective density are similar in terms of overall shapes. Furthermore, the absolute electron density at critical points is slightly smaller in ab initio calculations (0.032 and 0.019) than promolecular results (0.039 and 0.021). Therefore, promolecular density is qualitatively accurate to perform the aNCI analysis, which is also confirmed in previous single snapshot NCI analysis.</p><!><p>In Case I, we applied the aNCI analysis to two systems: water in water and benzene in water, which represent prototypical examples of highly fluctuating systems. While the normal NCI analysis shows information about time dependent noncovalent information and so doesn't reveal solvation interactions (see Fig. 5). s(r)¯ does. Figure 4 illustrates the evolution of the aRDG as a function of the number of snapshot, 1000 snapshots can achieve the converged aRDG at low-density regions in terms of number and positions of spikes. In the water-water system (Figure 4-a), the effective density value at the most negative ICP is −0.06 with 1 snapshot, −0.04 with 10 snapshots, and is converged with −0.038 under 100 and 1000 snapshots. In benzene-water system, the effective value at the most negative ICP increase from −0.038 with 1 snapshot to −0.014 with 1000 snapshots. These modifications of ICP's effective density indicate that the use of an ensemble of structures influence the interaction strength. For example, the actual hydrogen bonding would not be as strong as it can be in a single snapshot because fluctuations exist. More importantly, the visualized aNCI pictures can reveal the microscopic solvation structures as shown in Figure 5 for water-water and benzene-water system, respectively.</p><p>Figures 5-a) and 5-c) represent the single snapshot NCI analysis for both systems. In the NCI analysis, an increased number of weak interactions are observed which results in a nonsymmetrical geometry for the ICP's and in a concealment of the important interactions. They all are averaged out in aNCI calculations. Therefore, in aNCI, the main (or important) interactions are clearly visualized (as shown in Figure 5-b) and 5-d)): water-water hydrogen bond and benzene-water π-hydrogen bond. These comparisons explicitly show that aNCI is required in fluctuating environment. The detailed interactions of both systems are discussed below.</p><!><p>Many studies on water structures have been carried out27,28. In ideal condition, water molecules interact with each other through two hydrogen atoms as hydrogen bond donors and oxygen atom as hydrogen bond acceptor. The solute oxygen atom interacts with two other hydrogen atoms from other waters and a tetrahedral hydrogen bond pattern is formed. Both hydrogen bond donors/acceptors should have the same interaction strength due to its symmetry. The aNCI analysis results, which position the four hydrogen bond ICPs under tetrahedral shape, are consistent with the experimental symmetry. In addition, the hydrogen bond donor is observed to have larger effective density (−0.037) than the hydrogen bond acceptor (−0.021), with the negative sign indicating that both regions are attractive interactions between subsystem and environment. When hydrogen bond network forms, environment water hydrogen atoms can interact with subsystem water oxygen atom from any direction, and this fluctuation is reflected on the decreased density value. As pointed out by Kumar et al29, the average hydrogen bond number per water molecule in liquid state is 3.2 ~ 3.6 (less than 4), this corresponds to the strength decrease of the two hydrogen bond acceptors in our analysis. The aNCI picture is also similar with a spatial position function in the water-water environment as described in Ref. 30. This demonstrates the advantages of aNCI in revealing the spatial distribution, the interaction type and strength.</p><!><p>The benzene-water system has also been studied previously31,32. After the interesting π-hydrogen bonding31,32 was proposed, many theoretical and experimental studies33–39 have been carried out. The π-hydrogen bonding was found to be at both sides of benzene ring, and the benzene molecule was shown to act as a hydrogen bond acceptor. The interaction strength was much smaller and flexible than a water-water hydrogen bond networks. The π-hydrogen bond was also found to be highly unstable in nature. In the aNCI analysis, the π-hydrogen bond structure of benzene in 3D space appears clearly for the first time. The overlap between repulsive (effective density 0.013) and attractive (effective density −0.014) contributions have similar nature as the solute water oxygen atom (hydrogen bond acceptor): since the π-hydrogen bonding region is fairly broad, steric clash contributes to the positive effective density value.</p><p>The fluctuation indexes of both systems were depicted in Figure 6. For water-water system, the lowest fluctuations are encountered around the hydrogen bond donor interaction region (0.30 %). Therefore, these interactions are mostly stable compared to the hydrogen bond acceptor ones (0.61 %). Finally, the vdW interactions appear to be the most flexible ones (~1.2 %). For benzene-water system, the relative stability of interactions follows: π-hydrogen bonding (0.67%) > benzene planar vdW interaction (0.95%) > close to π-hydrogen bonding vdW interactions (1.3%). In these cases, the stronger the interaction is, the smaller the fluctuations are. It is also obvious that the π-hydrogen bonding is more flexible than all the polar interactions in water-water system, which is consistent with the experimental observations40,41. The fluctuation index is demonstrated to be physical and could be used to analyze the rigidity of different interactions.</p><!><p>In this case, the aNCI analysis is used to analyze the solvation effect through simple SN2 reaction of Cl− + CH3Cl -> CH3Cl + Cl−. The transition state is taken from a reaction path generated using QM/MM-MFEP, and the reactant state is fully optimized. In Figure 7, the aRDG vs. effective density is plotted for both reactant and transition states. The aNCI interactions and its fluctuation index between the subsystem and environment were visualized in Figure 8.</p><p>In Figure 7, along the nucleophile attack towards to the transition state, the spikes are getting narrower and the density of the strongest interactions (both repulsive and attractive) decreases. Both effects may be caused by a rearrangement of the electron density (and therefore the atomic charges) within the subsystem. Indeed, in the reactant state, the charge of the attaching/leaving Cl is −1.0/−0.23 a.u., respectively, while in the transition state, both of them have −0.71 a.u. and small positive charge migrates on the hydrogen atoms. The surrounding water molecules respond to the substrate charge change, and the radial distribution function (RDF) between Cl and water hydrogen atoms has been plotted in Figure 9. In the transition state, both Cl has a similar RDF with environmental water hydrogen atoms, which is consistent with the charge equity. In the reactant state, the attaching Cl attracts more first shell water hydrogen atoms than the transition state Cl. That indicates with a more negative charge, the interactions between ion-water pair are stronger, and our fluctuations index agrees with the understanding that the stronger interaction is less fluctuating. Although our approach is based on the static solute conformation, a prevoius ab initio molecular dynamics study42 on the same system has observed that the static and dynamics approaches have similar global features, and this further validates our conclusion.</p><!><p>Crystal structures are commonly used to characterize binding pockets and positions of specific ligands to guide drug design. However, a direct observation over crystal structures would be subjective and it could be insufficient to reveal critical binding interactions between ligand and protein. Furthermore, crystal structure is static and may not be helpful to extract binding information. Here, aNCI analysis is applied to analyze the binding patterns of ligand-protein systems.</p><p>We chose two ligands in our study. The first one is a pre-clinical drug molecule, BIIB02143–45. It is a small molecule inhibitor of the heat shock protein Hsp90 that binds competitively with geldanamycin in the ATP-binding pocket of Hsp90. The molecular structure is taken from PDB: 3O6O. The second one is an in silico drug design intermediate compound in ligand-protein binding study46, named VHD as from structure PDB: 2XAB. Both ligands have a purine-scaffold and aromatic moiety. The bound proteins, Hsp83 and Hsp90, have similar global folding and active site structures. The interactions between substrate and environments have been scratches in Figures 10-a) and 10-b). In both Figures, all polar atoms with a distance less than 4.0 Å from ligand are shown, with no explicit hydrogen atoms.</p><p>In both systems, the substrate occupies the same active site with different orientations. Two polar amino acids, Asp and Thr both directly and indirectly interact with the two substrates: BIIB021 over the purine-scaffold N1 and N2, and VHD over the aromatic moiety O2 and linkage O3. Three crystal water molecules were observed inside the active site for both systems, and they are confined in hydrogen bond networks. The BIIB021 aromatic moiety largely locates within the active site and surrounded by many nonpolar residues: Phe123, Tyr124, Trp147, Leu92, Leu88, and Val135. In contrast, most of the VHD purine part is solvent accessible, which may indirectly stabilize the substrate binding.</p><p>To reveal more information based on crystal structures, we carried out QM/MM-MFEP optimization for both systems. The optimized substrate structures have only 0.148 and 0.161 carbon atom root-mean-square-deviation with respect to those in crystal structures, respectively. This indicates QM/MM-MFEP optimization conserve the ligand binding poses. Based on this subsystem structure, the aNCI analysis with 1000 snapshots is calculated. The 3D aNCI density and aNCI fluctuation index for both substrates are shown in Figures 11 and 12, and the ICP effective density and fluctuation index are listed in Table II and Table III, respectively.</p><!><p>Polar interactions between BIIB021 and environments are summarized in Table II, and can be visualized in Figure 11. The two ICPs of N1 atom correspond to strong Hydrogen Bonds (HB) with Asp83 (−0.048) and a crystal water molecule XWAT1 (−0.031). The average HB distances are 1.73 Å and 2.01 Å, respectively. N2 behaves as the HB acceptor for both Thr169 (−0.016) and XWAT1 (−0.016). A detailed analysis over the 1000 structures demonstrates that the HB constantly switches between XWAT2-N2 and Thr169-N2. This is so frequent that, in average, N2 shares HBs with both groups. This effective density is about half as the N1 atom interactions, which goes along with the averaged HB distance between N2 and XWAT2 (2.62 Å), Thr169-OH group (2.87 Å). N3 has only one HB with XWAT3 (−0.031), which is consistent with the crystal structure observation. This water position is quite rigid with an averaged HB distance 2.03 Å. N4 atom, which has XWAT4 nearby from crystal structure, forms two directional hydrogen bonds with bulk water molecules (During the MD simulations, several water molecules moved around N4 atom and no stable HB exists). This difference between crystal structure observation and aNCI results may be caused by the local environment, which makes water molecules favor a two-side interaction direction, not along the co-plane of purine ring. On the other hand, N6 atom has only one strong interaction with bulk water molecules, and this is because the local environment around N6 atom restricts the bulk water molecules to approach it from this single direction. Besides aNCI density, the fluctuation index over these hydrogen bonds is observed to be close related with their interaction strength. For N2 and N4, the fluctuation index are 0.55%, 0.58% and 0.47%, 0.60%, respectively, and their effective densities are about half comparing with N1 and N3. The weaker the effective density amount is, which may be caused by interaction with versatile waters, the more flexible these interactions are.</p><p>The π-π stacking between aromatic moiety and Phe123 side chain creates the most stable vdW interaction with fluctuation index around 0.25%, which is even less fluctuated than the strongest hydrogen bond. The aromatic moiety ring appears therefore rigid and suitable for substrate binding. The purine ring's interaction with Met83 and Asn36 are also stable, with fluctuation index around 0.35%. However, since the latter are comparably weaker than HBs, the contribution to substrate binding may be limited.</p><!><p>The interactions between VHD and environment are listed in Table III, and visualized in Figure 12. The O2 atom is tightly bounded to Asp93. Asp93 not only directly stabilizes the O2 atom, but it also helps to confine the crystal waters XWAT2 and XWAT3. The O1 atom is hydrogen bonded with XWAT1 which is stabilized by Leu48, and this interaction is the second strongest polar interaction. O3 atom has two HBs: one with Thr184, and another with XWAT3, which is it stabilized by Asp93. The aNCI analysis is consistent with direct crystal structure observations, and it offers more information on interaction strength and stability.</p><p>Besides the polar interactions, the purine-scaffold ring makes a hydrophobic interaction with Ala55 which has effective density value of −0.009. A hydrophobic interaction between the aromatic ring and Asn51, Ser52 backbone has a similar strength and confine the substrate inside the active site.</p><p>According to our aNCI analysis, the purine scaffold is accessible to bulk solvent and only has fluctuated weak interactions with protein. If the binding mode does not change with the chemical modifications, the C5 atom of purine scaffold could be a promising site. Indeed, as found in experimental studies47, modifications over this region help to improve the biological property without disturbing the binding ability. In contrast, the modifications carried out on the aromatic moiety (change of O1 and isopropanol groups) can dramatically decrease the binding ability.</p><p>From aNCI analysis, both ligands were observed to bind the protein very well: the BIIB021 utilizes mainly nitrogen atoms to form hydrogen bond with the crystalized water and protein residues; the VHD has three oxygen atoms that form strong hydrogen bond with environment. The Asp residue is in a crucial role for the binding conformation: it not only forms direct hydrogen bond with substrate, but also helps the creation of an internal water hydrogen bond network. With these confined water molecules, the substrate can bind the active site strongly. A further design for potential drug molecule would maintain the critical interactions revealed by aNCI while ligand modifications should be performed on those unstable and weak interaction parts.</p><!><p>In this work, the NCI analysis is generalized to aNCI analysis, which characterizes the interactions from an ensemble of structures. With the aNCI density and aNCI fluctuation index, both averaged noncovalent interactions and fluctuations can be directly visualized. For solute-solvent system, the tetrahedral hydrogen interaction network in water-water, and the π-hydrogen bond in benzene-water system were characterized. The π-hydrogen bond appears as a weaker and more flexible interaction comparing with water-water counterpart. In SN2 reaction, a reorganization of partial charges changes the solvation structures, and as well as the interaction strengths with environment. Further applications in ligand-protein binding systems reveal the complex interaction networks of two drug molecules in heat shock protein active site. With aNCI analysis, the crystal structure was further illustrated and it could help the drug design process.</p><p>The aNCI analysis is general and can be coupled with any type of molecular dynamic simulation (QM, QM/MM, polarizable and classical MM force fields). In this work it was interfaced with the recently developed QM/MM-MFEP approach with classical force field. However, in system where it is difficult to represent by classical force field, such as metal cations, pure QM or polarizable force fields would play a critical role. The further works are currently ongoing to study the influence of these methods used on the aNCI results. Overall, aNCI is a useful tool to characterize noncovalent interaction patterns in fluctuating environments.</p>

PubMed Author Manuscript

Identification of FDA-approved drugs that computationally bind to MDM2

The integrity of the p53 tumor suppressor pathway is compromised in the majority of cancers. In 7% of cancers, p53 is inactivated by abnormally high levels of MDM2\xe2\x80\x94an E3 ubiquitin ligase that polyubiquitinates p53, marking it for degradation. MDM2 engages p53 through its hydrophobic cleft and blockage of that cleft by small molecules can re-establish p53 activity. Small molecule MDM2 inhibitors have been developed, but there is likely to be a high cost and long time period before effective drugs reach the market. An alternative is to repurpose FDA-approved drugs. This report describes a new approach, called Computational Conformer Selection, to screen for compounds that potentially inhibit MDM2. This screen was used to computationally generate up to 600 conformers of 3,244 FDA-approved drugs. Drug conformer similarities to 41 computationally-generated conformers of MDM2 inhibitor nutlin 3a were ranked by shape and charge distribution. Quantification of similarities by Tanimoto combo scoring resulted in scores that ranged from 0.142 to 0.802. In silico docking of drugs to MDM2 was used to calculate binding energies and to visualize contacts between the top-ranking drugs and the MDM2 hydrophobic cleft. We present 15 FDA-approved drugs predicted to inhibit p53/MDM2 interaction.

identification_of_fda-approved_drugs_that_computationally_bind_to_mdm2

Introduction<!>Creating the nutlin 3a-conformer model<!>Conformer generation and rapid overlay chemical structure software (ROCS) program<!>Docking measurements<!>A structural model for nutlin 3a bound to MDM2<!>Generation of nutlin 3a conformers<!>Screening for drugs that can adopt the similar conformations as nutlin 3a<!>Bepridil<!>Conclusion

<p>The p53 tumor suppressor is a transcription factor that enhances transcription of genes required for DNA repair, cell cycle arrest, apoptosis and angiogenesis in response to cellular stress [1]. The p53 gene is mutated in a wide range of human cancers at an overall frequency of approximately 50% [2]. In cancers with wild-type p53 genes, defects in the p53 pathway have been observed [3]. One defect is the overexpression of MDM2, a negative regulator of p53. MDM2 overexpression is due to gene amplification at an overall frequency of 7% of human cancers [4, 5]. MDM2 overexpression can occur in cancers without gene amplification due to increased transcription and increased expression from MDM2 promoter elements responsive to Smad3/4 and SP1 [6, 7]. MDM2 is an E3 ligase that binds to and ubiquitinates p53, targeting the tumor suppressor for degradation by the 26S proteasome [8–10]. As part of its feedback pathway, p53 mediates transcription of the MDM2 gene, the product of which maintains p53 at relatively low levels in normal proliferating cells [11]. If cells are stressed (by, for example, DNA damage or oncogene activation), p53 and MDM2 become phosphorylated and interact with factors that inhibit p53-MDM2 complex formation. In the absence of bound MDM2, the p53 level rises in the cell nucleus and activates transcription of appropriate genes that ultimately prevent cancer. When the p53 level is suppressed, cells undergo DNA replication in spite of the fact that the genome is damaged resulting in mutations [12]. If the mutations deactivate tumor suppressors or activate proto-oncogenes, cells can become incapable of responding to growth-restraint signals and cancer ensues.</p><p>In some cancers with a wild-type p53 genotype, abnormally high MDM2 levels accumulate and inhibit the p53 pathway. This scenario occurs in up to 20% of soft tissue tumors [4] and with an overall frequency of 7% in 19 different tumor types [5]. One approach to combating these cancers is to relieve the suppression of wild-type p53 by inhibiting MDM2. Early studies using antisense oligonucleotides showed that MDM2 may be a viable druggable target [13].</p><p>One well-studied small molecule MDM2 inhibitor is nutlin 3a, a cis-imidazoline compound with an IC50 of 90 nM for inhibition of p53 binding to MDM2 [14, 15]. Nutlin 3a treatment of wild type p53-expressing cultured cancer cells increases p53 levels and causes p53-mediated transactivation, cell cycle arrest, and apoptosis [16]. However, the magnitude and selectivity of these p53 downstream effects are variable [15]. Nutlin 3a and other small molecules that target the N-terminal p53 binding pocket of MDM2 decrease the rate of tumorigenesis of wild-type p53-expressing cancer cells in nude mouse-bearing human cancer xenografts [14, 16, 17].</p><p>Nutlin 3a was isolated by screening a library of cis-imidazoline compounds for inhibition of p53 binding to MDM2 by surface plasmon resonance [14]. Nutlin 3a belongs to a class of compounds composed of two halogen-derivatized phenyl rings in cis configuration attached to positions 4 and 5 of an imidazoline scaffold. Roche, the manufacturer of these compounds, has a nutlin 3a-like compound called RG7112 in Phase I clinical trials for the treatment of advanced solid tumors and hematologic cancers [18]. Although nutlins are a promising start for drug development, there may be many hurdles to overcome prior to approval for clinical use. An alternate approach to development of novel MDM2 inhibitors is to determine if FDA-approved compounds can be repurposed as MDM2 inhibitors [19].</p><p>In this report, we present an in silico approach to drug repurposing, called Computational Conformer Selection (CCS), to screen and identify compounds with similar shape and charge distribution as nutlin 3a across different nutlin 3a conformers. Conformers of FDA-approved drugs are computationally generated and compared with the shape and charge distribution of nutlin 3a. Highly ranked drugs are docked to the N-terminal p53-binding domain of MDM2 using the Autodock Vina docking program [20] yielding binding energies. Here we report the ranked order of the 3,244 drugs from the ZINC database of FDA-approved drugs [21].</p><!><p>Nutlin 2 from PDB #1RV1 structure was used as the starting point. The nutlin 2 isopropyl group was substituted for the ethyl group at position 2 on the methoxy-phenyl ring. Chlorine atoms were substituted for the bromine atoms at position 4 on the two phenyl rings. A ketone oxygen was added to position 3 of the piperazine ring and the ethanol group was removed from position 4 of the piperazine ring.</p><!><p>Conformers of nutlin 3a and ZINC database drugs were created by OMEGA with the maximum number of conformers for each drug set to 600 (-maxconfs 600) [22]. Rapid Overlay of Chemical Structures (ROCS) software was used to measure similarities between nutlin 3a and drug conformers [23, 24]. The number of random starting points for compound alignments was set to "-randomstarts 40". Default parameters of ROCS were changed to "rank by ColorTanimoto" to emphasize charge distribution. Preliminary analysis of ROCS-mediated FDA drug comparisons to nutlin 3a showed that the Tanimoto combo score was strongly dominated by the shape parameter. In the default setting, the ROCS software program produced Tanimoto shape scores ranging from 0.103 to 0.644 for FDA drug conformers when the nutlin 3a conformers were queried. On the other hand, for the same comparisons the color (charge distribution) scores for FDA drug conformers ranged from 0 to 0.350. To create a better balance between Tanimoto shape and Tanimoto color, the default ROCS matching setting was changed to "rank by ColorTanimoto" to place greater emphasis on the color parameter. With this change, the Tanimoto shape scores ranged from 0.071 to 0.538 and the Tanimoto color scores ranged from 0.006 to 0.373. Unexpectedly, ROCS generated repeats of a few drug conformers (see Additional Supporting Information online for details on repeats).</p><!><p>Drugs were prepared for docking by obtaining structure data format (sdf) files from the ZINC database [21]. The software program PyRx (version 0.7) was used to convert the sdf files to PDBQT files. PDBQT files were used for docking to MDM2 with AutoDoc Vina (version 1.1). The macromolecule docking target was amino acids 25–109 of MDM2 (Protein Data Bank #1RV1). Water molecules, heteroatoms (such as metals) and nutlin 2 were removed from 1RV1 prior to docking. The coordinates of the search space for MDM2 were maximized to allow the entire macromolecule to be considered for docking. The search space coordinates were: Center X: 13.5812 Y: 0.8458 Z: 19.5482, Dimensions (Å) X: 44.1592 Y: 34.3006 Z: 28.3009. Two-dimensional images highlighting the compounds and interacting MDM2 amino acids were created with LIGPLOT (version 4.5.3) [25]. RMSD values of compounds and solvent excluded areas within MDM2 were calculated by Chimera software (version 1.4.1) [26].</p><!><p>Our approach to computationally screen drugs that target MDM2 is based on the idea that such compounds would exhibit shape and charge distribution similar to nutlin 3a across its different conformers. To test if this approach was valid, a feasible structural model for nutlin 3a bound to MDM2 was required. Since no nutlin 3a structures are available, a model was developed using the structure of nutlin 2 bound to MDM2 [14]. Nutlin 2 (Figure 1A) shares, spatially, 90% of its heavy atoms with nutlin 3a (Figure 1B), suggesting that both compounds likely bind to MDM2 in a similar fashion. The nutlin 3a model was built from the crystal structure of nutlin 2 bound to MDM2 (PDB 1RV1-mean resolution = 2.18 ± 1.21 Å) (Figure 1C). The nutlin 3a model/MDM2 complex was energy minimized using WebLabViewer Pro software program (version 4.0). The energy minimized nutlin 3a model is called the nutlin 3a conformer model (nutlin 3a-CM) (Figure 3D).</p><p>A comparison of the structure of nutlin 2 and nutlin 3a-CM revealed that the root mean square deviation (RMSD) of bound molecules was 0.73 Å for shared heavy atoms indicating that energy minimization did not dramatically alter the nutlin 3a-CM conformation. Given the low RMSD value, one would expect nutlin 3a-CM to bind to MDM2 with a binding energy comparable to nutlin 2. Autodock Vina was used to separately dock the two nutlins to MDM2 (PDB 1RV1). The average binding energies reported (out of 9 dockings) was −6.8 ± 0.7 kcal/mol for nutlin 2 and −7.7 ± 0.6 kcal/mol for nutlin 3a-CM. The top-scoring docked nutlin 2 (from PDB 1RV1) and nutlin 3a-CM conformers generated after docking indicate that both nutlins bind with three functional groups inserted into the N-terminal p53-binding hydrophobic pocket of MDM2 (Figure 2). The three functional groups of nutlins simulate three p53 amino acid side chains that are critical for engagement with MDM2: Phe19, Trp23, Leu26 [27]. The three functional groups are two singly halogenated-substituted phenyl rings and an alkyloxy group at position 3 of the methoxy-phenyl ring. The slightly better binding energy of nutlin 3a-CM is likely due to increased van der Waals interactions resulting from its bulkier alkoxy group. The docked nutlins closely resemble the crystal structure showing nutlin 2 bound to MDM2 [14]. The docking software allows ligand flexibility for optimal fitting in the protein pocket. Nutlin-3a-CM docked without much perturbation from the original nutlin 3a-CM conformation (RMSD = 0.23 Å) (Figure 3A).</p><!><p>Our approach to drug repurposing is to identify small molecule conformers from the drug database that resemble low energy conformers of nutlin 3a. Low energy conformers of nutlin 3a were generated by the software program OMEGA [22] which uses a rule-based algorithm in combination with variants of the Merck force field 94 to produce conformations [28, 29]. Using default parameters, OMEGA generated 41 conformations of nutlin 3a from the nutlin 3a sdf file obtained from PubChem (PubChem ID: 11433190).</p><p>The ROCS software program was utilized to compare nutlin 3a-CM with 41 OMEGA-generated nutlin conformers. The Tanimoto shape scores ranged from 0.325 to 0.748 and the Tanimoto color (charge distribution) scores ranged from 0.278 to 0.472 resulting in a top Tanimoto combo score of 1.189. Structural alignment of the top scoring nutlin 3a generated by OMEGA (nutlin 3a-OG) and nutlin 3a-CM gave an RMSD of 2.52 Å (Figure 3B). The high RMSD value occurs because OMEGA rotated the phenyl methoxy ring of nutlin 3a 159° compared to nutlin 3a-CM to reduce steric hindrance between the isopropyloxy group and the piperizine ring. OMEGA also bent the piperizine ring so that carbon atoms 3, 4 and 5 on the piperizine ring are far from the phenyl methoxy ring. The high RMSD value obtained reflects these structure differences. ROCS analysis of the other 40 OMEGA-generated conformers confirmed that these OMEGA-generated conformers were more dissimilar to nutlin 3a-CM than nutlin 3a-OG (data not shown).</p><p>Autodock Vina was used to dock nutlin 3a-OG into MDM2. The average binding energy of nutlin 3a-OG docked to MDM2 was −7.1 kcal/mol ± 0.4 (9 dockings) compared to nutlin 3a-CM with a −7.7 kcal/mol binding energy. The less favorable binding energy is due to the fact that the binding conformation of nutlin 3a-OG differs from the binding conformation of nutlin 3a-CM (Figure 4). The bound nutlin 3a-OG isopropyloxy group is directed away from the hydrophobic pocket instead of into the hydrophobic pocket. This result calls into question how nutlin 3a is able to produce a conformation predicted to bind into the three sub-pockets of the hydrophobic cleft of MDM2. Because an experimental structure of nutlin 3a bound to MDM2 is unavailable, we must consider the possibility that nutlin 3a binds to MDM2 in a manner similar to nutlin 3a-OG. Another possibility is that Nutlin 3a-OG may in fact bind to MDM2 analogously to nutlin 2, but Auto Dock Vina is not capable of simulating the docking process accurately. Regardless of how nutlin 3a binds to MDM2's hydrophobic cleft, in vitro competition experiments shows that nutlin 3a prevents p53 peptide from binding to MDM2 with an IC50 of 90 nM [14] indicating that nutlin 3a is an effective MDM2 inhibitor.</p><!><p>Our approach was to compare conformers of compounds from a database of FDA-approved drugs to the 41 conformations of nutlin 3a generated from the nutlin 3a sdf file. Drug conformations that ranked relatively high by shape and color across 41 conformations of nutlin 3a would be considered good candidates for inhibitions of MDM2. The ZINC drug database contains molecules that are currently, or at one time were FDA-approved. The drug list shows some redundancy since it contains isomers as well as variably protonated versions of the same molecule. There are 1,125 drugs with unique molecular formulas in the ZINC database and 3,244 drugs when the isomers and alternate protonated forms are included.</p><p>The top 15 drugs were ranked by average Tanimoto combo scores, which ranged from 0.722 to 0.802 (Table 1). For all 3,244 drugs, the range was 0.142 to 0.802 (see Data S1, Supporting Information). The contribution of shape and color to the top 15 Tanimoto combo scores are also presented in Table 1. The calculated binding energies from docking of drugs to MDM2 (average of 100 dockings) are shown as well. The binding energies of the top 15 drugs ranged from −8.456 to −4.724 kcal/mol. The top ranking drug, S-bepridil, had an average binding energy of −6.6 kcal/mol, whereas the average binding energy of nutlin 3a-OG was −7.1 kcal/mol. The drug with the best binding energy is nadrolone phenyl propionate at −7.61 kcal/mol. Calculation of buried surface area of the hydrophobic revealed that nutlin 3a-CM buried 329 Å2 of MDM2. Of the top 15 drugs, the best in terms of shielding MDM2 from solvent is Nafronyl-d4, which buried 296 Å2 of solvent. In general, one expects the FDA-drugs to bury less MDM2 surface than nutlins because the former are generally smaller than the latter. To better understand the interactions of the top 15 drugs ranked by the ROCS program, we performed an analysis of the drug atom to MDM2 atom interactions that take place after docking.</p><p>Data for the number of top 15 drug atoms predicted to bind to MDM2, the number of hydrogen bonds predicted to bind to MDM2, the number of MDM2 amino acids predicted to bind to drugs and the percentage of nutlin 3a-CM-binding amino acids that bind to drugs were collected (Table 2) [25]. The number of drug atoms involved in MDM2 contacts ranged from 8 to 16. The top hit, S-bepridil has 14 drug atoms that bind to MDM2. This compares favorably to nutlin 3a-CM, which has 15 drug atoms that bind to MDM2. Four of the top 15 drugs make hydrogen bond contacts to MDM2 (nutlin 3a-CM has no hydrogen bond contacts). Nutlin 3a-CM is predicted to contact nine MDM2 amino acids—all without hydrogen bonds. The number of MDM2 amino acids that bind to the top 15 drugs ranged from seven to ten. Because these drugs were selected to mimic nutlin 3a CM one might expect that the nutlin 3a-CM interacting amino acids would overlap with the drug interacting amino acids. S-bepridil interacted with five amino acids that also interacted with nutlin 3a CM. It is worth noting that the fifth-ranked drug was R-bepridil—which is predicted to have 16 atoms in contact with 10 amino acids of MDM2. Seven MDM2 amino acids overlap those that contact nutlin 3a-CM. Of the top 15 drugs, 11 are predicted to interact with the majority of the same MDM2 residues as nutlin 3a-CM. The MDM2 residues that interact with nutlin 3a-CM and the top five drugs are Leu54, Gly58, Ile61, Val93, His96, and Ile99 (Figure 5). A recent NMR study of the Nutlin 3a-MDM2 complex confirmed that at least five of these six residues interact with Nutlin 3a [30]. X-ray crystallography structure analysis of the p53-MDM2 complex shows that all six MDM2 residues bind to p53 [27].</p><p>To further investigate the five highest scoring drugs, we show them computationally docked to MDM2 (Figure 6). We note that in each case, a minimum of two functional groups of the drug bind to two of the three hydrophobic subpockets of MDM2. The three MDM2 subpockets individually bind p53 amino acids Phe19, Trp23 and Leu26 [27]. S-bepridil, Caramiphen, and R-bepridil have functional groups that bind to all three subpockets of MDM2. The combination of the potential to bind to three subpockets and their relatively small sizes suggests they are good candidates for further exploration as MDM2 inhibitors.</p><!><p>The top hit, S-bepridil, marketed under the brand name Vascor, increases the contractile force of cardiac myofilament [31] by enhancing the responsiveness of the cardiac troponin C [32]. NMR and X-ray crystal structure analysis of bepridil-troponin C complexes show that S-bepridil binds to a hydrophobic depression within the N-terminal domain of troponin C [32, 33]. S-bepridil interacts with 13 hydrophobic residues in troponin C. In our computational models, S-bepridil interacts with 14 MDM2 residues and R-bepridil interacts with 16 MDM2 residues. This may make bepridil attractive as a potential inhibitor of MDM2.</p><!><p>Drug repurposing has had some success. Minoxidil, originally developed to treat hypertension, is used to combat hair loss. Viagra, designed to treat hypertension, is now indicated for erectile dysfunction and pulmonary hypertension. Rituxin was initially indicated for non-Hodgkin's lymphoma and it is now approved for treatment of chronic lymphocytic leukemia and rheumatoid arthritis [34]. Raloxifene, originally approved for osteoporosis, has been repurposed for some breast cancers. It modulates binding of estrogen to its receptor and leads to a decrease in estrogen-mediated transcription [35]. Bisphosphonates were originally approved to reduce skeletal-related effects of chemotherapy such as pathologic fractures, spinal cord compression, hypercalcemia, and the need for bone surgery. There is now mounting evidence that they create a micro-environment in the bone that is refractory to cancer growth [36]. None of these drug repurposing examples were discovered with a bioinformatics approach. Rather, clinical side effects were noted and measured. With the development of more extensive drug structure and protein structure databases, a computational approach to repurposing drugs should be feasible. Indeed, a database of 3,665 FDA-approved and investigational drugs and their known protein targets were used to predict potential new drug-target associations (Keiser et al., 2009 Nature). Importantly, of the 184 new predicted associations, 30 were experimentally tested and 23 of these were confirmed. This progress encouraged us to use an approach that combines bioinformatics and structure analysis to repurpose drugs to target MDM2.</p><p>Our Conformation Computational Selection method introduced here is a new approach to drug repurposing. This approach is based on knowledge of the high resolution drug-target structure. OMEGA is used to create putative several drug conformations. Those conformations are then compared to potential drug conformations from a database (in our case, we used a FDA-approved drug database). ROCS is used to quantify the drug similarities in terms of shape and atom type across the different conformations. In our studies, the top 15 scoring drugs all docked to the p53 binding domain in MDM2.</p>

PubMed Author Manuscript

ML327 Induces Apoptosis and Sensitizes Ewing Sarcoma Cells to TNF-Related Apoptosis-Inducing Ligand

Ewing sarcomas are rare mesenchymal-derived bone and soft tissue tumors in children. Afflicted children with distant metastases have poor survival despite aggressive therapeutics. Epithelial-to-mesenchymal transition in epithelial carcinomas is associated with loss of E-cadherin and resistance to apoptosis. ML327 is a novel small molecule that we have previously shown to reverse epithelial-to-mesenchymal transition features in both epithelial and neural crest-derived cancers. Herein, we sought to evaluate the effects of ML327 on mesenchymal-derived Ewing sarcoma cells, hypothesizing that ML327 initiates growth arrest and sensitizes to TNF-related apoptosis-inducing ligand. ML327 induced protein expression changes, increased E-cadherin and decreased vimentin, consistent with partial induction of mesenchymal-to-epithelial transition in multiple Ewing Sarcoma cell lines (SK-N-MC, TC71, and ES-5838). Induction of epithelial features was associated with apoptosis, as demonstrated by PARP and Caspase 3 cleavage by immunoblotting. Cell cycle analysis validated these findings by marked induction of the subG0 cell population. In vitro combination treatment with TRAIL demonstrated additive induction of apoptotic markers. Taken together, these findings establish a rationale for further in vivo trials of ML327 in cells of mesenchymal origin both alone and in combination with TRAIL.

ml327_induces_apoptosis_and_sensitizes_ewing_sarcoma_cells_to_tnf-related_apoptosis-inducing_ligand

1. Introduction<!>2.1. Cell Culture<!>2.2. Antibodies and Reagents<!>2.3. Chemical Synthesis<!>2.4. Western Blotting<!>2.5. Cell Cycle Analysis<!>3.1. ML327 Induced Partial MET in ES Cells<!>3.2. ML327-induced MET is Associated with Apoptosis Induction<!>3.3. ML327 Pre-Treatment Sensitizes ES Cell Lines to TRAIL-Induced Apoptosis<!>4. Discussion<!>Supplemental Figure 1. ML327 Treatment is Associated with Decreased p53 Expression

<p>Ewing Sarcomas (ES) are rare, aggressive childhood cancers thought to arise from mesenchymal stem cells of bone and soft tissues [1,2,3]. This family of cancers arises due to a balanced chromosomal translocation of the EWS gene and a member of the ETS family of genes, most commonly friend leukemia virus integration 1 (FLI1) [1,4]. Five-year survival rates are 60–70% for children with localized disease, and these rates fall to 30% for children afflicted with metastatic spread [1,4]. Multimodal treatment is the standard of care, encompassing systemic chemotherapy in combination with radiation or surgery [4]. Together, these findings highlight the need to identify novel biologic strategies for the treatment of this disease.</p><p>Epithelial-to-mesenchymal transition (EMT) is a dynamic developmental process that is exploited in carcinoma progression to mediate invasion, metastasis, and therapeutic resistance [5,6]. Loss of E-cadherin expression is considered the hallmark event of EMT [7]. We have previously conducted a phenotypic screen of 83,200 compounds to identify small molecules capable of inducing the re-expression of E-cadherin expression in colon and lung carcinoma cell lines [8]. Chemical optimization of our initial "hit" yielded a chemical probe, ML327 (N-(3-(2-hydroxynicotinamido) propyl)-5-phenylisoxazole-3-carboxamide), that partially reverses EMT in advanced epithelial cancers [9]. ML327 does not alter the cellular viability of colon and lung cancer cells, but is capable of blocking carcinoma cell invasiveness in vivo [10]. Intriguingly, we have recently reported that induction of partial mesenchymal-to-epithelial transition (MET) in neural crest-derived neuroblastomas blocks growth both in vitro and in vivo by inducing cell cycle arrest and necrosis, highlighting the therapeutic potential of this small molecule in cancers of non-epithelial origin [11].</p><p>Lack of progress in the treatment of children with ES has led to investigations into the efficacy of TNF-related apoptosis-inducing ligand (TRAIL) [12,13]. TRAIL is a pro-apoptotic cytokine of the TNF superfamily with appealing therapeutic potential given its ability to selectively induce apoptosis in cancer cells with minimal toxicity [14]. The majority of ES cell lines are sensitive to TRAIL in vitro [12]. TRAIL-based strategies have also been shown to block tumor growth and osteolysis and increase survival in in vivo ES models [15,16]. Resistance to TRAIL has been linked to acquisition of migratory mesenchymal characteristics and upregulation of anti-apoptotic proteins, including cellular FLICE-like inhibitory protein (cFLIP) [14].</p><p>The therapeutic potential of ML327-induced MET against cells of mesenchymal origin has not been explored. In the present study, we hypothesized that induction of MET using ML327 would block the growth of ES cells and sensitize to TRAIL-mediated apoptosis. Herein, we report that ML327 induces apoptosis in ES cells and has additive pro-apoptotic effects when used in combination with TRAIL in vitro. These findings provide a rationale to investigate the in vivo effects of small molecule-mediated MET agents, such as ML327, in the treatment of sarcomas, both alone and in combination with TRAIL-based therapeutic strategies.</p><!><p>SK-N-MC cell line was purchased from the American Type Culture Collection (ATCC, Manassas, VA). TC71 and ES-5838 were kindly provided as a gift from Jialiang Wang, PhD (Vanderbilt University; Nashville, TN). SK-N-MC and TC71 cells both exhibit a EWS-FLI1 translocation, while ES-5838 cells feature an EWS-ERG translocation. Cells were maintained in RPMI 1640 with 10% FBS at 37 °C in a humidified atmosphere consisting of 5% CO2 and 95% air.</p><!><p>E-cadherin antibody was from Cell Signaling Technology (Danvers, MA). Vimentin, Caspase 3 and PARP primary antibodies were obtained from Abcam (Cambridge, MA). cFLIP primary antibody was purchased from Enzo Life Sciences (Farmingdale, NY). TRAIL was purchased from Bio Vision (#4354-50, San Francisco, CA) and was also graciously provided by Dr. Avi Ashkenazi (Genentech, San Francisco, CA). All other reagents were obtained from Sigma (St. Louis, MO).</p><!><p>ML327 was synthesized as previously described through the Vanderbilt Institute of Chemical Biology [10]. ML327 was solubilized in DMSO.</p><!><p>Whole cell lysates were collected using cell lysis buffer (20 mM Tris, 150 mM NaCl, 1 mM EDTA, 1 mM EGTA, 0.1% SDS, 1% sodium deoxycholate, 1% Triton X-100, aprotinin, leupeptin, and 1 mM sodium orthovanadate) supplemented with proteinase inhibitors (Roche; Mannheim, Germany) and PMSF (1 mM). Protein (30–70μg) was run on a SDS-PAGE gel, transferred to a PVDF membrane, and probed with antibodies. Blots were developed using an enhanced chemiluminescence substrate (Perkin Elmer; Waltham, MA).</p><!><p>Cell cycle distribution was analyzed using flow cytometry. 1×106 cells were trypsinized, washed with PBS, and fixed in 70% ethanol. Fixed cells were incubated with RNAse (100 μg/mL), stained with propidium iodide (50 μg/mL), and analyzed on a 3-laser BD LSRII (BD Biosciences, San Jose, CA). Flow Cytometry experiments were performed in the VUMC Flow Cytometry Shared Resource.</p><!><p>We have previously reported partial reversal of TGF-β-induced EMT with ML327 (10 μM) in an immortalized mouse mammary epithelial cell line [10]. This EMT reversal was associated with upregulation of E-cadherin, a hallmark of epithelial cell fate, as well as a downregulation of the mesenchymal marker, Vimentin. We therefore tested whether ML327 would induce MET features in ES cell lines. We evaluated whether ML327 (10 μM) was capable of eliciting protein expression changes consistent with MET induction in three ES cell lines (TC71, ES-5838, and SK-N-MC) featuring EWS-fusion proteins. Similar to our observations in epithelial and neural crest-derived cell lines, ML372 successfully induced the expression of E-cadherin in all 3 ES cell lines tested (Fig. 1A) within 24h of treatment. Conversely, diminished Vimentin expression was noted by Western blotting in TC71 and ES-5838 ES cells 48h after ML327 treatment (Fig. 1A). Taken together, these results suggest that ML327 treatment induces protein expression changes consistent with partial MET features in ES cells.</p><p>We also performed light microscopy daily to observe whether morphologic changes elicited by ML327 treatment were consistent with an epithelialized phenotype. Surprisingly, we observed an increase in the proportion of rounded cells with cytoplasmic shrinking and non-adherent vesicles thought to represent apoptotic bodies (Fig. 1B) in all 3 experimental ES cells starting 48–72h following treatment with ML327 (10 μM). The remaining cell population lacking features of apoptosis induction did not exhibit morphologic changes consistent with epithelialization. Taken together, these findings demonstrate that ML327 is capable of inducing protein expression changes consistent with MET induction, but the predominant morphologic features elicited are most suggestive of apoptosis induction.</p><!><p>Our light microscopy observations led us to hypothesize that ML327 induces apoptosis in ES cells. We treated our 3 experimental ES cell lines with ML327 (10 μM) for up to 4 days and found increased Caspase 3 cleavage and PARP cleavage in all 3 cell lines. ES-5838 and SK-N-MC cells demonstrated increased Caspase 3 and PARP cleavage by 2 days of treatment that persisted out to 4 days (Fig. 2A and 2C). TC71 cells demonstrated increased Caspase 3 and PARP cleavage by 1 day of treatment that persisted out to 3 days (Fig. 2B). These findings demonstrate that ML327 consistently induces the proteolytic activation of two critical apoptosis mediators in ES cells.</p><p>To validate our findings, we treated ES-5838, TC71, and SK-N-MC cells with ML327 (10 μM) for 72 h and performed cell cycle analysis using flow cytometry with propidium iodide staining. In all 3 cell lines, ML327 treatment resulted in an increased percentage of cells in the Sub-G0 cell cycle phase (Fig. 3). Specifically, ML327 induced a 26-fold and 12-fold increase in Sub-G0 population in SK-N-MC and TC71 cells, respectively (Fig. 3). A more modest 4-fold induction of sub-G0 was observed in ES-5838 cells treated with ML327 (Fig. 3C). Overall, these findings demonstrate that ML327 successfully induces apoptosis in ES cells.</p><p>P53 is a critical tumor suppressor capable of activating apoptosis in the presence of cellular stressors, including but not limited to DNA damage and oxidative stress. Activation of P53 can lead to rapid protein stabilization. We performed a time course experiment in SK-N-MC cells to evaluate whether p53 stabilization was associated with the induction of a ML327-mediated apoptosis. Similar to our prior findings in epithelial and neural-crest derived cells, protein expression of p53 diminished during ML327-induced apoptosis (Suppl Fig. 1) [11]. These findings, together with our prior observations, suggest that ML327 likely induces apoptosis in a p53-independent manner.</p><!><p>EMT in epithelial cancers has been shown to be associated with resistance to apoptosis inducing ligands [14,17]. E-cadherin expression has been shown to be necessary for apoptosis induction by TRAIL [17], a death-receptor ligand that initiates apoptosis via Caspase 8 cleavage (Fig. 4A). We therefore examined whether ML327 would sensitize ES cell lines to TRAIL-induced apoptosis. Three ES cell lines were pre-treated with ML327 (10 μM) for 24 h followed by TRAIL (50ng/mL) for 6 h. ML327 pre-treated cells demonstrated increased Caspase 3 and PARP cleavage after TRAIL treatment compared to vehicle pre-treatment as well as ML327 alone (Figs. 4B–D).</p><p>ML327 has previously been shown to reduce cFLIPs expression in colon cancer cell lines, thereby sensitizing them to TRAIL-induced apoptosis [18]. We therefore analyzed the response of cFLIPs to ML327 treatment in 3 ES cell lines and found that cFLIPs protein was reduced (Figs. 4B–D). Taken together, these results suggest that MET in ES cell lines sensitizes to TRAIL-induced apoptosis in association with increased E-cadherin and reduced cFLIPs expression levels.</p><!><p>Therapeutic resistance is a consistent feature of tumors arising from cells of mesenchymal origin and carcinomas that have undergone EMT in their disease progression [14]. We originally identified ML327 as a small molecule that is capable of inducing the expression of E-cadherin, a hallmark of epithelial cell fate, in advanced carcinoma cells of the lung and colon [10]. We have systemically expanded our observations to encompass tumors of neural crest origin, neuroblastoma, and now cells of mesenchymal origin (ES) [11]. E-cadherin is consistently upregulated by ML327 in all tested cell lines (colon, lung, breast, neuroblastoma, and ES cells) with the exception of RKO (colon) and MDA-MB-231 (breast) cancer cells, whose E-cadherin promoters are silenced by DNA hypermethylation [10,18]. Intriguingly, the predominant cellular response observed within colon and lung carcinomas is blockade of cellular migration with no observed changes in cellular viability appreciated in vitro and in vivo [10]. The predominant feature of our trials in neuroblastoma is growth arrest and necrosis with minimal induction of apoptotic markers [11]. Herein, we report a marked induction of apoptotic markers, Caspase 3 and PARP cleavage, in association with MET induction in ES cells. Overall, these findings suggest an enhanced sensitivity of non-epithelial derived tumors, such as ES cells, to ML327.</p><p>Incremental progress has been made in improving the outcomes of children afflicted with ES, prompting considerable investigation into the potential use of death-inducing ligands, such as TRAIL [4,12,19]. Of these agents, TRAIL has received much attention given its lack of toxicity to non-transformed cells, highlighting a potential therapeutic window for exploitation [14]. Rapid clearance and resistance have plagued early clinical trials using TRAIL, highlighting the need to identify novel small molecules that sensitize to TRAIL-based therapeutics [14,20]. Reversal of mesenchymal or migratory features has been identified as one potential strategy for TRAIL sensitization [14]. Specifically, the HDAC inhibitor, MS-275, has been shown to both reverse EMT, attenuate metastasis, and sensitize breast cancer cells to TRAIL [21]. Our findings support these observations, as ML327 induces epithelial-like features and sensitizes ES cells to TRAIL-mediated apoptosis. We have previously reported the capacity of ML327 to mediate TRAIL sensitization in colon cancer cells, demonstrating this process to be in part mediated by down regulation of cFLIPs [18]. Our results are in support of our previous investigation, as we observe consistent downregulation of cFLIPs in association with ML327-mediated TRAIL sensitization. Further investigation into the therapeutic potential of the EMT reversal/MET induction agents, such as ML327, in combination with TRAIL-based strategies merits further investigation featuring studies to determine in vivo efficacy.</p><p>A persistent and critical deficiency of our current and previous characterizations of ML327 remains our inability to identify a direct intracellular effector of this small molecule. Modification of the structure of ML327 in ways that would facilitate affinity purification of bound proteins has resulted in loss of biological activity. We are currently undertaking alternative approaches to identify the mechanism. Identification of the direct intracellular target of ML327 offers both therapeutic promise and may provide mechanistic insights into the regulation of both EMT and therapeutic resistance to TRAIL.</p><p>In conclusion, we have validated the capacity of ML327 to elicit features of MET in ES cells. In contrast to prior characterizations in carcinoma and neural crest-derived tumors, ML327 elicits striking induction of apoptosis in all tested ES cell lines. Furthermore, partial MET induction in ES cells using ML327 sensitized ES cells to TRAIL-mediated apoptosis. Together, these findings support further in vivo characterization of ML327 in mesenchymal cancers, such as ES, both alone and in combination with TRAIL-based therapeutic strategies.</p><!><p>Immunoblotting demonstrates decreased levels of p53 in SK-N-MC cells treated with ML327 (10 μM) over a 4d time course.</p>

PubMed Author Manuscript

An integrated-molecular-beacon based multiple exponential strand displacement amplification strategy for ultrasensitive detection of DNA methyltransferase activity

DNA methylation is a significant epigenetic mechanism involving processes of transferring a methyl group onto cytosine or adenine. Such DNA modification catalyzed by methyltransferase (MTase) plays important roles in the modulation of gene expression and other cellular activities. Herein, we develop a simple and sensitive biosensing platform for the detection of DNA MTase activity by using only two oligonucleotides. The fluorophore labeled molecular beacon (MB) can be methylated by MTase and subsequently cleaved by endonuclease DpnI at the stem, giving a shortened MB. The shortened MB can then hybridize with a primer DNA, initiating a cycle of strand displacement amplification (SDA) reactions. The obtained SDA products can unfold new MB and initiate another cycle of SDA reaction. Therefore, continuous enlargement of SDA and exponential amplification of the fluorescence signal are achieved.Because the triple functions of substrate, template and probe are elegantly integrated in one oligonucleotide, only two oligonucleotides are necessary for multiple amplification cycles, which not only reduces the complexity of the system, but also overcomes the laborious and cumbersome operation that is always a challenge in conventional methods. This platform exhibits an extremely low limit of detection of 3.3 Â 10 À6 U mL À1 , which is the lowest to our knowledge. The proposed MTasesensing platform was also demonstrated to perform well in a real-time monitoring mode, which can achieve a further simplified and high-throughput detection. The sensing strategy might be extended to the activity detection of other enzymes, thus showing great application potential in bioanalysis and clinical diagnosis.

an_integrated-molecular-beacon_based_multiple_exponential_strand_displacement_amplification_strategy

Introduction<!>Chemicals and apparatus<!>MTase assisted strand displacement amplication<!>Fluorescence measurement<!>Gel electrophoresis assay<!>Detection of Dam MTase activity in a real sample<!>Design of the MTase biosensor<!>Feasibility of the designed biosensor<!>Optimization of the experimental conditions<!>Sensitivity and selectivity of the biosensor<!>MTase inhibitor screening<!>MTase activity assay in real sample<!>MTase activity assay in real-time SDA mode<!>Generality of the proposed sensing strategy<!>Conclusion

<p>Genomic DNA methylation is a crucial epigenetic DNA modication which commonly occurs in nature, and plays important roles in the regulation of gene expression. 1,2 The DNA methylation process is catalyzed by a series of methyltransferases (MTase) that transfer a methyl group from S-adenyl methionine (SAM) to the C-5/N-4 positions of cytosine (C) or the N-6 position of adenine (A). 3 To date, studies on cancer pathology have revealed that the abnormal pattern of DNA methylation induced by aberrant MTase activities is involved in various types of cancer, including breast, prostate, lung, liver and colon cancers. [4][5][6][7][8][9] Therefore, the aberrant MTase activity can be recognized as a new generation of biomarkers for the early clinical diagnosis of cancer progression. Notably, inhibiting MTase to block DNA methylation may provide useful information for anticancer therapeutic applications. [10][11][12] Hence, the development of a rapid and sensitive platform for the detection of MTase activity is of great signicance in both pharmacological and biochemical research.</p><p>Conventional analytical methods for the detection of MTase activity include methylation-target polymerase chain reaction (PCR), radioactive labeling-based gel electrophoresis, and highperformance liquid chromatography (HPLC). [13][14][15][16] However, most of these methods are time-consuming and harmful to biological systems because of their radioactive substances. 17 Recently, various alternative approaches have been developed for an MTase assay, such as uorescence, chemiluminescence, colorimetric and electrochemical methods. [18][19][20][21][22][23][24][25][26][27][28] The emerging methods have attracted intense attention due to their cheap instrumentation and low toxicity. DNA-based biosensors, in particular, can provide high sensitivity and selectivity because a State Key Laboratory of Medicinal Chemical Biology, Tianjin Key Laboratory of Biosensing and Molecular Recognition, Research Centre for Analytical Sciences, College of Chemistry, Nankai University, Tianjin 300071, P. R. China. E-mail: kongdem@nankai.edu.cn; Fax: +86-22-23502458 b Collaborative Innovation Center of Chemical Science and Engineering (Tianjin), Tianjin, 300071, P. R. China † Electronic supplementary information (ESI) available: Fig. S1 to S7 and Table S1.</p><p>See DOI: 10.1039/c8sc05102j</p><p>Cite this: Chem. Sci., 2019, 10, 2290</p><p>All publication charges for this article have been paid for by the Royal Society of Chemistry of their programmable target-specic sequence and their computational controlled signal amplication strategy. 29,30 A series of DNA-based biosensors have been applied in the analysis of MTase activity. For example, Yuan's group reported a uorescence method based on exonuclease III (Exo III)assisted isothermal cycling signal amplication; 31 Zhang's group reported a surface-enhanced Raman scattering method based on strand displacement amplication (SDA) strategy; 32 Zhu's group reported a chemiluminescence strategy combining hybridized chain reaction (HCR) and rolling circle amplication (RCA); 33 and recently, Zhang's group reported an RNase HIIassisted single-ribonucleotide repair-mediated ligationdependent cycling signal amplication method. 34 Nevertheless, the complicated signal amplication process usually requires a sophisticated design of the DNA template and probe sequences. 34 And numerous DNA oligos, which have to be involved in the system to assist the reaction, inevitably generate high background because of the nonspecic amplication. 33 Thus, a biosensor with a simple design and high performance for the detection of MTase activity is still in urgent demand.</p><p>In this paper, by integrating the triple functions of substrate, template and reporter into one molecular beacon probe, we propose a novel strategy to design a succinct DNA-based biosensor via MTase-triggered multiple cycles of strand displacement amplication (SDA) reactions for exponential signal amplication. Such a biosensor can achieve ultrasensitive MTase activity detection in a simple "one-step" way, and the limit of detection is 3.3 Â 10 À6 U mL À1 , which is the lowest we know of. Furthermore, the proposed sensing platform is demonstrated to perform well for the screening of MTase inhibitors, as well as the real-time monitoring of MTase activity.</p><!><p>All of the oligonucleotides used in this project (Table 1) were synthesized and puried by Sangon Biotech. Co. Ltd. (Shanghai, China). Dam and M. SssI methyltransferase (MTase), endonuclease DpnI, Klenow Fragment Polymerase (3 0 -5 0 exo-) (KFP), nicking enzyme Nb.BbvcI, S-adenyl methionine (SAM), EcoRI enzyme and the corresponding buffer solution were obtained from New England Biolabs (Beijing, China). Deoxyribonucleoside triphosphate (dNTP) was purchased from Tiangen Biotech. Co. Ltd. (Beijing, China). All other chemicals were applied in analytical grade and were ordered from Solarbio (Beijing, China). All solutions for the reaction were prepared with ultrapure water which was puried by a Milli-Q water purication system (>18.25 MU cm À1 ).</p><p>Fluorescence spectra were measured by a Hitachi RF-5301 uorescence spectrometer (Hitachi Ltd., Japan). The values of the melting temperature (T m ) of the DNA oligos were obtained from the Mfold website (University at Albany-State University of New York, USA). The gel electrophoresis results were obtained by a Gel Documentation System (Huifuxingye, Beijing, China). A real-time quantication PCR (RT-qPCR) assay was performed in a commercial StepOnePlus™ Real-Time PCR instrument (Applied Biosystems, USA).</p><!><p>The molecular beacon template was rst diluted in 1Â Dam MTase reaction buffer (50 mM Tris-HCl, 10 mM EDTA, 5 mM 2mercaptoethanol, p H7.5) to a nal concentration of 2 mM; then 160 mM SAM, and different concentrations of Dam MTase were added to the reaction system to a total volume of 25 mL. The mixture was reacted at 25 C for 2 h. Next, 10Â CutSmart buffer (500 mM potassium acetate, 200 mM Tris-acetate, 100 mM magnesium acetate, 1 mg mL À1 BSA, pH 7.9) and 10 U of DpnI were added for the cleavage reaction and the total volume was adjusted to 50 mL. The cleavage reaction was performed at 37 C for 1 h, followed by heat deactivation at 80 C for 20 min. Finally, 10 mL of the truncated molecular beacon (1 mM), 10 mL of primer DNA (1 mM), 10 mL of 10Â CutSmart buffer, 10 mL of 10 mM dNTP, 0.5 U of KFP and 1 U of Nb.BbvcI were mixed together to a total volume of 100 mL and incubated at 37 C for 1 h followed by 80 C heat deactivation for 20 min as the SDA process.</p><!><p>Aer the SDA process, the uorescence signal of the product was directly measured by the uorescence spectrometer. The excitation wavelength was set at 490 nm, and the emission spectrum from 500 nm to 650 nm was collected for further analysis.</p><!><p>The SDA product was analyzed through polyacrylamide gel electrophoresis (PAGE). Aer the reaction, 15 mL of the product was mixed with 3 mL of prepared-loading buffer. The mixture was loaded into a 10% polyacrylamide gel contained in 1Â TBE buffer (9 mM Tris base, 9 mM boric acid, 0.2 mM EDTA, pH 7.5).</p><p>The PAGE was performed under 120 V constant voltage at room temperature for 50 min. The gel was stained with ethidium bromide. The stained gel was visualized using the Gel Documentation Imaging System.</p><p>Inhibition of Dam MTase activity 10 mL of the molecular beacon substrate (10 mM) was rst mixed with different concentrations of gentamycin or 5-uorouracil, and pre-incubated in 1Â Dam MTase reaction buffer at 37 C for 30 min. Then 10 U mL À1 Dam MTase and 160 mM SAM were added into the reaction mixture and incubated at 37 C for 2 h. Next, 10Â CutSmart buffer and 10 U of DpnI were added into the mixture and another incubation was performed at 37 C for 1 h, followed by heat deactivation at 80 C for 20 min. Finally, the SDA reaction was performed at 37 C for 1 h followed by 80 C heat deactivation for 20 min. The uorescence signal was measured as described above, and the relative activity (RA) of the Dam MTase was calculated based on eqn (1):</p><p>where F G , F R , and F 0 represent the uorescence intensity in the presence of different concentrations of gentamycin, in the absence of gentamycin and in the absence of MTase, respectively.</p><!><p>A total volume of 100 mL of a sample containing 10% human serum spiked with various concentrations of Dam MTase was prepared for the Dam MTase activity assay. The procedure for the uorescent measurement was the same as that described above.</p><!><p>Herein, only two oligonucleotides are used. One is a hairpin-like molecular beacon (MB) oligonucleotide (molecular beacon, Table 1) which plays the triple roles of substrate for MTase recognition, template for MTase-triggered strand displacement amplication (SDA) and probe for amplied signal output. The other is a single-stranded linear oligonucleotide used as an SDA primer (primer, Table 1). The two ends of the MB are labeled with the uorophores FAM and TAMRA, respectively, and the FAM uorescence is efficiently quenched by TAMRA due to close contact. The double-stranded stem of MB contains a specic 5 0 -GATC-3'/3 0 -CTAG-5 0 region where the adenines (A) will be methylated in the presence of Dam MTase and subsequently be cleaved by endonuclease DpnI at the methylated positions. The cleavage reaction leads to two consequences: the separation of FAM from TAMRA to give recovered uorescence and the hybridization of the primer with the shortened MB. The hybridization would not affect the intact molecular beacon because the longer hairpin structure is much more stable (T m ¼ 69.1 C) compared to the shortened hairpin (T m ¼ 42.7 C). Such a hybridization will initiate the rst cycle of SDA reaction under the catalysis of Klenow Fragment Polymerase (KFP) and nicking enzyme Nb.BbvcI to give amplied SDA products. These products will in turn recognize and open the original MB substrates that were not previously methylated by Dam MTase. As a result, the uorescence of FAM, which was quenched by the TAMRA quencher, is recovered due to the separation between the two ends of the MB. In addition, primers will hybridize on these unfolded MBs to initiate the second cycle of SDA, resulting in the release of product strands, which could be repeatedly used to unfold new MBs. More importantly, this cycle of SDA could also provide new products. Collectively, since the product strands used to trigger new SDA reactions can be provided by three routes-products of the rst cycle of SDA, products of the second cycle of SDA and the repeatedly used products-SDA reactions will be continuously enlarged and exponential signal amplication can be achieved (Scheme 1). Integrating three functions into one MB substrate will not only simplify the design of the sensing system, but also avoid the high background signal induced by the nonspecic amplication because of the reduction in the number of DNA oligos included in the complicated signal amplication process reported before. In the low MTase concentration range, most of the MB substrates might not be cleaved, but these parts can also participate in signal amplication via the next cycle of SDA. Therefore, the issues of low utilization of probes and/or templates, which are oen suffered by most DNA-based biosensors, are completely overcome.</p><!><p>We employed polyacrylamide gel electrophoresis (PAGE) to monitor the reaction process of the proposed sensing platform.</p><p>As shown in Fig. 1A, aer the treatment with Dam MTase and DpnI to the MB, a band with a faster migration rate appeared, according to the cleavage of the template (Fig. 1A, line 6). In contrast, the MB treated with DpnI showed no signicant difference from the original MB (lines 5 & 7), indicating that the MTase was necessary in the MB-cleavage process. Moreover, the expected SDA products were only observed when the shortened MB was employed. As shown in Fig. 1A, lines 1-4, a new band with a slower migration rate was observed, indicating that the amplication proceeded and an unfolded signal reporter had been generated. However, the band of the product did not appear if we applied the original MB for the SDA (line 3), showing that the MTase-associated DNA cleavage was necessary for the initiation of the SDA process. In addition, compared to the one without nicking enzyme Nb.BbvcI, the band of the product was much brighter aer the enzyme was added in (lines 1 & 4), showing that the SDA process would be completed only in the presence of both Klenow and Nb.BbvcI. We also veried the mechanism of the proposed sensing platform by a uorescent assay. As shown in Fig. 1B, a signicant increase in the uorescence intensity could be observed only in the presence of all of the materials. No obvious signal could be observed in the absence of any of the enzymes or DNA oligos, indicating that the uorescent signal was obtained from the SDA process particularly initiated by the target enzyme. In addition, the value of the uorescence intensity (I t ) at 520 nm is about 6 times higher in the presence of nicking enzyme Nb.BbvcI (Fig. S1 †), and the value of DI t (subtracting the background from the signal intensity) is about 25 times higher, compared with the one without Nb.BbvcI. This result indicated that the nicking enzyme-mediated amplication cycles could efficiently increase the sensitivity of the sensing platform.</p><!><p>In order to obtain the best sensing performance, biosensor design and some important experimental conditions were optimized. First of all, the best primer was selected (Fig. 2). An ideal primer should efficiently hybridize and unfold the shortened MB but have no effect on intact MB. We noticed that when a short primer (e.g. Table 1, primer 0 with 11 nucleotides) was used, a very low uorescence signal was given even in the presence of 0.1 U mL À1 Dam MTase, indicating that an Scheme 1 Illustration of the probe-reporter integrated biosensor platform mechanism. exponential SDA reaction was not initiated because this primer was too short to hybridize with the shortened MB. When the primer length increased to 13 nucleotides, however, high background was given even in the absence of Dam MTase. The reason was that the primer was too long; it can hybridize and unfold intact MB, resulting in the occurrence of undesirable amplication reactions. By comparing several primers with different lengths, primer 1 of 12 nt length provided an acceptable background and the best signal-to-noise ratio (I t /I 0 , where I t and I 0 represent uorescence intensity with and without Dam MTase). Thus primer 1 was applied in subsequent investigations.</p><p>The reaction time of the SDA reaction was also critical because increasing reaction time will certainly give an enhanced signal intensity but might also increase the possibility of undesirable side reactions, thus inducing high background. By synchronously monitoring the time-dependent uorescence changes of the sensing platform with and without Dam MTase, 1 h was selected as the optimal reaction time due to its acceptable background and the highest I t /I 0 value. Similarly, using I t /I 0 as a criterion, the concentration of primer 1 was optimized at 100 nM (Fig. S2 †), and the amounts of KFP and Nb.BbvcI were selected at 0.5 and 1 units in a total volume of 100 mL, according to previous research. 35</p><!><p>To investigate the sensitivity of the biosensor, we measured the uorescence intensity at various concentrations of Dam MTase under the optimal conditions (Fig. 3A). The uorescence intensity was enhanced as a function of MTase concentration from 10 À5 to 10 U mL À1 (Fig. 3B). In logarithmic scales, the uorescence intensity exhibited a linear correlation with the concentration of Dam MTase over a wide range from 10 À5 to 1 U mL À1 (Fig. 3B). The regression equation is I t ¼ 178.38 + 26.79 Â log C MTase with a correlation coefficient of 0.973, where I t and C MTase represent the uorescence intensity and the Dam MTase concentration, respectively. The limit of detection (LOD) was estimated to be 3.3 Â 10 À6 U mL À1 based on 3 times the standard deviation over the blank response (3s/S). Notably, the LOD of this biosensor is the lowest to our knowledge. The sensitivity of this biosensor platform has improved by as much as 2 orders of magnitude compared with that of rolling circle amplication (RCA) based chemiluminescence assay, 21 by 3 orders of magnitude compared with that of the methylation response DNAzyme colorimetric assay, 22 by 3 orders of magnitude compared with that of quantum dots-mediated FRET assay, 36 and as well as being an improvement over other methods. [37][38][39] The extremely low LOD could be attributed to (i) the high amplication efficiency because of making full use of the integrated probe-reporter MB substrates in the SDA amplication process; (ii) multiple highly efficient SDA cycles were included to enhance the signal; and (iii) low background due to few oligos being applied in the whole system.</p><p>Selectivity is an important characteristic of biosensors. To evaluate the selectivity of this biosensor, we introduced M.SssI MTase as a potential interference enzyme. M.SssI MTase is also a methyltransferase; it can methylate the cytosine within a sequence of 5 0 -C-G-3'/5 0 -C-G-3 0 in double-stranded DNA. 20 As shown in Fig. 4, the uorescence intensity increased signicantly in the presence of Dam MTase. In contrast, no distinct signal change was observed in the presence of M.SssI MTase compared to the blank control even when the concentration of M.SssI MTase was 10 times higher than that of Dam MTase, thus revealing the high selectivity of the proposed biosensor toward Dam MTase.</p><!><p>DNA MTase was reported as a signicant biomarker as well as an important diagnostic target in recent research. 36 Pharmacologically inhibiting DNA MTase activity may alter the DNA methylation levels in cells, which is concerned in a variety of cancers. 39 Thus, the screening of inhibitors for MTase has attracted intense interest. To evaluate the potential of our biosensor for DNA MTase inhibition assay, we applied gentamycin, which is widely used as an inhibitor of methyl transferase, 40 as a model inhibitor. Collecting the uorescence intensity at 520 nm with different concentrations of gentamycin added into the reaction system, we dened the relative activity of the Dam MTase based on eqn (1). As shown in Fig. 5A, the relative activity of Dam MTase reduced gradually with increasing concentration of gentamycin from 0 to 60 mM. The half maximal inhibition (IC 50 ) is dened as the concentration of the inhibitor applied to achieve a 50% relative activity. According to the calibration curve in Fig. 5A, the IC 50 value of gentamycin is calculated to be 8.02 mM. Here, the value is consistent with the value (10.0 mM) obtained by an Exo IIImediated uorescence based assay. 31 Similarly, the inhibition function of 5-uorouracil was also investigated using the proposed platform. According to Fig. 5B, the IC 50 value of 5-uorouracil is calculated to be 0.71 mM. The value obtained by our sensing platform approximates those reported in recent research. 18,37 This result demonstrates that this biosensor can be used for the inhibitory capacity evaluation of DNA MTase inhibitors and thus for inhibitor screening, holding great potential in pharmacological applications.</p><!><p>We also investigated the capability of the proposed biosensor in the application of a Dam MTase assay in a human serum sample. Different concentrations of Dam MTase were spiked into a 10% serum sample and the recoveries were calculated by comparing the measured Dam MTase activity and the amount of Dam MTase added into the system (see Fig. S3 † for the spectrum result). As shown in Table 2, the recoveries were found to be 94.0% to 111.0% according to the measurement, with relative standard deviations (RSD) ranging from 3.4% to 9.0%. The result revealed that the new biosensor was reliable for the detection of MTase in a real sample.</p><!><p>The above MTase activity assay was conducted in an end-point detection mode, in which the uorescence intensity was detected aer the SDA reactions. To further simplify the experimental operations, the feasibility of MTase activity quantication in a real-time mode was investigated. In such a mode, the uorescence change of the sensing system was instantaneously recorded during the SDA reaction via a commercial, real-time quantitative PCR instrument (StepOnePlus™ Real-Time PCR system, ABI, United States). As shown in Fig. 6A, plots of uorescence (F) vs. reaction time (T) were obtained with different MTase concentrations. Similar to the conventional real-time PCR, a unique signal-processing method can be applied. That is, log(F)-T plots were constructed on the basis of the obtained F-T plots (Fig. 6B), and the RT t values, the reaction time at which the log(F) reaches a set threshold, were calculated. On a logarithmic scale, a linear relationship (R 2 ¼ 0.985) was obtained between the RT t value and the concentration of MTase, in the range of 1 Â 10 À5 to 1 U mL À1 . The linear regression equation was calculated as RT t ¼ 29.26 À 5.33 Â log[C MTase ]. From Fig. 6, it was found that the log(F)-T plots of the systems containing 1 Â 10 À5 U mL À1 Dam MTase could be easily discriminated from those of the blank control, thus conrming the excellent detection sensitivity of the real-time detection mode. Compared to the end-point mode, the real-time mode has simplied operations due to the elimination of a post-amplication signal detection procedure, thus endowing the method with increased high-throughput detection capability. In addition, since the signal amplication and detection are synchronously performed, the reaction tubes can be directly discarded aer SDA reactions without opening the lids, thus greatly reducing the risks of amplication product carryover contamination, a big challenge that exponential nucleic acid amplication reactions have to face.</p><!><p>By a slight modication of the sequence of the stem region of the MB, the proposed MTase-sensing strategy might be easily extended to the design of other sensors targeting different enzymes. For example, the design of a restriction endonuclease sensor can be achieved by simply replacing the MTase recognition sequence on MB with a restriction endonuclease recognition one. As a proof-of-concept, a sensing platform for EcoRI, a restriction endonuclease that can specically recognize and cleave 5 0 -GAATTC-3 0 /3 0 -CTTAAG-5 0 , was designed. Similar to the proposed Dam MTase-targeted sensing platform, the hairpin structure was less stable aer cleavage by EcoRI (T m decreased from 73.3 C to 42.0 C), and thus could be unfolded by a short primer, initiating the SDA cycles. We conrmed the feasibility of the designed platform by gel electrophoresis. (see Fig. S4 † for the details). The sensor was then demonstrated to work well for the highly sensitive detection of enzyme activity with good selectivity as well (Fig. S5 and S6 †). This result indicated that the proposed sensor design strategy might be used for designing a series of biosensors for the activity analysis of a broad spectrum of biologically important enzymes.</p><!><p>In summary, a novel exponential signal amplication strategy was developed and successfully used for biosensor design. By highly integrating the three roles of substrate, template and reporter into one oligonucleotide, the biosensor platform can be constructed by using only two oligonucleotides, thus greatly reducing the complexity of the sensing system and simplifying the sensing operations. The potential background generated by the numerous DNA oligonucleotides involved in the multiple amplication steps can also be avoided. The unique exponential signal amplication mode ensures the full and effective utilization of the oligonucleotides in sensing systems, thus endowing the biosensor with an extraordinarily high sensitivity. The constructed biosensor was demonstrated to perform well for the ultrasensitive detection of Dam MTase activity, as well as the screening of the potential inhibitor of Dam MTase, and the quanticational analysis of Dam MTase in a real sample. The LOD of Dam MTase was as low as 3.3 Â 10 À6 U mL À1 , which is the lowest so far, compared with reported methods (Table S1 †). In addition, by simply modifying the enzyme recognition sequence in the oligonucleotide, the sensor design strategy was demonstrated to be easily extendable to the detection of other enzymes (e.g. restriction endonuclease EcoRI), thus showing great potential in pharmacological assay and in clinical diagnosis.</p>

Royal Society of Chemistry (RSC)

Double Porphyrin Cage Compounds

The synthesis and characterization of double porphyrin cage compounds are described. They consist of two porphyrins that are each attached to a diphenylglycoluril‐based clip molecule via four ethyleneoxy spacers, and are linked together by a single alkyl chain using “click”‐chemistry. Following a newly developed multistep synthesis procedure we report three of these double porphyrin cages, linked by spacers of different lengths, i.e. 3, 5, and 11 carbon atoms. The structures of the double porphyrin cages were fully characterized by NMR, which revealed that they consist of mixtures of two diastereoisomers. Their zinc derivatives are capable of forming sandwich‐like complexes with the ditopic ligand 1,4‐diazabicyclo[2,2,2]octane (dabco).

double_porphyrin_cage_compounds

Introduction<!><!>Introduction<!>Results and Discussion<!><!>Results and Discussion<!><!>Results and Discussion<!><!>Results and Discussion<!><!>Results and Discussion<!><!>Results and Discussion<!><!>Results and Discussion<!><!>Results and Discussion<!><!>Conclusions<!>Experimental Section<!><!>Experimental Section<!>

<p>In the past decades, nature has served as a source of inspiration for scientists working on the design of new molecular systems that are capable of mimicking the action of enzymes, e.g. with respect to rate and substrate selectivity. A class of enzymes that stands at the basis of life are the DNA‐polymerases, which, together with exonucleases, replicate and break down the DNA present in organisms.[ 1 ] During DNA replication, the information stored in the pattern of nucleotides on the encoding (mother) strand is transferred to the new daughter strand. One of the factors contributing to the high replication fidelity of DNA‐polymerases is the fact that they make use of sequential processive catalysis, meaning that the enzyme stays bound to the substrate while it performs multiple rounds of consecutive reactions.[ 2 ] During these reactions the movement of the polymerase along the biopolymer chain is discrete and unidirectional, as it repeatedly moves one nucleotide further to carry out the next reaction. In this way, information is reliably copied and stored again by making use of the four nucleobases present in the DNA chain.</p><p>Taking the natural processive enzymes as a blueprint, our group has developed synthetic processive catalysts that are based on a glycoluril‐based manganese porphyrin cage, MnSC (Figure 1A).[ 3 ] This compound is capable of threading an alkene‐containing polymer chain and catalytically converting it to its polyepoxide in an efficient, processive fashion (Figure 1B).[ 4 ] Our current research focuses on achieving more control over the processive epoxidation catalysis, i.e., in terms of the directionality of performing the catalytic reactions along the polymer chain, and the ability to stereoselectively epoxidize trans‐double bonds in the polymers into (R,R)‐ or (S,S)‐epoxides. When such control can be realized, we envisage the use of the epoxidized polymers as a new type of data storage material,[ 5 ] with the two mirror image epoxides acting as the digits 0 and 1 in a binary code.</p><!><p>(A) Molecular structure of catalytic porphyrin cage compound MnSC and the axial ligand bupy. (B) Schematic representation of the processive catalytic conversion of cis‐polybutadiene (yellow) into its polyepoxide, carried out by MnSC (blue), to which a bupy ligand (green) is axially coordinated. (C) Schematic representation of a double porphyrin cage compound (yellow) in which the "writing" cage serves as a processive catalyst for the epoxidation of a polymer chain (red) and the "instructing" cage binds a Me2V cofactor (blue), which transfers information via the ditopic ligand dabco (orange) to the "writing cage". (D) Molecular structures of the double porphyrin cage compounds described in this paper.</p><!><p>In order to be able to manipulate the reactivity and/or selectivity of the manganese porphyrin cage catalyst, we have designed double porphyrin cage molecules, in which one of the cages is meant to serve as the "writing cage" and the other as the "instructing cage" (Figure 1C‐D). Here, the "writing cage" threads a polymer chain and catalytically converts its double bonds into epoxides. It receives its instructions via an external stimulus, e.g., a light‐induced binding event, coming from the "instructing cage". This stimulus is transferred via a ditopic ligand, e.g. dabco (1,4‐diazabicyclo[2,2,2]octane) that binds between the two cages via their porphyrin metal centers. An example of such an external stimulus could be the binding of a cofactor guest,[ 6 ] such as Me2V (N,N'‐dimethyl‐4,4‐bipyridinium dihexafluorophosphate). Me2V is known to have a high affinity for the receptor cavity of the porphyrin cages.[ 7 ] In this paper we report our synthetic efforts to covalently link two porphyrin cage compounds in a geometry that ensures close proximity of the metal centers in the two porphyrin planes (Figure 1D). We previously showed that the coordination of axial ligands such as dabco to the metal center in the porphyrin cage is allosterically influenced by the binding of a Me2V guest inside the cavity.[ 8 ] We here describe the synthesis of 3 double porphyrin cages, which differ in the length of the linker that connects the two separate porphyrin rings. Furthermore, we report the ditopic coordination of dabco to the zinc derivatives of the double cage compounds, yielding stable 1:1 sandwich complexes.</p><!><p>Synthesis. Conventionally, porphyrin cage compounds are prepared via a fourfold nucleophilic substitution reaction of tetratosyl‐functionalized molecular clip 7 (Scheme 1) with tetrakis‐meso‐ortho‐hydroxyphenyl porphyrin.[ 9 ] For the synthesis of the double porphyrin cages, the latter compound needed to be equipped on one of the meso‐aryl substituents with a functional group that would allow further conversion to a linker between the two cages. To synthesize such a mono‐functionalized porphyrin, we designed the synthetic route that is depicted in Scheme 1. It starts with an acid‐catalyzed condensation of paraformaldehyde with pyrrole to provide dipyrromethane 1 in 57 % yield.[ 10 ] A MacDonald [2+2] condensation of this compound with 2‐methoxybenzaldehyde and subsequent oxidation of the porphyrinogen gave 5,15‐bis(2‐methoxyphenyl)porphyrin 2 in 35 % yield.[ 11 ] The attachment of the third meso‐aryl substituent was accomplished by nucleophilic addition of o‐methoxyphenyllithium (obtained by a reaction of 2‐bromoanisole with n‐butyllithium) to one of the available meso‐positions of 2.[ 11 , 12 ] After oxidation with DDQ, compound 3 was obtained in 40 % yield. The remaining free meso‐position of this compound is unsuitable for a similar reaction with an appropriate lithium compound to prepare the desired mono‐functionalized porphyrin, because a reaction of the lithium salt with the free meso‐position yields a Meisenheimer‐type complex. The negative charge of this complex is localized and stabilized at the opposite meso‐position of the porphyrin, which therefore should be unsubstituted.[ 13 ] As an alternative, we selected a Suzuki cross‐coupling reaction to introduce the final meso‐aryl ring. To this end, the free meso‐position of 3 was first regioselectively brominated with NBS, giving compound 4 in quantitative yield.[ 14 ] A subsequent cross‐coupling of 4 with 5‐bromo‐2‐methoxyphenyl boronic acid, in the presence of a palladium catalyst, gave porphyrin 5 in 83 % yield.[ 15 ] To prevent additional cross‐coupling reactions at the bromo‐functionalized phenyl ring, product formation of 5 was monitored with the help of mass spectrometry (MS), and the reaction was carried out at a moderate temperature. At the moment that MS showed full conversion of 4, the reaction was stopped. As a final step, the methoxy groups of 5 were deprotected using boron tribromide to give the desired mono‐bromo‐functionalized tetrahydroxy porphyrin 6 in 89 % yield. Compound 6 was then coupled to clip 7 [ 9 ] to give the mono‐bromo‐substituted porphyrin cage compound 8 in 16 % yield. To protect the porphyrin of 8 from undesired metal insertion in following synthesis steps, it was metallated by inserting a zinc(II) center, providing compound 9 in quantitative yield. To convert 9 into a suitable precursor for a copper‐catalyzed azide‐alkyne 1,3‐dipolar cycloaddition (CuAAC) reaction, the bromide substituent was converted into an alkyne via a Suzuki–Miyaura cross coupling reaction. Using potassium triisopropylsilylacetylene trifluoroborate as the alkyne source,[ 16 ] compound 10 was obtained in 93 % yield. Subsequent deprotection of the alkyne with tetrakis‐n‐butylammonium fluoride provided mono‐acetylene‐functionalized porphyrin cage 11 in 73 % yield. Using CuAAC, this compound was subsequently reacted with α,ω‐diazidoalkane linkers of different length to obtain a series of three double cage compounds with various spacers lengths between the two cages. To facilitate the characterization of the new compounds by NMR, 15N‐enriched α,ω‐diazidoalkanes were synthesized according to a literature procedure using 15N‐enriched sodium azide, in which the 15N isotope is located at one of the two terminal azide positions.[ 17 ] CuAAC reactions of 11 with the respective diazides provided zinc double porphyrin cages Zn2C3DC, Zn2C5DC and Zn2C11DC in yields of 35, 31 and 62 %, respectively. The rather low yields of the double click reactions are attributed to steric hindrance between the two approaching porphyrin cages. During purification by column chromatography, it was possible to retrieve unreacted 11, which could be re‐employed in subsequent click reactions. After treatment of the Zn2CxDC (x = 3, 5 or 11) compounds with aqueous hydrochloric acid, free base double porphyrin cage compounds H4C3DC, H4C5DC and H4C11DC were obtained in yields of 68, 81, and 92 %, respectively.</p><!><p>Synthesis of double porphyrin cage compounds.</p><!><p>Structural characterization. The double porphyrin cage compounds were characterized with the help of 1H, 13C and 15N NMR spectroscopy in chloroform solution. All proton and carbon resonances could be readily assigned with the help of 2D techniques (see Figure 2 for the assignment of H4C3DC). The NMR measurements revealed the presence of several structural isomers of the double porphyrin cage compounds. The 15N spectrum of H4C3DC showed the presence of four singlets for the 15N‐labeled nitrogen atoms present in the triazole rings of the spacer between the two porphyrin cages (Figure 2B). The fact that two sets of two singlets are observed for the two different types of 15N‐labeled nitrogen atoms indicates that two different species of H4C3DC must be present. A 1H‐15N HBMC spectrum revealed couplings of both the nitrogen signals 15N‐61 at 346.54 and 346.45 ppm with triazole proton signal H‐58 at 7.77 ppm. Also the signals of 15N‐59 at 244.89 and 244.77 ppm coupled with this triazole signal, as well as to proton signals at 4.32 and 2.47 ppm belonging to CH2‐62 and CH2‐63, respectively. Via a ROE contact between H‐58 and the phenyl proton H‐22 in the 2D ROESY spectrum, the resonance of the latter proton in the 1H NMR spectrum could be assigned (8.40 ppm). C‐22 displays a 13C‐HMBC coupling to H‐24 (8.28–8.21 ppm), which in turn displays two COSY cross coupling peaks with H‐25. The fact that for this proton two well‐separated resonances are observed confirms the abovementioned existence of two different species. The two signals integrate equally, meaning that the two species are present in a 1:1 ratio. In addition to H‐25, also H‐45, H‐46, and H‐48 in the same quadrant of the molecule (denoted "I" in Figure 2) give two distinct signals in the NMR spectrum, all confirming the existence of two different species. Conversely, the analogous proton signals in the other three quadrants of the molecule (II, III and IV) do not show such a splitting of signals.</p><!><p>(A) 1H NMR spectrum of H4C3DC in CDCl3 and proton/carbon numbering of the compound. The Roman numbers I, II, III and IV and the related colors indicate the four symmetry quandrants within the porphyrin cage structure. (B) 15N spectrum of H4C3DC in CDCl3.</p><!><p>The observation that H4C3DC exists as two equally abundant species is the result of the fact that this compound, as well as the other double porphyrin cage compounds, are mixtures of two diastereoisomers. The monofunctionalized single porphyrin cage compounds 8–11 are all racemates of two enantiomers. Although their chiral nature is at first sight not easily recognizable, their macrocyclic three‐dimensional (3D) structure gives rise to the existence of two non‐superimposable mirror‐image structures, i.e., enantiomers (Figure 3A). The compounds exhibit planar chirality,[ 18 ] and the 3D chiral morphology leads in this specific case to the emergence of two chiral centers, i.e., the two quaternary bridgehead carbon atoms in the diphenylglycoluril framework (C‐52, see Figure 2A).[ 19 ] The subsequent connection of two molecules of (rac)‐11 via click chemistry leads to the expected formation of two diastereoisomers in equal quantities (Figure 3B). So far, we have not succeeded in separating the diastereoisomers via chromatographic methods. Besides H4C3DC, also H4C5DC displayed two sets of signals for protons H‐25, H‐45, H‐46 and H‐48 in its 1H NMR spectrum. In contrast, H4C11DC did not display double resonance sets for these protons. This is presumably the result of the spacer length in this double cage compound. The longer spacer increases the distance between the two cages so much that they no longer experience chemical environments that are different enough to cause double resonances for the same protons in the NMR spectrum.</p><!><p>(A) Representations of the two enantiomers of monofunctionalized single cage compounds 8–11. Top: 3D representations. Bottom: schematic view of the porphyrin (black square) seen from the top. The black lines left and right from the square indicate the location of the o‐xylylene side walls of the cage, the cyan dot the functional group at the top. (B) Representations (top views and 3D views) of the two diastereoisomers that are obtained by the coupling of the enantiomers of monofunctionalized single cage compound 11.</p><!><p>Insertion of a zinc center in the porphyrin led to broadening of the proton signals of the double zinc porphyrin cage compounds Zn2CxDC in their 1H NMR spectra in CDCl3. The addition of deuterated pyridine to these solutions caused a sharpening of the signals, which suggests that the signal broadening is caused by coordination interactions (intramolecular or intermolecular) between the triazole nitrogen or carbonyl oxygen atoms and the porphyrin zinc centers, which are broken by coordination of the deuterated pyridine to the zinc centers. To characterize the Zn2CxDC compounds, we recorded their NMR spectra at 60 °C in [D6]DMSO, a solvent that also prohibits such intramolecular or intermolecular coordination. The spectra indeed displayed sharp resonances, which could be completely assigned to the two diastereoisomers that were also observed for the H4CxDC compounds.</p><p>Coordination complexes of the double zinc porphyrin cage compounds with dabco. The ability of the double zinc porphyrin cages to coordinate the ditopic ligand dabco was investigated. The coordination of this ligand to zinc porphyrins and the typical spectral changes it induces in NMR and optical spectra have been reported extensively.[ 20 ] To enable future follow‐up studies with chloroform‐insoluble viologen guests, the ligand coordination studies were carried out in a solvent mixture of chloroform and acetonitrile (1:1, v/v). While in chloroform the Zn2CxDC compounds all displayed a Soret band at 421 nm, in CHCl3/CH3CN this band shifts to 426 nm, which we attribute to axial coordination of acetonitrile to the zinc centers. When dabco is added to a solution of Zn2C3DC, the band at 426 nm initially does not shift, but sharpens and increases in intensity (Figure 4A), which indicates the formation of a more well‐defined species. After the addition of several equivalents of dabco, the spectral changes remain virtually unchanged up to the addition of several hundreds of equivalents of the ligand (Figure 4B). In the presence of a larger excess of dabco, the band at 426 nm again decreases in intensity, with a concomitant emergence of a new band at 431 nm, which increases in intensity as up to 5 × 105 equivalents of dabco are added.</p><!><p>(A) Changes in UV/Vis spectra of Zn2C3DC upon the addition of dabco, in CHCl3/CH3CN, 1:1 (v/v). (B) Titration curves extracted from the spectral changes in (A), monitoring the absorbance of the bands at 426 (red) and 431 nm (blue). The solid lines through the data points represent fits assuming a standard 1:2 binding model.</p><!><p>The initial sharpening of the Soret band is attributed to the formation of a 1:1 sandwich complex between Zn2C3DC and dabco.[ 21 ] Such a complex is better defined and more rigid than the open‐folded complex between Zn2C3DC and axially coordinated acetonitrile molecules. Moreover, due to effective molarity and binding statistics effects, this complex is highly stable and only upon the addition of a large excess of the ligand, the emergence of a red‐shifted Soret band at 431 nm reflects the formation of a 1:2 Zn2C3DC:dabco complex, which is again open‐folded.[ 20 ] UV/Vis titrations of Zn2C5DC and Zn2C11DC with dabco gave similar results (see Supporting Information).</p><p>The obtained titration curves were fitted with a standard 1:2 binding model (Figure 4B). The association constants K a and binding free energies ΔG θ were extracted from the fits and are summarized in Table 1.[ 22 ] The K a‐values obtained for all the double zinc porphyrin cage compounds are comparable to those obtained previously for the coordination of dabco to other bis‐zinc porphyrin compounds, which range between 104 ‐ 108 m – 1 for K a (1:1) and 102 – 103 m – 1 for K a (1:2) in various solvents.[ 20 ] Due to possible competition for axial ligation to the zinc centers between dabco and the acetonitrile solvent, the association constants determined here are on the lower end of the commonly reported range.[ 23 ] While the K a(1:1)‐values of the complexes of dabco with Zn2C5DC and Zn2C11DC are quite similar, the K a(1:1)‐value of the complex with Zn2C3DC is significantly higher. We attribute this difference to a more favorable effective molarity effect for the second zinc porphyrin in the latter complex: once the first zinc porphyrin of Zn2C3DC has coordinated a dabco ligand, the second zinc porphyrin is in close proximity as a result of the relatively short spacer between the two porphyrins. This effect is favorable for 1:1 complexation and the generation of a high K a(1:1)‐value, but since the difference in K a(1:1)‐value between the complexes of dabco with Zn2C5DC and Zn2C11DC is much less prominent, probably also other factors play a role. Molecular modelling calculations of the complexes (Spartan™, PM3/semi‐empirical) revealed that the short C3‐spacer allows an almost perfect fit for the dabco ligand between the zinc porphyrins of Zn2C3DC. The longer C5 and C11 spacers of the other two double zinc porphyrin cages require progressive folding of their chain to still allow a ditopic binding of the ligand (Figure 5), which may be a factor that leads to a lowering of the binding strength of the ligand.</p><!><p>Association constants K a [M – 1] and binding free energies ΔG θ (kJ mol–1) for the 1:1 and 1:2 complexes between Zn2CxDC and dabco in CHCl3/CH3CN, 1:1 (v/v). The error represents the standard deviation (±1 S.D.) over a triplo of measurements</p><p>Computer‐modeled structures of the sandwich complexes of double zinc porphyrin cages (indigo) (A) Zn2C3DC, (B) Zn2C5DC and (C) Zn2C11DC with dabco (red). The alkyl spacers between the cages are indicated in yellow.</p><!><p>To corroborate the formation of the 1:1 and 1:2 complexes identified by the UV/Vis titration experiments, NMR studies were carried out for the complex formation between dabco and Zn2C3DC, at various temperatures in CDCl3/CD3CN, 1:1 (v/v). Figure 6 shows the changes in the 1H NMR spectra at 298 K when up to 8 equivalents of dabco were added to a 1 mm solution of Zn2C3DC. The addition of one equivalent of the ligand caused an upfield shift of the porphyrin β‐pyrrole signals from 8.9–8.5 to 8.5–8.3 ppm, while a new, very broad signal appeared at –4.5 ppm. The latter signal is characteristic for a dabco ligand bound in a sandwich‐like geometry between two zinc porphyrins.[ 20 ] Its broadness is probably caused by rapid exchange between the bound and unbound components on the NMR timescale. In the presence of two or more equivalents of dabco, a broad signal corresponding to uncomplexed ligand (at 2.6 ppm) appeared, while the signal at –4.5 ppm disappeared. Remarkably, no signal corresponding to a dabco ligand in an open‐folded 1:2 complex was observed. In such a complex, the signal of the methylene‐protons of bound dabco are expected around –3 ppm. This absence may again be due to signal broadening as a result of rapid exchange.[ 20 , 24 ]</p><!><p>1H NMR titration of Zn2C3DC with 0 to 8 equiv. of dabco (500 MHz, 298 K, CDCl3/CD3CN, 1:1 (v/v)). The blue box and arrow indicate the location of the signal of dabco in the 1:1 sandwich complex, and the red arrow the signal of non‐coordinated dabco.</p><!><p>To decrease the exchange dynamics, NMR studies were carried out at lower temperature (Figure 7). At temperatures lower than 278 K the broad resonance at –4.5 ppm sharpened into a well‐defined peak at –4.46 ppm, which shifted slightly upfield to –4.6 ppm at 238 K, while simultaneously the signals of Zn2C3DC gradually broadened.</p><!><p>Variable temperature 1H NMR spectra of the 1:1 sandwich complex between Zn2C3DC and dabco (400 MHz, CDCl3/CD3CN, 1:1 (v/v)). The arrow indicates the location of the signal of dabco in the 1:1 sandwich complex.</p><!><p>To investigate the binding of dabco into further detail, up to 4 equivalents of the ligand were added to the solution of Zn2C3DC and NMR spectra were recorded at 248 K (Figure 8). In the presence of 0.25 equivalents of dabco, the signals of Zn2C3DC sharpened, indicating a decrease in dynamics of the cage molecule. When the amount of dabco was increased to 1 equivalent, the resonance of the sandwiched ligand at –4.58 ppm slightly broadened. The addition of more equivalents of dabco led to the emergence of a signal of the mono‐coordinated ligand in the open‐folded 1:2 complex at –2.92 ppm (α‐protons of the ligand), while simultaneously the resonance at –4.58 ppm decreased in intensity. With the use of 2D NMR techniques we were unable to locate the signals of the β‐protons of the mono‐coordinated dabco ligand or the signal of the free ligand, which is probably due to the broadness of the signals and the poor signal‐to‐noise ratio under these conditions. Due to the gradual broadening of the dabco resonances, no reliable baseline correction could be applied and, therefore, no reliable ratio of the 1:1 and 1:2 complexes could be determined by integration of the signals. The signals of Zn2C3DC broadened and became less defined upon the addition of more equivalents of dabco, which indicates an increase in the dynamics of the double cage compound. Although the NMR signals become better resolved at temperatures lower than 288 K, it can be expected that the complexes also exist at 298 K, albeit more dynamic and in abundancies that are governed by the association constants at that temperature. Thus, the NMR results corroborate the binding geometries as proposed by the UV/Vis experiments.</p><!><p>1H NMR titration of Zn2C3DC with 0 to 4 equiv. of dabco (500 MHz, 248 K, CDCl3/CD3CN, 1:1 (v/v)). The blue and purple arrows indicate the locations of the signals of dabco in the 1:1 sandwich complex (–4.58 ppm) and in the 2:1 open‐folded complex (–2.92 ppm), respectively.</p><!><p>We have presented a multistep route for the synthesis of covalently linked double porphyrin cage compounds. For the synthesis of these compounds, a mono‐bromo‐substituted porphyrin was synthesized in 6 steps starting from paraformaldehyde and pyrrole. Upon connection of the porphyrin to a cavity molecule based on diphenylglycoluril, a racemate of two planar‐chiral porphyrin cage molecules was obtained. After converting the bromo‐substituent of these cage compounds into an acetylene function, double porphyrin cage compounds could be prepared via "click" reactions with α,ω‐diazidoalkanes of different lengths. With the help of extensive NMR spectroscopy experiments the double cage compounds could be fully characterized. The presence of multiple resonances for several protons of the double cages with C3 and C5 spacers proved their existence as two diastereoisomers, while the absence of such multiple resonances in the case of the double cage compound with the C11 spacer indicated that the two cavities of that compound are so remote that they are no longer affected by each other's chirality. The ability of the zinc double porphyrin cage compounds to form sandwich complexes with the ditopic ligand dabco was confirmed by UV/Vis and NMR titration studies. These revealed that the zinc double cage with the short C3 spacer formed the strongest 1:1 sandwich complex with the ligand, probably as a result of a favourable effective molarity effect and binding geometry. As expected, the 1:2 open complexes displayed association constants of about 3 orders of magnitude lower than those of the respective 1:1 complexes. The formation of the 1:1 and 1:2 complexes were corroborated by variable temperature 1H NMR titrations. Future research will focus on the possibilities to transfer information between the receptor cavities of the sandwich complexes during catalytic reactions. We will equip the double porphyrin cage compounds with one zinc and one catalytic manganese center, and investigate the influence of the binding of guests in the zinc porphyrin cage, via the coordinated dabco ligand, on the processive epoxidation of threaded polyalkenes by the manganese cage of the system.</p><!><p>Materials and methods. All commercially obtained chemicals were used without further purification, unless stated otherwise. Dry n‐pentane was stored under argon in a glovebox, THF was distilled under nitrogen from potassium, CH2Cl2 was distilled under nitrogen from calcium hydride, and MeCN was distilled under argon from calcium chloride. For TLC analysis, TLC Silicagel 60 F254 (Merck) and for column chromatography, Silica gel 0.035–0.070 mm 60A (Acros), SilicaFlash® P60 40–63 µm (SiliCycle) or Silicagel 60 H (Merck) were used. 1H, 13C NMR and 15N NMR spectra were recorded on Bruker Avance III 400 or 500 MHz spectrometers at 25 °C unless stated otherwise. Chemical shifts are reported in parts per million (ppm) relative to tetramethylsilane (TMS, 0.00 ppm) as the internal reference. NMR data are presented as follows: chemical shift (∂) in ppm, multiplicity (s = singlet, bs = broad singlet, d = doublet, t = triplet, q=quartet, td = triplet of doublets, m = multiplet and/or multiple resonances), integration, assignment and coupling constant (J) in Hertz (Hz). All NMR signals were assigned on the basis of 1H, 13C NMR, 15N NMR, COSY, ROESY, HSQC, and HMBC experiments. The numbering of the proton, carbon, and nitrogen atoms used in the assignments is depicted in Figure 9. Phase and baseline correction was applied to all NMR spectra. As several porphyrin compounds were obtained as a mixture of atropisomers that were inseparable by column chromatography, the reported chemical shifts represent the averaged shifts over all of these atropisomers. Assigning all inequivalent 13C‐signals was not possible due to the limited resolution of the 2D spectra and overlapping proton signals, therefore ranges are reported for several almost identical 13C atoms. LCQ mass spectra were recorded in methanol on a Thermo Finnigan LCQ Advantage Max mass spectrometer, and MALDI‐TOF mass spectra were measured in reflective mode with dithranol as matrix on a Bruker Microflex LRF MALDI‐TOF mass spectrometer. Accurate MALDI‐TOF mass spectra were obtained on a Brukerdaltonics autoflex (ST‐A2130) MALDI‐TOF mass spectrometer with cesium triiodide as matrix in reflective mode. Accurate masses were obtained from a solution of the compound in methanol on a JEOL AccuTOF CS JMS‐T100CS. Compound 7 was synthesized according to a literature procedure.[ 9 ]</p><!><p>Carbon, proton, and nitrogen numbering of 5‐(5‐bromo‐2‐methoxyphenyl)‐10,15,20‐tris(2‐methoxyphenyl) porphyrin 5 (left) and the double cage compounds M2CnDC (M = 2H or Zn, n = 3, 5 or 11) (right) used for all NMR analyses. For the double cage compounds, R represents the mirror image of the cage molecule, with the restriction that for Zn2C3DC the attachment starts at carbon number 63, for Zn2C5DC at carbon atom 64, and for Zn2C11DC at carbon atom 67 (the latter is shown). The CH2‐groups of the diphenylglycoluril framework (carbon atoms 45, 46, 48 and 50) all have two inequivalent geminal protons, marked as a and b. The location of the signals of these protons could only be identified for CH2‐50, with the help of ROESY experiments.</p><!><p>Syntheses. For the NMR assignments of proton, carbon and nitrogen signals, the atom numbering depicted in Figure 9 will be used.</p><p>Di(1H‐pyrrol‐2‐yl)methane (1). This compound was synthesized according to modified literature procedures.[ 10 ] To distilled pyrrole (153 mL, 2.21 mol) a 37 % aq. formaldehyde solution (9 mL, 90.42 mmol) was added. Argon was led through the solution for 15 min, after which TFA was added (0.82 mL, 10.21 mmol). The reaction mixture was stirred at r.t. for 5 min. The orange solution was diluted with CH2Cl2 (150 mL) and washed with sat. aq. Na2CO3 (3 × 100 mL) and water (100 mL). The organic layer was concentrated in vacuo. The excess of pyrrole was removed by vacuum distillation at 35 °C. Silicagel column chromatography using DCM/n‐heptane (60:40 (v/v)) as the eluent afforded 1 (7.469 g, 51 mmol, 57 %) as a white fluffy solid.</p><p>1H‐NMR (CDCl3, 400 MHz): ∂ 7.91 (bs, 2H, NH), 6.68 (td, 2H, ArH‐5, J = 2.6, 1.5 Hz), 6.15 (q, 2H, ArH‐4, J = 2.9 Hz), 6.05–6.02 (m, 2H, ArH‐3), 3.99 (s, 2H, CH 2‐1); 13C{1H}‐NMR (CDCl3, 100 MHz): ∂ 129.03 (ArC‐2), 117.24 (ArC‐5), 108.38 (ArC‐4), 106.36 (ArC‐3), 26.38 (CH2‐1).</p><p>5,15‐Bis(2‐methoxyphenyl)porphyrin (2). Argon was led through CH2Cl2 (1.85 L) for 30 min and compound 1 (1.205 g, 8.24 mmol), 2‐methoxybenzaldehyde (1.122 g, 8.24 mmol) and TFA (0.140 mL, 1.82 mmol) were added. The reaction mixture was stirred in the dark at r.t. for 16 h. Then a solution of DDQ (2.685 g, 11.83 mmol) in CH2Cl2 (70 mL) was added. After stirring for 2 h the mixture was quenched with Et3N (6 mL) and concentrated in vacuo. The residue was purified over a silica gel column using chloroform as the eluent. The crude product was dissolved in CH2Cl2, filtered, and precipitated in n‐heptane. After washing the precipitate with n‐pentane (4×), drying in vacuo yielded 2 (749.1 mg, 1.43 mmol, 35 %) as a purple solid containing 2 atropisomers.</p><p>1H‐NMR (CDCl3, 500 MHz): ∂ 10.23 (s, 2H, ArH‐1,11), 9.33 (d, 4H, β‐pyrrole‐H‐3,9,13,19, J = 4.5 Hz), 8.97 (d, 4H, β‐pyrrole‐H‐4,8,14,18, J = 4.5 Hz), 8.06 (d, 2H, ArH‐28,40, J = 7.2 Hz), 7.80 (t, 2H, ArH‐30,42, J = 7.9 Hz), 7.41 (t, 2H, ArH‐29,41, J = 7.5 Hz), 7.38 (d, 2H, ArH‐31,43, J = 8.0 Hz), 3.61 (s, 6H, OMe‐46,48), –3.07 (s, 2H, NH); 13C{1H}‐NMR (CDCl3, 126 MHz): ∂ 159.45 (ArC‐32,44), 147.28 (ArC‐5,7,15,17), 145.18 (ArC‐2,10,12,20), 135.80 (ArC‐28,40), 131.33 (ArC‐3,9,13,19), 130.68 (ArC‐4,8,14,18), 130.29 (ArC‐27,39), 129.89 (ArC‐30,42), 119.61 (ArC‐29,41), 115.01 (ArC‐6,16), 111.07 (ArC‐31,43), 104.78 (ArC‐1,11), 55.81 (OMe‐46,48); UV/Vis (CHCl3): λ/nm (log (ε/M–1 cm–1)) 408 (5.53), 502 (4.21); Accurate mass: m/z 523.21361 [M + H]+; calcd. for C34H27N4O2 523.213040; Anal. Calcd for C34H26N4O2: C, 78.14; H, 5.01; N, 10.72; found C, 77.99; H, 4.92; N, 10.68.</p><p>5,10,15‐Tris(2‐methoxyphenyl)porphyrin (3). 2‐Bromoanisole (2.763 g, 14.77 mmol) was added to dry n‐pentane (50 mL). n‐Butyllithium (1.6 m in hexanes, 8 mL, 12.79 mmol) was added and the reaction mixture was stirred for 30 min at r.t. until a white precipitate was formed. The mixture was filtered under Schlenk conditions and the white residue was washed with dry n‐pentane (4 × 40 mL), after which it was dissolved in distilled THF (150 mL). Compound 2 (390 mg, 0.75 mmol) was added and the reaction mixture was stirred at r.t. for 16 h, after which water (3.5 mL) was added, followed by DDQ (761.32 mg, 3.35 mmol). After stirring for 1 h at r.t. the reaction mixture was evaporated to dryness and the residue was purified over a silica gel column using chloroform as the eluent. The crude product was subsequently purified by silica gel column chromatography using DCM/n‐heptane (75:25, (v/v)) as the eluent to yield 3 (187.7 mg, 0.29 mmol, 40 %) as a purple solid containing 3 atropisomers.</p><p>1H‐NMR (CDCl3, 500 MHz): ∂ 10.11 (s, 1H, ArH‐1), 9.25 (d, 2H, β‐pyrrole‐H‐3,19, J = 4.6 Hz), 8.90 (d, 2H, β‐pyrrole‐H‐4,18, J = 4.4 Hz), 8.79–8.75 (m, 4H, β‐pyrrole‐H‐8,9,13,14), 8.04 (d, 1H, ArH‐34, J = 7.3 Hz), 8.02–7.92 (m, 2H, ArH‐28,40), 7.80–7.70 (m, 3H, ArH‐30,36,42), 7.39–7.35 (m, 3H, ArH‐29,35,41), 7.35–7.26 (m, 3H, ArH‐31,37,43), 3.61–3.51 (m, 9H, OMe‐46,47,48), –2.89 (s, 2H, NH); 13C{1H}‐NMR (CDCl3, 126 MHz): ∂ 159.56–159.37 (ArC‐32,38,44), 149.00–144.00 (broad, ArC‐2,5,7,10,12,15,17,20), 135.78–135.50 (ArC‐28,34,40), 131.49–130.77 (ArC‐27,33,39), 131.20–130.10 (broad, ArC‐3,4,8,9,13,14,18,19), 129.76–129.68 (ArC‐30,36,42), 119.46–119.26 (ArC‐29,35,41), 115.87–115.19 (ArC‐6,11,16), 111.05–110.81 (ArC‐31,37,43), 104.40 (ArC‐1), 55.81 (OMe‐46,47,48); UV/Vis (CHCl3): λ/nm (log (ε/M–1 cm–1)) 413 (5.74), 508 (4.25); Accurate mass: m/z 629.25449 [M + H]+; calcd. for C41H33N4O3: 629.25526.</p><p>5‐Bromo‐10,15,20‐tris(2‐methoxyphenyl)porphyrin (4). Compound 3 (385.4 mg, 0.61 mmol) was dissolved in chloroform (25 mL) and NBS (128.15 mg, 0.72 mmol) was added. After stirring at r.t. for 25 min the reaction mixture was quenched by the addition a solution of Na2SO3 (177.33 mg, 1.41 mmol) in water (10 mL). After 15 min of stirring the reaction mixture was diluted with chloroform (50 mL) and washed with water. The organic layer was dried with Na2SO4, filtered and the solvents evaporated to dryness. The crude product was purified by silica gel column chromatography using DCM/n‐heptane (50:50 going to 75:25, (v/v)) as an eluent to yield 4 (467.7 mg, 0.66 mmol, 100 %) as a purple solid containing 3 atropisomers.</p><p>1H‐NMR (CDCl3, 500 MHz): ∂ 9.60 (d, 2H, β‐pyrrole‐H‐3,19, J = 4.8 Hz), 8.80 (bs, 2H, β‐pyrrole‐H‐4,18), 8.68 (bs, 4H, β‐pyrrole‐H‐8,9,13,14), 8.03–7.89 (m, 3H, ArH‐28,34,40), 7.79–7.70 (m, 3H, ArH‐30,36,42), 7.38–7.33 (m, 3H, ArH‐29,35,41), 7.33–7.26 (m, 3H, ArH‐31,37,43), 3.62–3.52 (m, 9H, OMe‐46,47,48), –2.63 (s, 2H, NH); 13C{1H}‐NMR (CDCl3, 126 MHz): ∂ 159.50–159.20 (ArC‐32,38,44), 135.62–135.38 (ArC‐28,34,40), 130.74 (ArC‐27,33,39), 129.91–129.84 (ArC‐30,36,42), 119.41 (ArC‐29,35,41), 116.53–116.29 (ArC‐6,11,16), 110.97–110.77 (ArC‐31,37,43), 102.43 (ArC‐1), 55.79 (OMe‐46,47,48); UV/Vis (CHCl3): λ/nm (log (ε/M–1 cm–1)) 421 (5.54), 517 (4.21); Accurate mass: m/z 707.16436 (M(79Br)+H)+, 709.16321 (M(81Br)+H)+; calcd. for C41H32BrN4O3 707.16578 (79Br); 709.16373 (81Br); Anal. Calcd for C41H31BrN4O3: C, 69.59; H, 4.42; N, 7.92; found C, 69.16; H, 4.34; N, 7.76.</p><p>5‐(5‐Bromo‐2‐methoxyphenyl)‐10,15,20‐tris(2‐methoxyphenyl)porphyrin (5). Compound 4 (690 mg, 0.98 mmol), (5‐bromo‐2‐methoxyphenyl)boronic acid (1.125 g, 4.88 mmol), triphenylarsine (119.9 mg, 0.39 mmol), bis(triphenyl‐phosphine)palladium (II) dichloride (137 mg, 0.20 mmol), and tripotassium phosphate (1.035 g, 4.88 mmol) were dissolved in distilled THF (175 mL). The reaction mixture was stirred at 42.5 °C for 4 h (reaction progress monitored with the help of MALDI‐TOF). After cooling to r.t., the mixture was evaporated to dryness and the residue purified by a silica gel column using DCM as the eluent. Subsequently, the crude product was purified by silica gel column chromatography using DCM/n‐heptane (60:40 going to 80:20 (v/v)) as the eluent to yield 5 (660.5 mg, 0.81 mmol, 83 %) as a purple solid containing 8 atropisomers.</p><p>1H‐NMR (CDCl3, 500 MHz): ∂ 8.78–8.68 (m, 8H, β‐pyrrole‐H‐3,4,8,9,13,14,18,19), 8.20–8.07 (m, 1H, ArH‐22), 8.06–7.91 (m, 3H, ArH‐28,34,40), 7.84 (d, 1H, ArH‐24, J = 9.0 Hz), 7.75 (t, 3H, ArH‐30,36,42, J = 7.9 Hz), 7.36–7.32 (m, 3H, ArH‐29,35,41), 7.32–7.28 (m, 3H, ArH‐31,37,43), 7.21–7.14 (m, 1H, ArH‐25), 3.63–3.50 (m, 12H, OMe‐45,46,47,48), –2.65 (s, 2H, NH); 13C{1H}‐NMR (CDCl3, 126 MHz): ∂ 159.46 (ArC‐32,38,44), 158.73 (ArC‐26), 137.96–137.64 (ArC‐22), 135.83–135.40 (ArC‐28,34,40), 133.37 (ArC‐21), 132.38 (ArC‐24), 131.11 (ArC‐27,33,39), 129.72 (ArC‐30,36,42), 119.33 (ArC‐29,35,41), 115.71 (ArC‐6,11,16), 113.44 (ArC‐1), 112.51 (ArC‐25), 111.77 (ArC‐23), 110.91 (ArC‐31,37,43), 56.09 (OMe‐45,[46,47,48]), 55.85 (OMe‐[45],46,47,48); UV/Vis (CHCl3) λ/nm (log (ε/M–1 cm–1)) 419 (5.62), 514 (4.28); Accurate mass: m/z 813.20482 (M(79Br)+H)+, 815.20403 (M(81Br)+H)+; calcd. for C48H38BrN4O4: 813.20764 (79Br); 815.20560 (81Br); Anal. Calcd for C48H37BrN4O4: C, 70.85; H, 4.58; N, 6.89; found C, 70.55; H, 4.45; N, 6.80.</p><p>5‐(5‐Bromo‐2‐hydroxyphenyl)‐10,15,20‐tris(2‐hydroxyphenyl)porphyrin (6). Compound 5 (660.5 mg, 0.81 mmol) was dissolved in distilled DCM (40 mL) and this solution was cooled to –30 °C. Then, BBr3 (1.55 mL, 16.2 mmol) was added. The reaction mixture was warmed up to r.t. overnight before it was poured into an ice‐water mixture (100 mL). Ethyl acetate (110 mL) and sat. aq. NaHCO3 (250 mL) were added upon which the color of the mixture turned from green to red. The organic layer was washed with sat. aq. NaHCO3 (2 × 70 mL) and dried with MgSO4, filtered, and the solvents evaporated to dryness. The crude product was purified by silica gel column chromatography using ethyl acetate/chloroform (50:50 (v/v)) as an eluent, yielding 6 (549.4 mg, 0.73 mmol, 89 %) as a purple solid containing 8 atropisomers.</p><p>1H‐NMR (CDCl3, 500 MHz): ∂ 8.96–8.88 (m, 8H, β‐pyrrole‐H‐3,4,8,9,13,14,18,19), 8.13–8.08 (m, 1H, ArH‐22), 7.98–7.94 (m, 3H, ArH‐28,34,40), 7.83 (t, 1H, ArH‐24, J = 8.8 Hz), 7.73 (t, 3H, ArH‐30,36,42, J = 8.1 Hz), 7.37–7.32 (m, 6H, ArH‐29,31,35,37,41,43), 7.23 (d, 1H, ArH‐25, J = 2.5 Hz), 4.92 (bs, 4H, OH), –2.78 (s, 2H, NH); 13C{1H}‐NMR (CDCl3, 126 MHz): ∂ 155.40 (ArC‐32,38,44), 154.72 (ArC‐26), 137.07 (ArC‐22), 135.01 (ArC‐28,34,40), 133.57 (ArC‐24), 133.00–131.00 (broad ArC‐3,4,8,9,13,14,18,19), 130.76 (ArC‐30,36,42), 129.35 (ArC‐21), 127.22 (ArC‐27,33,39), 119.77–119.73 (ArC‐29,35,41), 117.36 (ArC‐25), 115.71–115.55 (ArC‐31,37,43), 114.01–113.73 (ArC‐6,11,16), 111.95–111.91 (ArC‐23), 111.53–111.30 (ArC‐1); UV/Vis (CHCl3): λ/nm (log (ε/M–1 cm–1)) 419 (5.32), 513 (4.20); Accurate mass: m/z 757.14343 (M(79Br)+H)+, 759.14233 (M(81Br)+H)+; calcd. for C44H30BrN4O4: 757.14504 (79Br), 759.14300 (81Br); Anal. Calcd for C44H29BrN4O4: C, 69.75; H, 3.86; N, 7.40; found C, 69.31; H, 3.91; N, 7.13.</p><p>Mono‐bromo porphyrin cage compound (8). To compound 7 (549.09 mg, 0.41 mmol), potassium carbonate (1.406 g, 10.19 mmol), compound 6 (306.97 mg, 0.41 mmol), and acetonitrile (750 mL) were added. Argon was led through the reaction mixture for 30 min, and the solution was subsequently refluxed for 16 h. After cooling, the mixture was filtered through Celite and the filtrate was evaporated to dryness. The crude product was purified by column chromatography (alumina Brockmann III) using chloroform as the eluent. Precipitation from DCM/n‐heptane followed by washing with n‐pentane (4 ×) and drying in vacuo yielded 8 (92.6 mg, 0.07 mmol, 16 %, racemate of two enantiomers) as a purple solid.</p><p>1H‐NMR (CDCl3, 500 MHz): ∂ 8.80–8.74 (m, 4H, β‐pyrrole‐H‐3,4,13,14), 8.70–8.64 (m, 4H, β‐pyrrole‐H‐8,9,18,19), 8.21 (d, 1H, ArH‐22, J = 2.5 Hz), 8.10–8.03 (m, 3H, ArH‐28,34,40), 7.86 (dd, 1H, ArH‐24, J = 8.9, 2.5 Hz), 7.78–7.73 (m, 3H, ArH‐30,36,42), 7.41–7.36 (m, 3H, ArH‐29,35,41), 7.35–7.32 (m, 3H, ArH‐31,37,43), 7.21 (d, 1H, ArH‐25, J = 8.9 Hz), 6.98–6.91 (m, 6H, ArH‐55,56), 6.83–6.80 (m, 4H, ArH‐54), 6.19 (s, 3H, ArH‐48(II, III, IV)), 6.18 (s, 1H, ArH‐48(I)), 4.31–4.23 (m, 4H, CH 2‐45a), 4.23 (d, 4H, CH 2‐50a, J = 15.9 Hz), 4.10–4.00 (m, 4H, CH 2‐45b), 3.74 (d, 4H, CH 2‐50b, J = 15.6 Hz), 3.55–3.48 (m, 4H, CH 2‐46a), 3.39–3.30 (m, 4H, CH 2‐46b), –2.75 (s, 2H, NH); 13C{1H}‐NMR (CDCl3, 126 MHz): ∂ 158.72 (ArC‐32,38,44), 158.01 (ArC‐26), 156.99 (C=O‐51), 146.58–146.49 (ArC‐47), 138.01 (ArC‐22), 135.77 (ArC‐28,34,40), 134.03 (ArC‐21), 133.64 (ArC‐53), 132.34 (ArC‐24), 131.80–131.76 (ArC‐27,33,39), 129.93–129.83 (ArC‐49), 129.64 (ArC‐30,36,42), 128.52–128.45 (ArC‐55,56), 128.11 (ArC‐54), 119.88–119.84 (ArC‐29,35,41), 115.59–115.39 (ArC‐6,11,16), 115.18 (ArC‐48), 113.51 (ArC‐25), 113.16 (ArC‐1), 112.33 (ArC‐23), 111.93–111.87 (ArC‐31,37,43), 84.77 (C‐52), 67.43 (CH2‐46), 66.86 (CH2‐45), 44.41 (CH2‐50); UV/Vis (CHCl3): λ/nm (log (ε/M–1 cm–1)) 421 (5.46), 516 (4.18); Accurate mass: m/z 1423.39108 (M(79Br)+H)+, 1425.39106 (M(81Br)+H)+; calcd. for C84H64BrN8O10: 1423.39288 (79Br), 1425.39083 (81Br).</p><p>Zinc mono‐bromo porphyrin cage compound (9). The mono‐bromo porphyrin cage compound (8, 220 mg, 0.15 mmol, 1 equiv.) was dissolved in chloroform (15 mL) and methanol (7.5 mL). Zinc(II) acetate dihydrate (119.60 mg, 0.54 mmol, 3.5 equiv.) was added and the reaction mixture was stirred at r.t. for 1 h. After evaporating the mixture to dryness, the residue was purified by silica gel column chromatography using chloroform as the eluent. Precipitation from DCM/n‐heptane followed by washing with n‐pentane (4 ×) and drying in vacuo yielded the product (9, 229.4 mg, 0.15 mmol, 100 %, racemate of two enantiomers) as a pink/purple solid.</p><p>1H‐NMR (CDCl3, 500 MHz): ∂ 8.94–8.87 (m, 4H, β‐pyrrole‐H‐3,4,13,14), 8.80–8.73 (m, 4H, β‐pyrrole‐H‐8,9,18,19), 8.23 (d, 1H, ArH‐22, J = 2.5 Hz), 8.12–8.03 (m, 3H, ArH‐28,34,40), 7.85 (dd, 1H, ArH‐24, J = 8.9, 2.6 Hz), 7.78–7.72 (m, 3H, ArH‐30,36,42), 7.40–7.35 (m, 3H, ArH‐29,35,41), 7.35–7.31 (m, 3H, ArH‐31,37,43), 7.21 (d, 1H, ArH‐25, J = 8.9 Hz), 6.99–6.90 (m, 6H, ArH‐55,56), 6.77–6.72 (m, 4H, ArH‐54), 6.11 (s, 3H, ArH‐48(II, III, IV)), 6.10 (s, 1H, ArH‐48(I)), 4.26–4.16 (m, 4H, CH 2‐45a), 4.09 (d, 4H, CH 2‐50a, J = 15.6 Hz), 4.06–3.94 (m, 4H, CH 2‐45b), 3.66 (d, 4H, CH 2‐50b, J = 15.8 Hz), 3.55–3.46 (m, 4H, CH 2‐46a), 3.32–3.23 (m, 4H, CH 2‐46b); 13C{1H}‐NMR (CDCl3, 126 MHz): ∂ 158.75 (ArC‐32,38,44), 158.04 (ArC‐26), 156.84 (O=C‐51), 150.23–149.45 (ArC‐2,5,7,10,12,15,17,20), 146.51–146.42 (ArC‐47), 137.89 (ArC‐22), 135.56 (ArC‐28,34,40), 134.78 (ArC‐21), 133.58 (ArC‐53), 132.53 (ArC‐27,33,39), 132.10 (ArC‐24), 131.70–130.50 (ArC‐3,4,8,9,13,14,18,19), 129.89–129.79 (ArC‐49), 129.41 (ArC‐30,36,42), 128.54–128.46 (ArC‐55,56), 128.05 (ArC‐54), 119.85 (ArC‐29,35,41), 116.51–116.28 (ArC‐6,11,16), 115.19 (ArC‐48), 114.11 (ArC‐1), 113.68 (ArC‐25), 112.33 (ArC‐23), 112.11–112.04 (ArC‐31,37,43), 84.68 (C‐52), 67.45 (CH2‐46), 66.94 (CH2‐45), 44.26 (CH2‐50); UV/Vis (CHCl3) λ/nm (log (ε/M–1 cm–1)) 422 (5.64), 549 (4.29); MALDI‐TOF; m/z: 1485.3 (M(79Br)+H)+, 1487.3 (M(81Br)+H)+; calcd. for C84H62BrN8O10Zn 1485.3 (79Br); 1487.3 (81Br).</p><p>Potassium triisopropylsilylacetylene trifluoroborate (TIPSA BF3K). This compound was synthesized according to a modified literature procedure.[ 16 ]A solution of triisopropylsilylacetylene (2.0 mL, 8.9 mmol) was dissolved in degassed THF (20 mL) and cooled to –78 °C. n‐Butyllithium (5.6 mL, 8.9 mmol) was added dropwise and the reaction mixture was stirred for 1 hour. Trimethoxyborate (1.5 mL, 13 mmol) was added dropwise, the mixture was stirred for 1 hour and then warmed up to –20 °C over the course of 1 hour. Subsequently, a saturated solution of potassium bifluoride (4.30 g, 55.1 mmol) in water was added under vigorous stirring. The mixture was stirred at –20 °C for 1 hour, after which it was warmed to room temperature. The solvent was evaporated and the resulting white residue was dried under vacuum for 2 hours. The residue was washed with subsequently acetone and hot acetone and the filtrate was evaporated to yield a white solid, which was re‐precipitated from hot acetone/petroleum ether (80–100). After cooling the suspension to –20 °C, the precipitate was filtered off and the residue was washed with n‐pentane. After drying in air and under vacuum, the product was obtained as a white, fluffy solid (1.0 g, 3.6 mmol, 40 %).</p><p>1H‐NMR ([D6]Acetone, 400 MHz): ∂ 1.08–1.10 (m, 18H, CH(CH 3)2), 0.93–1.03 (m, 3H, CH(CH3)2); 19F‐NMR ([D6]acetone, 380 MHz): ∂ 135.5 (1:1:1:1, J = 34 Hz); 29Si‐NMR ([D6]acetone, 80 MHz): ∂ (ppm) –5.63 (s); 11B‐NMR ([D6]acetone, 133 MHz) δ –2.15 (q, J = 36 Hz).</p><p>TIPS‐protected acetylene functionalized zinc porphyrin cage compound 10. Compound 9 (288 mg, 0.19 mmol), potassium triisopropylsilylacetylene trifluoroborate (111.7 mg, 0.39 mmol), cesium carbonate (315.8 mg, 0.97 mmol), and [1,1'‐bis(diphenylphosphino) ferrocene]dichloropalladium(II) (39.8 mg, 0.05 mmol) were dissolved in THF (42.75 mL) and water (2.25 mL) under an argon atmosphere. The reaction mixture was refluxed for 16 h. After cooling to r.t., the mixture was diluted with DCM and the solution was subsequently washed with water (4 × 50 mL). The organic layer was dried with MgSO4, filtered, and the solvents evaporated to dryness. The crude product was purified by silica gel column chromatography using 1 % MeOH in CHCl3 (v/v) as the eluent yielding 10 (285.4 mg, 0.18 mmol, 93 %, racemate of two enantiomers) as a pink/purple solid.</p><p>1H‐NMR (CDCl3, 500 MHz): ∂ 8.93–8.85 (m, 4H, β‐pyrrole‐H‐3,4,13,14), 8.74–8.67 (m, 4H, β‐pyrrole‐H‐8,9,18,19), 8.19 (d, 1H, ArH‐22, J = 2.1 Hz), 8.01–7.96 (m, 2H, ArH‐28,40), 7.95–7.91 (m, 1H, ArH‐34), 7.87 (dd, 1H, ArH‐24, J = 8.5, 2.1 Hz), 7.74–7.68 (m, 3H, ArH‐30,36,42), 7.33–7.29 (m, 3H, ArH‐31,37,43), 7.29–7.24 (m, 3H, ArH‐29,35,41), 7.22 (d, 1H, ArH‐25, J = 8.6 Hz), 6.97–6.91 (m, 6H, ArH‐55,56), 6.44 (bs, 4H, ArH‐54), 5.86–5.73 (m, 4H, ArH‐48), 4.17–4.07 (m, 4H, CH 2‐45a), 3.95–3.84 (m, 4H, CH 2‐45b), 3.55 (bs, 4H, CH 2‐50a), 3.47–3.39 (m, 4H, CH 2‐46a), 3.39–3.29 (m, 4H, CH 2‐50b), 3.17–3.05 (m, 4H, CH 2‐46b), 1.07 (bs, 21H, TIPS); 13C{1H}‐NMR (CDCl3, 126 MHz): ∂ 159.05 (ArC‐26), 158.84–158.81 (ArC‐32,44), 158.76 (ArC‐38), 156.40–156.37 (C=O‐51), 150.10–149.60 (ArC‐2,5,7,10,12,15,17,20), 146.32–146.08 (ArC‐47), 138.60 (ArC‐22), 135.54 (ArC‐28,40), 135.39 (ArC‐34), 133.51 (ArC‐24), 133.23 (ArC‐53), 132.85 (ArC‐21), 132.79–132.75 (ArC‐27,33,39), 131.50–130.50 (ArC‐3,4,8,9,13,14,18,19), 129.37 (ArC‐49), 129.31–129.22 (ArC‐30,36,42), 128.51 (ArC‐55,56), 127.86 (ArC‐54), 119.94–119.82 (ArC‐29,35,41), 116.07 (ArC‐11), 115.91–115.88 (ArC‐6,16), 114.87 (ArC‐23), 114.78 (ArC‐48), 114.38 (ArC‐1), 112.39–112.27 (ArC‐31,37,43), 111.92 (ArC‐25), 107.10 (ArC‐57), 88.95 (ArC‐58), 84.37 (C‐52), 67.37–67.19 (CH2‐46), 67.02–66.96 (CH2‐45), 43.76 (CH2‐50), 18.69 (TIPS‐C), 11.33 (TIPS‐C); MALDI‐TOF: m/z 1587.5 [M + H]+; calcd. for C95H83N8O10SiZn: 1587.5.</p><p>Acetylene functionalized zinc porphyrin cage compound 11. Argon was led through THF (75 mL) for 30 min and compound 10 (372 mg, 0.23 mmol) and tetrabutylammonium fluoride trihydrate (739.5 mg, 2.34 mmol) were added. The reaction mixture was stirred in the dark at r.t. for 4 h and was then evaporated to dryness. The residue was dissolved in DCM and this solution was washed with water (3 × 50 mL), dried with Na2SO4, filtered and the solvents evaporated to dryness. The crude product was purified by silica gel column chromatography using 0.5 % MeOH in CHCl3 (v/v) as the eluent. Precipitation from DCM/n‐heptane followed by washing with n‐pentane (4 ×) and drying in vacuo yielded 11 (244 mg, 0.17 mmol, 73 %, racemate of two enantiomers) as a pink/purple solid.</p><p>1H‐NMR (CDCl3, 500 MHz): ∂ 8.91–8.87 (m, 4H, β‐pyrrole‐H‐3,4,13,14), 8.77–8.71 (m, 4H, β‐pyrrole‐H‐8,9,18,19), 8.24 (d, 1H, ArH‐22, J = 2.2 Hz), 8.10–8.03 (m, 3H, ArH‐28,34,40), 7.89 (dd, 1H, ArH‐24, J = 8.6, 2.2 Hz), 7.77–7.72 (m, 3H, ArH‐30,36,42), 7.38–7.34 (m, 3H, ArH‐29,35,41), 7.34–7.32 (m, 3H, ArH‐31,37,43), 7.27 (d, 1H, ArH‐25, J = 8.7 Hz), 6.97–6.92 (m, 6H, ArH‐55,56), 6.74–6.68 (m, 4H, ArH‐54), 6.09–6.05 (m, 4H, ArH‐48), 4.28–4.15 (m, 4H, CH 2‐45a), 4.07–3.96 (m, 8H, CH 2‐45b, CH 2‐50a), 3.63 (d, 4H, CH 2‐50b, J = 16.0 Hz), 3.54–3.47 (m, 4H, CH 2‐46a), 3.32–3.22 (m, 4H, CH 2‐46b), 3.05 (s, 1H, alkyneH‐58); 13C{1H}‐NMR (CDCl3, 126 MHz): ∂ 159.22 (ArC‐26), 158.80–158.74 (ArC‐32,38,44), 156.80 (C=O‐51), 150.18–149.55 (ArC‐2,5,7,10,12,15,17,20), 146.49–146.35 (ArC‐47), 139.07 (ArC‐22), 135.58–135.59 (ArC‐28,34,40), 133.54 (ArC‐24), 133.50 (ArC‐53), 132.81 (ArC‐21), 132.67–132.59 (ArC‐27,33,39), 131.60–130.56 (ArC‐3,4,8,9,13,14,18,19), 129.83–129.67 (ArC‐49), 129.36 (ArC‐30,36,42), 128.53–128.47 (ArC‐55,56), 128.04 (ArC‐54), 119.85 (ArC‐29,35,41), 116.32–116.13 (ArC‐6,11,16), 115.16–114.97 (ArC‐48), 114.46 (ArC‐1), 113.37 (ArC‐23), 112.18–112.07 (ArC‐31,37,43), 111.70 (ArC‐25), 84.66 (C‐52), 83.76 (alkyneC‐57), 76.12 (alkyneC‐58), 67.45–67.40 (CH2‐46), 66.95–66.93 (CH2‐45), 44.21 (CH2‐50); MALDI‐TOF: m/z 1431.3 [M + H]+; calcd. for C86H63N8O10Zn: 1431.4.</p><p>1,3‐Diazidopropane. This compound was synthesized according to a modified literature procedure.[ 17 ] Sodium azide‐1‐15N (490 mg, 7.43 mmol) and 1,3‐dibromopropane (0.25 mL, 2.48 mmol) were dissolved in DMF (16 mL) and the reaction mixture was heated at 85 °C for 20 h. After cooling to r.t., water (100 mL) was added to the mixture and the resulting solution was extracted with diethyl ether. Subsequently the organic layer was washed with water (5 × 100 mL) and brine (2 × 100 mL), dried with Na2SO4, filtered, and evaporated to almost dryness to yield the product as a clear solution in diethyl ether (96 %, based on 1H NMR integrals). Due to its instability the product was directly used in the following synthesis step.</p><p>1H NMR (CDCl3, 500 MHz): ∂ 3.42 (t, 4H, N3CH 2CH2CH 2N3, J = 6.5 Hz), 1.87–1.80 (m, 2H, N3CH2CH 2CH2N3); 13C NMR (CDCl3, 126 MHz): ∂ 48.58 (N3 CH2CH2 CH2N3), 28.47 (N3CH2 CH2CH2N3); 15N NMR† (CDCl3, 51 MHz): ∂ 247.78 (s, N*NN*CH2CH2CH2N*NN*), 211.36 (s, N*NN*CH2CH2CH2N*NN*), 69.69 (s, N*NN*CH2CH2CH2 N*NN*); †50 % of the indicated N atoms are 15N‐labeled.</p><p>1,5‐Diazidopentane. This compound was synthesized according to a modified literature procedure.[ 17 ] Sodium azide‐1‐15N (486 mg, 7.37 mmol) and 1,5‐dibromopentane (0.35 mL, 2.57 mmol) were dissolved in DMF (15 mL) and the reaction mixture was heated at 80 °C for 16 h. After cooling to r.t., water (100 mL) was added to the mixture and the resulting solution was extracted with diethyl ether (2 × 150 mL). Subsequently the organic layer was washed with brine (2 × 100 mL), dried with Na2SO4, filtered and the evaporated to almost dryness to yield the product as a clear solution in diethyl ether (75 %, based on 1H NMR integrals). Due to its instability the product was directly used in the following synthesis step.</p><p>1H NMR (CDCl3, 500 MHz): ∂ 3.29 (t, 4H, N3CH 2(CH2)3CH 2N3, J = 6.8 Hz), 1.60–1.52 (m, 4H, N3CH2CH 2CH2CH 2CH2N3), 1.45–1.35 (m, 2H, N3(CH2)2CH 2(CH2)2N3); 15N NMR† (CDCl3, 51 MHz): ∂ 242.94 (s, N*NN*CH2(CH2)3CH2N*NN*), 205.87 (s, N*NN*CH2(CH2)5CH2N*NN*), 66.78 (s, N*NN*CH2(CH2)3CH2 N*NN*); †50 % of the indicated N atoms are 15N‐labeled.</p><p>1,11‐Diazidoundecane. This compound was synthesized according to a modified literature procedure.[ 17 ] Sodium azide‐1‐15N (500 mg, 7.58 mmol, 3 equiv.) and 1,11‐dibromoundecane (0.60 mL, 2.53 mmol, 1 equiv.) were dissolved in DMF (15 mL). The reaction mixture was stirred at 80 °C for 20 h. After cooling to r.t., water (150 mL) was added to the mixture and the resulting solution was extracted with diethyl ether (2 × 100 mL). Subsequently, the organic layer was washed with water (10 × 200 mL) and brine (2 × 200 mL), dried with Na2SO4, filtered and evaporated to almost dryness to yield the product as a clear solution in diethyl ether (45 %, based on 1H NMR integrals). Due to its instability the product was directly used in the following synthesis step.</p><p>1H‐NMR (CDCl3, 500 MHz): ∂ 3.26 (t, 4H, N3CH 2(CH2)9CH 2N3, J = 7 Hz), 1.62–1.56 (m, 4H, N3CH2CH 2(CH2)7CH 2CH2N3), 1.39–1.28 (m, 14H, N3(CH2)2 (CH 2 ) 7(CH2)2N3); 13C NMR (CDCl3, 126 MHz): ∂ 51.49 & 51.46 (N3 CH2(CH2)9 CH2N3), 29.44 & 29.41 (N3(CH2)3(CH2)5(CH2)3N3),, 28.8 (N3CH2 CH2(CH2)7 CH2CH2N3), 26.7 (N3(CH2)2 CH2(CH2)5 CH2(CH2)2N3; 15N NMR† (CDCl3, 51 MHz): δ = 248.5 (s, N*NN*CH2(CH2)9CH2N*NN*), 210.5 (s, N*NN*CH2(CH2)9CH2N*NN*), 71.9 (s, N*NN*CH2(CH2)9CH2 N*NN*); †50 % of the indicated N atoms are 15N‐labeled.</p><p>Double porphyrin cage compound Zn2C3DC. To a mixture of compound 11 (135.5 mg, 0.09 mmol) and copper(I) iodide (9.36 mg, 0.05 mmol) in freshly distilled THF (19 mL) and freshly distilled MeCN (19 mL), DIPEA (15.80 µL, 0.09 mmol) was added. A solution of 1,3‐diazidopropane‐15N‐enriched in diethyl ether (74.3 µL, 0.57 m, 0.04 mmol) was added and the reaction mixture was stirred in the dark at r.t. for 9 days. On days 2 and 4 additional CuI (12.08 mg and 21.15 mg, resp.) and DIPEA (2 × 15.80 µL) were added. The mixture was diluted with DCM (50 mL) and washed with water (3 × 50 mL). The organic layer was evaporated and the crude product was purified by silica gel column chromatography using a gradient of 1–5 % of MeOH in CHCl3 (v/v) as the eluent. Precipitation from DCM/n‐heptane followed by washing with n‐pentane (4×) and drying in vacuo yielded Zn2C3DC (44.8 mg, 0.02 mmol, 35 %) as a pink/purple solid.</p><p>1H‐NMR ([D6]DMSO at 333.15 K, 500 MHz): ∂ 8.79 (dd, 2H, β‐pyrrole‐H, J = 4.6, 1.2 Hz), 8.75–8.70 (m, 6H, β‐pyrrole‐H), 8.58–8.49 (m, 8H, β‐pyrrole‐H) 8.50–8.47 (m, 2H, Triazole‐H‐58), 8.33 (t, 2H, ArH‐22, J = 2.4 Hz), 8.19 (dd, 1H, ArH‐24*, J = 8.5, 2.3 Hz), 8.18 (dd, 1H, ArH‐24*, J = 8.5, 2.3 Hz) 7.92–7.82 (m, 6H, ArH‐28, 34, 40), 7.79–7.66 (m, 6H, ArH‐30, 36, 42), 7.56–7.48 (m, 6H, ArH‐31, 37, 43), 7.46 (d, 1H, ArH‐25*, J = 8.7 Hz), 7.39 (d, 1H, ArH‐25*, J = 8.9 Hz), 7.37–7.19 (m, 6H, ArH‐29, 35, 41), 7.05–6.94 (m, 12H, ArH‐55, 56), 6.84–6.78 (m, 6H, ArH‐48(II,III,IV)), 6.20 (s, 1H, ArH‐48(I)*), 6.19 (s, 1H, ArH‐48(I)*), 4.43 (t, 4H, CH 2‐62, J = 6.8 Hz), 4.22–4.03 (m, 24H, CH 2‐45, CH 2‐50a), 3.65 (d, 8H, CH 2‐50b, J = 15.7 Hz), 3.51–3.40 (m, 8H, CH 2‐46a), 3.28–3.22 (m, 8H, CH 2‐46b), 2.51–2.46 (m, 2H, CH 2‐63); 13C NMR‡ ([D6]DMSO at 333.15 K, 126 MHz): ∂ 158.40 (ArC‐32, 38, 44), 156.00 (C=O‐51), 155.86 (C=O‐51), 149.37 (ArC‐2, 5, 7, 10, 12, 15, 17, 20), 148.57 (ArC‐2, 5, 7, 10, 12, 15, 17, 20), 146.07 (C‐57), 145.69 (ArC‐47), 134.53 (ArC‐28, 34, 40), 133.06 (ArC‐53), 132.40 (ArC‐27, 33, 39), 131.70 (ArC‐22), 130.21 (ArC‐3, 4, 8, 9, 13, 14, 18, 19), 129.89 (ArC‐3, 4, 8, 9, 13, 14, 18, 19), 129.81 (ArC‐49), 128.83 (ArC‐30, 36, 42), 127.79 (ArC‐55, 56), 127.40 (ArC‐54), 125.67 (ArC‐24), 120.44 (CH‐58), 119.16 (ArC‐29, 35, 41), 114.79 (ArC‐48(II,III,IV)), 114.46 (ArC‐48(I)), 112.62 (ArC‐25), 112.60 (ArC‐31, 37, 43), 84.00 (C‐52), 83.83 (C‐52), 67.06 (CH2‐46), 66.66 (CH2‐45), 46.44 (CH2‐62), 43.18 (CH2‐50), 29.28 (CH2‐63); 15N{1H} NMR† ([D6]DMSO at 333.15 K, 51 MHz): ∂ 346.64 (Triazole‐N‐61*), 346.56 (Triazole‐N‐61*), 248.62 (Triazole‐N‐59*), 248.57 (Triazole‐N‐59*); UV/Vis (CHCl3/CH3CN 1:1, v/v) λ/nm (log (ε/M–1 cm–1)) 426 (5.71), 559 (4.34), 597 (3.91); Accurate MALDI‐TOF: m/z 2988.7613 (M)+; calcd. for C175H130N20 15N2O20 64Zn2: 2988.8355. *Peaks belonging to the two different diastereoisomers that could not be separated. ‡13C‐NMR shifts obtained via HSQC and HMBC spectra as no 13C{1H}‐NMR spectrum was recorded. †50 % of the indicated N atoms are 15N‐labeled. For UV/Vis spectra, see text.</p><p>Double porphyrin cage compound Zn2C5DC. To a mixture of compound 11 (134.1 mg, 0.09 mmol) and copper(I) iodide (10.2 mg, 0.05 mmol) in freshly distilled THF (19 mL) and freshly distilled MeCN (19 mL), DIPEA (15.80 µL, 0.09 mmol) was added. A solution of 1,5‐diazidopentane‐15N‐enriched in diethyl ether (51.2 µL, 0.83 m, 0.04 mmol) was added and the reaction mixture was stirred in the dark at r.t. for 7 days. On days 2 and 4 additional CuI (20.26 mg and 10.44 mg, resp.) and DIPEA (2 × 15.80 µL) were added. The mixture was concentrated in vacuo and the crude product was purified by silica gel column chromatography using a gradient of 0.7–5 % MeOH in CHCl3 (v/v) as the eluent. Precipitation from DCM/n‐heptane followed by washing with n‐pentane (4 ×) and drying in vacuo yielded Zn2C5DC (39.7 mg, 0.01 mmol, 31 %) as a pink/purple solid.</p><p>1H‐NMR ([D6]DMSO at 333.15 K, 500 MHz): ∂ 8.80 (dd, 2H, β‐pyrrole‐H, J = 4.7, 3.5 Hz), 8.76–8.72 (m, 6H, β‐pyrrole‐H), 8.59 (d, 2H, β‐pyrrole‐H, J = 4.6 Hz) 8.55–8.52 (m, 6H, β‐pyrrole‐H), 8.48–8.45 (m, 2H, Triazole‐H‐58), 8.34 (t, 2H, ArH‐22, J = 1.9 Hz), 8.20 (dd, 2H, ArH‐24, J = 8.6, 2.2 Hz), 7.91–7.86 (m, 6H, ArH‐28, 34, 40), 7.79–7.67 (m, 6H, ArH‐30, 36, 42), 7.54–7.46 (m, 6H, ArH‐31, 37, 43), 7.44 (dd, 2H, ArH‐25*, J = 8.8, 5.9 Hz), 7.38–7.27 (m, 6H, ArH‐29, 35, 41), 7.04–6.95 (m, 12H, ArH‐55, 56), 6.83–6.78 (m, 8H, ArH‐54), 6.27–6.23 (m, 6H, ArH‐48(II,III,IV)), 6.18 (s, 1H, ArH‐48(I)*), 6.16 (s, 1H, ArH‐48(I)*), 4.41 (t, 4H, CH 2‐62, J = 6.9 Hz), 4.20–4.00 (m, 24H, CH 2‐45, CH 2‐50a), 3.68–3.57 (m, 8H, CH 2‐50b), 3.53–3.40 (m, 8H, CH 2‐46a), 3.27–3.20 (m, 8H, CH 2‐46b), 1.85 (p, 4H, CH 2‐63, J = 7.3 Hz), 1.30–1.22 (m, 2H, CH 2‐64); 13C NMR‡ ([D6]DMSO at 333.15 K, 126 MHz): ∂ 158.27 (ArC‐32, 38, 44), 157.92 (ArC‐26), 155.88 (C=O‐51), 155.72 (C=O‐51), 149.20 (ArC‐2, 5, 7, 10, 12, 15, 17, 20), 148.47 (ArC‐2, 5, 7, 10, 12, 15, 17, 20), 145.84 (C‐57), 145.55 (ArC‐47(II, III, IV)), 145.42 (ArC‐47(I)), 134.60 (ArC‐28, 34, 40), 132.92 (ArC‐53), 132.42 (ArC‐21), 132.24 (ArC‐27, 33, 39), 131.75 (ArC‐22), 130.28 (ArC‐3, 4, 8, 9, 13, 14, 18, 19), 129.91 (ArC‐3, 4, 8, 9, 13, 14, 18, 19), 129.66 (ArC‐49(II, III, IV)), 129.51 (ArC‐49(I)), 128.89 (ArC‐30, 36, 42), 127.84 (ArC‐55, 56), 127.41 (ArC‐54), 125.68 (ArC‐24), 122.16 (ArC‐23), 120.17 (CH‐58), 119.21 (ArC‐29, 35, 41), 115.13 (ArC‐6, 11, 16), 114.82 (ArC‐48(II,III,IV)), 114.67 (ArC‐1), 114.47 (ArC‐48(I)*), 114.36 (ArC‐48(I)*), 112.67 (ArC‐25, 31, 37, 43), 83.83 (C‐52), 83.71 (C‐52), 67.07 (CH2‐46), 66.69 (CH2‐45), 48.68 (CH2‐62), 43.19 (CH2‐50), 28.33 (CH2‐64), 28.27 (CH2‐63); 15N{1H} NMR† ([D6]DMSO at 333.15 K, 51 MHz): ∂ 345.74 (Triazole‐N‐61*), 345.72 (Triazole‐N‐61*), 250.74 (Triazole‐N‐59*), 250.66 (Triazole‐N‐59*); UV/Vis (CHCl3/CH3CN 1:1, v/v) λ/nm (log (ε/M–1 cm–1)) 426 (5.95), 558 (4.57), 598 (3.94); Accurate MALDI‐TOF: m/z 3016.8603 (M)+; calcd. for C177H134N20 15N2O20 64Zn2: 3016.8668. *Peaks belonging to the two different diastereoisomers that could not be separated. ‡13C‐NMR shifts obtained via HSQC and HMBC spectra as no 13C{1H}‐NMR spectrum was recorded. †50 % of the indicated N atoms are 15N‐labeled.</p><p>Double porphyrin cage compound Zn2C11DC. To a solution of compound 11 (134.8 mg, 0.09 mmol) and copper(I) iodide (11.4 mg, 0.06 mmol) in freshly distilled THF (19 mL) and freshly distilled MeCN (19 mL), DIPEA (15.80 µL, 0.09 mmol) was added. A solution of 1,11‐diazidoundecane‐15N‐enriched in diethyl ether (304 µL, 0.14 m, 0.04 mmol) was added and the reaction mixture was stirred in the dark at r.t. for 6 days. On days 3 and 5 additional CuI (16.78 mg and 19.24 mg, resp.) and DIPEA (2 × 15.80 µL) were added. The mixture was concentrated in vacuo and the crude products was purified by silica gel column chromatography using a gradient of 0.5 %‐5 % MeOH in CHCl3 (v/v) as the eluent. Precipitation from DCM/n‐heptane followed by washing with n‐pentane (4 ×) and drying in vacuo yielded Zn2C11DC (81.9 mg, 0.03 mmol, 62 %) as a pink/purple solid.</p><p>1H‐NMR ([D6]DMSO at 333.15 K, 500 MHz): ∂ 8.80 (dd, 2H, β‐pyrrole‐H, J = 4.6, 2.1 Hz), 8.76–8.70 (m, 6H, β‐pyrrole‐H), 8.59 (t, 2H, β‐pyrrole‐H, J = 4.2 Hz) 8.56–8.49 (m, 6H, β‐pyrrole‐H), 8.47–8.44 (m, 2H, Triazole‐H‐58), 8.36–8.34 (m, 2H, ArH‐22), 8.23 (dt, 2H, ArH‐24, J = 8.7, 1.7 Hz), 7.92–7.88 (m, 6H, ArH‐28, 34, 40), 7.79–7.68 (m, 6H, ArH‐30, 36, 42), 7.56 (d, 2H, ArH‐25, J = 8.8 Hz), 7.54–7.41 (m, 6H, ArH‐31, 37, 43), 7.38–7.27 (m, 6H, ArH‐29, 35, 41), 7.04–6.95 (m, 12H, ArH‐55, 56), 6.83–6.79 (m, 8H, ArH‐54), 6.26–6.20 (m, 8H, ArH‐48), 4.24 (t, 4H, CH 2‐62, J = 6.9 Hz), 4.21–4.04 (m, 24H, CH 2‐45, CH 2‐50a), 3.69–3.59 (m, 8H, CH 2‐50b), 3.53–3.38 (m, 8H, CH 2‐46a), 3.29–3.18 (m, 8H, CH 2‐46b), 1.79–1.71 (m, 4H, CH 2‐63), 1.19–1.11 (m, 14H, CH 2‐64, 65, 66, 67); 13C NMR‡ ([D6]DMSO at 333.15 K, 126 MHz): ∂ 158.56 (ArC‐32, 38, 44), 158.23 (ArC‐26), 156.16 (C=O‐51), 156.03 (C=O‐51), 149.51 (ArC‐2, 5, 7, 10, 12, 15, 17, 20), 148.75 (ArC‐2, 5, 7, 10, 12, 15, 17, 20), 146.13 (C‐57), 145.80 (ArC‐47), 134.57 (ArC‐28, 34, 40), 133.23 (ArC‐53), 132.81 (ArC‐21), 132.56 (ArC‐27, 33, 39), 131.70 (ArC‐22), 130.23 (ArC‐3, 4, 8, 9, 13, 14, 18, 19), 129.93 (ArC‐49), 129.87 (ArC‐3, 4, 8, 9, 13, 14, 18, 19), 128.84 (ArC‐30, 36, 42), 127.79 (ArC‐55, 56), 127.34 (ArC‐54), 125.56 (ArC‐24), 122.52 (ArC‐23), 120.10 (CH‐58), 119.16 (ArC‐29, 35, 41), 115.52 (ArC‐6, 11, 16), 114.70 (ArC‐48), 112.66 (ArC‐25, 31, 37, 43), 84.08 (C‐52), 67.02 (CH2‐46), 66.61 (CH2‐45), 48.93 (CH2‐62), 43.20 (CH2‐50), 28.84 (CH2‐63), 28.08 (CH2‐62), 28.02 (CH2‐64), 27.67 (CH2‐65), 25.18 (CH2‐66, 67); 15N{1H} NMR† ([D6]DMSO at 333.15 K, 51 MHz): ∂ 345.55 (Triazole‐N‐61), 251.23 (Triazole‐N‐59); UV/Vis (CHCl3/CH3CN 1:1, v/v) λ/nm (log (ε/M–1 cm–1)) 426 (5.93), 559 (4.56), 597 (4.03); Accurate MALDI‐TOF: m/z 3100.9959 (M)+; calcd. for C183H146N20 15N2O20 64Zn2: 3100.9607. *Peaks belonging to the two different diastereoisomers that could not be separated. ‡13C‐NMR shifts obtained via HSQC and HMBC spectra as no 13C{1H}‐NMR spectrum was recorded. †50 % of the indicated N atoms are 15N‐labeled.</p><p>Double porphyrin cage compound H4C3DC. Compound Zn2C3DC (65.93 mg, 22 µmol) was dissolved in CHCl3 (100 mL) after which aq. HCl (6 m, 200 mL) was added and the reaction mixture was stirred at r.t. for 2 h. Subsequently, the organic layer was washed with sat. aq. NaHCO3 (200 mL) and water (200 mL), dried with Na2SO4, filtered, and the solvents evaporated to dryness. The crude product was purified by silica gel column chromatography using 1.5 % MeOH in CHCl3 (v/v) as the eluent. Precipitation from DCM/n‐heptane followed by washing with n‐pentane (4 ×) and drying in vacuo yielded H4C3DC (43.16 mg, 15 µmol, 68 %) as a purple solid.</p><p>1H‐NMR (CDCl3, 500 MHz): ∂ 8.80–8.76 (m, 4H, β‐pyrrole‐H), 8.76–8.70 (m, 4H, β‐pyrrole‐H), 8.69–8.65 (m, 4H, β‐pyrrole‐H), 8.65–8.59 (m, 4H, β‐pyrrole‐H), 8.40 (dd, 2H, ArH‐22, J = 4.3, 2.4 Hz), 8.28–8.21 (m, 2H, ArH‐24), 8.08–7.97 (m, 6H, ArH‐28, 34, 40), 7.79–7.75 (m, 2H, Triazole‐H‐58), 7.76–7.63 (m, 6H, ArH‐30, 36, 42), 7.38–7.22 (m, 12H, ArH‐29, 31, 35, 37, 41, 43), 7.26 (1H, ArH‐25#), 7.20 (1H, ArH‐25◊), 7.01–6.90 (m, 12H, ArH‐55, 56), 6.85–6.77 (m, 8H, ArH‐54), 6.23–6.17 (m, 6H, ArH‐48(II,III,IV)), 6.16 (s, 1H, ArH‐48(I)#), 6.13 (s, 1H, ArH‐48(I)◊), 4.32 (t, 4H, CH 2‐62, J = 6.1 Hz), 4.28–4.20 (m, 6H, CH 2‐45a(II, III, IV)), 4.23 (d, 8H, CH 2‐50a, J = 16.2 Hz), 4.20 (1H, CH 2‐45a(I)#), 4.12 (1H, CH 2‐45a(I)◊), 4.08–4.00 (m, 6H, CH 2‐45b(II, III, IV)), 3.96 (1H, CH 2‐45b(I)#), 3.88 (1H, CH 2‐45b(I)◊), 3.80–3.67 (m, 8H, CH 2‐50b, J = 15.4 Hz), 3.55–3.42 (m, 6H, CH 2‐46a(II, III, IV)), 3.47 (1H, CH 2‐46a(I)#), 3.39–3.28 (m, 6H, CH 2‐46b(II, III, IV)), 3.37 (1H, CH 2‐46a(I)◊), 3.33 (1H, CH 2‐46b(I)#), 3.27 (1H, CH 2‐46b(I)◊), 2.52–2.41 (m, 2H, CH 2‐63), –2.74 (s, 2H, NH); 13C NMR (CDCl3, 126 MHz): ∂ 158.86 (ArC‐26), 158.76 (ArC‐32, 38, 44), 157.07 (C=O‐51), 147.84 (C‐57), 146.62 (ArC‐47), 135.78 (ArC‐28, 34, 40), 133.21 (ArC‐22), 131.90 (ArC‐27, 33, 39), 132.20 (ArC‐21), 129.85 (ArC‐49), 129.63 (ArC‐30, 36, 42), 128.53 (ArC‐55, 56), 128.20 (ArC‐54), 127.13 (ArC‐24), 122.10 (ArC‐23), 120.07 (CH‐58), 119.83 (ArC‐29, 35, 41), 115.30 (ArC‐6, 11, 16) 115.27 (ArC‐48(II,III,IV)), 115.01 (ArC‐48(I)), 114.50 (ArC‐1), 112.20 (ArC‐25), 111.92 (ArC‐31, 37, 43), 84.78 (C‐52), 67.44 (CH2‐46(II, III, IV)), 67.20 (CH2‐46(I)#), 67.13 (CH2‐46(I)◊), 66.90 (CH2‐45(II, III, IV)), 66.87 (CH2‐45(I)#), 66.78 (CH2‐45(I)◊), 46.65 (CH2‐62), 44.43 (CH2‐50), 30.54 (CH2‐63); 15N{1H}‐NMR† (CDCl3, 51 MHz): ∂ 346.55 (Triazole‐N‐61*), 346.44 (Triazole‐N‐61*), 244.89 (Triazole‐N‐59*), 244.77 (Triazole‐N‐59*); UV/Vis (CHCl3/CH3CN 1:1, v/v) λ/nm (log (ε/M–1 cm–1)) 418 (5.81), 514 (4.50), 546 (4.00), 590 (4.00), 6.43 (3.67); Accurate mass: m/z 2867.01780 (M+2H)+; calcd. for C175H136N20 15N2O20: 2867.02419. ◊/#Peaks belonging to the two diastereoisomers. *Peaks belonging to the two different diastereoisomers that could not be separated. †50 % of the indicated N atoms are 15N‐labeled.</p><p>Double porphyrin cage compound H4C5DC. This compound was synthesized as described for H4C3DC, using Zn2C5DC (45.53 mg, 15.1 µmol), CHCl3 (75 mL) and aq. HCl (6 m, 200 mL) Compound H4C5DC was obtained as a purple solid (35.66 mg, 12.3 µmol, 81 %).</p><p>1H‐NMR (CDCl3, 500 MHz): ∂ 8.80–8.76 (m, 4H, β‐pyrrole‐H), 8.76–8.70 (m, 4H, β‐pyrrole‐H), 8.69–8.60 (m, 8H, β‐pyrrole‐H), 8.36 (t, 2H, ArH‐22, J = 2.0 Hz), 8.33–8.28 (m, 2H, ArH‐24), 8.06–7.98 (m, 6H, ArH‐28, 34, 40), 7.77–7.61 (m, 6H, ArH‐30, 36, 42), 7.65 (2H, Triazole‐H‐58), 7.37–7.23 (m, 12H, ArH‐29, 31, 35, 37, 41, 43), 7.28 (1H, ArH‐25#), 7.25 (1H, ArH‐25◊), 6.98 –6.91 (m, 12H, ArH‐55, 56), 6.84–6.78 (m, 8H, ArH‐54), 6.21–6.17 (m, 6H, ArH‐48(II,III,IV)), 6.16 (s, 1H, ArH‐48(I)#), 6.14 (s, 1H, ArH‐48(I)◊), 4.30 (t, 4H, CH 2‐62, J = 6.9 Hz), 4.27–4.15 (m, 6H, CH 2‐45a(II, III, IV)), 4.23 (d, 8H, CH 2‐50a, J = 16.0 Hz), 4.20 (1H, CH 2‐45a(I)#), 4.18 (1H, CH 2‐45a(I)◊), 4.08–3.93 (m, 6H, CH 2‐45b(II, III, IV)), 4.00 (1H, CH 2‐45b(I)#), 3.95 (1H, CH 2‐45b(I)◊), 3.74 (d, 8H, CH 2‐50b, J = 15.7 Hz), 3.53–3.41 (m, 6H, CH 2‐46a(II, III, IV)), 3.47 (1H, CH 2‐46a(I)#), 3.37–3.27 (m, 6H, CH 2‐46b(II, III, IV)), 3.44 (1H, CH 2‐46a(I)◊), 3.32 (2H, CH 2‐46b(I)*), 1.94–1.87 (m, 4H, CH 2‐63), 1.38–1.33 (m, 2H, CH 2‐64), –2.74 (s, 2H, NH); 13C NMR (CDCl3, 126 MHz): ∂ 158.75 (ArC‐26), 158.72 (ArC‐32, 38, 44), 156.90 (C=O‐51), 147.70 (C‐57), 146.61 (ArC‐47), 135.80 (ArC‐28, 34, 40), 133.31 (ArC‐22), 131.90 (ArC‐27, 33, 39), 131.33 (ArC‐21), 129.81 (ArC‐49), 129.51 (ArC‐30, 36, 42), 128.51 (ArC‐55, 56), 128.12 (ArC‐54), 126.80 (ArC‐24), 122.30 (ArC‐23), 119.79 (ArC‐29, 35, 41), 119.14 (CH‐58), 115.31 (ArC‐48(II,III,IV)), 115.30 (ArC‐6, 11, 16), 114.99 (ArC‐48(I)), 114.40 (ArC‐1), 112.04 (ArC‐25), 111.95 (ArC‐31, 37, 43), 84.77 (C‐52), 67.45 (CH2‐46(II, III, IV)), 67.21 (CH2‐46(I)*), 66.87 (CH2‐45), 49.90 (CH2‐62), 44.44 (CH2‐50), 29.62 (CH2‐63), 23.41(CH2‐64); 15N{1H}‐NMR† (CDCl3, 51 MHz): ∂ 344.66 (Triazole‐N‐61*), 344.61 (Triazole‐N‐61*), 247.82 (Triazole‐N‐59*); UV/Vis (CHCl3/CH3CN 1:1, v/v) λ/nm (log (ε/M–1 cm–1)) 420 (5.81), 515 (4.45), 5,46 (4.12), 590 (4.15), 642 (3.78); Accurate mass: m/z 2895.04174 (M+2H)+; calcd. for C177H140N20 15N2O20: 2895.05549. ◊/#Peaks belonging to the two diastereoisomers. *Peaks belonging to the two different diastereoisomers that could not be separated. †50 % of the indicated N atoms are 15N‐labeled.</p><p>Double porphyrin cage compound H4C11DC. This compound was synthesized as described for H4C3DC, using Zn2C11DC (39.03 mg, 12.6 µmol), CHCl3 (50 mL) and aq. HCl (6 m, 200 mL). Compound H4C11DC was obtained as a purple solid (34.59 mg, 11.6 µmol, 92 %).</p><p>1H‐NMR (CDCl3, 500 MHz): ∂ 8.80–8.77 (m, 4H, β‐pyrrole‐H), 8.76–8.72 (m, 4H, β‐pyrrole‐H), 8.71–8.61 (m, 8H, β‐pyrrole‐H), 8.35–8.33 (m, 2H, ArH‐22), 8.33–8.31 (m, 2H, ArH‐24), 8.07–8.01 (m, 6H, ArH‐28, 34, 40), 7.75–7.66 (m, 6H, ArH‐30, 36, 42), 7.56–7.52 (m, 2H, Triazole‐H‐58), 7.38–7.273 (m, 12H, ArH‐29, 31, 35, 37, 41, 43), 7.35 (2H, ArH‐25), 6.99 –6.91 (m, 12H, ArH‐55, 56), 6.84–6.78 (m, 8H, ArH‐54), 6.21–6.15 (m, 8H, ArH‐48), 4.29 (2H, CH 2‐45a(I)), 4.23 (6H, CH 2‐45a(II, III, IV)), 4.23 (d, 8H, CH 2‐50a, J = 15.6 Hz), 4.16 (t, 4H, CH 2‐62, J = 7.2 Hz), 4.12–3.97 (m, 6H, CH 2‐45b(II, III, IV)), 4.09 (2H, CH 2‐45b(I)), 3.74 (d, 8H, CH 2‐50b, J = 15.8 Hz), 3.57–3.44 (m, 6H, CH 2‐46a(II, III, IV)), 3.55 (2H, CH 2‐46a(I)), 3.39 (2H, CH 2‐46b(I)), 3.36–3.28 (m, 6H, CH 2‐46b(II, III, IV)), 1.73 (4H, CH 2‐63), 1.19–1.08 (m, 14H, CH 2‐64, 65, 66, 67), –2.73 (s, 2H, NH); 13C NMR (CDCl3, 126 MHz): ∂ 158.77 (ArC‐32, 38, 44), 158.72 (ArC‐26), 157.03 (C=O‐51), 147.48 (C‐57), 146.68 (ArC‐47), 135.88 (ArC‐28, 34, 40), 133.20 (ArC‐22), 131.99 (ArC‐21), 131.82 (ArC‐27, 33, 39), 129.82 (ArC‐49), 129.73 (ArC‐30, 36, 42), 128.66 (ArC‐55, 56), 128.25 (ArC‐54), 127.20 (ArC‐24), 122.54 (ArC‐23), 119.96 (ArC‐29, 35, 41), 119.04 (CH‐58), 115.47 (ArC‐48(II,III,IV)), 115.46 (ArC‐48(I)), 115.29 (ArC‐6, 11, 16) 114.54 (ArC‐1), 112.28 (ArC‐25), 112.15 (ArC‐31, 37, 43), 84.80 (C‐52), 67.62 (CH2‐46(II, III, IV)), 67.37 (CH2‐46(I)), 67.20 (CH2‐45(I)), 67.04 (CH2‐45(II, III, IV)), 50.40 (CH2‐62), 44.42 (CH2‐50), 30.34 (CH2‐63), 26.60 (CH2‐64, 65, 66, 67); 15N{1H}‐NMR† (CDCl3, 51 MHz): ∂ 343.68 (Triazole‐N‐61), 249.14 (Triazole‐N‐59); UV/Vis (CHCl3/CH3CN 1:1, v/v) λ/nm (log (ε/M–1 cm–1)) 420 (5.90), 515 (4.58), 548 (4.08), 590 (4.08), 644 (3.78); Accurate mass: m/z 2979.14800 (M+2H)+; calcd. for C183H152N20 15N2O20: 2979.14939. †50 % of the indicated N atoms are 15N‐labeled.</p><!><p>Supporting Information</p><p>Click here for additional data file.</p>

PubMed Open Access

Effect of β-Cyclodextrin on Physicochemical Properties of an Ionic Liquid Electrolyte Composed of N-Methyl-N-Propylpyrrolidinium bis(trifluoromethylsulfonyl)amide

Ionic liquids (ILs) are promising electrolyte materials for developing next-generation rechargeable batteries. In order to improve their properties, several kinds of additives have been investigated. In this study, β-cyclodextrin (β-CD) was chosen as a new additive in IL electrolytes because it can form an inclusion complex with bis(trifluoromethylsulfonyl)amide (TFSA) anions. We prepared the composites by mixing N-methyl-N-propylpyrrolidinium bis(trifluoromethylsulfonyl)amide/LiTFSA and a given amount of triacetyl-β-cyclodextrin (Acβ-CD). The thermal behaviors and electrochemical properties of the composites were analyzed by several techniques. In addition, pulse field gradient NMR measurements were conducted to determine the self-diffusion coefficients of the component ions. The addition of Acβ-CD to the IL electrolytes results in the decrease in the conductivity value and the increase in the viscosity value. In contrast, the addition of Acβ-CD to the IL electrolytes induced an improvement in the anodic stability because of the formation of an inclusion complex between the Acβ-CD and TFSA anions. CDs are potential candidates as additives in IL electrolytes for electrochemical applications.

effect_of_β-cyclodextrin_on_physicochemical_properties_of_an_ionic_liquid_electrolyte_composed_of_n-

Introduction<!><!>Materials<!>Measurements<!>Interaction Between Acβ-CD and TFSA Anion<!><!>Interaction Between Acβ-CD and TFSA Anion<!><!>Thermal Properties<!><!>Thermal Properties<!><!>Conductivity and Viscosity<!><!>Conductivity and Viscosity<!><!>Diffusion Coefficients<!><!>Diffusion Coefficients<!>Electrochemical Properties<!><!>Electrochemical Properties<!><!>Conclusions<!>Author Contributions<!>Conflict of Interest Statement

<p>Ionic liquids (ILs) have been attractive as electrolyte materials because of their unique properties such as high ionic conductivity at room temperature and wide potential window (Armand et al., 2009). In addition, as ILs have a low vapor pressure and low flammability, they will be suitable for developing safer electrolytes instead of organic solvents (Ohno, 2011). Among onium cations, pyrrolidinium-based ILs are primarily being used as electrolytes in rechargeable batteries (Ishikawa et al., 2006; Matsumoto et al., 2006; Seki et al., 2008; Yoon et al., 2015). Pyrrolidinium-based ILs are superior in thermal and electrochemical stability as compared to those of other onium-based ILs. However, it is difficult to realize target ion transport with ions such as lithium ions or sodium ions in ILs, because the component ions of ILs as solvents also migrate along the potential gradient. New designs of ILs that address such drawback has been proposed by many researchers. For example, one candidate is poly(IL)s, which fix cation or anion species on the polymer chain (Yuan et al., 2013; Nishimura and Ohno, 2014; Qian et al., 2017). Another candidate is zwitterions, which have the cation and anion in the same molecule (Yoshizawa et al., 2004; Narita et al., 2006; Yoshizawa-Fujita et al., 2011). Nevertheless, it is still difficult to achieve a high ionic conductivity over 10−2 S cm−1 at room temperature and a high lithium transference number (tLi+) over 0.5.</p><p>Rechargeable batteries, especially lithium-ion batteries (LIBs), employing ILs as the electrolyte materials have been developed (Ishikawa et al., 2006; Matsumoto et al., 2006; Seki et al., 2008; Ohno, 2011). For practical applications, a high energy density of LIBs is required. In order to improve the energy density of LIBs, the cells are needed to be operated at higher cut-off voltages. However, high cut-off voltages induce a significant decrease in the charge/discharge cycle stability of LIBs due to the decomposition of electrolytes. As a result, a passivation layer on the electrode is formed, even when ILs are used as electrolytes (Seki et al., 2008). The decomposition reaction of electrolytes should be suppressed at high cut-off voltages to allow the use of high-voltage cathode materials [e.g., LiCo1/3Ni1/3Mn1/3O2 (Yabuuchi and Ohzuku, 2003), LiNi0.5Mn1.5O4 (Zhu et al., 2014)]. Various additives have been used to improve the anodic stability of electrolyte materials (Franco, 2015).</p><p>Cyclodextrin (CD) is a circular oligosaccharide composed of α-D(+)-glucopyranose units. The CD, which possesses seven glucose units, is called β-CD. They have a three-dimensional funnel-shaped architecture with a narrower rim molded by a hydrogen-bonding network built by primary OH groups (one group per glucose unit), and with a broader rim composed of secondary OH groups (two groups per glucose unit) (Crini, 2014). The two rims of the molecules are hydrophilic, while the interior of their cavity is hydrophobic. It is known that β-CD tends to form inclusion complexes with guest molecules with suitable characteristics of polarity and dimension in aqueous solutions (Silva et al., 2008; Baâzaoui et al., 2016). CD is among the most frequently used host molecule in supramolecular chemistry; this ability has been widely used in food and pharmaceutical studies (Szejtli, 1998; Crini, 2014). It has also been widely used in lithium battery research as a surfactant to effectively disperse solid substances in liquids and as an agent to promote complexation reactions, which is beneficial to material dispersion and molding (Chen et al., 2016).</p><p>Recently, Amajjahe et al. (2008) found that the anion of 1-butyl-3-vinylimidazolium bis(trifluoromethylsulfonyl)amide exclusively formed a host–guest complex with β-CD (Amajjahe and Ritter, 2008; Amajjahe et al., 2008). He et al. (2009) investigated the interaction of hydrophobic ILs and β-CD in detail (He et al., 2009). They found that the imidazolium cation did not interact with β-CD while its long alkyl side chain did. In addition, hydrophobic anions with fluorine atoms could interact with β-CD, and the interaction between the bis(trifluoromethylsulfonyl)amide (TFSA) anion and β-CD was stronger than those of BF4 and PF6 anions. These results prompted us to investigate the effect of β-CD on the physicochemical properties of ILs, and we expected that the anion trap ability of β-CD would contribute to the enhancement of the Li-ion conductivity and the improvement of the electrochemical stability. In this study, a pyrrolidinium-based IL with a TFSA anion, N-methyl-N-propylpyrrolidinium bis(trifluoromethylsulfonyl)amide ([C3mpyr][TFSA]) (see Figure 1), was used as the electrolyte solution. Its LiTFSA composites were prepared and mixed with different amounts of β-CD, and their physicochemical and electrochemical properties were evaluated.</p><!><p>Chemical structures of [C3mpyr][TFSA], and Acβ-CD.</p><!><p>N-Methylpyrrolidine (Tokyo Chemical Industry Co., Ltd., > 98.0%), 1-chloropropane (Tokyo Chemical Industry Co., Ltd., > 99.0%), and lithium bis(trifluoromethylsulfonyl)amide (LiTFSA) (Kishida Chemical Co., Ltd., 99.0%) were purchased. N-Methylpyrrolidine and 1-chloropropane were purified by distillation in vacuo prior to use. Triacetyl-β-cyclodextrin (Acβ-CD) (Tokyo Chemical Industry Co., Ltd., > 97.0%) (see Figure 1) was used after drying.</p><p>[C3mpyr][TFSA] was prepared as follows. N-Methyl-N-propylpyrrolidinium chloride ([C3mpyr][Cl]) was synthesized according to a previously published procedure (Laus et al., 2008). [C3mpyr][Cl] and LiTFSA were separately dissolved in deionized water. LiTFSA aq. was added dropwise to [C3mpyr][Cl] aq. The resulting liquid was purified by washing repeatedly with deionized water until no residual chloride was detected with the use of AgNO3 aq. [C3mpyr][TFSA] was obtained as a colorless liquid at room temperature and characterized by 1H NMR, fast atom bombardment mass spectrometry (FAB-MS), and elemental analysis. 1H NMR (CD2Cl2, 300 MHz): δ (ppm) = 3.50 (2H, ddd, J = 10.22, 5.58, 3.01 Hz), 3.26 (1H, dt, J = 8.59, 4.04 Hz), 3.04 (1.5H, s), 2.27 (2H, s), 1.83 (1H, tt, J = 12.20, 5.61 Hz), 1.06 (1.5H, t, J = 7.39 Hz). MS (FAB+): m/z 128.2 [M], 536.4 [2M+X]+, MS (FAB−): m/z 280.0 [X], 688.0 [M+2X]−. Anal. Calcd. for C10H18F6N2O4S2 (%): C, 29.4; H, 4.44; N, 6.86; S, 15.7; Found (%): C, 29.2; H, 4.42; N, 6.78; S, 16.1.</p><p>A given amount of LiTFSA was dissolved in [C3mpyr][TFSA] [IL:LiTFSA = 18 : 1 (molar ratio)], and then a given amount of Acβ-CD was added into IL/LiTFSA composites at molar ratios LiTFSA:Acβ-CD = 1.0 : 0.5, 1.0 : 1.0, and 1.0 : 1.5. Four kinds of samples were prepared to investigate the effect of Acβ-CD on the properties of the IL electrolyte. Their composites are abbreviated as the molar ratio of Acβ-CD. For example, the abbreviation of [C3mpyr][TFSA] : LiTFSA : Acβ-CD = 18 : 1 : 1.5 is Acβ-CD1.5. These mixtures were stirred at 60°C for 24 h.</p><!><p>Fourier-transform infrared (FT-IR) measurements were performed on a Nicolet 6700 (Thermo Fisher Scientific) by using KRS-5.</p><p>1H and 19F NMR measurements were carried out with a Bruker Avance III HD 400 MHz at 25°C. Thermogravimetric analysis was conducted using a TG-DTA instrument (TG/DTA7200, Hitachi High-Technologies Corp.) under a nitrogen atmosphere at temperatures ranging from 25 to 500°C at a heating rate of 10°C min−1. The thermal behavior was examined using differential scanning calorimetry (DSC) (DSC7020, Hitachi High-Technologies Corp.) at temperatures between −150 and 100°C at a heating/cooling rate of 10°C min−1.</p><p>Impedance measurements were carried out using a VSP-300 (Bio-Logic Science Instruments) at frequencies ranging from 100 mHz to 1 MHz and temperatures ranging from 80 to −40°C. The temperature was controlled by a constant-temperature oven (SU-642, Espec Corp.). The composites were enclosed in a homemade glass cell having two platinum electrodes. The measurements were carried out by maintaining the cells at each temperature for 30 min. Viscosity measurements were carried out using a stabinger viscometer (SVM3000, Anton Paar) and temperature ranging from 80 to 20°C.</p><p>Pulse field gradient nuclear magnetic resonance (PFG-NMR) measurements were carried out with a Bruker Avance III HD 400 MHz at 80°C for 1H, 7Li, and 19F nucleus. The ILs were filled into 5-mm NMR tubes, which were sealed. The measurements were carried out in 16 gradient steps per diffusion experiment. The gradient strength was 1,700 G cm−1. The diffusion coefficients were calculated from the peak integration attenuation according to Equation 1 (Tanner and Stejskal, 1968):</p><p>where A is the signal at a certain gradient (G), A0 is the signal at a gradient of 0, δ is the width of the gradient pulse, Δ is the diffusion time, D is the diffusion coefficient, and γ is the gyromagnetic ratio of the nuclei.</p><p>Linear sweep voltammetry (LSV) measurements were carried out by using a VSP-300 (Bio-Logic Science Instruments) in the potential range of −0.2 and 6 V at 60°C at a scan rate of 1.0 mV s−1. Li foils were used as the reference and counter electrodes, while Ni and Pt plates were used as working electrodes in the potential ranges of −0.2–3.0, and 3.0–6.0 V, respectively. The electrodes were separated by a glass filter to prevent short-circuiting. The cyclic voltammetric measurements of [C3mpyr][TFSA]:LiTFSA:Acβ-CD = 18:1:1.0 composites were carried out using a VSP-300 (Bio-Logic Science Instruments) in the potential range of −0.25–1.0 V at 25, at a scan rate of 1.0 mV s−1, with Li foils as the reference and counter electrodes, and the Ni plate was used as the working electrode. The electrodes were separated by a glass filter to prevent short-circuiting.</p><!><p>A given amount of β-CD was initially added into [C3mpyr][TFSA] and its LiTFSA mixture. Unfortunately, the IL electrolytes could not dissolve β-CD at any concentration. β-CD possesses hydroxyl groups, which form hydrogen bonds. The Lewis basicity of the TFSA anion is weak, and the TFSA anion cannot break the hydrogen bond. In fact, ILs with anions such as chloride and acetate, which exhibit a stronger Lewis basicity, can dissolve cellulose (Ohno and Fukaya, 2009) as such anions interact with the hydroxyl groups of cellulose because of the strong electron-donating ability. The TFSA anion could not dissolve even oligosaccharides. Therefore, Acβ-CD was used in this study instead of β-CD. A given amount of Acβ-CD was added into the IL electrolytes. [C3mpyr][TFSA] with a weak Lewis-base anion could dissolve Acβ-CD, which has an acetyl group instead of a hydroxyl group.</p><p>FT-IR measurements were conducted, and each peak was assigned according to the literatures (Liu et al., 2009; Roy et al., 2016; Li et al., 2017; Wu et al., 2017). Figure 2 presents FT-IR spectra of [C3mpyr][TFSA]/LiTFSA and [C3mpyr][TFSA]/LiTFSA/Acβ-CD composites. The FT-IR spectrum of [C3mpyr][TFSA]/LiTFSA exhibits characteristic peaks for C-H stretching, CH2 bending, and S = O stretching bands etc. The peaks in the range from 2,978 to 2,882 cm−1 can be assigned to the C-H stretching and CH2 bending modes. In the case of TFSA anion, the peaks of S = O stretching band and C-SO2-N bond are observed at 1,349 and 1,136 cm−1, respectively. In addition, CF3 symmetric stretching modes are located in 1,195 cm−1 and 1,056 cm−1. For the spectrum of [C3mpyr][TFSA]/LiTFSA/Acβ-CD, a new peak is observed at 1,746 cm−1, which is assigned to C = O stretching mode for acetyl group, and the absorbance increases with increasing the Acβ-CD amount, indicating that the composites are formed by mixing [C3mpyr][TFSA]/LiTFSA and Acβ-CD.</p><!><p>FT-IR spectra of [C3mpyr][TFSA]/LiTFSA composites and their Acβ-CD composites.</p><!><p>19F NMR measurements were carried out to investigate the interaction between Acβ-CD and the TFSA anion. Figure 3 presents 19F NMR spectra of the CF3 group in the TFSA anion for the [C3mpyr][TFSA]/LiTFSA and [C3mpyr][TFSA]/LiTFSA/Acβ-CD composites. The chemical shifts of the CF3 group of the TFSA anion are −79.73, −79.80, −79.88, and −79.99 ppm for the Acβ-CD 1.5, 1.0, 0.5, and 0 composites, respectively. Zhang et al. (2014) performed 19F NMR measurements of 1-ethyl-3-methylimidazolium bis(trifluoromethylsulfonyl)amide to detect the host–guest interaction between the β-CD and TFSA anion (Zhang et al., 2014). As the molar ratio of CD increases, downfield shifts for the fluorine atom of the CF3 group in the TFSA anion are observed because of the formation of the complex for CD and the TFSA anion. In all the [C3mpyr][TFSA]/LiTFSA/Acβ-CD composites, the CF3 group chemically shifts to a lower magnetic field as compared to that of Acβ-CD 0, suggesting that Acβ-CD forms a complex with the TFSA anion.</p><!><p>19F NMR spectra of [C3mpyr][TFSA]/LiTFSA composites and their Acβ-CD composites.</p><!><p>To evaluate the thermal stability, onset thermal decomposition temperatures (Td) were measured. Figure 4 shows TGA traces for the [C3mpyr][TFSA]/LiTFSA and [C3mpyr][TFSA]/LiTFSA/Acβ-CD composites. Acβ-CD 0 shows a Td value at 386°C. Pyrrolidinium-based ILs with the TFSA anion are known to exhibit a higher thermal stability, and the Td value of Acβ-CD 0 is consistent with the literature value (Yang et al., 2014). All the composites with Acβ-CD show similar Td values, and their Td values are about 300°C. This is due to the decomposition of Acβ-CD. The amount of weight loss is consistent with the amount of added Acβ-CD in the IL electrolytes.</p><!><p>TGA curves of [C3mpyr][TFSA]/LiTFSA composites and their Acβ-CD composites.</p><!><p>DSC traces of the composites are presented in Figure 5. Acβ-CD 0 exhibits a melting point (Tm) of 8.0°C, which is consistent with the literature value (Wu et al., 2011). When the addition amount of Acβ-CD is 0.5 in the molar ratio, the glass transition temperature (Tg), two crystallization temperatures, and Tm are observed at −91, −33, −19, and 7.1°C, respectively. Acβ-CD 0.5 exhibits no crystallization temperature upon the cooling scan. In addition, the Tm value of Acβ-CD 0.5 slightly decreases as compared to that of Acβ-CD 0. The crystallization temperature and Tm cannot be observed in the composites in which the added amount of Acβ-CD is larger than the amount of Li salt in the molar ratio. Acβ-CD 1.0 and 1.5 exhibits Tg only and maintains low values below −79°C. These results suggest that the interaction between the Acβ-CD and TFSA anion prevents the crystallization of the IL.</p><!><p>DSC charts of [C3mpyr][TFSA]/LiTFSA composites and their Acβ-CD composites.</p><!><p>Arrhenius plots of the ionic conductivities for the composite electrolytes are presented in Figure 6. Figure 7 exhibits a typical Nyquist plot for Acβ-CD 1.5 at 0°C. The ionic conductivity values were calculated from the touchdown point on the Z′-axis which exhibits the resistance of the compound. For Acβ-CD 0 and 0.5, the conductivity values were not obtained by means of our impedance apparatus, because of the crystallization of the electrolytes below 0°C, which is consistent with the DSC results. According to the DSC results, as Acβ-CD 1.0 and 1.5 are liquid in a wide temperature range, they exhibit a higher ionic conductivity even below 0°C, as shown in Figure 5. The ionic conductivities of the Acβ-CD 1.5, 1.0, 0.5, and 0 composites are 3.2 × 10−4, 5.1 × 10−4, 1.6 × 10−3, and 2.1 × 10−3 S cm−1 at 25°C, respectively. The addition of Acβ-CD results in the decrease in the conductivity value, ascribable to the formation of an inclusion complex between Acβ-CD and the TFSA anion. This complex decreases the diffusivity of the component ions, thus decreasing the ionic conductivities. Roy and Roy (2017) used trihexyltetradecylphosphonium chloride as an IL, where a similar decrease in conductivity was observed as the amount of CD increased. The decrease in conductivity will be due to the encapsulation of guest molecules in the hydrophobic cavity of CD (Roy and Roy, 2017).</p><!><p>Arrhenius plots of ionic conductivities for [C3mpyr][TFSA]/LiTFSA composites and their Acβ-CD composites.</p><p>A typical Nyquist plot for Acβ-CD 1.5 at 0°C.</p><!><p>Figure 8 shows the ionic conductivity at 25°C and viscosity at 30°C as a function of the molar ratio of Acβ-CD. The ionic conductivity monotonously decreases with the Acβ-CD content as mentioned above. The viscosity values of the Acβ-CD 1.5, 1.0, 0.5, and 0 composites are 19,000, 1,700, 250, and 60 mPa s at 30°C, respectively. The viscosity values increase steeply as Acβ-CD is added to the composites. In addition, the conductivities and viscosities are inversely proportional (Salminen et al., 2007). The viscosity increases with CD concentration probably because of the IL and CD interactions and solvation (Roy et al., 2016). Thus, it is considered that the ionic conductivities decrease because of the increase in the viscosities of the composites.</p><!><p>Correlation between ionic conductivity and viscosity for [C3mpyr][TFSA]/LiTFSA composites and their Acβ-CD composites.</p><!><p>The self-diffusion coefficients of C3mpyr+ (DH), Li+ (DLi), and TFSA− (DTFSA) for the [C3mpyr][TFSA]/LiTFSA and [C3mpyr][TFSA]/LiTFSA/Acβ-CD composites were determined by means of PFG-NMR at 80°C, as shown in Figure 9. The DH and DTFSA values of these composites are almost the same at 80°C, while the DLi value is lower than those of the DH and DTFSA values. The increase in the Acβ-CD content induces a large difference between the DLi value and other values. The apparent lithium transfer number (tLi+) was calculated from the diffusion coefficient values using Equation (2):</p><!><p>Diffusion coefficients of [C3mpyr][TFSA]/LiTFSA composites and their Acβ-CD composites as a function of mole ratio of Acβ-CD.</p><!><p>Unlike electrochemical techniques, the diffusion coefficients obtained using the PFG-NMR mwcies but also from non-ionic species (Horiuchi et al., 2017). The tLi+ values of the Acβ-CD 1.5, 1.0, 0.5, and 0 composites were 0.06, 0.08, 0.12, and 0.22, respectively. The decrease in the tLi+ values with the increase in Acβ-CD content suggests that an inclusion complex will be formed between LiTFSA and Acβ-CD. In addition, Acβ-CD would form an inclusion complex with not only LiTFSA but also the aggregation including a Li cation, similar to the combination of two TFSA anions and one Li cation because the large surface charge density of a Li cation induces the formation of cluster ions (Appetecchi et al., 2016).</p><!><p>The electrochemical stabilities of the [C3mpyr][TFSA]/LiTFSA and [C3mpyr][TFSA]/LiTFSA/Acβ-CD composites were investigated by LSV on a Ni electrode (from 3 to −0.2 V) and Pt electrode (from 3 to 6 V) at 60°C. The LSV results are presented in Figure 10. The electrochemical window (EW = Eanodic-Ecathodic) of all the IL electrolytes was determined from the values for the cathodic (Ecathodic) limit at −0.1 mA cm−2 and anodic (Eanodic) limit at 0.1 mA cm−2. The EW value of Acβ-CD 0 is 4.6 V vs. Li/Li+ and that of Acβ-CD 0.5 is 4.6 V vs. Li/Li+, which is almost the same as that of Acβ-CD 0. The EW values are about 5.5 V vs. Li/Li+ for both Acβ-CD 1.0 and 1.5 composites. As the addition amount of Acβ-CD increases, the oxidation stability improves. This improvement should be based on the formation of an inclusion complex between Acβ-CD and the TFSA anion because the anodic stability significantly improves, and the cathodic stability is almost the same regardless of the addition of Acβ-CD.</p><!><p>Linear sweep voltammograms of [C3mpyr][TFSA]/LiTFSA composites and their Acβ-CD composites.</p><!><p>The reversible oxidation and reduction reactions of lithium were examined at room temperature for Acβ-CD 1.0. Figure 11 shows the cyclic voltammogram for Acβ-CD 1.0 on a Ni electrode. Acβ-CD 1.0 exhibits reduction and oxidation peaks for Li at about −0.1 and 0.1 V vs. Li/Li+, respectively. The current density decreases with the cycling number from 1st to 10th. After that, the current density maintains a constant value, and stable reversible redox reactions are observed during 20 cycles. At the initial anodic sweep, an anodic current is observed. This behavior is also observed for pyrrolidinium-based ILs (Towada et al., 2015; Horiuchi et al., 2016). In addition, the maximum current density of the anodic peak slightly shifts to a higher potential value with the increase in cycle number. These results suggest that a solid electrolyte interphase film is formed on the Ni electrode surface (Grande et al., 2015), even in the presence of Acβ-CD.</p><!><p>Cyclic voltammograms of Acβ-CD 1.0 at 25°C.</p><!><p>The effect of Acβ-CD on the properties of [C3mpyr][TFSA]/LiTFSA was investigated by means of several techniques. The chemical shift of the CF3 group of the TFSA anion shifted to a lower magnetic field with the increase in the Acβ-CD content. With the addition of Acβ-CD to the IL electrolyte, the Tm of the IL disappeared and the viscosity increased. These results suggest that an inclusion complex is formed between Acβ-CD and the TFSA anion. In contrast, the tLi+ and DLi values decreased with the increase in the Acβ-CD content in the composites. The anodic stability of [C3mpyr][TFSA]/LiTFSA was significantly improved after adding a certain amount of Acβ-CD. Li plating and stripping in the [C3mpyr][TFSA]/LiTFSA/Acβ-CD composite were repeatedly observed. According to these results, Acβ-CD will be an interesting additive for improving the electrochemical stability of ILs. It is known that there are three kinds of CD, α-CD, β-CD, and γ-CD, which have different cavity sizes. The physicochemical properties of various ILs with different anions could be controlled by choosing suitable CD derivatives.</p><!><p>MS and MY-F designed the research. MS prepared the samples and measured the properties. MS and NK carried out the NMR measurements and the data collection. YT, MR, and MY-F participated in the data analysis. MS and MY-F wrote 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

A Phase 1 dose escalation study of NEO-102 in patients with refractory colon and pancreatic cancer

Purpose NEO-102 is a novel chimeric IgG1 monoclonal antibody which recognizes a variant form of MUC5AC expressed specifically by human pancreatic and colorectal tumors. Preclinical models have demonstrated encouraging signs of antitumor activity through antibody-dependent cell-mediated cytotoxicity. Methods This is a Phase I, dose escalation trial of NEO-102 (Ensituximab) for patients with refractory pancreatic and colorectal cancer. The primary objective was to determine safety and tolerability of escalating doses of NEO-102. Secondary objectives were to assess pharmacokinetics, clinical benefit and biologic correlates. Patients whose tumors express NPC-1 antigen were eligible. Dose escalation was performed in a 3+3 design at doses of 1.5mg/kg, 2mg/kg, 3mg/kg and 4mg/kg. Results A total of 19 patients (4 pancreatic and 15 colon cancer) were enrolled at participating institutions in the treatment phase. Most common treatment-related adverse events (AEs) included anemia, fatigue, fevers, chills and flushing. Grade 3/4 AEs were hyperbiliruibinemia and anemia. There was no detectable hemolysis. Of twelve patients evaluable for disease response, the response rate at week 8 included 4 patients with stable disease and 8 patients with progressive disease (PD). The maximum tolerated dose was 3mg/kg. Of 74 patients who underwent tissue screening, positive NPC-1 expression was 47% in colon and 59% in pancreatic cancer Conclusions Treatment with the NEO-102, in this first-in-human study, is well tolerated with a manageable safety profile. A maximum tolerated dose of 3 mg/kg has been established. Toxicity profile is typical for this therapeutic class and allows for combination with conventional cytotoxic therapies.

a_phase_1_dose_escalation_study_of_neo-102_in_patients_with_refractory_colon_and_pancreatic_cancer

Introduction<!>Study design and patient selection<!>Patients were screened for inclusion in two phases<!>NEO-102 Immunohistochemistry<!>Treatment<!>Statistical Analysis<!>Results<!>Safety<!>Efficacy/ Response<!>Pharmacokinetics<!>IHC assay<!>Discussion<!>

<p>Cancer of the colon and pancreas represent two of the top four causes of cancer deaths among men and women in the United States[1]. NEO-102 (NPC-1C, Ensituximab; Precision Biologics, Inc.) is a novel chimeric IgG1 monoclonal antibody developed as a biological treatment for patients with pancreatic and colorectal cancers.</p><p>NEO-102 was one of several antibodies raised against an allogeneic colorectal cancer vaccine that had previously been tested in human clinical trials in the United States and not derived from MUC5AC antigen [2]. This original vaccine was screened from 79 patients with various stages of colon cancer whose tumor membranes fractions were pooled, separated by HPLC, and tested for delayed type hypersensitivity. The component that produced a strong DTH response upon screening was selected as the original vaccine and used in clinical trials. Results of that preliminary study with the vaccine in patients with refractory colorectal cancer revealed clinical benefits that correlated directly with patients developing IgG responses against the vaccine [2]. This formed the rationale for the vaccine to screen antibodies for sensitivity, specificity and anti-tumor function against colon cancer. NEO-102 was the first of 3 antibodies identified that met these criteria. Although unknown at the time of drug discovery, through protein purification and mass spectroscopy the homologous sequence to MUC5AC (NPC-1) was identified as the target for this compound. Preclinical studies demonstrated that NEO-102 antibody binds specifically to a novel MUC5AC-related antigen and directs antibody-dependent cellular cytotoxicity (ADCC) in the presence of normal human peripheral blood mononuclear cells (PBMCs)[3]. Additionally, antitumor efficacy was noted in preclinical pancreatic and colon cancer tumor xenograft models [3].</p><p>MUC5AC is a member of the mucin gene family and involves a carbohydrate structure on the amino acid backbone of a large (~100kDa) heavily glycosylated protein. The MUC5AC gene has been reported to be expressed mainly on the surface epithelium of normal gastric mucosa and normal airway epithelium [4]. MUC5AC is found in the GI tract and is preferentially expressed on colon and pancreatic cancer cells [5]. MUC5AC is predominantly expressed in the respiratory tract, in inflammatory conditions such as cystic fibrosis and COPD, producing increase levels of mucous production. Whereas MUC5AC expression has been demonstrated in fetal and pre-cancerous colonic mucosa, it is absent in normal adult colon [6,7]. Unlike the inflammatory conditions where MUC5AC is heavily glycosylated, in pancreas and colon tumors, MUC5AC is aberrantly glycosylated [8–11]. NEO-102 antibody can therefore discriminate between the native MUC5AC and the aberrantly glycosylate, NPC-1, variant of MUC5AC in tumors imparting tumor selectivity, which is exploited in this therapeutic strategy. As a companion diagnostic tool, an immunohistochemistry (IHC) based assay has been developed in parallel. This work provided the foundation for exploring NEO-102 as a therapeutic strategy for the management of pancreatic and colon cancer.</p><!><p>This is a Phase 1 open label, multi-institution, dose escalation clinical trial of the therapeutic monoclonal antibody, NEO-102. Eligible patients had histologically confirmed colorectal cancer that had progressed on at least two lines of systemic therapy or advanced adenocarcinoma of the pancreas that had progressed on at least one line of systemic therapy. Patients were preselected based upon IHC testing for NPC-1 antigen expression performed on archival formalin fixed paraffin embedded tissue (FFPE). A minimum of 20% of tumor tissue staining positive at ≥ 2 + intensity was required for eligibility. Patients were required to have good performance status (ECOG performance status ≤ 2), evidence of measureable disease per Response Evaluation Criteria in Solid Tumor (RECIST criteria v1.1 [12]), adequate hematologic (hemoglobin > 8.5g/dL, absolute neutrophil count ≥ 1500/mm3 and platelets ≥ 50,000/mm3), hepatic (total bilirubin <2.0, alanine transaminase and aspartate transaminase less than 3 times the upper limit of normal or 5 times the upper limit of normal in presence of liver metastasis) and renal (serum creatinine ≤ 1.5mg/dL, creatinine clearance of > 40mL/min/1.73 m2) function. Exclusion criteria included disseminated or uncontrolled brain metastases, ascites with clinically identifiable abdominal distention, major surgery within 4 weeks of enrollment, concomitant uncontrolled illness, concurrent antineoplastic systemic therapy, uncontrolled diabetes, history of grade 2 or above allergic reaction to cetuximab, prior hemolytic anemia, concurrent warfarin use, and anticipated life expectancy of less than 8 weeks.</p><!><p>IHC screening and treatment screening. Informed consents for both screening phases were performed separately. Patients with positive expression of NPC-1PC-1 target antigen by IHC were eligible to initiate the treatment screening phase.</p><p>The primary objective of this study was to determine the safety and tolerability of escalating doses of NEO-102 monoclonal antibody. Secondary objectives were determination of pharmacokinetics at each dose level, as well as clinical benefit as measured by overall survival (OS) and RECIST criteria v1.1[12]. Patients were enrolled at three participating institutions with approval from the ethics committees at respective institutions and regulatory authorities. The study followed the Declaration of Helsinki and the International Conference on Harmonization Good Clinical Practice guidelines. The study was supported by Precision Biologics, Inc., and registered at Clinicaltrials.gov (NCT01040000). All patients signed a written informed consent prior to starting study specific procedures.</p><!><p>Formalin-fixed paraffin embedded sections at 5 µM were obtained, placed on glass slides and stained with hematoxylin/eosin using an automated H/E stainer. IHC for NEO-102 was performed on formalin fixed paraffin embedded sections at 4 µM placed on positively charged slides. Following deparaffinization the antigen retrieval was performed at 115°C in a decloaking chamber. The endogenous peroxidase was blocked by incubating with 3% H2O2 for 10 mins. The slides were then loaded on to a (DAKO) Autostainer followed by endogenous biotin blocking. Following a brief protein blocking step, the sections were incubated for 60 minutes at room temperature with the NEO-102 antibody at a 1:200 dilution (Precision Biologics). Detection was performed using a commercially available Streptavidin-HRP antibody conjugate by incubating for 30 minutes.</p><!><p>NEO-102 was administered intravenously (IV) every 14 days. Three to six patients were treated at each of the following dose levels: 1.5 mg/kg, 2 mg/kg, 3 mg/kg and 4 mg/kg. NEO-102 was initially started at a rate of 0.5 mg/min and the rate increased as tolerated in 0.5 mg/min increments every 30 minutes to a maximum rate of 4 mg/min. Patients were monitored as inpatient for 24 hours after the first infusion and remaining infusions were administered as an outpatient. Premedication with dexamethasone (10 mg IV), ranitidine (50 mg IV) and diphenhydramine (25–50 mg IV) was administered prior to each dose. Additional treatment could be offered in the absence of dose limiting or unacceptable toxicity, progressive disease, or per investigator discretion.</p><p>Adverse events were graded for severity using the National Cancer Institute Common Terminology Criteria for Adverse Events (CTCAE version 4.0). Dose limiting toxicity (DLT) was defined as any one of the following in the first 30 days: Any grade 3 or 4 hematologic/non-hematologic toxicity or severe infusion related reaction/allergic reaction/hypersensitivity to NEO-102. Transient toxicity related to infusion including fatigue, infusion reactions, flu-like symptoms, fever, and headache that recover to grade 1 or less within 8 hours after standard supportive treatment was not considered a DLT. A standard three-plus-three dose escalation design was followed for dose escalation. [13] The starting dose of 1.5 mg/kg for this trial was based on clinical experience with an earlier version of ensituximab (NEO-101).[14] Hemoglobin levels were checked 24–48 hours after completion of infusion to assess for hemolysis. Patients who experienced greater than 1 gm/dL drop in hemoglobin after dosing with NEO-102 underwent additional testing to rule out possible hemolysis including the following tests: direct coombs, haptoglobin, fibrinogen, D-Dimer, thrombin time or peripheral blood smear review, as indicated.</p><p>Tumor assessments were performed by conventional CT scans at baseline and then every 8 weeks. Response assessment was performed based on RECIST criteria v1.1[12]. Blood for pharmacokinetic (PK) analysis was collected at the following time points: prior to the start of infusion, at the end of infusion (EOI), and 1, 4, 24, 72 and 168 hours after the end of the first infusion. Pre-treatment and EOI blood samples were also were drawn for doses 2–4. Where possible, a sample was also collected 14 days after the fourth dose. PK analysis was performed using quantitative self-sandwich ELISA assay for NEO-102 PK developed by using anti-NPC-1C idiotype antibody (4B6). Individual concentration-time profiles were constructed for each patient for the first cycle. Peak and trough concentrations for each dose were reported as the concentration of NEO-102 within three minutes after the end of infusion and the NEO-102 concentration immediately prior to the next treatment (approximately 14 days later). Additional samples were drawn on day 1, 4, 15 and 57 for human antichimeric antibody (HACA) and cytokines. These samples were batched and testing was performed at BioReliance Corporation (Rockville, MD, USA) using a commercially available multiplex 96-well enzyme-linked immunoabsorbent assay kit (MS6000 Human Pro-Inflammatory 9-Plex Ultra-Sensitive Kit K11007; Meso Scale Diagnostics, Gaithersburg, MD, USA) on a Sector Imager 6000 according to the manufacturer's recommendation (Meso Scale Diagnostics). Pro-inflammatory cytokines (IL1β, IL-2, IL-6, IL-8, IL-10, IL-12p70, TNF-α, INF-γ, GM-CSF) were evaluated at baseline (BL), day 4, day 15 (prior to dose 2), and day 57, when available. Serum samples were stored at −70°C. The standards and serum samples were run in triplicate with 4 time points of each subject in the same plate to avoid plate variation.</p><!><p>Demographic and baseline disease characteristics for study patients were summarized by means, standard deviations (SD), medians, and ranges for continuous variables and counts, and proportions for categorical variables. Unless otherwise noted, all statistical testing was 2-sided and was carried out at the 0.05 significance level. All analyses and tabulations were carried out by SAS (version 8.2 or higher; SAS Institute) on a PC platform.</p><!><p>A total of 19 patients received at least one dose of NEO-102. Fifteen had colorectal cancer and 4 had pancreatic cancer. Median age was 58 years (32–69 years), 58% were male, and 68% had an ECOG performance status of 1. A total of 47% of patients had received four or more prior lines of systemic therapy.</p><!><p>All patients who received at least one dose of NEO-102 and were eligible for safety assessment. Adverse events (AEs) attributed to being related to NEO-102 are summarized in table 2. Most common AEs were anemia (32%), fatigue (32%), fever (21%), chills (16%), flushing (16%), increased bilirubin (16%), congestion (16%). Of these, the following were grade 3/4 AEs: anemia in 4 patients (21%), increased bilirubin in 2 patients (11%) and transient hypoxia (grade 3) in one patient (5%). Hypoxia was accompanied by confusion, and radiologic evidence of new groundglass opacities. Pathology after bronchoscopy revealed diffuse alveolar damage, many neutrophils and negative for malignant cells and hypoxia resolved on oral prednisone and 1 L oxygen by nasal cannula.</p><p>Dose titration to 4 mg/kg was completed with no observed dose limiting toxicity in the first 3 patients and preplanned expansion cohort at 4 mg/kg began. The sixth and seventh patients at 4 mg/kg experienced grade 3 anemia and transient asymptomatic grade 3 hyperbilirubinemia, respectively as mentioned above. Additionally, the fifth patient at this dose experienced a transient asymptomatic grade 3 hyperbilirubinemia after the third dose of NEO-102. Although these events occurred after completion of dose escalation phase of the clinical trial, the study cohort review committee decided to de-escalate to 3 mg/kg to ensure safety of subsequent patients. Three additional patients were then treated at the 3 mg/kg dose level with 1 out of 6 patients experiencing grade 3 reversible hypoxia (deemed to be a DLT). Therefore 3 mg/kg was established as the maximum tolerated dose (MTD).</p><!><p>Patients who received at least two doses of NEO-102 were considered evaluable for response. Sixteen patients were eligible for tumor assessment measurements. There were 5 patients with stable disease and 6 patients with PD at week 8. Five patients were removed from treatment prior to week 8 evaluation (2 for symptoms of clinical progression, 2 for treatment-related toxicity, 1 for unrelated toxicity). As of September 1, 2015 10 patients died secondary to progressive disease (7 colorectal). Patients with colorectal cancer had an overall median survival of 51 weeks (12.0 months) (range 6–108+ weeks); patients with pancreatic cancer had an overall median survival of 20 weeks (5 months) (range 10–33 weeks) in this phase I study.</p><!><p>A total of 14 patients were evaluable for pharmacokinetics. First doses ranged from 104 to 531 mg (1.5 to 4.0 mg/kg), administered over 2.0–6.33 h. Mean dose and duration at the MTD of 3 mg/kg was 240.8 mg infused over 4.17 hours. Mean concentration-time profile by dose level following the first dose of NEO-102 for the four dose levels assessed are shown in Figure 2. Maximal serum concentrations of NEO-102 were typically observed at the end of infusion, with a rapid decline in concentrations over the first 24–48 hours, followed by a slower terminal elimination phase. Pharmacokinetic parameters for NEO-102 are shown in Table 3. Drug remained detectable in all patients fourteen days after the first drug administration, resulting in some drug accumulation. At the MTD of 3 mg/kg, serum concentration immediately prior to the second dose ranged from 1.79 to 15.62 ug/mL (n = 4). The mean accumulation ratio observed for each dose level ranged from 1.08 to 1.51 (not shown). The pharmacokinetics of NEO-102 appear to be non-linear with a less than dose proportional increase in exposure with increasing doses.</p><p>There were no elevations from baseline of serum IL-1β, IL-2, IL-10, IL-12p70, GMCSF, IFNγ, and TNFα in 13 subjects at D4, D15 and D57 samples. Increased serum levels of IL-6 were observed at D57 in two subjects, D4 in one subject, and at D15 and D57 in one subject. Three-fold higher serum concentrations of IL-8 were observed at D57 compared to baseline in two subjects. There was no evidence of clinical cytokine storm phenomena in any patients treated. HACA concentrations were less than the assay lower limit of detection (LOD) of 3.9 ng/mL in 12 patients evaluated.</p><!><p>Up to four unstained formalin-fixed paraffin embedded (FFPE) tumor tissue slides from 74 potential subjects were tested using biotinylated NEO-102 antibody and a streptavidin-HRP detection system to determine IHC of the NPC-1 antigen. Figure 1 depicts tumor immunostains with NEO-102 on patient samples from the clinical trial demonstrating membrane and luminal signal. Rate of positive NPC-1 IHC (n=74) screening was 47% for colon cancer (21 of 45 tested) and 59% for pancreatic cancer (17 of 29 tested).</p><!><p>In this phase 1 study of patients preselected for target antigen expression, an aberrantly glycosylated MUC5AC-related antigen, treatment with NEO-102 is well tolerated with an encouraging safety profile. A maximum tolerated dose of 3 mg/kg has been established. Commonly experienced adverse events were mild and well-tolerated. For patients treated at the 1.5 mg/kg, 2 mg/kg and 3 mg/kg dose levels, grade 3 or 4 adverse events were anemia and one case of hypoxia. Other adverse events including fatigue, diarrhea, nausea, mucositis, weight loss and abdominal pain were mild and partly reflect characteristics of the tumors being evaluated. NEO-102 infusion reactions with standard premedications were rare (1 out 19 patients).</p><p>Two cases of grade 3 hyperbilirubinemia in patients with concomitant liver metastasis were noted at the 4mg/kg dose level. In one case, bilirubin elevation was deemed to be from biliary obstruction which improved with biliary stenting. The other patient following the 2nd dose of NEO-102 had a total bilirubin increase from 1.4 mg/dL to 4.0 mg/dL occur 24hrs after treatment, which resolved to grade 1. A 3rd dose was administered at a lower dose, but the patient developed grade 3 hyperbilirubinemia and was removed from study according to per-protocol criteria. Notably, no events of elevated bilirubin were seen at lower dose levels including the MTD of 3 mg/kg. All patients that developed anemia tested negative for hemolysis. One patient with grade 3 anemia was determined to have anemia of chronic disease combined with possible myelosuppression from multiple prior cytotoxic therapy. One case of hypoxia occurred in a patient with colon cancer with extensive lung metastases at the 2mg/kg dose level. The patient was admitted 5 days following the 1st dose of NEO-102 with hypoxia and shortness of breath and condition improved following medical management. Patient proceeded to receive a second dose of NEO-102 but was subequently taken off treatment due to declining functional status. Due to the temporal relationship with drug infusion and increasing recognition of pneumonitis related to immunological agents, hypoxia was deemed to be possibly drug related.</p><p>Overall NEO-102 demonstrated a favorable toxicity profile at the 3 mg/kg dose level. This has allowed for combination of NEO-102 with other chemotherapeutic agents in ongoing and planned trials.</p><p>The NPC-1C antibody exhibits cell-specific binding and ADCC activity against human colorectal and pancreatic tumor cells, but not against control tumor cell lines which do not express this variant of MUC5AC [15]. In vivo, the anti-tumor efficacy of NPC-1C was tested using pre-established subcutaneous tumor xenograft models [3]. Data showed significant and reproducible anti-tumor activity which provided the foundation for human studies. No partial responses were seen in the patients treated in this phase 1 study, which enrolled a heavily pretreated group of patients. This likely reflects the patient population being treated as the currently FDA approved agents for refractory colon cancer (regorafenib and TAS-102) have also show very low overall response rate[16] [17]. For agents that impart their effect through immunomodulation, survival end points are largely considered more reliable as opposed to tumor shrinkage/response rates. Although the overall survival of patients treated on this study is encouraging, no definitive conclusions can be drawn due to the small sample size. Positive selection of patients who potentially had indolent disease to begin with and an effect of additional lines of therapy received by these patients may be favorably skew results. Details on additional lines of therapy are not available for assessment. An immunohistochemistry (IHC) based companion diagnostic assay has been developed as an eligibility selection criteria to ensure that patients' tumors express the NPC-1 target, which correlated preclinically to anti-tumor responses. In this assay, the NEO-102 antibody is biotinylated and tested for the ability to detect the MUC5AC Tumor Associated Antigen (TAA) expressed in normal and malignant human tissues. The rate of expression of target antigen in this clinical trial is 47% for colon cancer and 59% of pancreatic cancers.</p><p>Following infusion of NEO-102, maximal serum concentrations are observed and distribution appears to be rapid. Cytokine evaluation demonstrated elevations in IL-6 and IL-8 post NEO-102 infusion. In contrast, no elevations in the other proinflammatory serum cytokines IL-1β, IL-2, IL-10, IL-12p70, GMCSF, IFN-γ, and TNFα were observed. This correlated with the antibody being well tolerated post infusion without clinical evidence of cytokine release syndrome.</p><p>In summary, NEO-102 is well tolerated with a predictable pharmacokinetic profile. Current treatment strategies for colon and pancreas cancer lack predictive biomarkers and clinical toxicities limit the prospect of potential combination strategies [18–20]. The favorable toxicity profile of NEO102 as observed in this study, has allowed the exploration of the role of NEO-102 for the treatment of NPC-1 positive colon and pancreatic cancer as monotherapy, as well as in combination with cytotoxic chemotherapy. A maximum tolerated dose of 3 mg/kg has been established as the recommended phase 2 dose.</p><!><p>Prior Presentations: 'A Phase I/IIA multicenter clinical trial of the chimeric monoclonal antibody NEO102 (NPC-1C) in adults with refractory pancreatic and colorectal cancer' AACR 2014 Annual Meeting</p><p>Disclaimers: None</p><p>Conflicts of Interest:</p><p>NA: The authors declare that they have no conflict of interest.</p><p>JT: The authors declare that they have no conflict of interest.</p><p>MM: The author declares that they have no conflict of interest.</p>

PubMed Author Manuscript

Five vs Six Membered-Ring PAH Products from Reaction of o-Methylphenyl Radical and two C 3 H 4 Isomers

Gas-phase reactions of the o-methylphenyl (o-CH3C6H4) radical with the C3H4 isomers allene (H2C=C=CH2) and propyne (HCºC-CH3) are studied at 600 K and 4 Torr (533 Pa) using VUV synchrotron photoionisation mass spectrometry, quantum chemical calculations and RRKM modelling. Two major dissociation product ions arise following C3H4 addition: m/z 116 (CH3 loss) and 130 (H loss). These products correspond to small polycyclic aromatic hydrocarbons (PAHs).The m/z 116 signal for both reactions is conclusively assigned to indene (C9H8) and is the dominant product for the propyne reaction. Signal at m/z 130 for the propyne case is attributed to isomers of bicyclic methylindene (C10H10) + H, which contains a newly-formed methylated five-membered ring. The m/z 130 signal for allene, however, is dominated by the 1,2-dihydronaphthalene isomer arising from a newly created six-membered ring. Our results show that new ring formation from C3H4 addition to the methylphenyl radical requires an ortho-CH3 group -similar to omethylphenyl radical oxidation. These reactions characteristically lead to bicyclic aromatic products, but the structure of the C3H4 co-reactant dictates the structure of the PAH product, with allene preferentially leading to the formation of two six-membered ring bicyclics and propyne resulting in the formation of six and five-membered bicyclic structures.

five_vs_six_membered-ring_pah_products_from_reaction_of_o-methylphenyl_radical_and_two_c_3_h_4_isome

Introduction<!>Multiplexed Photoionisation Mass Spectrometry<!>Computational Chemistry<!>Reaction Products<!>Reaction Mechanisms<!>Figure 4. Potential energy schematic for the o-CH3C6H4 + propyne reaction starting at P0. Adduct formation and isomerization between adduct P1 and adduct P2 is included, and the remaining scheme follows the lowest addition barrier pathway via adduct P1. P denotes stationary points unique to o-CH3C6H4 + propyne and C stationary points common to both C3H4 reactions. G3X-K 0 K enthalpies reported in kcal mol −1 relative to o-CH3C6H4 + allene (A0).<!>Figure 4. Potential energy schematic for the o-CH3C6H4 + allene reaction starting from A0. Adduct formation and isomerization between adduct A1 and adduct A2 is included, and the scheme follows the lowest addition barrier pathway via A1. A denotes stationary points unique to the o-CH3C6H4 + allene reaction and C stationary points common to both C3H4 reactions. G3X-K 0 K enthalpies reported in kcal mol −1 relative to o-CH3C6H4 + allene (A0<!>RRKM Analysis<!>Supporting Information<!>Author contributions<!>Conflicts of interest

<p>Identifying the key chemical pathways responsible for the formation of fused-ring structures and polycyclic aromatic hydrocarbons (PAHs) is required to accurately model soot formation, [1][2][3][4] molecular weight growth in the interstellar medium [5][6] and planetary atmospheres. 7 For PAH formation, the progression from the first aromatic ring to bicyclic PAHs is an important step as this same ring-expansion process can be replicated to form increasingly larger PAHs. [8][9] The production of bicyclic aromatics with fused "six-six" membered rings (e.g., naphthalene) versus "six-five" membered rings (e.g., indene) influences soot formation by changing the structural characteristics of the subsequent PAH products. 1, [10][11][12] Specifically, these six-five combinations introduce curvature into the PAH, whereas purely six-membered ring structures result in planar graphitic sheets. For phenyl radicals (C6H5), the textbook PAH formation process is the hydrogenabstraction/C2H2-addition (HACA) framework, which leads primarily to six-membered ring formation, [13][14][15] but this is just one contributing mechanism and it is recognized that further insights are required to better model the soot formation from the combustion of aromatic fuels -particularly involving branched aromatics (e.g., toluene, xylenes, trimethylbenzenes).</p><p>Toluene and other methylated benzenes are major components of typical gasoline fuel blends, often included in high concentrations (>30% v/v) due to their high energy density and anti-knock rating. [16][17] Upon thermal activation, toluene will usually decompose by H atom loss, producing the resonance-stabilized benzyl radical (C6H5CH2). [17][18][19][20] At high temperature conditions (>1200 K) other reactive intermediates, including the higher energy o-methylphenyl (o-CH3C6H4) radical isomer, are also formed. [21][22][23] Although no direct detection of a methylphenyl radical has been reported from such environments, existing chemical models of combustion incorporate the reactions of both benzyl and methylphenyl radicals. [24][25][26][27][28] Additionally, through H atom or CH3 loss, common during combustion, methylated and other functionalized PAHs could produce larger PAH analogues of the methylphenyl radical. [29][30] Once generated these radical species can undergo subsequent reactions with other hydrocarbons leading to unique product channels. Chemical models of soot formation still struggle with aromatic fuels 31 although there is significant ongoing progress in the area. [26][27]31 Previous studies on the ortho-methylphenyl radical (o-CH3C6H4) + O2 reaction demonstrated that, following O2 addition, the o-methylphenyl peroxyl (o-CH3C6H4-OO) structure provides an accessible and labile H-atom for intramolecular abstraction by the neighboring peroxyl group, ultimately leading to the characteristic products: o-quinone methide (o-CH2=C6H4=O) + OH. [32][33][34] Similar pathways are inaccessible for the m-methylphenyl (m-CH3C6H4) and p-methylphenyl (p-CH3C6H4) isomers. Cross-molecular beam single-collision experiments of m-and p-CH3C6H4 radical reactions with C3H4 and C4H4 show that the methyl substituents are essentially spectators to the unimolecular rearrangements after adduct formation, [35][36][37][38] but no study has investigated the effect of an o-methyl group on these or other hydrocarbon growth reactions.</p><p>To explore the possible ring-growth pathways of o-CH3C6H4 with small hydrocarbons, we have observed its reactions with two C3H4 isomers: propyne (HCºC-CH3) and allene (H2C=C=CH2) using multiplexed photoionisation (PI) mass spectrometry at 600 K and 4 Torr. This synchrotronbased technique combines a flow reactor, time-of-flight mass spectrometry, and VUV photoionisation to enable the detection of reaction products with kinetic and isomeric resolution.</p><p>We show that the addition reactions of o-CH3C6H4 with allene and propyne result in both H atom and CH3 loss product pathways. Moreover, the major products of these reactions are seen to change depending on the C3H4 isomer, favoring the formation of a new six-membered ring for the allene case and a new five-membered ring for propyne. Reported reaction pathways are developed with quantum chemical calculations and Rice-Ramsperger-Kassel-Marcus master equation (RRKM-ME) rate theory, supporting our experimental product assignments and confirming the active role of the ortho methyl substituent.</p><!><p>Gas-phase experiments were performed using multiplexed photoionisation mass spectrometry [39][40] with VUV synchrotron radiation at the Chemical Dynamics Beamline, [38][39][40] Advanced Light Source (ALS), Lawrence Berkeley National Laboratory, USA. The multiplexed photoionisation mass spectrometer comprises a quartz flow tube reactor, differentially pumped vacuum chamber, photoionisation source (i.e., synchrotron radiation), and an orthogonal time-of-flight mass spectrometer as described in more detail elsewhere. [39][40] A 650 μm diameter hole, positioned 37 cm along the tube, allows gas to continuously effuse from the quartz slow-flow reactor into a differentially pumped vacuum chamber. This gas is sampled by a skimmer to create a molecular beam that is intersected orthogonally by quasi-continuous vacuum-ultraviolet (VUV) synchrotron radiation and ions generated here are sampled by a 50 kHz pulsed orthogonal-extraction time-offlight mass spectrometer. In the experiments described here, o-iodotoluene (o-CH3C6H4-I), diluted in He, is photolyzed in the quartz flow tube at 248 nm by a pulsed KrF excimer laser (4 Hz, ca. 50 mJ cm -2 ), generating o-CH3C6H4 radicals from carbon-iodine bond homolysis thus initiating the gas-phase radical chemistry.</p><p>All reactions considered here were conducted with the reactor maintained at 600 K and 4 Torr (533 Pa). The gas flow within the heated reactor was such that a reaction time of 0-50 ms could be monitored for a gas flow velocity of 4 m s -1 . The o-iodotoluene, C3H4 gas (either allene or propyne), and He gas was supplied to the reactor through separate mass-flow controllers resulting in a combined flow rate of 50 sccm. The o-iodotoluene sample was entrained in He gas using a fritted bubbler with the liquid sample maintained at 19 °C (292 K) and under a pressure of 400 Torr (53 kPa). The resulting vapor pressure of o-iodotoluene was calculated to be 295 mTorr (39.3 Pa) at 292 K using Antoine parameters derived from Stull. 37 The number densities within the reactor at 600 K and 4 Torr were 6.2 × 10 12 molecule cm -3 for o-iodotoluene, 2.1 × 10 16 molecule cm -3 of either allene or propyne and 4.3 × 10 16 molecule cm -3 for the He buffer gas. The total gas flow number density was 6.4 × 10 16 molecule cm -3 at 600 K and 4 Torr.</p><p>All photoionisation data are normalized to the ALS photocurrent measured with a NIST-calibrated photodiode (SXUV-100, International Radiation Detectors Inc.). Ion signal acquired 20 ms before laser photolysis is averaged and subtracted from the post-laser signal so that positive ion signal is the result of ion signal created as a result of the laser photolysis. The PI spectra presented herein are the average of three individual acquisitions normalized over the integral of the spectrum. Mass spectra and kinetic traces represent the co-addition of at least three separate acquisitions of at least 900 laser pulses each.</p><p>For the 1,2-dihydronaphthalene reference PI spectrum, the purity of the sample was confirmed with conventional NMR and the spectrum was remeasured using a second commercial sample, finding no change in the spectrum (Figure S1).</p><!><p>Reaction enthalpies and adiabatic ionization energies (AIEs) are calculated from electronic and zero-point energies computed in Gaussian 09. 45 All reaction enthalpies and barriers were calculated using the composite G3X-K method at 0 K. 46 The CBS-QB3 method [47][48][49] was used to calculate AIEs and relative enthalpies for sets of 78 Da (C6H6), 116 Da (C9H8), and 130 Da (C10H10) isomers. Enthalpies are reported in kcal mol -1 and AIE are reported in electron volts (eV).</p><p>Both are calculated with 0 K electronic energies and include the zero-point energy correction.</p><p>Stationary points were classified as either minima (no imaginary frequencies) or saddle points (one imaginary frequency). The assignment of a transition state between minima was verified by IRC calculations. 50 RRKM-ME analysis was performed on both reaction systems using the MultiWell 2020 software package. [51][52][53] Geometries, vibrational frequencies and calculated enthalpies for stationary points were taken from Gaussian 09 calculations, as described above, and internal degrees of freedom were approximated as either harmonic oscillators or hindered rotors. Hindered rotor potentials were determined using Fourier analysis of relaxed coordinate scans of the relevant rotor using a step size of 20°. Collisional energy transfer was modelled with the biexponential model DEdown = 200 cm -1 , as implemented in MultiWell 2020, using He as the collider. The Lennard-Jones parameters used for He were σ = 2.5 Å and ε/kB = 9.9 K, while Lennard-Jones parameters for all potential wells were set to σ = 5.92 Å and ε/kB = 410 K (obtained from experimental results from toluene). 54 The energy grain was 10 cm -1 for all calculations and the number of trials was 10 7 for all simulations. Master equation simulations of these systems were performed at 600 K and 4 Torr, in accord with the experimental conditions. Based on the number density of allene and propyne in the reactor, pseudo-first order rate coefficients (k1st) for m/z 116 and m/z 130 growth can (in the absence of a pseudo first-order analysis) be used to estimate an upper-bounds for the second-order rate coefficients for the title reactions under the 600 K and 4 Torr conditions. These estimates are (2.3 ± 0.3) × 10 −14 molecule cm 3 s −1 for o-CH3C6H4 + allene and (1.7 ± 0.1) × 10 −14 molecule cm 3 s −1 for o-CH3C6H4 + propyne (where uncertainties are simply 1s from the exponential fit). These values are similar in magnitude to the values predicted by Vereecken et. al. for phenyl + allene (5 × 10 −14 cm 3 molecule −1 s −1 ) 55 and phenyl + propyne (9 × 10 −14 cm 3 molecule −1 s −1 ) at 600 K. 56 Experiments performed at 421 K for phenyl + allene report the k2nd at (5.08 ± 1.07) ×10 9 cm 3 mol −1 s −1 (8.44 ×10 −15 cm 3 molecule −1 s −1 ). 57 These studies also predict that rates of H atom abstraction (i.e., C6H5 + C3H4 à C6H6 + C3H3) will be at least an order of magnitude slower than radical addition at this temperature. [55][56] The kinetic traces of all labelled ions in Figure 1 are included in the Supplementary Information (Figure S3 and Figure S4, Table S1) and this also assists with identifying species not relevant to the title reactions.</p><!><p>The mass spectra presented in Figure 1 are consistent with C10H10 + H and C9H8 + CH3 comprising the major primary product channels from the o-CH3C6H4 + C3H4 reactions and identifying these product isomers and the mechanisms leading to them is the main emphasis of this work (vide infra).</p><p>However, a number of other products are identified in the mass spectra, which we first address briefly. Note that these products may arise from a number of processes, including side and secondary reactions, wall chemistry, photolysis, and dissociative photoionization.</p><p>For the allene case, side processes lead to the formation of m/z 39, 41 and 78 and this is based on their kinetic traces and fitted growth rates as shown in Figure S3 and Table S1. The m/z 39 signal is assigned as propargyl (C3H3), presumably a product of 248 nm allene photolysis. Propargyl is not a bimolecular reaction product, as the true reaction products at m/z 116 and m/z 130 have comparatively slower growth rates (Table S1). Although allene has a relatively low 248 nm absorption cross section 58 at 298 K, it likely has an elevated cross section at 600 K. It is known that propargyl recombines to form m/z 78 and this is in accord with the measured kinetic trace of m/z 78 (Figure S3). 59 The source of the m/z 41 signal is less clear, but its fast rise appears to indicate that it is related to the allene photochemistry (Figure S3, Table S1). Similarly, based on the kinetic trace (Figure S4) the m/z 39 signal in Figure 1 (b) appears to arise from side reactions, likely form the 248 nm photolysis of propyne.</p><p>Besides the addition channel for the o-CH3C6H4 + C3H4 reactions, bimolecular products toluene (m/z 92) + propargyl (m/z 39) could be formed by a direct or indirect H-abstraction mechanism.</p><p>Such a mechanism might be expected to proceed in minor amounts based on previous work on the phenyl radical. [55][56] Unfortunately, here these m/z channels have interference from other processes.</p><p>We have shown previously that the photolysis of o-iodotoluene in the absence of other reactants gives rise to m/z 90 and 92; 17 thus, the abstraction reaction is obscured in these experiments.</p><p>However, experiments with isotopically labelled propyne-d4 discussed later in the manuscript provide evidence that abstraction is only a minor channel.</p><p>Returning to the reaction products of the title reactions, it is evident from the ion intensities shown in Figure 1 that, following addition, H atom loss (m/z 130) is favored for the allene case and that CH3 loss (m/z 116) is favored for propyne. At 9.3 eV the m/z 116:130 peak ratio for the propyne reaction is ca. 4:1 whereas the ratio is ca. 1:5 for the allene case. As discussed later this difference in branching ratio implies distinct mechanistic pathways between the two reactions, but first the isomeric assignments of m/z 116 and m/z 130 need to be considered. Da) acquired on the same instrument at 600 K. The match is very close for the propyne reaction, where indene is assigned as the primary reaction product, and the signal-to-noise ratio is superior to the allene reaction, where indene is formed in lower abundance. The PI signal onset at ~8.1 eV agrees with the calculated (8.2 eV) and literature (8.15 eV) 60 adiabatic ionization energy (AIE) for indene in Table S2. Other plausible isomers are ruled out on the basis of their ionization energy and relative stability (Table S2). This information leads us to conclude that indene is the sole m/z 116 isomer detected between 7.8-9.3 eV for both reactions.</p><p>For assignment of the m/z 130 product, there are several plausible candidates arising from H atom loss from the initial C10H11 adducts. The adiabatic ionization energy (AIE) values and relative enthalpies for a set of C10H10 isomers were calculated (Table 1) and out of these isomers, both 2propynyl-toluene and 2-propadienyl-toluene are deemed unlikely as they have relative enthalpies 28 -32 kcal/mol higher than the lowest energy C10H10 isomer, placing them near the energy of the reactants. However, based on relative stability the remaining C10H10 isomers (1,2dihydronaphthalene, 1,4-dihydronaphthalene, 1-methylindene, 2-methylindene, and 3methylindene) are all plausible candidates, with accessible reaction pathways (vide infra). Aside from 1,4-dihydronapthalene (AIE = 8.6 eV), the candidate C10H10 isomers have ionization energies in the range of ca. 7.8 -8.3 eV, complicating their discrimination. The m/z 130 PI spectrum obtained from each reaction is reported in Figure 3, along with measured reference spectra for 1,2-dihydronaphthalene (AIE = 8.0 eV) and 2-methylindene (AIE = 7.8 eV).</p><p>Unfortunately, high-purity samples for 1,4-dihydronaphthalene, 1-methylindene, and 3methylindene are difficult to acquire and so reference PI spectra for these isomers are unavailable.</p><p>We also checked the PI spectrum of m/z 130 for both reactions at 500 K, compared with 600 K, and the traces are essentially the same (Figure S5).</p><p>Compared to the 1,2-dihydronaphthalene reference spectrum, the m/z 130 PI spectrum for allene tracks the reference spectrum well, including at the important onset region, but deviates at photon energies ≥8.6 eV. This deviation points to the presence of another isomer with a higher energy ionization onset and aligns with the AIE of 1,4-dihydronaphthalene (AIE = 8.6 eV in Ref. 63).</p><p>C Thus, we tentatively assign the m/z 130 product signal from the o-CH3C6H4 + allene reaction as a mixture of 1,2-and 1,4-dihydronaphthalene.</p><p>For the propyne case, the match of m/z 130 signal with the known spectrum for 1,2dihydronaphthalene looks rather good at first pass. However, the experimental signal is above baseline between 7. In summary, signal at m/z 116 can be conclusively assigned as indene for both studied reactions.</p><p>Assignment of products at m/z 130, however, is less certain. For o-CH3C6H4 + allene, where C10H10 + H is the major reaction channel, the reaction forms a mixture of 1,2-dihydronaphthalene and 1,4dihydronaphthalene. However, these dihydronaphthalene isomers do not appear to be products of the o-CH3C6H4 + propyne reaction, where C10H10 + H is a minor reaction channel. In this case, we assign 2-methylindene as a C10H10 product with contributions from some other isomer(s).</p><p>Importantly, these are not unequivocal assignments however, but by comparing the two reactions it is clear that CH3 loss to yield indene is favored for the propyne reaction whereas H atom loss to form dihydronaphthalene isomers is favored for the allene reaction. To develop a rationale for x 5</p><p>these differences, and provide more insight into the product assignment, the reaction mechanisms are required and therefore the potential energy landscapes for both reactions will now be discussed.</p><!><p>To develop a mechanism consistent with the experimental results, and in return provide insight into unresolved assignments of the product detection experiments, enthalpies for key intermediates and transition states were calculated for both o-CH3C6H4 + allene and o-CH3C6H4 + propyne reactions using the G3X-K method. From previous studies on ortho-substituted phenyl radical oxidation, [32][33][34] we expect that the ortho-substituent will influence the mechanism and the final product distribution compared to the analogous phenyl + C3H4 reactions.</p><p>In the schemes reported below, A denotes stationary points from o-CH3C6H4 + allene, P for reactions with propyne, and C for stationary points common to both C3H4 reactions. All 0 K enthalpies are reported in kcal mol −1 relative to o-CH3C6H4 + allene (A0). Starting with Figure 4, two adducts are formed from o-CH3C6H4 + propyne (P0, −0.2 kcal mol −1 ), with enthalpies at −40.7 kcal mol −1 for adduct P1 and −35.7 kcal mol −1 for adduct P2. Isomerization between P1 and P2 is mediated by P3 (−16.0 kcal mol −1 ). If adduct P1 is formed, rearrangement and decomposition leads to indene + CH3 (C4) via first a 1,5-H atom shift (TS P1→P4 −37.8 kcal mol −1 ), then cyclisation (TS P4→P5, −30.1 kcal mol −1 ) and finally CH3 loss to form indene with a reaction enthalpy of −43.4 kcal mol −1 . Alternatively, from P5, H-atom loss can produce 2-methylindene + H (C7) via a transition state at −32.9 kcal mol −1 , which is 4.9 kcal mol −1 higher in energy than TS P5→C4. This higher H atom loss barrier is consistent with the favored loss of CH3 observed experimentally for o-CH3C6H4 + propyne (Fig. 1). Adduct P1 forms following the lower barrier for propyne addition and is therefore expected to be the major entrance channel, suggesting the m/z 130 product for the propyne case is 2-methylindene based on the exclusivity of the end product of this pathway. The direct H-atom abstraction from o-CH3C6H4 produces toluene (C6) and the propargyl radical at −21.5 kcal mol −1 but appropriate treatment of this abstraction pathway, and the barrier along this coordinate, would require a more detailed computational investigation that is outside the scope of this study. Experimentally we observe ion signals consistent with this H-atom abstraction mechanisms for both reactions (m/z 39 and m/z 92 in Figure 1), although, as discussed above, there are several side processes that will also produce these signals.</p><!><p>To test the Figure 4 mechanism, the o-CH3C6H4 + propyne reaction was repeated with propyne-d4 (44 Da) and the resulting PI mass spectrum is shown in Figure 5. Two products are detected at m/z 117 and 134 and, accordingly, the m/z 117 product ion (+1 Da shift from m/z 116) is consistent with CD3 loss to form C9H7D (and the small m/z 118 signal is the expected natural 13 Cisotopologue of C9H7D). The m/z 134 product is consistent with H loss from the adduct to form C10H6D4. These pathways are included in the scheme in Figure 5 Deuteration also allows us to monitor the direct H-abstraction pathway for this reaction as the Dabstraction from propyne-d4 by o-CH3C6H4 would generate singly deuterated toluene at m/z 93 (shifted away from the m/z 92 background signal). The signal at m/z 93 in Figure 5 is a mixture of the 13 C-isotopologue of the background m/z 92 signal and the deuterated toluene adduct. Although the m/z 93 peak is not time resolved, the signal is 2.5% larger in ion intensity compared to the expected 13 C-abundance of o-iodotoluene and this may point to a small product signal resulting from D-abstraction by o-CH3C6H4. For the propyne case, the direct abstraction pathway appears to be minor under these conditions even accounting for the kinetic isotope effect in the labelling experiment. Returning to the calculations, now for the allene reaction, Figure 6 shows the formation of the two adducts from o-CH3C6H4 + allene (A0) located at −36.1 kcal mol −1 (=CH2 addition, adduct A1) and −59.1 kcal mol −1 (=C= addition, adduct A2). Isomerization between adducts A1 and A2 is via A3 (−24.7 kcal mol −1 ). Pathways leading to the formation of 1,2-dihydronaphthalene + H (A8, −34.6 kcal mol −1 ) and 1,4-dihydronaphthalene + H (A9, −31.7 kcal mol −1 ) commence with a 1,5 H-atom shift via TS A1→A4 (−24.1 kcal mol −1 ), the cyclisation from A4 to form the 1,2,4trihydronaphthalen-3-yl radical (A5, −63.0 kcal mol −1 ) followed by subsequent H-atom loss forming either A8 (1,2-dihydronaphthalene) or A9 (1,4-dihydronaphthalene). Recall that from the experimental results in Figure 3, the m/z 130 PI spectrum for the allene reaction, the assignment of 1,2-dihydronaphthalene was based on the fit to a reference spectrum up to ca. 8.6 eV, where the deviation from the reference suggests the presence of the 1,4-dihydronaphthalene isomer. The production of both 1,2-and 1,4-dihydronaphthalene is supported by the mechanism in Figure 6, where pathways to both end products are accessible and competitive. There are other pathways to these products. A cyclisation and rearrangement process from A4 (A4→A7→A10→A6) or a 1,2-H atom shift from A5 (A5→A6), forms the A6 intermediate, which dissociates to exclusively form 1,2-dihydronaphthalene + H, which could add to the yield of 1,2dihydronaphthalene relative to 1,4-dihydronaphthalene. Quantitative insight will come from the RRKM-ME calculations below, however even at this point it is apparent that the allene reaction pathways favor H atom loss, and this is in accord with the experimental results (Figure 1a).</p><!><p>Figure 7 shows the alternative addition pathways for propyne (adduct P2) and allene (adduct A2), which lead to a common intermediate (C1, −58.2 kcal mol −1 ) and so after this point the mechanism is shared. Cyclisation of C1 produces C2 (−75.2 kcal mol −1 ), via TS C1→C2 (−34.2 kcal mol −1 ) then produces 1-methylindene + H (C5). Previous studies report that 1-methylindene and 3methylindene readily isomerize at 600 K, [64][65] so both isomers would result by this pathway in our experiment.</p><p>The 1-methylindene (C5) product is also formed from C3 following a 1,2-H atom shift from C2.</p><p>Alternatively, C3 may eliminate CH3 to form indene (C4). The likely rate limiting step for CH3 loss is the C2→C3 1,2-H atom shift, which is 6 kcal mol −1 higher than direct H atom loss from C2. Therefore, indene + CH3 production from C1 is predicted to be minor, but non-zero, and is nevertheless assigned as the pathway responsible for the small indene + CH3 signal observed for the allene experiments. Based on entrance barrier energetics, however, it is expected that flux through C1 will be overall minor compared to the formation of either P1 or A1. Based on these potential energy schemes, the majority of the H atom loss product for the allene case is assigned as a mixture of 1,2-dihydronaphthalene and 1,4-dihydronaphthalene. As the potential energy scheme in Figure 7 shows, it is possible that some of the m/z 130 signal originates from the formation of 1-methylindene (and, as noted above, some subsequent thermal isomerization to 3-methylindene at 600 K), however, as 1-methylindene has a similar AIE to 1,2dihydronaphthalene (both ~ 8.0 eV) this assignment cannot be verified experimentally. For o-CH3C6H4 + propyne, the signal at m/z 130 (corresponding to H atom loss) is plausibly a combination of 1-, 2-, and 3-methylindene. The PI spectrum for m/z 130 does exhibit an onset at 7.8 eV, consistent with the AIE of 2-methylindene. No plausible mechanism leading to "six-six" bicyclic ring formation was identified for the o-CH3C6H4 + propyne reaction. To extract more insight from these calculations, RRKM-ME modelling was performed using these stationary points.</p><!><p>To improve on the predictions based on potential energy schemes, we examined the kinetics of both reactions with RRKM-ME simulations as implemented in the MultiWell program. 51 These simulations predict branching ratios for each available product set, as well as the formation of stabilized C10H11 adducts. Two sets of model runs were required for each reaction, starting at each of the two initial adducts (adduct P1 or adduct P2 for the propyne case and adduct A1 or adduct A2 for the allene case -structures of these adducts are in Figure 4 and 6). As small errors in the enthalpies can result in large deviations in the predicted branching ratio, the G3X-K method was used for all stationary points. 46 An accurate method is particularly important for the reliable determination of key competing channels, such as between adduct A1 and adduct A2. All stationary points in the simulation for each reaction are included in the Electronic Supplementary Information (Figures S6 and S7). The H-abstraction pathway was not included in these simulations. Table 2 lists the branching ratio results of these calculations depending on the starting adduct. When P1 is the starting adduct, the majority of product signal is predicted to be indene + CH3 (89%), with a minor contribution originating from H atom loss (2%) -comprising 1-methylindene (1%) and 2-methylindene (1%). In contrast, simulations from adduct P2 predict that CH3 loss is minor (2%), and 1-methylindene + H will dominate (98%). When adduct A1 is the starting point, no CH3 loss product is predicted whereas H atom loss again dominates at 94%, comprising 1,2dihydronaphthalene (58%) and 1,4-dihydronaphthalene (35%). Similar to adduct P2, branching from adduct A2 has a minor contribution from CH3 loss (4%) and H atom loss dominates with 1methylindene and 3-methylindene products (96%).</p><p>To estimate the branching ratio for each reaction, the integrated densities of states for the corresponding transition states leading to adduct formation were compared at an activation energy of 18 kcal mol -1 (the estimated internal energy of A0 at 600 K). The result of this is that 72% of propyne addition leads to adduct P1 and 28% to adduct P2, while 70% of allene addition forms adduct A1 and 30% forms adduct A2. Absolute branching ratios were then determined by normalizing the calculated branching ratios in Table 2 to the calculated partition between the two entrance channels for each reaction. The results for this comparison are shown in Table 3. Overall, comparing the calculated branching ratios, the o-CH3C6H4 + propyne reaction is predicted to produce indene + CH3 in much higher yields compared to o-CH3C6H4 + allene (65% vs 1%), consistent with the experimental findings. For the propyne case the predicted H atom loss branching ratio is 29% mostly arising via the formation of 1-methylindene (28%). A small portion of this H atom loss channel is predicted to form 2-methylindene (<1%). Therefore, the majority of the H atom loss products (m/z 130) from o-CH3C6H4 + propyne are assigned as a mixture of 1methylindene (and thus some 3-methylindene from thermal isomerization), while a smaller fraction is predicted to form 2-methylindene resulting in the onset at 7.8 eV as seen in Figure 3. In agreement with the assignment from section 3. investigating the reaction of p-CH3C6H4 with allene and propyne found a shared general mechanism for both reactions: H loss to 6-methyl-1H-indene and 5-methyl-1H-indene. 35 Similarly, for the phenyl radical previous studies indicate that reactions with allene and propyne both produce indene as the major product. [66][67] This difference in the present study directly results from the presence of the ortho methyl substituent, which alters the reaction mechanism and thus leads to distinct major products across both reactions. Together these results indicate that the molecular configuration of the hydrocarbon reactant can alter the formation of the second aromatic ring and reinforces previous studies which suggested that an ortho-methyl substituent influences the reaction mechanism and changes the final product distribution. "six-five" bicyclic rings -thus demonstrating that by only changing the reactant isomer the branching ratio between five-and six-membered ring growth, one of the primary factors influencing PAH formation, can be altered.</p><!><p>• Photoionisation spectra of second 1,2-dihydronapthalene sample.</p><p>• Product mass spectra of allene + o-CH3C6H4 and propyne + o-CH3C6H4 at 600 K and 500K.</p><p>• Kinetic traces and derived first-order rate coefficients of major m/z channels for both reactions at 600 K.</p><p>• Calculated CBS-QB3 adiabatic ionization energies and relative enthalpies for possible C9H8 product isomers.</p><p>• Photoionisation spectra of m/z 130 for allene + o-CH3C6H4 and propyne + o-CH3C6H4 at 600 K and 500K.</p><p>• Complete potential energy schemes used for the RRKM-ME simulations of allene + o-CH3C6H4 and propyne + o-CH3C6H4.</p><!><p>The manuscript was completed with contributions of all authors. All authors have given approval to the final version of the manuscript.</p><!><p>There are no conflicts to declare.</p>

ChemRxiv

Rad6 upregulation promotes stem cell-like characteristics and platinum resistance in ovarian cancer

Ovarian cancer is the fifth most deadly cancer in women in the United States and despite advances in surgical and chemotherapeutic treatments survival rates have not significantly improved in decades. The poor prognosis for ovarian cancer patients is largely due to the extremely high (80%) recurrence rate of ovarian cancer and because the recurrent tumors are often resistant to the widely utilized platinum-based chemotherapeutic drugs. In this study, expression of Rad6, an E2 ubiquitin-conjugating enzyme, was found to strongly correlate with ovarian cancer progression. Furthermore, in ovarian cancer cells Rad6 was found to stabilize \xce\xb2-catenin promoting stem cell-related characteristics, including expression of stem cell markers and anchorage-independent growth. Cancer stem cells can promote chemoresistance, tumor recurrence and metastasis, all of which are limiting factors in treating ovarian cancer. Thus it is significant that Rad6 overexpression led to increased resistance to the chemotherapeutic drug carboplatin and correlated with tumor cell invasion. These findings show the importance of Rad6 in ovarian cancer and emphasize the need for further studies of Rad6 as a potential target for the treatment of ovarian cancer.

rad6_upregulation_promotes_stem_cell-like_characteristics_and_platinum_resistance_in_ovarian_cancer

1. Introduction<!>2.1. Cell Lines and reagents<!>2.2. Clonogenic survival assays<!>2.3. Western blotting<!>2.4. Sphere formation assays<!>2.5. Generation of carboplatin-resistant cells<!>2.6. Immunohistochemistry<!>2.7. 3-D culture invasion assay<!>2.8. Statistical analysis<!>3.1. Rad6 expression correlates with ovarian cancer stage<!>3.2. Rad6 levels correlate with platinum drug resistance and stem-like characteristics<!>3.3. Ectopic overexpression of Rad6 induces carboplatin-resistance and stem-like properties<!>4. Discussion

<p>Ovarian cancer (OC) is a serious problem worldwide and is the most deadly gynecological malignancy in women in the USA [1]. OC is typically asymptomatic during early development and thus is frequently detected at late stages where prognosis is poor. Currently, the five year survival rate is only 45.6%, a number which has changed little in decades [2]. This highlights the need to find new treatment options for patients with ovarian cancer.</p><p>Rad6 is an E2 ubiquitin-conjugating enzyme originally identified in yeast where it is required for DNA repair, induced mutagenesis and proliferation [3]. Humans have two homologs, Rad6A and Rad6B, which were able to complement mutant Saccharomyces cerevisiae Rad6 in DNA repair [4]. In humans, Rad6 has been shown to regulate gene transcription through modulation of chromatin conformation by histone modification and degradation [5–7]. Furthermore, Rad6 plays a pivotal role in choosing which DNA repair pathway is used [8,9]. Loss of Rad6 sensitizes cells to DNA damage and chemotherapeutic drugs [7]. Conversely, overexpression of Rad6 corresponds with mitotic abnormalities and can lead to transformation [10]. High levels of Rad6 correlate with melanoma development and progression [11]. Elevated Rad6 was also found in breast cancer, where it promotes malignant progression through stimulation of the Wnt/β-catenin signaling pathway [10,12].</p><p>In this report we examined the expression of Rad6 in ovarian cancer patient tissue and found that its expression correlated with tumor stage. In OC-derived cell lines increased Rad6 expression led to increased expression of stem cell markers and components signaling pathways that promote stemness. Cells expressing higher levels of Rad6 were also more capable of anchorage-independent growth, a key property of cancer stem cells (CSCs). Furthermore, ectopic overexpression of Rad6 increased resistance to carboplatin in ovarian cancer cells. Therefore, Rad6 is important for the progression of ovarian cancer and promotes stem cell characteristics that can provide the tumor with increased capacity for chemoresistance, proliferation and metastasis.</p><!><p>Fallopian tube epithelial cells (FTSEC or FTEC) were used as normal ovarian cells (generously provided by Dr. Amir Jazaeri) [13]. OV90 and SKOV3 cells were purchased from ATCC, isogenic A2780 (cisplatin-sensitive) and A2780/CP70 (cisplatin-resistant) cells were previously described [14]. OV90 and SKOV3 cells were cultured in 1:1 DMEM/F12 (Mediatech). FTSEC, A2780 and A2780/CP70 cells were cultured in RPMI [14]. All media were supplemented with 10% FBS and 1x Penicillin/Streptomycin. Carboplatin was from Sigma and Rad6 expression vector was obtained from Addgene [15]. The siRNAs used in this study were purchased from Dharmacon and the transfections were done using Lipofectamine 2000 (Invitrogen) following the manufacturer's protocol. Antibodies specific to the following proteins were used: Gli1 (Cell Signaling Technology); GAPDH, ALDA1H1, BMI1, Nanog, OCT4, Myc, and β-Catenin (Santa Cruz Biotechnology); Rad6 (Bethyl Laboratories); H2B, H3K79me3 and SOX2 (Abcam); and Ub-H2B (Millipore).</p><!><p>For clonogenic survival assays, 500 cells were plated in 6-well culture dishes in triplicate [16]. Cells were allowed to attach overnight and treated with indicated concentrations of carboplatin or vehicle control (DMSO) overnight. After the drug treatment cells were washed three times with PBS and two times with growth medium and allowed to form colonies in complete growth medium. After 8 to 12 days colonies were fixed in methanol, stained with crystal violet (0.5% w/v) and counted as described [16]. Only colonies containing more than 25 cells were counted.</p><!><p>For Western blot analysis cell lysates were prepared after washing three times with ice-cold PBS. Cells were lysed in ice-cold cytoskeletal (CSK) buffer freshly supplemented with protease and phosphatase inhibitors (Roche) as described [17]. After quantitation of the protein concentrations, samples were normalized and prepared in 5x Laemmli buffer and heated to 100°C for 15 min. Denatured samples were resolved by SDS-PAGE gel followed by immunoblotting as described [17].</p><!><p>The adherent ovarian cancer cells growing in log phase were harvested by trypsinization, counted and seeded in ultra-low attachment 6 well dishes at 1×104 to 1000 cells/well. The cells were allowed to grow and form spheres as detailed elsewhere [18]. In brief, the counted cells were carefully dispersed as single cells, cultured in stem cell-specific serum free media (2 mL) in an ultra-low attachment six well plates for 10–12 days. This defined media (1:1 DMEM/F-12) was supplemented with 1% penicillin-streptomycin, B27 and N2 supplements (all from Gibco), and growth factors [recombinant human epidermal growth factor (EGF) and fibroblast growth factor (FGF), both from Invitrogen]. These conditions support stem cell growth, and with time, cells proliferate to form floating single cell cloned spheres, known as OC spheres. After seeding the cells were observed daily to ensure that spheres were forming as a result of multiplication from a single cell and not due to cell adherence. Fresh media containing growth supplements EGF (20 ng/ml) and FGF (20 ng/ml) was added every 72 h. The spheres containing ≥ 50 cells were scored as large (true stem cell spheres), while spheres <50 but >15 cells were considered to be small spheres.</p><!><p>The carboplatin-resistant SKOV3 cell lines were generated by intermittently exposing the original SKOV3 cell line to sequentially increased concentrations of carboplatin. The SKOV3 cells were exposed to different concentrations of carboplatin for 72 hours and then allowed to recover in the drug free medium. This process was repeated to get resistant cell lines at a range of concentrations of carboplatin (1 to 20 μM). Resistance was confirmed by clonogenic survival assay. The cell lines were named according to level of carboplatin resistance. The cell line widely used here is SKOV3/CP20 which is resistant to 20 μM carboplatin.</p><!><p>Normal ovarian tissue and ovarian cancer tumor samples were purchased from US BIOMAX and expression of Rad6 protein was measured by immunohistochemistry. Tissue sections were incubated with anti-Rad6 antibody, followed by a specific biotinylated secondary antibody (1:250 dilution), and then conjugated HRP streptavidin and DAB chromogen and tissues were counterstained with hematoxylin. Stained sections were analyzed by Zeiss Axioscope 2 microscope and images captured by AxioCam camera at 400x magnifications were analyzed by NIS Element AR software (Nikon). This process was carried out without any knowledge of the identity of each tissue sample to prevent bias in scoring the samples.</p><!><p>OC cell spheres generated as described above were harvested and immobilized in Matrigel according to manufacturer's instructions (Cultrex). Briefly, 4-chambered glass slides were coated with thin layer of Matrigel (80 μL/well). Approximately 100 spheres in PBS were mixed (1:1 ratio) with Matrigel and 500 μl was layered on top of the solidified basement Matrigel-coated chamber slide. The slide was incubated at 37°C for 30 min. After Matrigel had solidified complete growth media supplemented with EGF (20 ng/ml) was added. The spheres were allowed to grow in this 3-D culture media for 72 hr. The spheres were monitored and imaged by Zeiss Axioscope 2 microscope and images were captured for every 24 hr by AxioCam camera at 100x magnification.</p><!><p>Clonogenic survival data and sphere formation data presented are averages of three independent experiments. The experiments performed in triplicate each time. Error bars represents the ± SEM. Data were analyzed either by GraphPad Prism 6 or Excel 2010 (Microsoft).</p><!><p>Rad6 levels correlate with severity of melanoma and breast cancer [10,11]. To determine if Rad6 is also associated with disease progression in OC, normal and staged OC tissue samples were stained for Rad6 (Fig. 1A). The identities of the samples were concealed to limit any bias in scoring for Rad6 signal intensity. Consistent with the findings in aforementioned cancers, Rad6 levels correlated well with disease progression (Fig. 1B). Aberrant Hedgehog (Hh)/Gli signaling is commonly associated with tumor growth, metastasis, and resistance to chemotherapy [16,19] and we previously demonstrated that ALDH1A1 regulates DNA repair and checkpoint progression to enhance resistance to platinum drugs [14]. Therefore, we examined stage 3 and 4 OC and adjacent normal tissue for expression of these proteins in addition to Rad6. All three proteins were highly expressed in the tumor samples compared to normal tissue (Fig. 1C). In order to determine if the high levels of these proteins is due to enhanced transcription of their genes in the OC tumors, mRNA levels of each were quantitated. Gli1 and Rad6 message correlated with OC stage similarly to protein, with Gli1 mRNA being 4 to almost 10-fold higher in grade 3 and nearly 10-fold higher in stage 4 and Rad6 3 to 7-fold more in stage 3 and 13-fold greater in stage 4 tissue compared to normal ovarian tissues (Fig. 1D).</p><p>To further establish the correlation between Rad6 and OC we analyzed a normal, immortalized cell line derived from fallopian tube epithelial cells, a proposed OC precursor [13], and several OC cell lines, including 2 isogenic pairs of platinum-sensitive (A2780 and SKOV3) and platinum-resistant (A2780/CP70 and SKOV3/CP20) lines. As expected the OC cell lines had higher Rad6 expression than normal cells (Fig. 1E). Interestingly, Rad6 levels were also significantly higher in the cell lines with platinum drug resistance.</p><!><p>Carboplatin-resistant A2780/CP70 cells exhibited greater Rad6 expression than its isogenic carboplatin-sensitive counterpart (Fig. 1E). SKOV3/CP20 cells were generated by sequential exposure to increasing concentrations of carboplatin and analyzed for Rad6 expression. Rad6 levels increased with the level of platinum resistance in these cells (Fig. S1). Together these results show that Rad6 correlates with platinum resistance in ovarian cancer cells; a finding that is consistent with earlier studies reporting that Rad6 can promote chemoresistance [7]. Rad6 is also known to stabilize β-catenin and stimulate Wnt signaling [12], a pathway associated with stem cell properties [20]. Thus, we examined the capability of the isogenic platinum-sensitive and –resistant A2780 cells for anchorage-independent growth as spheres, an indicator of stem cell-like properties. The resistant A2780/CP70 cell line generated substantially more spheres in culture (Fig. 2A & B). To further confirm the correlation between Rad6, stemness and platinum resistance we examined the expression levels of several stem cell markers (ALDH1A1, BMI1, SOX2, Nanog, OCT4) and signaling proteins (Gli1 and β-catenin) and found that all were upregulated in the platinum-resistant cells (Fig. 2C and not shown). An increase in histone 2B (H2B) ubiquitination and formation of histone H3K79me3 were also seen in these cells. Alterations in H2B ubiquitination have previously been shown to enhance transcription in cancer cells [21] and Rad6 has been implicated in H2B ubiquitination [5]. Histone H3K79 trimethylation is dependent upon H2B monoubiquitantion by Rad6 and is a marker for an open transcriptionally active region [22,23]. Additionally, Matrigel invasion assays showed that the more stem cell-like, platinum resistant A2780/CP70 cells were also more invasive than their platinum-sensitive counterpart (Fig. 2D).</p><p>At least 80% of OC cases in the US involve serous tumors [24]; therefore, it is important to confirm the correlation between stemness, platinum resistance and Rad6 expression using cell lines derived from serous tumor (SKOV3). As seen with the isogenic A2780 cell lines, the platinum-resistant SKOV3/CP20 cells were more proficient at anchorage-independent growth (Fig. 3A & B), and expressed higher levels of Rad6, stem cell markers, and Wnt and Hh signaling molecules (β-catenin and Gli1) than the isogenic platinum-sensitive SKOV3 (Fig. 3C). The SKOV3/CP20 generated for this work were confirmed to be resistant to the platinum chemotherapy drug carboplatin (to 20 μM, hence the name CP20 - Fig. 3D). Importantly, knocking down Rad6B using siRNAs attenuated SKOV3 cells' clonogenic potential (Fig. 3E). This shows that the effects are not cell line specific but are found in cells derived from several types of ovarian cancers.</p><!><p>Rad6 expression correlates with OC severity (Fig. 1A–D), resistance to platinum-based chemotherapeutic drugs (Fig. 1E & S1) and with anchorage-independent growth as spheres and expression of stem cell markers and the signaling molecules β-catenin and Gli1 in OC cell lines (Fig 2 & 3). In order to determine if Rad6 is responsible for the enhanced platinum resistance and the development of stem cell properties, it was overexpressed in SKOV3 cells. Increasing Rad6 expression in these cells was accompanied by increased sphere formation, which is indicative of stem cell characteristics (Fig 4A & B). This was accompanied by increased expression of stem cell markers (SOX2, BMI1, ALDH1A1, Nanog, and OCT4) and ubiquitinated histone 2B (Fig. 4C and not shown) in the SKOV3/Rad6 cells. Rad6 overexpression in these normally platinum-sensitive cells significantly increased resistance to carboplatin in clonogenic survival assays (Fig. 4D). Taken together this data shows that overexpression of Rad6 in ovarian cancer cells can lead to increased resistance to platinum-based chemotherapeutic drugs and enhanced stem cell characteristics potentially leading to more aggressive tumor growth and metastasis. This is significant in light of evidence showing that chemotherapy drug-resistant CSCs can lead to tumor recurrence [25].</p><!><p>Ovarian cancer remains a serious health issue worldwide and in the United States alone there are more than 22,000 new cases annually. Because of its largely asymptomatic progression OC typically goes undetected until reaching an advanced stage with widespread intra-abdominal involvement and/or metastatic spread. While survival rates of many cancers have improved with current detection and treatment methods, OC outcomes have changed little in decades [2]. Therefore, there is a need for better understanding of the nature and pathology of OC in order to develop better treatment options. Thus in this study we have explored the role of Rad6 in ovarian cancer. Expression of the E2 ubiquitin-conjugating enzyme Rad6 correlates well with progression of OC tumors (Fig. 1A–D). This finding correlates with previous studies of Rad6 expression in melanoma and breast cancer [10,11]. This is important because Rad6 has been implicated in regulating proliferation, DNA repair and resistance to chemotherapeutic drugs [3,7–9].</p><p>The expression of Rad6 in laboratory cell lines derived from OC was also high and correlated with resistance to platinum-based drugs (carboplatin and cisplatin – Fig. 1D). When an OC cell line (SKOV3) was made resistant to increasing levels of the chemotherapy drug carboplatin, Rad6 expression increased with resistance (Fig. S1), suggesting Rad6 expression is important for developing drug resistance. This was confirmed by measuring carboplatin resistance in OC cells overexpressing Rad6. These cells showed significantly greater resistance to carboplatin than vector control (Fig. 4D). This confirms previous studies correlating Rad6 expression with drug resistance [7].</p><p>Rad6 has previously been shown to stabilize β-catenin a member of the Wnt signaling pathway [12]. Wnt signaling through β-catenin promotes stemness and self-renewal [20]. There is a wealth of evidence that cancer stem cells (CSCs) promote tumor initiation, cellular heterogeneity, avoidance or resistance to treatment, recurrence and metastasis [26]. Chemotherapy resistance, recurrence and metastasis are major limitations in the current treatment of OC that need to be addressed; therefore, OC cells were analyzed for expression of stem cell markers and capacity for anchorage-independent growth (stem cell characteristic). Analysis of two sets of isogenic OC cell lines showed that the platinum-resistant cell lines that express higher levels of Rad6 (A2780/CP70 & SKOV3/CP20) were enriched in protein markers of stemness compared to platinum-sensitive counterparts (A2780 & SKOV3) which express lower levels of Rad6 (Fig. 2C & 3C). Furthermore, downregulation of Rad6B attenuated clongenic potential of OC cells (Fig 3E). On the other hand, when OC cells were transfected with a plasmid expressing Rad6B, these cells exhibited increased levels of stem cell markers (Fig. 4C) and Rad6 levels correlated with histone modifications suggesting open, actively transcribed chromatin. Each of these cell lines expressing higher levels of Rad6 and stem cell markers were more capable of anchorage-independent growth as OC spheres than their isogenic counterparts that had less Rad6 (Fig. 2A & 2B). Rad6 overexpressing SKOV3 cells formed more spheres than the vector control cells (Fig. 4A & B). Together these findings suggest that Rad6 promotes stemness in OC cells by stabilizing β-catenin. A model representation of these findings is presented in the Supplementary material (Fig. S2). Increased Rad6 expression in OC cells means there is more Rad6-mediated ubiquitin signaling leading to β-catenin stabilization (increased Wnt signaling) and enhanced Gli1 expression (increased Hh signaling). Enhanced activity of these signaling pathways combined with Rad6-directed chromatin remodeling leads to increased expression of proteins that promote stem cell-like properties leading to chemoresistance. Rad6 also partners with Rad18, an E3 ubiquitin ligase to regulate multiple DNA repair pathways including translesion synthesis and the Fanconi anemia/BRCA pathway [27,28].</p><p>Since most ovarian cancer cases are diagnosed at later stages and Rad6 expression correlates with progression of ovarian cancer, many patients will already have high levels of Rad6 expressing tumor cells. Rad6 promotes chemoresistance and stem cell characteristics, which could lead to enhanced disease recurrence and metastasis. These findings suggest that development of new therapies targeting Rad6 may be warranted in the treatment of ovarian cancer.</p>

PubMed Author Manuscript

p38 MAPK Activation, JNK Inhibition, Neoplastic Growth Inhibition and Increased Gap Junction Communication in Human Lung Carcinoma and Ras-Transformed Cells by 4-Phenyl-3-Butenoic Acid

Human lung neoplasms frequently express mutations that down-regulate expression of various tumor suppressor molecules, including mitogen-activated protein kinases such as p38 MAPK. Conversely, activation of p38 MAPK in tumor cells results in cancer cell cycle inhibition or apoptosis initiated by chemotherapeutic agents such as retinoids or cisplatin, and is therefore an attractive approach for experimental anti-tumor therapies. We now report that 4-phenyl-3-butenoic acid (PBA), an experimental compound that reverses the transformed phenotype at non-cytotoxic concentrations, activates p38 MAPK in tumorigenic cells at concentrations and treatment times that correlate with decreased cell growth and increased cell-cell communication. H2009 human lung carcinoma cells and ras-transformed liver epithelial cells treated with PBA showed increased activation of p38 MAPK and its downstream effectors which occurred after 4 h and lasted beyond 48 h. Untransformed plasmid control cells showed low activation of p38 MAPK compared to ras-transformed and H2009 carcinoma cells, which correlates with the reduced effect of PBA on untransformed cell growth. The p38 MAPK inhibitor, SB203580, negated PBA\xe2\x80\x99s activation of p38 MAPK downstream effectors. PBA also increased cell-cell communication and connexin 43 phosphorylation in ras-transformed cells, which were prevented by SB203580. In addition, PBA decreased activation of JNK, which is upregulated in many cancers. Taken together, these results suggest that PBA exerts its growth regulatory effect in tumorigenic cells by concomitant up-regulation of p38 MAPK activity, altered connexin 43 expression, and down-regulation of JNK activity. PBA may therefore be an effective therapeutic agent in human cancers that exhibit down-regulated p38 MAPK activity and/or activated JNK and altered cell-cell communication.

p38_mapk_activation,_jnk_inhibition,_neoplastic_growth_inhibition_and_increased_gap_junction_communi

<!>Materials<!>Cell Culture<!>Cell Growth Assay<!>Apoptosis Assay<!>Preparation of Membrane-Enriched/Alkali-Resistant Fraction for Western Blot Analysis<!>Western Blot Analysis of Connexin 43<!>Western Blot Analysis of Signaling Pathway Proteins<!>Fluorescence Dye Transfer Assay<!>Statistical Analyses<!>PBA activates p38 MAPK in ras-transformed epithelial and human lung carcinoma cells<!>PBA increases phosphorylation of the p38 MAPK downstream effectors, HSP-27 and ATF-2, in ras-transformed epithelial and human lung carcinoma cells<!>PBA-Me also increases p38 MAPK phosphorylation in H2009 human carcinoma and WB-ras1 cells<!>The p38 MAPK inhibitor, SB203580, negates the effect of PBA on the p38 MAPK downstream effector, HSP-27<!>The p38 MAPK inhibitor, SB203580, reduces PBA-induced P2 phosphorylation of connexin 43 and PBA-enhanced cell-cell communication<!>PBA inhibits growth of ras-transformed and human lung carcinoma cells<!>PBA decreases activation of JNK<!>PBA has no effect on activation of Akt, cdc2, CDK2, PKC or p44/42 MAPK<!>DISCUSSION

<p>p38 MAPKs are stress-activated members of the mitogen-activated protein kinase (MAPKs) family that play important roles in the control of cell proliferation in a wide variety of cell types [Raingeaud et al., 1995; Wang et al., 1997; Lewis et al., 1998; Schaeffer et al., 1999; Pearson et al., 2001; Engelberg et al., 2004]. p38 MAPK transmits signals from the cell membrane to the nucleus in response to oxidative or osmotic stress, cytokines, and radiation [Zarubin et al., 2005; Dhillon et al., 2007]. Activation of p38 MAPK via phosphorylation activates the transcription factors ATF-2 and HSP-27 [Raingeaud et al., 1995; Cobb et al, 1995; Rouse et al., 1994], among others [Cohen, 1997; Yang et al., 1999], and other kinases such as MAPKAPK-2 [Kyriakis and Avruch, 2001; Kumar et al., 2003]. Inhibition of tumorigenesis by p38 MAPK activation has been shown to occur in mice that express mutant Erb2 or H-Ras [Bulavin et al., 2004]. p38 MAPK activation also correlated with increased apoptosis [Cao et al., 2004; Liberto et al., 2004; Lowe et al., 2005], and p38 MAPK has been described as a tumor suppressor [Dhillon et al., 2007; Timofeev et al., 2005; Bradham and McClay, 2006]. In addition, regulation of p38 MAPK has been correlated with changes in phosphorylation of the gap junction protein, connexin 43, and the kinase may directly phosphorylate this protein [Lee et al., 2004; Ogawa et al, 2004; Aranvindakshan and Cyr, 2005]. Thus, there appears to be a link between p38 MAPK activity, connexin phosphorylation, and neoplastic transformation.</p><p>Lung cancer is the leading cause of cancer death for both men and women, with over 200,000 new cases occurring in the United States in 2010 [American Cancer Society, 2010]. Five year survival rates for lung cancer are below 10% [Brognard and Dennis, 2002]. Current therapies for lung cancer are poorly effective and there is a need to identify additional therapeutic agents. Human lung neoplasms frequently express mutations in ras-genes and inactivation or deletion of p53 and other tumor suppressor genes [Mitsudomi et al., 1992; Ruch et al., 1998]. Clinical trials of inhibitors of the EGF receptor, which activates the Ras pathway, showed some therapeutic effectiveness in a small subset of individuals [Levitzki, 2003; Comis, 2005; Cho et al., 2007; Wong et al., 2010]. However, these anti-tumor agents were not effective in patients with ras-mutations or defects in downstream effectors [Levitzki, 2003]. Thus, additional agents are needed to treat this common and deadly form of cancer.</p><p>PBA is an irreversible turnover-dependent inhibitor of peptidylglycine-alpha-monooxygenase (PAM) in vitro [Katopodis and May,1990; Katopodis et al., 1990], with anti-inflammatory effects in vivo mediated by a non-COX inhibitory pathway [Bauer et al., 2007]. PBA decreased lung cancer cell proliferation, presumably by inhibiting the synthesis of amidated growth factors [Iwai et al., 1999]. We previously demonstrated that PBA inhibited the growth of ras-transformed WB-F344 cells and up-regulated gap junction-mediated cell-cell communication and connexin 43 [Sunman et al, 2004]. The WB-F344 cells are a diploid rat liver epithelial cell line that exhibits contact inhibition of growth and is non-tumorigenic [Tsao et al, 1984]. These cells also exhibit high levels of gap junction cell-cell communication and connexin 43 expression [Matesic et al., 1994]. A highly tumorigenic and poorly communicating derivative of the WB-F344 cell line known as WB-ras was generated by stable retroviral transduction with the v-Ha-Ras oncogene [de Feijter et al., 1990]. Gap junction-medicated cell-cell communication is involved in cellular growth control and differentiation, and can inhibit tumorigenesis [Trosko and Ruch, 1998]. We now report that PBA activates p38 MAPK in both ras-transformed WB and human lung carcinoma cells at concentrations and treatment times that correlate with decreased cell growth and altered connexin expression. The methylated derivative of PBA, PBA-Me, also inhibits growth and activates p38 MAPK, but at a 5-fold lower concentration. PBA's effects on p38 MAPK and cell-cell communication in ras-transformed cells were prevented by a specific p38 MAPK inhibitor and suggest a link between p38 MAPK activation and cell-cell communication. In addition, PBA inhibited activation of another stress-activated MAPK, JNK, which is over-activated in selected cancers [Antonyak et al, 2002; Rennefahrt, et al., 2004]. These studies underscore the therapeutic potential of PBA and PBA-Me in lung and other cancers that have reduced cell-cell communication and p38 MAPK activity, and/or increased JNK activity.</p><!><p>WB-neo and WB-ras cells were derived from WB-F344 rat liver epithelial cells [De Feijter et al., 1990] and were obtained from Dr. James Trosko at Michigan State University. H2009 human lung carcinoma cells were obtained from the American Type Culture Collection (Manassas, VA). Alpha Modification of Eagle's medium was purchased from Mediatech (Herndon, VA). RPMI medium, L-glutamine, trypsin, and phosphate buffered saline (PBS), were from Fisher Scientific (Pittsburgh, PA). Fetal bovine serum (FBS) was from Invitrogen (Carlsbad, CA). PBA, Lucifer Yellow-CH fluorescent dye, phenylmethylsulfonyl fluoride (PMSF), iodoacetamide (IAA), protease inhibitor cocktail, G418, trypan blue solution, and Ponceau Red solution were from Sigma Chemical Co. (St. Louis, MO). Connexin43 monoclonal antibody (MAb3086) was obtained from Chemicon International (Temecula, CA). Phospho-p38 MAP kinase (Thr180/Tyr182) polyclonal antibody, p38 MAP kinase polyclonal antibody, phospho-ATF2 (Thr71) polyclonal antibody, phospho-HSP27(Ser82) polyclonal antibody, JNK polyclonal antibody, phospho-JNK (Thr183/Tyr185) polyclonal antibody, phospho-MAPKAPK-2 (Thr334) polyclonal, Akt polyclonal antibody, phospho-Akt (Ser473) polyclonal antibody, cdc2 polyclonal antibody, phospho-cdc2 (Thr161) polyclonal antibody, CDK2 polyclonal antibody, phospho-CDK2 (Thr160) polyclonal antibody, phospho-PKC pan (βII ser660) polyclonal antibody, β-actin polyclonal antibody, and anti-rabbit IgG alkaline phosphatase-conjugated antibody were from Cell Signaling Technology (Beverly, MA). Tween-20, TRIS-HCl, DC Protein Assay, SDS, nonfat dry milk, 25x alkaline phosphatase color development buffer, 5-bromo-4-chloro-3-indolyl phosphate/nitroblue tetrazolium (BCIP/NBT), protein molecular mass standards, and all electrophoresis and transfer buffer components were from Bio-Rad (Hercules, CA). LumiPhos chemiluminescence reagent was from Thermo Scientific (Rockford, IL). Biotin-(GT) anti-mouse IgG antibody and alkaline phosphatase-conjugated streptavidin were from MP Biomedicals, LLC (Irvine, CA). Dieldrin was from Accustandard (New Haven, CT). Polyvinylidene difluoride (PVDF) membranes were from Millipore (Bedford, MA). SB203580, p38 MAPK inhibitor III and TdT-FragEL DNA fragmentation detection kit were obtained from EMD Biosciences (La Jolla, CA).</p><p>PBA-methyl ester (PBA-Me) was synthesized as follows: 10 g of PBA were dissolved in 100 mL HPLC-grade methanol acidified with 10 drops of concentrated HCl. The reaction mixture was heated to reflux and allowed to proceed overnight. The reaction was then quenched via the addition of 100 mL water and the pH adjusted to 7.0 using NaOH. The solution was then extracted 4 times with ethyl acetate, and the organic layer was dried over magnesium sulfate, filtered, and evaporated under reduced pressure to yield 9.2 g (85%) of the final product. PBA was re-purified from the Sigma stock by re-crystallization. All other chemicals, reagents, and solvents used were of analytical grade.</p><!><p>WB-neo and WB-ras rat liver epithelial cells were subcloned from single cells to obtain WB-neo3 and WB-ras1 lines, and were used between passages 5 and 18. Cells were grown in alpha Modification of Eagle's Medium supplemented with 2 mM/L l-glutamine and 5% FBS. G418 antibiotic was diluted 1:2 in PBS and added to the cell growth media at a concentration of 500 µM, but was omitted for experiments. H2009 human lung carcinoma cells were grown in RPMI medium supplemented 2 mM/L l-glutamine and 10% FBS. Confluent cells were subcultured by trypsinization and plated at 5–25% confluence during each passage. Cells were incubated in an atmosphere of 5% CO2 at 37°C.</p><!><p>Cells were plated at 5–10% confluence in 2 ml of media on 35 mm dishes and allowed to acclimate for 24 hours. Cells were then treated with vehicle or drug dissolved in sterile H2O (PBA, 0.1 mg/ml = 613 µM) with or without 2 µM SB203580 and incubated at 37°C for the duration of the experiment. Cells were washed once with PBS, treated with 0.5 ml of trypsin until cells no longer adhered to the dish, quenched with 1.5 ml of media, and cells in solution were counted using a hemocytometer.</p><!><p>H2009 cells plated in 35 mm dishes were treated with vehicle or 0.1 mg/ml PBA twice (at t=0 and 24 h). Apoptotic cells were fixed with 4% formaldehyde in PBS and stained using a TdT-FragEL DNA fragmentation detection kit according to the manufacturer's protocol for fixed cell preparations.</p><!><p>Cells were grown in 15 ml of media with vehicle or drug(s) in 75 cm2 flasks to 90–100% confluence. Media was removed and the cells were washed with PBS, then 0.375 ml of TRIS-IAA buffer (10 mM TRIS (pH 7.5), 10 mM IAA, and 1 mM PMSF) was swirled over the cells and 0.55 ml of 40 mM NaOH was added. The cells were then scraped and transferred to microcentrifuge tubes on ice. Each sample was sonicated (two 15 s pulses at 35% maximum power with a Bronson Cell Disruptor 185 Sonicator, allowing 1–2 minutes between pulses). Samples were centrifuged at 14,000 × g for 30 minutes at 4°C. The supernatants were discarded and the pellets were washed with 1 ml of TRIS-PMSF buffer (10 mM TRIS (pH 7.5), and 1 mM PMSF). Samples were centrifuged at 14,000 × g for 15 minutes at 4°C and the supernatants were again removed. Each pellet was re-suspended in 75 µl of TRIS/PMSF buffer and three 5 µl aliquots were removed for total protein assay using the Bio-Rad DC protein assay. The remaining sample was frozen in liquid nitrogen and stored at −20°C.</p><!><p>Membrane-enriched/alkali-resistant protein samples were loaded onto 1 mm, 10 well, 12.5% acrylamide gels and run at 60V until the samples had passed through the stacking gel, then at 120–150V. Proteins were transferred on to PVDF membranes at 20V overnight in the presence of 0.05% SDS. PVDF membranes were washed in water and stained with Ponceau Red solution for 2–3 min. Membranes were incubated for 1 h in block buffer (4% nonfat dry milk, 40 mM TRIS, pH 7.5, and 0.1% Tween-20) and overnight with anti-connexin43 monoclonal antibody (2 µl/10 ml block buffer) with shaking at 4°C. Membranes were washed with block buffer and incubated at room temperature for 1 h with a secondary antibody (anti-mouse biotinylated antibody 25 µl /10 ml block buffer) on a shaker, followed by washing and incubation for 1 h at room temperature with alkaline phosphatase-conjugated streptavidin (diluted 1:400 in block buffer containing 0.5 M NaCl). After washing, bands were visualized using BCIP/NBT.</p><!><p>Cells were grown to 80–90% confluence in 25 cm2 flasks, washed with 10 ml of PBS and extracted with 250 µl 2% SDS, 1 mM PMSF, and 1:100 dilution of protease inhibitor cocktail. Lysed cells were scraped, transferred to microcentrifuge tubes, and sonicated for two, 15 second pulses at room temperature. Protein concentrations were determined by Bio-Rad DC assay and proteins were separated on 12.5% acrylamide SDS gels and transferred to PVDF membranes overnight at 20V or 1 h at 180 V. Membranes were stained with Ponceau Red, then incubated in block buffer for 1–2 hours. P38 MAPK, phospho-p38 MAPK, phospho-HSP-27, phospho-ATF-2, JNK, phospho-JNK, Akt, phospho-Akt, cdc2, phospho-cdc2, CDK2, phosphoCDK2, phospho-PKC or β-actin polyclonal antibodies were incubated separately with blots in block buffer overnight at 4°C. Immunopositive bands were detected using alkaline phosphatase-linked anti-rabbit secondary antibody and development with BCIP/NBT as substrates, or Lumi-Phos where noted. Blots were scanned on an HPscanjet 4400C scanner and band densities determined using UN-SCAN-IT software (version 5.1) from Silk Scientific, Inc. (Orem, UT). Two to five replicate blots were analyzed for each experiment.</p><!><p>To quantify gap junction-mediated cell-cell communication, a fluorescence dye transfer assay was performed, modified slightly from that previously described [Jou et al., 1993]. Cells were grown to 90–100% confluence in 35mm culture dishes and treated with vehicle or drug as indicated. Two dishes in each group were treated with 10 µM dieldrin, a gap junction inhibitor, for 30 min prior to assay. Dishes were then washed once with Ca2+/Mg2+ PBS and twice with PBS. One ml of Lucifer Yellow dye (0.5 mg/ml in PBS) was added to each dish and six to eight score lines were made in the cell monolayer with a surgical blade. Dishes were kept in the dark for 2 min, then washed 3 times with PBS and once with Ca2+/Mg2+ PBS. Cells were fixed with 1.5 ml of 4% paraformaldehyde for 30 minutes, and then washed again with PBS. Fluorescence was observed using a Leitz microscope with a 10× objective lens. Several randomly selected fields on each dish were digitally photographed and the number of fluorescent cells adjacent to score lines was counted in a defined unit area. The number of communicating cells was determined by subtracting the average number of fluorescent cells per unit area in the dieldrin treated dishes (non-communicating cells) from the number of fluorescent cells per unit area in vehicle and drug-treated dishes.</p><!><p>Data are presented as the mean ± the standard deviation (S.D). One-way analysis of variance (ANOVA) was used to test for significance between repeated measures. Tukey's post-hoc test was used following one-way ANOVA to determine significant differences within a group. A probability of P < .05 was considered statistically significant in all calculations. Statistical analyses were performed using Statistix for Windows v8.1.</p><!><p>Treatment of WB-ras1 cells with 0.1 mg/ml PBA for 48 hr increased phosphorylation of p38 MAPK on key activation sites, Thr180/Tyr182 (Figure 1A). Lanes 4 and 5 (separately treated, replicate samples) of Figure 1A show increased immunoreactive band density of phospho-p38 MAPK (top panel), compared to vehicle-treated control lanes 2 and 3 (also separately treated, replicate samples). PBA did not substantially alter total p38 MAPK content in the cells (Figure 1A lower panel). Densitometric scans of blots (Figure 1 B) revealed a 3-fold increase in the density of phospho-p38 MAPK (normalized to total p38 MAPK density) in PBA-treated WB-ras1 cells compared to vehicle-treated control. PBA similarly increased p38 MAPK phosphorylation in H2009 human lung tumor cells (Figure 1 C, top panel, lanes 4 and 5 compared to vehicle-treated control lanes 2 and 3), and had no effect on total p38 MAPK levels (Figure 1C, bottom panel, lanes 2–5). Densitometric scans revealed a ~2.5 fold increase in phospho-p38 MAPK in PBA–treated H2009 cells (Figure 1D). This increased p38 MAPK phosphorylation by PBA was sustained for as long as 14 days post treatment (Figure 1E, top panel, PBA-treated lanes 4 and 5 compared to vehicle-treated control lanes 2 and 3 and quantification in F). In non-transformed plasmid control WB-neo3 cells, a substantially smaller increase in p38 MAPK phosphorylation was observed after treatment with PBA at 0.1 mg/ml for 48 hours (Figure 1 G and H) compared to WB-ras1 cells treated over the same time (Figure 1 A and B). No change in p38 MAPK phosphorylation occurred in WB-ras1 cells treated for 2 h (not shown). A small increase was seen at 4 hr which was not significant (Figure 1 I, p>0.05), while at 24 h treatment, a ~1.7 fold increase in p38 MAPK phosphorylation was observed (Figure 1 J, p<0.05). Similarly, treatment of H2009 cells for 4 h showed no significant change in p38 MAPK phosphorylation (not shown). In the absence of PBA, basal levels of activated p38 MAPK were lower in WB-ras1 and H2009 cells compared to the non-transformed WB-neo3 cells (Figure 2).</p><!><p>Treatment of WB-ras1 and H2009 cells with 0.1mg/ml PBA for 48 h increased the phosphorylation of HSP-27 and ATF-2 at key activation sites by approximately 2–6 fold in identical samples (Figure 3A and B, compare lanes 4 and 5 with vehicle control lanes 2 and 3 in the top and 3rd panels and quantification in B,C,E and F). Treatment of H2009 cells with PBA (0.1mg/ml PBA for 48 h) also significantly increased phosphorylation of MAPKAPK-2, a substrate for p38 MAPK and HSP-27 kinase (8.6 ± 0.13 for PBA-treated cells versus 0.3 ± 0.16 for vehicle-treated cells in relative density units, p< 0.01). PBA had no significant increase on phosphorylation of HSP-27 in WB-ras1 cells treated with 0.1 mg/ml PBA for 2 h or 6 h, or in H2009 cells treated for 4 h (p> 0.05, not shown).</p><!><p>The methylated form of PBA, PBA-Me, also increased the phosphorylation of p38 MAPK in H2009 cells at 10-fold lower concentration (10 µg/ml) than the parent compound. (Figure 4A, compare lanes 4 and 5 with vehicle control lanes 2 and 3) and this effect was further increased at 20 µg/ml PBA-Me (Figure 4A, lanes 6 and 7 with lanes 4 and 5, and quantification in Figure 4B). Similarly, 20 µg/ml PBA-Me stimulated p38 MAPK and HSP-27 phosphorylation in WB-ras1 cells by more than two-fold (data not shown).</p><!><p>The p38 MAPK active site inhibitor, SB203580, inhibits the actions of the kinase on downstream effectors such as HSP-27. When WB-ras1 cells were treated with 2 µM SB203580, basal levels of p38 MAPK phosphorylation (Figure 5A, top panel, lanes 2 ,3, 6 and 7) were detected and the stimulation of p38 MAPK phosphorylation by PBA was not substantially affected (Figure 5A, top panel, lanes 8 and 9 compared to lanes 6 and 7, quantification on 5B). This was expected, since the inhibitor does not affect phosphorylation at the regulatory Thr180/Tyr182 activation sites. However, SB203580 greatly reduced PBA-stimulated phosphorylation of the p38 MAPK downstream target, HSP-27 (Figure 5A, 4th panel from the top, lanes 8 and 9 compared to lanes 4 and 5 and quantification in Figure 5C). β-actin content and Ponceau staining indicate equivalent protein loading of the blots. SB203580 also reduced PBA-stimulated phosphorylation of HSP-27 in H2009 cells. PBA-stimulated phosphorylation of HSP-27 was also reduced in WB-ras1 cells by treatment with another p38 MAPK-specific inhibitor, p38 MAPK inhibitor III (data not shown).</p><!><p>PBA increased the content of the connexin 43-P2 phosphoform relative to levels of the Po (non-phosphorylated) and P1 phosphoforms in WB-ras1 cells (Figure 6A,B), as we previously reported [Sunman et al., 2004]. SB203580 prevented the increase in connexin 43-P2 and had no effect on the basal levels of this phosphoform (Figure 6A,B). PBA also increased cell-cell communication approximately 3.5 fold in these cells, but this increase was negated by SB203580 (Figure 6C). SB203580 had no effect on the basal level of cell-cell communication in these cells. Treatment of H2009 cells with PBA (0.1 mg/ml or 0.2 mg/ml for 48 h or 5 d) did not increase cell cell-communication or connexin 43 P2 content. However, a 3-fold increase in the amount of Po connexin 43 was observed (3.1 ± 0.4 for vehicle treated cells versus 9.3 ± 0.6 for 0.1 mg/ml PBA-treated cells at 5 days in relative density units, p<0.01).</p><!><p>We previously demonstrated that 0.1 mg/ml PBA inhibited the growth of WB-ras1 cells and that PBA-Me was inhibitory at ~10-fold lower concentration [Sunman et al., 2004]. As seen in Figure 7A, PBA at 0.1 mg/ml also inhibited the growth of H2009 cells over 14 days. Growth was significantly inhibited as early as 2 days of treatment (day 3 of growth). PBA also increased apoptosis in these cultures treated for 2 d with 0.1 mg/ml from 5.5 ± 0.5 % apoptotic cells in control dishes to 8.1 ± 0.6 % apoptotic cells in PBA-treated dishes (p<0.05). SB203580 alone also inhibited growth of WB-ras1 cells (Figure 7B) and H2009 cells (not shown) and was therefore not efficacious in preventing PBA's inhibitory effect on cell growth.</p><!><p>Treatment of WB-ras1 cells with 0.1 mg/ml PBA for 48 h decreased phosphorylation of JNK on key activation sites (Thr183/Tyr185) by ~2 fold (Figure 8A, B). PBA (0.1 mg/ml PBA for 48 h) similarly decreased JNK phosphorylation in H2009 human lung tumor cells (Figure 8C, D) No significant effect on JNK Thr183/Tyr185 phosphorylation was seen in WB-ras1 or H2009 cells treated for 4 h (not shown).</p><!><p>We previously demonstrated that PBA had no effect on activation of the p42/44 MAPK pathway after 4 h, 48 h, or 5 days treatment [Sunman et al., 2004]. To further test the specificity of PBA for the p38 MAPK pathway, we monitored activation of other signaling pathways. PBA had no effect on the activation of Akt, cdc2, CDK2, and PKC (Figure 9). Densitometric evaluations revealed no differences between controls and PBA-treated cells (p > 0.1; data not shown).</p><!><p>The results presented above demonstrate that PBA and PBA-Me increased phosphorylation of p38 MAPK on Thr180/Tyr182 in both ras-transformed epithelial cells and H2009 human lung carcinoma cells. Furthermore, PBA increased phosphorylation on activation sites of HSP-27, MAPKAPK-2, and ATF-2, which are downstream effectors of p38 MAPK. Increased phosphorylation of p38 MAPK on Thr180/Tyr182 and increased phosphorylation of downstream effectors HSP-27, MAPKAPK-2, and ATF-2 are known indicators of activation of the p38 MAPK signaling pathway [Raingeaud et al., 1995; Cobb et al, 1995; Rouse et al., 1994]. However, in our experiments, increased activation of p38 MAPK and downstream effectors was only observed with treatment durations of greater then 4 h (Figure 1 I). The time-course of these effects suggests that PBA did not directly activate the prototypic p38 MAPK signaling cascade since this would have occurred within minutes. We therefore hypothesize that PBA and PBA-Me activate p38 MAPK in ras-transformed and human lung carcinoma cells by an alternative mechanism. In support of this, we found no effect of PBA on activation of MKK3/MKK6, which are upstream kinases that phosphorylate p38 MAPK [Enslen et al., 1998; Brancho et al., 2003], or MLK-3, an upstream kinase of MKK3/MKK6 in the p38 MAPK signaling cascade (unpublished observations). However, MKK4 was also shown to be capable of activating p38 MAPK in vivo [Brancho et al., 2003; Dhillon et al., 2007]. In addition, a MAPKK-independent mechanism of p38 MAPK activation has been reported that involves the TAK1 binding protein, TAB1, [Ge et al., 2002]. Alternatively, PBA-induced inhibition of a p38 MAPK phosphatase, Wip1, [Bulavin et al., 2004] that acts on Thr 180 and/or Tyr 182 could result in the observed enhanced p38 MAPK activation. Other possible mechanisms include metabolic, structural, or gene expression changes in the cells that alter p38 MAPK activation over time which would correlate with PBA-enhanced phosphorylation of p38 MAPK (Figure 1 E and F) and effects on cell growth that occurred over 2–14 d (Fig 7A). PBA also decreased phosphorylation of JNK, which correlated in time with p38 MAPK activation. This suggests a link between these two pathways following PBA treatment, that may occur via crosstalk mechanisms proposed by Wagner and Nebreda [2009].</p><p>There are four isoforms of p38 MAPK: alpha, beta, gamma, and delta [Kumar et al, 2003]. HSP-27 activation can be mediated by the alpha or beta isoforms whereas all four isoforms can activate ATF-2. Our data suggest the alpha and beta isoforms are affected by PBA and PBA-Me, but the results do not allow a definitive answer.</p><p>Specificity of PBA for the two related stress-activated MAPK signaling pathways, p38 MAPK and JNK is demonstrated by its lack of effect on activation of enzymes in other key signaling pathways, as shown in Figure 9. PBA also showed a higher degree of activation of p38 MAPK in WB-ras1 cells compared to the plasmid control WB-neo3 cells (Figure 1), which correlates with its greater growth inhibition in WB-ras1 cells versus WB-neo3 cells [Sunman et al., 2004]. The higher level of endogenous p38 MAPK activation in WB-neo3 cells compared to WB-ras1 may explain why PBA is less effective in up-regulating p38 MAPK in the non-transformed cells compared to the ras-transformed cells.</p><p>Our data demonstrate an approximately 3-fold increase in p38 MAPK phosphorylation at concentrations of PBA or PBA-Me that caused significant decreases in cell growth [Sunman et al., 2004] and Figure 7. While the Western blot signal for phospho-p38 MAPK cannot be correlated directly with kinase activity, it suggests that a moderate change in phosphorylation of p38 MAPK correlates with a large reduction in neoplastic cell growth. Timofeev et al. [2005] also noted suppression of in vivo tumorigenesis that was related to modest changes in p38 MAPK activity. HSP-27 and MAPKAPK-2 activation by PBA, on the other hand, were as high as 6-fold and 8-fold greater than controls, respectively, suggesting amplification of downstream signals or differences in the balance of upstream kinases and phosphatases acting on p38 MAPK compared to its downstream effectors. Furthermore, PBA-enhanced phosphorylation of p38 MAPK (Figure 1 E and F) and its effects on cell growth were observed as long as 14 d following treatment (Fig 7A). This suggests that PBA and PBA-Me may be effective anti-tumor agents with a long duration of action despite the lack of more dramatic changes in activation of p38 MAPK. The approximately 2-fold increase in apoptosis elicited by PBA suggests cell death contributes to this growth inhibition.</p><p>Reduced gap junction-mediated cell-cell communication, as seen in WB-ras1 cells [De Feijter et al., 1990], is a phenotypic characteristic of many neoplastic cells that allows them to avoid the growth regulatory influences of adjacent cells [Yamasaki and Naus, 1996; Trosko and Ruch, 1998]. Restoration of gap junction-mediated communication in such cells often decreases their growth and tumorigenicity. The results of our present and previous experiments [Sunman et al., 2004] demonstrate that PBA strongly increases gap junction-mediated communication between WB-ras1 cells. A key finding of our present study is that treatment of cells with the p38 MAPK specific inhibitor, SB203580, prevented this enhanced cell-cell communication and also reduced connexin43 P2 formation (Figure 6). The P2 phosphoform has been associated with high level gap junction communication and the occurrence of large gap junction plaques in WB cells [Musil and Goodenough, 1991; Matesic et al, 1994]. These results suggest PBA stimulates p38 MAPK or a downstream effector to phosphorylate connexin43 to the P2 phosphoform which increases gap junction-mediated communication. Whether p38 MAPK can directly phosphorylate connexin 43 in WB-ras1 cells, as seen in other cells [Lee et al, 2004; Ogawa et al., 2004], remains to be determined. While H2009 cells are also deficient in cell-cell communication, PBA did not increase cell-cell communication in these cells at 48 h or 5 d treatments. This correlated with a lack of increased connexin 43 P2 phosphorylation and suggests gap junction regulation and connexin 43 phosphorylation pathways are different in H2009 and WB-ras1 cells. Additionally, H2009 cells do not appear to be good indicators of SB203580 actions on cell-cell communication.</p><p>Lee et al. [2004] reported that SB203580 treatment increased connexin 43 phosphorylation in similar H-ras-transformed rat liver epithelial cells and this was correlated with increased gap junction communication. This result conflicts with our observation that SB203580 prevented PBA-enhanced cell-cell communication and P2 phosphorylation in WB-ras1 cells. However, Lee et al. used substantially higher concentrations (5–20 µM) of the inhibitor and shorter incubation times (1 h), which may account for the differences in results.</p><p>Inhibition of p38 MAPK, and concomitant activation of JNK, play a critical role in ras-induced transformation that is independent of Raf activation [Pruitt et al., 2002]. Conversely, activation of p38 MAPK resulted in cancer cell cycle inhibition or apoptosis initiated by retinoids, cisplatin, and other chemotherapeutic agents [Losa et al., 2003; Olson and Hallahan, 2004; Iyoda et al., 2003]. In addition, Iyoda et al. [2003] reported that increased p38 MAPK activity in hepatocarcinoma cells transfected with a MKK6 mutant gene decreased the growth of these cells. Hui, et al. [2007] demonstrated that in chemical-induced liver cancer development, mice carrying a liver-specific deletion of p38α showed enhanced hepatocyte proliferation and tumor growth that correlated with activation of the JNK signaling pathway, and provided further evidence suggesting that p38α may suppress cancer cell proliferation by antagonizing the JNK pathway. Our data demonstrate that PBA and PBA-Me activate p38 MAPK, suppress neoplastic cell growth, inhibit JNK, and enhance gap junction communication and thus may be a similarly effective therapeutic agent for cancers with down-regulated p38 MAPK and/or over-activated JNK. The increased gap junction communication may in turn enhance other forms of cancer therapy including bystander cytotoxicity of radiation and chemotherapeutic agents [Prise and O'Sullivan, 2009].</p>

PubMed Author Manuscript

Unzipping Natural Products: Improved Natural Product Structure Predictions by Ensemble Modeling and Fingerprint Matching

The Omics revolution, powered by computational biology, has transformed the natural products landscape. In part due to logistic advantages (high diversity, small genome sizes, ease of isolation) and available automation for annotation, large numbers of bacterial genomes now populate public databases. Several programs have been developed to group similar sequences and their products into clusters. When expressed, these genomes produce small molecule products with a common core. This then has led to a drive for reliable and predictive in silico-based mining approach for finding new putative structures. While these programs are great at identifying the complex genetic pathways encoding natural products, there remains a gap in translating genetic messages to the products (predict small molecule structures). The understanding of the rules around polyketides synthethases (PKSs) and nonribosomal peptide-synthethases (NRPSs) do allow software to predict the small molecule core scaffolds to an extent, but not any of the currently available programs in this space predict any tailoring modifications.

unzipping_natural_products:_improved_natural_product_structure_predictions_by_ensemble_modeling_and_

<!>Methods:<!>Figure 2 -The deconstruction of the molecule of interest to a scientist is subject to a recursive application of graph editing. All edits are applied to the desired molecule which creates an ensemble. This ensemble is then expanded by repeatedly subjecting it to the same reactions which generate additional potential untailored precursors. Through repeated, recursive, application, our method should create all potential untailored precursors. In our scripts, encoded reaction SMARTS are used with the RDKIT to cut ring bonds, remove ring substitutions, change atom types. The result is an ensemble of molecules which may contain a structure similar to one predicted by Anti-SMASH<!>Figure 3 -The deconstruction of Barbamide by a judicious application of recursive editing of the molecule produced an ensemble<!>shown at the top) from the expanded set (B) associated with the desired compound (A). The initial filtering would represent atoms at an untyped atom "A" and then subsequent scoring would include explicit atom types.<!>Table 1 -List of public domain natural compounds used to calibrate the methodology<!>Discussion<!>Table 2 -Results returned from a representative set of public-domain molecules used to demonstrate our algorithm. The "Sought<!>Limitations and Future Directions<!>Conclusion

<p>Drug discovery encourages exploration of vast molecular space 1 to discover new molecules to treat disease. As synthetic biology 2 can provide an efficient mechanism to produce drug-like derivatives of natural products, 34 it is natural that a drug discovery scientist would want to harness or control this science. To accomplish this, the practicing scientist must know how to properly deduce the target molecule from a primary genetic sequence. While the number of identified biosynthetic gene clusters (BGC) -which control primary assembly of core structures -numbers in the thousands, many tailoring reactions remain unknown. Researchers detect gene clusters utilizing sequencing data and comparative genomics between evolutionary-conserved sequences. These methods find groups of genes on chromosomes linked by common attributes, and, as a result of the unknown tailoring reactions, define only core products of molecular assembly.</p><p>Programs such as ClustScan 5 , Anti-SMASH 6 , GARLIC/GRAPE 7 , and ClusterFinder 8 exist to facilitate this prediction, confounding side pathways resulting from tailoring reactions, along with lack of overall genomic data, often interfere with predictive accuracy. To further complicate the problem, transferring a set of natural product assembly gene modules into another organism often yields different final products. For example, in erythromycin, 16 different tailoring pathways result in a diverse set of final structures. 9 Prediction of final chemistry product structures struggles with the variety of post-assembly modification which results from tailoring reactions.</p><p>Figure 1 -Prediction of the barbamide core scaffold 10 (left) from anti-SMASH 3.0 omits expected tailoring reactions, such as thiazole condensation, oxidation, elimination, and methylation. At right, the actual final structure of Barbamide as produced by the expressed Biosynthetic Gene Cluster.</p><p>Chemists utilizing synthetic biology must, therefore, understand the interplay of microbial host, primary sequence, tailoring reactions, and available starting synthons to predict products accurately. With thousands of biosynthetic clusters producing core structures and a yet-undefined suite of tailoring reactions, an end-to-end model which predicts all final products for all clusters will be a significant undertaking requiring complete characterization of all reactions and 10,000+ representative compounds.</p><p>Our approach attempts to be simple, general and fast while enabling a search not only from a putative small product from a cluster but more importantly for drug discovery, a search for a putative genomics source (cluster) which will yield a novel small molecule with a defined substructure. We rely on Anti-SMASH for core prediction and implement a reverse-skip of the tailoring reactions to bypass the final product desired by a chemist to match a related untailored compound to a possible cluster product from Anti-SMASH. We achieve this "skip" through a simple exhaustive enumeration of generic types of bond breaking and bond addition. Our insight was to create an ensemble of possible structures to reverse search the cores output from Anti-SMASH, using the pregenerated and pre-fingerprinted output as a fast way to link a chemist's desired compound to a cluster worth testing.</p><p>Many common reactions expected during tailoring include methylation, acylation, glycosylation, oxidation, cyclization, condensation, and other reactions where the mechanism is relatively well understood and readily generalized. We, therefore, disclose a model to "unzip" natural products until they match a putative antiSMASH product, thus matching them with this reverseskip of the final product to downstream isolates. In short: we take an input structure and create an ensemble of intermediate products before the application of tailoring reactions. These intermediates are then used to rank order the putative BGCs. The result is a fast and interactive method for taking a chemist from the idea of a compound to the known core molecules.</p><!><p>Anti-SMASH 3.0 11 predicts untailored small molecule products of secondary metabolism. To suggest clusters which when expressed might yield a desired finished molecule, we needed a method which could relate the predicted unfinished molecules to a final, tailored target molecule.</p><p>We thus needed to either finish the anti-SMASH predicted molecules and/or identify pre-tailored precursors to a molecule of interest. As the identification and cataloging of finishing reactions is an active object of study, 12 there is no simple way to know which reaction nature has encoded for each cluster. As the second strategy of finding putative precursors can be implemented comprehensively and is simple to encode (but computationally more intensive), we chose the latter unzipping method based on the manipulation of the compound of interest. We encoded reactions which fragment the molecule according to a generalization of a reversed set of finishing reactions. The application of this molecule deconstruction resulted in an ensemble of (40-50) structures which we could use to identify clusters containing a compound similar to at least one compound of the unzipped ensemble.</p><p>The method, as currently construed, is intended to be a proof-of-concept: fast enough to run on more than 2,000 molecules, and readily able to scale to cluster / parallelized Cloud processes as desired. A requirement was that the prediction for one compound should require no more than 15 seconds on one CPU.</p><p>To fragment the molecules, encoded reactions were run recursively. That is, each unzipping reaction was run on each product from the previous reaction application; a product of any reaction is both put in the ensemble of potential untailored compounds AND a copy of it subjected to all encoded reactions. The expansion of a compound into an ensemble is run until no more unique compounds are produced. Thus, we explored all possible combination of reactions enacted at all possible sites to produce the ensemble of molecules. From this enumerated library, we generated an initial ensemble. From this set, small fragments were discarded, and the molecule was then standardized and validated as a chemically viable molecule using standard valence rules. In this version of the algorithm, we support the following reaction: cyclization, methylation, acylation, oxidation, and glycosylation. Each reaction is performed as followed. For cyclization, each ring bond is cut (as in Figure 2 below for the thiazole.) For methylation: methyl groups and heteroatoms were removed from rings. For acylation, oxygens associated with rings were deleted. For oxidation, hydroxyl groups were converted to carbonyl groups. For glycosylation, glycan bonds were broken. Other reaction classes are remarked upon in the Discussion section (vide infra).</p><!><p>Thus, for Barbamide (Figure 3, compound A), the reference ensemble becomes dozens of variations formed by cutting bonds and replacing atoms (B). The predicted structures from Anti-SMASH are given as queries where most atoms are specified while some (connection points) are represented as untyped atoms. An initial filtering of compounds which may match Anti-SMASH predictions is made by counting the atoms which overlap of bonding patterns of Anti-SMASH predictions. Subsequent steps, described below, use Morgan fingerprint 13 typing.</p><!><p>which contains a molecule which matched the untailored product predicted by Anti-SMASH 3.0. This editing includes the breaking of bonds within rings in addition to other listed transmutations. This ensemble (B) is first 'search' for a compatible framework by the topological description of the Anti-SMASH predictions(C). Thus, a scientist wishing to identify the Barbamide producing cluster would draw "A" and the system would loop to create an ensemble of untailed compounds (B). These compounds would be 'reverse searched' by predictions from AntiSMASH. Thus, the antiSMASH prediction (shown at the bottom of (C) would identify an untailed compound (</p><!><p>Molecules we identified via fingerprint similarity were then compared to the predicted structures using a Maximum Common Substructure (MCS) metric calculated using FMCS 14 . Clusters are advanced in the search if the number of atoms in common was greater than 80% for molecules with a molecular weight greater than 125 Dalton, to filter out a prohibitively large amount of small aromatic rings and fragments that would not correspond to the natural product of interest.</p><p>The method was qualitatively calibrated with 12 compounds of interest with known clusters (Table 1). These compounds were chosen to inhabit a wide range of molecular weights and possess a variety of types and numbers of rings. With the method outlined here, we were able to consistently perform a reverse search with the 6,881 structures and many internally expressed clusters in less than a minute on a typical modern workstation.</p><!><p>This ensemble creation can be thought of like the reverse of applying tailoring modifications to anti-SMASH predictions. Once putative BGCs are ranked, it is useful to try and identify the necessary tailoring reactions required for transforming the predicted structures into the final product. To accomplish this from the subset of ranked BGCs, we take the forward approach, using a catalog of tailoring modifications, to try and identify the necessary chemical routes to the final product (see Figure 3). These modifications are encoded with Reaction SMILES and tailoring reactions such as chain elongation, acyclic control for regiospecific cyclization reactions from the broad and general reaction is then available for electronic post-assembly modification of specific gene clusters. 15 For example (see Figure 4), the oxidative and O-methylation-tailoring of pre-rapamycin will now yield the known-structure of the post-assembly product, rapamycin. "Cuts" in each molecule to determine what constitutes a fragment follow specific rules, such as recognition of common functional group motifs (C=OOH, CH3, etc.) and ensuring only valid, covalent bonds are cut (no radical fragmentation). The application was integrated with anti-Smash 3.0 and published as a RESTful web service. The code was implemented in the graphical prototyping environment KNIME which also served as the UI for delivering the search forms to the users. The chemical informatics RDKit package was accessed both through dedicated KNIME nodes and by use of Python within a KNIME Python node. The Python code is provided as an attachment. 16 In this production code for our search tool, chemical sketches of structures of interest (Figure 5) become one of an ensemble of possible untailored compound (such as #1, #2, #3) which are linked (using pattern fingerprint similarity methods) to gene clusters via the Anti-SMASH predicted core structures. Hence, knowledge of the tailoring reactions, and precisely how to reverse them, allows enumeration of potential cores which can be used to identify the set which will include the gene cluster. By applying simple reversesynthetic operations, we unfold the tailoring reactions into something more similar to what might be expected from the assembly stage.</p><p>The existing Anti-SMASH model predicts structural cores from genetic sequences as shown below (with a neural network annotated as a hidden Markov model. We will implement a fingerprint-based indexing method for transformations. This fingerprint will initially have independent bits for the curated and cataloged reaction. The fingerprints will use an index of existing clusters (to facilitate cluster look-up and similarity) and then be extended to the prediction of clusters from sequences. This approach mimics the development and use of chemical fingerprints such as MACC keys or Daylight fingerprints. The chemical structure fingerprints which were created for the database look-up are then used in combination with information theory and data analysis technique to build similarity and then predictive models.</p><!><p>Searching for gene clusters to build new molecules leads to a significant limitation that is not entirely solved by simple molecule similarity searching. Namely, the compound predictions of known gene clusters do not match known metabolic products. For example, the predicted structure of coelibactin 17 is not, in effect, coelibactin. Anti-SMASH attributes this with the notification "tailoring reactions not taken into account." To overcome this, we developed a brute-force method to attempt to make chemists' queries look more like the predictions of anti-SMASH. The current method has a rudimentary knowledge of a few tailoring reactions that are used to retro-synthetically modify the user's query into the core molecule assembly as predicted by Anti-SMASH. In effect, this reverses some known tailoring reactions to search Anti-SMASH for relevant gene clusters.</p><p>Compounds such as coelibactin (below) are rationally and programmatically decomposed to remove rings, methyl groups, and rich moieties such as -CCl 3 . This method creates an ensemble of structures that are then matched against predicted and known geneclusters products of Anti-SMASH and increases the likelihood of finding the assembly pre-modification (i.e., before the tailoring reactions are applied). The example below shows the unmodified known product (barbamide) and the matching Anti-SMASH core that was identified after performing the retrosynthetic analysis: Developing a general predictive model for tailoring reactions in a specific gene cluster will not be straightforward. 18 However, we believe it cannot be done without knowledge both of tailoring reactions and how to apply and rank them from a scientific standpoint. We plan to separate the application of these tailoring reactions from the creation of a searchable index for reactions.</p><p>Applying our brute-force post-assembly reactions to a core allows proper ranking of known tailored products predicted. Note that the predicted core structure of both ranked gene-clusters is different from the final product, the bacterial siderophore coelibactin (Figure 6)</p><p>Figure 6 -Specifying the coelibactin final molecule leads back to the corresponding aS 3.0 gene cluster, through ensemble-modeled detection Since we will often know the organism (and its sequence) that produced the molecule, which means that we will have a limited number of predicted Anti-SMASH molecules. For the sake of computational efficiency, one could imagine that we could have performed all possible tailoring reactions to all Anti-SMASH results pre forma, and then matching all results against the desired molecule of interest. We have chosen, instead, to take a computationally lighter approach, as we feel it grants the user a right balance of speed and accuracy without the need to cache or compute billions of reaction pathways each time a search is run. Our speed objective, which we met, was to have the system implemented on one CPU to respond with predictions within 60-90 seconds.</p><!><p>Compound" is the common name of a structure that a chemist might want to synthesize with a biosynthetic cluster. This "Sought Structure" is an antibiotic which is likely to exist in the public database underlying Anti-SMASH. The "Anti-SMASH Core Predicted" is the (pre-tailored) product of Anti-SMASH BGC which predicted by Anti-SMASH and suggested by this method. The untailored version of the sought structure (which was used to link to the BGC) is the "Untailored Sought Structure". The last column ("Anti-SMASH Most Similar Core") shows what the user would have received if the method did not exist AND instead they were using a fingerprint method and Tanimoto index to look for a cluster.</p><p>Table 3 -The method only gives a prediction when an expanded structure is within a Tanimoto similarity range and contains a number of atoms in an MCSS between the expanded structure and the core from an Anti-SMASH cluster. By examination of the MCSS of the 'best' fitting structures (that were well below the Tanimoto cut-off), we see that small fragments (shown on the right above) characterize structures which could not and would not be used to identify an Anti-SMASH cluster</p><p>The method has been calibrated with antibiotics which are annotated in the public database which were regressed to yield cut-offs for similarity and a minimum MCCS fit. As part of our analysis, we ran a large sample (1,498) of naturally-produced compounds to see if the method scaled and did not yield false predictions. We used 200 compounds from the PRISM paper 19 and 1,298 diverse natural products. 96.8% did not yield predictions above the cut-off. This does not mean that we missed identifying clusters -but rather than the public database is yet to contain the far majority of the clusters for known natural products. Using a Tanimoto-based similarity comparison of the compounds being sought and the cores in AntiSMASH, we would have identified 11 cores. The use of this method increased the cores found by 55 -for a total of 66. Thus, the method, as primitive as it is, was useful in providing a 6x enrichment (Figure 7).</p><p>Compound similar to Anti-SMASH predicted core Expanded compound similar to Anti-SMASH predicted core The use of expanded compounds in this method can guide you to an AntiSMASH core you would not find by eye or with a Tanimotobased search. This is best understood by comparing structures being sought to the predicted cores which were identified via the 'expanded compounds' created by the method. From Figure 8, it is seen that the 'expanded compounds' introduced a reverse tailored structure which is closer or identical to the untailored Anti-SMASH predicted cores.</p><!><p>The method outlined above (Figure 8), while applicable to PKS assemblies and small, polycyclic products, still suffers from some significant limitations. Many anti-SMASH gene clusters do not produce a Molfile of a predicted structure, either due to frame-shifts, calling errors, or just unknown function in 1 or more clusters. "R-groups" and other Markush wildcards, used in SMARTS to deal with structural uncertainty, cannot be replaced if representing >50% of the molecule. Tailoring reactions which ablate large portions of the precursor assembly product -especially large leaving groups like indole, phenyl, or coumarin -are especially hard to predict. We do not always capture redox events, as those absent a specific transferase or cytochrome may be overlooked. Importantly, the method will break, but not form, rings, and it does not add chemical moieties from cross-domain reactions or condensations. Thus, the method will be most successful on molecules which have similar frameworks to those known to Anti-SMASH.</p><p>We envision a refinement where the final filtering of clusters would be possible with a bolt-on prediction based on emerging knowledge of the full tailoring catalog. Another potential enhancement could be tracing the lineage of the reactions to obtain the closest match. If we associated each of these reactions with a particular enzyme, it would be trivial. However, this would most likely require additional development time and annotation that does not readily exist in biosynthetic databases today. Though we have identified condensation reactions as an important class to support, we have not developed them for this publication. As the code will be open-source, other researchers could assist in these modifications at a later date.</p><!><p>In conclusion, we present an algorithmic improvement to predicted Anti-SMASH structures, in which structures derived from gene clusters can be modified by predicted tailoring reactions to derive the desired target molecule. Future enhancements may be able to upgrade the prediction, such that we can begin to predict unrecognized molecules based on related strains and homology of gene clusters. We could then generate a reaction aware fingerprint that can be used to search a core scaffold against the ensemble of possible products given the known tailoring reactions.</p>

ChemRxiv

An in-silico study on selected organosulfur compounds as potential drugs for SARS-CoV-2 infection via binding multiple drug targets

The emerging paradigm shift from 'one molecule, one target, for one disease' towards 'multi-targeted small molecules' has paved an ingenious pathway in drug discovery in recent years. This idea has been extracted for the investigation of competent drug molecules for the unprecedented COVID-19 pandemic which became the greatest global health crisis now. Perceiving the importance of organosulfur compounds against SARS-CoV-2 from the drugs under clinical trials, a class of organosulfur compounds effective against SARS-CoV were selected and studied the interaction with multiple proteins of the SARS-CoV-2. One compound displayed inhibition against five proteins (both structural and non-structural) of the virus namely, main protease, papain-like protease, spike protein, helicase and RNA dependent RNA polymerase. Consequently, this compound emanates as a potential candidate for treating the virulent disease. The pharmacokinetics, ADMET properties and target prediction studies carried out in this work further inflamed the versatility of the compound and urge to execute in-vitro and in-vivo analysis on SARS-CoV-2 in the future.

an_in-silico_study_on_selected_organosulfur_compounds_as_potential_drugs_for_sars-cov-2_infection_vi

Introduction<!>Ligand preparation<!>Molecular docking<!>Physicochemical properties<!>ADMET studies<!>Molecular target prediction<!>Results and Discussion<!>Docking studies of the organosulfur compounds with the SARS-CoV-2 Mpro<!>Docking studies of the organosulfur compounds with the SARS-CoV-2 PLpro<!>Docking studies of the organosulfur compounds with the SARS-CoV-2 Spike protein (Spro)<!>Docking studies of the organosulfur compound with the SARS-CoV Helicase<!>Docking studies of the organosulfur compounds with the SARS-CoV-2 RdRp<!>Physicochemical properties study based on the Lipinski's rule<!>Prediction of the absorption, distribution, metabolism, excretion, and toxicity (ADMET) profile<!>Identification of target class for compound 1 via target prediction studies<!>Conclusion

<p>The deplorable situation of the present world aroused by the dreadful behavior of an RNA virus named the Severe Acute Respiratory Syndrome Corona Virus-2 (SARS-CoV-2) is originated in the City of Wuhan, China in late 2019. The Corona Virus Disease (COVID- 19) pandemic caused by the novel coronavirus (later named as SARS-CoV-2) massacred about four lakhs lives leaving more than eight million people in the infection with fever, dry cough, short breath and other respiratory ailments across 215 countries. Similar infections were reported in 2012 by the Middle East Respiratory Syndrome Corona Virus (MERS-CoV) and the Severe Acute Respiratory Syndrome Corona Virus (SARS-CoV) in 2003 but they were less contagious with 10,590 cases and 1632 fatalities collectively 1 . According to the World Health Organization (WHO), there are no approved medicines or vaccines for COVID-19 as of now while extensive researches are undergoing to explore the treatment proceeding including the clinical trials of more than 300 compounds 2 . The leading approach for the development of curative medication is drug repurposing as it allows for the rapid acceptance with a profit of low cost, known and optimized synthetic route and often facile to leapfrog the preliminary stages of clinical trial 3 .</p><p>In hand the list of various known drugs for repurposing, the promising approach is to go down the line of SARS-CoV drugs. The phylogenetic analysis of SARS-CoV-2 revealed about 89.1% genomic similarity with SARS-CoV which is also a beta corona virus 4 . Closely scrutinizing the proteins involved in the SARS-CoV-2 infection, the Spike Protein (Spro) promotes the entry of the virus into the human cell by binding with the type 1 transmembrane metallocarboxypeptidase known as Angiotensin Converting Enzyme-2 (ACE-2). The receptor binding motif of SARS-CoV and SARS-CoV-2 spike proteins are the same and they possess the sequence similarity of 76% by showcasing more similar adherence in the receptor binding domain 5 . Spro is a structural protein located on the periphery of the virus and other main structural proteins are envelope protein, membrane proteins and nucleocapsid protein. The nonstructural proteins are more in number, sixteen, and they are responsible for the viral multiplication and other specific purposes for infection. Once inside the cell, the viruses commence the synthesis of its RNA by the enzyme called RNA-dependent RNA polymerase (RdRp). Knowing the fact that SARS-CoV and SARS-CoV-2 share RdRp sequence with about 96% similarity, the inhibition of this protein is a prospective strategy of drug action 6 . The other promising drug targets are Chymotrypsin Like Protease (3CLpro) otherwise called Main Protease (Mpro) and Papain Like Protease (PLpro) which help in virus replication. These two protease enzymes of the SARS-CoV-2 exhibit 96% and 83% percentage similarity respectively with that of SARS-CoV with similar active site sequences 7,8 . Helicase is another target for antiviral drugs as it plays a vital role in replication and the central dogma of the virus 9 . Spotlighting the structural similarity between SARS-CoV and SARS-CoV-2, we selected organosulfur compounds as drug candidates against SARS-CoV-2 which were previously found to be effective against SARS-CoV infection.</p><p>Organosulfur compounds are an important class of molecules with the sulfur-containing functional groups such as sulfones, sulfonamides, disulfides, sulfoxides, thiophene, thiazole etc 10 . The impact of organosulfur compounds in the pharmaceutical sector is impeccable right from the example of penicillin. Clinical trial of a range of organosulfur compounds such as Ritonavir, arbidol 11 , baricitinib 12 etc. is currently underway against SARS-CoV-2. Given the potential of organosulfur compounds as an antiviral drug, we selected eight organosulfur compounds (Figure 1) which are already reported to show antiviral activity against SARS-CoV, a very close analogue of SARS-CoV-2, and studied the inhibitory action against several druggable targets of SARS-CoV-2 to investigate the possibility of multiple targets binding of the selected candidates. As the mutation rate and thus evolution rate is more for RNA viruses 13 , multiple target binding increases the efficiency of the drug by reducing the effect of viral resistance against one protein 14 . Therefore, the molecular docking study of each compound is carried out with five different target proteins. The ADMET properties, target prediction and Lipinski's rule are also predicted for the selected compounds to explore more about the pharmacokinetics and druggability.</p><!><p>The structure of all the organosulfur compounds was drawn in ChemDraw and the 3D structure was generated by UCSF Chimera 15 from the SMILE string. The structures of all the reference compounds were obtained in UCSF Chimera through their PubChem ID. All the structures were energy minimized through the same software and converted the PDB structure to PDBQT format by using AutoDock Tools.</p><!><p>Molecular docking study was carried out by using AutoDock Vina 16 to explore the binding affinity and the involved interactions in between all eight organosulfur compounds and the five druggable protein targets of SARS-CoV-2 namely Main proteases (Mpro, Chain A), Papainlike proteases (PLpro, Chain A), Spike-protein (Spro, Chain B), Helicase protein, RNA dependent RNA polymerase (RdRp). The crystal structure of Mpro (PDB ID: 6Y84), PLpro (PDB ID: 6W9C), Spro (PDB ID: 6LZG), RdRp (PDB ID: 6M71) and helicase (PDB ID: 6JYT) were retrieved from the protein databank (http://www.rcsb.org) 17 . The hydrogen atoms and gasteiger charges were added to each protein, subsequently, all the proteins were saved in PDBQT format by using the AutoDock v4.2 program 18 . For Mpro protein grid box (30 Å × 30 Å × 30 Å) centered at (X12, Y-8, Z20 Å), for PLpro grid box (30 Å × 30 Å × 30 Å) centered at (X-42, Y29, Z30 Å), SPro grid box (30 Å × 30 Å × 30 Å) centered at (X-36, Y33, Z12 Å), RdRp grid box (30 Å × 30 Å × 30 Å) centered at (X120, Y122, Z127 Å at 0.375 Å spacing) and for helicase protein grid box (42 Å x 30 Å x 86 Å) centered at (X424, Y29, Z25 Å at 0.375 Å spacing) were prepared and saved the output grid file in txt format. A docking run was given from the command prompt. Best docked conformation and minimum binding energy were considered for further analysis. UCSF chimera was used for the visualization of the docked conformation and results. The results and 2D interaction plots were analyzed by using Discovery studio visualizer 19 .</p><!><p>The physicochemical properties according to Lipinski's rule were calculated for all the selected organosulfur compounds to predict the pharmacokinetics property. SwissADME tool was used to calculate the properties from the SMILES structures of each compound. (http://www.swissadme.ch/) 20 .</p><!><p>Predicting in-silico pharmacokinetic properties of a new drug is very crucial for further studies. ADMET (Absorption, Distribution, Metabolism, Excretion and Toxicity) prediction provides some important information for new compounds. ADMET studies have been carried out by using the computational pkCSM tool (http://biosig.unimelb.edu.au/pkcsm/prediction) 21 .</p><!><p>For the validation of targets, we used molecular target studies by using the Swiss Target Prediction tool (http://www.swisstargetprediction.ch/) 22 which is a web server that predicts the putative targets of the given molecule by utilizing 2D and 3D similarity index with known ligands. The smile formats of the compounds were entered to obtain the targets.</p><!><p>In the current study, a set of organosulfur compounds known for targeting SARS-CoV Mpro were selected and to carry out the molecular docking studies to assess their potential against SARS-CoV-2. To examine the possibility of binding with multiple targets, we selected five SARS-CoV-2 proteins namely Mpro, PLpro, Spro, RdRp and helicase, and docked with the selected compounds along with their known inhibitors as the reference compounds such as indinavir for Mpro, darunavir for PLpro, arbidol for Spro, remdesivir for RdRp and ivermectin for helicase. Utilizing molecular docking study, the binding energy of the organosulfur compounds with the reference drug compounds is calculated and the results are tabulated in Table 1. Ivermectin -----8.5</p><!><p>Molecular docking study of the organosulfur compounds with the Mpro of SARS-CoV-2 exhibited promising results with several of them. Indinavir, a well-known drug that has already been reported to inhibit Mpro of the SARS-CoV-2 23 was studied as a reference compound. Docking score of the organosulfur compounds along with the reference compound is tabulated in Table 1 and docking conformations of the reference compound and the organosulfur compound with the highest binding affinity i.e. 1 is represented in Figure 2. All the compounds were found to bind in the active site of Mpro where Cys145 and His41 are the catalytic residues. Indinavir docked with a conformation that makes five hydrogen bonds with Thr26, Ser46, Asn142, Cys145 and Gln189 along with pi-alkyl interaction with Pro168, van der Waals interaction with His41 and other residues (see Figure 2b). When the lowest binding energy for the reference compound was observed to be -7.7 kcal/mol, the organosulfur compound, 1 docked with a minimum binding energy of -8.6 kcal/mol with three conventional hydrogen bonds with His41 (catalytic residue), Cys44 and Gly143. The interactions include two pi-sulfur interactions with Met49 and Met165; eight van der Waals interaction with Thr25, Thr45, Ser46, Leu141, Ser144, His164, Glu166 and Gln189; carbon-hydrogen bond with Asn142 and Cys145 which is also an active site residue (Figure 2d). The compound 6 binds with the binding energy -7.6 kcal/mol which is close to the binding energy of the reference compound with conventional hydrogen bond with both the active site residues His41 and Cys145 along with other van der Waals interaction. A pi-sulfur interaction is also observed with His41 making the binding stronger (Figure S1b in the Supplementary Information). Thus, 6 claims to be a fairly good candidate to inhibit Mpro of SARS-CoV-2. 3 docked with Mpro with a binding energy -7.3 kcal/mol even though hydrogen bonding with the catalytic dyad is absent. Nevertheless, 3 forms a pi-pi stacking interaction with His41, pi donor hydrogen bond with Cys145, conventional hydrogen bond with Gly143, pi-sulfur bond with Met165, pi-sigma bond with Glu166 and van der Waals interaction with other residues (Figure S1d in the Supplementary Information). Based on these observations, 1, 6 and 3 can be a potential drug against the SARS-CoV-2 acting via Mpro inhibition.</p><!><p>The molecular docking study of organosulfur compounds with the SARS-CoV-2 PLpro revealed that the three compounds, namely 1, 6 and 7 exhibited reasonably strong interaction with binding energy in the range of the reference compound darunavir which is a known organosulfur drug for SARS-CoV-2 8 (Table 1). Darunavir docked into the binding pocket of the protein with catalytic triad Cys111, His272 and Asp286. Darunavir bound with the PLpro with the lowest binding energy of -6.6 kcal/mol when all the selected compounds could achieve only higher energy than this. The interactions of Darunavir with the protein include four hydrogen bonds with Asn109, Gln269, Lys274 and Asp 286; two pi-sigma interactions with Trp106 and His272 besides the pi-alkyl and van der Waals interaction with other residues as depicted in the Figure 3b. Compounds 1, 6 and 7 showed the lowest binding energy of -6.2, -6.1 and -6.1 kcal/mol, respectively among the set. Out of the three potent molecules, 1 bound with protein by forming a hydrogen bond with the residues Trp106 and Ala288; pi-pi stacking with His272 and Trp106 and also van der Waals interaction with Asp286, Lys105, Gly287 and Leu289 (Figure 3d). Although 6 and 7 exhibited the same binding energy with PLpro, 6 made more interactions with the catalytic residues. The interactions are two hydrogen bonds with Trp106 and Cys111; pi-pi stacking interaction with Trp106, pi-alkyl interaction with Leu289 and also van der Waals interaction with His272 and other neighboring residues (Figure S2b in the Supplementary Information). Whereas 7 made no interaction with the catalytic triad but it exhibited hydrogen bonding with Asp37, Lys94 and Tyr97; amide pi stacked interaction with Gly142 along with few pi-alkyl and van der Waals interaction (Figure S2d in the Supplementary Information). Hence 1 and 6 found to be better candidates to inhibit PLpro of the SARS-CoV-2.</p><!><p>The docking scores of organosulfur compounds selected for the study of inhibition of Spro of the SARS-CoV-2 are shown in Table 1. Antiviral organosulfur drug for Influenza virus, arbidol, which has been repurposed against the SARS-CoV-2, is taken as the reference compound and is already in the clinical trial 24 . The docking score of arbidol with Spro is -6.1 kcal/mol when five compounds namely 1, 2, 3, 6, and 7 in our list showed lower binding energy values than the reference compound. The active site of the protein comprised the amino acid residues Phe486, Gln493 and Asn501. Arbidol displayed one hydrogen bonding with Gly496, pi-pi interaction with Tyr449, van der Waals interaction with the catalytic residues Gln493, Asn501 and also with other residues (Figure 4b). Compound 1 interacted with Spro more strongly with the binding energy -7.2 kcal/mol by making a hydrogen bond with Gly496; pi-pi stacking with Tyr 505 along with the van der Waals interaction with Glu406, Tyr453, Tyr495, Asn501 and Gly502 (Figure 4d). 3 has the strongest interaction with the protein, after 1, by -6.8 kcal/mol. The compound made two hydrogen bonding with Gln498 and Tyr449; pi-sulfur and pi-pi T shaped interaction with Tyr505; the active site residue Gln493 and Asn501 interacted with the protein through van der Waals interaction (Figure S3b in the Supplementary Information). Compound 2 binds with the binding energy -6.7 kcal/mol and made pi-sulfur interaction with Tyr505; hydrogen bonding with Tyr449 and Gln498, and it managed to make a van der Waals interaction with the catalytic residue Asn501 and other residues (Figure S3d in the Supplementary Information). 7 is another compound that also showed good results with the docking score -6.7 kcal/mol. This compound is interacting with the catalytic residues Gln493 and Asn501 through van der Waals interaction by making the hydrogen bond with Glu406 (Figure S3f in the Supplementary Information). Analyzing these results, it can be observed that organosulfur compounds exhibited considerably low binding energy with Spro of the SARS-CoV-2 warranting further in vitro and in vivo investigation to consider them as potential drugs for COVID-19.</p><!><p>For this study, we have taken the crystal structure of SARS-CoV helicase protein which shares almost 99.83 % similarity over the complete length of sequences with the helicase protein of SARS-CoV-2 virus 25 . The docking scores of the eight organosulfur compounds and the reference compound are represented in Table 1. The docked structure with the lowest binding energy and best conformation of the top-scoring organosulfur compound namely 1 along with one reference drug ivermectin, with their corresponding 2D interaction plots are shown in Figure 5. The docked conformation indicates that these molecules bind within the active site of the helicase protein of SARS-CoV. Figure 5b illustrates that the reference compound ivermectin binds via two carbon-hydrogen bonds with residues Arg178, Asp534, four alkyl hydrophobic interactions with Pro408, Ala312, Ala313, Ala316 and van der Waals interactions and these conformation resulted in the lowest binding energy of -8.5 kcal/mol. Compound 1 binds with two conventional hydrogen bonds involving Lys202 and Arg178, two Pi-alkyl interactions with Lys202 and Ala520, one pi-anion interaction with Glu201, and van der Waals interactions as shown in Figure 5d giving -7.7 kcal/mol binding energy. Compound 2 stabilizes the complex through one hydrogen bond with Lys202, one pi-anion with Glu201, one carbonhydrogen interaction with Asn177, pi-alkyl interaction with Lys202 and van der Waals interactions with other residues as represented in the 2D plot Figure S4b in Supplementary Information, Compound 2-helicase complex resulted in -6.9 kcal/mol binding energy. 3 was found to bind within the active site of the helicase through one hydrogen bond with Lys202, alkyl hydrophobic interaction with Arg178, pi-alkyl interaction with Lys202, pi-anion interaction with Glu201 and van der Waals interactions with other residues as shown in Figure S4d in Supplementary Information providing -7.2 kcal/mol binding energy. If we compare the values of the docking score, we can observe the reference compound ivermectin has the lowest binding energy with -8.5 kcal/mol followed by organosulfur compound 1, then 3 and finally 2. Three organosulfur compounds (1, 2 and 3) show good inhibitory activity towards the helicase protein of the SARS-CoV suggesting that these three organosulfur compounds might also be potent against helicase protein of the SARS-CoV-2.</p><!><p>Molecular docking studies of our organosulfur compound library against the RdRp protein of the SARS-CoV-2 revealed that the three organosulfur compounds, namely 1, 2 and 3 exhibited lowest binding energy along with the reference compound remdesivir 26 . The docked conformations of the RdRp-organosulfur compounds are depicted in Figure 6 and the docked scores are mentioned in Table 1. Remdesivir, a potent SARS-CoV-2 RdRp inhibitor binds in the active site (see Figure 6a) through hydrogen bonding with Tyr619 and Asp760, pi-sulfur interaction with Asp618, Asp761, pi-alkyl interactions with Lys621 and Pro620, alkyl hydrophobic interaction with Phe793, respectively, along with other interactions such as pisigma and van der Waals interactions with other residues as depicted in the 2D plot and this significant number of interactions resulted in the lowest binding energy of -7.4 kcal/mol. The docked structure showed that 1 formed four hydrogen bonds with Cys622, Arg553, Ser682, Asn691, and different non-covalent interaction such as pi-alkyl interaction with Lys621, pianion interaction with Asp623 and van der Waals interactions as shown in Figure 6d resulting lowest binding energy of -6.8 kcal/mol among all eight organosulfur compounds. 2 is involved in three hydrogen bond interaction with Trp800, Ser814, Cys813, two pi-anion interactions with Asp761, Glu811 and van der Waals interaction as represented in Figure S5b in the Supplementary Information consequently resulting -6.5 kcal/mol binding energy. Compound 3 formed two H-bond interactions with Lys621 and Ser795, one pi-alkyl interaction with Val166, one pi-sigma interaction with Pro620, two pi-cation interactions with Asp618, Lys798 and van der Waals interactions as depicted in Figure S5d eventually this complex resulted in minimum binding energy -6.8 kcal/mol. Here we observed that among the eight organosulfur compounds, three compounds namely 1, 2 and 3 showed promising binding activity with RdRp of the SARS-CoV-2. Based on these observations, the above mentioned three organosulfur compounds can be potential RdRp inhibitors to combat the SARS-CoV-2 infection. In order to identify the potential inhibitors for the SARS-CoV-2, molecular docking studies were carried out over eight potential organosulfur compounds against multiple target proteins namely Mpro, PLpro, Spro, RdRp and helicase. Among these compounds, 1 exhibited the lowest binding energy against all five proteins, which suggests that 1 could be the potential drug candidate for treating COVID-19. Apart from 1, 3 and 6 exhibited promising binding affinities towards the above mentioned five proteins which suggests that these two organosulfur also can act as antiviral drugs against the SARS-CoV-2. We were keen to do further investigation on the physicochemical, pharmacokinetic properties and target prediction studies of 1 which showed the lowest binding energy among all the eight organosulfur compounds against five target proteins of the SARS-CoV-2. The predicted pharmacokinetic properties of other organosulfur compounds are tabulated in Table S1-S5 in the Supplementary Information.</p><!><p>The physicochemical properties of the compounds were studied to predict the pharmacokinetics of the drug by the Lipinski's rule. The guidelines for an orally active drug according to the Lipinski's rule are (i) molecular weight (MW) <500 Daltons, (ii) octanol-water partition coefficient (clogP) <5, (iii) polar surface area (PSA) <150 Å 2 , (iv) number of hydrogen bond donors (HBD) <5, (v) number of hydrogen bond acceptors (HBA) <5 and (vi) Number of rotatable bonds (RB) <10 27 . The calculated values for the same for the selected organosulfur compounds are tabulated in Table 2 and the result shows that all the compounds strictly follow Lipinski's rule with zero violation. This indicates that the compounds have the potential for drug-like activities.</p><!><p>We carried out ADMET property profiling to explore the drug likeliness of compound 1 which exhibited efficient binding energy among the eight organosulfur compounds against RdRp, PLpro, Mpro, Spro and helicase proteins of SARS-CoV_2 in the molecular docking study. Insilico pharmacological prediction of 1 was performed using the pkCSM server to assess the overall ADMET properties (see Table 3). A favorable ADMET profile is necessary for the molecules in drug discovery. 1 showed water solubility and high Caco-2 permeability, which indicates that this drug can be absorbed orally. 1 showed good human intestinal absorption and skin permeability. Compound 1 was predicted to be a substrate of P-glycoprotein as well as Pglycoprotein I and II inhibitor</p><p>The Volume of distribution at steady state (VDss) prediction indicates a low theoretical dose of 1 will be required to get it uniformly distributed in blood plasma. Blood-brain barrier (BBB) permeability prediction showed that 1 readily cross the BBB and drug can penetrate the central nervous system.</p><p>It is well known that cytochrome P450s can regulate the metabolism of various drugs. In that respect, it is worth noting that inhibitors of CYP2D6/CYP3A4 can hamper the pharmacological properties of drugs. 1 inhibits neither CYP2D6 nor CYP3A4, whereas it is predicted to be a substrate of CYP3A4. Further, it was observed that 1 is not a substrate of ROCT-2 which means that this drug can be excreted through other routes such as bile, sweat and breathe.</p><p>We have also assessed the toxicity index of the organosulfur compound 1. The toxicity prediction from the Ames test (Salmonella typhimurium reverse mutation assay) revealed that 1 could be considered as a mutagenic agent. High toxicity was predicted in Tetrahymena pyriformis. 1 was shown to inhibit the human ether-a-go-go-related gene II (hERG II). However, 1 was found to be associated with hepatotoxicity. The maximum recommended tolerated dose (MRTD) in human prediction shows that 1 does not violate MRTD. 1 is predicted to be a high acute toxic compound as it falls under minnow toxicity. Additionally, compound 1 is not associated with skin sensitivity.</p><!><p>Most of the drug performs its mechanism of action by interacting with the proteins, enzymes and other biomacromolecules. However, many drugs have more than one target. In-silico predictions of drug targets based on resemblance with known drugs are very useful to find out the number of targets. Here we observed that 1 has 68% kinases as a target. As shown in Figure 7, compound 1 interacts with a broad range of proteins and enzymes. The detailed information on the target, common name, UniProt ID, ChEMBL ID, target class, probability and known actives in 2D/3D are shown in Table S6 in the Supplementary Information.</p><!><p>When the entire world fight against the global pandemic of SARS-CoV-2, the major challenge towards the scientists is to annihilate the viral effect. This study is based on the identification of potential drug molecules against the deadly virus from the list of known drugs against SARS-CoV which has a very similar structure of SARS-CoV-2. As the viral drug targets are susceptible to mutations at higher rates, our aim was to investigate the compounds which could bind with multiple targets. From the list of selected organosulfur compounds, we could find compounds that interacted with multiple targets and surprisingly one compound, 1 found to be very efficient on inhibiting all the five SARS-CoV-2 targets namely RdRP, helicase, Mpro, PLpro and Spro with a significant binding affinity. Hence, this compound can be an effective candidate against SARS-CoV-2 for a longer term as it is capable of binding with multiple targets and inhibiting its activity, thus reducing the effect of drug resistance. The physicochemical properties of all the compounds are studied and found that all the compounds are druggable with zero violations from the Lipinski's rule. The ADMET profile and the target prediction studies were also carried out for the most potential candidate 1 and observed that this can be a promising drug against the SARS-CoV-2. In-silico ADMET studies of 1 revealed that it has promising pharmacokinetic properties and does not fall under high-risk chemical groups. Target prediction analysis also showed that compound 1 exhibits excellent drug-like properties. Based on the results obtained, we look forward to performing the in vitro and in vivo studies to evaluate the potency of compound 1 and other hits as plausible therapeutic agents for the pandemic COVID-19 through multi-target binding.</p>

ChemRxiv

Effects of the Salt-Processing Method on the Pharmacokinetics and Tissue Distribution of Orally Administered Morinda officinalis How. Extract

Salt processing, which involves steaming with salt water, directs herbs into the kidney channel. After being salt processed, kidney invigorating effects occur. However, the underlying mechanism of this method remains elusive. The compounds monotropein, rubiadin, and rubiadin 1-methyl ether are the major effective components of Morinda officinalis How. To clarify the pharmacokinetics and tissue distribution of these three compounds, we employed liquid chromatography-tandem mass spectrometry (UPLC-MS/MS) to determine the contents of the three components in rat plasma and tissues. Separation was achieved on an Acquity UPLC HSS T3 column (100 mm × 2.1 mm, 1.8 μm, Waters). Formic acid aqueous solution (0.1%; A) and acetonitrile (containing 0.1% formic acid; B) were used as the mobile phase system with a programmed elution of 0∼5 min with 70% A and then 5∼7 min with 60% A. All analytes were measured with optimized multiple reaction monitoring (MRM) in negative ion mode. Geniposide and 1,8-dihydroxyanthraquinone were used as the internal standards (IS). The linear ranges were 1.2∼190, 1.3∼510, and 0.047∼37.5 μg/mL, respectively. Compared with the Morinda officinalis without wood (MO) group, the Cmax and AUC0-t parameters of rubiadin and rubiadin 1-methyl ether elevated remarkably for the salt-processed Morinda officinalis (SMO) groups, which indicates that steaming by salt could increase the bioavailability of rubiadin and rubiadin 1-methyl ether. The Tmax for monotropein is shorter (0.5 h) in SMO groups than that in MO group, which means that monotropein was quickly absorbed in the SMO extract. Moreover, the contents of three compounds in the small intestine were the highest.

effects_of_the_salt-processing_method_on_the_pharmacokinetics_and_tissue_distribution_of_orally_admi

1. Introduction<!>2.1. Materials and Chemicals<!>2.2. Animals<!>2.3. LC-MS Analytical Conditions<!>2.4. Preparation of Reference Substances<!><!>2.5. Preparation of MO and SMO Extract<!>2.6. Preparation of Biological Samples<!>2.7.1. Specificity<!>2.7.2. Calibration Curve and Sensitivity<!>2.7.3. Precision, Accuracy, and Stability<!>2.7.4. Extraction Recovery and Matrix Effect<!>2.8. Analysis of Pharmacokinetics<!>2.9. Analysis of Tissue Distribution<!>3.1. Liquid Chromatography Optimization<!>3.2. Mass Spectrometry Optimization<!>3.3. Extraction Procedure Optimization<!>3.4.1. Specificity<!>3.4.2. Linearity and Sensitivity<!>3.4.3. Precision, Accuracy, and Stability<!>3.4.4. Extraction Recovery and Matrix Effect<!>3.5. Pharmacokinetics Study<!>3.6. Tissue Distribution<!>4. Conclusion

<p>Morinda officinalis is a commonly used traditional Chinese medicines (TCMs) [1], which has been used in China for many years. People use it as a tonic or tonifying kidney product to protect against and cure depression, rheumatoid arthritis, osteoporosis, and kidney-yang deficiency syndrome [2–6]. Its main active components include oligosaccharides, polysaccharides, iridoid glycosides, and anthraquinones [7–9].</p><p>Chinese medicinal herb processing is a unique pharmaceutical technique, which can turn Chinese medicinal materials into decoction pieces using processing methods like stir-frying, steaming, and frying with sand or wheat bran [10]. Decocting pieces are the main form prescribed in TCM clinics [1]. The pharmacological actions and energy properties (nature, flavor, and channel tropism) might be changed after processing, side effects and disagreeable odors can be eliminated. It is essential to use the proper processing method to ensure the quality and safety of traditional Chinese medicine decocting pieces. Steaming or stir-frying with salt can change the activity direction of Chinese herbal medicines and can also improve their efficacy [11–14].</p><p>The commercially available products of Morinda officinalis in herbal markets are M. officinalis without wood (MO) and salt-processed products (SMO). It was revealed that MO extract exerts tonifying kidney and supporting yang effects by regulating the functions of the hypothalamus-pituitary-adrenal axis [15, 16]. The medicinal efficacy of MO and SMO is different because of their different processing methods. After being salt processed, the effects of strengthening kidney-yang can be dramatically enhanced. The chemical composition changes during processing, for example, glycosides can be hydrolyzed into aglycone or converted into other constituents, the contents of toxic ingredients can be decreased or converted into other constituents [17, 18].</p><p>Monotropein, which possess anti-inflammatory, analgesia, and antiosteoporotic activities, is the main iridoid glycoside isolated from M. officinalis [19, 20]. Anthraquinones, especially alizarin-anthraquinones, like rubiadin and rubiadin 1-methyl ether, can inhibit osteoclastic bone resorption in vitro and invigorate the kidney-yang [21–24]. Therefore, in the present study, we chose monotropein, rubiadin, and rubiadin 1-methyl ether as the representative compounds to explain the effects of the different processing methods on the pharmacokinetics of M. officinalis.</p><p>There have been reports on the pharmacokinetics or tissue distribution of iridoids and oligosaccharides using HPC-DAD and LC-MS [25–28]. However, the abovementioned studies paid more attention to pure iridoid, inulin-type fructo-oligosaccharides and neglected interactions among iridoids and other compounds. There still has been no report about the pharmacokinetics and tissue distribution characteristics of alizarin-anthraquinones in M. officinalis. Also, the effect of processing methods on the pharmacokinetics, bioavailability, and tissue distribution characteristics of iridoids and anthraquinones has barely been reported.</p><p>In this study, we developed a selective and sensitive UPLC-MS/MS method for simultaneous determination of monotropein, rubiadin, and rubiadin 1-methyl ether in rat plasma and tissues. The analysis method was successfully applied to the pharmacokinetics study of MO and SMO extracts.</p><!><p>M. officinalis were collected from Gaoliang County, Guangdong Province, and identified by Prof. Feng Li of the Liaoning University of Traditional Chinese Medicine. For reference substance, rubiadin (J0111AS), rubiadin 1-methyl ether (J0307AS), monotropein (00605AS) were purchased from Meilun Biological Technology Co. Ltd. (Shanghai, China). The internal standards (ISs) geniposide (1203A023) and 1,8-dihydroxyanthraquinone (100398–200701) were provided by the National Institute for Food and Drug Control (Beijing, China).</p><p>Merck KGaA (Darmstadt, Germany) supplied LC-MS-grade formic acid and acetonitrile (HPLC-grade). The ultrapure water was generated with a Milli-Q water purification system (18.2 MΩ, Millipore, Billerica, USA). Other reagents were of analytical grade from Tianjin Kermol Chemical Reagent Co., Ltd. (Tianjin, China) and Waters Xevo TQD Mass Spectrometer (Massachusetts, USA).</p><!><p>Sprague-Dawley (SD) rats (200 ± 20 g, weight) were purchased from Changsheng Biotechnology Co. Ltd. (License No. SCXK (Liao) 2015-0001, Benxi, Liaoning Province). The experimental animals were kept at 25 ± 2°C and 60% ± 10% humidity with a 12 h light/dark cycle. Water and chow were provided ad libitum. All animal pharmacological experiments followed the ethical regulations of the Liaoning University of Traditional Chinese Medicine.</p><!><p>For LC-MS/MS analysis, data acquisition and instrument control were performed using MassLynx 4.1 software (Waters Corp., Milford, MA, USA). The analysis column was an Acquity UPLC HSS T3 column (100 mm × 2.1 mm, 1.8 μm, Waters) with a temperature of 40°C. The mobile phase was acetonitrile containing 0.1% formic acid (A) and 0.1% formic acid aqueous solution (B). The elution gradient was 0.00–0.50 min with 2% B; 0.51–5.00 min with 2%–30% B; and 5.01–7.00 min with 40% B. The flow rate was 0.2 mL/min.</p><p>The Waters triple quadruple mass spectrometer (Xevo TQD, Waters Corp., Milford, MA, USA) equipped with an electrospray ionization source (ESI) was used in negative ion mode. The desolvation gas was nitrogen with a flow rate of 500 L/h at a temperature of 250°C. All detected compounds were measured in multiple reaction monitoring (MRM) mode. Figure 1 shows the chemical structures of the three analytes and the internal standards.</p><p>The precursor and product ion pairs for MRM were m/z 389.01 ⟶ 191.01 for monotropein (collision energy 15 eV); m/z 253.05 ⟶ 195.21 for rubiadin (collision energy 40 eV); m/z 267.07 ⟶ 252.04 for rubiadin 1-methyl ether (collision energy 18 eV); m/z 387.03 ⟶ 225.03 for IS geniposide (collision energy 8 eV); and m/z 238.94 ⟶ 166.99 for IS 1,8-dihydroxyanthraquinone (collision energy 35 eV).</p><!><p>Monotropein, rubiadin, and rubiadin 1-methyl ether were dissolved in methanol to prepare a stock solution with a concentration of 0.5 mg/mL. The stock solution was diluted with methanol to get the appropriate concentrations for the working standard solutions. The IS stock solution (1 mg/mL) and IS working solution (200 ng/mL) were also prepared in methanol, as described above. All prepared solutions were stored at 4°C before use.</p><p>The preparation of calibration standards was done as follows: The appropriate working solution was spiked into blank plasma or tissues to obtain concentrations of 75∼0.047 μg/mL for monotropein, 396∼1.54 μg/mL for rubiadin, and 596∼1.16 μg/mL for rubiadin 1-methyl ether. The QC concentrations for monotropein were 0.14 μg/mL, 2 μg/mL, and 28.1 μg/mL; for rubiadin, they were 1.6 μg/mL, 50.6 μg/mL, and 712.5 μg/mL; and for rubiadin 1-methyl ether, they were 3.9 μg/mL, 54.6 μg/mL, and 765 μg/mL. The QC samples used for the recovery, matrix effect, intra- and interday accuracy, precision, and stability studies were prepared in the same way as that of the calibration standards. The solutions were stored for one week at −20°C.</p><!><p>MO: steam the clean Morindae officinalis Radix for 0.5 h, remove the woody cores while hot, cut into sections, and dry.</p><p>SMO: steam the clean Morindae officinalis Radix with salt water for 4.0 h, until it is steamed thoroughly, remove the woody cores while hot, cut into sections and dry.</p><!><p>Then, MO and SMO were refluxed with 80% ethanol three times for 1 h each to make a combined filtrate. Then, the extracts were concentrated to 4 g·mL−1.</p><p>The contents of the three constituents in the MO and SMO extracts were detected by the same UPLC-MS/MS method, including the same column, mobile phase system, and column temperature as used for the pharmacokinetics and tissue distribution study. The contents of monotropein, rubiadin, and rubiadin 1-methyl ether in the MO and SMO extracts were 6.44, 0.088, and 0.174 and 3.95, 0.091, and 0.178 mg/g, respectively. The concentrated extract was redissolved in distilled water.</p><!><p>10 μL of IS working solution and 400 μL of methanol were added to 100 μL plasma. The solution was vortexed for 10 min and then centrifuged at 12,000 rpm for 5 min. The supernatant was shifted to another centrifugation tube and dried with nitrogen at 37°C. The residue was redissolved in 100 μL acetonitrile and centrifuged for 5 min at 12,000 rpm. Five microliters of supernatant was drawn and analyzed by UPLC-MS/MS.</p><p>The whole tissue was cut into pieces and homogenized with ice-cold 0.9% (m/v) sodium chloride solution (1 : 9, w/v). Using 50 μL of tissue homogenate, the tissue samples were prepared in the same way as the plasma samples.</p><!><p>Six individual sources of blank plasma were used to measure the specificity of this analysis method. Meanwhile, we also analyzed the lower limit of quantification (LLOQ) of samples, the IS in blank plasma samples, and plasma samples within 2 h of being subjected to the MO extract (40 g/kg).</p><!><p>The peak-area ratio of the compounds to the IS was plotted to obtain the linearity. The weighted least-squares linear regression (weighting factor 1/X2) was applied to determine the regression equations. The lowest concentration in the calibration curve was determined as the LLOQ with signal-to-noise of ≥10.</p><!><p>Through evaluating three levels of QC samples (n = 5) after one day and then for three days in a row, we obtained the intra- and interday precision, along with the accuracy data. The permitted range was within 15%.</p><p>The QC samples used for the recovery QC samples were used to judge the stability, and the storage conditions were as follows: (1) freeze-thaw stability, (2) at room temperature for 4 h, and (3) put into the autosampler for 8 h, and (4) at −20°C for 30 days. The permitted range was within 15%.</p><!><p>Three different QC sample concentrations were used for the recovery experiment. The absolute extraction recovery was measured as the ratio of the QC sample concentration extracted from the blank plasma/tissue to those in the QC samples.</p><p>The peak areas of the biosamples extracted in blank plasma/tissue versus those dissolved in methanol solution were measured to obtain the matrix effect.</p><!><p>The SD rats (6, male) were housed at 22 ± 2°C and fasted for 12 h with drinking water available ad libitum. MO and SMO ethanol extract were orally administered at a dose of 40 g/kg (equivalent to 6.44 and 3.95 mg/kg of monotropein, 0.088 and 0.091 mg/kg of rubiadin, and 0.174 and 0.178 mg/kg of rubiadin 1-methyl ether). Blood samples (0.5 mL) were collected from the venous plexus of the eye socket at 0.083, 0.25, 0.5, 1.0, 1.5, 2.0, 2.5, 4.0, 6.0, 12.0, and 24 h under anesthesia. Then, the blood samples were centrifuged at 4500 rpm for 10 min, and then the plasma was immediately transferred to new tubes and stored at −80°C until analysis. The pharmacokinetic parameters were calculated using DAS software (3.0 version, China Food and Drug Administration) based on noncompartmental method. All data were recorded as mean ± SD.</p><!><p>MO and SMO ethanol extract (40 g/kg) was orally administered to SD rats (weight, 180–220 g, n = 6), at 0.25, 1, 3, and 6 h, respectively. Heart, liver, spleen, stomach, kidney, brain, and small intestine tissues were extracted, washed with saline, blotted with filter paper, weighed, and then stored at −80°C.</p><!><p>We tested mobile phase and gradient elution programs to determine the best chromatographic performance. With the mobile phase of water (containing 0.1% formic acid) and acetonitrile (containing 0.1% formic acid), the responses of the three components and two ISs were considerably better. To reach a better and more rapid separation effect, we optimized the gradient elution program.</p><!><p>We carried out both positive and negative modes in mass spectrometry. Three components and the ISs all had high responses under the negative mode. The MS/MS transitions and parameters are given in Table 1.</p><!><p>Protein precipitation (PP) was used for the pretreatment of biological samples. We applied methanol and acetonitrile for PP. The samples had better peak shapes and recovery after being treated by acetonitrile, so we chose acetonitrile for further PP in this study.</p><!><p>The MRM chromatograms of three components and the ISs are shown in Figure 2. The peak separation was better for the three components and the ISs under the established UPLC-MS/MS conditions, with no significant interference and no cross interference.</p><!><p>The internal standard method was applied for the establishment of calibration curves, which showed good linearity (r2 > 0.9907) in the linear ranges. The regression equations, correlation coefficients, linear ranges, and LLOQs of the three components in the plasma and tissues are listed in Table 2.</p><!><p>The range of intra and interday precisions was 0.21% to 5.96%, and the accuracy range was from −8.06% to 1.33% for QC samples in plasma and tissues. These values are in the acceptable range, and the results are shown in Table 3.</p><p>Table 4 shows the results for the stability of the plasma and tissue samples in different conditions. They are also in the acceptable range.</p><!><p>The range of extraction recoveries was 80.07∼98.64% for the three QC sample levels (Table 5). These data indicate that the sample treatment method is reasonable. No significant effect of endogenous substances was identified.</p><!><p>After oral administration of MO and SMO extract (40.0 g/kg), we successfully determined the concentrations of monotropein, rubiadin, and rubiadin 1-methyl ether in rat plasma using the established method. Figure 3 shows the mean plasma concentration-time profiles. Table 6 displays the pharmacokinetic parameters.</p><p>The concentration of monotropein in SMO reached Cmax in 0.5 h, whereas the Tmax of monotropein in the MO groups was 1.0 h, which indicates that monotropein was quickly adsorbed into the blood after oral administration. The Tmax for rubiadin was 1.5 h, longer than that for rubiadin 1-methyl ether, whereas the Cmax for rubiadin 1-methyl ether was the highest, especially in the SMO groups. In the plasma concentration-time curves of monotropein, an obvious double-peak phenomenon was found, which is related to enterohepatic recirculation. The pharmacokinetics properties shown in the present assay could be helpful for further studies on the pharmacokinetics of MO extract and further applications in different processing methods.</p><!><p>The concentrations of monotropein, rubiadin, and rubiadin 1-methyl ether in the liver, kidney, lung, and small intestine were determined at 0.25, 1, 3, and 6 h after administration. The results are shown in Figure 4. All compounds were detected in tissues, except for monotropein, which could not be quantified in the lung tissue. This result indicates that monotropein might be rapidly transformed into its metabolites in the lungs, or the content of monotropein in the lung was lower. Meanwhile, all three components had higher concentrations in the small intestine, especially rubiadin and rubiadin 1-methyl ether in the SMO groups.</p><!><p>In this study, we established an efficient and accurate UPLC-MS/MS method for the determination of three components from MO and SMO in plasma and tissues after oral administration in rats. Meanwhile, this is the first simultaneous determination of monotropein, rubiadin, and rubiadin-1-methyl ether in rat plasma and tissues. The study examined the pharmacokinetics and tissue distribution. The information found might partially illustrate their metabolic mechanisms in vivo, as well as provide a scientific basis for the strengthening of kidney-yang using the salt-processed principle of M. officinalis because of better bioavailability.</p>

PubMed Open Access

Calcineurin is an important factor involved in glucose uptake in human adipocytes

Calcineurin inhibitors are used in immunosuppressive therapy applied after transplantation, but they are associated with major metabolic side effects including the development of new onset diabetes. Previously, we have shown that the calcineurin inhibiting drugs tacrolimus and cyclosporin A reduce adipocyte and myocyte glucose uptakes by reducing the amount of glucose transporter type 4 (GLUT4) at the cell surface, due to an increased internalization rate. However, this happens without alteration in total protein and phosphorylation levels of key proteins involved in insulin signalling or in the total amount of GLUT4. The present study evaluates possible pathways involved in the altered internalization of GLUT4 and consequent reduction of glucose uptake provoked by calcineurin inhibitors in human subcutaneous adipose tissue. Short- and long-term treatments with tacrolimus, cyclosporin A or another CNI deltamethrin (herbicide) decreased basal and insulin-dependent glucose uptake in adipocytes, without any additive effects observed when added together. However, no tacrolimus effects were observed on glucose uptake when gene transcription and protein translation were inhibited. Investigation of genes potentially involved in GLUT4 trafficking showed only a small effect on ARHGEF11 gene expression (p < 0.05). In conlusion, the specific inhibition of calcineurin, but not that of protein phosphatases, decreases glucose uptake in human subcutaneous adipocytes, suggesting that calcineurin is an important regulator of glucose transport. This inhibitory effect is mediated via gene transcription or protein translation; however, expression of genes potentially involved in GLUT4 trafficking and endocytosis appears not to be involved in these effects.

calcineurin_is_an_important_factor_involved_in_glucose_uptake_in_human_adipocytes

Introduction<!>Subcutaneous adipose tissue (SAT)<!><!>Culture of adipose tissue and isolated adipocytes<!><!>Culture of adipose tissue and isolated adipocytes<!>Assessment of cell viability<!>Glucose uptake in adipocytesz<!>Adipose tissue gene expression<!><!>Adipose tissue gene expression<!>Statistical analysis<!>Calcineurin inhibition reduces glucose uptake<!><!>Calcineurin inhibition reduces glucose uptake<!>Inhibition of calcineurin, but not other protein phosphatases, reduces adipocyte glucose uptake<!><!>The inhibitory effect of tacrolimus on glucose uptake requires gene transcription and protein translation<!><!>Effects of tacrolimus on the expression of genes involved in GLUT4 translocation<!><!>Discussion

<p>The International Diabetes Federation has shown that the number of people with type 2 diabetes is increasing worldwide. In 2014, there were about 400 million individuals with diabetes mellitus, and this number is expected to increase to approx. 600 million in 2035 [1]. Major clinical trials show that glucose control is not sufficient to prevent comorbidities and its excess mortality, associated with type 2 diabetes [2]. Hence, there is an unmet need to identify and evaluate novel and innovative therapeutic concepts and approaches based on previously unknown molecular pathways.</p><p>Calcineurin is a serine/threonine phosphatase controlled by cellular calcium concentrations [3]. Calcineurin has been implicated in a variety of biological responses, including lymphocyte activation and cardiovascular and skeletal muscle development [3]. Since their introduction, calcineurin inhibitors have become the cornerstone of immunosuppressive therapy in solid organ transplantation. However, they are associated with the development of cardiovascular and metabolic complications, like dyslipidemia, hypertension and diabetes melitus [4]. New onset diabetes after transplantation (NODAT) is a common metabolic complication with reported incidence rates up to 50% during the first years after transplantation [5, 6]. Similar to type 2 diabetes, both impaired insulin secretion and insulin resistance in peripheral tissues and liver are the principal pathogenic components of NODAT [7]. However, the mechanisms are not known. Calcineurin inhibitors have been shown to cause adverse effects in white adipose tissue metabolism that can contribute to the development of insulin resistance and diabetes mellitus [8–12]. Our previous work has shown that cyclosporin A and tacrolimus are able to reduce glucose uptakes in human adipocytes and L6 muscle cells. The reduction of glucose uptake was achieved by decreasing the total amount of the glucose transporter type 4 (GLUT4) at the cell surface, mainly due to increased internalization rate [10]. Furthermore, we have also shown that the calcineurin inhibitors increased GLUT4 internalization without affecting total protein levels or phosphorylation of key insulin signalling proteins, including insulin receptor substrate 1 (IRS1), protein kinase B (PKB), AS160 and GLUT4. GLUT4 is the major glucose transporter in muscle and adipose tissue, which constantly cycles between the plasma membrane and intracellular membranes due to the presence of insulin. The GLUT4 endocytic and exocytic itineraries involve a complex interplay of trafficking events and intracellular signalling cascades [13, 14]. Calcineurin can dephosphorylate cytoskeletal proteins, such as actin and tubulin [15–20], and proteins involved in endocytosis, such as dynamin and assembly protein 180 kDa (AP)180 [21]. Also, it is known that calcineurin can induce endocytosis in neurons and other cell types in response to increased cytosolic calcium concentration [21–23].</p><p>A role of calcineurin in glucose uptake has also emerged from studies in skeletal muscle in mice. These studies demonstrate that, in transgenic mice overexpressing an activated form of calcineurin, there is an elevation of insulin-stimulated skeletal muscle glucose uptake [24, 25]. Furthermore, several studies have shown that the calcium–calcineurin pathway directly affects insulin-stimulated glucose transport in adipocytes [26, 27] and elevated levels of cytosolic calcium are associated with insulin resistance [28].</p><p>Therefore, the aim of the present study was to further investigate possible molecular mechanisms underlying our previous findings, with respect to increased internalization of GLUT4 at the plasma membrane and consequent reduction of glucose uptake induced by calcineurin inhibitors in human subcutaneous adipocytes.</p><!><p>Human abdominal SAT biopsies were obtained from nondiabetic subjects (10 males/32 females; age 50 ± 16 years; body mass index (BMI) 26.1 ± 3.2 kg/m2). Due to limited amount of adipose tissue obtained, not all experiments were performed in the same biopsies. The number of experiments is indicated in each section below. Subjects were fasted overnight (> 10 h), and fasting venous blood samples were collected in the morning for analysis of glucose, insulin and lipids by routine methods at the Department of Clinical Chemistry at the Sahlgrenska University Hospital and the Uppsala University Hospital. SAT biopsies were performed by needle aspiration of subcutaneous fat from the lower abdomen (n = 31) after intradermal local anaesthesia with lidocaine (Xylocain; AstraZeneca, Södertälje, Sweden), or by elective abdominal surgery (n = 11) after induction of general anaesthesia.</p><p>The clinical and biochemical characteristics of the adipose tissue donors are described in Table 1. Anthropometric measurements including body composition, assessed by bioimpedance, were measured in all subjects [29]. Subjects with diabetes, other endocrine disorders, systemic illnesses or malignancy, as well as ongoing medication with systemic glucocorticoids, beta blockers and immune-modulating therapies were excluded from the study. The study protocol was approved by the Regional Ethics Review Boards in Gothenburg and Uppsala. Written informed consent was obtained from all subjects.</p><!><p>Clinical and biochemical characteristics of adipose tissue donors (n = 42)</p><p>HbA1c glycosylated haemoglobin, HOMA-IR homeostatic model assessment-insulin resistance, LDL low-density lipoprotein, HDL high-density lipoprotein</p><p>aCalculated as fasting insulin (mU/L) × fasting glucose (mM)/22.5</p><!><p>Adipocytes were isolated from SAT obtained from needle biopsies after collagenase type II digestion (Roche, Mannheim, Germany) in Hank's medium (Invitrogen Corporation, Paisley, UK) containing 6 mM glucose, 4% BSA and 150 nM adenosine (Sigma Chemical Co., MO, USA) (pH 7.4) for 60 min at 37 °C in a shaking water-bath. Isolated adipocytes were filtered through a 250-μm nylon mesh and pre-incubated for 15 min (short-term) or 20 h (long-term) with tacrolimus (100 nM), cyclosporin A (100 nM), deltamethrin (1 μM), okadaic acid (250 nM), actinomycin D (5 μg/ml) or cycloheximide (25 μM)—alone or in combination (see the Results section).</p><p>The time points and the concentrations were chosen according to previous studies [10, 12, 30–34]. Tacrolimus binds to FK506-binding proteins, and cyclosporin A binds to cyclophilins-forming complexes that inhibit calcineurin [6, 35]. The concentration (100 nM) of tacrolimus and cyclosporin A was previously shown to induce maximum reduction of glucose uptake in adipocytes and to be at therapeutic concentrations commonly used in clinic [10, 12]. Deltamethrin is a type II synthetic pyrethroid insecticide that can also inhibit calcineurin [32], but the mechanism of action is unknown. Deltamethrin was used to test the effect of a different calcineurin inhibitor on glucose uptake for comparison. Actinomycin D and cycloheximide are well-known gene-transcription and protein-translation inhibitors, respectively [33, 34]. They were used to test whether transcription and/or translation is involved in the inhibitory effects of the calcineurin inhibitors on glucose uptake. The concentrations of deltamethrin, actinomycin D and cycloheximide were shown to maximally inhibit calcineurin, gene transcription and protein translation, respectively, without significantly reducing cell viability [32–34] (Fig. 1). Okadaic acid is a phosphatase inhibitor that, at 250 nM concentration, can inhibit the phosphorylated myosin light-chain (PMLC) phosphatase, phosphatase 1 and phosphatase 2A, but not calcineurin (protein phosphatase 2B) [30, 31] .</p><!><p>The incubations with tacrolimus, deltamethrin, actinomycin D and cycloheximide do not alter the viability of human subcutaneous adipocytes. After isolation, adipocytes were incubated for 20 h with either tacrolimus 100 nM, deltamethrin 1 μM, actinomycin D 5 μg/ml or cycloheximide 25 μM, and the cell viability was measured. The results were calculated relatively to untreated cell values and represent the means ± SEM of four subjects</p><!><p>For short-term incubations, isolated adipocytes were washed three times in glucose-free Krebs Ringer media (KRH) supplemented with 4% BSA, 150 nM adenosine and pH 7.4. Adipocytes were then diluted ten times in supplemented KRH medium and pre-incubated for 15 min with the described conditions for further glucose uptake analysis. For long-term incubations, isolated adipocytes were washed three times in Hank's medium that contained 6 mM glucose, 4% BSA and 150 nM adenosine and placed in DMEM (Invitrogen) with 6 mM glucose and 10%  FCS (Invitrogen) in the different conditions described and at 37 °C under a gas phase of 5% CO2 in a culture chamber for 20 h. After incubation, cells were washed and diluted ten times in KRH medium (4% BSA, 150 nM adenosine, pH 7.4) for further glucose uptake analysis. The average cell diameter was measured in isolated adipocytes from all subjects [36].</p><p>Effect of long-term incubation (20 h) with tacrolimus on gene expression of possible intermediates of GLUT4 trafficking was analysed in SAT samples. For this, 100 mg of adipose tissue explants were incubated for 20 h without or with tacrolimus (100 nM) in 24 well polystyrene plates containing 1 ml of DMEM (6 mM glucose, 10% FCS) (Invitrogen Corporation, Paisley, USA) in a humidified atmosphere of 5% CO2 at 37 °C. Adipose tissue was thereafter snap-frozen for gene expression analysis.</p><!><p>After 20 h incubation of subcutaneous adipocytes (n = 4) with tacrolimus (100 nM), cyclosporin A (100 nM), deltamethrin (1 μM), okadaic acid (250 nM), actinomycin D (5 μg/ml) or cycloheximide (25 μM), cell viability was assessed with the water soluble tetrazolium-colorimetric reagent (WST-1, Roche, Mannheim, Germany) according to manufacturer instructions. The viability of adipocytes was not significantly affected with any treatment compared with untreated cells (Fig. 1).</p><!><p>Glucose uptake in isolated subcutaneous adipocytes was performed according to a previously validated technique for human adipocytes, which reflects rate of transmembrane glucose transport [37]. Briefly, after long-term (20 h) or short-term (15 min) incubation without changing the media, adipocytes were incubated with or without insulin (25 and 1000 mU/ml, Actrapid, NovoNordisk, Bagsvaerd, Denmark) for 15 min, followed by an additional 45 min of incubation with D-[U–14C] glucose (0.26 mCi/L, 0.86 mM, Perkin Elmer, Boston, MA, USA). The reaction was stopped by transferring the cells into pre-chilled vials followed by separation from the medium by centrifugation through Dow Corning Xiameter PMX 200/100cC silicone fluid (BDH Prolabo Chemicals, Leuven, Belgium). Radioactivity associated with the cells was then determined using a scintillation counter. Cellular glucose uptake was calculated using the following formula: Cellular clearance of medium glucose = (cell-associated radioactivity × volume)/(radioactivity of medium × cell number × time). Using this experimental setup, glucose uptake is mainly determined by the rate of transmembrane glucose transport. Adipocyte size and number were measured as described previously [38]. Glucose uptake was normalized per cell number for each experimental condition and expressed relative to control. All experiments were performed in triplicates.</p><!><p>The aim of the adipose tissue gene expression analyses was to verify whether previously reported effects of calcineurin inhibitors on GLUT4 trafficking by increasing the rate of GLUT4 endocytosis [10], could be due to effects on expression of key genes directly involved with these mechanisms.</p><p>Total RNA from adipose tissue was isolated with RNeasy Lipid Tissue Mini Kit (Quiagen, Hilden, Germany), and used for cDNA synthesis using High-Capacity cDNA Reverse Transcriptase kit (Applied Biosystems, CA, USA). The protocol was carried out in accordance with manufacturer's instructions. Total RNA concentration and purity were measured using the Nanodrop 2000 Spectrophotometer (ThermoFisher Scientific, Rockford, USA). Gene expression was analysed using the QuantStudio™ 3 Real-Time PCR Systems (Applied Biosystems, CA, USA).</p><p>First, TaqMan® Array—96-well plates (Applied Biosystems, CA, USA) having four housekeeping genes [18S ribosomal RNA, low-density lipoprotein receptor-related protein 10 (LRP-10), glyceraldehyde-3-phosphate dehydrogenase (GAPDH) and glucuronidase beta (GUSB)] plus 92 selected genes (see Table 2 for the list of the genes)— were used (n = 3). The selected genes encode for proteins involved in cytoskeleton organization and potentially in GLUT4 trafficking, specially endocytosis, therefore, putative downstream gene targets of calcineurin signalling. The normalization of the gene expression of the 92 analysed genes was performed with the geometrical mean [(Ct value of LRP-10 × Ct value of GUSB)1/2] of the Ct values of the two housekeeping genes, GUSB and LRP-10, with lower coefficient of variation (CV = standard deviation of Ct values of all samples/mean of all samples). Subsequently, the expression of each gene was normalized to control, and calculated as a relative fold change (\documentclass[12pt]{minimal} sepackage{amsmath} sepackage{wasysym} sepackage{amsfonts} sepackage{amssymb} sepackage{amsbsy} sepackage{mathrsfs} sepackage{upgreek} \setlength{\oddsidemargin}{-69pt} egin{document}$${{ ext{2}}^{ - \Delta \Delta {C_{ ext{t}}}}}$$\end{document}2-ΔΔCt method).</p><!><p>Gene expressions in subcutaneous adipose tissue after 20-h treatment with tacrolimus compared with no treatment (control) and analysed with a PCR microarray (n = 3)</p><p>In italic are the genes selected for standard qRT-PCR</p><p>Bold values indicate statistical significance: *p < 0.05</p><!><p>Second, a standard qRT-PCR (n = 23) was performed to confirm the gene expressions of the two highest- and the two lowest-expressed genes after tacrolimus treatment (fold change) plus the genes that were statistically different between control and tacrolimus treatment using specific TaqMan gene expression assays [assay on demand: Rho guanine nucleotide exchange factor 11 (ARHGEF11, Hs01121959_m1, FAM); NCK adaptor protein 2 (NCK2, Hs02561903_s1, FAM); LIM domain kinase 1 (LIMK1, Hs00242728_m1, FAM); related protein 2/3 complex, subunit 5 (ARPC5, Hs00271722_m1, FAM); RAB4A, member RAS oncogene family (RAB4A, Hs01106488_m1, FAM); lethal giant larvae homolog 1 (LLGL1, Hs01017181_m1, FAM); hepatocyte growth factor-regulated tyrosine kinase substrate (HGS, Hs00610371_m1, FAM); vasodilator-stimulated phosphoprotein (VASP, Hs01100128_m1, FAM); MAP/microtubule affinity-regulating kinase 2 (MARK2, Hs00997759_m1, FAM); and vesicle-associated membrane protein-associated protein A (VAPA, Hs00427749_m1, FAM), Applied Biosystems, CA, USA]. The relative quantification of mRNA levels was plotted as the fold change (\documentclass[12pt]{minimal} sepackage{amsmath} sepackage{wasysym} sepackage{amsfonts} sepackage{amssymb} sepackage{amsbsy} sepackage{mathrsfs} sepackage{upgreek} \setlength{\oddsidemargin}{-69pt} egin{document}$${{ ext{2}}^{ - \Delta \Delta {C_{ ext{t}}}}}$$\end{document}2-ΔΔCt method) compared with control and normalized to the housekeeping gene GUSB (Hs00939627_m1, VIC) (Applied Biosystems, CA, USA) previously shown to have the lowest coefficient of variation between control and tacrolimus treated samples. Samples were run in duplicates.</p><!><p>Data were expressed relative to control (without treatment) as means ± standard error of the mean (SEM) of measurements performed in duplicate or triplicate. Statistical significance analysis was determined using the two-tailed paired t test. Comparisons were performed within the same individual to minimize confounding variables. The differences were considered significant for p values < 0.05. Statistical analysis was performed using the GraphPad Prism Software (San Diego, CA, USA).</p><!><p>Short-term (75 min) or long-term (20 h) incubation of subcutaneous adipocytes with the different calcineurin inhibitors—tacrolimus (100 nM) and deltamethrin (1 μM)—have similar inhibitory effects on basal and insulin-stimulated glucose uptake [short-term by 10–14% (p < 0.05) and long-term by 30–60% (p < 0.05)], compared with control (Fig. 2a, b). In addition, the long-term incubation of adipocytes with cyclosporin A (100 nM), another calcineurin inhibitor, induced a very similar decrease in glucose uptake compared with tacrolimus (Fig. 2c). The coincubations of adipocytes for 75 min or 20 h with tacrolimus and deltamethrin did not have additive effects on glucose uptake (Fig. 2a, b) suggesting that they may undergo the same mechanism.</p><!><p>Calcineurin inhibitors decrease glucose uptake in human subcutaneous adipocytes. After isolation, adipocytes were incubated for 75 min (a) or 20 h (b, c) with 100 nM of tacrolimus (a–c) and/or 1 μM of deltamethrin (a, b) or 100 nM of cyclosporin A (c) and the glucose uptake was measured in the absence or presence of 25 or 1000 μU/ml of insulin for 1 h. The results were calculated relatively to untreated cell values and represent the means ± SEM of at least 4 subjects. (a) p < 0.05 compared with control and no insulin; (b) p < 0.05 compared with control treated with insulin 25 μU/ml, (c) p < 0.05 compared with control treated with insulin 1000 μU/ml, (d) p < 0.05 compared with tacrolimus and no insulin, (e) p < 0.05 compared with cells treated with deltamethrin and no insulin and (f) p < 0.05 compared with cells treated with deltamethrin and tacrolimus and no insulin with paired t test</p><!><p>The adipogenic differentiated 3T3-L1 cells were also incubated for 20 h with tacrolimus and deltamethrin, but no significant differences were found (data not shown), suggesting that 3T3-L1 cells may not be a good cellular model to represent the effect of calcineurin inhibitors in human adipocytes.</p><!><p>Okadaic acid at 250 nM inhibits protein phosphatases 1 (PP1) and 2A (PP2A), but not calcineurin [30, 31]. Long-term incubation of adipocytes with okadaic acid (250 nM) significantly increased basal and insulin-stimulated glucose uptake by about 50% compared with control (Fig. 3). Addition of tacrolimus decreased basal and insulin-stimulated glucose uptake in the presence of okadaic acid compared with control (Fig. 3) suggesting that okadaic acid and tacrolimus may act through different mechanisms on glucose uptake.</p><!><p>Tacrolimus inhibits okadaic acid-stimulated glucose uptake in human subcutaneous adipocytes. After isolation, adipocytes were incubated for 20 h with 100 nM of tacrolimus and/or 250 nM of okadaic acid and the glucose uptake was measured in the absence or presence of 25 or 1000 μU/ml of insulin for 1 h. The results were calculated relatively to untreated cell values and represent the means ± SEM of at least 4 subjects. (a) p < 0.05 compared with control and no insulin, (b) p < 0.05 compared with cells treated with insulin 25 μU/ml, (c) p < 0.05 compared with cells treated with insulin 1000 μU/ml, (d) p < 0.05 compared with cells treated with tacrolimus and no insulin, (e) p < 0.05 compared with cells treated with tacrolimus and insulin 25 μU/ml, (f) p < 0.05 compared with cells treated with tacrolimus and insulin 1000 μU/ml, (g) p < 0.05 compared with cells treated with okadaic acid and no insulin, (h) p < 0.05 compared with cells treated with okadaic acid and insulin 25 μU/ml with paired t test</p><!><p>Long-term (20 h) incubation of adipocytes with the gene transcription inhibitor, actinomycin D (5 μg/ml), and the protein-translation inhibitor, cycloheximide (25 μM), decreased both basal and 25 μU/ml insulin-stimulated glucose uptake by about 30–50% (p < 0.05), compared with control (Fig. 4). Addition of tacrolimus did not further affect the basal and insulin-stimulated glucose uptake. This suggests that the inhibitory effect of tacrolimus on glucose uptake might be mediated by the regulation of gene and/or protein expression.</p><!><p>Combinatorial effects of tacrolimus with gene transcription inhibitor or protein-translation inhibitor on adipocyte glucose uptake. After isolation, adipocytes were incubated for 20 h with 5 μg/ml of actinomycin D, 25 μM of cycloheximide and/or 100 nM of tacrolimus and the glucose uptake was measured in the presence or absence of 25 or 1000 μU/ml of insulin for 1 h. The results were calculated relatively to untreated cell values and represent the means ± SEM of at least 6 subjects. (a) p < 0.05 compared with control and no insulin, (b) p < 0.05 compared with cells treated with insulin 25 μU/ml, (c) p < 0.05 compared with cells treated with insulin 1000 μU/ml, (d) p < 0.05 compared with cells treated with tacrolimus and no insulin, (e) p < 0.05 compared with cells treated with actinomycin D and no insulin, (f) p < 0.05 compared with cells treated with actinomycin D and tacrolimus and no insulin, (g) p < 0.05 compared with cells treated with cycloheximide and no insulin, (h) p < 0.05 compared with cells treated with cycloheximide and tacrolimus and no insulin with paired t test</p><!><p>To evaluate whether tacrolimus could affect expression of genes involved in GLUT4 vesicle translocation, 92 genes were analysed in SAT treated or untreated with tacrolimus for 20 h (n = 3, Table 2). These genes correspond to proteins that regulate the cellular cytoskeleton and vesicular trafficking, and might potentially be involved in GLUT4 trafficking, especially endocytosis. ARHGEF11 (fold change = 1.29, p = NS) and NCK2 (fold change = 1.24, p = NS) were the two genes with the greatest increase in gene expression after tacrolimus treatment, compared with non-treated tissue, whereas VAPA (fold change = 0.81, p = NS) and MARK2 (fold change = 0.87, p = NS) were the two genes with the greatest decrease in gene expression after tacrolimus incubation, compared with control. Furthermore, LIMK1 (fold change = 1.22, p < 0.05), ARPC5 (fold change = 1.17, p < 0.05), RAB4A (fold change = 1.17, p < 0.01), LLGL1 (fold change = 1.10, p < 0.05) and HGS (fold change = 1.07, p < 0.01) were significantly increased by tacrolimus treatment, while VASP (fold change = 0.89, p < 0.05) was significantly decreased by tacrolimus compared with untreated SAT (Table 2).</p><p>The expression of these genes was confirmed by qRT-PCR using SAT from a larger cohort (n = 23) that was treated in similar conditions (with or without 100 nM tacrolimus for 20 h). Among the genes selected, only ARHGEF11 was shown to be significantly increased by tacrolimus treatment (Table 3), but with a small effect (fold change = 1.06, p < 0.05).</p><!><p>Gene expression in subcutaneous adipose tissue after treatment with tacrolimus compared with no treatment (control) and analysed by qRT-PCR (n = 23)</p><p>ARPC5 actingnalli protein 2/3 complex, subunit 5, ARHGEF11 Rho guanine nucleotide exchange factor (GEF) 11, HGS hepatocyte growth factor-regulated tyrosine kinase substrate, LIMK1 LIM domain kinase 1, LLGL1 lethal giant larvae homolog 1, MARK2 MAP/microtubule affinity-regulating kinase 2, NCK2 NCK adaptor protein 2, RAB4A RAB4A, member RAS oncogene family, VAPA VAMP (vesicle-associated membrane protein)-associated protein A, VASP vasodilator-stimulated phosphoprotein</p><p>*p < 0.05</p><!><p>The specific inhibition of calcineurin by tacrolimus, cyclosporin A and deltamethrin, but not the inhibition of other protein phosphatases 1, 2A and phosphorylated myosin light-chain, reduced glucose uptake in subcutaneous adipocytes, suggesting that calcineurin plays an important role in glucose uptake in human, as well as in rodent adipocytes, as previously described [8, 39]. This effect required at least in part gene transcription and/or protein synthesis, as we described. Analysis on the effect of tacrolimus on expression of genes involved in cytoskeleton function and potentially in GLUT4 trafficking suggests that ARHGEF11 could be a putative downstream gene target of calcineurin signalling associated with GLUT4 trafficking.</p><p>Tacrolimus and cyclosporin A have different biochemical structures, but they inhibit calcineurin through similar mechanisms of action: both bind to immunophilins forming a complex in the cytosol that binds and blocks calcineurin [6, 35]. Tacrolimus binds mainly to FK506-binding proteins (FKBP) and cyclosporin A binds to cyclophylins [6]. Both immunophilins interacts with calcineurin in absence of ligands. Deltamethrin is a type II synthetic pyrethroid insecticide known to be a potent specific inhibitor of calcineurin [32]. This is the first study showing important effects of deltamethrin on human adipocyte glucose uptake. In this study, short- or long-term incubation with tacrolimus, cyclosporin A and the alternative calcineurin inhibitor deltamethrin decreased basal and insulin-stimulated glucose uptake in subcutaneous adipocytes in a similar way, indicating that calcineurin plays an important role for regulation of glucose uptake in human adipocytes. Further evidence comes from the lack of additive effect on glucose uptake when coincubating adipocytes with tacrolimus and deltamethrin, suggesting that their effects on glucose uptake may be mediated by the same mechanism, the inhibition of calcineurin.</p><p>Okadaic acid inhibits PP1 and PP2A at nanomolar concentrations, but has no effect on calcineurin (PP2B) with the concentration used in this work [30, 31]. Okadaic acid is also known to stimulate adipocyte glucose uptake mainly through PP2A inhibition [6, 40] and independently of phosphoinositide 3-kinase activation [41, 42]. Tacrolimus reduced okadaic acid-stimulated glucose uptake to a similar extent as in control, suggesting that okadaic acid and tacrolimus effects on glucose uptake could be mediated through different pathways.</p><p>The degree of inhibition of basal and insulin-stimulated glucose uptake was similar by both calcineurin inhibitors, suggesting that this effect is independent of the early steps of the insulin signalling. These data is in agreement with our previous findings showing that calcineurin inhibitors, tacrolimus and cyclosporin A, decrease glucose uptake in isolated adipocytes by removing GLUT4 from the plasma membrane, via an increased rate of endocytosis [10], but with no apparent defects on insulin signalling including expression and phosphorylation and total protein levels of GLUT4 and GLUT1 [10]. Inhibitory effects of cyclosporin A on adipocyte glucose uptake have also been shown in adipocytes isolated from long-term cyclosporin A-treated rats (up to 9 weeks) or ex vivo treated with cyclosporin A [8, 39]. However, long-term treatment of rats reduces the expression of genes and proteins involved in glucose uptake, such as IRS1 and GLUT4 [39]. Furthermore, overexpression of an activated form of calcineurin in skeletal muscle of mice, induce changes in the expression of genes involved in lipid and glucose metabolism, including GLUT4, with concomitant elevation of insulin-stimulated skeletal muscle glucose uptake [24].</p><p>On the other hand, our data suggest that treatment of the 3T3-L1 adipocyte mouse cell line with tacrolimus or deltamethrin does not affect glucose uptake, which is also in agreement with previous findings [43]. Thus, it seems that the effect of the calcineurin inhibitors on glucose uptake might vary between species and exposure times and therefore it is important to use a human model to study the mechanisms involved in calcineurin inhibition on glucose uptake.</p><p>Coincubation of adipocytes with the gene transcription inhibitor, actinomycin D, or with the protein-translation inhibitor, cycloheximide, and tacrolimus prevented the inhibitory effect of tacrolimus on glucose uptake. This suggests that gene transcription and/or protein translation are required and important for the inhibitory effect of tacrolimus on glucose uptake. Furthermore, the inhibitory effects of the calcineurin inhibitors were more evident with longer (20 h) than with shorter pre-incubation time (15 min), suggesting that the calcineurin inhibitors are more likely to affect gene and/or protein expression rather than acute phosphorylation events. This supports the hypothesis that calcineurin is an important factor involved in glucose uptake in human adipocytes and this effect likely requires gene transcription and protein synthesis.</p><p>In the current analysis, we evaluated effects of calcineurin inhibition on the expression of genes that encode proteins involved in the regulation of cytoskeleton and potentially in GLUT4 trafficking (endocytosis and exocytosis) by gene microarray in SAT explants previously treated with tacrolimus (n = 3). The genes with the greatest increase (ARHGEF11 and NCK2) and greatest decrease (VAPA and MARK2) after tacrolimus treatment, and genes significantly affected by tacrolimus treatment (LIMK1, ARPC5, RAB4A, LLGL1, HGS and VASP) were further analysed using a larger cohort of subjects (n = 23). However, only ARHGEF11 was significantly increased by chronic tacrolimus treatment. Some variants of ARHGEF11 have been associated with type 2 diabetes and schizophrenia in several ethnic populations [44–47]. ARHGEF11 acts as a guanine nucleotide exchange factor for RhoA GTPase and mediates the interaction with the actin cytoskeleton [48]. It is involved in the regulation of G protein signalling, actin cytoskeletal organization [49] and other processes such as insulin signalling [50], insulin secretion [51], and lipid metabolism [52]. Nevertheless, the increase in ARHGEF11 in gene expression of about 6% found in this study is unlikely to have biological relevance and explain the differences shown on glucose uptake. Altogether our data suggest that the inhibitory effects of calcineurin inhibitors on glucose uptake likely requires gene transcription and protein synthesis, but not the expression of the studied genes potentially involved in GLUT4 trafficking and glucose uptake. However, it could include effects on other genes yet unstudied. Hence, more work is needed to find the mechanisms involved in glucose uptake inhibition by calcineurin inhibitors and more importantly identifying the mechanism of calcineurin regulation on glucose uptake in adipose tissue.</p><p>In conclusion, the specific inhibition of calcineurin by tacrolimus, cyclosporin A or deltamethrin, decreased glucose uptake in human subcutaneous adipocytes, suggesting that calcineurin is an important mechanism in the regulation of glucose transport. This effect likely requires gene transcription and protein synthesis, but not via effects on GLUT4 or classical genes known to regulate vesicular trafficking, such as dynamin and RAB proteins. These data suggest that calcineurin is an important regulator of glucose uptake in human adipocytes and its inhibition might contribute to impaired glucose handling in peripheral tissues, as reported with calcineurin modifying therapy in organ-transplanted patients.</p>

PubMed Open Access

Novel sulphonamide benzoquinazolinones as dual EGFR/HER2 inhibitors, apoptosis inducers and radiosensitizers

AbstractA series of sulphonamide benzoquinazolinones 5–18 was synthesized and evaluated for cytotoxic activity against MDA-MB-231 cell line. The compounds showed IC50 ranging from 0.26 to 161.49 µM. The promising compounds were evaluated for their inhibitory profile against epidermal growth factor (EGFR) and HER2 enzymes. Compound 10 showed more potent activity on both EGFR and HER2 than erlotinib (IC50 3.90 and 5.40 µM versus 6.21 and 9.42 µM). The pro-apoptotic activity of 10 was evaluated against caspase-3, Bax, B-cell lymphoma protein 2 (Bcl-2) expression levels, and cell cycle analysis. Compound 10 increased the level of caspase-3 by 10 folds, Bax level by 9 folds, decreased the level of the Bcl-2 by 0.14 and arrested the cell cycle in the G2/M phase. The radio-sensitizing activity of 10 was measured using a single dose of 8 Gy gamma radiation (IC50 decreased from 0.31 to 0.22 µM). Molecular docking was performed on EGFR and HER2 receptors.

novel_sulphonamide_benzoquinazolinones_as_dual_egfr/her2_inhibitors,_apoptosis_inducers_and_radiosen

Introduction<!><!>Introduction<!>Materials and methods<!>General procedure<!>MTT assay<!>In vitro enzymatic activity assay<!>The effect on caspase-3<!>The effect on BAX and Bcl-2 levels<!>Analysis of the cell cycle distribution<!>Radiosensitizing activity<!>Molecular docking<!>Chemistry<!><!>In vitro cytotoxic activity against MDA-MB-231 cell line<!><!>EGFR and HER2 inhibition<!><!>Activation of caspase-3<!><!>Effects on Bcl-2 family proteins<!><!>Effects on Bcl-2 family proteins<!>Cell cycle analysis<!><!>Cytotoxicity against normal breast cells<!><!>Radiosensitizing evaluation<!><!>Molecular docking<!>Docking on EGFR<!><!>Docking on HER2<!><!>Conclusion

<p>The major challenge in cancer therapy is the induction of apoptosis through anticancer agents1–3. Apoptosis is a crucial process in maintaining normal tissue homeostasis in the human body, mediated by signal transduction pathways. The two major apoptotic pathways are extrinsic and intrinsic. The extrinsic pathway is induced by the trans-membrane death receptors, while the intrinsic is through mitochondrial stress caused by DNA damage and heat shock4. Activated caspases are the executioners of apoptosis5. So, more effective therapeutic strategies for better understanding of signaling pathways and molecular targets should be further provided.</p><p>Breast cancer is the world's second leading cause of cancer-related death6. The overexpression of the HER2 enzyme in breast cancer is correlated with poor prognosis and drug resistance7. HER2 belongs to the epidermal growth factor family (EGFR), also called ErbB. It is a member of receptor tyrosine kinases (TKs) involved in signaling pathways controlling angiogenesis, cell differentiation, and proliferation8. The EGFR consists of a subfamily of EGFR (HER1), HER2, HER3, and HER4, that are only expressed at low levels in normal human tissues9. Although most patients with EGFR mutant cancers respond to therapies, the patients develop resistance after an average of one year on treatment10. The resistance to HER2 targeted therapies is associated with the overexpression of EGFR family enzymes11. It is obvious that HER family is interdependent and shows functional redundancy. The blockage of one HER receptor can be compensated by another HER family member9,12. The cross-linking and compensatory activities of the HER family members can provide a strong rationale for co-targeting of both EGFR and HER2 enzymes.</p><p>Molecular hybridization is a simple and effective tool to combine covalently two drug pharmacophores into a single molecule13. Lately, it has been observed that benzo[g]quinazoline and sulphonamides demonstrated profound growth inhibitory activity against different cancer cells and TK enzymes14,15. The quinazoline is a privileged scaffold that constitutes an important class of heterocyclic compounds owing to its varies pharmacological properties16,17. Afatinib, lapatinib, gefitinib, and erlotinib (Figure 1) are the representative drugs in this class in clinical use for targeted anticancer therapies18–21. The use of them has paved the way to develop new quinazoline-based molecules acting as EGFR inhibitors. Also, it is well-known that sulphonamides are strongly related to anticancer activity22,23. They have several targets, most of which are directly connected to oncogenesis24. They proved to exhibit good activity through many mechanisms as carbonic anhydrase24, matrix metalloprotienase25, NADPH reductase26, histone deacetylase27, and PI3K inhibition28.</p><!><p>Examples of dual EGFR/HER2 inhibitors.</p><!><p>In this context, we desire to exploit newer leads with tuneable anticancer activity and low toxicity14,29. A series of substituted benzo[g]quinazolinone benzenesulfonamide hybrids were designed, synthesized, and evaluated as dual EGFR/HER2 inhibitors. The apoptotic activity of the most potent compound was evaluated through the activation of the proteolytic caspase-3, Bax and B-cell lymphoma protein 2 (Bcl2) expression levels, cell cycle analysis, and radio-sensitizing activity. Molecular docking was carried out inside the binding site of EGFR and HER2 receptors in order to confirm their possible mechanism of action.</p><!><p>Melting points were uncorrected and measured on a Gallen Kamp melting point apparatus (Sanyo Gallen Kamp, UK). Precoated silica gel plates (Kieselgel 0.25 mm, 60 F254, Merck, Germany) were used for TLC with a developing solvent system of chloroform/methanol (7:3) and detected by the UV lamp. IR spectra were recorded using FT-IR spectrophotometer (Perkin Elmer, USA). NMR spectra were scanned on an NMR spectrophotometer (Bruker AXS Inc., Switzerland) operating at 500 MHz for 1H and 125.76 MHz for 13C. Chemical shifts are expressed in δ-values (ppm) relative to TMS as an internal standard, using DMSO-d6 as a solvent. Mass spectra were recorded on ISQ LT Thermo Scientific GCMS model (Massachusetts, USA). Elemental analyses were performed on a model 2400 CHNSO analyser (Perkin Elmer, USA). All the values were within ±0.4% of the theoretical values. All reagents were obtained from Sigma-Aldrich of AR grade.</p><!><p>A mixture of 4 (0.383 g, 0.001 mol) and 2-chloro-N-substituted acetamide derivatives (0.001 mol) in dry acetone (30 ml) and anhydrous K2CO3 (0.138 g, 0.001 mol) was stirred at room temperature for 10 h. The mixture was filtered and the product formed was crystallized from ethanol to give 5–18.</p><p>N-(5-Methylisoxazol-3-yl)-2-[(4-oxo-3–(4-sulfamoylphenyl)-3,4-dihydrobenzo[g]quinazolin-2-yl)thio]acetamide (5): Yield, 68%; m.p. 292.4 °C. IR: 3403, 3305, 3190 (NH2, NH), 3095 (arom.), 2980, 2922 (aliph.), 1693, 1665 (2CO), 1631 (CN), 1340, 1161 (SO2). 1HNMR: 2.10 (s, 3H, CH3), 4.21 (s, 2H, S-CH2), 7.02 (s, 1H, CH isoxazole), 7.61–8.20 (m, 10H, Ar-H), 8.81 (s, 2H, SO2NH2), 9.50 (s, 1H, NH). 13CNMR: 18.5, 30.2, 92.7, 119.3, 119.9 (2), 124.1, 126.8 (2), 126.9, 127.7 (2), 128.0, 128.4, 130.6, 131.8, 133.7, 135.8, 145.9, 149.1, 161.2, 162.5, 169.7, 170.2. MS m/z (%): 521 (M+), 383 (100). Anal. Calcd. for C24H19N5O5S2 (521.08): C, 55.27; H, 3.67; N, 13.43. Found: C, 55.49; H, 3.98; N, 13.76.</p><p>2-[(4-Oxo-3-(4-sulfamoylphenyl)-3,4-dihydrobenzo[g]quinazolin-2-yl)thio]-N-(thiazol-2-yl)acetamide (6): Yield, 73%; m.p. 304.0 °C. IR: 3410, 3381, 3111 (NH2, NH), 3100 (arom.), 2970, 2881 (aliph.), 1741, 1693 (2CO), 1601 (CN), 1365, 1163 (SO2). 1HNMR: 4.20 (s, 2H, S-CH2), 7.01–8.20 (m, 12H, Ar-H), 8.82–8.88 (m, 3H, SO2NH2+NH). 13CNMR: 27.3, 113.3, 119.4, 123.3 (2), 124.4 (2), 126.6, 128.1, 128.7 (2), 129.4, 129.9, 131.0, 136.8, 137.9, 139.1 (2), 142.8, 155.4, 161.2, 167.1, 168.2. MS m/z (%): 523 (M+) (0.72), 156 (100). Anal. Calcd. for C23H17N5O4S3 (523.61): C, 52.76; H, 3.27; N, 13.38. Found: C, 52.98; H, 3.48; N, 13.74.</p><p>N-(6-Ethoxybenzo[d]thiazol-2-yl)-2-[(4-oxo-3-(4-sulfamoylphenyl)-3,4-dihydrobenzo[g]quinazolin-2-yl)thio]acetamide (7): Yield, 78%; m.p. 255.9 °C. IR: 3336, 3210, 3169 (NH2, NH), 3059 (arom.), 2978, 2931 (aliph.), 1680, 1678 (2CO), 1602 (CN), 1355, 1161 (SO2). 1HNMR: 1.32 (t, 3H, J = 10 Hz, CH3 ethoxy), 3.90 (s, 2H, S-CH2), 4.12 (q, 2H, J = 10.5 Hz, CH2 ethoxy), 6.99–8.10 (m, 13H, Ar-H), 8.82–8.86 (m, 3H, SO2NH2+NH). 13CNMR: 15.2, 27.3, 63.9, 105.6, 114.1, 119.3, 120.0, 123.4 (2), 126.5, 127.5 (2), 128.1, 128.6 (2), 129.2, 129.8, 130.9, 131.1, 134.0, 136.9, 139.5, 143.0, 144.4, 154.4, 156.5, 161.4, 170.3, 172.9. MS m/z (%): 617 (M+), 383 (100). Anal. Calcd. for C29H23N5O5S3 (617.09): C, 56.39; H, 3.75; N, 11.34. Found: C, 56.68; H, 4.09; N, 11.71.</p><p>N-(6-Nitrobenzo[d]thiazol-2-yl)-2-[(4-oxo-3-(4-sulfamoylphenyl)-3,4-dihydrobenzo[g]quinazolin-2-yl)thio]acetamide (8): Yield, 70%; m.p. 278.3 °C. IR: 3363, 3274, 3220 (NH2, NH), 3071 (arom.), 2929, 2840 (aliph.), 1710, 1695 (2CO), 1597 (CN), 1566, 1336 (NO2), 1336, 1165 (SO2). 1HNMR: 4.30 (s, 2H, S-CH2), 7.51–8.20 (m, 13H, Ar-H), 8.71 (s, 2H, SO2NH2), 8.90 (s, 1H, NH). 13CNMR: 31.1, 119.1, 119.3, 121.8 (2), 122.4 (2), 126.0, 127.4 (2), 128.8, 129.5 (2), 129.8 (2), 131.1 (2), 139.1 (3), 143.0 (2), 157.6 (2), 161.0, 169.2 (2). MS m/z (%): 618 (M+) (4.78), 124 (100). Anal. Calcd. for C27H18N6O6S3 (618.04): C, 52.42; H, 2.93; N, 13.58. Found: C, 52.78; H, 3.21; N, 13.82.</p><p>2-[(4-Oxo-3-(4-sulfamoylphenyl)-3,4-dihydrobenzo[g]quinazolin-2-yl)thio]-N-(5-(trifluoromethyl)-1,3,4-thiadiazol-2-yl)acetamide (9): Yield, 81%; m.p. 257.0 °C. IR: 3444, 3284, 3246 (NH2, NH), 3091 (arom.), 2910, 2835 (aliph.), 1715, 1695 (2CO), 1600 (CN), 1400, 1174 (SO2). 1HNMR: 4.20 (s, 2H, S-CH2), 7.63–8.10 (m, 10H, Ar-H), 8.81 (s, 2H, SO2NH2), 11.83 (s, 1H, NH). 13CNMR: 26.9, 119.4 (2), 123.5 (2), 126.5, 127.4 (2), 128.1, 128.6, 129.2 (2), 129.8 (2), 131.1, 136.9, 139.4, 145.7, 156.2 (2), 161.4 (2), 172.4. MS m/z (%): 592 (M+) (2.11), 350 (100). Anal. Calcd. for C23H15F3N6O4S3 (592.03): C, 46.62; H, 2.55; N, 14.18. Found: C, 46.30; H, 2.21; N, 13.93.</p><p>N-(3,4-Dimethylphenyl)-2-[(4-oxo-3-(4-sulfamoylphenyl)-3,4-dihydrobenzo[g]quinazolin-2-yl)thio]acetamide (10): Yield, 77%; m.p. 232.8 °C. IR: 3416, 3289, 3143 (NH2, NH), 3063 (arom.), 2948, 2842 (aliph.), 1718, 1691 (2CO), 1631 (CN), 1390, 1160 (SO2). 1HNMR: 2.15 (s, 3H, CH3), 2.18 (s, 3H, CH3), 4.12 (s, 2H, S-CH2), 7.03–8.21 (m, 13H, Ar-H), 8.80 (s, 2H, SO2NH2), 10.31 (s, 1H, NH). 13CNMR: 19.2, 20.0, 27.9, 117.2, 119.4, 120.9, 123.4 (2), 126.6, 127.4 (2), 128.1, 128.8, 129.4 (2), 129.9 (2), 130.0, 131.0, 131.7, 136.8 (2), 136.9, 137.1, 145.8, 155.4, 161.3, 165.6. MS m/z (%): 544 (M+) (1.24), 310 (100). Anal. Calcd. for C28H24N4O4S2 (544.12): C, 61.75; H, 4.44; N, 10.29. Found: C, 62.04; H, 4.69; N, 10.56.</p><p>N-(2,5-Dimethylphenyl)-2-[(4-oxo-3-(4-sulfamoylphenyl)-3,4-dihydrobenzo[g]quinazolin-2-yl)thio]acetamide (11): Yield, 78%; m.p. 279.3 °C. IR: 3388, 3269, 3212 (NH2, NH), 3051 (arom.), 2982, 2844 (aliph.), 1693, 1655 (2CO), 1600 (CN), 1328, 1157 (SO2). 1HNMR: 2.02 (s, 3H, CH3), 2.21 (s, 3H, CH3), 4.20 (s, 2H, S-CH2), 7.18–8.34 (m, 13H, Ar-H), 8.86 (s, 2H, SO2NH2), 11.16 (s, 1H, NH). 13CNMR: 19.3, 22.6, 30.2, 110.7, 119.2, 119.9 (2), 122.7, 124.6, 125.2 (2), 127.0, 127.4, 128.6, 128.9 (2), 129.0 (2), 129.9, 130.9, 133.8, 134.6, 135.9, 136.5, 145.2, 155.8, 161.4, 169.0. MS m/z (%): 544 (M+) (2.88), 340 (100). Anal. Calcd. for C28H24N4O4S2 (544.12): C, 61.75; H, 4.44; N, 10.29. Found: C, 61.62; H, 4.11; N, 10.07.</p><p>N-(2,6-Dimethylphenyl)-2-[(4-oxo-3-(4-sulfamoylphenyl)-3,4-dihydrobenzo[g]quinazolin-2-yl)thio]acetamide (12): Yield, 89%; m.p. 300.5 °C. IR: 3361, 3269, 3132 (NH2, NH), 3049 (arom.), 2972, 2871 (aliph.), 1699, 1653 (2CO), 1600 (CN), 1355, 1155 (SO2). 1HNMR: 1.78 (s, 6H, 2CH3), 4.22 (s, 2H, S-CH2), 7.54–8.32 (m, 13H, Ar-H), 8.81–8.85 (m, 3H, SO2NH2+NH). 13CNMR: 15.0 (2), 31.1, 119.4, 123.3 (2), 126.6 (2), 127.4 (4), 128.1, 128.8 (2), 129.4, 129.8, 131.0 (2), 136.9 (2), 139.1 (2), 142.9, 145.4, 155.0, 161.3, 166.2. MS m/z (%): 544 (M+) (1.80), 340 (100). Anal. Calcd. for C28H24N4O4S2 (544.12): C, 61.75; H, 4.44; N, 10.29. Found: C, 61.42; H, 4.18; N, 9.99.</p><p>N-(2-Methyl-4-nitrophenyl)-2-[(4-oxo-3-(4-sulfamoylphenyl)-3,4-dihydrobenzo[g]quinazolin-2-yl)thio]acetamide (13): Yield, 85%; m.p. 293.8 °C. IR: 3441, 3358, 3240 (NH2, NH), 3057 (arom.), 2978, 2916 (aliph.), 1697, 1664 (2CO), 1627 (CN), 1539, 1340 (NO2), 1357, 1161 (SO2). 1HNMR: 2.04 (s, 3H, CH3), 4.30 (s, 2H, S-CH2), 7.53–8.25 (m, 13H, Ar-H), 8.81 (s, 2H, SO2NH2), 10.03 (s, 1H, NH). 13CNMR: 18.3, 27.4, 105.2, 119.4, 121.2, 123.4 (2), 123.7, 123.9, 126.7 (2), 127.4, 128.2, 128.8 (2), 129.4, 129.9, 131.8, 136.8, 134.7, 139.3, 142.8, 143.8, 145.9, 155.3, 161.3, 167.0. MS m/z (%): 575 (M+) (8.50), 79 (100). Anal. Calcd. for C27H21N5O6S2 (575.09): C, 56.34; H, 3.68; N, 12.17. Found: C, 56.72; H, 3.77; N, 12.50.</p><p>N-(2,4-Dioxo-1,2,3,4-tetrahydropyrimidin-5-yl)-2-[(4-oxo-3-(4-sulfamoylphenyl)-3,4-dihydrobenzo[g]quinazolin-2-yl)thio]acetamide (14): Yield, 76%; m.p. 280.0 °C. IR: 3409, 3261, 3217 (NH2, NH), 3100 (arom.), 2972, 2841 (aliph.), 1741, 1701, 1681, 1653 (4CO), 1582 (CN), 1396, 1159 (SO2). 1HNMR: 4.13 (s, 2H, S-CH2), 5.20 (s, 1H, CH uracil), 7.51–8.22 (m, 10H, Ar-H), 8.75 (s, 2H, SO2NH2), 9.42 (s, 2H, 2NH), 10.81 (s, 1H, CONHCO). 13CNMR: 28.2, 78.2, 119.3, 123.9 (2), 126.6, 127.4 (2), 128.2, 128.5 (2), 129.3, 129.8, 131.0, 136.8 (2), 139.0, 142.7, 146.1, 155.2, 161.4 (2), 165.4 (2). MS m/z (%): 550 (M+) (4.50), 79 (100). Anal. Calcd. for C24H18N6O6S2 (550.07): C, 52.36; H, 3.30; N, 15.26. Found: C, 52.72; H, 3.67; N, 15.50.</p><p>N-(1,3-Dimethyl-2,6-dioxo-1,2,3,6-tetrahydropyrimidin-4-yl)-2-[(4-oxo-3-(4-sulfamoylphenyl)-3,4-dihydrobenzo[g]quinazolin-2-yl)thio]acetamide (15): Yield, 83%; m.p. 307.7 °C. IR: 3410, 3334, 3171 (NH2, NH), 3086 (arom.), 2963, 2831 (aliph.), 1708, 1691, 1678, 1645 (4CO), 1618 (CN), 1398, 1155 (SO2). 1HNMR: 3.41 (s, 6H, 2CH3), 4.13 (s, 2H, S-CH2), 6.58 (s, 1H, CH uracil), 7.50–8.22 (m, 10H, Ar-H), 8.81 (s, 2H, SO2NH2), 11.30 (s, 1H, NH). 13CNMR: 26.4, 28.6, 31.2, 73.8, 119.3, 123.4 (2), 126.7, 127.4 (2), 128.1, 128.8 (2), 129.4, 129.9, 131.0, 136.8, 139.1 (2), 142.7, 145.8, 155.1, 158.5, 161.3, 166.6, 170.1. MS m/z (%): 578 (M+) (3.42), 89 (100). Anal. Calcd. for C26H22N6O6S2 (578.10): C, 53.97; H, 3.83; N, 14.52. Found: C, 53.68; H, 3.59; N, 14.31.</p><p>2-[(4-Oxo-3-(4-sulfamoylphenyl)-3,4-dihydrobenzo[g]quinazolin-2-yl)thio]-N-(pyrazin-2-yl)acetamide (16): Yield, 81%; m.p. 205.7 °C. IR: 3429, 3325, 3246 (NH2, NH), 3060 (arom.), 2959, 2825 (aliph.), 1695, 1681 (2CO), 1629 (CN), 1338, 1157 (SO2). 1HNMR: 4.21 (s, 2H, S-CH2), 7.53–8.42 (m, 13H, Ar-H), 8.87 (s, 2H, SO2NH2), 9.24 (s, 1H, NH). 13CNMR: 28.9, 119.4, 123.4 (2), 126.6, 127.4 (2), 128.0, 128.8 (2), 129.4, 129.9 (2), 131.0, 136.7, 136.8 (2), 139.1, 142.7, 145.9, 149.2, 155.2, 161.3, 167.5. MS m/z (%): 518 (M+) (1.09), 129 (100). Anal. Calcd. for C24H18N6O4S2 (518.08): C, 55.59; H, 3.50; N, 16.21. Found: C, 55.28; H, 3.19; N, 16.03.</p><p>N-(Naphthalene-1-yl)-2-[(4-oxo-3-(4-sulfamoylphenyl)-3,4-dihydrobenzo[g]quinazolin-2-yl)thio]acetamide (17): Yield, 78%; m.p. 241.6 °C. IR: 3412, 3296, 3166 (NH2, NH), 3059 (arom.), 2981, 2860 (aliph.), 1741, 1658 (2CO), 1627 (CN), 1348, 1161 (SO2). 1HNMR: 4.33 (s, 2H, S-CH2), 7.45–8.24 (m, 17H, Ar-H), 8.86 (s, 2H, SO2NH2), 10.34 (s, 1H, NH). 13CNMR: 31.3, 108.2, 119.0, 121.2, 121.4, 122.4 (2), 123.4, 126.0 (2), 126.2, 126.5, 126.7 (2), 127.4, 128.1, 128.4 (2), 128.6, 129.5, 129.9, 131.1, 136.9 (2), 141.0 (2), 143.8, 157.9, 161.4, 166.9. MS m/z (%): 566 (M+) (7.12), 114 (100). Anal. Calcd. for C30H22N4O4S2 (566.11): C, 63.59; H, 3.91; N, 9.89. Found: C, 63.78; H, 4.02; N, 10.12.</p><p>N-(9,10-Dioxo-9,10-dihydroanthracen-2-yl)-2-[(4-oxo-3-(4-sulfamoylphenyl)-3,4-dihydrobenzo[g]quinazolin-2-yl)thio]acetamide (18): Yield, 85%; m.p. 256.7 °C. IR: 3442, 3279, 3134 (NH2, NH), 3061 (arom.), 2976, 2833 (aliph.), br. 1693, 1670 (4CO), 1629 (CN), 1332, 1161 (SO2). 1HNMR: 4.15 (s, 2H, S-CH2), 7.23–8.16 (m, 17H, Ar-H), 8.87 (s, 2H, SO2NH2), 10.43 (s, 1H, NH). 13CNMR: 31.1, 114.2, 119.6, 119.9 (2), 123.0, 125.6 (2), 125.7, 126.8 (2), 126.9, 127.1, 127.5, 128.8 (2), 128.9, 129.7, 130.4 (2), 131.0, 131.9 (2), 132.9, 133.3, 133.7, 136.0, 143.0, 144.1, 160.6, 160.9, 167.4, 187.1 (2). MS m/z (%): 646 (M+) (5.29), 128 (100). Anal. Calcd. for C34H22N4O6S2 (646.10): C, 63.15; H, 3.43; N, 8.66. Found: C, 63.41; H, 3.70; N, 9.02.</p><!><p>MDA-MB-231 breast cancer cells and 184A1 normal breast cells of American Type Culture Collection were obtained from VACSERA, Egypt. Cells were cultured using Dulbecco's Modified Eagle's Medium (Invitrogen/Life Technologies) supplemented with 10% FBS (Hyclone), 10 µg/mL of insulin, and 1% penicillin–streptomycin. Cells were seeded in 96-well plate with cells density 1.2–1.8 × 10,000 cells/well, in a volume of 100 µL complete growth medium + 100 µL of the tested compound per well and the plate was incubated for 24 h before the MTT assay. The cell layer was rinsed with 0.25% (w/v) trypsin, 0.53 mM EDTA solution, incubated for 2 h, then the absorbance was measured at a wavelength of 570 nm30. IC50 was calculated according to the equation of Boltzmann sigmoidal concentration-response curve using Graph Pad Prism 5.</p><!><p>EGFR and HER2 kinase kits were purchased from Invitrogen. EGFR (PV3872), 0.200 mg/mL and HER2 (PV3366), 0.192 mg/mL were used. ATP solution and a kinase/peptide mixture were prepared. The plate was incubated for 1 h at room temperature. About 5 mL of the developing solution was added to each well. The plate was incubated for 1 h and then read by ELISA Reader (PerkinElmer, USA). Every experiment was repeated three times. Data represented as means ± SE from three independent experiments. Curve fitting was performed using Graph Pad Prism 5.</p><!><p>The Quantikine-Human active caspase-3 immunoassay (R&D Systems Inc., USA) is used to measure the active caspase-3 level, by adding 100 µL of the standard diluent to the zero standard wells. Cover and incubate for 2 h at room temperature. Add 100 μL of caspase-3 (active) detection antibody solution into each well except the chromogen blank. Incubate for 1 h then add 100 µL anti-rabbit IgG HRP working solution to each well and incubate for 30 min. The absorbance of each well was measured at 450 nm.</p><!><p>Cells were grown in RPMI 1640 containing 10% foetal bovine serum at 37 °C, stimulated with the compounds to be tested for Bax, and lysed with cell extraction buffer. This lysate was diluted in the standard diluent buffer over the range of the assay and measured for human active Bax and Bcl2 content according to the reported method31.</p><!><p>To determine the effect of compound 10 and erlotinib on the cell cycle distribution of MDA-MB-231 cell line; cell cycle analysis was performed using the CycleTEST™ PLUS DNA Reagent Kit (Becton Dickinson Immunocytometry Systems, San Jose, CA, USA). Control cells with known DNA content (peripheral blood mononuclear cells) were used as a reference point for determining the DNA index for the test samples. The cells were stained with propidium iodide stain following the procedure provided by the kit then incubated at room temperature for 5 min in the dark and run on the DNA cytometer. Cell cycle distribution was calculated using CELLQUEST software (Becton Dickinson Immunocytometry Systems, San Jose, CA, USA).</p><!><p>Irradiation was performed at the National Center for Radiation Research and Technology (NCRRT), Egyptian Atomic Energy Authority (EAEA), using gamma cell-40 (137Cs) source. Compound 10 was selected to be re-evaluated for the in vitro antiproliferative activity in combination with γ-irradiation using MTT assay. Cells were incubated with compound 10 in molar concentrations of 0.01, 0.1, 1.0, and 10 µM. After 2 h, cells were subjected to a single dose of 8 Gy of γ-radiation at a dose rate of 0.758 rad/s for 17.73 min, and then the anti-proliferative activity was measured 48 h after irradiation. The IC50 of the tested compounds was calculated after irradiation.</p><!><p>Molecular modeling was performed using the Molecular Operating Environment (MOE, 10.2008) software. The protein data bank files (PDB: 1M17 and 3RCD) were selected for this purpose. Water molecules were ignored and hydrogen atoms were added. The co-crystallized ligands in both receptors were re-docked into the active site for method standardization. The structure of compound 10 was drawn on ChemDraw and copied as smiles to MOE. Energy minimizations were performed for compound 10 using MMFF94X force field and the partial charges were calculated. Docking of 10 inside the active site of the enzyme to generate one hundred conformations. Top-scored conformation was captured by 2D and 3D images.</p><!><p>The synthesis of the target compounds 5–18 was described in Scheme 1. The starting compound 4-(2-mercapto-4-oxobenzo[g]quinazolin-3(4H)-yl) benzenesulfonamide 429was prepared from the reaction of 3-amino-2-naphthoic acid 3 with 4-isothiocyanatobenzenesulfonamide 222. The reaction of 4 with 2-chloro-N-substituted acetamide derivatives in dry acetone containing an equimolar amount of anhydrous K2CO3 yielded the appropriate N-(substituted)-2-[(4-oxo-3-(4-sulfamoylphenyl)-3,4-dihydrobenzo[g]quinazolin-2-yl)thio]acetamides 5–18 (Scheme 1). IR spectra of 5–18 revealed NH, CH aliphatic, and CO bands at their specified regions. 1H-NMR spectra of 5–18 revealed two singlets at 3.90–4.33 ppm attributed to the CH2 and 8.81–11.83 ppm attributed to the NH protons and the disappearance of SH singlet at 2.01 ppm of 4. 13C-NMR spectra of 5–18 exhibited two downfield signals attributed to the C–S and CO carbons. The 1HNMR spectrum of 5 revealed two singlets at 2.10 and 7.02 ppm corresponding to the CH3 and CH isoxazole. 13C-NMR of 5 showed an up-field signal at 18.5 ppm due to CH3. 1HNMR of 7 showed triplet at 1.32 ppm and quartet at 4.12 ppm due to the ethoxy group. 13C-NMR of 7 showed two up-field signals at 15.2 and 63.9 ppm due to the ethoxy carbons. IR of 8 revealed the NO2 peaks at 1566 and 1336 cm−1. 13C-NMR of 9 showed a signal at 119.4 ppm for the CF3 carbon. 1HNMR of 10–12 revealed singlets at the range of 1.78–2.21 ppm due to the 2CH3 and 13C-NMR showed two signals in the range of 15.0–22.6 ppm. IR of 13 showed the NO2 peaks at 1539 and 1340 cm−1. 1HNMR of 13 revealed a singlet at 2.04 ppm for the CH3. IR of 14 and 15 showed 4 CO peaks in their specified regions. 1HNMR of 14 revealed two singlets at 5.20 and 10.81 ppm corresponding to the CH uracil and CONHCO, respectively. 1HNMR of 15 revealed two singlets at 3.41 and 6.58 ppm due to 2CH3 and CH uracil, respectively. 13C-NMR of 15 showed two signals at 28.6 and 31.2 ppm for the 2CH3. 13C-NMR of 18 revealed a signal at 187.1 due to the 2CO of anthraquinone.</p><!><p>Synthesis of the benzo[g]quinazolinone derivatives 4–18.</p><!><p>The in vitro cytotoxicity of the targeted compounds 5–18 was measured using MTT assay against human breast cancer cell line (MDA-MB-231), and erlotinib was used as the reference drug. Table 1 indicates that compounds 5–18 showed IC50 values in the range of 0.26–161.49 µM, in comparison to erlotinib (IC50= 0.48 µM). Compounds 10, 11, 13, 14, 16, and 18 were more active than the reference drug, with IC50 values in the range of 0.26–0.40 µM. The 9,10-dioxo-9,10-dihydroanthracene derivative 18 was the most active followed by the 2,5-dimethyl phenyl 11, the 3,4-dimethyl phenyl 10, the pyrazinyl 16, the 2,4-dioxopyrimidinyl 14, and the 2-methyl-4-nitrophenyl derivative 13. The EGFR inhibitory profile of the synthesized compounds 5–18 was measured and reported in Table 1. The results showed that the tested compounds exhibited inhibitory activity towards EGFR, ranging from 72.90% to 8.71%. The most cytotoxic compounds showed the highest inhibitory profile. Compound 14 showed the highest percentage inhibition followed by 11, 10, erlotinib, 13, and 18 (percentage inhibition ranging from 72.90% to 67.26%).</p><!><p>The cytotoxic activity and percentage inhibition of compounds 5–18 on EGFR against MDA-MB-231 breast cancer cell line.</p><p>*The values represent the mean ± SE of three independent experiments.</p><!><p>The IC50 values of the compounds showing the highest percentage inhibition towards EGFR were determined. Compounds 10, 11, 13, 14, and 18 were screened on both EGFR and HER2 enzymes in reference to erlotinib. From the results in Table 2, we can conclude that all the tested compounds together with erlotinib have better inhibitory activity and lower IC50 on EGFR than HER-2 enzyme except for compound 18 (IC50 ranges from 2.55 to 10.20 µM towards EGFR versus 3.20–31.31 µM towards HER2). The 3,4-dimethyl phenyl derivative 10 was more potent than erlotinib on both EGFR and HER2 (IC50 3.90 and 5.40 µM versus 6.21 and 9.42 µM, respectively). Compound 11 was the most potent towards EGFR (IC50 2.55 µM), while compound 18 was the most potent towards HER2 (IC50 3.20 µM).</p><!><p>IC50 of compounds 10, 11, 13, 14, and 18 against EGFR and HER2 enzymes.</p><p>*The values represent the mean ± SE of three independent experiments.</p><!><p>Caspase-3 is a member of the cysteine-aspartic acid protease family that plays a crucial role in apoptosis32. It is an inactive proenzyme converted to the active form through caspases 8, 9, and 1033. Caspase-3 is activated in the apoptotic cell by both extrinsic (death ligand) and intrinsic (mitochondrial) pathways34 by cleaving multiple proteins in the cells leading to cell death35. The effect of compound 10 on caspase-3 was evaluated in reference to erlotinib. Compound 10 showed an increase in the level of the active caspase 3 by 10 folds compared to the control cells. While erlotinib increases the level of caspase 3 by 9 folds (Table 3).</p><!><p>The effect of compound 10 on the level of caspase-3.</p><!><p>The Bcl-2 family plays a central role in tumour progression or inhibition of mitochondrial intrinsic apoptotic pathway36. The pro-apoptotic Bax is essential for cell apoptosis. However, the anti-apoptotic Bcl-2 overexpression enhances cell survival by suppressing apoptosis37. Thus, the balance between these two different proteins determines the cell fate38,39. Increments in the Bax/Bcl2 ratio trigger a cascade of caspases that leads to the activation of caspase 3; the apoptosis executioner40. In this study, MDA-MB-231 breast cells were treated with compound 10 and their effect on the expression levels of Bcl2, and Bax were illustrated in Table 4.</p><!><p>The effect of compound 10 on Bax/Bcl2 expression levels.</p><!><p>Compound 10 and erlotinib boosted the level of the pro-apoptotic protein Bax by 9 and 11 folds, respectively, compared to the control cells. On the other hand, they markedly reduced the levels of the anti-apoptotic proteins Bcl2 by 0.14 and 0.07 folds, respectively. The results showed that both compound 10 and erlotinib markedly boosted the Bax level and down-regulated Bcl2 level proving their pro-apoptotic effect.</p><!><p>Cell cycle progression is responsible for normal cell growth and proliferation. DNA damage can lead to either DNA repair or cell death through apoptosis. The condition of the cells is assessed at certain checkpoints that act as control mechanisms to ensure the proper cell division. Cell cycle checkpoints are the G1 (restriction), the S (metaphase), and the G2/M41. The role of anticancer agents is to stop the cell division at these checkpoints. Treatment with the anticancer agents can determine at which phase apoptosis occurs in the cell cycle. In this study, MDA-MB-231 cells were treated with compound 10 at its IC50. The results in Table 5 indicate that compound 10 arrested the cell cycle at the G2/M phase when compared to the untreated control (17.52% and 6.44%, respectively; Figure 2(A,C)). While erlotinib arrested the cell cycle at the G2/M phase by 24.81% (Figure 2(B)). Also, the cell population in G1 and S phases decreases after treatment (49.36% and 18.28% versus 69.55% and 23.04%, respectively) in case of compound 10 compared to control. While in the case of erlotinib, the cell population in G1 and S phases markedly decreases after treatment to (41.55% and 16.31%, respectively). These results reveal that in MDA-MB-231 cells, cell cycle arrest occurs in the G2/M phase in the case of compound 10 and erlotinib.</p><!><p>The effect of inhibitors on the phases of the cell cycle (A) compound 10, (B) erlotinib, and (C) control MDA-MB-231 cells.</p><p>The effect of compound 10 and erlotinib on the phases of cell cycle.</p><!><p>The cytotoxicity of compound 10 compared to erlotinib was measured against 184A1 normal breast cells using MTT assay in order to determine the relative safety of compound 10 on normal tissues. Compound 10 and erlotinib showed mild cytotoxic effect with an IC50 of 84.5 and 101.9 µM, respectively (Table 6).</p><!><p>The cytotoxicity of compound 10 and erlotinib on 184A1 normal breast cells</p><!><p>Most cancer patients receive radiation therapy during the course of treatment. Gamma rays are high energy radiation used in therapy to shrink tumours and kill malignant cells by damaging their DNA either directly or indirectly through free radicals formation. The major drawback of radiation therapy is that they cannot differentiate between normal and cancerous cells. So, the use of radiotherapy and selective chemotherapy are required in order to eliminate normal tissue damage42.</p><p>The cytotoxicity of compound 10 was measured on MDA-MB-231 cell line before and after being subjected to a single dose of 8 Gy γ-irradiation. The ability of compound 10 to enhance the cell-killing effect of γ-irradiation was examined. The results showed that compound 10 is able to sensitize the cancer cells to the lethal effects of gamma radiation (Table 7).</p><!><p>IC50 of compound 10 on MDA-MB-231 cells before and after being subjected to a single dose of 8 Gy γ-radiation.</p><!><p>Molecular docking was performed using MOE 10.2008 inside the active site of EGFR (PDB ID: 1M17)43 and HER2 receptors (PDB ID: 3RCD)44. In order to rationalize the biological results and to gain insight into the SAR of the target compounds, an attempt to interpret the observed enzymatic activities of the tested compounds on the basis of the ligand-protein interactions was done. The enzymatic activity of EGFR and HER2 inhibitors depends on the ability of the compound to properly dock into the binding site and to establish interactions with the key amino acids. Accordingly, the active compound in this study should attain the same binding mode observed for the ligand.</p><!><p>The EGFR catalytic domain consists of an N-terminal lobe, which consists mainly of one α-helix and C-helix. The C-terminal lobe is mainly α-helical, and a short strand termed the hinge region connects the two lobes45. The N-(3-ethynylphenyl)-6,7-bis(2-methoxyethoxy)quinazolin-4-amine (erlotinib) is the co-crystallized ligand inside the EGFR receptor46. Erlotinib was located well in the ATP pocket and interacts with Met 769 by a hydrogen bond of 2.70 A° length, and hydrophobic interactions with Leu 694, Leu 820, Lys 721, and Thr 766 (hinge region; Figure 3). Compound 10 was docked in the active site of the enzyme and bound in the same manner as the ligand. Compound 10 binds with energy score (S = −9.88 Kcal/mol) and interact with the active site through Met 769 by a hydrogen bond of 0.85 A°, Cys 773 with the CO of quinazolinone and Phe 699 with the phenyl ring of the acetamide through a π–π interaction (Figure 4, Table 8).</p><!><p>2 D and 3 D ligand interactions of erlotinib inside the active site of 1M17.</p><p>2 D and 3 D interaction maps of compound 10 inside the active site of 1M17.</p><p>Docking results of compound 10 inside 1M17 and 3RCD active sites.</p><!><p>The crystal structure of HER2 complexed with TAK-285 (PDB ID: 3RCD) showed that Ala 751, Leu 800, Met 801, Leu 852, and Asp 863 are the key amino acids. The X-ray co-crystallized structure of TAK-285 with HER2 demonstrated that it binds to the ATP pocket through an H-bond with Met 801 and to the hinge region by a series of hydrophobic interactions with Leu 852, Leu 726, Phe 1004, Thr 798, Thr 862, and Leu 78547 (Figure 5). Compound 10 pursued the similar binding pattern in HER2 with Met 801 by the SO2 of the sulphonamide group, Thr 862 and Asp 863 by the CO of the acetamide and Lys 753 with the N-1 of quinazolinone (Figure 6, Table 8).</p><!><p>2 D and 3 D interaction maps of TAK-285 inside the active site of 3RCD.</p><p>2 D and 3 D interaction maps of 10 inside the active site of 3RCD.</p><!><p>An array of new 3,4-dihydrobenzo[g]quinazolinone derivatives containing sulphonamide moiety was designed, synthesized, and evaluated for their cytotoxic effect on MDA-MB-231 breast cancer cell line. The tested compounds showed IC50 values ranging from 0.26 to 161.49 µM on MDA-MB-231. The new compounds were tested for their inhibitory profile against EGFR and HER2 enzymes. The 3,4-dimethyl phenyl derivative 10 was more potent than erlotinib on both EGFR and HER2 (IC50 3.90 and 5.40 µM versus 6.21 and 9.42 µM, respectively). The 2,5-dimethyl phenyl derivative 11 was the most potent towards EGFR, while the anthraquinone derivative 18 was the most potent towards HER2. Compound 10 was evaluated as an apoptosis inducer through the activation of the proteolytic caspase-3, Bax and Bcl-2 expression levels, and cell cycle analysis. It was found that compound 10 increases the level of caspase-3 by 10 folds, Bax level by 9 folds, decreases the level of Bcl-2 by 0.14 folds and arrested the cell cycle in the G2/M phase. The radiosensitizing activity of 10 was measured on MDA-MB-231 cell line after being irradiated by a single dose of 8 Gy. IC50 decreased from 0.31 to 0.22 µM after being irradiated. Docking of 10 inside the active site of EGFR and HER2 receptors revealed that it binds in the same manner as that of the co-crystallized ligand.</p>

PubMed Open Access

Unravelling three-dimensional adsorption geometries of PbSe nanocrystal monolayers at a liquid-air interface

The adsorption, self-organization and oriented attachment of PbSe nanocrystals (NCs) at liquid-air interfaces has led to remarkable nanocrystal superlattices with atomic order and a superimposed nanoscale geometry. Earlier studies examined the NC self-organization at the suspension/air interface with time-resolved in-situ X-ray scattering. Upon continuous evaporation of the solvent, the NC interfacial layer will finally contact the (ethylene glycol) liquid substrate on which the suspension was casted. In order to obtain structural information on the NC organization at this stage of the process, we examined the ethylene glycol/NC interface in detail for PbSe NCs of different sizes, combining in-situ grazing-incidence smalland-wide-angle X-ray scattering (GISAXS/GIWAXS), X-ray reflectivity (XRR) and analytical calculations of the adsorption geometry of these NCs. Here, we observe in-situ three characteristic adsorption geometries varying with the NC size. Based on the experimental evidence and simulations, we reveal fully three-dimensional arrangements of PbSe nanocrystals at the ethylene glycol-air interface with and without the presence of rest amounts of toluene.

unravelling_three-dimensional_adsorption_geometries_of_pbse_nanocrystal_monolayers_at_a_liquid-air_i

<!>Results and discussion<!>Methods

<p>T he self-assembly of semiconductor nanocrystals (NCs) on a liquid substrate has been pioneered by Murray et al. 1 , who confined the formation of NC superlattices at a liquid-air interface. By drop casting a suspension of NCs in an apolar, volatile solvent on top of diethylene glycol, the NCs are forced to adsorb at the liquid-air interface and self-organize into large-area superlattices upon solvent evaporation. The exact superlattice structure that forms depends on interactions between the NCs 2,3 and the interaction of the particles with the interface. Recently, this method has been extended to form atomically connected NC solids through a process called oriented attachment: PbSe NCs adsorbed at the liquid/air interface align their atomic lattices and fuse via epitaxial {100}/{100} connections into an atomically coherent superlattice of one NC in thickness [4][5][6] . By tuning the synthesis conditions, various superlattice geometries could be obtained, such as a square superlattice geometry, where the NCs have a {100} facet pointing upwards, and a honeycomb superlattice geometry, where the NCs have a {111} facet pointing upwards 6,7 . In both superlattice allotropes, the NCs attach via their {100} facets 8 .</p><p>In situ synchrotron X-ray scattering techniques are nowadays being used more often to resolve the dynamics of self-assembly processes of lead chalcogenide NCs [9][10][11][12] . Recently, we 13 and others 14 have studied the formation mechanism of two-dimensional (2-D) square PbSe NC superlattices in situ and showed that the NCs undergo a remarkable sequence of phase transitions. Upon solvent evaporation, the NCs adsorb at the liquid-air interface and form a monolayer with hexagonally packed particles. Upon ligand detachment from the {100} facets, the NCs align crystallographically with a {100} facet pointing upwards. During this process, the hexagonal geometry of the superlattice is gradually changed to a square geometry. Finally, the particles attach epitaxially via a necking process, where surface atoms move to form the connection between the NCs. There are still a large number of open questions regarding the described self-assembly process: how is the honeycomb superlattice formed? At which moment in the process is there a bifurcation toward either a square or a honeycomb geometry, and which factors decide the geometry? Which factors determine the amount of disorder on both atomic and NC length scales and how can we reduce this 10,[15][16][17][18][19][20][21] ? A recently introduced simulation model to predict the self-assembly of NCs at fluid-fluid interfaces provided fundamental insights that will help to answer these questions 22,23 .</p><p>The adsorption geometry of the NCs at a liquid-air interface 24 should have a large impact on the final geometry of the NC superlattices. The pathway and the formation mechanism of silicene-like structures 7 with two distinctly different heights of two subsets of NCs remains a mystery. Ultimately, the NC interactions with the two liquids will determine the way the NCs adsorb at the interface 22,23,25 . Moreover, the interfacial adsorption of the NCs will also create a capillary distortion of the liquid. Soligno et al. showed that cubes can create a hexapolar distortion of a liquid-liquid interface that could, in principle, induce directional "capillary" interactions between the NCs 26 , but very probably these directional forces are too weak to dominate the assembly 22,23 . Recently, the relation between the surface chemistry and the shape of PbSe NCs has been studied by Peters et al. 8 . They showed that chemisorption and surface reconfiguration results in a transformation of the NC shape from a truncated nanocube with rough surface facets to a truncated octahedron with larger and smooth {111} facets.</p><p>To study the three-dimensional (3-D) adsorption geometry of our NC monolayers, we perform three different X-ray scattering techniques quasi-simultaneously, which are shown schematically in Fig. 1. For the in situ grazing-incidence small-angle X-ray scattering (GISAXS) and grazing-incidence wide-angle X-ray scattering (GIWAXS) experiments, the incoming X-ray beam glances the liquid-air interface at an incident angle, α i , of 0.14 o , i.e., the critical angle for total external reflection of the X-ray photons at 22 keV for PbSe. The GISAXS pattern is recorded in the forward direction and reveals information on the periodicity and order in the NC monolayer. The atomic diffraction is recorded at a detector placed closer to the sample under a higher angle. The collected GIWAXS signal allows us to obtain the crystallographic orientation of the NCs at the liquid-air interface. The before mentioned techniques are complemented with specular X-ray reflectivity (XRR) measurements. Using the doublecrystal deflection diffractometer at the ID10 beamline of the European Synchrotron Radiation Facility (ESRF), the angle of incidence is varied and the intensity of the specular beam is recorded on a one-dimensional (1-D) detector on the diffractometer arm. The scattering vector for this specular reflection only has a component in the vertical direction, i.e., q z , which allows us to obtain information on the average density profile of the NC monolayer in the direction perpendicular to the liquid-air interface. When a monolayer of NCs is present at the liquid-air interface, the signal is modulated owing to constructive and destructive interference upon scattering and so-called Kiessig fringes can be observed. These fringes modulate with a period 2π/ Δ in reciprocal space, where Δ is the thickness of the densified colloidal system formed at the interface, for instance, a NC monolayer 27 .</p><p>The focus of this work is on the nearly last stage of the process of PbSe superlattice formation, i.e., when the NC monolayer rests on the ethylene glycol (EG) substrate, possibly before atomic attachment has taken place. We adsorb PbSe NCs with different sizes at the EG-air interface and study the in-plane NC geometry with GISAXS and their crystallographic orientation with GIWAXS. Furthermore, we extend these two techniques with specular XRR measurements 28,29 , which allowed us to obtain the density profiles of the NC monolayers in the direction X-ray beam z x y Fig. 1 Schematic of the in situ GISAXS, GIWAXS, and XRR experiments performed at ID10 of the ESRF. The GISAXS/GIWAXS experiments, which reveal the order of the NCs in the plane of the monolayer, are done using an angle of incidence of 0.14 o with respect to the liquid surface. A detector is placed in the forward scattering direction to collect the GISAXS data, which reveals information on the inter-nanocrystal order. A second detector is placed at a higher angle and closer to the sample to collect the GIWAXS signal, which reveals information on the crystallographic orientation of the NCs with respect to the NC monolayer. Upon completion of solvent evaporation, the specular XRR is collected using a dual crystal deflection scheme; the angle of incidence is varied and the intensity of the specular reflection is recorded. The inset shows a schematic of a PbSe truncated nanocube.</p><p>perpendicular to the EG/air interface. The data presented here is (also) an important step forward in the understanding the formation mechanism of 2-D superlattices, but the main goal of this work goes beyond this: we unravel the 3-D adsorption profile of PbSe NCs at fluid-fluid interfaces using a unique and novel combination of different experimental and numerical techniques.</p><!><p>In situ synchrotron X-ray scattering. We synthesized NCs with varying sizes in the range of 4-10 nm, as outlined in Supplementary Methods. From the literature, it is already known that the truncation of the NCs is decreased when increasing the NC size 30 . The size of the {100} facets increases at the expense of the {110} and {111} facets giving more cubic-shaped NCs. We divided the used NCs in three size ranges: small-sized NCs with a diameter <5.5 nm, medium-sized NCs with diameter in the range 5.5-7.6 nm, large-sized NCs with diameter in the range 8.2-9.1 nm. Transmission electron microscopic (TEM) images and absorption spectra of the synthesized NCs can be found in Supplementary Methods (see Supplementary Table 1 and Supplementary Figures 1-3 for a summary of the NCs used throughout this study).</p><p>The NC dispersion in toluene is dropcasted inside the liquid cell on EG, which acts as an immiscible liquid substrate. To the EG substrate, we add 100 μL of a 31.7 μM oleic acid solution in EG solution for two reasons: first, to increase the wettability of the toluene droplet with NCs on the EG. The second reason is that addition of oleic acid to the EG sub-phase has proven to stop oriented attachment of the NCs 4 . The single NC adsorption geometry is key to understand the behavior of these NCs at the EG-air interface, which is why we attempt to block the epitaxial fusing of the NC monolayers.</p><p>First, we present and discuss GISAXS data, see Fig. 2. The scattering patterns are acquired after 2 h of solvent evaporation from the casted NC suspensions, so that most toluene should be evaporated from the NC dispersion. The GISAXS signal for the small NCs, shown in Fig. 2a, shows Bragg rods at positions of 0.94 nm −1 , 1.62 nm −1 , and 1.88 nm −1 in the horizontal scattering direction q y with a full width at half maximum (FWHM) of 0.06 nm −1 . The peak positions of 1:√3:2 indicate that the NCs are ordered in a 2-D hexagonal lattice with an NC center-to-center distance of 7.7 ± 0.4 nm, roughly the NC diameter plus interdigitated oleic acid ligands.</p><p>The GISAXS pattern of the medium-sized NCs, depicted in Fig. 2b, shows Bragg rods at 0.86 nm −1 and 1.71 nm −1 in the q y direction with an FWHM of 0.08 nm −1 . The 1:2 relative peak positions indicate that there is a preference for NC ordering in one dimension, e.g., the formation of linear structures. The NC center-to-center distance is calculated to be 7.3 ± 0.7 nm.</p><p>The relatively small center-to-center distance of the NCs compared to the NC diameter is not fully understood. Hanrath and co-workers showed through molecular dynamics simulations that interparticle distances as small as 0.5 nm can be achieved when the ligand density of oleic acid on PbSe is low enough 31 . This is a hard parameter to quantify during these experiments, as the ligand density on the NC surface likely changes throughout the self-assembly process. For the large NCs, the GISAXS signal depicted in Fig. 2c shows Bragg rods at 0.75 nm −1 , 1.52 nm −1 , and 2.18 nm −1 in the q y direction with an FWHM of 0.05 nm −1 . Again the 1:2:3 relative peak positions indicate that there is a preference for NC ordering in one dimension. The center-tocenter distance is calculated to be 8.4 ± 0.5 nm. This is smaller than the center-to-center distance expected for two NCs with two layers of oleic acid between them (length oleic acid ∼1.8 nm 32 ). Further discussion follows below. The GISAXS experiments do not imply that we are monitoring linear structures, as a small degree of disorder and deviation of a 90°(square NC ordering) or 60°(hexagonal NC ordering) bond angle removes the in-plane correlation peaks 13 . The occurrence of oriented attachment cannot be excluded, which would explain the decreased centerto-center distances of the NCs, which will be discussed further in the sections on the GIWAXS and XRR data. The full GISAXS q z q z q y q y q y dataset on all NC sizes can be found in Supplementary Information (Supplementary Figs. 4-6, a summarized version is presented in Supplementary Table 2). The crystallographic orientation of the NCs is obtained by measuring the diffraction from their atomic lattices, which is measured simultaneously with the GISAXS data and is presented in Fig. 3. We calculated the expected GIWAXS pattern for NCs having a [001] axis pointing perpendicular to the liquid-air interface in Fig. 3a. Figure 3b-d show the GIWAXS patterns corresponding to the small-, medium-, and large-sized NCs, respectively. For the small NCs, we observe powder rings in the GIWAXS pattern, which means that the NCs do not have a preferential orientation at the liquid-air interface. Either the NCs are still freely rotatable or the ensemble of NCs have random, static orientations, which will average out in a ring in the GIWAXS pattern.</p><p>The GIWAXS pattern of the medium-sized NCs, shown in Fig. 3c, shows a series of well-defined diffraction spots. An azimuthal intensity trace over the reflection originating from the {222} planes shows that it has an orientation of 36.0°with respect to the liquid-air interface, corresponding to NCs having a {001} facet pointing upwards. The FWHM of the reflection, which is an indication of the degree of rotational freedom of the <100> axis perpendicular to the liquid surface the NCs still have at the liquid-air interface, is 6.5°. The GIWAXS pattern of the large NCs, shown in Fig. 3d, matches the calculated GIWAXS pattern very well. It shows that the NCs have the same orientation as the medium-sized NCs, i.e., a {001} facet pointing upwards. The variation in orientations is slightly smaller, as the FWHM of the 222 reflection is 6.1°. This means that the large NCs have less rotational freedom at the liquid-air interface. We show the width of the Bragg peaks along the azimuthal directions for the 222, 420, 422, and 311 reflections versus the NC diameter in Supplementary Table 3. The apparent decrease in peak width with increasing NC size indicates an increase in crystallographic alignment of the NCs. The GIWAXS patterns of all NCs are presented in Supplementary Figures 7-10.</p><p>The orientation of the NCs is further verified by looking at intensity traces in the 2θ direction along the scattering horizon (azimuthal angle ~1°), which we presented for the same samples in Supplementary Information (Supplementary Fig. 19). For the medium-and large-sized NCs, only reflections originating from {hk0} PbSe lattice planes are observed along the scattering horizon. These particular atomic planes are oriented perpendicular to the liquid-air interface when the NCs have a [001] axis pointing upwards and hence scatter horizontally. Previously observed superlattices from PbSe NCs have been shown to show attachment of their {100} facets 4,7 . We are able to verify whether or not the NCs are attached by looking at the FWHM of the 400 reflection in the horizontal scattering direction and estimating the crystalline size of the NCs. The results are presented in Table S10. Since the diameters obtained using the Scherrer equation are not significantly larger than the NC diameters determined by TEM, attachment does not occur. Possibly, there are some regions on the sample with and without attached NCs. We have also taken this into account in the analysis of the reflectivity data, which will be discussed below.</p><p>NC alignment was also observed by Van der Stam et al., who recently showed that, upon addition of oleic acid, 11 nm polyhedral ZnS bifrustum NCs align atomically with their {002} facet pointing upwards at the EG-air interface 33,34 . Hanrath et al. showed that large cubic PbSe NCs with an edge length of 25 nm do align a [111] axis perpendicular to the toluene-air interface, but there the NCs formed 3-D body-centered cubic superlattices 25 . This is also the orientation that PbSe NCs are required to have before silicene-type honeycomb superlattices can form 7,22 . We will show in the section below that the full adsorption geometry of the PbSe NCs studied here is determined by an interplay between the adsorption energy, mostly dictated by the NC size, and the degree of truncation of the NCs.</p><p>To obtain quantitative information in the direction perpendicular to the liquid-air interface, we performed specular XRR measurements. When one measures the specular reflection, the scattering vector q only has a component perpendicular to the interface z. The interference of the X-ray photons reflected from a stratified surface will give rise to periodic intensity oscillations, the Kiessig fringes. The periodicity of the oscillations contains information on the layer thickness, whereas the scattered intensity depends on the averaged scattering density profile across the interface. This means that one can fit the acquired reflectivity curve to get detailed information on a materials density gradient in the z-direction, e.g., how NCs adsorb at liquid-air or liquid-liquid interfaces.</p><p>The adsorption behavior of NCs at interfaces has already been studied using a combination of X-ray techniques. Vorobiev et al. studied self-assembly of iron oxide nanoparticles of different sizes at the water-air interface under different surface pressures to show the optimal conditions for making monolayer films of the particles 35 . XRR has been used recently in a combination with GISAXS to study various types of Au nanoparticles on different substrates 36,37 . Kosif et al. were able to produce rigid NC films at the water-air interface by connecting gold nanoparticles together with thiol-group containing linker molecules 38 . They studied the film structure under different surface pressures using grazingincidence X-ray diffraction and XRR to see how much force was necessary to buckle the NC film.</p><p>We continue the discussion here with a comparison of the XRR data from the small-and medium-to-large-sized PbSe NCs presented in Fig. 4 (the full set of XRR measurements are presented in Supplementary Figure 11-19, and fitted parameters are displayed in Supplementary Tables 4-9). For the fitting of the data, we apply a recursive fitting procedure based on a Parratt formalism 39 . The fit takes into account the position and orientation of the particle with respect to the liquid-air interface, the degree of truncation of the NCs, the thickness of the ligand corona around the NC, and several density and roughness parameters (see Supplementary Information for an extensive discussion on the XRR data analysis). We observe three characteristic density profiles perpendicular to the EG-air interface, which can be categorized into those for (1) smallsized NCs, (2) large-and medium-sized NC monolayers, and (3) large-and medium-sized NC monolayers, where a certain degree of buckling is observed; sometimes, one NC is displaced downwards by roughly half an NC diameter. The observation of these buckled layers is striking, since their presence is not readily deduced from the GISAXS data, which shows the necessity of the specular XRR measurements to fully characterize the adsorption behavior of these NCs.</p><p>Figure 4a shows a representative XRR curve from the small NCs as blue dots, with the corresponding fit as a red solid line. The clear oscillations of the signal only show one period, proving that we truly are looking at a monolayer of NCs. The value of the real part of the refractive index of each layer j, δ j , is proportional to the electron density of that layer. This is plotted in Fig. 4d as the blue curve; the yellow curve is the first derivative of this density profile, which we use to identify the position of the particle with respect to the liquid-air interface. The middle of the scattering length density (SLD) profile corresponds to a mean value for the center of mass of the NCs above the liquid-air interface level, which lies 3.8 nm above the EG-air interface, i.e., the NCs do not penetrate into the EG. This can be rationalized by speculating that toluene is still adsorbed in the NC ligand corona (a more detailed argument follows in the theory section). Figure 4b presents a representative XRR curve from the medium-/large-sized NCs as blue dots, with the corresponding fit as a red solid line. Again, only one periodicity of the XRR signal is observed, confirming that we are measuring a NC monolayer again. The corresponding SLD plot, shown in Fig. 4d, shows that the center of mass of the large NC monolayer is sticking out 3.0 nm above the EG-air interface. This is one out of two typically observed adsorption geometries for the medium/large sized NCs. This also explains the easy transfer to solid substrates using a Langmuir-Schaefer-type stamping technique. In Fig. 4c, we show the XRR curve of the second typically observed adsorption geometry, which we roughly observed in 20% of our experiments on the medium-/large-sized NCs. The density profile is displayed in Fig. 4f. Two NCs layers are observed in the SLD profile, where one layer is displaced by roughly half an NC diameter downwards with respect to the EG-air interface. This observed "buckling" of the NC monolayers resembles the expected buckling of the honeycomb superlattice 5,7 , with the exception that these NCs are oriented with a {100} facet pointing upwards from the liquid-air interface. Here there is truly the penetration of some NCs through the interface. We note that the observed NC adsorption geometry could differ strongly from the initial adsorption geometry at the toluene-air interface. Future in situ experiments will be focused on obtaining a stable (i.e., non-evaporating) toluene-air interface to see how the NCs adsorb and align during the early stages of the self-assembly process.</p><p>Analytical calculations on the NC adsorption geometry. We corroborate and clarify the presented experimental data with analytical calculations to predict the equilibrium adsorption geometry of the PbSe NCs at the toluene/air and interface. An internal energy γA is associated with a fluid-fluid interface of surface area A, with γ the fluid-fluid surface tension. A microparticle or nanoparticle can reduce the internal energy of the system by adsorbing at the fluid-fluid interface, since this reduces A. Treating the two fluids as homogeneous, the total internal energy associated with a particle staying at the interface between the two fluids (say fluid 1 and fluid 2) is 40,41</p><p>where A 1 and A 2 are the surface areas of the particle in contact with fluid 1 and fluid 2, respectively, and γ 1 and γ 2 is the surface tension of the particle surface with fluid 1 and fluid 2, respectively. In Eq. ( 1), we assume that the fluid-fluid interface is flat far away from the particle, so the fluid pressure-volume terms can be neglected in U (see Supplementary Information). Gravity is not included in Eq. ( 1), since this is negligible for nanoparticles compared to the surface energy terms. At equilibrium, the particle stays at the interface with the position and orientation that minimize U 40 . Such a minimum-U orientation is fundamental for directing the self-assembly of NCs at fluid-fluid interfaces 22 . To improve our understanding of the self-assembly of PbSe NCs in honeycomb and square superstructures 4,6,7,13 , we theoretically predict the equilibrium orientation of a single PbSe NC at the various fluid-fluid interfaces involved in the self-assembly experiments, that is toluene-air and EG-air. In principle, also the EG-toluene interface is involved in the experiments, but the NCs are not expected to adsorb at this interface (see Supplementary Information).</p><p>We numerically compute U for various orientations φ, ψ of a NC at a flat fluid-fluid interface, where the z = 0 plane corresponds to the interface without NC, φ is the polar angle of the NC vertical axis with respect to the z-axis, and ψ is the Euler internal angle of the NC around its vertical axis, as shown in Fig. 5a. As NC surface, we consider both a rhombicuboctahedron and a cantellated rhombicuboctahedron (the exact definition Fig. 4 Specular XRR reveals three characteristic density profiles perpendicular to the ethylene glycol-air interface. a Representative XRR curve from a monolayer of small PbSe NCs. The red line is the best fit of the data, in which we approximate the NCs as truncated nanocubes. b Representative XRR curve from a monolayer of medium-/large-sized PbSe NCs. The red line is the best fit of the data. c For large and medium-sized PbSe NCs, we observe a second type of adsorption geometry, where the nanocrystals slightly buckled, i.e., two layers on top of each other. d SLD plot from the fit for the small NCs displayed in a. The blue curve represents the density profile; the yellow curve is the first derivative of this density profile. The nanocrystal monolayer floats on top of the ethylene glycol. e SLD plot from the fit displayed for the medium-/large-sized NCs displayed in b. Again, the NC monolayer floats on top of the ethylene glycol. f SLD plot from the fit for the second adsorption geometry observed for medium-and large-sized PbSe NCs displayed in c. As can be seen, there are two NC layers, which are displaced by roughly half a particle diameter from each other in the direction perpendicular to the ethylene glycol-air interface. is given in Supplementary Information). These shapes are approximations meant to capture the key geometrical features of PbSe NCs, which typically have a highly truncated cubic shape when their size is around or <6 nm, and a slightly truncated cubic shape for larger sizes. For each considered orientation φ, ψ of the NC, the unknowns A, A 1 , A 2 in Eq. ( 1) are computed with the numerical method introduced by Soligno et al. [40][41][42] , where a simulated annealing algorithm is used to find the equilibrium shape of the fluid-fluid interface for the given NC orientation (see Supplementary Information for technical details on our calculations). Our approach ensures that the effects due to the capillary deformations induced by the adsorbed NC in the equilibrium shape of the fluid-fluid interface are included in U. As shown in ref. 17 , neglecting capillary deformations (the so-called Pieranski approximation 34,[43][44][45] , corresponding to assuming that the fluid-fluid interface is flat everywhere also in the presence of the NC) can lead to qualitative errors in the predictions for the NC equilibrium orientation.</p><p>Our theoretical results are presented in Fig. 5. For convenience, we show E(φ, ψ) instead of U(φ, ψ), where</p><p>is just U shifted by a constant (see Supplementary Information), such that E = 0 corresponds to the NC desorbed from the interface and fully immersed in fluid 2. In Eq. ( 2), Α 0 is the area of the fluid-fluid interface when no NC is present, and the parameter θ is Young's contact angle, defined by Young's Law, that is 34</p><p>To represent a PbSe NC (capped by oleic acid ligands) at an EG/air and toluene/air interface, we set cosθ = 0.05 and cosθ = 0.64, respectively (see in Supplementary Information a justification for these numbers). In Fig. 5, E is shown both in units of Σγ (with Σ the NC total surface area) and, for a given value of the NC size and of γ, in units of k B T r (with k B the Boltzmann constant and T r room temperature).</p><p>A highly truncated cubic NC with size 6 nm, see Fig. 5c, e, is bonded by an energy well ∼−340 k B T r at the EG/air interface and ∼−30 k B T r at the toluene/air interface. Therefore, the NC prefers to adsorb at the EG/air interface. However, in the experiments, we expect the NC to first adsorb at the toluene/air interface, since the EG/air interface does not exist until all the toluene around an NC is evaporated (and the NCs do not adsorb at the EG/toluene interface, see Supplementary Information). Once the NC is at the toluene/air interface, it remains there until the toluene is completely evaporated, since the energy barrier to spontaneously desorb from the toluene/air interface (∼30 k B T r ) is too high at room temperature. The SLD plots of Fig. 4d, e show that the small-and medium-sized NCs did not adsorb in the EG phase. We hypothesize that a thin layer of toluene was still on top of the EG phase when the X-ray scattering measurements were done, such that the NCs were confined at the toluene-air interface (or at least, toluene was still present in the ligand corona of the NCs). Indeed, if all the toluene was evaporated, then the NCs should have been adsorbed at the EG/air interface and be half-immersed in the EG phase and half-immersed in the air, as shown in the insets of Fig. 5c, which is in contrast with the experimental results, see Fig. 4d, e. Possibly, the evaporation of the last amount of toluene from inside the ligand corona of the NCs is much slower than the evaporation of the bulk toluene liquid.</p><p>Figure 5e shows that the 6-nm highly truncated NC has multiple metastable orientations at the toluene/air interface, separated by energy barriers of a few k B T r , suggesting that the NC has essentially random orientation at this interface, in agreement with the experimental results in Fig. 3b, c.</p><p>A slightly truncated cubic NC with size 8 nm, see Fig. 5b, d, is bonded by an energy well ∼−840 k B T r at the EG/air interface and ∼−100 k B T r at the toluene/air interface. Therefore, for the same argument previously illustrated, we expect the NCs to remain at the toluene/air interface as long as the toluene is not evaporated. The SLD plots of Fig. 4f show that some of the large-sized NCs are adsorbed in the EG phase, while others remain on top of it. This suggests that toluene was still present close to the NCs on top of the EG phase, while it was essentially evaporated around the NCs that managed to adsorb in the EG phase.</p><p>The larger-sized NCs experimentally are found oriented with a {100} facet parallel to the interface plane, see Fig. 3d. For the NCs still adsorbed at the toluene-air interface, i.e., not immersed in the EG phase, this orientation matches with our predictions for a slightly truncated cubic NC at a toluene-air interface, see Fig. 5d.</p><p>For the NCs partially immersed in the EG phase, so staying at the EG-air interface, we would expect as equilibrium orientation a {111} facet parallel to the interface plane, see Fig. 5b. However, a metastable orientation with a {100} facet parallel to the interface is also predicted for a slightly truncated NC at the EG-air interface, see Fig. 5b, with energy barrier ∼5 k B T r . This seems to explain why the larger-sized NC adsorbed in the EG phase are also found oriented with a {100} facet parallel to the interface: since this is their orientation when they are adsorbed at the toluene-air interface, they remain trapped in the same orientation when the toluene completely evaporates and they adsorb at the EG/air interface.</p><p>Combining the data from the GISAXS, GIWAXS, and XRR experiments, and strengthening them with our calculations, we can create a size-and shape-dependent 3-D model of how PbSe NCs adsorb at the EG-(toluene)-air interface. A schematic representation is shown in Fig. 6. The smaller NCs (with sizes ≤5.5 nm) are oriented randomly, which results from the smaller adsorption energy and the reduced size of the NC facets compared to larger NCs. As the NCs grow in size, their adsorption energy is increased, the truncation of the particles is reduced, and they get a more cubic shape with larger {100} facets. The delicate interplay between adsorption energy, which is directly related to the NC size, and truncation parameter of the NCs gives the specific adsorption geometries that we determined experimentally.</p><p>Future X-ray scattering experiments should focus on different aspects of the self-assembly process: (1) Varying the ligand density and different ratios of facet sizes, which are intimately related, to identify key parameters driving the formation of either honeycomb or square 2-D superlattices. (2) Unraveling the adsorption geometry of the PbSe NCs at the toluene-air interface. These experiments are more difficult, as the evaporation of toluene has to be stopped to perform the reflectivity measurements. A solution could lie in using specially developed liquid cells, which controllably saturate the atmosphere inside the cell with solvent vapor 46 .</p><p>To summarize, we studied the adsorption behavior of PbSe NCs at the EG-air interface with and without the presence of residual amounts of toluene, using a combination of X-ray scattering techniques. Furthermore, we combine GISAXS and GIWAXS with specular XRR measurements to obtain full 3-D pictures of how these NCs adsorb at the EG-(toluene)-air interface. We show that larger PbSe NCs align crystallographically with a [001] axis perpendicular to the liquid-air interface, due to an interplay between adsorption energy and the degree of truncation of the NCs. These experiments were corroborated with analytical calculations of the NC equilibrium adsorption geometry to rationalize the NC behavior at the various fluid-fluid interfaces as a function of NC size and truncation. The adsorption geometry of the NCs in the early stages of oriented attachment are expected to have great impact on the atomically connected 2-D superlattices. The experiments presented throughout this work show that it is possible to reveal 3-D arrangements of NC monolayers adsorbed at liquid-air interfaces, from the NC to atomic length scales. This particular combination of techniques increases our understanding of self-assembly processes on liquid substrates and will help guide the fabrication of novel 2-D superstructures.</p><!><p>NC synthesis. The PbSe NCs used for the oriented attachment experiments in this study were prepared using an adapted method described by Steckel et al. 47 . Details can be found in Supplementary Methods.</p><p>In situ synchrotron X-ray scattering. The in situ X-ray scattering experiments under grazing incidence were performed at beamline ID10 of the ESRF, Grenoble. The energy of the incident X-ray beam was set at 22.0 keV, above the Pb and Se. We observed little to no beam damage at this X-ray energy. We optimized the grazing angle to 0.14°for the best signal-to-noise ratio on both GIWAXS and GIWAXS detectors. The scattering was recorded by two Pilatus detectors. The GIWAXS patterns were recorded on a Pilatus 300 K detector with 619 × 487 pixels, each 172 × 172 μm 2 in size, positioned approximately 25 cm from the sample. The GISAXS patterns were recorded on a Pilatus 300K-W detector with 1475 × 195 pixels, each 172 × 172 μm 2 in size, positioned 0.988 m from the sample. Before drop casting the dispersion of NCs on top of the EG substrate, the X-ray beam was aligned to the surface. The XRR patterns were collected on a 1-D Mythen detector. The self-assembly of the NCs was performed in a home-built liquid cell, which can be flushed with nitrogen repeatedly to lower the oxygen and water levels (Supplementary Fig. 21). A Teflon Petri dish (Ø 64 mm) was filled with 28 mL of EG. To EG, we added 100 μL of a 31.7 μM oleic acid solution in EG. The cell was then flushed five times with vacuum/nitrogen cycles. Next, the PbSe NC solution (1 mL; 1.2 × 10 −6 mol L −1 for all solutions) was deposited on top of the liquid substrate. A photograph of the experimental setup, penetration depth calculations and additional GISAXS simulations can be found in Supplementary Figures 21-23.</p>

Nature Communications Chemistry

Insights into the Structural Patterns of the Antileishmanial Activity of Bi- and Tricyclic N-Heterocycles

Influence of various structural patterns in a series of novel bi- and tricyclic N-heterocycles on the activity against Leishmania major and Leishmania panamensis has been studied and compounds that are active in the low micromolar region have been identified. Both quinolines and tetrahydrooxazinoindoles (TOI) proved to have significant antileishmanial activities, while substituted indoles were inactive. We have also showed that a chloroquine analogue induces Leishmania killing by modulating macrophage activation.

insights_into_the_structural_patterns_of_the_antileishmanial_activity_of_bi-_and_tricyclic_n-heteroc

Introduction<!>Preparation of the compounds used in the antileishmanial tests<!>Screening of compounds against L. major and L. panamensis promastigotes<!>Antileishmanial activity of compounds on intracellular amastigotes<!>Discussion<!>Conclusions<!>Materials and methods<!>General procedure for the synthesis of 1,2-oxazines<!>Mice<!>Parasites<!>Promastigote Inhibition Assay<!>Amastigote Inhibition Assay<!>Macrophage cytotoxicity assay<!>Statistical Analysis

<p>Leishmaniasis is a tropical disease with a significant global health burden that is caused by Leishmania flagellate protozoa.1 Twenty Leishmania species that are pathogenic for humans have been identified. They are transmitted by several sand flies species, Phlebotomus in the Old World and Lutzomya in the New World.2 Leishmaniasis has a wide spectrum of clinical manifestations depending on the Leishmania species and the immunological status of the host. These include localized and diffused cutaneous leishmaniasis, mucocutaneous form and visceral disease.3</p><p>Leishmania species differ in virulence, vectors preferences and geographic distribution. However, all species have a similar life cycle involving a motile, flagellated stage in the midgut of vector (promastigote) and an intracellular non-motile stage (amastigote) in host macrophages.4 Macrophages are the most important effector cells in Leishmania infection, and their appropriate activation is required to eliminate the parasite. The destruction of the parasite by macrophages depends on the production of nitric oxide (NO), tumor necrosis factor (TNF), interleukin (IL)-1 among other mediators, and is negatively affected by a variety of factors including IL-10.5</p><p>Current treatments for leishmaniasis are based in drugs whose specific mechanisms of action are poorly understood. The most used drugs (e.g. pentavalent antimonials, pentamidine, amphotericin B and miltefosine) require lengthy treatments and have high toxicity and serious side effects.6 Furthermore, resistance to some of these drugs has been reported for a diversity of Leishmania strains.7 Consequently, the search for new drugs for the treatment of the disease that carries a multitude of health and socioeconomic problems in endemic countries is an enduring challenge.</p><p>Several families of compounds have been tested for antileishmania activity.8 Both, natural products and synthetic compounds have been recently identified as promising leads against leishmaniasis. Particularly promising scaffolds include quinolines and indoles. Antiparasitic,9 antibacterial,10 antineoplastic,11 and antiviral12 activities have been reported for quinoline derivatives. For instance, naphthylisoquinoline alkaloids showed low micromolar activities against Leishmania donovani,13 as well as against intracellular amastigote stage of Leishmania major.14 Similarly, the naturally-occurring hypocrellin A was found to be more active against L. donovani in vitro than amphotericin B and pentamidine.15 Synthetic antileishmanial 1,4-anthraquinones have also been described.16 Recently, abietane-type diterpenoids have emerged as potent antileishmanial agents.17</p><p>In this study we evaluated the effects on intracellular amastigotes and promastigotes of L. panamensis and L. major of three families of bi- and tricyclic N-heterocycles: tetrahydrooxazinoindoles (TOIs) 1, quinolines 2, and indoles 3 (Fig. 1). Since quinolines, e.g. amodiaquine, chloroquine, mefloquine, and primaquine have been successfully used as antimalarials, they may hold promise as a new class of antileishmanial agents. Indeed, amodiaquine and its basic side chain-modified analogues have been found to have a significant antileishmanial activity.18 In addition, 7-chloro-4-quinolinyl hydrazones have shown strong activity against the intracellular parasite.19 We have recently developed an efficient method of synthesis of 2-susbtituted quinolines from quinoline N-oxides that allows for a simple access to various substituted quinolines, including 2-substituted derivatives of amodiaquine and chloroquine.20 Using this method, we have prepared a series of quinolines bearing substituents in 2, 4, 5, 7 and 8 positions, including novel amodiaquine and chloroquine analogues. The 1,2-oxazine moiety in tetrahydrooxazinoindoles (TOIs) 1 resembles pyridine in amodiaquine and chloroquine. In addition to structural similarity, quinolines and 1,2-oxazines both contain a weekly basic nitrogen atom that may be important for the antileishmanial activity. Hence, it was of interest to compare these two classes of basic N-heterocycles. Our recently described method of tetrahydrooxazinoindoles (TOIs) synthesis offered a facile entry to the novel 1,2-oxazine-containing framework in racemic and enantioselective fashion.21 Since TOI framework contains an indole moiety, a series of substituted indoles were also prepared, and their antileishmanial activities have been compared with the quinolines and TOIs.</p><p>Our results show that most of the tested compounds are more active against intracellular amastigotes than on promastigotes of both Leishmania species assayed. L. panamensis amastigotes appear to be more sensitive to our active compounds than L. major amastigotes. Interestingly we also found that one of the compounds inhibited the production of IL-10 by macrophages infected with either L. panamensis or L. major.</p><p>Further, we describe herein that some TOIs are competent antileishmanial agents, and their activity is related to the presence of the 1,2-oxazine moiety. This result, along with the data on the antileishmanial activity of several substituted quinolines, provide important insights into the antileishmanial activity of N-heterocycles and point to the potential of 1,2-oxazines22 as new structural frameworks for biomolecular applications.23</p><!><p>The tetrahydrooxazinoindole compounds 4 bearing substituents in positions 3, 4, 4a, 5, 6, and 9 were accessed using the inverse electron demand [4+2] cycloaddition reaction of indoles 5 with transient nitrosoalkenes that were generated in situ from α-chlorooximes 6 (Scheme 1).21 The indoles 5 were prepared by means of N-alkylation and a reductive C3-alkylation (See Supporting Information for details).</p><p>The enantiomerically enriched TOIs were prepared using Cu(DM-Binap)OTf as a catalyst, while racemic TOIs 4 were prepared using Cu(rac-Binap)OTf as a catalyst, in the presence of silver carbonate as a chloride scavenger and a base. The TOI products were obtained in good to excellent yields. The 4-chloro-substituted TOIs were isolated as single diastereomers, in line with the previously observed results.21</p><p>X-ray crystal structures were elucidated for compounds 7–11 (Fig. 2), aiding in the confirmation of the structure of the TOI products and their precursors. Interestingly, although TOI compounds 9 and 10 were prepared with >90% ee, only a small amount of racemic crystals was obtained, indicating that the racemate is less soluble than both enantiomers, as previously observed for other scalemic mixtures.24 The indoles, quinolines and TOIs were selected based on the combination of ready synthetic availability and structural diversity.</p><!><p>Libraries of TOIs, indoles and quinolines were tested against L. major and L. panamensis extracellular promastigotes. To evaluate the effect of compounds on promastigotes of both Leishmania species we performed a first screening at a fixed concentration of 10 μM. Any compound inducing a growth inhibition of 50% or more, was further tested using a four concentration points including 1, 3, 10 and 30 μM. Viability of promastigotes was assessed by using an ATP-bioluminescence assay after 24 hours of incubation with the compound. 25 Our results showed that none of the tested compounds were active for L. major promastigotes (Tables 1 and 2), whereas four compounds from the TOI library (10% of the total number of the tested compounds) were effective in the killing of L. panamensis promastigotes, with IC50 values ranging from 8 to 12 μM (Table 2). Several compounds, e.g. 11–14, showed some toxicity to uninfected macrophages.</p><!><p>The evaluation of the effect of compounds against L. major and L. panamensis intracellular amastigotes was performed by using the Giemsa staining method.26 As described above for promastigotes, a first screening was performed at a compound concentration of 10 μM. Accordingly, active compounds were subjected to a four-point dose response evaluation. Active compounds and their IC50 values are presented in Table 1 (quinolines 12 and 15) and Table 2 (TOI compounds 9, 11, 13, 14, 16–20).</p><p>Compounds from the quinoline family showed similar effect on both Leishmania species. Only two quinoline derivatives (10 % of the compounds tested) were active and exhibited similar IC50 values for L. major and L. panamensis (Table 1 and Fig S2). Compound 12 exhibited cytotoxic effects on macrophages with a 50% cytotoxic concentration (CC50) of 14.03 μM. However, that cytotoxic concentration is still tenfold higher than the IC50 calculated for L. panamensis and L. major (1.07±0.51 and 1.65±0.3 respectively). Both quinoline derivatives, however, have similar values of selectivity index (Table 1).</p><p>Compounds from TOIs family showed differential activity against both Leishmania species. Compounds 17, 18 and 19 were exclusively active against intracellular L. major whereas compounds 11, 14, and 20 showed effect only for L. panamensis, – both promastigotes and amastigotes (Table 2). These differences in sensitivity to some compounds were previously described for Leishmania species.27 Compounds 9, 13 and 16 were active for amastigotes of both species but the effect on L. panamensis was higher (IC50= 0.8, 1.22, 0.87 μM respectively) than on L. major (IC50= 12.27, 4.30, 3.84 μM) (Table 2). It is important to note that SI values are consistently higher for L. panamensis than for L. major (Tables 1 and 2), suggesting that L. panamensis is more sensitive to both families of compounds than L. major. Compounds of the indole series showed no activity against any Leishmania species or stage tested (Table S1).</p><p>Considering the specificity of the most active compounds for the amastigote form, we evaluated the possible immunomodulatory effect of compounds 9, 12, 13, 15, and 17–19. Our results showed that compound 12 inhibited production of IL-10 by macrophages infected with L. panamensis and L. major (Fig. 3) in a dose dependent manner. These results suggest that the compound-induced parasite killing mechanism may include a regulation of the macrophage activation.</p><!><p>Leishmaniasis is recognized as one of the most neglected diseases, and the development of new drugs against leishmaniasis is an important therapeutic goal.28</p><p>There is evidence of recent appearance of Leishmania resistance to antimonials – the current first line of treatment.7b,d Molecular and biochemical differences among species influence the sensitivity of Leishmania species to different chemical agents, complicating the search for antileishmanial drugs with broad activity profile. Here, we studied synthetic compounds of three classes and identified positive hits that inhibit the intracellular amastigotes and promastigotes of L. panamensis and L. major. We also showed that L. panamensis is more sensitive than L. major to these compounds. Quinoline compounds were previously identified as efficient antimalarial and antileishmanial agents.18 Several quinoline derivatives showed inhibitory capacity against different Leishmania species that is comparable to reference drugs.8 In view of the lack of antileishmanial activity of substituted indoles, we further focused on the study of the influence of substituents in the oxazine ring of the TOI compounds. In the TOI series, presence of the bulky aromatic rings in the N9 and C4a positions generally led to the loss of antileishmanial activity. Allyl groups in N9 and C4a resulted in higher activities than both larger benzylic and smaller (methyl) groups. On the other hand, both aromatic substituents and ester groups in C3 were well tolerated. Displacement of the O1 with TsN (S14, see Table S2) led to the loss of activity, further highlighting the importance of substitution in the C ring of the TOI system.</p><p>The C4 position in 1,2-oxazines is generally difficult to access synthetically. Hence, influence of the substituents in C4 position was tested with TOI compounds bearing a chlorine atom anti to the C4a substituent. Interestingly, compounds 17 and 19 were found to be active only against intracellular L. major amastigotes, indicating that significant selectivity can be achieved through modulation of the 1,2-oxazine moiety.</p><p>In general, TOI framework has provided more hits than quinolines. In the quinoline series, only two compounds (12 and 15) that are structurally related to amodiaquine and chloroquine exhibited significant activity against intracellular L. major and L. panamensis amastigotes, with no activity against promastigotes (Table 1).</p><p>It has been known that there are important differences in the sensitivity towards chemical agents between both parasite stages, and between different Leishmania species.25,29 We showed herein that, of the 39 assayed compounds, none was active against L. major promastigotes, while 20% of the compounds were active against L. major intracellular amastigotes. Promastigotes differ biologically from amastigotes in metabolism, morphology and surface composition and these differences have an impact on the sensitivity of parasites to chemical agents.30 Moreover, to be active against amastigotes, compounds must cross the cellular membrane and maintain its stability in the intracellular environment. Additionally, some compounds may be toxic to the parasite only when metabolized inside the macrophage, then showing the behavior of being inactive in promastigotes and active in the amastigote. Another interesting possibility may be that some of the compounds, instead of being toxic directly to the parasite, may be able to activate the macrophage to fully develop their antiparasitic activity. We also showed here that compound 12 inhibits the production of IL-10 by macrophages infected with L. panamensis and L. major. It is well-known that IL-10 is an antiinflammatory mediator that inhibits a variety of macrophage functions including phagocytosis, expression of co-stimulatory molecules and production of pro-inflammatory cytokines, with important consequences in macrophage activation.31 Interleukin 10 has been implicated as a key factor in the survival of Leishmania infection both in vitro and in vivo. High levels of IL-10 have been linked to leishmaniasis progression and parasite persistence.32 The addition of IL-10 to L. major-infected macrophages results in uncontrolled parasite replication.32a Our results suggest that the antileishmanial activity of compound 12 might be, at least in part, mediated by the modulation of the macrophage activation. The selectivity index of this chloroquine analogue was 3 to 4 times higher (8.5/13.1 L. major/L. panamensis) than the SI of chloroquine (3.1) previously reported under similar experimental conditions33 Since chloroquine is an approved drug, these SI values suggest that compound 12 is a promising candidate for further therapeutic investigation.</p><p>We have shown here that novel synthetic derivatives of several families of compounds are promising antileishmanial hits, opening new possibilities for further development of 1,2-oxazine-based antileishmanial agents as a way towards new and effective drugs against this neglected disease. Due to the differences in Leishmania species sensitivity to drugs, the search of species-specific antileishmanial drugs has been encouraged. The higher values of SI for L. panamensis than for L. major for the active compounds described herein indicate that they are good leads in the development of L. panamensis-specific drugs.</p><!><p>In conclusion, a targeted library of tetrahydrooxazinoindoles (TOIs) was synthesized, and antileishmanial activities were discovered for a number of the TOI compounds. The activity was compared with indoles that were found to be inactive, and with quinolines. For quinolines, only amodiaquine and chloroquine analogues were found to be active. We have identified that the antileishmanial activity of the chloroquine analogue 12 may also be due to a modulation of macrophage activation. The activity of TOI compounds opens an avenue for the search of structurally novel antileishmanial agents and for further elucidation of their mechanism of action.</p><!><p>Dichloromethane was dried and purified under an argon atmosphere using an LC technology Solutions' SP-1 Solvent Purifier All oximes were synthesized according to the literature procedure.21 All heterocyclic N-oxides were synthesized according to reported procedures.20b N′-(2-chloro-1-phenylethylidene)-4-methylbenzenesulfonohydrazide was synthesized according to literature procedure.34 All other reagents were purchased and used without further purification. Column chromatography was performed using CombiFlash Rf-200 (Teledyne-Isco) automated flash chromatography system. 1H, 13C, 19F NMR spectra were recorded at 500 (1H), 125 (13C), and 282 MHz (19F) on Varian Mercury VX 300 and Agilent Inova 500 instruments in CDCl3 solutions. Chemical shifts (δ) are reported in parts per million (ppm) from the residual solvent peak and coupling constants (J) in Hz. Proton multiplicity is assigned using the following abbreviations: singlet (s), doublet (d), triplet (t), quartet (quart.), quintet (quint.), septet (sept.), multiplet (m), broad (br). Infrared measurements were carried out neat on a Brüker Vector 22 FT-IR spectrometer fitted with a Specac diamond attenuated total reflectance (ATR) module.</p><!><p>To an oven dried flask was added 3Å MS (5 scoops), CuOTf ½ PhMe (10–20 mol%), rac-BINAP or (S)-DM-BINAP (10–20 mol%) and dichloromethane (0.1–0.2M). The reaction was stirred for 15 min and then cooled to –78 °C under argon. Indole (1 equiv.), oxime (1 equiv.) and silver carbonate (3 equiv.) were added sequentially. The reaction was allowed to warm to either −20 or −15 °C and was stirred for the specified time. The reaction mixtures were then filtered, concentrated under reduced pressure, and purified by column chromatography [hexanes/EtOAc, silica gel] to yield the desired products.</p><!><p>Female and male Balb/c mice, 8 weeks of age, were provided by INDICASAT's animal facility. Animals were maintained with 12 hours light/dark cycle, at a constant temperature of 24 °C with free access to food and water. All experimental procedures were approved by the Institutional Animal Care and Use Committee of INDICASAT (IACUC-14-002) and were based in the strict observance of the ethic guidelines related to the handling of lab animals in accordance with international regulations and those established by INDICASAT.</p><!><p>Promastigotes of L. panamensis (MHOM/PA/94/PSCI-1) and L. major (Restrepo et al., 2013) were cultured at 25°C, in Schneider medium (Sigma) supplemented with 20% FBS (Gibco). Parasite virulence for both strains was maintained by inoculating them previously in hamster.</p><!><p>L. panamensis and L. major parasites from stationary phase culture were washed with PBS 1X and centrifuged at 1700xg for 10 minutes. Parasites were diluted in Schneider media supplemented with 20 % FBS and seeded in 96 well white opaque plate (Thermo Scientific, Nunc) at a density of 2×106 parasites per well in a volume of 99 μL. Each well was treated with 1 μL of compound at a concentration of 10 μM in screening assays and later at 1, 3, 10 and 30 μM in dose-response assays. The parasites were incubated at 25°C during 24 hours. After incubation period, 50 μL of CellTiter-Glo® reagent (Promega) was added to each well for lysing the parasites and the plate was incubated at room temperature for 10 minutes to stabilize the luminescent signal. The resulting ATP was recorded in relative-light units (RLU) in a multi-detection microplate reader (Synergy HT-Biotek).</p><!><p>Peritoneal resident macrophages from Balb/c mice were collected by peritoneal lavage with cold PBS 1X (AppliChem). Cells were seeded in RPMI (Gibco) with 10% FBS (Gibco) at a density of 1×106 cells per well in 24 well plates with a round glass coverslip in each well and cultured for 2 h at 37°C in an atmosphere of 5% CO2. Non-adherent cells were removed by washing and adherent macrophages were infected with late stationary phase promastigotes at 1:30 ratio (cell:parasite) for L. panamensis and 1:10 for L. major during 1 hour at 37°C, 5% CO2. Non-internalized promastigotes were removed by washing with RPMI media. Infected macrophages were treated with the compounds at a final concentration of 10 μM. Dose-response curves were produced for active compounds using concentrations of 1, 3, 10 and 30 μM. Amphotericin B (Sigma) was used as positive control. Infected macrophages were incubated for 24 hours at 37°C in 5% CO2. All negative controls and stimulus were performed in the presence of 0.1% DMSO (Sigma) since compounds are solubilized in this solvent. After incubation, supernatants were collected for evaluation of the presence of IL-10 and coverslips were washed once with PBS, fixed with Methanol (Merck), and stained with Giemsa (Sigma). The infection rate was calculated by counting the number of amastigotes per cell in a total of 250 cells. The percentage of parasite inhibition was calculated as</p><!><p>Peritoneal resident macrophages from Balb/c mice were cultured at 37°C in 5% CO2 in the presence of 3, 10, 30 and 100 μM of active compounds. Twenty-four hours later incubation supernatants were removed, then 100 μl of MTT (3-(4,5-dimethylthiazol-2-yl)-2,5-diphenyltetrazolium bromide) (Sigma) (0.5 mg/mL) dissolved in RPMI were added to each well and cells were incubated by 4 hours at 37°C. The MTT is reduced in living cells mitochondria to purple formazan crystals. The supernatants were discarded and formazan crystals were dissolved in 100 μl of 0.04 M HCl in isopropanol. The optical density was analyzed at 570 nm using an ELISA plate reader. The percentage of viable cells was calculated as % viability = (OD sample/OD control) × 100%. All experimental cells were cultured in the presence of medium plus 10% FCS and 0.1% DMSO.</p><!><p>Results were analyzed using the GraphPad Prism 5 statistical software package (GraphPad software, La Jolla). Half maximal inhibitory concentrations (IC50 and CC50) were calculated adjusting a sigmoidal dose-response curve following GraphPad Prism 5 procedure.</p>

PubMed Author Manuscript

Switching from zoledronic acid to denosumab increases the risk for developing medication-related osteonecrosis of the jaw in patients with bone metastases

PurposeSwitch from zoledronic acid (ZA) to denosumab may increase the risk of medication-related osteonecrosis of the jaw (MRONJ) owing to the additive effect of denosumab on the jawbone and residual ZA activities. We evaluated the risk of developing MRONJ in patients who received ZA, denosumab, or ZA-to-denosumab for the treatment of bone metastases.MethodsThe medical charts of patients with cancer who received denosumab or ZA for bone metastases were retrospectively reviewed. Patients who did not undergo a dental examination at baseline were excluded. Primary endpoint was the evaluation of the risk of developing MRONJ in the ZA-to-denosumab group. Secondary endpoints were probability of MRONJ and the relationship between risk factors and the time to the development of MRONJ.ResultsAmong the 795 patients included in this study, 65 (8.2%) developed MRONJ. The incidence of MRONJ was significantly higher in the ZA-to-denosumab group than in the ZA group [7/43 (16.3%) vs. 19/350 (5.4%), p = 0.007]. Multivariate Cox proportional hazards regression analysis revealed that denosumab treatment [hazard ratio (HR), 2.41; 95% confidence interval (CI), 1.37–4.39; p = 0.002], ZA-to-denosumab treatment (HR, 4.36; 95% CI, 1.63–10.54, p = 0.005), tooth extraction after starting ZA or denosumab (HR, 4.86; 95% CI, 2.75–8.36; p < 0.001), and concomitant use of antiangiogenic agents (HR, 1.78; 95% CI, 1.06–2.96; p = 0.030) were significant risk factors for MRONJ.ConclusionOur results suggest that switching from ZA to denosumab significantly increases the risk for developing MRONJ in patients with bone metastases.Supplementary InformationThe online version contains supplementary material available at 10.1007/s00280-021-04262-w.

switching_from_zoledronic_acid_to_denosumab_increases_the_risk_for_developing_medication-related_ost

Introduction<!>Materials and methods<!>Study design, setting, and patient population<!>Treatment procedure for bone metastases<!>Data collection and assessment<!>Statistical analysis<!><!>Discussion<!>

<p>Bone metastases are common in advanced cancers, resulting in clinically important complications, such as cancer-related pain, fractures, spinal cord compression, and hypercalcemia [1]. Skeletal-related events (SREs) remarkably decrease the quality of life of patients with bone metastasis. The effectiveness of bone-modifying agents (BMAs), such as zoledronic acid (ZA) and denosumab, in the treatment of bone metastases has been established. The results of randomized controlled trials comparing denosumab and ZA for the prevention of SREs in metastatic bone diseases have shown that denosumab is superior in cases of breast [2] and prostate cancer [3] and not inferior in cases of solid tumors and multiple myeloma [4, 5]. In addition, side effects, such as acute kidney injury, sometimes require the discontinuation of ZA. Thus, ZA has to be replaced with denosumab for some patients [6, 7].</p><p>Despite the effectiveness of BMAs, these medications can increase the risk of medication-related osteonecrosis of the jaw (MRONJ). MRONJ causes significant pain and reduces patient quality of life; therefore, multidisciplinary team care that enables appropriate monitoring and referral to a dental specialist for close follow-up and assessment of early stage MRONJ is recommended [8–10]. Several risk factors for MRONJ have been reported, including medication-, patient-, and oral health-related risk factors [8–10]. However, the risk of MRONJ in this patient population has not been fully evaluated. Both ZA and denosumab have been associated with MRONJ, but their pharmacological mechanisms are completely different. ZA has a high affinity for bone hydroxyapatite and specifically inhibits osteoclastic bone resorption and is therefore used for the treatment of bone metastases [11]. In contrast, denosumab is a fully humanized monoclonal antibody with high affinity and specificity for the nuclear factor-kappa B (NFκB) ligand RANKL. The effect of ZA on bone is long-lasting, whereas that of denosumab is temporary. We hypothesized that the risk of MRONJ may additively increase after switching from ZA to denosumab. An article had shown that switching from ZA-to-denosumab was one of the risk factors for developing MRONJ by logistic regression analysis, but not shown that switching increases risk directly [12].</p><p>In this study, therefore, we evaluated whether switching from ZA increase risk for developing MRONJ in cancer patients with bone metastases, comparing it to that in patients who received ZA/denosumab alone.</p><!><p>We retrospectively reviewed the medical records of patients with cancer who received denosumab and/or ZA for the treatment of bone metastases after dental examinations by dentists between Jul 2011 and Oct 2019.</p><!><p>This study was conducted in accordance with the Declaration of Helsinki. The study protocol was approved by the Ethics Committee of the Kobe City Medical Center General Hospital (approval number: zn171010). Patients were eligible if they were ≥ 20 years of age, diagnosed with solid tumors or multiple myeloma, had at least one bone metastasis or osteolytic lesion, and received denosumab and/or ZA treatment at Kobe City Medical Center General Hospital between Jul 1, 2011 and Oct 31, 2019. The exclusion criteria were as follows: no dental examination before the initiation of denosumab or ZA treatment, use of ZA for the treatment of hypercalcemia, lack of follow-up for at least 1 month after the treatment, received denosumab followed by ZA, or history of radiation therapy of the jaws.</p><!><p>Following dental examination, when needed, patients underwent dental procedures (including tooth extraction) to minimize the risk of developing MRONJ before the initiation of BMAs. All patients were subcutaneously administered 120 mg denosumab every 4 weeks or 4 mg ZA intravenously every 3 to 4 weeks. Patients with impaired kidney function (creatinine clearance of ≤ 60 mL/min) were given a manufacturer-recommended reduced dose of ZA (3–3.5 mg), according to the same administration schedule as that for patients with normal kidney function. We divided the study subjects into three groups as follows: patients who received only ZA (ZA group), only denosumab (denosumab group), and ZA followed by denosumab (ZA-to-denosumab group).</p><!><p>All data were collected from the electronic medical record system. We evaluated information regarding sex, age, weight, type of cancer, comorbidities, tooth extraction before and after starting BMA treatments, concomitant medications, type of BMAs, number of treatment courses, and outcomes of treatment for MRONJ. To reduce the potential bias in evaluating patient and treatment characteristics associated with the development of MRONJ, we limited the study participants to those examined by dentists before starting BMA treatments because poor oral health status has been reported as a significant risk factor for developing MRONJ [8–10]. Furthermore, all patients were recommended to visit dental clinics routinely after BMA initiation. If the patients were considering invasive dental procedures, including tooth extraction, after the initiation of treatment with BMAs, they were asked to consult with dentists in our hospital. After the initiation of BMA treatment, patients who complained of dental symptoms, such as pain or oral discomfort, consulted with a dentist following the attending physician's request. Tooth extraction was performed in unavoidable situations, including accidental root fracture or acute exacerbation of periodontal disease. MRONJ was diagnosed by dentists in our hospital based on clinical and radiographic findings, according to the criteria stated in the American Association of Oral and Maxillofacial Surgeons (AAOMS) position paper [13], and the cutoff date for diagnosing MRONJ was Dec 31, 2019. The primary endpoint was the evaluation of the risk of developing MRONJ in the ZA-to-denosumab group, whereas secondary endpoints included the probability of MRONJ and the relationship between risk factors and the time to the development of MRONJ.</p><!><p>Categorical data are presented as numbers (percentage) and were compared between groups using the Chi-square test or Fisher's exact test, as appropriate. Continuous data are presented as medians (interquartile ranges), and the Mann–Whitney U test was used to compare the groups. Univariate and multivariate Cox proportional hazards regression models were used to identify the risk factors for MRONJ. Variables with a p value < 0.05 in the univariate analysis were evaluated as potential covariates in the multivariate analysis. The time to the development of MRONJ was determined using the Kaplan–Meier method with the log-rank test. All statistical analyses were performed using JMP 13.0.0 (SAS Institute Inc., Cary NC, USA). A p value < 0.05 was considered statistically significant. For comparisons of the incidences of MRONJ between anti-resorptive treatment groups, the Bonferroni correction was applied to determine the level of significance for each group (p < 0.0167).</p><!><p>Patient characteristics</p><p>For continuous values, data are presented as the median [interquartile range (IQR)]</p><p>BMA bone-modifying agent, MRONJ medication-related osteonecrosis of the jaw, ZA zoledronic acid</p><p>aIncludes axitinib, bevacizumab, everolimus, lenvatinib, pazopanib, ramucirumab, regorafenib, sorafenib, sunitinib, and temsirolimus</p><p>Incidences of medication-related osteonecrosis of the jaw in patients receiving denosumab or zoledronic acid for bone metastases. Kaplan–Meier curves of cumulative incidences of MRONJ (a) and the incidences of MRONJ (b) in patients of the ZA alone (n = 350), denosumab alone (n = 402), and ZA-to-denosumab (n = 43) groups are shown. In the ZA-to-denosumab group, patients received a median (IQR) of 8 (2–17) ZA infusions before the first dose of denosumab. *Statistical significance was considered at a p value < 0.0167 for the Chi-square test (the criteria for significance were adjusted using Bonferroni correction). IQR interquartile range, MRONJ medication-related osteonecrosis of the jaw, ZA zoledronic acid</p><!><p>In the present study, we showed for the first time that among all patients who received dental examinations before BMA treatment for bone metastases, ZA-to-denosumab treatment significantly increased the risk of developing MRONJ, when compared to that with ZA. Concomitant use of antiangiogenic agents and tooth extraction after starting BMA treatment were also significant risk factors. Our study results clearly showed that the highest incidence of MRONJ was observed in the ZA-to-denosumab group. This information is important for minimizing the toxicity of anti-resorptive treatments in cancer patients with bone metastasis, since BMA treatment needs to be switched from ZA to denosumab in some patients, such as with skeletal disease progression or ZA-induced acute kidney injury [6, 7]. Bisphosphonates, including ZA, are known to have a high affinity for hydroxyapatite of bone [11], leading to prolonged drug action and excessive toxic effects. Therefore, in patients who switched from ZA to denosumab treatment, the additive effects of denosumab on the jawbone and the residual effect of ZA may increase the risk of MRONJ. Higuchi et al. conducted a single center, retrospective chart review and revealed that switching from ZA to denosumab is one of the risk factors in logistic regression analysis [12]. However, that study did not directly show the risk itself by comparing the switching group to ZA/denosumab alone group. In contrast, in an extended observation of two phase III trials, the incidence of MRONJ did not increase in patients who received ZA followed by denosumab [14]. Our study, for the first time, fully evaluated the risk of developing MRONJ in patients who received ZA followed by denosumab, comparing it to that in patients who received ZA alone, in a clinical practice setting. As part of a comprehensive pharmacovigilance plan, a prospective, post-marketing drug surveillance of cancer patients with bone metastases receiving antiresorptive therapies is ongoing in Denmark, Sweden, and Norway [15]. The observational period of this surveillance is up to 5 years, and the results will be reported for three treatment cohorts as follows: denosumab-naïve patients, ZA-naïve patients, and patients who switch from bisphosphonate treatment to denosumab. The results of the study will further clarify the relationship between the characteristics of BMAs and their effects on MRONJ.</p><p>The reported incidence of MRONJ is 1–17% [2–5, 12, 14, 16–19]. The incidence of MRONJ in the present study was within this range for ZA (5.4%), denosumab (9.7%), and ZA-to-denosumab (16.3%) groups. Importantly, none of the patients in the ZA-to-denosumab group developed MRONJ while receiving ZA, but seven of these 43 patients developed MRONJ after switching to denosumab. Our multivariate analysis revealed that patients in both the denosumab and ZA-to-denosumab groups had a significantly higher risk of developing MRONJ than those treated with ZA. In contrast, previous randomized controlled trials showed that the incidence of MRONJ in patients treated with denosumab was not significantly different from that in patients treated with ZA, although it tended to be higher [3, 17, 18]. This discordance might be attributed to the scheduled periodic dental examinations (e.g., at baseline and every 6 months thereafter) in previous randomized clinical trials, which decreased the risk of developing MRONJ [2–4, 17]. In fact, a recent meta-analysis of eight randomized controlled trials found a remarkably higher risk of developing MRONJ in patients treated with denosumab than in those treated with ZA [20]. The higher incidence of denosumab-associated ONJ seems to reflect the superior effect of denosumab in preventing skeletal-related events, compared to that with ZA [2, 3].</p><p>The median number of infusions of ZA was 8. Since most patients received ZA every 4 weeks in our study, patients received ZA treatment for approximately 8 months. Subsequently, BMAs were usually switched to denosumab treatment, and the cumulative incidence of MRONJ in the ZA-to-denosumab group was higher than that in the denosumab or ZA alone groups from 4 months after the first administration of denosumab. Therefore, the difference was evident at 12 months from the first administration of ZA, which seems early. We speculate that a relative lack of awareness of dental follow-up in clinical practice compared to that with prospective intervention studies might have influenced this marked difference in the cumulative incidence of MRONJ.</p><p>The other independent risk factors for developing MRONJ in this study were concomitant use of antiangiogenic agents and tooth extraction after starting BMAs, which were consistent with the findings of previous reports [10, 16, 19]. Our result may support the notion that tooth extraction before starting BMAs is a useful prophylactic intervention to reduce the risk of developing MRONJ. However, because tooth extraction before starting BMAs significantly increased the risk of developing MRONJ in the univariate analysis, early dental consultation should be considered after patients are diagnosed with cancer.</p><p>This study has some limitations. First, oral health status, such as periodontal disease, dental prosthesis, dental implants, and periodontal surgeries, was not fully investigated in our retrospective observational study design. To reduce the effect of these factors, we limited the study participants to those examined by dentists before starting BMA treatments. Second, we did not evaluate the effect of other risk factors, such as denture use and tobacco use [8–10]. Despite our best attempt to obtain clinical information, we were not able to collect all these data with this retrospective study design. To our knowledge, however, these missing data should have similar impacts among the groups. Lastly, the IQR of the number of infusions of ZA in the ZA-to-denosumab group varied from 2 to 17, indicating that patients with various backgrounds were included in this group. Despite these limitations, this real-world observational study demonstrated that the risk of developing MRONJ was significantly higher in patients with advanced cancer treated with ZA followed by denosumab. In conclusion, the results of this study suggest that switching from ZA to denosumab significantly increases the risk of developing MRONJ in patients with bone metastases.</p><!><p>Supplementary file1 (PDF 85 KB)</p><p>Publisher's Note</p><p>Springer Nature remains neutral with regard to jurisdictional claims in published maps and institutional affiliations.</p>

PubMed Open Access

Protofibril–Fibril Interactions Inhibit Amyloid Fibril Assembly by Obstructing Secondary Nucleation

AbstractAmyloid‐β peptides (Aβ) assemble into both rigid amyloid fibrils and metastable oligomers termed AβO or protofibrils. In Alzheimer's disease, Aβ fibrils constitute the core of senile plaques, but Aβ protofibrils may represent the main toxic species. Aβ protofibrils accumulate at the exterior of senile plaques, yet the protofibril–fibril interplay is not well understood. Applying chemical kinetics and atomic force microscopy to the assembly of Aβ and lysozyme, protofibrils are observed to bind to the lateral surfaces of amyloid fibrils. When utilizing Aβ variants with different critical oligomer concentrations, the interaction inhibits the autocatalytic proliferation of amyloid fibrils by secondary nucleation on the fibril surface. Thus, metastable oligomers antagonize their replacement by amyloid fibrils both by competing for monomers and blocking secondary nucleation sites. The protofibril—fibril interaction governs their temporal evolution and potential to exert specific toxic activities.

protofibril–fibril_interactions_inhibit_amyloid_fibril_assembly_by_obstructing_secondary_nucleation

<!>Introduction<!>Results and Discussion<!><!>Results and Discussion<!><!>Results and Discussion<!><!>Results and Discussion<!>Conclusion<!><!>Conclusion<!>Conflict of interest<!>

<p>F. Hasecke, C. Niyangoda, G. Borjas, J. Pan, G. Matthews, M. Muschol, W. Hoyer, Angew. Chem. Int. Ed. 2021, 60, 3016.</p><!><p>Amyloid fibrils are cross‐β structured protein assemblies that represent the hallmark of many protein aggregation disorders. [1] For several disease‐related proteins, amyloid fibrils correspond to the thermodynamic minimum of the free energy landscape for folding and aggregation. [2] For example, Aβ amyloid fibrils are the core components of the senile plaques found in Alzheimer's disease (AD)‐affected brains. [3] Aβ fibrils are polymorphic, variably constructed from in‐register parallel β‐sheets.[ 4 , 5 , 6 ] They form by nucleated polymerization, where initial fibril nuclei grow by monomer addition to the fibril ends. [7] A frequent contributor to the typical sigmoidal growth profile of amyloid fibrils is fibril‐mediated secondary nucleation. In this process, the fibril surface acts as the preferential site for new fibril nucleation, leading to the autocatalytic proliferation of amyloid fibrils. [7]</p><p>A second type of assemblies that Aβ is prone to form are metastable globular oligomers with a molecular weight >50 kD, and their associated curvilinear fibrils with typical lengths up to 200 nm.[ 8 , 9 , 10 , 11 , 12 , 13 , 14 ] These oligomers are collectively referred to as AβO or protofibrils.[ 8 , 12 , 15 ] As these oligomers are formed in a reaction distinct from fibril formation (i.e., off‐pathway),[ 8 , 11 , 13 , 16 ] the term protofibril can be misleading. Similarly, the term AβO is used interchangeably for on‐pathway oligomers. Below we use the designations globular oligomer (gO) and curvilinear fibril (CF) to refer specifically to the off‐pathway, metastable assemblies. GO/CFs form in a lag‐free oligomerization reaction with a much higher reaction order than that observed for fibril formation. [11] Like amyloid fibrils, gO/CFs are rich in β‐sheets, but their structure has not been resolved to the same level of detail yet. [17] GO/CFs have been reported for several amyloidogenic proteins, suggesting that they are a general alternative assembly type of this class of proteins.[ 16 , 18 , 19 , 20 ]</p><p>Aβ gO/CFs may represent the main toxic species in AD, as they are more effective than amyloid fibrils at inducing synaptic dysfunction, inhibiting long‐term potentiation, triggering inflammation, and disrupting membranes.[ 8 , 13 ] Several receptors that mediate toxic signaling of extracellular Aβ gO/CFs have been identified. [21] In addition, intracellular Aβ gO/CFs show cytotoxic effects. [8] Aβ gO/CFs are enriched in a halo surrounding senile plaques, pointing to a potential role of gO/CF‐fibril interactions.[ 22 , 23 ] For example, fibril plaques have been suggested to serve as a reservoir, or buffer, of Aβ oligomers.[ 22 , 23 ] However, gO/CF–fibril interactions have not been characterized in detail.</p><p>We have recently reported that the high concentration dependence of gO/CF formation results in a threshold monomer concentration required for gO/CF formation, denoted critical oligomer concentration (COC), which is significantly higher than the threshold for fibril formation.[ 11 , 20 ] Above the COC, the assembly kinetics are biphasic, with an initial lag‐free gO/CF formation phase, followed by a sigmoidal phase representing the nucleation and growth of fibrils which slowly replace the metastable gO/CFs. Surprisingly, we observed that gO/CF formation above the COC progressively increased the lag period for subsequent fibril nucleation and growth, revealing that gO/CFs inhibit fibril formation not only by competing for monomers, but also in an active fashion. These observations were made with two distinct amyloid proteins, a dimeric variant of Aβ40 (dimAβ) and hen egg‐white lysozyme (hewL). [11]</p><p>Here, we investigate how gO/CFs actively inhibit fibril formation. We first show that the inhibitory effects of off‐pathway gO/CF formation on subsequent fibril nucleation and growth are similarly present in the two dominant AD peptides Aβ40 and Aβ42. We then demonstrate for Aβ as well as for hewL that gO/CFs bind to fibril surfaces. GO/CF binding also promotes fibril bundling, thereby further reducing fibril surface area. We finally take advantage of the Aβ‐dimAβ system to show that the gO/CF‐fibril interaction interferes with secondary nucleation and blocks the proliferation of amyloid fibrils.</p><!><p>To investigate gO/CF formation of Aβ, we have generated dimAβ, a dimeric Aβ variant in which two Aβ40 units are linked in one polypeptide chain through a flexible glycerin‐serine‐rich linker. [11] The conformational properties of the Aβ40 units in dimAβ are the same as those of unlinked Aβ40. [11] However, due to the increased local Aβ concentration, gO/CF formation of dimAβ is strongly promoted, which is reflected in the comparatively low COC of ≈1.5 μM at neutral pH. [11] Above the COC, Thioflavin T (ThT) fluorescence indicates biphasic assembly kinetics of dimAβ (Figure 1 A). During the first phase, gO/CFs form (Figure 1 C) in an oligomerization reaction with a high reaction order of ≈3. [11] After a lag‐time, amyloid fibril formation is observed, in agreement with a nucleation‐polymerization reaction (Figure 1 A,C). [11] Upon prolonged incubation, the metastable gO/CFs are slowly replaced by amyloid fibrils. [11] Above the COC, the lag‐time of amyloid fibril formation develops an inverse dependence on protein concentration, i.e., the lag‐time increases with protein concentration (Figure 1 B), indicating that gO/CFs actively interfere with amyloid fibril formation. [11]</p><!><p>Biphasic assembly kinetics of Aβ. A), D), G) Transition from sigmoidal (orange) to bimodal (blue) amyloid growth kinetics of dimAβ, Aβ40, and Aβ42, monitored by ThT fluorescence. Concentration dependent time traces of A) dimAβ assembly in 50 mM Na‐phosphate, 50 mM NaCl, pH 7.4, 37 °C, and D) Aβ40 or G) Aβ42 assembly in 50 mM Na‐phosphate, pH 7.4, 27 °C. ThT fluorescence is plotted logarithmically to highlight the stable low signal during the lag‐time under sigmoidal growth conditions. B), E) Dependence of the lag‐time of the second kinetic phase on protein concentration. C), F) AFM images corresponding to the early oligomeric and subsequent fibril‐dominated kinetic phases observed above the COC.</p><!><p>We tested if these observations, previously made for dimAβ and hewL, are reproduced for Aβ40 and Aβ42. A logarithmic plot of the ThT time course of Aβ40 assembly at a concentration of 20 μM or below shows a sigmoidal curve with a lag‐time of several hours. This is in agreement with amyloid formation by a nucleation‐polymerization reaction with prominent contributions from secondary nucleation (Figure 1 D). In contrast, for Aβ40 concentrations of 40 μM or above, an additional, lag‐free kinetic phase occurred during which gO/CFs assembled (Figure 1 D,F). These gO/CFs were replaced by amyloid fibrils during a second kinetic phase (Figure 1 D,F). Aβ40 assembly thus follows the same pattern as dimAβ assembly, albeit with an approximately 20‐fold higher COC (≈30 μM), which is expected considering the lack of a covalent connection between Aβ monomers in unlinked Aβ40. ThT kinetics recorded with Aβ40 by the deGrado and Prusiner lab, for concentrations at or above those used here, generated similar biphasic kinetics and produced long‐lived Aβ gOs. [24] As with dimAβ and hewL, the lag‐time of amyloid fibril formation of Aβ40 started to increase above the COC (Figure 1 E). This indicates that Aβ40 gO/CFs share the ability to interfere actively with fibril formation. For Aβ42, the ThT time courses indicated a transition to biphasic kinetics at a concentration between 10 and 30 μM (Figure 1 G), in line with previous observations. [25] The short lag times of Aβ42 amyloid fibril formation undermined our efforts of correlating biphasic ThT kinetics with the onset of gO/CF formation in that system. Nevertheless, the data for Aβ40 and Aβ42 show that the observations made for dimAβ and hewL extend to the two prevalent Aβ variants, with higher COCs of the unlinked peptides.</p><p>One possible mechanism by which gO/CFs might actively inhibit amyloid formation would be by interfering with secondary nucleation. GO/CFs might bind to amyloid fibril surfaces, where they could block the sites capable of catalyzing fibril nucleation. To test this hypothesis, we first investigated if gO/CFs bind to amyloid fibril surfaces. Fibrils were formed from Aβ40 at a concentration of 10 μM. Since this concentration is below the COC of Aβ40, only fibrils but no gO/CFs were formed. Upon centrifugation, the fibrils were found in the pellet (Figure 2 A, left). GO/CFs were formed by quiescently incubating dimAβ at a concentration of 10 μM for 24 hours. Under these conditions dimAβ assembled into gO/CFs whereas amyloid fibrils were still absent. The gO/CFs were collected from the supernatant after centrifugation (Figure 2 A, middle). When Aβ40 fibrils and dimAβ gO/CFs were mixed and subsequently centrifuged, the pellet contained amyloid fibrils decorated with gO/CFs (Figure 2 A, right). This indicates that the fibril surfaces have an affinity for gO/CFs, leading to co‐precipitation of the two species. The experiment was repeated for hewL. HewL amyloid fibrils grown under sigmoidal (sub‐COC) conditions (Figure 2 B, left) and hewL gO/CFs formed during the early phases of biphasic growth (Figure 2 B, middle) were mixed, resulting in binding of gO/CFs to the lateral surfaces of the fibrils (Figure 2 B, right). In addition, mixing of hewL gO/CFs with fibrils at growth temperatures dramatically increased lateral bundling and precipitation of fibrils, while isolated fibrils remained unchanged (Figure 2 C). Both binding and bundling reduce the fibril surface area available for secondary nucleation.</p><!><p>GO/CFs bind to amyloid fibril surfaces. AFM images of assemblies of A) dimAβ and Aβ40 or B),C) hewL. A) Amyloid fibrils formed from 10 μM Aβ40 were found in the pellet upon centrifugation at 14 000 g (left); gO/CFs formed from 10 μM dimAβ remained in the supernatant (middle). Upon mixing equimolar amounts, dimAβ gO/CFs co‐precipitated with Aβ40 fibrils and decorated fibril surfaces (right). B) Amyloid fibrils and gO/CFs formed from 1.75 mM hewL were grown below (50 mM NaCl) or above (250 mM NaCl) the COC, respectively. After isolation and adjusting NaCl for both to 450 mM,100 μM of fibrils were mixed with 1 mM of gO/CFs at room temperature and in 450 mM NaCl. C) Mixing hewL gO/CFs and fibrils at growth temperature (52 °C), instead, induced rapid fibril bundling and precipitation while, under the same conditions, fibrils themselves remained unchanged.</p><!><p>In order to isolate the consequences of this gO/CF and fibril interaction on fibril growth mechanisms we performed seeded fibril growth experiments with increasing gO/CF admixtures. To do so, we took advantage of the different COCs for dimAβ vs. Aβ40: at low μM concentrations dimAβ assembles into gO/CFs, whereas Aβ40 continues to exhibit the sigmoidal kinetics of nucleated‐polymerization with secondary nucleation. Furthermore, dimAβ gO/CFs possess high kinetic stability and persist even for several hours after dilution to sub‐COC concentrations, thereby allowing to investigate effects of gO/CFs down to sub‐μM concentrations. [26] Amyloid fibril formation is a multistep reaction (Figure 3 G). [27] To test the effects of gO/CFs specifically on fibril elongation and secondary nucleation, we seeded Aβ40 monomers with different concentrations of sonicated Aβ40 fibrils in the presence of increasing concentrations of dimAβ gO/CFs (Figure 3 A). When 10 % Aβ40 seeds were added to 2.5 μM Aβ40 monomers, fibril elongation was the dominant reaction as evident from the immediate linear increase in ThT signal (Figure 3 B). Addition of 1.25 μM dimAβ gO/CFs (corresponding to an Aβ40 subunit concentration of 2.5 μM) did not have a substantial effect, showing that gO/CFs do not actively interfere with amyloid fibril elongation (Figure 3 B). When a lower amount, that is, 0.1 %, of Aβ40 seeds was applied, sigmoidal time traces were obtained, indicating the importance of autocatalytic amplification of amyloid fibrils by secondary nucleation (Figure 3 C). In this case, addition of dimAβ gO/CFs led to a concentration‐dependent increase in lag‐time (Figure 3 C). Since primary nucleation does not contribute to the ThT signal on this time scale at this Aβ40 monomer concentration (Figure 1 D) and fibril elongation is not affected by gO/CFs (Figure 3 B), we conclude that gO/CFs inhibit secondary nucleation. The inhibitory effect was already discernible at a concentration of 60 nM gO/CFs, which corresponds to a gO/CF:monomer ratio of 1:20 in numbers of Aβ40 units. Such a substoichiometric effect is compatible with inhibition of an autocatalytic process. To confirm that inhibition of Aβ40 fibril formation is in fact caused by gO/CFs and not due to any other activity of dimAβ on Aβ40, we compared the effects of i) dimAβ gO/CFs prepared above the COC and diluted to a sub‐COC concentration of 0.3 μM with those of ii) dimAβ monomers that were freshly eluted from size exclusion chromatography and kept at a sub‐COC concentration of 0.3 μM. The dimAβ preparation that contained gO/CFs due to incubation above the COC exhibited a much stronger effect on fibril formation than the one kept below the COC (Figure 3 D). The inhibition is not an unspecific effect of any polypeptide assembly in the size range of gO/CFs, as it is not observed for ferritin, a 24‐mer of helical bundles with a molecular weight of 440 kD (Figure S1).</p><!><p>GO/CFs inhibit secondary nucleation of amyloid fibrils. A) Scheme of the kinetics assays. The effects of dimAβ gO/CFs on secondary nucleation and elongation of Aβ40 amyloid fibrils were probed. B) Elongation of Aβ40 fibril seeds by Aβ40 monomers in the absence and presence of dimAβ gO/CFs. C) Secondary nucleation‐elongation of Aβ40 fibril seeds by Aβ40 monomers in the absence and presence of dimAβ gO/CFs. D) Secondary nucleation‐elongation of Aβ40 fibril seeds by Aβ40 monomers in the absence (grey) and presence of dimAβ gO/CFs formed above the COC and diluted below the COC (orange) or dimAβ monomers below the COC (blue). E) Global fits to the data using a nucleation‐elongation model. All parameters were shared apart from the elongation rate constants. F) Global fits to the data using a secondary nucleation‐elongation model. All parameters were shared apart from the secondary nucleation rate constant. G) Nucleation‐growth model including binding of gO/CFs to amyloid fibril surfaces, which inhibits secondary nucleation. P, fibril particle concentration; M, fibril mass concentration; m, monomer concentration; n c, nucleus size; k n, primary nucleation rate constant; k 2, secondary nucleation rate constant; k +, elongation rate constant; K D, affinity of gO/CF for the fibril surface. H), I) Numerical simulations applying the model outlined in G), using the rate constants obtained for the nucleation‐elongation model in F) (uninhibited trace) and a K D of 160 nM. Duplicate or triplicate measurements per condition are shown in panels (C), (E), (F), (H), (I).</p><!><p>To further confirm that the kinetics data are in agreement with inhibition of secondary nucleation, we computed global fits to the gO/CF concentration‐dependent data for two different models of fibril formation using the software package Amylofit. [27] First, we applied a nucleation‐elongation model and performed global fits that attributed the effects of gO/CFs to an altered rate constant of either primary nucleation or fibril elongation (all parameters were shared among the data sets apart from the rate constants of primary nucleation or fibril elongation, respectively). These fits showed clear deviations from the experimental data (Figures 3 E and S2A,B). Second, we applied a secondary nucleation‐elongation model and performed global fits that attributed the effects of GO/CFs to altered rate constants of either primary nucleation, secondary nucleation, or fibril elongation (again, keeping all other fitting parameters the same among the data sets). The global fit to this model using a variable rate constant of primary nucleation did not reproduce the decreasing slope during the exponential growth phase with increasing gO/CF concentration (Figure S2C). In contrast, when the rate constants of secondary nucleation or fibril elongation were variable, good agreement with the data was obtained (Figures 3 F and S2D,E). These fits do not differentiate between effects on secondary nucleation and fibril elongation, as the rate constant of secondary nucleation occurs in the regression equation only in the form of its product with the rate constant of fibril elongation. [28] However, as we can exclude any substantial effect of gO/CF on fibril elongation (Figure 3 B), the global fits further strengthen the case for gO/CFs inhibiting amyloid fibril formation through an effect on secondary nucleation. As gO/CFs bind to amyloid fibril surfaces, they likely inhibit secondary nucleation by blocking the sites capable of catalyzing secondary nucleation (Figure 3 G). This mode of inhibition of Aβ fibril formation has previously been described for the BRICHOS chaperone. [29] The reduction in the number of active sites effectively corresponds to a reduction in the fibril surface available for autocatalytic amplification rather than to a decrease in the secondary nucleation rate constant. We extended the nucleation‐polymerization model by including an equilibrium of gO/CF binding to fibrils that reduces the fibril mass engaged in secondary nucleation (Figure 3 G). Numerical simulations with the modified model were performed, using the rate constants obtained by Amylofit for the uninhibited case of nucleation‐polymerization with variable secondary nucleation (black fit in Figure 3 F). In particular, the same secondary nucleation rate constant was used for all gO/CF concentrations, attributing the gO/CF concentration dependence of the kinetics solely to changes in the fibril mass available for secondary nucleation according to the gO/CF:fibril interaction equilibrium. The gO/CF:fibril interaction was treated as a 1:1 interaction in the number of Aβ subunits. When applying a dissociation constant of K D=160 nM the numerical simulations yielded good agreement with data obtained both at 2.5 μM and 5 μM Aβ40 monomer concentration (Figure 3 H,I).</p><!><p>We previously observed a remarkable inversion of the scaling relation between increasing protein concentration and decreasing lag‐times for dimAβ and hewL amyloid fibril formation upon crossing the COC. [11] Here, we reproduced the surprising increase in lag‐time with increasing protein concentration for Aβ40, which indicates that gO/CFs actively inhibit fibril formation (Figure 1 E). Collectively, the AFM data (Figure 2) and chemical kinetics data (Figure 3) provide strong evidence that gO/CFs inhibit Aβ amyloid fibril formation by binding to amyloid fibril surfaces, blocking the sites that would otherwise promote secondary nucleation. The same mode of inhibition was observed for the BRICHOS chaperone, but not for a set of control proteins. [29] This suggests that this inhibitory activity is rather specific. It is also in line with the relatively high affinity of the gO/CF:fibril interaction, as indicated by the observed inhibition at low nM gO/CF concentration.</p><p>Our observations provide insight into the structure specificity of secondary nucleation. Decoration of amyloid fibril surfaces with gO/CFs formed from the same protein results in less efficient secondary nucleation. This demonstrates that gO/CF surfaces do not possess the same capacity as amyloid fibril surfaces to catalyze fibril nucleation, suggesting that the cross‐β structure of amyloid fibrils is essential for efficient secondary nucleation. This is consistent with the distinct structural signatures of gO/CFs vs. fibrils seen in the amide‐I bands of their respective infrared spectra that we have shown for hewL and that have been reported for Aβ, as well.[ 20 , 30 ]</p><p>Figure 4 shows an updated Scheme of oligomer and amyloid fibril formation. GO/CFs are an alternative (off‐pathway), metastable assembly type and form rapidly and extensively above the COC. GO/CFs inhibit amyloid formation by competing for the monomers that are required for amyloid fibril nucleation and elongation. [11] In addition, as we show here, GO/CFs actively inhibit the autocatalytic amplification of fibrils by blocking secondary nucleation sites on amyloid fibrils.</p><!><p>Scheme of oligomer and amyloid fibril formation. GO/CFs constitute an alternative (off‐pathway) assembly type that competes with amyloid fibrils for monomers and that inhibits the autocatalytic amplification of amyloid fibrils by secondary nucleation. GO/CFs interfere with secondary nucleation by binding to amyloid fibrils surfaces and blocking the sites that catalyze nucleation.</p><!><p>Recently, protofibril–fibril interactions were observed under conditions of biphasic Aβ42 assembly, and the protofibrils were interpreted to represent nuclei formed by secondary nucleation. [31] This interpretation is in conflict with the off‐pathway nature of protofibrils.[ 11 , 13 ] The results reported here show that protofibril–fibril interactions do not represent, but rather interfere with secondary nucleation.</p><p>The interplay between gO/CFs and amyloid fibrils has a high relevance for AD pathogenesis: GO/CFs, which are thought to represent the main toxic Aβ species,[ 8 , 13 , 21 , 32 ] were shown to associate with amyloid fibril plaques in vivo, with potential consequences for the neurotoxic activities of both assembly types.[ 22 , 23 ] For example, amyloid fibril plaques might serve as reservoir of toxic gO/CFs.[ 22 , 23 ] Our results demonstrate that the interaction of gO/CFs with amyloid fibrils affects the kinetics of formation and depletion of the two species. By binding to amyloid fibrils, gO/CFs inhibit formation of new fibrils and thereby delay their own replacement by amyloid fibrils. The dimAβ‐Aβ40 system may serve as a valuable tool for further elucidation of the interplay between gO/CFs and amyloid fibrils.</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

Rhodium(I)-Catalyzed Benzannulation of Heteroaryl Propargylic Esters: Synthesis of Indoles and Related Heterocycles**

A de novo synthesis of benzene ring allows the preparation of a diverse range of heterocycles including indoles, benzofurans, benzothiophenes, carbazoles and dibenzofurans from simple heteroaryl propargylic esters using a unified carbonylative benzannulation strategy. Multiple substituents can now be easily introduced to the C4\xe2\x80\x93C7 positions of indoles and related heterocycles.

rhodium(i)-catalyzed_benzannulation_of_heteroaryl_propargylic_esters:_synthesis_of_indoles_and_relat

<p>The importance of indole as one of the most abundant heterocycles in natural products and pharmaceutical agents continues to inspire the development of general methods for their preparation.[1] The majority of indole syntheses focused on the formation of pyrroles from substituted benzene derivatives by forming bonds a–d or their combinations via cyclization or cycloaddition reactions (X = NR, Scheme 1). Substituents on the C4–C7 positions of indoles need to be derived from polysubstituted benzene derivatives, which are often not readily available. Synthesis of indoles from substituted pyrroles by forming bonds e–i would be different from most methods in the literature and allows the introduction substituents to the C4–C7 positions on indoles from the corresponding reactants. This approach, however, is challenging as it requires a de novo synthesis of the benzenoid ring.[2]</p><p>We herein report a carbonylative benzannulation strategy for the synthesis of highly substituted indoles 3 and 4 from pyrrolyl propargylic esters 1 and 2, respectively. Multiple substituents on C4–C7 positions of the indole products can be easily introduced from the corresponding propargylic esters via the de novo synthesis of the benzenoid ring.[2] We also demonstrated that the dearomative benzannulation strategy could be extended to the synthesis of other heterocycles such as benzofurans, benzothiophenes, carbazoles and dibenzofurans.</p><p>Among the benzannulation methods, Dötz benzannulation[3] of alkenyl Fischer carbene with alkyne proved to be extremely versatile for the synthesis of phenols with a para-alkoxy group.[4] Wulff and Merlic developed a complimentary benzannulation of dienyl Fischer carbenes for the synthesis of phenols with an ortho-alkoxy group.[5] Barluenga extended the benzannulation strategy to the synthesis of benzofurans from alkynyl Fischer carbene and furfural derivatives.[6] Very recently, Katukojvala reported a [4+2] benzannulation of pyrroles with enalcarbenoid derived from diazo compounds for the synthesis of indoles.[7]</p><p>Compared to Fischer carbenes[4] and diazo compounds,[8] propargylic esters are more desirable starting materials as precursors of metal carbenes through a 1,2-acyloxy migration process.[9] We[10] and others[11] have demonstrated that vinyl propargylic esters (i.e. 3-acyloxy-1,4-enynes) can be employed as a five-carbon building block in [5+1] cycloadditions through carbene intermediates for the synthesis of resorcinols. We also showed that the same 1,4-enyne five-carbon building block could be employed in [5+2] cycloadditions with alkynes or alkenes through carbene intermediates for the synthesis of 7-membered rings.[12] We envisioned that a variety of highly substituted heterocycles might be prepared if propargylic esters 1 or 2 could undergo benzannulation with carbon monoxide through the same metal carbene intermediate. However, the feasibility of this transformation is not obvious before we initiate our investigation because a dearomatization event is required for the formation of indoles 3 or 4 from 1 or 2, respectively,[13] while all previous studies involve 1,4-enyne systems.[10–12]</p><p>Propargylic ester 1a (Table 1) was prepared from the corresponding aldehyde in a one-pot procedure and 83% yield. Treatment of this ester with strong π-acidic metals that are capable of promoting 1,2-acyloxy migration of propargylic esters[9] did not yield any desired product (entries 1 and 2). We were pleased to find that a 83% yield was observed for indole 3a when [Rh(CO)2Cl]2 was employed as the catalyst (entry 3). Most rhodium catalysts could promote the benzannulation reaction at room temperature (entries 4–6) with the exception of Wilkinson's catalyst (entry 7). High yields were observed in most solvents (entries 8–12) and the yield was slightly better in DCM (entry 10). The catalyst loading could be lowered down to 3 mol%, while the reaction was still worked at room temperature (entry 13).</p><p>We then set out to examine the scope of this new benzannulation for the synthesis of highly substituted indoles (Table 2). The type of esters did not impact the reaction much (entries 1–3). The tosyl group could also be replaced by p-nitrobenzenesulfonyl or Boc- groups (entries 4–6), though slightly higher temperature was required in the latter case. The reaction could tolerate various functional groups, including bromine, ketone, free alcohol, and ester (entries 7–10). It is worth to point out that substituents on the 4-position of the pyrrole did not interfere with annulations that occurred on the adjacent 3-position.</p><p>A carboxylic ester could be introduced to the C5-position of the indole by starting with the corresponding ynoate substrate (entry 11). Substrate 1l with a methyl group did not yield any desired indole product (entry 12), which is consistent with the preferred 1,3-acyloxy migration mode for propargylic esters bearing an internal alkyne.[9]</p><p>We next turned out attention to substrates with a propargylic ester on the 3-position of the pyrrole. Slightly lower yields were obtained for pyrrolyl propargylic esters with either a N-tosyl or N-Boc group (entries 13 and 14). Cationic rhodium catalyst gave the best yield in the latter case. The reaction also worked with substrates bearing a tertiary ester (entry 15). An additional methyl group was introduced to the C4-position of the indole in this case. The Boc-protecting group in products 3f and 4b can be easily removed to yield the corresponding free indoles.[14]</p><p>A unified strategy may be realized for various indole-related heterocycles by simply replacing the pyrrole ring with other heteroaryl groups (Table 3). Indeed, benzofurans and benzothiophenes with multiple substituents on the C4–C7 positions were prepared in good yields from the corresponding aryl propargylic esters (entries 1–4). It is worth to mention that substrates 5a and 5b were prepared from bio-renewable feedstock furfural in high yields.[15] Tricyclic carbazoles 6e and 6f were easily prepared from the corresponding 2-indolyl propargylic ester 5e and 3-indolyl propargylic ester 5f, respectively (entries 5 and 6). These two carbazoles have complementary substituents on the newly formed benzene ring. Tricyclic dibenzofuran 6g was synthesized from the corresponding aryl propargylic ester 5g (entry 7). Multiple substituents could be introduced to the indolyl propargylic ester starting materials and highly substituted carbazoles on both benzene rings were prepared efficiently (entries 8–11). No benzannulation product was observed for phenyl propargylic ester 5l (entry 12). Instead, carboxylic acid 7 was obtained in 52% yield. We also tried to replace the phenyl group in 5l by p-methoxyphenyl group and no desired benzannulation product was observed. A 36% yield of product 6m could be obtained from nathphyl propargylic ester 5m.</p><p>The pivalate and hydroxyl groups derived from the benzannulation can serve as handles for the introduction of more functional groups. For example, the former may undergo direct Ni-catalyzed cross-coupling reactions;[16] and the latter can be converted to triflate for Pd-catalyzed cross-coupling reactions. As shown in Scheme 2, a Pd-catalyzed reduction afforded indole product 8; and a Sonogashira coupling with a terminal alkyne yielded indole product 9. An allyl substituent could be introduced to the C5- or C6-position of indoles 10 and 11, respectively, by a sequence of allylation of the phenolic OH group followed by Claisen-rearrangement. The Ts-group in product 6k can be easily removed to afford carbazole 12 while retaining the pivalate group.</p><p>In Rh-catalyzed cycloadditions involving 3-acyloxy-1,4-enynes, the 1,2-migration of ester was the rate-determining step and the reaction was significantly faster for esters bearing an electron-donating group.[10, 12d, 17] Under our standard conditions, both products 6n and 6o could be prepared in 70% and 66% yields, respectively. Surprisingly, we found that the benzannulation of 5n with an electron-rich ester was completed in less than 1 min at room temperature. We then mixed substrates 5n and 5o in a 1:1 ratio and tried to examine the rate difference between them. After running the reaction for just around 10 s, we observed over 95% yield of 6n and less than 5% yield of 6o based on 1H NMR using CH2Br2 as the internal standard. The first 1,2-acyloxy migration step is likely the rate-determining step for the benzannulation of heteroaryl propargylic esters.</p><p>Previously, a concerted 1,2-acyloxy migration oxidative cyclization was proposed as the initial step for Rh-catalyzed cycloadditions of 3-acyloxy-1,4-enyne 13 to form metallacycle 14 and metal carbene intermediate 14′ (Scheme 4).[17a] The double bond in the aryl group of 15, however, is unlikely to coordinate to rhodium. A sequence of 1,2-acyloxy migration, carbene formation, and CO insertion may afford ketene 18. A dearomatic 6π-electrocyclization of ketene 18 followed by aromatization may yield product 19. Although metal catalysts (e.g. Au- or Pt-based complexes) can promote the 1,2-acyloxy migration of propargylic esters to generate metal-carbene intermediates, very few of them have the ability to undergo CO insertion to yield ketenes.[9] This may explain why only Rh(I)-complexes worked for the carbonylative benzannulation of aryl propargylic esters among the π-acidic metals in Table 1.</p><p>To further understand the details of the mechanism, we performed Density Functional Theory (DFT) calculations for several potential pathways. The preferred one is shown in Scheme 5 and other pathways are outlined in the Supporting Information. DFT calculations indicate that [Rh(CO)3Cl] is the resting state of the catalyst (Scheme 5). Substrate 5b undergoes ligand exchange with [Rh(CO)3Cl] to form complex 20. This 16-electron rhodium complex adopts a square-planar geometry, where only the alkyne coordinates to the Rh centre. This is different from the Rh(I)-catalyzed [5+n] cycloaddition of 3-acyloxy-1,4-enyne with alkyne and CO, where the rhodium binds to both alkyne and alkene simultaneously in the initial π-complex.[17] Promoted by the Rh(I) catalyst, the 1,2-acyloxy migration occurs stepwise with an overall barrier of 21.9 kcal/mol, leading to the formation of zwitterion intermediate 22, where the positive charge is stabilized by both the alkene and the furan moieties. The 1,2-acyloxy migration requires the highest free energy of activation (TS1), thus being rate determining step, which is consistent with faster reactions for electron-rich esters such as 5n.[17]</p><p>Intermediate 22 can be isomerized a four-membered metallacycle, which can undergo CO insertion and subsequent reductive elimination to afford ketene intermediate similar to 18 for 6π-electrocyclization and aromatization to furnish benzofuran product 6b, as shown in the Supporting Information (Figure S1). In contrast to reactions involving 3-acyloxy-1,4-enynes, where the 1,4-enyne serves as a bidentate ligand,[17] the furan ring does not coordinate to rhodium at any time. The benzannulation involves much higher energy barriers when the furan moiety coordinates to the metal centre to form metallacyles related to 14 or 14′ as shown in Figures S1 and S2). It is clear that aryl propargylic esters behave very differently from 1,4-enynes we studied before.</p><p>In summary, we have developed a practical and unified method for the synthesis of indoles and related heterocycles. This method differs from most previous approaches by allowing the introduction of various substituents to the benzene-portion of indoles and related heterocycles. We demonstrated that indoles and related heterocycles with substituents at any of C4–C7 positions could be prepared by starting with appropriately substituted heteroaryl propargylic esters. We anticipate that this method will find immediate applications in many areas of organic, medicinal and material chemistry. Computational studies suggest that a zwitterion intermediate instead of a metallacycle intermediate is formed after the 1,2-acyloxy migration. This finding may have broad implications in many transition metal-catalyzed reactions.</p>

PubMed Author Manuscript

Ligand mediated nanocluster formation with classical and autocatalytic growth

We present a systematic study of ligand-mediated nanocluster (NC) formation using a kinetic model which provides atomic insight to sub-nanometer cluster (S-NC) and NC formation. Our model describes the role of ligand-mediated nucleation and growth in obtaining monodisperse NCs. Nucleation includes metal ion reduction, reversible ligand association to the metal ion/atom, and formation of dimer nuclei. Growth can occur through autocatalytic surface growth and ligand-associated monomer addition to the cluster depending on the rate of metal ion to neutral metal atom conversion. Furthermore, we studied the effect of the initial concentration of metal ion on NC formation using fast and slow reducing agents in the presence of slowly and rapidly binding ligands. The model shows fast nucleation, slow growth, and a high molar ratio of rapidly binding ligand to metal ion promotes the formation of S-NCs and NCs. Our results can guide experiments in the synthesis of ultra-small clusters. the reaction rates. [19][20][21] Mozaffari et al. developed a model of ligand-mediated nucleation and autocatalytic growth of nanoparticles, 22 while Lazzari et al. reported a kinetic study of ligandmediated CdSe nanoparticle formation. The authors fit their model to the experimental data and extracted temperature dependent kinetic parameters. 23 We adapted the method of Lazzari et al 23 and developed a ligand-mediated model of NC formation in which we investigate the parameter space of NC formation. We explore the initial conditions and rate constants that allow the synthesis of stable S-NCs and NCs. Our kinetic model involves a precursor conversion of ions to neutral atoms associated to ligands and then formation of dimer nuclei followed by ligand-mediated growth through ligand-associated monomer addition and autocatalytic surface growth of seed clusters. We derived kinetic equations for our model and solved them numerically using an ordinary differential equation (ODE) solver in MATLAB. We do not model diffusion and assume NC formation is controlled by the reaction kinetics of the homogenously mixed solution. Experimentally, rapid mixing of reagents can be achieved with micromixers. 24 We use the model to predict the size distributions of NCs for a range of kinetic parameters, starting conditions, and reaction schemes.The kinetic model explicitly shows the important role of ligands in NC nucleation and growth.The NC size distribution shifts to larger sizes with increasing ligand-associated monomer growth or autocatalytic surface growth rates. The model confirms fast nucleation forms small clusters. A high rate of monomer formation, strong association of ligand with metal ion/atom, and fast nucleation results in sub-nanometer clusters. We show rapidly binding ligands kinetically stabilize NCs for both strong and weak reducing agents. 24,25 However, we find fast autocatalytic growth occurs with a weak reducing agent. Consequently, large (> 1.5 nm) polydisperse NCs are formed. In contrast, autocatalytic growth is prevented with a strong reducing agent and stable

ligand_mediated_nanocluster_formation_with_classical_and_autocatalytic_growth

<!>METHODS<!>ML<!>D M i<!>RESULTS AND DISCUSSION<!>Figures 6a and 6b<!>CONCLUSIONS<!>ASSOCIATED CONTENT<!>Corresponding Author

<p>Nanoclusters (NC) are clusters of metal atoms with diameters smaller than 2 nm and contain unique properties due to their small sizes. 1 Understanding their formation is useful for a variety of applications in catalysis, [2][3][4][5][6] bio-imaging and sensing, [7][8][9][10] and medical therapies. 11 The synthesis of atomically precise NCs is possible with meticulous control over experimental conditions which requires knowledge of the variables that effect their sizes. 12 Advanced experimental techniques, such as transmission electron microscopy (TEM), have recently allowed researchers to observe nanoparticle growth with unprecedented resolution. 13 Theoretical models have also unveiled many factors that affect the size of nanoparticles. 14,15 However, many key factors remain to be investigated. For example, what is the role of ligands in nanocluster nucleation? Are they merely spectators, or do they play an important role in the nucleation process? Here, we take steps toward finding the optimal conditions for sub-nanometer cluster (S-NC) formation and provide mechanistic insight into their ligand-mediated nucleation and growth. To pursue our intention, we developed a ligand-mediated kinetic model to investigate the formation of metal NCs starting from individual atoms (monomers) to the final stages of formation.</p><p>For several years, nanoparticle nucleation and growth were explained by LaMer's burst nucleation mechanism followed by nanoparticle growth. 16,17 LaMer's burst nucleation is based on Becker and Doring's classical nucleation theory (CNT). 16,17 In LaMer's nucleation mechanism, the concentration of monomers reaches a critical supersaturation point after which they nucleate by overcoming the energy barrier for nucleation. 16 This is followed by growth of nanoparticles as a separate step. However, experiments show that CNT fails to adequately describe nucleation and growth of transition metal nanoparticles. 19,20 Finke and Watzky (FW) considered nucleation and growth occurring simultaneously through a two-step mechanism. In the FW model, slow, continuous nucleation occurs simultaneously with autocatalytic growth which is controlled by monodisperse S-NCs are formed. Besides its predictive power, the model provides atomic insight into the mechanism of NC formation which informs experiments in developing and optimizing methods to produce monodisperse NCs stabilized at early stages of growth in a homogenously mixed solution.</p><!><p>The detailed mechanism of the kinetic model is where the cluster C i,j = M i L j is composed of i metal atoms M and j ligands L. The reaction scheme starts with Eq. 1a representing the reduction of the precursor metal ion (M + ) to zero-valent metal atom (M) with rate constant k p, 1 . Examples of M and M + are gold (Au) and silver (Ag) atoms and their corresponding monovalent cations. The precursor and neutral metal atom are typically solvated and coordinated with one or more ligands e.g. thiolate or phosphine. The scheme could reasonably start at Eq. 1d, but we explicitly consider ligand binding in our model. Oligomers are excluded from our reaction scheme for the sake of simplicity.</p><!><p>(1b)</p><p>C i,j + ML kg,i,j * )</p><p>C i,j + ML + kg,i,j,ac * )</p><p>C i,j + L ka,i,j * ) ke,j+1</p><p>< l a t e x i t s h a 1 _ b a s e 6 4 = " k P m C + k 9 h 0 t 9 n p F J d 1 v D m Q X Y 3 w D w = " > A A A E W 3 i c j V N d b 9 M w F P X S A q M M 6</p><p>In experiments, the concentration of the reducing agent is often many times the concentration of metal precursor. [26][27][28][29][30][31] 24,33,34 and neutral 35 dimers have been detected in the early stages of silver nanocluster formation. Furthermore, experiments indicate the dimer may be the nuclei for growth and is sometimes referred to as the kinetically effective nucleus. 36 Equations 1g and 1h show reversible growth occurs through two pathways: addition/dissociation of ML and ML + to/from the cluster C i,j to form cluster C i+1,j+1 / C i-1,j-1 with rate constants k g,i,j /k d,j</p><p>and k g,i,j,ac /k d,j+1,ac respectively. Equation 1h is also composed of two steps: addition of ligandassociated ion and reduction of the cluster, therefore, k g,ac is an effective rate constant accounting for both steps. Reduction of the positively charged cluster is irreversible but there's a small probability that the ligand-associated ion dissociates from the cluster before reduction occurs.</p><p>We incorporate this into the scheme above with k d,ac . Equation 1i includes reversible ligand association/elimination step of the cluster C i,j to form C i,j+1 and C i,j-1 with rate constants k a,i,j and k e,j+1 . These steps outline our kinetic model from which we extract three reaction schemes by modifying the rate coefficients to study NC formation.</p><p>For growth of the cluster, the model incorporates the addition of ligand-associated metal ion (autocatalytic surface growth) [20][21][22][37][38][39] and neutral ligand-associated metal atom (classical growth) 23 to the cluster C i,j to form the cluster C i+1,j+1 increasing the number of monomers i and ligands j by one. These two pathways, besides coalescent growth, 13,[40][41][42] have been extensively discussed in the literature for nanoparticle formation. Furthermore, ligand binding strength and concentration are important factors in kinetically stabilizing S-NCs and NCs in their metastable stages of growth. [26][27][28][29][30]43 In this regard, ligand adsorption on the cluster and elimination from it is implemented in the model through which cluster C i,j gains or loses a ligand to form C i,j+1 or C i,j-1 , respectively. Figure 1 presents a schematic of the different ways a cluster C i,j can change its indices i and j and Table 1 lists a summary of the rate coefficients used in the model. is the total number of sites on a cluster with i monomers, the number of vacant sites is (N s,i -j)</p><p>where j is the number of ligands that are already in occupation. Assuming that the rate coefficients for ligand addition and elimination are linearly dependent on the respective numbers of vacant (N s,i -j) and occupied sites j, so that</p><p>(5) (6)</p><p>The number of binding sites on a nanocluster (Eq. 8) originates from an empirically derived scaling relation for the number of ligands on a cluster with i monomers. 44 (8)</p><p>The brackets in Eq. 8 round the number to the nearest integer. Using the defined rate coefficients (Eqs. 2-7), we derived Eq. 9 for the rate of change of concentration of cluster C i,j with i monomers and j ligands. The minimum and maximum numbers of monomers in Eq. 9 are i min = 3</p><p>and i max = 400, respectively. The kinetic rate equations for other species M, M + , L, ML, ML + , C 2,2 are listed in the Supporting Information.</p><p>(</p><p>The rate equations contain two internal coordinates i and j, corresponding to the number of monomers i and ligands j in a cluster C i,j . To simplify the equations, we used the method of moments by summing over j to convert each 2-D equation to two 1-D equations. 23 Equations 10-12 define the zeroth, first, and second moments respectively.</p><p>(10) (11) k e, j = k e j</p><p>)</p><p>The zeroth moment (Eq. 10) provides the concentration of clusters with i monomers irrespective of the number of ligands on the surface of the cluster. The first moment (Eq. 11)</p><p>calculates the total concentration of ligands on clusters with i monomers. The second moment (Eq. 12) contains information about the shape of the distribution of ligands on the clusters and can be determined if we further assume the j ligands are binomially distributed on a cluster with i monomers (Eqs. 15 and 16), for which there is good experimental evidence. 23,45 Summing Eq. 9 over the number of ligands j before and after multiplying by j (Eqns. 10 and 11), we get two differential equations (Eqns. 13 and 14) for C " (𝑖, 𝑡) and L i (t) with the time dependence suppressed for brevity.</p><p>This conversion reduces the number of equations and saves computing time in solving them numerically. To perform the sum, we assume the number of ligands bound to a cluster with i monomers follows a binomial distribution, (</p><p>where p i is the probability of finding a bound ligand with respect to the total number of available sites (N s,i ) on a cluster with i monomers. 45 (16)</p><p>The concentration of clusters with i monomers and j ligands is then (17) Finally, Eq. 18 is obtained with the assumption of binomially distributed ligands on the clusters. 23,45 (</p><p>The coupled ordinary differential equations (ODEs) were solved numerically using ode15s in MATLAB. The solution to the kinetic rate equations provides the population of clusters C " (𝑖, 𝑡)</p><p>with i monomers as a function of time. Using a method described elsewhere, the monomeric distribution of clusters is transformed to a size distribution using Eq. 19, where D M = 0.25 nm is the monomer diameter. 23 The factor 0.45 originates from a scaling relation between the dimensionless mass of the clusters (i) and the clusters' diameter. 23 (19) Table 1 presents the rate coefficients used in our model, and Table 2 displays reasonable magnitudes for three reaction schemes that are discussed in the Results and Discussion.</p><!><p>Table 2. Rate coefficients used to create Schemes 1, 2, and 3.</p><!><p>To understand the role of ligands in stabilizing NCs we developed a ligand-mediated kinetic model in which growth is governed by ligand-associated monomer addition to the cluster and autocatalytic surface growth (see Methods). Experiments indicate that an excess of ligand, e.g.</p><p>thiolates 46,47 or phosphines 48,49 , can shift the NC size distribution to smaller sizes and may trap NCs in their early stages of growth. [26][27][28][29][30] Inspired by these experiments, our reaction schemes utilize a large ligand to metal ion molar ratio of 6.00 mM/0.05 mM = 120.</p><p>We investigated three reaction schemes. Scheme 1 ( and slow growth in the presence of a rapidly binding ligand. The rapidly binding ligands act as a barrier for growth. We elaborate on this observation by calculating the probability of a NC with 10 monomers having ligands on its surface with slowly and rapidly binding ligands. Scheme 3</p><p>shows a strong reducing agent results in small NCs as it promotes fast monomer formation and nucleation 50 , which we observe favors classical growth over autocatalytic growth. Conversely, with a weak reducing agent, monomer formation occurs slowly while many charged monomers are available for autocatalytic growth. As a result, the NCs have larger diameters compared to the results with a strong reducing agent in Scheme 3.</p><p>Scheme 1: Ligand-mediated classical growth.</p><p>< l a t e x i t s h a 1 _ b a s e 6 4 = " z W 7 q B p K j 2 S V R N J g W 4 q h y 2 c y b</p><p>x o j I 8 h e w h U / i I / g G d o g t i P E j a Z x 4 J F t X 9 5 5 z 7 p n r a z 9 m V E j L + t n R 9 B s 3 b 9 3 e u 2 P c v X f / w c P 9 g 0 First, we discuss results for a small ligand association rate k a (10 -3 M -1 s -1 ), assuring the NCs have a bare surface. These results are contrasted with a large ligand association rate (10 6 M -1 s -1 )</p><p>providing NCs with surfaces covered in ligands. The results of Figure 2 were calculated with a ligand association rate coefficient of k a = 10 -3 M -1 s -1 . We also investigated the dependence of NC sizes on k g and k n with a large ligand association rate constant (k a = 10 6 M -1 s -1 ). Small and large k a correspond to slowly and rapidly binding ligands, respectively. Figure 3a again indicates increasing k n results in smaller NCs, as discussed in Figure 2. Figure 3c displays the NC size distribution for k a = 10 6 M -1 s -1 . For the same rate constants k g and k n , rapidly binding ligands form smaller NCs. Figure 3b shows increasing the growth rate k g with a slowly binding ligand shifts the size distribution to larger NCs. In contrast, Figure 3d displays the NC size distribution obtained with a rapidly binding ligand for increasing k g . With a rapidly binding ligand, the NCs do not grow significantly with an increase in k g by two factors of 10. This trend reflects the limited number of binding sites on the surface of the NCs.</p><p>Autocatalytic surface growth involves the addition of a charged monomer (ML + ) to the growing NC, while simultaneously being reduced. Therefore, we set the neutral monomer (ML) dimerization k n and growth and dissociation rate constants k g and k d equal to zero which provides a model for ligand-mediated autocatalytic NC formation. Table 1 lists the rates constants used for this reaction scheme.</p><p>Similar to Figure 2, Figure 4 shows the trend of decreasing average NC diameter for increasing dimerization rate. ML + is formed through M + + L and reaches a maximum of approximately 0.05 mM and then is consumed through autocatalytic growth and conversion to ML. Formation of ML mostly occurs through the reduction of ML + , but as k n,ac increases from 10 1 M -1 s -1 to 10 3 M -1 s -1 the equilibrium concentration of ML decreases. Figure 4 also shows [C " ) ] evolution with no transient maximum which implies autocatalytic dimerization no longer conforms to the classical LaMer paradigm in this case.</p><p>< l a t e x i t s h a 1 _ b a s e 6 4 = " y p Q d s c F t K c X 8 P n T 3 g k j L 1 u l 5 the average NC diameter is larger when a weak reducing agent is used. Furthermore, the size distribution of NCs is broadened (Figure 5) in comparison to the single monomer growth pathway with a strong reducing agent (Figure 3).</p><p>In this scheme both neutral monomer (ML) addition and autocatalytic growth pathways are active which is a more likely scenario in real systems. 24,34 Table 2 lists the rate constants used for Scheme 3. In the presence of a strong reducing agent, many neutral monomers are produced.</p><p>Consequently, many nuclei (dimers) are available for growth and the average diameter of NCs are relatively small. However, in the presence of a weak reducing agent, fewer neutral monomers are produced but many charged monomers are available for autocatalytic growth. Therefore, in the presence of a weak reducing agent, the average diameter of NCs are expected to be large. In light of this, we investigated the effect of k p,1 as a switch between classical and autocatalytic growth. Classical growth predominates with large k p,1 (strong reducing agent). Conversely, autocatalytic growth predominates with small k p,1 (weak reducing agent).</p><!><p>< l a t e x i t s h a 1 _ b a s e 6 4 = " To elaborate on the contrasting results obtained with strong and weak reducing agents for Scheme 3, we also discuss the effects of changing the concentration of metal ion precursor (M + ) and the ligand association rate to the NCs. Similar to Schemes 1 and 2 we observe smaller NCs in the presence of a rapidly binding ligand. Figures 7b and 7d show NCs with diameters less than 1 nm can be obtained with a strong or weak reducing agent in the presence of a rapidly binding ligand. 25 However, when a weak reducing agent is used, the NCs are more sensitive to an increase in concentration of metal ion precursor and the average diameter shifts to larger sizes as the concentration increases (Figures 7c and 7d). Figures 7a and 7b indicate the concentration of metal ion does not affect the NC size distribution using a strong reducing agent, which is likely due to the lack of NC growth through the addition of a non-ligand associated metal atom in our model. To illustrate the effect of ligand association rate to the NC, Figures 7e-h show the probability of finding j ligands on the surface of the M 10 L j NC with a diameter near 0.7 nm. Figures 7e and 7g indicate the probability of finding a single ligand on the surface of clusters with 10 monomers is approximately zero. On the other hand, Figures 7f and 7h show that 100% of NCs with 10 monomers have more than 5 ligands on their surface. Furthermore, 40% of NCs with 10 monomers have 9 of the possible 10 binding sites covered in ligands. These observations emphasize the role of ligands in occupying binding sites on the NC surface and thereby inhibiting NC growth.</p><p>As an example, we compare the results of our calculations to experimental results obtained through radiolytic reduction of metal ions in solution. Belloni et al 24 used gamma-radiation to produce solvated electrons and free radicals which then reduced monovalent silver atoms in solution and induced silver nanocluster nucleation and growth. The researchers observed smaller metal clusters at high radiation dose rates compared to low radiation dose rates. At high dose rates most of the ions were reduced to neutral monomers in solution, providing many nuclei for growth. At low rates relatively few neutral monomers were produced, but many monovalent metal atoms were available for autocatalytic growth resulting in larger nanoparticles. Our model supports these findings and illustrates the NC sizes that could be obtained using a chemical reducing agent in solution (Figures 6 and 7). In this case, we emphasize microfluidic mixers may be needed to ensure fast mixing of reactants.</p><p>As a second example, we compare the model to recent experiments on gold thiolate nanoclusters produced through gold precursor reduction by carbon monoxide which show a contrasting trend 51 to the trend presented here. The reduction kinetics of carbon monoxide was modified by adjusting the pH of the solution. At pH 11 the reduction kinetics of carbon monoxide is faster than at pH 7. The researchers observed larger nanoclusters (M 25 L 18 ) with faster reduction kinetics at pH 11 than at pH 7 (M 10-12 L 10-12 ). 51 Our model would suggest the opposite trend should be observed where smaller clusters are obtained with faster reduction kinetics because more nuclei are produced. To clarify the contrasting observations, we point out that gold thiolate forms oligomers in solution, and the size of the oligomer species is pH dependent with smaller structures observed at a higher pH. 52 In other words, more gold thiolate nuclei are present at higher pH and thus smaller nanoclusters are obtained after growth completes. 52 Our model does not incorporate the formation of oligomeric structures, but it is qualitatively consistent with the observation that a greater number of nuclei produce smaller nanoclusters.</p><p>In summary, the model shows fast nucleation, slow growth, high molar ratio of rapidly binding ligand to metal ion in a well-mixed solution promotes the formation of small nanoclusters.</p><p>Examples are the formation of gold Au and silver Ag nanoclusters (NCs) and sub nanoclusters (S-NCs) in mixing experiments. The use of microfluidic devices could provide well-mixed solutions to facilitate NC formation. [53][54][55][56][57][58] Finally, the model can be improved by incorporation of coalescent growth, diffusion, and kernels (rate constants) that account for the interaction between species in the system (e.g. DLVO theory). 30,40</p><!><p>We combined ligand-mediated monomer addition and autocatalytic surface growth in a kinetic model to understand the mechanism of NC formation which, to our knowledge, is the first study to do so. Our detailed investigation explicitly showed fast nucleation and slow growth promotes the formation of small nanoclusters. Our results suggest that even with a slow reducing agent the NCs can be kinetically stabilized by ligands binding to the NC surfaces. This implies ligands stabilize and facilitate the formation of NCs by suppressing growth and isolating NCs in their metastable states. Finally, the kinetic model showed, in a well-mixed solution, a high molar ratio of ligand to metal e.g. 6.00 mM/0.05 mM = 120 inhibits growth and promotes the stabilization of small nanoclusters.</p><!><p>The following files are available free of charge.</p><p>Additional equations and details about the methods (PDF)</p><!><p>*rasaiah@maine.edu</p>

ChemRxiv

VoteDock: Consensus Docking Method for Prediction of Protein\xe2\x80\x93Ligand Interactions

Molecular recognition plays a fundamental role in all biological processes, and that is why great efforts have been made to understand and predict protein\xe2\x80\x93ligand interactions. Finding a molecule that can potentially bind to a target protein is particularly essential in drug discovery and still remains an expensive and time-consuming task. In silico, tools are frequently used to screen molecular libraries to identify new lead compounds, and if protein structure is known, various protein\xe2\x80\x93ligand docking programs can be used. The aim of docking procedure is to predict correct poses of ligand in the binding site of the protein as well as to score them according to the strength of interaction in a reasonable time frame. The purpose of our studies was to present the novel consensus approach to predict both protein\xe2\x80\x93ligand complex structure and its corresponding binding affinity. Our method used as the input the results from seven docking programs (Surflex, LigandFit, Glide, GOLD, FlexX, eHiTS, and AutoDock) that are widely used for docking of ligands. We evaluated it on the extensive benchmark dataset of 1300 protein\xe2\x80\x93ligands pairs from refined PDBbind database for which the structural and affinity data was available. We compared independently its ability of proper scoring and posing to the previously proposed methods. In most cases, our method is able to dock properly approximately 20% of pairs more than docking methods on average, and over 10% of pairs more than the best single program. The RMSD value of the predicted complex conformation versus its native one is reduced by a factor of 0.5 \xc3\x85. Finally, we were able to increase the Pearson correlation of the predicted binding affinity in comparison with the experimental value up to 0.5.

votedock:_consensus_docking_method_for_prediction_of_protein\xe2\x80\x93ligand_interactions

Introduction<!>Materials and Methods<!>Benchmark Dataset<!><!>Docking Software<!>Docking Accuracy<!>Initial Ligand Conformations Ensembles<!>The binding affinity<!>Selection of the optimal pose<!>The strongly binding poses<!>The consensus docking method<!>Results<!>Quality of pose generation<!>Evaluation of binding affinity prediction<!>Consensus docking strategy<!>Conclusions<!>

<p>A typical drug design campaign requires substantial costs and is time consuming due to the fact that for thousands of chemical compounds biochemical screening has to be performed before proceeding to a more refined analysis. The in silico methods promise to shorten the time and decrease the amount of work needed when searching for a new inhibitor. One of the most important methods used here is the molecular docking that predicts a preferable conformation of a ligand when bound to a receptor molecule. Docking is used frequently in a high-throughput virtual screening where large libraries of commercially available compounds are searched to find the most active compound for a selected protein target. The aim of docking procedure is to predict the correct pose of a ligand in the binding site of the protein as well as to score it according to the strength of interaction in a reasonable time frame. As all programs exploit empirically based scoring functions and algorithms, docking results are sometimes far from reality.</p><p>Among the most frequently reported is the docking accuracy of small organic compounds to a given protein,1–6 yet the nucleic acids can also be considered as a target for ligand molecules.7,8 In the protein–protein docking,8–10 the interactions between two identical or different proteins are studied. In the case of protein– ligand docking, various algorithms address different representations of a ligand and a receptor, their intrinsic chemical properties, and detailed characteristics of intramolecular interactions between their atoms. As in recent years, the crystallography and multidimensional NMR provided a wealth of structural information about various biological targets, several protein–ligand docking programs have been proposed.11,12 Usually, the receptor is treated as a rigid molecule because of high computational costs, whereas conformational flexibility of ligands is taken into account leading to different placement algorithms.13 The scoring procedure of such docked conformers is still regarded as one of the most difficult tasks in molecular docking because of their empirical nature. In our work, we used only software that considers flexibility of ligands, not proteins, and thus structure of protein before docking was not changed in comparison with original pdb file, assuring that protein is already in bounded state.</p><p>There are three major goals of docking simulations: (1) the native conformation of ligand in the active site should be predicted; (2) the binding energy should be estimated allowing for arrangement of the tested set of ligands according to their affinity toward the protein target; (3) in addition, it should be fast enough to screen large collections of small chemical molecules. The typical docking procedure is performed in two steps. The first step is focused on placing a small molecule into the binding site of the protein using mostly geometrical features and searching for its best three-dimensional (3D) conformation inside the cavity. The second step is performed using different scoring functions and it leads to the estimation of the binding affinity between the protein and the ligand.</p><p>During the last two decades, a set of different docking programs has become available both for commercial and academic use, such as DOCK,14 AutoDock,15 FlexX,16 Surflex,17 GOLD,18 ICM,19 Glide,20 CDocker,21 LigandFit,22 MCDock,23 and many others. They are based on different algorithms and can be grouped into four general categories: stochastic Monte Carlo, fragment-based, genetic algorithms, and, finally, shape complementary methods. None of those programs uses systematic search to fully explore all degrees of freedom in both receptor and ligand molecules because of enormous computational cost of such a procedure.2 That is why docking programs avoid systematic search and perform only guided search in conformational space. Our consensus algorithm attempts to combine those independent docking approaches into a single and powerful prediction method. We select a set of representative conformations from each docking algorithm to efficiently inspect different guided search algorithms for correct conformation of a protein–ligand complex.</p><p>The binding affinity of generated output protein–ligand conformations is calculated here by using different scoring functions. More than 30 different scoring functions were published until 20092,24–40 and they can be classified into three major categories. The first group applies force fields functions to calculate the energy of a complex as the sum of the ligand and the receptor internal interaction energies and also the energy of intermolecular interactions between those two molecules. Typically, the force fields such as Assisted Model Building With Energy Refinement (AMBER)41 or Tripos42 are employed, considering two energy terms, i.e., van der Waals and electrostatic interactions between molecules. Additionally, to improve the accuracy of those functions, sometimes the solvation energy term is also included, usually using a distance-dependent dielectric function.43 Most of the docking programs do not support ligand binding to protein via covalent bond. However, when applied to protein–ligand complexes, the force fields are often found to overestimate the binding affinity,2 even when using very precise and time-consuming procedures. Therefore, the scaling coefficients multiplying both terms are used to resolve this problem. The second group, i.e., the empirical scoring functions, describes interactions between a protein and a ligand as scalable parameters. Almost all of the proposed parameters exploit hydrogen bonds, hydrophobic interaction, metal bonds energy, typical force fields energies, and finally, the solvation energy term. The scaling parameters together with the empirical functions are trained on the selected dataset of complexes with known binding affinity for which scaling factors for each energy term can be optimized. Empirical scoring functions are often able to recreate binding affinities of original training dataset with very high accuracy,24 yet the results on previously unconsidered protein– ligand complexes are not always successful. The third group, namely knowledge-based scoring functions, is developed from the statistical analysis of X-ray and NMR structures of protein– ligand complexes. The distribution of different pairs of atom types is gathered using a set of pairs of atoms, one from a protein and the other from a ligand, and then converted into pair-wise atom–atom statistical potentials. The final interaction energy is calculated as the sum of all pairwise interactions between atoms from a ligand and a protein lying within the sphere of the given cutoff (usually from 6 Å up to 12 Å).</p><p>The consensus is a novel technique recently used in various applications, mostly in bioinformatics. The main rationale behind is that although individual approaches can generate some misleading results, yet the distribution of those errors is random.32 That is why even a simple majority voting of a set of programs providing different results can be in principle much closer to the correct answer, than even the best single program. In the context of docking problem, several attempts to transfer that approach have been made. However, as the general opinion is that posing is not the main drawback of docking programs, typically consensus approach is applied in prediction of ligands activities. Nevertheless, some cases where authors applied this technique to poses selection were also reported. For example, Wolf et al.44 merged two docking algorithms, namely genetic-and fragment-based method into single AutoxX protocol. The software used FlexX and AutoDock algorithms for choosing optimal ligand conformation, and it was able to decrease the mean root mean square distance (RMSD) of top score conformations by 0.3 Å in comparison with best individual program from those two. This approach allowed to predict correct conformation of ligand for 126 pairs of the 206 tested (RMSD below 2 Å), more than six for AutoDock alone. However, no consensus scoring was proposed there, thus, scoring functions were omitted and reported separately from those two programs.</p><p>Up to now, the research community focused mostly on improving scoring predictions, because in common opinion, calculating a ligand in vitro activity is very difficult task. Therefore, typical strategy is to gather data from diverse set of scoring functions representing different approaches to create new function using simple linear regression technique. Typically, this procedure allows for development of the function working for specific protein families; therefore, it cannot be transferred from one family to another. Similar approach was used by Teramoto et al.28 where authors used the support vector regression performed on three protein families, acetylcholine esterase, thrombin, phosphodiesterrase 5, and proliferator-activated receptor gamma. New functions were used as an input scoring results obtained from F-score, D-score, Potential of Mean Force (PMF), G-score, and ChemScore. Those authors in 2007 used "rank-by-vote" approach, where instead of the absolute scores values, each ligand was given the rank based on its position in ligand list ordered by particular scoring function. Ligand with lowest average rank from the set of scoring functions was then chosen in this method as the most active one. Similar approach is also used in Sybyl's Consensus Score (CS) model. The successful modification of "rank-by-vote" approach was implemented in SeleX-CS algorithm developed by Bar-Haim.32 Here, the Monte Carlo simulated annealing is used to choose functions that can vote for a particular ligand. Two types of votes are allowed: "primary" rank-by-vote value, and "secondary" rank-by-number value. Authors reported three times increase in enrichment factor value obtained for studied small set of proteins. Summarizing, according to our knowledge, no single workflow that combines consensus both in pose prediction and score prediction has been introduced up to now.</p><p>Here we propose the consensus docking protocol that allows for massive docking of ligands into their corresponding protein targets using several independent docking algorithms and scoring functions running in parallel. Our approach combines the results from various programs into a single consensus prediction of the 3D protein–ligand complex structure. The clustering of results from those several docking algorithms is performed to select the poses that are close to the corresponding native conformation, and then the consensus scoring is performed using the multivariate linear regression to select the strongly binding conformations. The consensus docking method is evaluated here in terms of both posing and scoring abilities on the large dataset of protein–ligand complexes with known 3D structures and binding affinities.</p><!><p>Here, we present a novel docking method for selecting potent inhibitors using the results of docking performed by several programs. Seven docking software packages were used to perform the consensus procedure (AutoDock 4.2.1, Glide 4.5, GOLD 3.2, Surflex 2.2, FlexX 2.2.1, eHiTS 9.0, and LigandFit2.3). This selection covers a variety of types of docking algorithms, thus representing a rich data source for optimizing the consensus between the most popular docking programs. The method is optimized on a large benchmarking dataset of 1300 protein– ligand pairs to provide the accurate posing (RMSD value for each predicted conformation versus the corresponding native one and the percentage of successfully docked pairs in the whole collection of inhibitors) and the scoring ability (correlation of the obtained score with experimental pKd or pKi value). The predicted consensus pose for a given ligand on the protein target has on an average lower RMSD value in comparison with that obtained by any individual program. Furthermore, compounds predicted by us as the active ones, on an average have higher correlation coefficient calculated using the experimental binding value than scores predicted by any particular scoring functions.</p><!><p>To benchmark our method, we selected the PDBbind 2007 data-base45–47 containing 3124 protein–ligand complexes with known 3D structure and the corresponding ligand-binding affinity. The dataset is selected as the richest and diverse dataset used in various evaluation studies.30,48 From it, authors of PDBbind extracted a subset of 1300 protein–ligand pairs creating its "refined" set that was used in this work (see Supporting Information Table S1). The requirements for a given complex from the Protein Data Bank (PDB) database 45,46 to be included in refined, and in consequence, in our benchmark dataset were as follows.</p><p>The experimental resolution of a crystal structure has to be lower than 2.5 Å. Other studies performed previously on the GOLD original benchmark set by Jones et al.18 confirm that selecting structures with poor resolution may produce false predicted conformations. Both a ligand and a protein structure have to be complete without any chain breaks or unsolved regions. No NMR-solved structures are involved in creating the refined data-set. In addition, only complexes with known binding affinity are considered for the refined set. The activity should be given as either pKi (an inhibition constant), or pKd (a dissociation constant). The results given as IC50 are rejected as such values depend on the design of a binding assay. However, recently, the accuracy of this data was questioned49 as for 36% of the complexes the calculations of binding affinities were affected by crystal artifacts such as water molecules. We excluded complexes with ligands containing other than standard atom types (like Be or Si). Ligands that are not covalently bound with protein were discarded. Ligand mass should not be larger than 1000 amu. A complex is rejected if the distance between its ligand and the protein heavy atoms is closer than 2 Å. Finally, only complexes with a single ligand in the active site were chosen for docking simulations.</p><p>The native conformation of each ligand was extracted from the protein–ligand complex; similarly, the corresponding protein target's 3D structure was prepared. Each ligand was then converted into a two-dimensional representation; later the Simplified Molecular Input Line Entry Specification (SMILES)50 chemical name of each inhibitor was created by using OpenBabel (http://openbabel.sourceforge.net/) and Marvin software (http://www.chemaxon.com). Then, the 3D input ligands structures were generated ab initio from their SMILES names using two typically used programs, namely the Corina51 and Omega252. Therefore, four different datasets were created. First, as a starting point, a single low-energy conformation for each ligand was generated using the Corina program. The second dataset consists of 10 conformers for each ligand generated by Corina. The third dataset contains only a single low-energy conformation for each ligand generated using Omega2 software. Finally, the fourth input dataset for docking was created using 10 of the conformers generated by Omega2 for each ligand of the PDBbind 2007 database. In addition to performing the optimization and evaluation of our docking procedure, we tested whether the docking results depend on the ligand size, type, and structure, such as the number of rotatable bonds in a molecule or its chemical properties such as hydrophobic and hydrophilic features given by the ligand partition coefficient between water and octanol: (1)logPoct/wat=log(CoctCwat). Proteins extracted from crystal structures undergo the following preparation steps. First, hydrogen atoms are added with the protonation state simulated to pH = 7. Therefore, aspartate and glutamate amino acids were negatively charged, histidine was neutral, and arginine and lysine amino acids were positively charged. The terminal carboxyl groups were deprotonated, whereas amine groups were protonated. Atoms and bonds types were inspected by using Sybyl software, yet no geometry optimization was performed. In addition, we decided to remove all water molecules and metal atoms from the protein pdb files, as no significant changes in docking accuracy of our method were observed (for more details see Supporting Information Table S2).</p><!><p>Fragment-based incremental methods: Surflex (Jain, A.N. et al.), eHiTS (SimBioSys Inc.) and FlexX (BioSolveIt) that splits ligand into pieces which are docked in an incremental way;</p><p>Evolutionary methods: GOLD (CCDC) and AutoDock (The Scripps Research Institute) that use genetic algorithms;</p><p>Force field-based method: Glide (Schrodinger Inc.), which has implemented Monte Carlo based engine;</p><p>Shape complementary-based method: LigandFit (Accelrys Software Inc.), which exploits grids to fit the shape of the ligand to the target.</p><!><p>All docking programs were tested using their default parameters (see Supporting Information Table S3 for more details), although we are aware that proper selection of the scoring function and docking algorithm parameters can impact the obtained results. However, in the case of our diverse and highly populated benchmark dataset, the additional computational time used for such testing and further optimization of those parameters would make it impossible to perform using our limited hardware resources.</p><!><p>The ability of docking software to predict the correct ligand poses close to the native one (found in X-ray complexes) is crucial in achieving success in docking experiment. Typically, two approaches are used for evaluating the success. The first one describes how many specific contacts between the ligand molecule and the protein are recreated (for example, hydrogen bonds or hydrophobic interactions) rather than focusing on exact placement of all ligands atoms. In our studies, millions of poses are generated; therefore, it is impossible to follow such a detailed protocol. Instead, we decided to use the second approach, i.e., calculation of the RMSD value for heavy atoms between the predicted pose and the corresponding native conformation of the cocrystallized ligand. Such a measure is well known and accepted as a reliable structural quality parameter by the whole docking community. The RMSD value between two poses is given by the equation: (2)RMSD=1N∑i=1Ndi2, where N is the number of atoms and di is the distance between the corresponding atoms.</p><p>Of all poses that are predicted by a given docking program, one can select two poses of crucial importance. The first one is the conformation returned with the highest score by the corresponding scoring function. Here, such a pose is called the top score pose and its RMSD value to the native ligand conformation is calculated for each docking program and all analyzed protein–ligand complexes. In our approach, each program was run to predict 10 different poses for each input molecule. Therefore, for each program, we select the best pose conformation of all predicted ones, which has the lowest RMSD value to the native conformation. Those two RMSD values are very useful for benchmarking purposes, yet the best pose cannot be used in real-life experiments, where the native structure of the protein–ligand complex is not known. Usually, the best pose conformation is not returned with the highest score by a program, so it is not the top score pose.</p><p>The second useful quality parameter is the percentage of protein–ligand complexes for which the RMSD value of the top score pose is lower than 2 Å, a threshold commonly assumed as the acceptable accuracy by the docking community. In this work, we use two values of this measure, one calculated using the top score poses, and the other using the best pose conformation. Contrary to previously described RMSD values, the percentage of successfully docked complexes does not depend much on the results obtained for wrongly predicted complexes, which can significantly change the mean RMSD value averaged over the whole benchmarking dataset.</p><!><p>The number of selected initial conformers for docking evaluation study may impact the docking results, and, in fact, many previous benchmarks reported that docking the native structure of a ligand extracted from X-ray crystallographic structure provides better results. Therefore, we decided to find out what is the optimal number of the prepared conformers that should be used in docking procedure and further in the consensus approach. Typically, two programs are used in research community: Corina and Omega2 to recreate crystallographic structure of molecule. Therefore, we have used them both to generate the predefined number of conformers for each input 2D ligand chemical representation. We tested the quality of those predicted 3D structures by comparing them to the known 3D structure. We have compared three ensembles of initial ligand conformations: a single most optimal structure, 10 low-energy conformers, and, finally, the 100 of predicted conformers. We have noticed that 10 conformers ensemble seems to be the optimal number, because for almost all ligands, we were able to find at least one conformer within those predicted that was almost identical to the native 3D ligand structure (RMSD below 1 Å). This observation was confirmed for both Corina and Omega2 programs. Therefore, we decided to perform four different docking experiments for each protein–ligand pair using four different ensembles of ligands' initial conformations. In two cases, we have selected only single low-energy conformations (comprising corina one and omega one datasets). In next two cases we docked two ensembles of 10 conformations generated by either Corina or Omega2 (comprising corina ten and omega ten datasets). Therefore, we collected the docking results in those four independent experimental setups; each performed docking using different ensembles of the input ligand 3D conformations. The evaluation parameters (such as the RMSD values or the percentage of correctly docked complexes) were then averaged over those datasets, as no significant changes between those were observed. In addition, we carefully checked the docking results performed on the subset of complexes using hundreds of input conformers. The results were quite similar (in terms of both the minimal RMSD value and the percentage of accepted pairs below 2 Å). However, it takes on an average almost 10 times longer computational time in comparison with the docking of 10 conformers; therefore, the experiment was not repeated on the whole benchmarking dataset.</p><!><p>Another important issue in prediction of protein–ligand complexes is the ability of docking programs to correctly predict the strength of ligand binding to its protein target, i.e., the binding affinity. It describes the strength of intermolecular binding between the ligand and its receptor, and it can be described by the number of parameters, such as dissociations constant Ki, concentration of ligands that decrease the activity of particular enzymes by 50% IC50, or by the Gibbs free energy ΔG. The calculation of binding affinity was done here by each docking program internal scoring function. The PDBbind database reports the experimental values of the activity for all evaluated protein–ligand complexes. Therefore, we are able to compare the docking score with the corresponding experimental value of binding affinity, and to calculate the Pearson correlation coefficient between those two values. Scoring functions should order the list of poses in accordance with their binding strength to select those that are close to the native structure (hopefully the strongest bound conformation).</p><p>In our benchmark, we used the following scoring functions: GoldScore, LigScore1, GlideScore SP, Total score (the combination of several scoring functions used by Surflex: Chem-Score, F-Score, PMF-Score, and others), FlexX score, eHiTS score, and AutoDock scoring function. For each scoring function, we calculated Pearson and Spearman correlations for four different sets of conformers (generated using Corina and Omega2 software). The Spearman correlation is much less sensitive to a few outsiders, and on the contrary, a few wrong protein–ligand complexes can significantly affect the Pearson correlation coefficient.</p><!><p>Our benchmarking results show that no single docking software is more reliable than others. Therefore, the consensus approach seems to be the right tool to boost the overall docking accuracy. Several consensus approaches were successfully applied in the context of bioinformatics,54 chemoinformatics,55 and general computer sciences. For example, the 3D-Jury53 algorithm predicts a 3D protein structure using several autonomous methods. Here, we present our novel method MetaPose that is able to improve the selection of the best pose using a set of conformations obtained from various docking programs. The n predicted poses from seven tested programs are used, neglecting the scores obtained using their scoring functions (n = 70 for Corina and Omega2one, and n = 700 for Corina and Omega2 ten input datasets). Those 3D conformations are compared with each other by calculating the RMSD value between them. Even if a subset of the predicted poses is obviously wrong, or contaminated (for example, as a result of weak docking program), yet this does not affect our method because the majority of molecules would be placed correctly in a protein active site. In this way, the similarity matrix is created, where the ith row represents the similarity of the pose i to all other poses. The pose score 3DScore for this conformation is calculated as the arithmetical mean of RMSD values from the selected row: (3)3DScorei=1n∑j=1nwij, where n is the number of conformations and wji is the RMSD value calculated between i and j poses.</p><p>The pose with the lowest value of the score is selected by our method as the final result. We search here for the conformation that represents the entire set of possible poses; therefore, it is most similar to others. The conformational search of the MetaPose method is based only on the geometrical similarity between poses, without taking into account the scores given by docking programs.</p><!><p>The second step of our analysis is designed to improve the correlation between the docking score and its experimental values of the binding affinity. The MetaScore algorithm builds the consensus scoring function using multivariate linear regression optimization guided by the experimental results. We assume that the consensus scoring function is described by a linear combination of scores from seven docking programs with the weights describing the influence of each scoring function on the final result.</p><p>The optimization method has to be constructed and tested separately using different and not overlapping datasets of protein–ligand complexes. Therefore, we divided our benchmark dataset into two parts. The first dataset contains 400 pairs and was used for optimization of the new metascoring function. The second dataset (the rest of database) was used as the independent testing dataset. Each protein–ligand complex can be represented as an ensemble of seven hundreds or tens of poses (depending on the size of input conformers used in docking); therefore, each pose can be described by seven different docking scores of top score poses. The multivariate linear regression was then used to obtain the single MetaScore function for our four benchmarking subsets, namely Corina one, Corina ten, Omega2 best, and Omega2 ten generated conformers. However, surprisingly, the coefficients for each consensus scoring function were tested to be almost equal for each of those four optimization datasets, therefore the single MetaScore function can be proposed, and it is given by the equation: (4)MetaScore=−0.378*eHiTSScore+0.015*FlexXScore−0.358*GildeSPScore+0.014*GoldScore+0.004*LigScore+0.15*SurflexScore, where each scoring function is described by the name of the corresponding docking program. Result of this function is value that represents ( −logKd) value for a given protein–ligand complex.</p><p>The MetaScore subset of poses, i.e., all top score conformations from seven docking programs, are used for consensus procedure. We created the 7 × 7 similarity matrix calculating the RMSD values between those seven top score conformations. Then, for each of them, the 3DScore was calculated and the conformation with the lowest value was chosen as the predicted pose. The results of this procedure are summarized below and are presented in Tables 1 and 2 (the MetaScore row).</p><p>In addition, having all conformations ordered by their corresponding 3DScore value, we chose the top representative for all seven docking program, each with the lowest 3Dscore value of all predicted poses by this program. MetaScore eq. (4) can be calculated using docking scored of those six conformations. Similarly, the Pearson correlation between those values and experimentally determined binding affinities can be calculated. The results are presented in Table 1.</p><!><p>In the third step of our analysis, we designed the consensus method that is able both to predict the correct conformation of a protein-ligand complex, and its binding affinity value. Previously described methods, namely MetaPose and MetaScore, focus on different goals for virtual screening. The first algorithm selects the pose inside the protein active site that is close to the native one, and the second one focuses on the calculation of correct binding affinity for the analyzed ligand. Here, we introduce VoteDock algorithm that predicts both the correct pose and the strength of binding between the ligand and the receptor. The VoteDock uses modified MetaPose algorithm to select the correct conformation. The MetaPose neglects the information about the source of analyzed conformations (namely the docking programs from which they are taken). Therefore, although it is proven to be more effective than any individual program, yet the influence of each docking program is similar for all used algorithms, even if there is a single one among them that has a very weak ability to pose the ligand inside the active site, or to score it.</p><p>For each protein-ligand complex, we create subsets of poses, each containing the poses selected independently by a certain number of docking programs. Therefore, we can assign to each predicted conformation the number from one to six depending on the number of programs that confirmed that pose as the correct one. The similarity matrix for each subset of poses is calculated separately for all poses that were confirmed by the predefined number of docking programs, and called vote2, vote3, up to vote7. For example, if a conformation from GOLD was also predicted by eHiTS, i.e., if there exists at least one conformation from eHiTS's predicted poses that has the RMSD value between it and the GOLD's conformation lower than the threshold value of 2 Å, such a conformation is qualified to be included in vote2 subset. If the next docking program (for example FlexX) has another predicted conformation closer than 2 Å from the original pose, then the pose is also included in vote3 subset. In the case of vote7 dataset, the highest quality predicted conformations; all seven docking programs predicted each of them as the possible ones in our test. The voting procedure not only narrows down the number of predicted conformations but also eliminates incorrect poses that influence the center of the solutions domain. In the case of some protein–ligand complexes, the vote7 dataset is empty within the given RMSD threshold, therefore, the hybrid approach is proposed. For every protein–ligand pair, we select the subset of predicted conformations, which have the highest available vote order, and subsequently, we use MetaPose approach on such highest vote dataset. The similarity matrix is constructed and the 3DScore [see eq. (3)] is assigned to those poses. The conformation with the lowest value of 3DScore is chosen as the best one. If no vote subset is present (even the vote2 subset), as the result for such a protein–ligand complex, the pose selected by original MetaPose algorithm is returned.</p><p>The conformations selected by the first step of VoteDock procedure are then scored using the MetaScore scoring function, as each pose is now described by more than one docking score. A detailed analysis of vote subsets allows one to select a single program that is eliminated from each vote. In the case of vote7, the scores from all seven programs describe each predicted conformation. In the case of vote6, the AutoDock predictions are excluded for most of the complexes, vote5 eliminates Glide, vote4 FlexX docking program, vote3 removes LigandFit, and finally, in vote2, in most cases, eHiTS is lacking, leaving only two programs, namely Surflex and GOLD. The MetaScore procedure is here modified by optimizing weights of each docking program for each vote dataset separately, using multivariate linear regression (details on scoring function used here are listed in Supporting Information Table S4). Therefore, six new Meta-Score functions are calculated, each for the particular subset of docking programs. To compare correlation values from the VoteDock with individual docking programs, here we use the hybrid approach similarly to the previously described optimal conformation selection prediction procedure. We select the highest vote subset for each protein–ligand complex and apply the MetaScore optimized function suitable for this particular vote order. If no vote is present or a conformation is described by a different combination of programs, we apply the original Meta-Score procedure. The highest scores from the available program are collected, and the eq. (4) is used to calculate the predicted value of the ligand binding affinity. The workflow of data is presented in Figure 1, together with the schematic diagram showing how consensus methods work.</p><!><p>In this section, we summarize the quality of the three proposed consensus methods, namely MetaPose, MetaScore, and Vote-Dock and compare them with the results of seven diverse docking programs: AutoDock, eHiTS, FlexX, Glide, GOLD, LigandFit, and Surflex. First, we describe the ability of all used algorithms to correctly predict the ligand binding poses. In addition, we test whether the consensus docking results depend on the ligand size, type, and structure, such as the number of rotatable bonds in a molecule or its chemical properties like hydrophobic and hydrophilic features given by the ligand partition coefficient between water and octanol. Then, we report the ability of all programs' scoring functions to accurately predict the experimental binding affinities (pKi and pKd). The proposed novel methods that are based on the consensus between various docking algorithms are proved below to considerably increase docking accuracy and proper sorting of predicted poses.</p><p>In Tables 1 and 2, we present the evaluation of our consensus algorithms in both, the correct poses and the binding affinities prediction. Each consensus method is compared with the results obtained by the best and the second best docking program on the whole benchmarking dataset. In addition, we provide the mean docking result calculated by averaging the results of all seven docking programs on the same dataset.</p><p>We also explored several physicochemical features of the ligand, which are often used in various docking programs evaluations. First, we divided our datasets using the number of rotatable bonds in a ligand molecule. We created subsets of small compounds, which have five bonds or less, and large compounds, which have more than five rotatable bonds. It is obvious that for smaller molecules, the results will be better, yet our main goal was to identify, which program depends less on ligand size. Next, the hydrophobic/hydrophilic properties of ligands were analyzed. As previously, two datasets are created using the logp values. This parameter also covers the number of possible hydrogen bonds, which a ligand can create with a protein, as for more hydrophobic ligands, fewer contacts are usually built. Another ligands subset that we explored contains proteinlike molecules. This subset is interesting because of the growing number of protein-like drugs that are introduced to the market. We wanted to evaluate the quality of prediction for those types of molecules. Finally, the benchmarking dataset was divided based on the strength of the binding between the ligand and its corresponding receptor. Here, our goal was to determine if there is a preferable compound type that docking programs could handle more precisely, for example, small and strongly binding molecules. The results of those evaluations are presented for two preselected conformations: top score conformation and best pose conformation. The top score pose for each docking program, or consensus method, is the conformation that achieved the highest docking score of all generated by the program, whereas best pose is the one with the lowest RMSD value to the native conformation. In the case of MetaScore, top score conformation has the lowest 3DScore value of all conformations with the highest docking score from individual programs, whereas best pose has the lowest RMSD value of those conformations. MetaPose and VoteDock algorithms similarly have top score pose as the pose with the lowest value of 3DScore function among all poses generated by those consensus algorithms (MetaPose), or among those from the highest order vote subset of conformations, preferably vote7. In the case of MetaPose and VoteDock, the best pose was selected as the pose with the lowest RMSD value of the first 10 (corina one and omega one), or 100 (corina ten and omega ten) poses ordered by the 3DScore. Those limits in the number of analyzed poses simulates the use of our consensus algorithms as the stand-alone programs; therefore, they would generate only tens or one hundreds of poses.</p><p>In Table 1, we present the Pearson and Spearman correlations between the experimentally determined binding affinities and the scores from all scoring function. In our work, those correlations are calculated for both top score conformation and the best pose. To calculate the correlation between the score of best pose for MetaScore and MetaPose algorithms, the scores of best poses for individual docking programs are used, and the total score is calculated using eq. (4). However, MetaPose uses only the first ten or one hundred conformations, and sometimes some docking programs are not represented in this dataset, therefore, not always all six scores were used in eq. (4), and zero value was used for such missing docking scores. In the case of VoteDock, instead of the highest 3DScore conformations, we used individual docking scores for conformations with the lowest RMSD values of the first ten or one hundred conformations as ordered by their 3DScore values. Our evaluation was done separately on four different datasets (corina one, corina ten, omega one, and finally omega ten), therefore, the values in Tables 1 and 2 are averaged over those subsets.</p><!><p>The best docking program is the GOLD software, which in top score prediction outperforms other programs. However, the GOLD program uses the slowest, time-consuming algorithm that takes more than 5 min to dock a ligand to the receptor. In Table 1, GOLD was not chosen as the best program only twice, for the hydrophobic dataset of ligand eHiTS is the first one, and GOLD is the second one. The weakness of GOLD program in the case of hydrophobic ligands was already pointed out in other benchmarks.1 The GOLD scoring function was also not the first one in terms of the correlation with the experimental binding affinity (see Table 1). We report the eHiTS and Surflex scoring functions as the best and the second best program. In Table 2, we present the results of the pose generation that are analyzed in terms of ligand binding strength, where we report GOLD as the first program; however, in the case of strongly binding and small ligands, eHiTS achieves better results. In most cases, eHiTS is the second best docking program and sometimes switches its position with GOLD. In the case of protein-like ligands, AutoDock is chosen as the second program.</p><p>In the case of best pose presented in Tables 1 and 2, the best docking programs are usually the same as in the case of top score conformations. More diversity is observed for the second best program, namely Surflex (large ligand and hydrophobic ligands datasets) and LigandFit (protein dataset) are better than others. For the entire benchmark dataset of 1300 complexes, MetaPose algorithm is nearly 18% more accurate than averaged docking results, and its mean RMSD value drops by more than 1 Å. A smaller but still significant change can be observed when comparing MetaPose to the best docking program, where the increase is almost 5%, and the RMSD value is improved by almost 0.2 Å. Even more accurate is the VoteDock consensus docking method where the average docking accuracy increases by almost 23% in comparison with the average docking programs accuracy, and by more than 10%, when compared with the best result obtained by GOLD. The mean RMSD value is also increased by 0.5 Å in comparison with GOLD. In the case of MetaScore, structural results are above average docking, yet less than a 10% increase in successfully docked pairs can be observed. However, when compared with the best and the second best program, the obtained results are not so good as before. Therefore, the results on top score subset prove that MetaScore is very efficient in the prediction of binding affinities, yet this does not correspond to the good overall structural prediction.</p><p>The difference between MetaPose and the best docking program for all analyzed types of physicochemical features is typically around 5% increase in docking accuracy and 0.3 Å in the mean RMSD value. The VoteDock is usually even more accurate with a 10% increase in comparison with the best docking program, and the mean RMSD value usually drops by more than 0.6 Å. A similar difference can be seen when comparing the dataset composed of small molecules with one composed of large molecules. The number of ligand rotatable bonds describes its flexibility. The worst docking results are obtained for large ligands (the high number of rotatable bonds), which is due to the fact that the size of explored conformation space increases dramatically. MetaPose is on average 3 and 4% more accurate than the best docking program for, respectively, the small and large molecules subsets, whereas for VoteDock, 9 and 11% increase is observed. The gap in docking accuracy between small and large dataset of ligands is smaller when a consensus-docking algorithm is used. In the case of the best docking algorithm, there is an almost 20% drop in accuracy between small and large datasets. In the case of VoteDock, the drop of accuracy is close to 15%, MetaPose achieves intermediate results of around 17%.</p><p>The same conclusions can be observed when dividing the entire benchmarking dataset using hydrophobic and hydrophilic characteristics of ligands. Those features result from some important aspects of ligand behavior, mostly the ability to create hydrogen bonding between a ligand and a protein, as well as forming interactions with the hydrophobic pocket in the protein active site. As expected, hydrophilic ligands are predicted with a much higher rate of success than hydrophobic ones. However, it should be remembered that for the hydrophilic dataset, GOLD was chosen as the best program, and for the hydrophobic data-set, eHiTS was most successful. MetaPose and VoteDock are more accurate than any individual docking software. For hydrophilic ligands, there is, respectively, a 2 and 7% increase between single docking and the consensus result. Those two metaalgorithms are even more accurate for hydrophobic ligands with a 9 and 14% increase in docking accuracy. Those results clearly suggest that the consensus approach can be very effective in avoiding the weaknesses of individual docking programs. Similarly, as in the case of the large and small molecules subsets, there is a much smaller gap when comparing hydrophilic ligands with hydrophobic ones. For the consensus approach, the difference is close to 10% in the number of successfully docked pairs, and 0.4 Å for the mean RMSD value, whereas for the single docking programs, those differences are significantly higher, reaching almost 20% and 0.7 Å.</p><p>The increasing role of short peptides as potential drugs such as antibiotics, antihistamine, or antitumor agents create a unique opportunity to benchmark available docking software on known protein–peptide complexes. We have created a small benchmark dataset containing proteins with cocrystallized peptides, or other protein-like molecules. The complexes were extracted from PDBbind 2007 dataset, assuming that a selected ligand contains at least one amino acid-like substructure, and therefore, it is not always identical to the naturally occurring peptides. In the case of such polymers, we have checked the presence of protein bonding between molecule substructures, and structures with nontypical atoms (all types except oxygen, hydrogen, carbon, and sulfur) were discarded. Fourteen complexes contain phosphate atoms, and for six complexes, some fluorine atoms were found. Following this procedure, we have created the peptide benchmark dataset that contains in total 143 complexes for which ligand size may vary from single amino acids up to longer peptide chain created from tens of mers. The best docking software was able to find the correct top score pose only for 46% of those complexes within 2 Å cut off from the native ligand structure. The mean RMSD value for the top score pose was typically higher than 4 Å. eHiTS as the second best docking program achieved a very similar result. Our consensus approaches are able to increase the accuracy up to 50% for MetaPose and 57% for VoteDock. The mean RMSD value decreased to almost 3.5 Å and 3.3 Å, respectively. Therefore, we prove that our method is better in predicting the correct conformations for small proteins inside the receptor active site. Finally, we divided the ligands from the whole benchmarking dataset into three groups, according to their experimentally measured binding affinities to the corresponding protein receptors. The first group (strong) contains ligands for which their concentration necessary to inhibit the enzyme is lower than 45 nM; the second (medium) has their pKi or pKd between 45 nM and 3.6 µM, and, finally, the third group of inhibitors (weak) with the concentration of a compound greater than 3.6 µM. For all those groups, we calculated how many small and large molecules fall into each category, to check if the dependence of the benchmarking results is based only on ligand binding strength and not on its size. In the case of strong dataset, there are 271 large ligands and 159 small ligands; for medium dataset, there are 213 and 222; and finally for weak dataset, there are 165 and 270, respectively. In Table 2, we summarize the results for each subset, additionally divided into small and large molecules. In the case of individual programs, there is a small difference between particular datasets, however, small&weak and small&-medium molecules are usually better predicted than small&-strong ones. A similar trend is observed when looking for large molecules where large&weak molecules are predicted to be 10% more accurate than large&strong ligands, in case of GOLD, best docking program. This result is very unfortunate as in typical drug design strong-binding molecules are searched for. Our consensus method follows individual docking programs trend, however, both MetaPose and VoteDock seem to be less affected when comparing strong to medium, or medium to weak subsets. In the case of VoteDock, the drop of accuracy between for example small&weak and small&strong bound molecules is only 3%, whereas the results for large&weak and large&strong are almost identical. The mean RMSD value seems to change between those datasets only marginally. Similar behavior can be observed for MetaPose algorithm.</p><p>Summarizing, the consensus structural methods (MetaPose and VoteDock) seem to be more effective than any individual docking program. The percentage of successfully docked pairs is 5 and 10% higher than the best individual docking program, and the observed result is even higher when comparing the consensus with the averaged docking results. The significant improvement in the mean RMSD value is observed with VoteDock close to the cut-off value of 2 Å. Similar results are observed when dividing our dataset based on physicochemical properties of a ligand. Interestingly, the consensus methods prove to be successful even when individual docking programs fail. The hydrophobic dataset can be given as an example here, where there is more than 10% increase between the best program and VoteDock algorithm.</p><!><p>The second important issue in the prediction of protein–ligand complexes is the ability of docking programs to predict the strength of a ligand binding to its protein target. The best docking program with the highest correlations is eHiTS for which the correlations are 0.39 and 0.47 for Pearson and Spearman correlations, respectively. Surflex follows the eHiTS scoring function closely, with the correlation equal to 0.3 and 0.34, respectively. The averaged results for all seven scoring functions are around 0.2 for both Pearson and Spearman correlations, proving that programs have significant problems when binding affinities have to be predicted. The consensus docking procedure, namely Meta-Score [see eq. (4)] scoring function significantly improves the Pearson correlation between the final docking metascore and the experimental value of the binding affinity (see Table 1). In the case of training dataset (randomly selected subset of 400 protein–ligand complexes), the MetaScore reaches the value of 0.46 for the Pearson correlation. The optimized MetaScore was later tested on the rest of the protein–ligand complexes from the whole benchmarking dataset that were not used in the training. The Pearson correlation dropped slightly to 0.44, yet still it is much higher than any single docking program. If the MetaScore was trained on the whole PDBbind dataset, the Pearson correlation for the whole benchmarking dataset is equal to 0.48. Similar values were observed for Spearman correlation. What is more, the values of weights multiplying each used docking programs scores were found to be equal for all analyzed subsets used in optimization of the MetaScore equation, for example, for smaller and larger ligands if used separately for optimization procedure. Therefore, the MetaScore scoring function is the universal one, and it is applicable to various types or classes of inhibitors.</p><p>VoteDock is able to increase the Pearson correlation up to 0.49 and Spearman up to 0.5. However, in the case when protein–ligand complexes have at least vote2 or higher order subset and only the optimal combination of single docking programs is found (such conditions are fulfilled for half of the entire benchmarking dataset), the correlation is even higher and is close to 0.6. Summarizing, the drop in the values of Pearson and Spearman correlation for the whole dataset is explained by the fact that only half of the whole subset meets our selection criteria.</p><p>The third algorithm MetaPose achieved the worst results of all consensus approaches close to 0.4 for both Pearson and Spearman correlations, yet it is still higher than the correlation achieved by any single docking program. Nevertheless, MetaPose is designed to be the pose-prediction algorithm and should not be used for binding affinities prediction, as more accurate consensus docking algorithms are optimized and presented in this article.</p><!><p>In Table 3, we present the results for each step of VoteDock procedure. VoteDock is the hybrid approach that uses for an individual protein–ligand pair, the highest possible vote order dataset created by our consensus procedure. In Table 3, we present how many pairs can be classified as each vote order in the hybrid VoteDock procedure based on the selected threshold. The less strict threshold is selected, the more pairs pass to higher vote order dataset. For example, for the threshold of 1 µ for only 43 pairs, the vote7 subset of poses exists, whereas for the less restricted threshold of 3 Å, more than 220 pairs have vote7 as the final VoteDock set of solutions.</p><p>On the other hand, when the threshold increases the docking accuracy decrease is observed. For the threshold of 1 Å, the accuracy is 97%; in the case of the threshold of 1.5 Å, almost 95% is achieved; the threshold of 2 Å has 87%, the threshold of 2.5 Å 81.5%; and finally for the threshold of 3 Å ,only 78% of pairs is docked successfully. A similar trend can be observed for the mean RMSD value, which increases from 0.6 Å for the threshold of 1 Å up to 1.42 Å for the threshold of 3 Å. Therefore, we selected for VoteDock algorithm the optimal threshold equal to 2 Å, as it maximizes both the mean RMSD and the percentage of successfully docked pairs.</p><!><p>Very low values of correlation coefficients between docking score and pKd (pKi) indicate that all molecular docking programs that have been tested are unable to predict binding affinities correctly. Our results clearly show that still there is the lack of scoring function that would be universal for all kinds of ligands and protein families. Although reports of increasing accuracy of scoring functions have already been published,30 yet none of the functions we studied could be classified as reliable, and so further analysis is necessary. On the other hand, protein– ligand docking programs can predict with high accuracy poses of ligands in the binding site of protein. As they are usually very fast, this capability is extremely valuable.</p><p>Our novel methods, namely MetaPose, MetaScore, and Vote-Dock, use as the source data the results of individual programs, and by consensus approach, they attempt to overcome the weaknesses of each docking program to better predict both a ligand conformation in the active site and the strength of that interaction. The VoteDock, which is a combination of MetaPose and MetaScore algorithms, outperforms each individual docking program, both as concerns the correct pose predictions and the scoring. Those observations lead to the conclusion that applying metaapproach is a very successful procedure and worth exploring in the near future when even more docking programs will be available. Although individual programs in some particular cases can be close to our consensus methods, none of them can reach the quality of VoteDock on the large dataset of more than a thousand ligand–protein complexes. The future improvements of the existing software will strongly and positively affect the accuracy of VoteDock algorithm. The consensus will benefit from those advances, and its quality will be further increased.</p><p>Consensus algorithms increase the number of successfully docked pairs up to 70% for VoteDock and 63% for MetaPose, whereas the best docking software in our evaluation reached less than 60% docking accuracy. The mean RMSD of top score pose for VoteDock is equal to 2 Å, and for MetaPose is nearly 2.5 Å, which confirms that the consensus approach is a powerful tool in predicting ligand conformation inside a protein active site. In 93% of the protein–ligand complexes, at least one program was able to predict a single conformation with the RMSD value to the ligand native structure less than 2 Å. We were unable to exceed that value because a consensus method does not create ligands or poses de novo but only allows for selection of poses out of those that were previously generated. Moderate success was achieved in terms of the binding affinity prediction; the correlations for VoteDock and MetaScore are close to 0.5. Although the correlation for the best scoring function in our evaluation is substantially lower and equal to 0.38, we are aware that our results are still about halfway to achieve the perfect 1.0 value. Correlation values for docking programs show that there is still a lot to be done to increase the accuracy of scoring functions. Further improvements of docking software will significantly improve VoteDock accuracy. In addition, if we could generate with VoteDock procedure at least to the level of vote2 for all benchmarked ligands, the overall correlation for them would improve substantially up to 0.6. This value is twice as good as the best docking program correlation (eHiTS score).</p><p>Furthermore, our method could be an important contribution to the vHTS. vHTS is a computational method, which is widely applied to in silico screening of commercial collections of compounds to select the most potent inhibitors for a selected protein receptor. Typically, because of its speed and prediction accuracy several ligand-based methods make use of the information provided by already known inhibitors, such as pharmacophore matching, 3D shapes matching methods,56 or clustering and machine-learning techniques.55,57,58 However, when the target structure is known and there is no prior knowledge about inhibitors, typically docking techniques have to be used.</p><p>In addition, libraries of compounds for vHTS, such as Ligand.Info,59 can contain millions of molecules. The time required for docking large datasets of compounds with the programs we used in this article on typical 100 nodes cluster would be close to 3 months (AutoDock 192 s/molecule, eHiTS 168 s/ molecule, FlexX 34 s/molecule, LigandFit 110 s/molecule, Surflex 100 s/melocule, GOLD 305 s/molecule, and Glide 660 s/molecule). The time of running our consensus docking procedure to analyze all predicted poses is marginal in comparison with the time of individual docking. We need only a few seconds per molecule to process all poses for a selected compound, therefore, the overall increase in time is less than a week. The extra time needed to process docking results will improve the accuracy of predictions by almost 15%, and eliminates many false positives from vHTS experiment saving months of experimental work on biochemical analysis. Recently, our consensus method was successfully applied for screening possible drugs against H1N1 influenza virus; the details will be provided in a forthcoming paper.</p><p>Summarizing, our VoteDock consensus docking algorithm is able to predict a structure of a protein–ligand complex within 1400 s/molecule on the 2-GHz single-core processor. When a set of input ligands or multiple protein targets are used, the time needed to perform the prediction is scaling linearly with the size of the input dataset. However, because of the parallelization on the linux cluster, those calculations can be submitted at once, the overall time is similar to that for a single submission. In the case of large datasets of proteins or ligands (for example, whole proteomes and metalobomes), we suggest performing the vHTS experiment using MLdock service (machine learning–based algorithm55,57,60 combined with ICM-Pro docking58) instead of slower VoteDock algorithm.</p><p>The above VoteDock consensus method will be used in our new internet server that is now under development. Although most of docking programs used in our work cannot be distributed under academic license agreement, we have decided to combine VoteDock approach for pose prediction (using conformations obtained using AutoDock and DOCK software) with scoring prediction using the statistical function SMOG (Small Molecule Growth). Moreover, the VoteDock pipeline allows user with access to local versions of software described above to perform pose selection by combining several files with decoys prepared using those different docking programs.</p><!><p>Additional Supporting Information may be found in the online version of this article.</p>

PubMed Author Manuscript

Plasmon-assisted click chemistry at low temperature: an inverse temperature effect on the reaction rate

Plasmon assistance promotes a range of chemical transformations by decreasing their activation energies.In a common case, thermal and plasmon assistance work synergistically: higher temperature results in higher plasmon-enhanced catalysis efficiency. Herein, we report an unexpected tenfold increase in the reaction efficiency of surface plasmon-assisted Huisgen dipolar azide-alkyne cycloaddition (AAC) when the reaction mixture is cooled from room temperature to À35 C. We attribute the observed increase in the reaction efficiency to complete plasmon-induced annihilation of the reaction barrier, prolongation of plasmon lifetime, and decreased relaxation of plasmon-excited-states under cooling. Furthermore, control quenching experiments supported by theoretical calculations indicate that plasmon-mediated substrate excitation to an electronic triplet state may play the key role in plasmon-assisted chemical transformation. Last but not least, we demonstrated the possible applicability of plasmon assistance to biological systems by AAC coupling of biotin to gold nanoparticles performed at À35 C.

plasmon-assisted_click_chemistry_at_low_temperature:_an_inverse_temperature_effect_on_the_reaction_r

Introduction<!>Experimental procedures<!>Computational methods<!>Results and discussion<!>Mechanistic studies<!>AAC coupling of biotin<!>Conclusions

<p>The illumination of metal nanostructures at wavelengths corresponding to the plasmon resonance maximum leads to collective excitation of free electron oscillations and generation of hot electrons. 1,2 The excited electrons quickly reach nonequilibrium Fermi energy distribution through electron-electron scattering and then lose energy at a lower rate due to electron-phonon scattering. 3,4 Within this time-window, a direct injection of electrons from the metal nanostructure or intramolecular electron excitation may occur in nearby organic molecules possessing suitable acceptor molecular orbitals. [5][6][7][8][9][10][11][12] Excited organic molecules are then involved in (plasmonassisted) chemical transformations. [8][9][10][11][12][13][14] According to current theories, such plasmon-assisted reaction pathways include several events: (i) plasmon excitation of organic molecules by external electron injection or HOMO-LUMO intramolecular electron transition, (ii) vibrational relaxation of molecules in the excited state, and (iii) relaxation of an injected or excited electron, resulting in an "activated" molecule, which has a lower residual barrier for the desired chemical transformation. 15 This residual barrier is then overcome by conventional heating. 6 Under these assumptions, temperature and light act in synergythe light excites plasmons, which activates the nearby molecule, while the temperature is responsible for overcoming the residual thermodynamic barrier. 4,16,17 The relative importance of non-thermal and thermal effects in plasmon nanocatalysis is, however, far from being fully understood. 18 The temperature dependence of plasmon-assisted reaction kinetics could differ signicantly, depending on the value of the residual activation barrier and temperature impact on different stages of plasmon-assisted transformation(s). [19][20][21][22][23][24][25] This may even result in the non-monotonic temperature dependence of the reaction rate as has been recently demonstrated for CO oxidation. 26 In the special case of a completely vanished residual barrier, temperature effects on the lifetimes of excited electrons or molecular excited states can be expected to lead to inverse temperature dependence of the reaction rates: rstly, higher temperature could increase the electron-photon scattering and thus decrease the plasmon lifetime and probability of the reactant excitation. Secondly, it could decrease the lifetime of the electronically excited molecules, making the desired reaction less probable. To the best of our knowledge, this phenomenon has not been demonstrated yet.</p><p>Despite the fact that organic plasmon nanocatalysis is a rapidly evolving eld, detailed mechanisms of the initial stages of plasmon nanocatalysis remain elusive. [27][28][29] We previously demonstrated that plasmon assistance can induce the azide-alkyne cycloaddition (AAC) reaction. 30 AAC is an important biochemical coupling reaction performed either under heating or with the utilization of metal catalysts 31,32 (Cu, Ru, etc.). Carrying out the AAC reaction at low temperatures could make it cleaner and accessible to more delicate substrates. In this study, we report unprecedented inverse temperature dependence observed for the plasmon-assisted AAC reaction and attempt to explain its origin by exploring the reaction mechanisms using the combination of theoretical and experimental approaches.</p><!><p>Used materials, preparation of initial reagents and graed nanoparticles, measurement techniques and control experiments are described in the ESI. † Plasmon-driven azide-alkyne cycloaddition 1 ml of suspension of AuNPs with attached 4-ethynylphenyl groups (further designated as AuNPs-C^CH) in acetonitrile was mixed with 2 ml of 1 mM solution of 4-azidobenzoic acid, and the resulting mixture was illuminated for 3 hours with a light-emitting diode (LED) (660 nm emission wavelength, irradiance on the rst reactor window -20 mW mm À2 ) in a controlled temperature chamber under magnetic stirring. The intensity of the LED irradiation was measured by using Photodiode Power Sensors (Thorlabs, S142C). Aer the reaction, the modied AuNPs were separated by centrifugation (7000 rpm, 20 min), washed with acetonitrile and methanol, and dispersed in methanol to a nal volume of 0.5 ml. 0.1 ml of the resulting suspension was deposited on a silica substrate (0.5 cm  0.5 cm), dried at room temperature and subjected to further SERS analysis. The AAC experiments were performed separately ve times for each temperature. In turn all SERS measurements were also repeated 5 times at different spots on the sample surface. The obtained SERS results were evaluated as averaged peak intensity with error bars corresponding to the deviation of characteristic signals from separated samples and uncertainty in the framework of single samples. Spectra were plotted according to ref. 33.</p><p>Plasmon-driven azide-alkyne cycloaddition with biotin azide 1 ml of AuNPs-C^CH suspension in acetonitrile was mixed with 0.5 ml of 1.6 mM solution of biotin azide and the suspension was illuminated for 3 hours with LEDs (660 nm, 20 mW mm À2 ) at À35 C in a controlled temperature chamber. Aer the reaction modied AuNPs were separated by centrifugation (7000 rpm, 20 min), washed with acetonitrile, methanol and dispersed methanol to a nal volume 0.5 ml and 0.1 ml was deposited on a silica substrate (0.5 cm  0.5 cm) and dried at room temperature for further analysis.</p><!><p>The reaction pathways were calculated with the Gaussian 16 rev. A.03 program, [34][35][36][37][38] employing density functional theory (DFT). Specically, we used the B3LYP 39 functional with D3 Grimme's dispersion correction 40,41 and the 6-311++G(d,p) basis set. All reported minima and transition state structures were conrmed by the calculation and diagonalization of their Hessian matrices. The reported energies are Gibbs free energies calculated with the SMD method in acetonitrile solvent 42 at 298.15 K and the 1 bar standard state. The singlet excited state energies were calculated employing the TD-DFT method. 43 The energy of S 0 heat was calculated with the non-equilibrium geometry and is therefore only a crude estimate.</p><!><p>Efficiency of plasmon-assisted azide-alkyne cycloaddition increases at lower temperature</p><p>The schematic representation of our experimental setup is shown in Fig. 1A. To perform the plasmon-assisted AAC, we graed the alkyne reagent onto the surface of spherical gold nanoparticles (AuNPs). We chose the "gold standard" spherical AuNPs as a plasmonic support because they are resistant to oxidation, show only insignicant catalytic activity and exhibit homogeneous distribution of plasmon evanescent wave around their surfaces. The nanoparticles were graed through the diazonium approach with 4-ethynylphenyl groups (Fig. 1A), which keeps these chemical moieties near the plasmonic evanescent wave. The immobilization of 4-ethynylphenyl units was conrmed by Raman measurements (Fig. 1C and S1 †appearance of characteristic and pronounced SERS vibrational bands; assignment of individual bands is given in Table S1 †), whereas the conservation of the AuNP morphology and size aer the graing procedure was conrmed by TEM (Fig. 1D). The UV-Vis spectra of the graed AuNPs in acetonitrile exhibit a pronounced plasmon absorption band near the 650 nm wavelength. To maximize the absorption efficiency, we utilized LEDs with a 660 nm emission wavelength for plasmon assistance (Fig. 1E, absorption spectra in Fig. S2 † show that the 660 nm light is not absorbed by other compounds added to the reaction mixture).</p><p>The modied AuNPs were dispersed in acetonitrile solution with a large excess of 4-azidobenzoic acid and subjected to plasmon triggering under different temperatures between room temperature and the solvent melting point. Aer the reaction, the AuNPs were removed from the reaction mixture by centrifugation, and washed and dried to remove any absorbed azide. The AuNP sediment was then subjected to SERS measurements (Table S1 †) in the dry phasesuch a route allows us to exclude the side effects of solvent or Raman laser illumination of the reaction mixture. The reaction progress was tracked by SERS measurements through the decrease of the C^C characteristic SERS band (at 2000 cm À1the maximum peak intensity in the 1990-2050 cm À1 spectral area was considered) and appearance of the triazole ring characteristic band (at 2235 cm À1 ). In addition, the Raman band from the benzene ring (which is not involved in the reaction) was used as the intrinsic SERS marker, i.e. the reaction progress was monitored (semiquantitative) by employing the ratios of the peaks at 2000/1590 cm À1 and 2235/ 1590 cm À1 . For the cross-check, an Au-C related SERS band was used as a control (I 2235 /I 401 instead of I 2235 /I 1590results are given in Fig. S3 †). It should also be noted that two triazole isomers (1,4 and 1,5) could form during the AAC; however, Raman spectroscopy does not allow us to distinguish between them.</p><p>The reaction kinetics of the AAC at room temperature (RT) is presented in Fig. S4 † and related discussion in the ESI. † At RT, the reaction takes place during the rst 12 hours and is nished aer 20 hours, where the consumption of 4-ethynylphenyl groups signicantly decelerates. To simplify further analysis, we monitored the reaction conversion at 3 hours, which should correspond to the initial stages of the reaction, where the conversion should depend on the reaction rate rather than the availability of surface attached reaction centres.</p><p>In the next step, we examined the temperature effect on the AAC plasmon-assisted reaction. Experiments were performed in the +25 to À35 C temperature range (melting point of acetonitrile is À45 C, whereas viscosity at À35 C is 0.9 mPa s and at +25 C 0.4 mPa s). The SERS spectra recorded aer the reaction performed at different temperatures are shown in Fig. 2A. The appearance of characteristic triazole ring SERS peaks and the simultaneous decrease of acetylene peak intensity are well evident at all temperatures. In turn, Fig. 2B and S3 † show the evaluation of the characteristic SERS band ratio, corresponding to the consumption of acetylene groups and formation of triazole rings. The amounts of reacted acetylene groups as well as formed triazole rings slightly increase during the gradual reduction of the reaction temperature up to À10 C. A further decrease in the reaction mixture temperature leads to a rapid increase of observed conversion, which becomes especially pronounced in the À20 C to À30 C temperature range. At temperatures near À30 C the increase of the reaction rate reaches a plateau. Thus, we observed almost one order of magnitude increase of the reaction rate upon cooling the reaction from RT up to À35 C. Control measurements indicate the absence of 4-ethynylphenyl group degradation as well as no formation of side-products (Fig. S5 † and related discussion). In addition, the illumination of the reaction mixture, performed in the absence of plasmon-active nanoparticles, does not indicate any chemical conversion (see Fig. S6 and S7 † and related discussion). Therefore, our results clearly show the increase of the AAC rate under the plasmon excitation and the fact that the decrease in temperature leads to a larger conversion of the alkyne reactant.</p><!><p>We postulate that the plasmon assistance could signicantly decrease the residual activation energy of AAC according the mechanism presented in Fig. 1Bi.e. via electronic excitation of reactant(s), their vibrational relaxation in an electronic excited state, and product formation through the residual activation barrier, overcoming of which oen requires heating. 44,45 However, heating can also increase the electron-phonon scattering and shorten the lifetime of plasmon-excited molecular states. Therefore, the heating of the reaction mixture can negatively affect the reaction rate, if quantum effects prevail. So, the main question is which stage limits the reaction progressprobability of plasmon-assisted excitation of molecules, lifetime of excited molecules or overcoming the residual potential barrier?</p><p>To nd the dominant rate-limiting effects in our case, we assessed the potential impact of viscosity increase, plasmon heating (which can be locally enhanced due to nanoparticle agglomeration with temperature decrease), damping of electron scattering, and prolongation of hot-electron lifetime at low temperatures (Fig. S8-S10, † and related discussion), as well as performed DFT calculations of potential reaction pathways. First, experimental and theoretical estimations of nanoparticle agglomeration and related plasmonic local heating indicate the absence of apparent nanoparticle agglomeration and a moderate (approximately 1 increase of temperature in the close vicinity of metal nanoparticles under our experimental conditions (Fig. S9 and S10 †). Indeed, the collective heating of the suspension of AuNPs, as well as a large temperature gradient near AuNP surface, 46,47 does not allow us to completely exclude contributions from plasmon heating. At the same time, the inverse temperature dependence observed in our case is not consistent with plasmon heating as the main driving force of this plasmon-assisted chemical transformation.</p><p>Secondly, control experiments with the addition of PEG, performed at RT, do not result in comparable reaction acceleration, which could prove negligible viscosity inuence on AAC progress (Fig. S11 and S12, † and related discussion). Thirdly, the prolongation of the plasmon-excited electron lifetime was estimated from the temperature dependency of the Drude scattering parameter, taking into account the electron-phonon scattering in the bulk of AuNPs. 48 Simple estimation of excited electron relaxation, performed on the basis of Holstein approximation, indicates the increase of their lifetime with temperature decrease by $25%. Thus, the prolongation of the plasmonic excitation lifetime can explain only a fraction of the observed reaction speedup.</p><p>In DFT calculations, aimed at exploring potential AAC pathways, we focused on comparing the reactivity of differently activated phenylacetylene (Fig. 3A, reaction coordinates are given in the ESI †). Since the acetylene moieties were spatially separated from AuNPs (Fig. 1), we did not consider their interactions with the gold surface. Two common modes of plasmonic activation were considered: injection of hot electrons from metal nanoparticles or intramolecular HOMO-LUMO electron excitation. As a baseline, we show the thermal pathway of the AAC reaction (designated as PW0) in Fig. 3A. This reaction type represents Huisgen pericyclic [3 + 2] dipolar cycloaddition, where the product is formed via a single cyclic transition state TS 0 through an energy barrier of 124 kJ mol À1 in a concerted manner.</p><p>Next, we considered the plasmon-assisted electron excitation of the alkyne to a triplet T 1 state (PW1), which corresponds to energy-transfer catalysis. The energy needed to form the T 1 state is larger than the energy of a single 660 nm photon (181 kJ mol À1 ). Indeed, the dependence of the reaction rate on the laser power was found to be nonlinear (Fig. S13 †), which indicates the contribution from multiphoton absorption processes. This contribution becomes even more pronounced at À35 C. Product formation from the T 1 (Fig. 3A) state would start by C-N bond formation via TS T1 , accompanied by the spin density transfer to the azide unit. Int T formed in this process can undergo ring closure via TS T2 to give a triplet state of the product, which nally undergoes intersystem crossing (ISC) to P. The ISC might also take place in earlier steps in the reaction path, for example aer the formation of TS T1 , because they are all higher in energy than the TS in the ground state and would thus have sufficient energy to overcome the barrier leading to P. If the triplet excitation is transferred to the azide molecule via the Dexter mechanism without the concomitant C-N bond formation, it would lead to instant elimination of the nitrogen molecule and formation of triplet nitrene (TS T1nitrogen ), which would react with the acetonitrile solvent and would not stay bound to the nanoparticles. If the PW1 reaction coordinate is indeed operational, it could contribute to the observed reaction rate temperature dependence, since decreasing temperature, besides the prolongation of the plasmon lifetime, slows down the relaxation of the triplet state T 1 49 and therefore increases the probability that it encounters a suitably positioned azide reactant.</p><p>In the case of plasmon assistance through hot electron injection (photoredox catalysis), we considered two possibilities. The rst one (PW2 in Fig. 3A) involves the reaction of the thus formed phenylacetylene radical-anion with azide in a stepwise manner. We found different transition states for the formation of the rst bond differing in the orientation of reactants in the TSR1À and TSR3À transition states. Subsequently, we were able to nd the remaining intermediate Int R and transition state TSR2À leading to the observed product from TSR1À. However, the almost isoenergetic TSR3À leads to the expulsion of the nitrogen molecule, which indicated that the reaction pathway PW2 should lead to the formation of sizeable amounts of by-products.</p><p>The second possible reaction pathway (PW3 in Fig. 3A) considers the activation of the phenylacetylene molecule by rst forming the phenylacetylene radical-anion and subsequent loss of the injected electron and reaction of vibrationally excited phenylacetylene S 0 heat with azide. In this case, a vibrational energy of 59 kJ mol À1 could lower the reaction barrier of the concerted process, but since it is much less than 124 kJ mol À1 required to reach TS, and this path cannot be responsible for the observed temperature dependence. We also briey considered the activation of the phenylacetylene substrate with hot holes, 19 analogous to the recently proposed photoredox AAC mechanism 50 (PW4 in Fig. 3A), and this pathway also did not lead to a substantial decrease of the reaction barrier and is thus likely not operative.</p><p>To get further insight, and see whether PW1 or PW2 are in operation, we performed control experiments with the addition of TEMPO (2,2,6,6-tetramethylpiperidin-1-yl)oxyl, a scavenger of radicals that should block both PW2 and PW3, and addition of cyclooctatetraene (COT), a scavenger of the triplet excited state, 51 that should block PW1 (Fig. 3B and C). As can be seen in Fig. 3C, the addition of both compounds decreases the conversion at lower temperature. The effect of COT, however, was found to be signicantly higher (Fig. 3 and S14 †). Even the addition of a negligible amount of COT (10 À6 M) results in almost complete suppression of triazole formation at RT and at À35 C. Additional control experiments involving cyclohexene and 4-cyanobenzoic acid (see Fig. S15 and S16, † and related discussion) indicate that the impact of COT should be attributed mostly to excited state quenching. However, we also observed moderate inhibition of the reaction with 4-cyanobenzoic acid, which may be attributed to the partial blockage of surface reaction sites. Overall, the results of plasmon-assisted experiments performed with scavengers of radicals or excited states do not exclude the hot-electron injection related reaction pathway but indicate that PW1 should be viewed as the dominant mechanism. In addition, the different effects of the individual additives also run counter to the dominant effect of viscosity or the local heating effect.</p><p>We also checked the ability of the "reverse" AAC initiation, through the graing of azide chemical moieties onto the AuNP surface in reaction with phenylacetylene dissolved in acetonitrile under plasmon excitation at À35 C (Fig. S16 †). In this case, we did not detect the appearance of a reaction product. The singlet-triplet gap in the azide is much smaller than in the alkyne (ca. 150 kJ mol À1 ) and thus triplet azide does not have enough energy to overcome the energy barrier of TS T1 .</p><p>Finally, we compared our results with the catalytic efficiency of common Cu catalysts (Fig. 2B, C, and S7B †). It can be seen that plasmon assistance has similar efficiency to the Cu catalyst at RT and signicantly greater efficiency at À35 C. Moreover, control experiments with COT addition and Cu-based catalyst do not indicate any changes in the reaction rate. Therefore, the measured effect of reaction quenching by COT points to the involvement of the plasmon-excited molecular triplet state (or triplet state transfer).</p><p>Taken together, we nd that the PW1 mechanism is the most plausible, because it couples the release of a large amount of thermal energy (up to 280 kJ mol À1 ) from the triplet state of the alkyne with the physical proximity of the other reacting partner (azide). As a result, there is effectively no residual barrier, which is fully consistent with the observed reverse temperature behaviour of the reaction rate.</p><!><p>High reaction temperatures or utilization of toxic catalysis leads to lower reaction selectivity, restricts the reaction scope, and thus restricts the applicability for bioorganic transformations. [52][53][54] To demonstrate the advantages of plasmonassisted AAC at low temperature, we performed the coupling of biotin azide to AuNPs-C^CH. Biotin is widely used for the detection of cellular alkyne cholesterol, 55 preparation of a novel multifunctional benzophenone linker for pull-down assays, 56 and photoaffinity labelling 57 and double-click stapling techniques. 58 Classically, the AAC of biotin is performed with copper-based catalysts, which can sometimes be undesirable due to the presence of traces of copper in reaction products.</p><p>The schematic representation of biotin coupling with 4ethynylphenyl moieties on AuNPs is presented in Fig. 4 (top) and the results of plasmon-assisted AAC performed at À35 C without the addition of catalysts are shown in Fig. 4 (bottom). The SERS spectrum of biotin-azide is also presented for comparison (peak assignment is given in Table S2 †). The comparison of the Raman spectra of biotin-azide powder and AuNPs coupled with biotin by plasmon-induced AAC shows that the reaction proceeds with a high yield (estimated as the decrease of the C^C vibration band and well visible appearance of a biotin-related SERS band) at sub-zero temperature under copper-free conditions. This shows that our protocol is applicable to the AAC reactions of more complex substrates without the addition of metal catalysts or heating.</p><!><p>Plasmon assistance promotes a range of chemical transformations through lowering of the activation energy and the synergy between the heating of the reaction mixture and plasmon assistance is well described in the eld of plasmonic chemistry. Here, we theoretically and experimentally demonstrated the opposite situation, where the decrease of reaction temperature led to a signicant increase of plasmon-assisted catalytic efficiency. To model and explain the details of the complex process, we utilized an AAC reaction on the surface of spherical gold nanoparticles. The maximum reaction rate is observed at temperatures signicantly below zero and it signicantly exceeds the efficiency of common Cu-based catalysis. Density functional theory calculations identify several barrierless AAC pathways under plasmon assistance. Control experiments with cyclooctatetraene (triplet quencher) point to the involvement of the plasmon excited triplet state in the mechanism of the AAC reaction. The observed reverse dependence of the reaction rate on temperature can be explained by the decrease of electron-phonon scattering and retardation of organic molecule relaxation aer the plasmon-induced excitation. Finally, we carried out a plasmon-assisted AAC reaction at À35 C employing a bio-relevant compoundbiotin. The presented results demonstrate the potential of plasmonic chemistry in the synthesis of complex biomolecules and hybrid compounds.</p>

Royal Society of Chemistry (RSC)

A GC/MS method for the quantitation of N-nitrosoproline and N-acetyl-S-allylcysteine in human urine

Biomarkers in urine can provide useful information about the bioactivation of chemical carcinogens and can be used to investigate the chemoprotective properties of dietary nutrients. N-nitrosoproline (NPRO) excretion has been used as an index for endogenous nitrosation. In vitro and animal studies have reported that compounds in garlic may suppress nitrosation and inhibit carcinogenesis. We present a new method for extraction and sensitive detection of both NPRO and N-acetyl-S-allylcysteine from urine. The latter is a major metabolite of S-allyl cysteine which is abundant in garlic. Urine was acidified and the organic acids extracted by reversed phase extraction (RP-SPE) and use of a polymeric weak anion exchange (WAX-SPE) resin. NPRO was quantified by isotope dilution gas chromatography-mass spectrometry using 13C5NPRO and N-nitrosopipecolic acid (NPIC) as internal standards. This method was used to analyze urine samples from a study that was designed to test whether garlic supplementation inhibits NPRO synthesis. Using this method, 2.4 to 46 ng of NPRO per mL urine was detected. The method is straightforward, reliable and can be performed with readily available GC/MS instruments. N-acetyl-S-allylcysteine was quantified in the same fraction and detectable at levels of 4.1 to 176.4 ng per mL of urine. The results suggest that 3 to 5 grams of garlic supplements inhibited NPRO synthesis to an extent similar to a 0.5 g dose of ascorbic acid or a commercial supplement of aged garlic extract. Urinary NPRO concentration was inversely associated with the N-acetyl-S-allylcysteine concentration. It is possible that allyl sulfur compounds found in garlic may inhibit nitrosation in humans. .

a_gc/ms_method_for_the_quantitation_of_n-nitrosoproline_and_n-acetyl-s-allylcysteine_in_human_urine

Introduction<!>Human Subjects Protocol<!>Collection and storage of samples<!>Nitrosoproline extractions from urine<!>Gas chromatography/mass spectrometry protocols<!>Quantification of urinary nitrosoproline<!>Synthesis and quantification of urinary N-acetyl-S-allylcysteine<!>Statistical analysis<!>Results<!>Discussion<!>

<p>The nitrosoproline (NPRO) load test was developed to monitor in vivo nitrosation capacity, a process considered to play a role in the formation of carcinogenic nitrosamines. Nitrosamines can form in the stomach when nitrate and amine-rich foods are consumed together (1). Although nitrosamines are highly mutagenic, the role of nitrate exposure in causing certain cancers has been debated (2). Nitrosamine excretion is correlated with nitrate levels in drinking water and salivary nitrate levels (3). Interstingly, it has been suggested that exposure to nitrate from drinking water, but not vegetable sources, is closely correlated with cancer risk(4). This may be due to the concomitant presence of ascorbic acid (vitamin C) in vegetables. For example, it has been shown that low ascorbic acid intake is significantly associated with increased risk of colorectal cancer in subjects highly exposed to nitrate from well water (5). Ingested nitrate is mainly excreted in urine without causing adverse effects, but approximately 20% of the absorbed dose of nitrate is metabolized to nitrite by bacteria that reside in the mouth and stomach, and 25% of the dose undergoes enterosalivary recirculation (6). At low pH, nitrite can be converted to nitrous acid, which in reaction with oxygen forms the nitrosating intermediates dinitrogen trioxide (N2O3) and nitrosylonium ion (NO+) (7). N2O3 and NO+ are potent nitrosating molecules (8). In the stomach, ascorbic acid rapidly reacts with N2O3 to generate nitric oxide (NO)(6,9). Therefore, it is possible that phytonutrients in garlic may have similar antioxidant effects as vitamin C, inhibiting N-nitrosamine production from nitrate exposure (10).</p><p>NPRO formation has been used as a biomarker to study potential dietary interventions for preventing formation of nitrosamines after nitrate ingestion. Recent studies suggest that nitrite has important vasodilatory effects by mediating nitric oxide signaling under hypoxic conditions and that NO can circulate by forming conjugates with heme, thiols (GSNO), and secondary amines (GNNO) (9, 11). Since NPRO formation is a product of the reaction between proline and nitrite, it is possible that NPRO could be developed as a biomarker of the systemic reserve of N-nitrosation capacity (12). Moreover, NPRO is a non-carcinogenic nitrosaminoacid (13, 14).</p><p>Gas chromatography-thermal energy analysis (GC-TEA) has been used to quantify NPRO concentrations in urine (12, 15). This detector pyrolyzes the nitrosamine and releases nitric oxide, which is detected by a chemiluminescent detector. The advantage of TEA is that it is highly sensitive and selective towards nitroso-containing molecules; however, the main disadvantage is that it is not sensitive to other compounds of potential interest, and the availability of these detectors has become quite limited. Garland et al. (15) previously described an isotope dilution method employing gas chromatography, negative chemical ionization mass spectrometry (GC-NCI-MS) for the detection of basal NPRO excretion following derivatization with pentafluorobenzyl bromide (PFBBr). However, this method required a thin layer chromatographic step which was time-consuming and restricted the analysis to only NPRO-PFB conjugates. Therefore, there is a need for a robust analytical method that could be used to analyze NPRO with a similar low limit of detection (0.1ng NPRO/ml urine), but also used for the detection of other compounds that are of interest.</p><p>Herein, we describe a rapid stable isotope dilution method for the quantification of NPRO by GC-NCI-MS. This analytical method was applied to a pilot study to determine if consumption of daily dosages of encapsulated raw garlic or Kyolic, a commercial aged garlic supplement, can decrease NPRO urinary excretion similar to a 0.5 gram dose of vitamin C. These treatment groups were compared to a control group that received a placebo. Finally, N-acetyl-S-allylcysteine, which is the major metabolite of S-allyl cysteine, was quantified as a marker of absorption of a garlic-derived thiol compounds concomitant with the NPRO analysis (16).</p><!><p>The study protocol for human subjects was approved by the Institutional Review Board of The Pennsylvania State University. All subjects gave signed informed consent for participation in this study. The diet was prepared and served by the nutritional sciences research staff at Penn State University. Figure 1 outlines the dietary feeding protocol. Subjects received a 1-week run-in diet (3 meals per day), which was devoid of foods high in nitrate or containing garlic. On the morning of the first treatment day all subjects drank a bolus dose of 300 mg of sodium nitrate dissolved in 500 ml water followed by 500 mg of l-proline dissolved in 500 ml water. A 24-hour urine sample was collected the following morning. The next day subjects were randomized into one of 5 treatment arms or the control group. Subjects in the treatment arms received a capsule containing 1, 3, or 5 grams of fresh garlic, 3 grams of aged-garlic extract (Kyolic), or 500mg ascorbic acid. The control group received a placebo capsule. Subjects continued to receive nitrate and proline as well as the treatment for seven days. Urine was collected every other day for seven days.</p><!><p>Urine specimens were collected into 250 ml Nalgene containers, and 25g of ammonium sulfamate was added to each container. The urine was stored at -20°C for several years prior to analysis. Under these conditions, NPRO has been found to be quite stable (17). The urine samples were thawed and aliquotted into 50 mL volumes for storage and the total volume of the urine samples was measured. Additional, multiple aliquots of 5mL were prepared for extraction and analysis.</p><!><p>Authentic standards of nitrosoproline, 13C5-N-nitrosoproline, and N-nitrosopipecolinic acid (NPIC) were obtained from Dr. Shantu Amin at the Penn State School of Medicine Department of Pharmacology (Hershey, PA) and the NCI Chemical Carcinogen Reference Standards Repository program. Purity of the compounds was assessed using LC-MS and found to be >99.0%. Purified standards were maintained at - 20°C prior to use. Stock solutions of 13C5NPRO and NPIC were prepared in anhydrous ethanol and maintained at 4°C during the analysis period. Each 5 ml aliquot of urine was spiked with 250ng of 13C5NPRO (50ng/ml), and vigorously mixed with 3.8 ml of 0.9M HCl in a 10 ml glass test tube. Sample clean-up was performed on a Sep-Pak Vac tC18 reversed-phase SPE cartridge (500mg/6ml) (Waters Corp., Westfield, MA), using an 18-port vacuum manifold (Alltech Associates, Deerfield, MI). The column was conditioned with 3 ml methanol followed by 3 ml 50mM formic acid in water (pH 3). The acidified urine samples were applied to the sorbent bed and the breakthrough volume was discarded. The acidic compounds were eluted using 6 ml of 100% HPLC grade water (Mallinckrodt, Hanover, Germany). A 3 ml volume of 0.1M sodium phosphate buffer (pH 6.5) was added to the collected aqueous fraction and mixed vigorously. A polymeric weak anion exchange (Strata X-AW) SPE cartridge (60mg/3ml) was used to further clean-up the sample and exchange solvents prior to derivatization (Phenomenex, Torrance, CA). This column was conditioned with 3 ml of methanol containing 2% formic acid followed by 3 ml of deionized HPLC grade water. The buffered sample was added to the SPE cartridge, and potential interfering compounds were subsequently washed away with 3 ml of water followed by 3 ml methanol. The column was dried under a vacuum of 10 inches Hg for 2 minutes. The acidic compounds were then eluted with 6 ml of methanol containing 2.5% ammonium hydroxide. The eluent was dried to approximately 0.5 ml under nitrogen at 40°C. The sample was transferred to a 1.8 ml brown borosilicate glass vial, where it was dried to completion. Derivatization was carried out on the dried sample by adding 200 ul of a 0.25% N,N-diisopropylethylamine (DIPEA) and 200 ul of 0.25% pentafluorobenzyl bromide (PFBBr) followed by heating at 65°C in aluminum blocks for 3 hours (Pierce, Rockford, IL). The derivatized samples were dried under nitrogen at 65°C. The derivatives were reconstituted in a final volume of 200 ul isooctane for analysis by GC-MS.</p><!><p>An injector with autosampler was used to introduce 1 ul of derivatized sample into an Agilent 6890 gas chromatograph (GC) that was interfaced with a 5970N mass selective detector (Agilent Technologies, Wilmington, DE). Injections were made in the cool-on-column injection mode with the inlet programmed to track the oven temperature. There was a 3.5 minute solvent delay before turning on the mass selective detector (MSD). The flow rate of helium through the column was constantly maintained at 1 ml/min with an average linear velocity of 38 cm/s. The starting inlet pressure was 10.04 psi. Separations were performed on a 30 meter DB-5 column, with a 0.25 mm i.d. and 0.25 um film thickness (J&W Scientific, Rancho Cordova, CA). The GC temperature protocol was 100°C for 1 minute and the temperature was increased to 280°C at a linear rate of 10°C/min. The final oven temperature was held for 5.5 minutes. The method cycle time was 24.5 minutes with a 5 minute equilibration period following each run.</p><p>Molecules eluting from the column were ionized and detected in the negative chemical ionization (NCI) mode using methane as the reagent gas. The vacuum manifold pressure was 2.0x10-4 Torr with the reagent gas mass flow control set at 40 (flow rate=2 ml/min). The MSD transfer line temperature was held constant at 280°C, the source temperature was 150°C, and the quad temperature was 106°C. For quantitative determination of each urine sample, both full scan and selective ion monitoring (SIM) were performed. The full scan traces were collected for mass-to-charge ratios (m/z) between 50-300 amu. In a separate run, ion traces were set to selectively monitor ions at m/z=143 amu (NPRO, MW=144g/mole), m/z=148 amu (13C5NPRO, MW=149 g/mole), and m/z=157 amu (NPIC, MW=158 g/mole). The dwell time for each ion was 100 milliseconds.</p><!><p>The urinary NPRO concentration was estimated using relative response factors. Standard curves were run to determine the relative response ratio of unlabeled NPRO to 13C5 NPRO. Varying concentrations of NPRO (0.05pg/ul to 250pg/ul) were spiked with 6.25pg/ul 13C5NPRO, the expected final concentration of 13C5 NPRO following the extraction and dilution steps outlined above. The response ratios were plotted versus amount ratios; a second set of standard curves was generated daily by adding derivatized NPIC (50 pg/ul) to varying concentrations of 13C5NPRO (0.05pg/ul to 250 pg/ul). Response factors were calculated from the linear regression slope of the response ratio (y-axis) versus the amount ratio (x-axis); these were used to verify the analytical precision as well as calculate the concentration of NPRO in the urine specimens.</p><!><p>N-acetyl-S-allylcysteine was synthesized according to the method described by Jandke et al.(18). Briefly, 159.5 mg (1 mmol) of N-acetyl cysteine was weighed and stirred for 10 min. at room temperature with 115.8 mg (1 mmol) allyl bromide in 10 ml water (Sigma-Aldrich, St. Louis, USA). Sodium hydroxide (2 M) was added dropwise to reach a final pH of 12.0. Approximately 60 ml anhydrous ethanol was added to the solution which was continuously stirred for 4 hours at room temperature. The solution was dried under a stream of N2 gas at 40°C. Twenty milliliters of water was added to the dried residue and 2 M HCl was added dropwise to reach pH 2.0. This solution was extracted 2 times with 40 ml volumes of ethyl acetate. The organic layer was collected and dried under N2 gas at room temperature. The residue was redissolved in 55 ml ethyl acetate and esterified with 0.010 ml of 6.8 M PFBBr along with an equimolar volume of DIPEA (0.124 ml). This reaction was carried out at room temperature overnight. The esterified residue was dried under N2 gas at room temperature, and brought up in a final volume of 22.5 ml of ethyl acetate. A standard curve was generated for external calibration. The standard curve was linear over a concentration range of 0.1 ng/ul to 56 ng/ul.</p><!><p>Differences across groups were assessed using the non-parametric Kruskal-Wallis test. Spearman's rank test was used to test correlations. Linear regression was used to examine the association between the excretion of N-acetyl-S-allylcysteine and NPRO.</p><!><p>Figure 2 shows a total ion current (TIC) chromatogram collected for a five minute period of the run that contains the NPRO peak. The average retention times for NPRO, 13C5NPRO, and NPIC were 12.466 min. (+/-SD=0.007), 12.450 min. (+/-SD=0.003), and 12.888 min. (+/-SD=0.004), respectively. For quantitative determinations, the maximum difference in the retention times of the molecular ions was restrained to less than +/-0.02 minutes, which is three times the standard of deviation from the mean (n=3). Since the NPRO and 13C5NPRO peaks were nearly indistinguishable chromatographically, concentrations of endogenous NPRO in the sample were measured by estimation of ratios of molecular ion abundance for the labeled standard (m/z 148) that was added to the abundance of the molecular ion (m/z 143) that was present at the same retention time. An example of a typical standard curve for 0.1 to 80 pg/ml concentrations is shown in Figure 3.</p><p>The Residual Standard Deviation (RSD) for repeated injections was 10% for 2.3 pg 13C5NPRO (n=8) and was 3% for injections of 20 pg 13C5NPRO (n=4). The precision and recovery of the method was analyzed by performing the extraction (n=3) of 5 ml urine samples that contained 25 ng, 50 ng, and 75 ng of 13C5NPRO (Fig. 4). The recovery of standard was 77%, 84% and 88%, respectively. The availability of two internal standards made it possible to compare NPIC (m/z 157) with 13C5NPRO for validation of the internal method. As seen in figure 5, the quantification of NPRO using either internal standards was in excellent agreement.</p><p>The method was then applied to the analysis of samples from the human feeding study. In that study each subject's urinary NPRO excretion was evaluated at three different times after initiation of treatment. There was no significant difference between any of the treatment group concentrations over time; therefore, all three measurements for each person were averaged and treated as a single measurement. Comparisons were made between these averages for each of the treatment groups and the control groups (Fig. 6). Although we observed lower mean NPRO concentrations in the 3 and 5 gram garlic groups, Kyolic group and vitamin C group (when compared to the control group) the differences in NPRO concentrations between any of the groups was not statistically significant. High within-group variation in concentrations of NPRO and the low subject numbers contributed to the high standard error of these measurements.</p><p>We next determined the concentration of N-acetyl S-allylcysteine in urine as a biomarker of effective garlic intake. The quantity of N-acetyl-S-allylcysteine in urine was estimated by mass spectrometry. As can be seen from figure 7, the abundance of the extracted ion current at an m/z 202 from the 90pg/ul standard (upper panel) was nearly identical to that of the quantity detected in one of the 5 gram fresh garlic supplement urine specimens. Interestingly, the double peak that is seen in the urine sample may represent a racemization-product of the N-acetyl transferase enzymes (18).</p><p>Linear regression was used to analyze associations between the concentrations of N-acetyl-S-allylcysteine and NPRO excreted in urine (Fig. 8). The data points are identified by group. From the regression model, a 100 pg/ml decrease of NPRO concentration was found to be significantly associated (p<0.02) with each 1ng/ml unit increase in urinary N-acetyl-S-allylcysteine. The data were skewed so a non-parametric Spearman rank correlation test was also conducted. Using this test, the correlation coefficient was -0.80 and the p-value < 0.003. In five subjects, the N-acetyl-S-allylcysteine was below the limit of quantification; those data points were not included in this analysis.</p><!><p>The purpose of the current study was to develop a protocol using mass-spectrometry that is as sensitive as GC-TEA but also allows other urinary metabolites to be analyzed concurrently. Solid state extraction is a practical clean-up method for urinary NPRO analysis, and the negative chemical GC-MS method adapted from the Garland et al. yields excellent quantitative precision and sensitivity(19). Negative chemical ionization offers a highly sensitive mode for MS analysis because interference from background ions is reduced and the molecular ion can be determined since molecule fragmentation is minimal. When esterification reagents such as PFBBr are used, derivatized molecules form highly electronegative molecular ions. However, in order to obtain sufficient sensitivity, urine samples (especially those which have been frozen and thawed) must be cleaned-up adequately and purified. The Garland study applied a laborious thin-layer chromatography step for this purpose; therefore, a solid phase extraction protocol was developed to isolate the total organic acid fraction in our urine samples. An initial reversed phase extraction was successful in removing both non-polar molecules and other inorganic ions and collecting the organic acids. A second anion exchange extraction was used to elute the organic acids contained in the aqueous fraction. These molecules are selectively displaced from the polymeric WAX resin by the eluting solvent. This clean-up procedure allowed the sufficient lowering of background such that GCMS electron capture detection sensitivity became comparable to that which has been previously reported for thermal energy analysis.</p><p>The results of the feeding study suggest that accurate determinations of NPRO and N-acetyl-s-allylcysteine can be made simultaneously. The mean concentrations of these biomarkers have an inverse correlation with each other. This suggests that the ingestion of s-allylcysteine (SAC) may mitigate nitrosation reactions in the stomach and may be an important bioactive component of garlic (20). Certain antioxidants, such as ascorbic acid are known to mediate the toxicokinetics of nitrate metabolism (21). For example, 300 mg ascorbic acid has been shown to significantly inhibit gastric formation of NPRO in numerous studies (22). However, the role of nitrate/nitrite interconversion is complex and can generate nitric oxide as an end-product which results in increased gastric perfusion and vasodilation. Similar to ascorbic acid, SAC is thought to inhibit conversion of nitrous acid to dinitrogen trioxide or nitrosylonium ion. These compounds may also interfere with genotoxicity by inhibiting the bioactivation of pro-carcinogens, or by reducing the DNA damage caused by reactive intermediates. SAC has been shown to directly inhibit nitrosation of N-morpholine and to prevent metabolic activation of N-nitrosomorpholine through inhibition of the enzyme P4502E1 (23). Moreover, these compounds have been shown to induce phase II detoxification enzymes such as quinone reductase and glutathione-S-transferases, further reinforcing their potent antioxidant activity. S-allylcysteine and N-acetyl-S-allylcysteine have both been shown to mitigate oxidative and nitrosative stress (24, 25).</p><p>The feeding study suggests that intake of 5 g garlic reduces the quantity of NPRO excreted. N-acetyl-S-allylcysteine was significantly inversely correlated with NPRO. Intra-group variance was high, making statistical judgments across days unreliable in this small group of subjects. To address these issues, data from all three measurements for each subject were averaged for group-wise comparisons. The largest decrease in NPRO formation, compared to controls, was in the 5 gram fresh garlic group. Interestingly, these subjects also excreted the highest concentrations of N-acetyl-S-allylcysteine (Figure 7). Although the bioavailability of S-allycysteine is 98% in rats, the conversion rate to N-acetyl-S-allylcysteine is highly variable in humans, and may have considerable dependence on N-acetyltransferase isoforms or exposure to inducers that may alter the disposition of N-acetyl-S-allylcysteine (26). The half-life can range from not detectable to 5 hrs with considerable intra-individual variation. However, subjects in the Vitamin C group also appear to have detectable levels of N-acetyl-S-allylcysteine, which raises some concern as to whether subjects adhered to the dietary protocol throughout the course of the study. There are other exposures, such as allyl chloride, that can generate N-acetyl-S-allylcysteine, but exposure to allyl halides is expected to be rare except in certain occupations (27, 28). Moreover, researchers have documented that consumption of onions may also generate N-acetyl-S-allylcysteine (18).</p><p>We describe a sensitive and accurate method for combined determination of NPRO and N-acetyl-S-allylcysteine. The method makes possible determinations of biomarkers of garlic intake and the potential for using these biomarkers in future investigations to examine the biological and toxicological role of nitrosation. Our controlled feeding study did suggest that garlic may play some role in inhibiting NPRO formation and that the excretion of N-acetyl-S-allylcysteine, one of the major excreted metabolites of S-ally-cysteine (SAC) in garlic, is inversely correlated with NPRO excretion.</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>Feeding study design - nitrate dosing began on day 7 following a run-in diet and continued for seven days. Subjects in the treatment arms received supplementation on day 9 and continued for five days. Urine was collected on day 10, day 12, and day 14.</p><p>Total ion chromatogram (TIC) for a 5ml urine sample prepared as described in the methods. Panel (A) shows a 5 minute window of the ion current intensity with 20pg/ul NPRO spiked into the sample. The peak for NPRO is identified with an arrow. Panel (B) shows the mass spectra of a 2pg/ul quantity of NPRO standard. Quantitation was carried out on the molecular ion (m/z 143) at a retention time of 12.466 minutes.</p><p>Representative linear standard curve of the NPRO-PFB response for a concentration range of 0.1-80 pg/ul. The linear regression equation for this standard curve was: y=-54.8 + 403.6(x). The R2 for the fit was 0.9996.</p><p>The graph shows the concentration of 13C5 NPRO that was added at 5, 10 and 15 ng/ml of urine (x-axis) versus the 13C5 NPRO concentration (ng/ml) found following extraction and derivatization (y-axis). The percent return from these levels was 77, 84 and 89%, respectively. Data points and error bars are the mean (± SD) of 3 replicate measurements. The regression coefficient for linear fit was 0.944 and the goodness-of-fit for the regression model was R2= 0.9997.</p><p>The graph shows the correlations between the NPRO concentrations quantified by the internal standard method versus the isotope dilution method for each sample. The regression coefficient was 0.9 and the adjusted R2 equaled 1.</p><p>The graph shows the mean (±SEM) of the NPRO output in each of the groups. The concentrations were averaged for each subject. The groups, identified on the x-axis, are as follows: control (no supplement); 1 gram garlic supplement; 3 gram garlic supplement; 5 gram garlic supplement; 3 gram Kyolic supplement; 0.5 gram ascorbic acid supplement. The NPRO concentration (ng/ml) is on the y-axis.</p><p>The abundance of the extracted ion current for N-acetyl-S-allylcysteine (y-axis), and the mass spectra (inset) for the molecular ion m/z=202 is shown. Panel (A) shows the EIC of approximately 90ng/ul of synthesized standard. Panel (B) shows a similar concentration quantified in 5 ml of extracted urine which was collected from a subject that consumed a 5 gram quantity of a fresh garlic supplement for one day.</p><p>The graph plots the average total NPRO concentrations versus the average total concentration of N-acetyl-S-allylcysteine concentrations that were excreted in the urine samples of each subject. The regression line is for all the data points.</p>

PubMed Author Manuscript

Carbonate-promoted C–H carboxylation of electron-rich heteroarenes†

C–H carboxylation is an attractive transformation for both streamlining synthesis and valorizing CO2. The high bond strength and very low acidity of most C–H bonds, as well as the low reactivity of CO2, present fundamental challenges for this chemistry. Conventional methods for carboxylation of electron-rich heteroarenes require very strong organic bases to effect C–H deprotonation. Here we show that alkali carbonates (M2CO3) dispersed in mesoporous TiO2 supports (M2CO3/TiO2) effect CO32−-promoted C–H carboxylation of thiophene- and indole-based heteroarenes in gas–solid reactions at 200–320 °C. M2CO3/TiO2 materials are strong bases in this temperature regime, which enables deprotonation of very weakly acidic bonds in these substrates to generate reactive carbanions. In addition, we show that M2CO3/TiO2 enables C3 carboxylation of indole substrates via an apparent electrophilic aromatic substitution mechanism. No carboxylations take place when M2CO3/TiO2 is replaced with un-supported M2CO3, demonstrating the critical role of carbonate dispersion and disruption of the M2CO3 lattice. After carboxylation, treatment of the support-bound carboxylate products with dimethyl carbonate affords isolable esters and the M2CO3/TiO2 material can be regenerated upon heating under vacuum. Our results provide the basis for a closed cycle for the esterification of heteroarenes with CO2 and dimethyl carbonate.

carbonate-promoted_c–h_carboxylation_of_electron-rich_heteroarenes†

Introduction<!>Acidity calculations<!>Carbonate-promoted C–H carboxylation reactions<!>Kinetic isotope effects and DFT calculations to probe C–H carboxylation mechanism<!>Carboxylate esterification and M2CO3/TiO2 regeneration<!>Conclusion<!>Conflicts of interest

<p>C–H carboxylation (Scheme 1) is a compelling alternative to conventional syntheses of carboxylic acids that utilize oxidative transformations or more functionalized substrates and has attracted attention as a way to expand the use of CO2 in chemical production.1–3 However, carboxylation faces the challenge of overcoming the low reactivity of C–H bonds and CO2, and it lacks the large intrinsic driving force of other C–H functionalizations such as oxidation or amination. The insertion of CO2 into C–H bonds to form a carboxylic acid is actually endergonic on account of the small ΔH and negative ΔS, while C–H carboxylation is exergonic (depending on base strength) because of the driving force from deprotonation. Increased interest in this transformation over the last several years has led to a number of methods that encompass both acid–base2,4–9 (ionic) and radical mechanisms2,10–14 for C–H activation. Despite these recent advances, most methods for C–H carboxylation under conventional, solution-phase conditions require highly reactive, resource-intensive reagents to activate C–H bonds. As such, the development of alternatives that use benign, regenerable reagents is critical to create opportunities for scalable CO2 utilization.</p><p>The carboxylation of aromatic substrates is of particular interest for the synthesis of a wide variety of both fine and commodity chemicals.2,15–19 Because of the high bond dissociation enthalpy (BDE) of aromatic C–H bonds, acid–base (ionic) activation of the substrate has been the most commonly employed strategy. For some substrates, deprotonation of an X–H bond (X = heteroatom) generates a nucleophilic intermediate that undergoes C–H carboxylation via an electrophilic aromatic substitution (EAS) mechanism. The classic example is the Kolbe–Schmidt reaction used for aspirin synthesis, in which phenol is transformed into salicylate by reaction with hydroxide and CO2.20 While the carboxylation of indoles and pyrroles has been achieved similarly,6,8,21 these reactions have required the use of superstoichiometric LiOtBu to deprotonate the N–H bonds.</p><p>Apart from these special cases, carboxylation of (hetero)arene substrates via acid–base chemistry requires direct activation of the C–H bond to generate a reactive carbon-centered nucleophile. Within the last decade, several groups have demonstrated Brønsted-base-promoted carboxylation of (hetero)arenes in organic solvents at near ambient CO2 pressure.4,6–9,21 Hu et al. have shown that relatively acidic heteroarenes (pKa up to 28 in organic solvent) can be carboxylated using Cs2CO3 as the base in refluxing DMF.4,7 The carboxylation of electron-rich heteroarenes beyond this pKa threshold, however, has required much stronger bases. For example, the carboxylation of benzothiophene (pKa of 33 in THF) was not possible under these same conditions.4 Recently, Kondo et al. have demonstrated the carboxylation of a diverse set of (benzo)thiophenes and (benzo)furans by reaction with excess LiOtBu, CsF, and crown ether at 160 °C under a CO2 atmosphere.9 However, these conditions were not able to carboxylate protected indoles, such as 1-methylindole, whose C2 carbon has a pKa near 38. Carboxylation of electron rich heteroarenes functionalized with an amide directing group has been achieved using Ni catalysis with stoichiometric KOtBu and Mn0.22</p><p>Arenes have pKas that generally lie beyond what can be measured (pKa > 40). Researchers have developed methods to carboxylate arenes that are functionalized with a directing group by using a Rh or Pd species to catalyze C–H activation.23–25 In addition to the directing group and catalyst, these methods also require a strong base (KOtBu) or Lewis acid activator (AlMe(OMe)2) to engender reactivity. In the absence of a directing group, solution-phase arene C–H carboxylation requires an extremely strong base such as Schlosser's base,26 or stoichiometric aluminum reagents.25,27</p><p>Apart from acid–base strategies, a very recent report by König et al. has described a photoredox method to carboxylate of (hetero)arenes under mild conditions in which the substrate is activated by one-electron photoreduction and Cs2CO3 serves as the stoichiometric base.28 This method affords moderate to high yields across a variety of substrates, although it is presently incompatible with some classes of (hetero)arene substrates and uses relatively high loadings of a photocatalyst requiring multi-step synthesis.</p><p>We previously showed that simple alkali carbonates (M2CO3) can promote C–H carboxylation of very weakly acidic substrates in solvent-free, alkali salt media at elevated temperature.29–32 This transformation is particularly useful for converting a monocarboxylate substrate into a dicarboxylate product, where the substrate enables the formation of a molten reaction medium.33 More recently, we demonstrated that M2CO3 dispersed into mesoporous TiO2 (M2CO3/TiO2, Fig. 1a) promotes the carboxylation of benzene and other aromatic hydrocarbon C–H bonds in gas–solid reactions (Scheme 1b).34 Dispersion in mesopores disrupts the bulk M2CO3 crystal structure, creating an amorphous material that can attain superbase reactivity, even in the presence of CO2. This carbonate-promoted C–H carboxylation of aromatic hydrocarbons takes place at moderate pressures and temperatures of ∼400 °C. In this study, we begin to assess the generality and selectivity of this strategy using electron-rich heteroarenes, which have somewhat more acidic C–H bonds. We show that gas–solid carbonate-promoted C–H carboxylation occurs at substantially lower temperatures for these substrates and that selective reactions are possible in the presence of multiple C–H bonds (Scheme 1c, Fig. 1b). For thiophene substrates, the selectivity and mechanistic studies support a carboxylation pathway that proceeds via C–H deprotonation by the amorphous CO32−, as seen previously with arenes. For more nucleophilic indole substrates, however, carboxylation proceeds via electrophilic aromatic substitution, which provides a new pathway for CO2 utilization enabled by dispersed carbonate materials.</p><!><p>Thiophene- and indole-based heterocycles were selected as C–H carboxylation substrates to probe the effects of C–H acidity and π-nucleophilicity. The C–H acidities were evaluated by using density functional theory (DFT) to calculate the standard enthalpy change for heterolytic bond dissociation in the gas phase (ΔacidH°, also known as the gas phase acidity) (Fig. 2 and S1†).35,36 ΔacidH° provides a way to compare the thermodynamics of deprotonation irrespective of whether the pKa can be measured. Benzene, which reacts with dispersed carbonates at ∼400 °C, has a ΔacidH° of 401 kcal mol−1; its pKa is too large to be measured but has been estimated to be >43.36 The most acidic C–H bonds in each heterocycle were found to be more acidic (lower ΔacidH°) than benzene by 15–23 kcal mol−1, while the separation between the two most acidic C–H bonds in each substrate was 6–11 kcal mol−1. For comparison, the experimental pKa values of benzothiophene (C2), thiophene (C2), and 1-methylindole (C2) are 32, 33, and 38 according to measurements performed in THF.36,37 Additional DFT calculations to determine solution state pKa values showed good agreement to these experimental values (Fig. S2†).</p><!><p>C–H carboxylation reactions were performed in a sealed vessel containing M2CO3 dispersed on TiO2 (M2CO3/TiO2, M+ = Cs+, K+, Na+), heterocycle substrate, and CO2 (see ESI† for detailed experimental procedures). In most cases, the substrate was placed within a glass culture tube in the reactor to ensure that only volatilized substrate would be able to react with the M2CO3/TiO2 material (Fig. S3†). The products were isolated by aqueous extraction from the TiO2 support and quantified by 1H NMR (Fig. S4–S9†). In all cases, control experiments using M2CO3 without the TiO2 support showed no reactivity, whereas M2CO3/TiO2 promoted C–H carboxylation in varying degrees depending on the identity of M+. Additional control experiments showed minimal reactivity with the mesoporous TiO2 support alone.</p><p>We first assessed the temperature dependence of C–H carboxylation under a common set of conditions using 1.5 mmol substrate, a CO2 loading corresponding to 4–5 bar at the reaction temperature, and a reaction time of 3 h (Fig. 3 and S10†). The relatively low substrate loading corresponded to a maximum pressure of ∼2.5 bar at the highest temperature evaluated (320 °C). Thus, the overall pressure of the reactor at temperature was <8 bar for all of the reactions in this temperature screen. For benzothiophene (Fig. 3a), the onset of carboxylation reactivity was observed at 200 °C. Optimal results were seen at 280 °C, where 190 μmol of benzothiophene carboxylation product was obtained per gram of TiO2 (190 μmol g−1 TiO2) with a 20 : 1 ratio of 2-carboxylate to 3-carboxylate isomers for Cs2CO3/TiO2. Using K2CO3/TiO2, 207 μmol g−1 TiO2 of benzothiophene carboxylate product was obtained with a 25 : 1 product ratio (Fig. 3a). While both the yield and selectivity declined at higher temperatures, the carboxylation selectivity followed the C–H acidities, consistent with a mechanism gated by C–H deprotonation (see below). In contrast to Cs+ and K+, much lower reactivity was observed with Na2CO3/TiO2, suggesting that this material is a weaker base in gas–solid reactions.</p><p>Comparison of benzothiophene carboxylation with our previous results for benzene carboxylation further highlights the effect of C–H acidity on carboxylation. Whereas >200 μmol g−1 TiO2 of carboxylate products were obtained for benzothiophene at 280 °C and <8 bar total pressure, the maximum yields for benzene carboxylation using the same M2CO3/TiO2 materials were ∼100 μmol g−1 TiO2 at 420–440 °C and ∼30 bar total pressure. Thus, reducing the C–H bond acidity (ΔacidH°) by 23 kcal mol−1 enables higher yielding carboxylation reactions under substantially milder conditions (100 °C lower temperature, 1/3 the total pressure). Furthermore, the benzothiophene results also demonstrate that a 7 kcal mol−1 separation in C–H acidity (C2 vs. C3 position) is sufficient for selective C–H carboxylation.</p><p>Because of its high boiling point (221 °C), the vapor pressure of benzothiophene is expected to reach its saturation pressure at T ≤ 240 °C under the conditions used for the data in Fig. 3a (see Table S1† for saturation vapor pressures calculated using the Clausius–Clapeyron equation). As a result, the vapor pressure of benzothiophene varies by ∼5× over the 200–320 °C range examined. To deconvolute temperature dependence from substrate pressure dependence, a series of carboxylation reactions were performed at 280 °C for 3 h using different amounts of benzothiophene corresponding to calculated pressures ranging from 0.5 bar to 3.5 bar, which is approximately the saturation pressure at 280 °C. The total benzothiophene carboxylate yield showed a modest variation from 150 μmol g−1 TiO2 to 210 μmol g−1 TiO2 over this range (Fig. S17†). Thus, the temperature dependence of the benzothiophene carboxylation yield in Fig. 3a is primarily a result of the temperature effect on the rate constant.</p><p>Phenylthiophene reacted in a very similar manner to benzothiophene. The onset of carboxylation was observed at 200 °C with very high selectivity for the 5-phenylthiophene-2-carboxylate isomer (derived from the most acidic C–H bond) observed up to 280 °C. Comparable yields were observed for Cs2CO3/TiO2 and K2CO3/TiO2, while substantially lower yields were seen for Na2CO3/TiO2 (Fig. 3b). The carboxylate yield varied by ∼50% over a 7-fold variation in phenylthiophene pressure (0.5–3.5 bar) at 320 °C (Fig. S17†). The similarity in the temperature- and pressure-dependent yields for both benzothiophene and phenylthiophene is reflected in their nearly identical ΔacidH° values for their two most acidic C–H bonds, suggesting that the same mechanism is operative for both substrates. Notably, although separating the substrate with a culture tube in the reactor ensures that it can only interact with the M2CO3via the gas phase, the carboxylation reactions with low-volatility substrates like benzothiophene and phenylthiophene proceed in comparable or better yield when the two are combined directly (Fig. S11†).</p><p>In contrast to the heterocycle pressure dependence, increasing CO2 pressures were found to significantly inhibit C–H carboxylation for both substrates (Fig. S18†). The CO2 pressure dependence was evaluated for benzothiophene and phenylthiophene at 280 °C and 320 °C, respectively. Interestingly, inspection of the culture tubes for both substrates post-reaction revealed increasing amounts of un-vaporized heterocycle with increasing CO2 partial pressure (Fig. S18†). While the calculated saturation pressures indicate that all of the 1.5 mmol of substrate should be vaporized at these temperatures, this observation suggests that CO2 dissolves in the substrate upon melting and lowers its vapor pressure substantially. An additional contributing factor may be that higher CO2 pressure results in the formation of polycarbonate species (e.g. C2O52−) on the M2CO3/TiO2 material, which are weaker bases than CO32−, thereby reducing the rate of C–H deprotonation.</p><p>C–H carboxylation was also possible with thiophene itself. In the temperature screen performed with 1.5 mmol substrate (Fig. 3c), all of the thiophene is expected to be volatilized over the 200–320 °C range because of its relatively low boiling point (84 °C). The corresponding thiophene pressures range from 2–2.5 bar. In contrast to benzothiophene and phenylthiophene, no thiophene carboxylates were observed at 200 °C, which is consistent with the 6 kcal mol−1 higher ΔacidH° for its C(2)–H bond (Fig. 2). The formation of thiophene-2-carboxylate was observed beginning at 240 °C, with optimal results at 280 °C, where 96 μmol g−1 TiO2 was formed along with 9 μmol g−1 TiO2 thiophene-2,5-dicarboxylate when using Cs2CO3/TiO2. Comparable yields were obtained with K2CO3/TiO2, while Na2CO3/TiO2 was much less effective. The observation of thiophene-2,5-dicarboxylate indicates that the initially formed monocarboxylate product undergoes a second C–H carboxylation on the support. In addition to the thiophene carboxylates, ∼25 μmol g−1 TiO2 of propionate was produced across the temperature range of 240–320 °C (Table S12†). This product arises from an unknown decomposition pathway starting from thiophene or a thiophene carboxylate. The yield of thiophene carboxylates was improved by increasing the thiophene pressure to 5 bar, with a comparable proportion of propionate byproduct (Fig. S17†). In contrast to benzothiophene and phenylthiophene, essentially no CO2 pressure dependence was observed for thiophene at 280 °C. Given the much higher volatility of thiophene, CO2 has a negligible effect on its vapor pressure at this temperature.</p><p>We next investigated the effects of increasing the nucleophilicity of the heterocycle by switching from thiophene to indole substrates.38 To avoid the complication of an acidic N–H bond, we first evaluated 1-methylindole. The most acidic C–H position of this substrate is C(2)–H, whose ΔacidH° (384 kcal mol−1) is very close to the C(2)–H bond of thiophene (Fig. 2). The most nucleophilic position, however, is C3,39 which has a much less acidic C–H bond (ΔacidH° of C(3)–H is 11 kcal mol−1 higher than C(2)–H). Surprisingly, C–H carboxylation occurred readily at 200 °C with a strong preference for the C3 position (Fig. 3d). Moreover, the yield increased substantially as the alkali cation size was decreased, resulting in the highest yields for reactions using Na2CO3/TiO2. Optimal results were obtained using Na2CO3/TiO2 at 200 °C, with a yield of 250 μmol g−1 TiO2 for the C3-carboxylate (Fig. 3d and S8†). At higher temperatures (T > 240 °C) the C2-carboxylate was observed as an additional minor product. The selective formation of the C3 carboxylate is consistent with an EAS mechanism in which C–C bond formation precedes C–H deprotonation. Further support was found in the kinetic isotope effect for C–H carboxylation and DFT calculations (see below). Previously reported methods have achieved selective C3 carboxylation of 1-methylindole with CO2, but have required the use of stoichiometric organoaluminum reagents.25,40,41 Na2CO3/TiO2 provides a benign and much less resource-intensive alternative.</p><p>Selective C3 carboxylation was also observed with indole at 200 °C using M2CO3/TiO2 (Fig. 3e). The M2CO3 dependence followed the same trend as for 1-methylindole, with optimal results obtained using Na2CO3/TiO2. Because the N–H functionality on indole is much more acidic than the C–H bonds (≥35 kcal mol−1 difference in ΔacidH°), it is likely that indole is rapidly deprotonated by M2CO3/TiO2 to form indolide, which can react reversibly with CO2 to form indole-1-carboxylate (N–CO2−). Given the very low acidity of the C–H bond at C3 (ΔacidH° = 397 kcal mol−1, Fig. 2), the selectivity for C3 carboxylation at 200 °C is consistent with an EAS mechanism in which deprotonated indole is the reactive nucleophile.39 Beyond 200 °C, however, the reaction yielded a mixture of C3, C2, and C7 carboxylation products. At 280 °C, C2 carboxylation accounted for 44–64% of the total carboxylation products depending on the choice of M2CO3. Substitution at C2 is commonly seen alongside C3 in solution-phase EAS reactions with indole.42 Both methylindole and indole showed very similar pressure dependences on both heterocycle and CO2 partial pressure (Fig. S17 and S18†).</p><p>Finally, 3-methylindole (skatole) was evaluated to assess the effects of blocking carboxylation at C3. Carboxylation was observed at C2 with a similar temperature dependence as seen for C2 carboxylation of indole (Fig. 3f). Using Na2CO3/TiO2, 94 μmol g−1 TiO2 of the C2-carboxylate was obtained at 200 °C. Increasing the temperature to 240 °C boosted the yield to 138 μmol g−1 TiO2, although minor amounts of additional carboxylates were observed at this temperature, including the product of methyl carboxylation. To our knowledge, C2 carboxylation of 3-methylindole with CO2 has not previously been achieved.</p><!><p>To better understand the differences in thiophene- vs. indole-based heteroarene C–H carboxylation, kinetic isotope effects (KIEs) were measured using intermolecular competition experiments.43 C–H carboxylation reactions were performed for 1-methylindole (200 °C, 1.5 h) and benzothiophene (260 °C, 0.5 h) using various ratios of protiated and deuterated substrate (Fig. 4).34 KIE values of 2.0 and 1.7 were observed for C2 carboxylation of benzothiophene using K2CO3/TiO2 and Cs2CO3/TiO2, respectively. These values are consistent with a mechanism in which C–H deprotonation is slow and the resulting carbanion reacts rapidly with CO2 (Scheme 2) and does not support an EAS mechanism. In addition, previous studies of benzothiophene substitution with strong electrophiles have shown selective substitution at C3, indicating that this is the preferred position for EAS reactivity.39,44,45 The KIE values for benzothiophene are similar to what we have previously observed for benzene C–H carboxylation using the same M2CO3/TiO2 materials,34 as well as solid base-catalyzed reactions that feature rate-determining deprotonation.46</p><p>In contrast to benzothiophene, a KIE value of 1.1 was observed for C3 carboxylation of 1-methylindole, which is within NMR quantification error of 1.0. The disparity in KIE values for these two substrates indicates distinct mechanisms for their C–H carboxylation reactivity. The lack of a KIE for 1-methylindole is consistent with an EAS mechanism at 200 °C in which attack of the π system on CO2 precedes C–H deprotonation (Scheme 2). To our knowledge, an EAS reaction between CO2 and a neutral substrate has not previously been reported. DFT calculations were performed to assess the feasibility of such a pathway with 1-methylindole. Calculations performed using either vacuum or low dielectric solvents (ε < 9) failed to identify a transition state or putative EAS intermediate, suggesting that a gas-phase reaction between 1-methylindole and CO2 is unlikely. With a higher dielectric (ε > 20), however, an EAS transition state was identified that is ∼30 kcal mol−1 higher in energy than the substrates (Fig. S19†). Interestingly, the zwitterionic intermediate resulting from CO2 addition was very close in energy to the transition state, indicating that the reverse reaction is extremely rapid. Together, the KIE and DFT results suggest that the carboxylation of methylindole takes place via an EAS mechanism with substrate that is adsorbed onto the M2CO3/TiO2 material. The amorphous carbonate provides a dielectric to stabilize the transition state for CO2 addition and a proximal base that can immediately deprotonate the putative zwitterionic intermediate. The higher yield for Na2CO3/TiO2 may reflect a stronger adsorption of 1-methylindole because of the higher charge density for Na+. Further studies incorporating atomistic modeling of the amorphous carbonate surface are needed to assess this pathway more thoroughly. Nonetheless, the DFT results indicate that an EAS-like mechanism is possible.</p><!><p>In our previous study of arene C–H carboxylation, we showed that arene carboxylates could be isolated as methyl esters with concomitant regeneration of the M2CO3/TiO2 material by subjecting the carboxylation product to flowing CO2 and methanol at elevated temperatures.34 The same procedure was unsuccessful for isolating heteroarene carboxylate esters because their high boiling points (>300 °C) necessitated temperatures that led to decomposition under the reaction conditions. The use of flowing CO2 and dimethyl carbonate enabled isolation of methyl esters, but the yields were <50% (Fig. S20†). Instead, it was found that we could isolate the ester at near quantitative yields by heating the supported heteroarene carboxylate ((RCOOM)/TiO2) in neat dimethyl carbonate at 160 °C within a stainless-steel batch reactor (Fig. S21†). Subsequent heating of the support material under vacuum resulted in regeneration of M2CO3/TiO2. After establishing optimal carboxylation and methylation conditions, we assessed the ability of M2CO3/TiO2 to catalyze a closed heteroarene esterification cycle over multiple iterations (Fig. 5). When a single sample of Cs2CO3/TiO2 was used for 5 cycles, methyl benzothiophene-2-carboxylate was isolated as the only detectable product by NMR (Fig. S16†) from each cycle with an average yield of 150 μmol g−1 TiO2. In each cycle following the methylation step, an aliquot of the support (∼50 mg) was analyzed by aqueous extraction and 1H NMR to detect unreacted, supported carboxylate. In all cases, no supported carboxylates were observed, indicating complete methylation. Over the five cycles, no indication of catalyst degradation was observed (Fig. 5). These results support previous observations of the ability for dispersed carbonates to catalyze a closed esterification cycle,34 and extend the substrate scope to include heteroarenes.</p><!><p>Conventional solution-phase methods for C–H carboxylation of aromatic substrates with low C–H acidity have relied on the use of highly reactive and resource-intensive organic bases. Our results show that CO32− can serve as a benign, regenerable base for C–H carboxylation via a gas–solid reaction utilizing a dispersed, amorphous carbonate material. Compared to reactions with benzene and other arenes using the same M2CO3/TiO2 materials, the heteroarene carboxylations investigated here reach higher yields (up to 250 μmol g−1 TiO2) under substantially milder conditions (200 °C lower temperature, 1/3 the total pressure). Thiophene-based heterocycles react preferentially at the most acidic C–H bond. The temperature-dependent selectivity and KIE measured for benzothiophene are consistent with a mechanism in which C–H deprotonation is followed by C–C bond formation. In contrast, indole-based heterocycles react preferentially at the most nucleophilic position (C3). DFT calculations and the absence of a significant KIE support an EAS mechanism for the carboxylation of 1-methylindole, which nonetheless requires dispersed carbonate. The combination of CO32−-promoted C–H carboxylation and methylation with dimethyl carbonate provides a two-step cycle to convert aromatic heteroarenes into methyl esters with regeneration of M2CO3/TiO2. Ongoing work seeks to improve the efficiency of this cycle by using alternative supports to increase the loading of reactive carbonate and access reactivity at lower temperatures.</p><!><p>The authors declare no competing financial interests.</p>

PubMed Open Access

Enzyme-Directed Functionalization of Designed, Two-Dimensional Protein Lattices

The design and construction of crystalline protein arrays to selectively assemble ordered nanoscale materials has potential applications in sensing, catalysis and medicine. Whereas numerous designs have been implemented for the bottom-up construction of novel protein assemblies, the generation of artificial functional materials has been relatively unexplored. Enzyme-directed post-translational modifications are responsible for the functional diversity of the proteome and thus, could be harnessed to selectively modify artificial protein assemblies. In this study, we describe the use of phosphopantetheinyl transferases (PPTases), a class of enzymes that covalently modify proteins using coenzyme A (CoA), to site-selectively tailor the surface of designed, two-dimensional (2D) protein crystals. We demonstrate that a short peptide (ybbR) or a molecular tag (CoA) can be covalently tethered to 2D arrays to enable enzymatic functionalization using Sfp PPTase. The site-specific modification of two different protein array platforms is facilitated by PPTases to afford both small-molecule- and protein-functionalized surfaces with no loss in crystalline order. This work highlights the potential for chemoenzymatic modification of large protein surfaces towards the generation of sophisticated protein platforms reminiscent of the complex landscape of cell surfaces.

enzyme-directed_functionalization_of_designed,_two-dimensional_protein_lattices

Introduction<!>Protein expression and purification<!>Preparation of ybbR-N3 peptide<!>Preparation of modified CoA conjugates<!>Chemical conjugation of self-assembled RIDC3 crystals<!>Preparation of ybbRRIDC3 and ybbRRhuA crystals<!>Gel digestion and MS-MS analysis of RIDC3-ybbR crystals<!>Enzymatic labeling<!>Confocal microscopy<!>Negative-stain transmission electron microscopy (ns-TEM)<!>Atomic force microscopy (AFM)<!>Fluorescence microplate reader measurements<!>Enzymatic labeling of peptide-tagged RIDC3 arrays<!>Generation of RIDC3 arrays with CoA tags<!>Design and characterization of 2D protein arrays with genetically incorporated ybbR tags<!>Conclusions

<p>The structural and chemical diversity of proteins positions them as Nature's premier tools in facilitating a myriad of cellular processes. This diversity, in turn, makes them attractive building blocks for the construction of functional synthetic materials.1–8 Proteins exhibit functional complexity in a crowded cellular environment due to their self-assembly into multidimensional supramolecular assemblies, giving rise to emergent properties inaccessible to individual proteins.9 While the noncovalent self-assembly of multicomponent protein architectures with other components, including small molecule cofactors, inorganic complexes (e.g., metal clusters10), or other biomolecular species (e.g., nucleic acids11), is programmed at the genetic level, the structural and chemical diversity of the proteome is further enhanced and indeed, enabled, by enzyme-mediated post-translational covalent modifications. Such modifications play vital roles in the biosynthesis of complex biomolecules (e.g., fatty acids, peptides, polyketides)12,13 or in large-scale cellular functions such as cell-cell signaling and recognition (e.g., via glycosylation, phosphorylation, or membrane anchoring through the attachment of lipid tails).14 For example, neuronal microtubule filaments are found to be more stable in depolymerizing environments due to polyamination of glutamic acid residues by transglutaminases, which may affect aspects of brain development, neuronal regeneration and aging.15 Bacterial and archaeal S-layers, prototypical two-dimensional (2D) self-assembling protein lattices, perform critical roles in protection, virulence, cell morphology and surface recognition.16 For many S-layer proteins, translocation across the cell membrane is facilitated through post-translational modifications such as the N-glycosylation of select asparagine residues or lipid modifications.16,17 Inspired by diverse structure-function relationships in nature, the hierarchical construction of biological machines to perform challenging biological and chemical tasks remains a prominent goal of bionanotechnology.6,18 The development of protein-based biomaterials amenable to post-translational modifications19–21 presents potential applications in diagnostic sensors, vaccines, drug-delivery, or biomineralization matrices.22,23</p><p>Crystalline 2D protein arrays are a promising biotechnological platform due to their high-density display of polypeptides with nanoscale tunability and reconfigurability.19 Engineering the surfaces of 2D protein lattices could provide an opportunity to selectively organize target molecules or particles of interest in a site-dependent manner. This bottom-up route would furnish the ability to not only create periodic patterns with sub-10-nm precision (i.e., in a diffraction-unlimited fashion) but also to hierarchically assemble complex, multicomponent architectures that cannot be obtained through self-assembly alone. In addition, 2D protein crystals represent robust and inherently functional platforms24 that provide high surface area-to-volume ratios and thus could provide distinct advantages in engineering heterogeneous biocatalysts or immunotherapeutic agents, as well as in the fundamental study of biological/enzymatic reactions on 2D surfaces.</p><p>One can envision two major paths to generate functionalized 2D protein materials: repurposing natural protein assemblies like S-layer proteins or designing artificial assemblies from scratch. Naturally occurring extended structures provide a ready-made template for chemical or genetic manipulation; however, structural characterization and facile manipulation of these structures in vitro and in vivo is key to designing function. To this end, Sleytr and coworkers performed formative experiments to demonstrate the potential uses of S-layer lattices, reconstituted in vitro, for applications in membrane filtration25, drug delivery26, and the spatial organization of immunogenic biomolecules or inorganic nanoparticles.22,27 More recent reports describe genetic modification of S-layer substrates with short peptide tags (e.g., the split-protein system SpyTag-SpyCatcher28) to generate hierarchical 3D materials29 and high-density engineered displays on living cells.30 Many of these examples rely on chemical biology tools to facilitate biomolecule immobilization on solid substrates (e.g., biotinylation, azide-alkyne cycloaddition, gene fusions, incorporation of reactive peptides, split-protein tags).31–33 However, given a dearth of atomic-resolution structural information on S-layer proteins, it has remained challenging to manipulate them at will to create functional materials, and in vitro manipulation of S-layer proteins requires protein truncations to afford solubility and stability.</p><p>In contrast with the limited applicability of reconstituted or genetically altered S-layer proteins in vivo, the bottom-up assembly of protein architectures enables the construction of tailor-made nanomaterials with well-known structures and desired properties. There has been considerable progress in recent years to create artificial 2D protein assemblies using chemical design strategies including genetic fusions of symmetric protein modules,34–36 computational design of protein-protein interfaces,37 metal coordination,3,38 or reversible covalent bonding39 towards generating novel biological machines. While a number of these strategies have been employed to generate extended 2D arrays with crystalline order, there has been little exploration thus far in using these arrays as functional platforms.40,41 Interfacing chemical/biological conjugation tools with such ordered assemblies would allow for the facile incorporation of functionality onto crystalline substrates. In this work, we used two crystalline assemblies previously designed in our lab, RIDC3 and C98RhuA, as 2D platforms whose surfaces can be tagged with functional peptides as recognition elements to allow enzyme-mediated modification. Our results show that the incorporation of functional peptides onto the 2D crystal surfaces can be carried out genetically or chemically and does not disrupt the underlying lattice packing, which, in turn, enables site-specific surface modification with synthetic small molecules as well as large proteins.</p><!><p>The ybbR peptide was cloned into a pET20b(+) plasmid housing the RIDC3 vector using Quikchange mutagenesis protocols. RIDC3 and ybbRRIDC3 were expressed and purified as previously described.42 Briefly, cells harboring the RIDC3 or ybbRRIDC3 plasmid were grown in a lysogeny broth (LB, BioPioneer Inc.) medium containing 100 mg/L ampicillin and 34 mg/L chloramphenicol in 2.8-L flasks and shaken at 200 rpm for 16-20 h at 37 °C. Cells were harvested by centrifugation at 5,000 × g for 10 min, resuspended in a buffered solution containing 5 mM sodium acetate (pH 5.0) and 100 mg lysozyme (VWR) and lysed by sonication on ice. The crude lysate was subject to pH titration with 40% sodium hydroxide to pH 10 immediately followed by the addition of 50% (v/v) hydrochloric acid to pH 4.5 and centrifuged at 10,000 × g for 30 min. The cleared lysate was applied to a CM sepharose gravity column and eluted using a stepwise gradient (0 – 500 mM NaCl). Collected fractions were concentrated using an Amicon Stirred Cell (Millipore) and dialyzed against a buffered solution containing 10 mM sodium phosphate (NaPi) (pH 8.0) overnight. The protein was collected and purified using a DuoFlow workstation station equipped with a Bio-Scale Mini Macro-prep High Q-cartridge column (BioRad) and eluted using a linear gradient (0 – 500 mM NaCl). The separation of intact and truncated ybbRRIDC3 proteins (Figure S11a) was observed in this ion-exchange purification step, with intact ybbRRIDC3 eluting at ca. 80 mM NaCl and truncated ybbRRIDC3 eluting at ca. 110 mM NaCl. Protein purity was assessed by ultraviolet-visible (UV-vis) spectroscopy and samples with a RZ ratio (A415/A280) > 6 were pooled, concentrated and flash frozen for storage at −80 °C.</p><p>A pJ414 plasmid housing the ybbRRhuA gene was purchased from DNA 2.0. ybbRRhuA was expressed and purified in a similar fashion to C98RhuA, as previously described.39 Cells were grown in 30-mL LB cultures containing 100 mg/L ampicillin and shaken overnight at 37 °C. From these starter cultures, 5 mL was inoculated into 1-L LB media and shaken at 200 rpm at 37 °C to an optical density (OD) of 0.8 – 1. Protein expression was inducted with 1 mM isopropyl β-D-1-thiogalactopyranoside (IPTG, BioPioneer Inc.) for 12 h and cells were harvested by centrifugation at 5,000 × g for 10 min. Cells were resuspended in a buffered solution containing 10 mM tris(hydroxymethyl)aminomethane hydrochloride (Tris) (pH 7.5), 10 mM β-mercaptoethanol (βME) and 1 mM ZnCl2 supplemented with 100 mg lysozyme and 50 mg PMSF and lysed by sonication on ice. Cell lysate was centrifuged at 10,000 × g for 30 min, the cleared lysate was subject to the addition of Polymin-P (Acros) to a final concentration of 0.15% (w/v), and stirred for 30 min at 4 °C prior to centrifugation at 10,000 × g for 30 min. The supernatant was loaded onto a DEAE sepharose gravity column and eluted using a stepwise gradient (0 – 500 mM NaCl). Fractions containing ybbRRhuA were subject to the addition of ammonium sulfate to a final concentration of 1.7 M to precipitate the protein and stirred for 30 min prior to centrifugation at 10,000 × g for 30 min. The precipitated protein was resuspended in a buffered solution containing 10 mM Tris (pH 7.5), 10 mM βME, and 1 mM ZnCl2 and dialyzed against the same solution three times. The protein was collected and purified using a DuoFlow workstation station equipped with a Bio-Scale Mini Macro-prep High Q-cartridge column (BioRad) and eluted using a linear gradient (0 – 500 mM NaCl). Protein purity was assessed by SDS PAGE and pure fractions were dialyzed against a buffered solution containing 10 mM Tris (pH 7.5), 10 mM βME, and 1 mM ZnCl2 and flash frozen for storage at −80 °C.</p><p>B. subtilis Sfp, superfolder GFP (sfGFP), and ybbRGFP (EGFP-ybbR N-terminal fusion) variants were expressed and purified as previously described.43,44 Briefly, cells were inoculated into terrific broth containing 100 mg/L kanamycin and shaken at 200 rpm at 37 °C to an OD of 0.8. Protein expression was induced with 1 mM IPTG and shaken at 200 rpm overnight at 16 °C. Cells were harvested by centrifugation at 5,000 × g for 10 min, resuspended in a buffered solution containing 50 mM Tris (pH 7.5) and 250 mM NaCl supplemented with 100 mg lysozyme, 5 μg/mL DNAse I (Sigma), and 5 μg/mL RNAse (Worthington Biochemical Corp.) and lysed by French pressure cell press (500 – 1,000 psi). The cell lysate was centrifuged at 12,000 × g for 45 min and the cleared supernatant was subject to a Ni-NTA (Sigma) gravity column. The column was washed with a buffered solution containing 50 mM Tris (pH 7.5), 250 mM NaCl and 10 mM imidazole prior to protein elution at 300 mM imidazole (Sigma). Pure fractions were desalted into a buffered solution containing 50 mM Tris (pH 7.5) using a PD-10 desalting column (GE Healthcare Life Sciences), concentrated using an Amicon spin filter (Millipore), exchanged into a buffered solution containing 50 mM Tris (pH 7.4), 150 mM NaCl and 20% glycerol and flash frozen at −80 °C.</p><!><p>The desired sequence of the synthesized ybbR-N3 peptide is: DSLEFIASKLAG-K(N3). The Fmoc-Lys(N3) amino acid was purchased separately from Anaspec. Solid phase peptide synthesis using Rink amide MBHA resins was used to generate the peptide. A solution of 20% 4-methylpiperidine in dimethylformamide (DMF) was used for Fmoc deprotection (2 × 5 min) and peptide coupling was performed using 1-[Bis(dimethylamino)methylene]-1H-1,2,3-triazolo[4,5-b]pyridinium 3-oxide hexafluorophosphate (HATU) and N,N-diisopropylethylamine (DIPEA). Deprotection and coupling cycles were repeated for the addition of every amino acid and the final peptide was cleaved from the resin using a solution of trifluoroacetic acid (TFA) and dichloromethane (DCM) at a 9:1 ratio. Peptides were precipitated using cold ether, pelleted by centrifugation at 5,000 rpm for 15 min and dried in vacuo. The peptide was purified by reverse-phase high pressure liquid chromatography (HPLC) using a Hitachi-Elite LaChrom L2130 pump with a binary gradient and a UV-vis detector and eluted using a linear gradient of acetonitrile in water. Peptide purity was assessed by matrix-assisted laser desorption/ionization (MALDI).</p><!><p>TAMRA-CoA was synthesized and purified as previously described.45 GFP-CoA was prepared by first exchanging sfGFP into a buffered solution containing 20 mM 3-(N-morpholino)propanesulfonic acid (MOPS) (pH 7.5). Approximately 5 equiv. of sulfosuccinimidyl-4-[N-maleimidomethyl]cyclohexane-1-carboxylate (sulfo-SMCC, Fisher) dissolved in a solution of 100 μL DMF was added to a 1 mL solution containing 100 μM GFP and allowed to react at 37 °C for 1.5 h. The sample was buffer exchanged into a fresh solution of 20 mM MOPS (pH 7.5) using a 10 D/G desalting column to remove any unreacted SMCC and 5 equiv. of coenzyme A, free acid dissolved in a buffered solution containing 20 mM MOPS (pH 7.5) was added and allowed to gently shake overnight at room temperature on an orbital shaker. The sample was buffer exchanged into a fresh solution of 20 mM MOPS (pH 7.5) using a 10 D/G desalting column and characterized by UV-vis spectroscopy.</p><!><p>RIDC3 crystals were self-assembled by first exchanging RIDC3 into a buffered solution of 20 mM 2-(N-morpholine)ethanesulfonic acid (MES) (pH 5.5) using a 10 kDa Amicon spin filter. A concentrated stock solution of RIDC3 was supplemented with a solution of 20 mM MES (pH 5.5) and 5 equiv. ZnCl2 dissolved in water to a final protein concentration of 100 μM. After mixing with ZnCl2, the solution turned cloudy within 5 min and crystals matured in 7-10 days. Mature crystals were buffer exchanged × 5 into a fresh solution containing 20 mM MOPS (pH 7.5) by pelleting suspensions of crystals at 3,500 rpm in 30 s bursts for 2 min and carefully pipetting out the clear supernatant. Approximately 10 equiv. of a dibenzocyclooctyne-N-hydroxysuccinimidyl ester (DBCO-NHS) dissolved in DMF (25 mg/mL) was added to a suspension of RIDC3 crystals and allowed to gently shake at room temperature for 2 h. Crystals were buffer exchanged × 5 into a fresh solution containing 20 mM MOPS (pH 7.5) via centrifugation to remove any unreacted DBCO. A stock solution of ybbR-N3 dissolved in DMF was added to the suspension of RIDC3 crystals to a final ratio of 10:1 peptide:crystals and gently shaken overnight at room temperature. RIDC3-ybbR crystals were again buffer exchanged × 5 into a fresh solution containing 20 mM MOPS (pH 7.5) and analyzed by UV-vis spectroscopy and electrospray ionization mass spectrometry (ESI-MS).</p><p>The preparation of RIDC3-CoA crystals was performed similarly to RIDC3-ybbR crystals, starting with buffer exchanging mature crystals into a solution containing 20 mM MOPS (pH 7.5) via centrifugation. Approximately 10 equiv. of a stock solution of sSMCC dissolved in DMF was added to the suspension of crystals and allowed to gently shake at room temperature for 1.5-2 h. Crystals were buffer exchanged × 5 into a fresh solution containing 20 mM MOPS (pH 7.5) via centrifugation and 10 equiv. of coenzyme A, free acid dissolved in a buffered solution containing 20 mM MOPS (pH 7.5) was added and gently shaken overnight at room temperature. Crystals were buffer exchanged one final time and analyzed by UV-vis spectroscopy.</p><!><p>ybbRRIDC3 solutions were stored at 4 °C to limit degradation of the appended peptide. A stock solution of ybbRRIDC3 was combined with Zn2+ in buffered solutions at a pH range of 5.5 – 8.5. In general, a 50-μL sample contained 100 μM protein at [Zn2+]:[protein] ratios ranging from 1:1 to 10:1. Buffered solutions containing 20 mM MES were used at pH 5.5 and pH 6.5, 20 mM MOPS at pH 7.5 and 20 mM CHES at pH 8.5. Crystalline materials were found in solutions at pH 5.5 and pH 7.5. ybbRRhuA crystals were prepared by buffer exchanging protein into a fresh solution containing 10 mM T ris, 10 mM βME, and 1 mM ZnCl2 using a 10 kDaAmicon spin filter to a final protein concentration of 125-150 μM ybbRRhuA tetramer. Samples were gently shaken at 4 °C for 1-2 weeks until crystals matured.</p><!><p>A suspension of RIDC3-ybbR crystals was dissolved using a solution containing 1 mM ethylenediaminetetraacetic acid (EDTA), separated from any possible contaminants on a SDS-PAGE gel and stained with Coomassie Brilliant Blue. The band corresponding to RIDC3-ybbR was cut out into 1 mm × 1 mm cubes and destained × 3 with a solution containing 100 mM ammonium bicarbonate for 15 min followed by the addition of 100 μL acetonitrile (ACN) for 15 min. The supernatant was removed and gel pieces were dried in vacuo and chemically reduced with 200 μL of a solution containing 100 mM ammonium bicarbonate and 10 mm dithiothreitol (DTT) at 55 °C for 30 min. The supernatant was removed and gel pieces were incubated with 200 μL of a solution containing 100 mM ammonium bicarbonate and 55 mM iodoacetamide and incubated for 20 min at room temperature in the absence of light. Gel samples were washed first with a fresh solution of 100 mM ammonium bicarbonate and finally acetonitrile prior to dehydrating gel pieces using a Speedvac. Gel pieces were digested by first covering them in an ice-cold solution containing 50 mM ammonium bicarbonate and trypsin (0.1 μg/μL) and incubated on ice for 30 min. After the gel was completely rehydrated, excess solution was removed and replaced with a fresh solution containing 50 mM ammonium bicarbonate overnight at 37 °C. Peptide extraction was performed by the addition of a 50 μL solution containing 0.2% formic acid (v/v) and 5% ACN (v/v) in water and mixing at room temperature for 30 min. The supernatant was collected and the extraction procedure was repeated again, combining the 2nd supernatant with the previous solution. Samples were analyzed by liquid chromatography (LC) with tandem mass spectrometry (MS/MS) using electrospray ionization.</p><p>Trypsin-digested peptide solutions were analyzed using ultra high pressure liquid chromatography coupled with LC-MS/MS using nanospray ionization. Ionization experiments were performed using a TripleTof 5600 hybrid mass spectrometer (ABSCIEX) interfaced with nano-scale reverse-phase UPLC equipped with a 20-cm 75-micron ID glass capillary packed with 2.5-μm C18 CSHTM beads (Waters corporation). Peptides were eluted using a linear gradient (5-80% ACN) at a flow rate of 250 μL/min for 1 h. The UPLC solutions used were: Buffer A - 98% H2O, 2% ACN, 0.1% formic acid, and 0.005% TFA and Buffer B - 100% ACN, 0.1% formic acid, and 0.005% TFA. MS/MS data were acquired in a data-dependent manner; the MS1 data was acquired for 250 ms at m/z of 400 to 1250 Da and the MS/MS data was acquired from m/z of 50 to 2,000 Da. The independent data acquisition (IDA) parameters were as follows: MS1-TOF acquisition time of 250 ms, followed by acquisition of 50 MS2 events of 48 ms for each event. The threshold to trigger a MS2 event was set to 150 counts for ion charge states of +2, +3 and +4. The ion exclusion time was set to 4 s. Finally, the collected data were analyzed using Protein Pilot 5.0 (ABSCIEX) for peptide identifications.</p><!><p>Samples for enzymatic labeling were buffer exchanged via centrifugation (for crystal suspensions) or using Amicon spin filters (for soluble protein solutions) into a solution containing 20 mM MOPS (pH 7.5). Crystal suspensions were dissolved via the addition of a stock solution of EDTA for a final concentration of 1 mM and buffer exchanged × 5 into a fresh buffer (without EDTA) for experiments with soluble proteins. In general, a 20-μL reaction consisted of 10 μM Sfp PPTase, 15 mM MgCl2, 100-200 μM modified CoA and 50 μM protein-ybbR conjugate in a buffered solution containing 20 mM MOPS (pH 7.5). The solution was gently shaken on a gel rocker for 16-24 h at room temperature. Completed reactions were buffer exchanged × 5 via centrifugation (3500 rpm) to remove unbound dye and enzyme into a solution containing 20 mM MOPS (pH 7.5). A homogeneous solution of proteins (e.g., ybbRRIDC3 monomers or ybbRRhuA tetramers) were enzymatically labeled in an identical fashion. Completed reactions were buffer exchanged using 10 kDa Amicon spin filters × 5 to remove unbound dye.</p><!><p>A 5-μL suspension of crystals was pipetted onto a glass slide and covered with a cover slip (Fisher), sealing the edges with clear nail polish to prevent sample drying. Samples were imaged with a 100x oil objective on a spinning-disk confocal Zeiss Axio Observer inverted microscope equipped with a pair of Roper Quantum 5125C cameras. Samples were excited at 488 nm for green fluorescence and 564 nm for red fluorescence. Differential interference contrast and fluorescence images were captured at 1-s and 100-ms exposures, respectively. Images were collected in Slidebook 6 (Intelligent Imaging Innovations) and analyzed using Fiji (http://fiji.sc/Fiji).</p><!><p>A 3-μL suspension of crystals was pipetted onto formvar/carbon-coated Cu grids (Ted Pella, Inc.) that had been glow discharged for 45-60 s. Samples were incubated for 5 min, washed with 50 μL of MilliQ water and blotted with filter paper. A 3-μL drop of 1% uranyl acetate in water was pipetted onto the grid and incubated for 30-60 s and blotted dry with filter paper. Grids were imaged using a FEI Sphera transmission electron microscope operating at 200 keV, equipped with a LaB6 filament and a Gatan 4K charged-coupled device (CCD). Micrographs were collected using objective-lens underfocus settings ranging from 250 nm to 2 μm and analyzed using Fiji (http://fiji.sc/Fiji).</p><!><p>A 10-μL suspension of crystals was deposited onto freshly cleaved mica (Ted Pella, Inc.) and incubated for 10 min. The mica disc was gently dried using a stream of nitrogen with care not to push the drop over the edge of the disc. AFM measurements were performed on a Bruker Dimension Icon ScanAsyst atomic force microscope using a ScanAsyst-Air tip (Bruker) operating in tapping mode. Images were analyzed using NanoScope Analysis (v.1.5, Bruker).</p><!><p>Fluorescence measurements were performed using a 96-well plate (Falcon) containing 50 μL solutions of each sample. Excitation/emission wavelengths of 485/510 nm and 557/583 nm were used for green and red fluorescence respectively with a 2 nm slit width, a 0.2 s integration time and a gain of 100.</p><!><p>As our first 2D platform, we chose RIDC3, a variant of the monomeric protein cytochrome cb562, which we previously engineered for assembly into 1-, 2- and 3D protein lattices via Zn2+ coordination (Figure 1a).3 Self-assembled RIDC3 arrays have been previously shown to tolerate a broad pH (pH 5 – 9), temperature (4 – 80 °C) and solvent range (H2O, tetrahydrofuran, isopropanol), providing a robust platform for the surface display of biomolecules.24 To selectively functionalize RIDC3 array surfaces, we chose 4'-phosphopantetheinyl transferase (PPTase), a post-translational modification enzyme. Such enzymes perform selective labeling of biological architectures while maintaining efficiency and specificity under mild solution conditions (e.g., pH 7 at 25 °C in aqueous buffers), all desirable characteristics for interfacing nanomaterials with biological molecules. Native enzymes have been employed to direct protein-protein ligation or surface immobilization in aqueous environments46 (e.g., sortase47,48, PPTase49, farnesyl transferase50, transglutaminase51) to functionalize polymeric nanoparticles, gold surfaces, hydrogels and even to tailor cell surfaces. In particular, PPTases have shown remarkable flexibility as a tool for site-selective attachment of chemical probes onto proteins or peptides.52 PPTases covalently modify acyl carrier proteins (ACPs) at a conserved serine residue via the transfer of phosphopantetheine (PPant) from coenzyme A (CoA), which serves a crucial role in various biosynthetic pathways.13 Previous studies by Walsh and coworkers report the use of phage display to discover an 11 amino acid peptide (ybbR: DSLEFIASKLA) as a surrogate for the native ACP.53 The ybbR peptide acts as a minimal recognition sequence for the surfactin phosphopantetheinyl transferase (Sfp) from Bacillis Subtilis and can be used as a short peptide tag for site-specific protein labeling (Figure 1b).43,49 Sfp is known to be promiscuous towards many substrates and is capable of enzymatically transferring different sets of biomolecules tethered to CoA in a site-specific manner to a serine residue of ACP or ybbR (Figure 1c, Figure S1).45,54,55 The promiscuity of Sfp for a range of CoA analogs, and its recognition of short peptide substrates, provides a model system to explore the enzymatic surface functionalization of 2D crystalline protein arrays. As described below, we developed two strategies for the incorporation of a functional peptide handle onto RIDC3 crystals (Figure 1d, e): chemical modification of pre-formed 2D arrays with the peptide and the genetic incorporation of the peptide onto RIDC3 monomers followed by 2D self-assembly.</p><p>We first chose to chemically modify RIDC3 array surfaces with a synthesized ybbR peptide (1) to verify that crystallinity was retained upon the addition of the peptide and (2) ensure that chemoenzymatic labeling of ybbR remained possible on the surface. Suspensions of pre-formed RIDC3 crystals were treated with a 10-fold excess of a dibenzocyclooctyne-N-hydroxysuccinimidyl ester (DBCO-NHS) to modify surface-exposed lysine (Lys) residues (Figure 2a).56 The strained cyclooctyne, DBCO, was used to avoid using Cu(I) with Zn-coordinating RIDC3 arrays. Following treatment with DBCO, 10 equiv. of a synthesized ybbR peptide with a Gly spacer and azide-terminated non-natural amino acid, DSLEFIASKLA-G-K(N3) (Figure S2), was added to the RIDC3-DBCO crystal suspension and allowed to react overnight. Negative-stain transmission electron microscopy (ns-TEM) snapshots of chemically modified RIDC3 arrays confirmed that crystallinity was retained (Figure S3). To quantify ybbR conjugation, suspensions of RIDC3-ybbR crystals were washed and dissolved with the addition of the metal chelator ethylenediaminetetraacetic acid (EDTA) prior to characterization with electrospray ionization mass spectrometry (ESI-MS) and UV-vis spectroscopy (Figure 2b, c). ESI-MS revealed peaks corresponding to 1, 2, and 3 additions of DBCO and ybbR per RIDC3 monomer. These results were corroborated by UV-vis spectroscopy measurements monitoring the addition and consumption of DBCO (ε309 = 12000 M−1 cm−1)57, with 2.5 ± 0.35 ybbR peptides added per RIDC3 monomer (Figure S3c). A close examination of the RIDC3 crystal packing (PDB ID: 3TOM) revealed up to 7 surface-exposed Lys residues accessible for chemical modification (Figure S4). Thus, we performed tandem mass spectrometry (MS/MS) analysis of trypsin digested RIDC3-ybbR samples and identified three sites of modification on RIDC3: Lys19, Lys28, Lys85 (Figure S4, S5). Spectroscopic and ESI-MS data closely matched MS/MS results and confirmed the covalent attachment of ybbR onto RIDC3 crystals at surface-exposed Lys residues.</p><p>Enzymatic labeling of RIDC3-ybbR crystals was performed using Sfp PPTase and fluorescent CoA analogs, which provide a facile visual handle for positively identifying crystal modification.58 We first tested a small molecule CoA analog (TAMRA-CoA) in a one-pot reaction43,59, adding the dye with Sfp and MgCl2 to RIDC3-ybbR arrays and incubating the mixture overnight at room temperature with gentle shaking to facilitate enzymatic conjugation. Control samples were similarly prepared in the absence of Sfp and all solutions were thoroughly washed with buffer to remove any unbound dye (see Materials and Methods). Confocal microscopy measurements after enzymatic labeling showed brightly fluorescent RIDC3-ybbR crystals when incubated with Sfp and CoA whereas no fluorescence signals were detected in a negative control sample devoid of Sfp (Figure 3a). Importantly, differential interference contrast (DIC) images show a perfect overlap between areas of fluorescence and RIDC3-ybbR crystals in the field-of-view (Figure 3a). Corresponding ns-TEM micrographs of the labeled samples confirm that the arrays remain intact after enzymatic treatment and retain crystallinity (Figure S6). Furthermore, PPTases are promiscuous in their recognition of CoA analogs, so we tested enzymatic labeling using a larger biomolecule, GFP-CoA. In order to enable the enzymatic transfer of a GFP-PPant group onto ybbR, a GFP-CoA analog was generated by first incubating sfGFP with the bifunctional linker, sulfosuccinimidyl 4-(N-maleimidomethyl) cyclohexane-1-carboxylate (sSMCC)60, to label Lys residues with a NHS-ester moiety followed by the addition of CoA-SH for thiol-maleimide coupling (Figure S7a). CoA addition was quantified using UV-vis measurements, yielding 1.4 CoA per sfGFP (Figure S7b). As with TAMRA-CoA modification, incubation of RIDC3-ybbR crystals with Sfp and GFP-CoA resulted in brightly fluorescent crystals (Figure 3a) and the absence of Sfp showed minimal fluorescence at the location of the arrays. Based on the successful chemoenzymatic transfer of a small molecule dye (TAMRA-PPant) and a protein (sfGFP-PPant) onto RIDC3-ybbR crystals, we posited that a solution containing TAMRA-CoA and GFP-CoA could be used to colocalize both labels on the arrays. Suspensions of RIDC3-ybbR crystals incubated with both CoA substrates and Sfp displayed distinct fluorescence at TAMRA and GFP wavelengths, and an overlay of both images showed an unambiguous overlap of both signals on the crystal surface (Figure 3b).</p><p>While identifying fluorescently labeled crystals was straightforward using confocal microscopy, quantification of enzymatic labeling of RIDC3-ybbR crystals proved challenging. RIDC3-ybbR crystals after enzymatic modification could be dissolved using EDTA to yield soluble protein amenable for UV-vis spectroscopic measurements; however, the absorbance values for covalently tethered TAMRA or GFP were too low relative to that of RIDC3 for unambiguous detection. This is due in part to the intense absorbance of the c-type heme covalently bound to the RIDC3 monomer at the Soret and Q bands (415 and 527 nm, respectively), partially overlapping with the absorption maxima of GFP (485 nm) and TAMRA (554 nm) (Figure S8a). Instead, we directly probed crystal suspensions using a microplate reader to measure fluorescence intensities for labeled vs. unlabeled samples (Figure 3c). A standard curve of [fluorophore] vs. fluorescence intensity was prepared using TAMRA-CoA or GFP-CoA solutions to determine [RIDC3-ybbR + TAMRA-PPant] or [RIDC3-ybbR + GFP-PPant] based on fluorescence values (Figure S9a). Crystals were dissolved after fluorescence measurements to obtain protein concentrations via UV-vis spectroscopy. Our calculations suggest <0.5% of RIDC3-ybbR is enzymatically modified by Sfp using either CoA analog (Figure S9c). This unexpectedly low labeling efficiency prompted further analysis.</p><p>Since enzymatic reactivity at the surface may be lower than in solution, we obtained RIDC3-ybbR monomers (from dissolving chemically modified RIDC3-ybbR crystals with EDTA). Functionalization of these monomers revealed approximately 0.33 TAMRA per RIDC3, significantly higher than labeling on the crystal surface (Figure S8b). Previous studies on RIDC3 arrays have shown that solution-assembled crystals are multilayered24, likely limiting enzyme access to "interior" RIDC3 proteins within a 3D stack. Indeed, dry-state AFM characterization of RIDC3-ybbR crystals confirms that they consist of ~4 nm thick 2D layers that stacked to an average height of 133 ± 40 nm (Figure S9b). Considering that only ybbR peptides decorating the exterior of RIDC3-ybbR crystals are likely available sites of modification using Sfp, we used a combination of spectroscopic data (to quantify ybbR conjugation onto RIDC3) and AFM measurements (to gather crystal dimensions) to determine the enzyme-accessible fraction of the average RIDC3-ybbR crystal. From 46 crystals, we determined average dimensions of a = 3.2 ± 0.84 μm, b = 8.1 ± 2.4 μm, c = 0.133 ± 0.04 μm (Figure S9b, c). We next used the RIDC3 crystal packing to determine the protein "step" in each crystal dimension: 1 protein per 3.79 nm x 3.46 nm x 2.3 nm volume (PDB ID: 3TOM). Based on these measurements, we calculated the surface coverage of RIDC3 proteins to be ~3.5% of total protein contained in a given crystal. Correlating fluorescence intensities to absorbance values, we computed ~3% and ~10% enzymatic labeling of RIDC3-ybbR crystal surfaces for GFP-CoA and TAMRA-CoA respectively. Although these values are lower than those observed for labeling of free proteins in a homogenous solution, our experiments show that it is possible to enzymatically tailor the surfaces of "solid", micron-scale 2D protein materials with non-negligible yields. To our knowledge, this is the first demonstration of the generation of functional protein arrays directed by enzymatic processes with the use of two distinct covalent tags.</p><!><p>In addition to chemically conjugating ybbR tags onto 2D RIDC3 arrays as a means for CoA-mediated enzymatic labeling with small fluorophores, we wondered whether the 2D protein surface could first be functionalized with CoA molecules instead and subsequently labeled with guest proteins bearing ybbR tags. Thus, RIDC3-CoA arrays were prepared using a similar conjugation strategy to that used for generating GFP-CoA (Figure 4a). Pre-formed RIDC3 arrays were first treated with sSMCC to covalently modify Lys residues with the NHS-ester moiety followed by the addition of CoA-SH (unmodified, natural substrate) to react with the maleimide portion of SMCC. CoA addition was quantified via UV-vis spectroscopy λmax = 259 nm) and determined to be ~0.8 per RIDC3 (Figure 4b). Although a 25-fold excess of CoA was added after SMCC coupling, lower efficacy of thiol-maleimide chemistry at the array surface (especially compared to the strained cyclooctyne click chemistry used for ybbR conjugation) and the potential for CoA self-dimerization perhaps contributed to a lower CoA:RIDC3 ratio. Nevertheless, RIDC3-CoA crystals were incubated with Sfp and a genetically encoded ybbRGFP (EGFP-ybbR N-terminal fusion) to generate fluorescent GFP-labeled arrays (Figure 4c, d). Control experiments in the absence of PPTase or ybbRGFP showed no fluorescence, indicating that there is little to no nonspecific association of GFP to the RIDC3-CoA arrays. Interestingly, ybbRGFP-labeled RIDC3-CoA arrays consist of punctate patches of GFP forming an outline along the edges of the array (Figure 4d) unlike RIDC3-ybbR samples, which showed uniform fluorescence across the surface. This could arise from a combination of (1) slower reaction kinetics for maleimide-thiol coupling (in contrast to the rapid copper-free click chemistry employed for ybbR conjugation previously) resulting in (2) a greater density of CoA molecules covalently tethered onto the edges of the array and (3) easier access of Sfp and ybbRGFP to the edges relative to the interior. Quantification of ybbRGFP labeling onto RIDC3-CoA labels revealed ~8% of surface exposed proteins were labeled, on par with that of RIDC3-ybbR arrays (Figure 4c, Figure S10). These findings illustrate the feasibility of a tripartite protein material platform, in which the surface of the 2D arrays of one protein (RIDC3) can be processed by a second, catalytic protein (Sfp) to site-specifically attach a third protein (GFP).</p><!><p>Our results thus far indicated that enzymatic labeling only engages a fraction of the surface sites containing ybbR or CoA tags. While this can, in large part, be attributed to the inefficient mass transport of crystalline substrates and the immobility of the proteins embedded therein, we posited that the non-optimal positioning of surface attachment sites may also play a role. Thus, we constructed ybbRRIDC3 fusions with the ybbR peptide genetically appended to the polypeptide termini of RIDC3 to allow for (1) a uniform distribution of peptide upon forming crystals and (2) to bypass the post-assembly conjugation of ybbR or CoA to generate enzymatically-addressable 2D arrays (Figure 1e).</p><p>Upon close inspection of the N- and C-termini of RIDC3, we determined that only the C-terminus of RIDC3 faces the exterior of the crystal surface. Thus, RIDC3 was genetically modified at the C-terminus with a Gly-Gly spacer followed by the ybbR peptide (RIDC3 – GG – DSLEFIASKLA) to generate the ybbRRIDC3 plasmid, which was expressed and purified in the same manner as RIDC3. However, in the course of extensive studies, we found that ybbRRIDC3 fusions underwent a degradation process (of unknown mechanism) in which the final four residues (SKLA) of ybbR fusion were cleaved and thus rendered inactive for Sfp-mediated labeling (Figure S11). Although the fraction of uncleaved ybbRRIDC3 fusions was amenable to enzymatic labeling and self-assembly into extended arrays (Figure S11c, d), the cleavage process persisted even upon self-assembly. Thus, we set out to test the genetic-tagging strategy on another protein building block, with the expectation that an altered protein environment may prevent ybbR auto-cleavage.</p><p>In a recent report from our laboratories, a C4-symmetric protein – L-rhamnulose-1-phosphate aldolase (RhuA) – was modified with cysteine (Cys) residues at its corners (C98RhuA) to facilitate the self-assembly of unsupported 2D arrays in solution under controlled oxidation conditions via disulfide bond formation.39,61 The C98RhuA proteins tessellate to form an alternating arrangement of tetramers in the crystalline lattice. Furthermore, rotations about the individual axes of symmetry at the flexible disulfide linkages create a coherently dynamic lattice that display auxeticity (i.e. a longitudinal expansion when stretched in the transverse direction) with a Poisson's ratio of −1. Aside from these unique dynamic properties, the larger size (274 residues per monomer vs. 106 residues in RIDC3) and inherent symmetry of C98RhuA could provide a more robust scaffold for the genetic incorporation of ybbR while retaining its self-assembly properties. As with RIDC3, ybbR was genetically appended to the C-terminus of C98RhuA with a short spacer (C98RhuA – SGSG – DSLEFIASKLA) so that the peptide is displayed at the surface of 2D arrays minimally interfering with disulfide formation (Figure 5a, b). Importantly, no truncation of ybbR was observed during the purification and isolation in this construct (Figure S12a). ybbRRhuA self-assembles into a suspension of ordered crystals at the same solution conditions as C98RhuA (20 mM Tris pH 7.5, 10 mM βME, 1 mM ZnCl2), maintaining crystallinity and lattice dynamics (Figure 5c). Furthermore, ybbRRhuA tetramers are successfully modified after incubation with Sfp PPTase and TAMRA-CoA, containing ~0.22 TAMRA per polypeptide (i.e. 0.89 TAMRA perybbRRhuA tetramer) after enzymatic labeling (Figure 5d). Enzymatic labeling of ybbRRhuA crystals with Sfp PPTase and modified CoA was observed via confocal microscopy and ns-TEM. Brightly fluorescent crystals were found with both TAMRA-CoA and GFP-CoA-incubated solutions whereas ybbRRhuA arrays in the corresponding DIC images were harder to discern likely due to the increased porosity and minimal 3D layering of 2D crystals relative to RIDC3 arrays (Figure 5e). TEM analysis of the labeled samples confirmed that arrays remained intact after enzymatic modification (Figure 5f). Confocal microscopy and ns-TEM analysis of control samples devoid of Sfp PPTase were minimally fluorescent and remained crystalline (Figure S12b). Quantification of ybbRRhuA labeling was performed using fluorescence measurements on a plate reader as previously described (Figure S12c). We calculated ~0.7% and ~5.6% enzymatic labeling of ybbRRhuA crystal surfaces for GFP-CoA and TAMRA-CoA respectively. As seen with chemoenzymatic labeling of a solution containing ybbRRhuA tetramers, labeling of the array surfaces was less efficient than with their RIDC3 counterparts. However, in contrast to genetically modified ybbRRIDC3, the ybbRRhuA construct was found to be intact and stable and could be enzymatically modified without loss in crystallinity. Our results thus illustrate the robust enzymatic labeling of 2D protein arrays that are genetically modified with peptide recognition tags.</p><!><p>In this study, we have illustrated several strategies for the site-selective chemoenzymatic functionalization of artificial protein assemblies. Previous studies have investigated bioconjugation using Sfp on polymeric nanoparticles54, hydrogels62, antibodies55, protein-DNA chimeras63 or in conjunction with sortases for dual labeling capabilities.64 In our own work, we have utilized bacterial enzymes of this kind, and their peptide substrates, to assemble amphiphilic nanomaterials.65 Our study now expands this strategy to artificial, 2D crystalline protein arrays. Using two different protein systems that we previously designed for 2D self-assembly, we demonstrated that a) molecular or peptidic molecular tags could be attached to the 2D arrays both before or after self-assembly, b) the tag attachment could be done both chemically and genetically, and c) the selective enzymatic labeling of these recognition tags could proceed with small molecules as well as with proteins as substrates. Importantly, in all cases, the structural fidelity of the proteins as well as the crystalline order of their 2D assemblies were maintained. Although the overall chemoenzymatic labeling yields of the 2D protein arrays are far from quantitative, this is possibly due to the inherent physical limitations posed by a ternary reaction involving an enzyme, a suspended "solid-state" protein array with limited mobility as a template, and large fluorescent molecules (even a protein) as a substrate. This poses an interesting challenge for future studies in which we can envision optimizing chemoenzymatic labeling by engineering the protein array surfaces, the peptide substrates (leveraging novel machine learning and optimization tools45), and the Sfp enzyme for complementary electrostatic interactions or improved reaction sterics. Another exciting possibility is the use of secondary enzymes (like acyl carrier protein hydrolases),43,66 which have been shown to selectively cleave ybbR peptide tags, opening the path for the reversible tailoring of protein array surfaces.</p><p>Enzyme-mediated post-translational modification of proteins is ubiquitous in nature and greatly increases the complexity of the proteome. Here, we have reported one of the first examples for leveraging biological enzymes to selectively modify solid substrates with small molecules and proteins, with an eye toward the hierarchical construction of multi-component protein systems. The versatility and inherently modular nature of the strategies described herein offer promise in generating functional and hybrid materials for use in sensing, catalysis, immotherapeutics, or lab-on-a-chip design frameworks.</p>

PubMed Author Manuscript

Structure-Based Approach to the Identification of a Novel Group of Selective Glucosamine Analogue Inhibitors of Trypanosoma cruzi Glucokinase

Glucokinase and hexokinase from pathogenic protozoa Trypanosoma cruzi are potential drug targets for antiparasitic chemotherapy of Chagas\xe2\x80\x99 disease. These glucose kinases phosphorylate D-glucose with co-substrate ATP and yields glucose 6-phosphate and are involved in essential metabolic pathways, such as glycolysis and the pentose phosphate pathway. An inhibitor class was conceived that is selective for T. cruzi glucokinase (TcGlcK) using structure-based drug design involving glucosamine having a linker from the C2 amino that terminates with a hydrophobic group either being phenyl, p-hydroxyphenyl, or dioxobenzo[b]thiophenyl groups. The synthesis and characterization for two of the four compounds are presented while the other two compounds were commercially available. Four high-resolution X-ray crystal structures of TcGlcK inhibitor complexes are reported along with enzyme inhibition constants (Ki) for TcGlcK and Homo sapiens hexokinase IV (HsHxKIV). These glucosamine analogue inhibitors include three strongly selective TcGlcK inhibitors and a fourth inhibitor, benzoyl glucosamine (BENZ-GlcN), which is a similar variant exhibiting a shorter linker. Carboxybenzyl glucosamine (CBZ-GlcN) was found to be the strongest glucokinase inhibitor known to date, having a Ki of 0.71 \xc2\xb1 0.05 \xce\xbcM. Also reported are two biologically active inhibitors against in vitro Trypanosoma cruzi culture that were BENZ-GlcN and CBZ-GlcN, with intracellular amastigote growth inhibition IC50 values of 16.08 \xc2\xb1 0.16 \xce\xbcM and 48.73 \xc2\xb1 0.69 \xce\xbcM, respectively. These compounds revealed little to no toxicity against mammalian NIH-3T3 fibroblasts and provides a key starting point for further drug development with this class of compound.

structure-based_approach_to_the_identification_of_a_novel_group_of_selective_glucosamine_analogue_in

1. Introduction<!>2.1. Materials<!>2.2. Cloning<!>2.3. Standard Methods<!>3.1. Structural Comparison between TcGlcK and HsHxKIV<!>3.2. X-ray Crystallography of TcGlcK Inhibitor Complexes<!>3.3. Inhibition Studies of TcGlcK and HsHxKIV by Glucosamine Analogues<!>3.4. Analysis of Inhibitor Binding in TcGlcK<!>3.5. Effects of Glucokinase Inhibitors on In Vitro Amastigote Viability<!>3.6. Structural Factors for Selectivity of Glucosamine Analogues in TcGlcK and TcHxK<!>4. Conclusions

<p>Trypanosoma cruzi protozoa are human pathogenic Kinetoplastid parasites that cause Chagas' disease. There are approximately 6–7 million people infected worldwide with T. cruzi parasites and a majority are from 21 Latin American countries (1). The Centers for Disease Control and Prevention estimates that there are over 300,000 people infected in the United States (2). The endemic regions span from Argentina to Mexico (3–5) and T. cruzi parasites are an increasing danger to the U.S. public health (6) since 10 states (primarily in the Southwest) have reported T. cruzi-infected specimens of triatomine insect vector species (7). First-line clinically available drugs used in Latin America, such as benznidazole and nifurtimox (Figure 1), which have still been in use since 1985 and earlier, are highly effective against T. cruzi parasites in the acute stage with parasitological cures upwards to 80% (8).</p><p>However, in the chronic stage, the same antiparasitic therapy has substantially lower efficacy for reasons that are not well understood (9). These drugs have tolerability issues because they produce harsh side effects, including allergic dermopathy, anorexia, peripheral polyneuropathy, and vomiting (10, 11), and as a result, there is weak patient compliance. Furthermore, different T. cruzi strains were observed to respond much differently to certain inhibitors from in vitro biological inhibition assays, such as posaconazole, indicating that a drug that gives rise to good biological activity from one strain may not have the same effectiveness towards another (12). In order to develop new and improved drugs, the major challenges such as drug efficacy of chronic stage Chagas' disease, drug tolerability, and universal compound effectiveness simply need to be overcome.</p><p>Glycolysis and the pentose phosphate pathway (PPP) are implicated as targets in T. cruzi for antiparasitic drug development (13, 14). Glycolytic enzymes have been studied as potential drug targets for the treatment of trypanosomatid diseases (13–15), especially the hexokinases from T. cruzi (16–18) and T. brucei (19–22). Metabolic studies have demonstrated that T. cruzi trypomastigotes and amastigotes use glucose as their preferred carbon source in glucose-rich media (23). Additionally, a metabolic control analysis study involving RNA interference (RNAi) against five important T. brucei glycolytic enzymes was assessed for the role of glycolytic flux (i.e., as glucose consumption and pyruvate production) on the growth trend of the parasite's bloodstream life-stage and good evidence was revealed for T. brucei hexokinase (TbHxK) as a validated drug-target (24). T. cruzi parasites are highly dependent on glucose for cell growth and differentiation (25–27) and inhibition of mostly any of the glycolytic enzymes in T. cruzi will likely lead to prompt cell death (14).</p><p>Two enzymes responsible for the catalysis of D-glucose to glucose 6-phosphate (G6P), in the presence of ATP and Mg2+, include hexokinase and glucokinase. The inhibition of these enzymes appears to be a good strategy for targeting since glycolysis and the PPP should encounter a direct impact by functioning with lower metabolic flux. T. cruzi hexokinase (TcHxK) has higher activity and is found at a higher percentage than T. cruzi glucokinase (TcGlcK) in the parasite (28), it is currently proposed that TcHxK is a potential drug target (16), but the status of TcGlcK is unclear. Genetic validation by RNAi cannot be carried out in T. cruzi parasites due to a nonfunctional RNAi pathway (29, 30). However, the structure of TcGlcK was solved in recent years through X-ray crystallography by Cordeiro and colleagues (31) and there is still no solved crystal structure for any trypanosomal (T. cruzi or T. brucei) hexokinase.</p><p>Hexokinases have a broad substrate range where various monosaccharides can be phosphorylated at appreciable rates, including D-glucose, D-mannose, D-galactose, and D-fructose; in addition, hexokinases are inhibited by G6P. Hexokinases also have a higher affinity for D-glucose than glucokinases. Glucokinases exhibit high specificity for D-glucose catalysis, they are not inhibited by G6P, and they have lower affinity for D-glucose (compared to hexokinases) (32). Glucose kinases (hexokinases or glucokinases) are categorized into two nonhomologous families, the hexokinase family and the ribokinase family. Within the hexokinase family (the subject of the work herein), there are three groups where glucose kinases are further subdivided on the basis of conserved amino acid 1° structure, which include (1) the hexokinase group, (2) group A, and (3) group B. Between these groups, conservation is lost when 1° structure sequence alignments are performed (e.g., protein BLAST). Additionally, the molecular mass of a hexokinase group enzyme (50–54 kDa per subunit) is generally higher on a subunit basis compared to group A (35–42 kDa per subunit) or group B (30–36 kDa per subunit) enzymes (32). T. cruzi has a hexokinase (from the hexokinase group; subunit MW of ~52 kDa) and a glucokinase (from group A; subunit MW of ~42 kDa), which is especially unusual because trypanosomatids are the only organisms known to have both enzymes present; various Leishmania spp. also have both a hexokinase and a group A glucokinase. Moreover, the functionality of both TcGlcK and TcHxK are quite similar because they perform as true glucokinases (as mentioned above) (16, 28). On the basis of percent sequence identity from a BLAST against 1° structure of hexokinases found in the GenBank database, TcHxK aligns with various existing conserved motifs, but this is not the case when aligning against glucokinases, such as TcGlcK. TcHxK and TcGlcK also have an approximate 17-fold difference in their KM values for D-glucose catalysis of 0.060 mM (16) and 1.00 mM (28), respectively.</p><p>E. coli glucokinase (EcGlcK) has a representative X-ray crystal structure for group A glucokinases (32). A structure-based sequence alignment between TcGlcK (PDB entry 2Q2R) (31) and EcGlcK (PDB entry 1SZ2) (33) was found to be consistent with secondary structural elements (31) even though the 1° sequence alignment was less than 16% identical. The quaternary structure of TcGlcK (dimeric form) is also very similar to that of EcGlcK. Additionally, there are four Homo sapiens hexokinase (HsHxK) isoenzymes I–IV and HsHxKIV has a representative structure among them. The crystal structures were solved for all four isoenzymes and they are among the hexokinase group even though HsHxKIV has been termed glucokinase based on its in vivo physiological activity (31). The structural comparison provided by Lunin et al. (33) for EcGlcK and the C-terminal half of HsHxKI (residues 475–917) showed that the tertiary structures and active sites were similar. The 3° structure of EcGlcK (the group A structural prototype) compares very well to both TcGlcK and HsHxKI (C-terminal half).</p><p>The basis of the presented work was to use structure-based drug design (SBDD) and a surrogate X-ray crystallography approach, in which we conceived of competitive inhibitors by using SBDD on TcGlcK (PDB entry 2Q2R) with the intention of inhibiting TcGlcK (only) or both TcGlcK and TcHxK. In order to achieve selectivity from an inhibitor design (e.g., the inhibitor can bind stronger to TcGlcK and avoid binding with any of the four human hexokinase isoenzymes), visual assessments were made after the crystal structure of TcGlcK was superimposed onto the crystal structure of HsHxKIV. Another goal of the study was to determine if the TcGlcK inhibitors would also inhibit TcHxK because both enzymes phosphorylate D-glucose through catalysis. The active site regions in TcGlcK and TcHxK are assumed to compare well structurally, albeit a 19% global pairwise 1° structural alignment between the two enzymes. Furthermore, we sought to determine if TcGlcK was a chemically-validated drug-target by using glucosamine analogue inhibitors.</p><p>Here, we report the X-ray crystal structures of TcGlcK in its complexes with inhibitors BENZ-GlcN, CBZ-GlcN, HPOP-GlcN, and DBT-GlcN (Figure 1). These inhibitors present the molecular details for affinity in the active site and may disclose more inhibitor design possibilities that can be used for optimized selectivity and affinity. We present inhibitor kinetic parameters of the aforementioned inhibitors with enzymes TcGlcK and HsHxKIV. The synthesis and characterization are shown for HPOP-GlcN and DBT-GlcN whereas inhibitors BENZ-GlcN and CBZ-GlcN were commercially available. Finally, we show effective in vitro inhibition assays against T. cruzi amastigote parasites for BENZ-GlcN and CBZ-GlcN.</p><!><p>N-Cbz-D-glucosamine (CBZ-GlcN), 2-benzamido-2-deoxy-D-glucopyranose (BENZ-GlcN), and isopropyl β-D-thiogalactopyranoside (IPTG) were purchased from Carbosynth. Dulbecco's modified Eagle's medium (DMEM) was purchased from CellGro. N-Benzyl-2-nitro-1H-imidazole-1-acetamide (benznidazole, 97%), hexadeuterodimethyl sulfoxide (D6-DMSO, 99.96 atom%), tetrahydrofuran (anhydrous grade), 3-(4-hydroxyphenyl)propionic acid N-hydroxysuccinimide ester (≥97.0%), ammonium formate (HPLC grade, ≥99.0%), acetonitrile (HPLC grade), chlorophenol red-β-D-galactoside (CPRG), ethylenediaminetetraacetic acid tetrasodium salt hydrate (>99.0%), imidazole (99+%), bovine pancreas deoxyribonuclease I (DNase I), bovine pancreas ribonuclease A (RNase A), D-(+)-glucosamine hydrochloride (≥99%), triethanolamine (≥99.0%), sodium citrate tribasic dihydrate (≥99.0%), carboxymethyl (CM) cellulose, Saccharomyces cerevisiae glucose 6-phosphate dehydrogenase (type XV), β-nicotinamide adenine dinucleotide phosphate hydrate (NADP+, ≥95%), adenosine 5′-triphosphate disodium salt hydrate (ATP, 99%), D-glucose 6-phosphate sodium salt (≥98%), and all buffer salts (≥98%) were purchased from Sigma. Cobalt-nitrilotriacetic acid (Co-NTA) resin, methanol, ethyl acetate, lysozyme (type VI), triethylamine, DL-dithiothreitol, 2xYT broth, lysogeny broth (LB), kanamycin sulfate, protease inhibitor tablets (EDTA-free), 1,1-dioxobenzo[b]thiophen-2-ylmethyl N-succimidyl carbonate (95%), and all other chemicals were purchased from Fisher Scientific.</p><!><p>The following genes: Trypanosoma cruzi glucokinase, strain CL Brener (GenBank accession number XP_821474) and Homo sapiens hexokinase IV (UniProtKB accession code P35557) were cloned into separate kanamycin–resistant pET-28a(+) Escherichia coli expression vectors at restriction sites 5′ NcoI and 3′ HindIII at Genewiz, Inc. (South Plainfield, NJ). Codons in all plasmids were optimized for protein expression. These plasmid constructs were designated as pET-TcGlcK-xtal and pET-HsHxKIV-xtal, respectively. Each construct encodes for an N-terminal hexahistidine tag. The pET-TcGlcK-xtal plasmid encodes the segment MGRGSHHHHHHGMA that precedes the start methionine and the 9-residue segment VGKKQKAQL at the C-terminal region was not included. The CL Brener strain of T. cruzi glucokinase differs at eight positions compared to the previously reported construct used by Cordeiro and colleagues in the first X-ray crystal structure determination of TcGlcK; see PDB entry 2Q2R. These positions in the CL Brener strain of TcGlcK include the following changes: A22V, I65L, M81I, H125R, L213I, F232L, H327R, and S344T. The pET-HsHxKIV-xtal plasmid encodes the segment MGHHHHHHENLYFQGM that precedes residue K12 (N-terminal segment M1 – A11 of HsHxKIV was not inlcuded) and the 8-residue segment KKACMLGQ at the C-terminal region was also not included. This pET-HsHxKIV-xtal construct was based off of one used in X-ray crystallography experiments (34).</p><!><p>Standard methods are presented in the Supplementary Information section for the following: (1) the expression and purification of TcGlcK and HsHxKIV, (2) activity assays of TcGlcK and HsHxKIV, (3) crystallization of TcGlcK, (4) X-ray crystal structure determinations, (5) general procedures for the synthesis of glucosamine analogues, (6) the synthesis of HPOP-GlcN, (7) the synthesis of DBT-GlcN, and (8) biological assays. X-ray diffraction data collection and refinement statistics for the TcGlcK inhibitor complexes are shown in Table 1.</p><!><p>TcGlcK is a dimeric enzyme with two identical subunits. Each monomer has a large α + β domain (residues 145 – 352) and a small α/β domain (residues 1 – 130 and 353 – 367) (Figure 2).</p><p>TcGlcK selective inhibitor designs were based on structural details of the HsHxKIV and TcGlcK active sites. HsHxKIV has a large loop segment (L165 – G178) and a shorter loop segment (S281 – Q287) found adjacent to the glucose binding site. Various residue side chains stem off of these loops and cluster near each other in addition with some other residues not part of the loops; and they all include residues: K56, P153, N166, T168, and Q286 (Figure 3a). When hydroxyphenyloxopropyl glucosamine (HPOP-GlcN) (Figure 1) is modeled into the glucose binding site (without having any energy minimization performed), the p-hydroxyphenyl moiety is extended by a linker into the clustered side chains region, and the linker is positioned between a narrow channel, which is bordered by P153 and T168 (Figure 3b). From the proximity of all of these residues, we predict a steric clash situation would result. On the other hand, the same region and channel observed in TcGlcK that is next to the glucose binding site (Figure 3c) has some residue side chains, P103 and N105, on the large loop segment (G102 – I110) that are not tightly clustered together; in addition, the side chains of P94 and N105 make up the borders of a much wider channel for the linker to extend through (Figure 3d). Modeling HPOP-GlcN in TcGlcK (by the same method implemented for HsHxKIV) reveals that the linker and p-hydroxyphenyl moiety are not involved in any significant steric clashes, based on our visual assessment.</p><p>HPOP-GlcN may not favor binding in HsHxKIV, but will most likely favor binding in TcGlcK and thus give rise to selectivity. The clustered side chains in HsHxKIV are mostly flexible, and if HPOP-GlcN truly binds in this active site, as a consequence, we expect the flexible side chains of residues K56, N166, T168, and Q286 to reposition (see orange dashes in Figure 3b). However, an inhibitor with a bulkier tail group such as dioxobenzylthiophenyl glucosamine (DBT-GlcN) (Figure 1) is proposed to have substantial steric effects in that region regardless of side chain repositioning because of little available space. In regard to the channels that the linkers extend through, residue P153 (of HsHxKIV) has a rigid side chain that we predict will not be able to avoid close contact interactions with the linker groups of any of the glucosamine analogues (Figure 1) because there are key glucose-moiety binding interactions that must be maintained at the glucose binding site. P94 (of TcGlcK) is the comparable residue to P153 (Figure 4), but it is positioned differently to avoid steric hindrance of an extending linker (Figure 3d), and thus, gives rise to a wider channel to make easy access for glucosamine analogues in TcGlcK.</p><p>The determination of a good competitive inhibitor for hexokinase was previously accomplished by the early work of Anderson et al. (39) and Steitz et al. (40), in which an o-toluoyl glucosamine compound was complexed with yeast hexokinase through X-ray crystallography (PDB entry 2YHX). This finding proceeded to the development of modestly potent benzoyl glucosamine-based inhibitors for TbHxK (22, 41), including the m-bromo-benzoyl glucosamine inhibitor that had a determined Ki of 2.8 μM (22). We also tested one inhibitor that was shorter than the rest, benzoyl glucosamine (BENZ-GlcN), because the chemical structure for this compound was very similar to o-toluoyl glucosamine and this was a class of compound that was never tested on a T. cruzi hexokinase/glucokinase.</p><!><p>The glucosamine analogues all share a common glucose moiety that preserves key enzyme-substrate hydrogen bonding interactions with the monosaccharide hydroxyl groups from C1, C3, C4, and C6. The C2 hydroxyl is replaced by a NH group and also participates in the key hydrogen bonding. However, each compound differs in its linker and tail group designated to interact with the intersubunit region adjacent to the glucose binding site. Crystal growth of TcGlcK in the absence of D-glucose followed by soaks of glucosamine analogues at room temperature allowed inhibitors to bind in the active site as confirmed by X-ray crystallography. All of the TcGlcK inhibitor complexes in this work crystallized as a homodimer (in the asymmetric unit) in the P21 space group and were found in a near fully-closed conformation between the large and small domains. The homodimeric subunit-subunit interface is stabilized by a large contact surface area of 1,912 Å2 (~12% of the total surface area) as determined by the Protein Interfaces, Surfaces, and Assemblies (PISA) server (42). This is in contrast to the loose tetramer that is observed for the TcGlcK-D-Glucose-ADP complex (PDB entry 2Q2R), which crystallizes in a P21212 space group and was found as a fully-closed conformation (31). This tetramer is stabilized by having large contact surface areas as well as by having intermolecular ionic interactions (e.g., salt bridges) and hydrogen bonds. In the P21 space group the intermolecular ionic interactions that normally result from residues I219, D220, and Q223 are essentially absent. A superposition of the TcGlcK P21 crystal form (dimer) onto the P21212 crystal form (tetramer) reveals that the dimers having higher surface area contact are essentially identical (results not shown).</p><p>For structural comparisons, all four ligand complexes (including subunits) are well conserved. With the TcGlcK-CBZ-GlcN complex (PDB entry 5BRE) as a reference structure for superposition, the overall fold of the dimer is essentially identical to that of the TcGlcK-BENZ-GlcN complex (PDB entry 5BRD), the TcGlcK-HPOP-GlcN complex (PDB entry 5BRF), and the TcGlcK-DBT-GlcN complex (PDB entry 5BRH). This resulted in a median root-mean-square deviation (r.m.s.d.) of 0.75 Å (range 0.43 – 0.78 Å, N=3) for 726 Cα atoms. The dimer of the TcGlcK-CBZ-GlcN complex superimposes onto the dimer (in the asymmetric unit) of the TcGlcK-Glc-ADP complex (PDB entry 2Q2R) with a r.m.s.d. of 1.02 Å for 720 Cα atoms, indicating that their overall folds are similar. A superposition of the crystal structures reveals conserved features within the active site (Figure 5a). The monomeric structures of each TcGlcK-glucosamine analogue complex are generally similar. The r.m.s.d. values, when monomers of each enzyme are superimposed, are as follows: (1) TcGlcK-BENZ-GlcN vs. TcGlcK-D-glucose-ADP complex (r.m.s.d. of 0.92 Å for 352 Cα atoms); (2) TcGlcK-CBZ-GlcN vs. TcGlcK-BENZ-GlcN complex (r.m.s.d. of 0.29 Å for 367 Cα atoms); (3) TcGlcK-HPOP-GlcN vs. TcGlcK-CBZ-GlcN complex (r.m.s.d. of 0.40 Å for 364 Cα atoms); (4) TcGlcK-DBT-GlcN vs. TcGlcK-D-glucose-ADP complex (r.m.s.d. of 1.00 Å for 362 Cα atoms). R.m.s.d. values were determined through the MacPyMOL software (43).</p><p>Simulated annealing omit maps are shown for each glucosamine analogue inhibitor in Figure 5 (panels b – e) with an occupancy of 100% for all ligand atoms. BENZ-GlcN, CBZ-GlcN, and DBT-GlcN prefer the β anomer (orientation of the C1 hydroxyl of the glucose moiety) for active site binding in TcGlcK; however, HPOP-GlcN binds as the α anomer (Figure 5a). TcGlcK has a 4-fold higher binding preference for the β anomer of D-glucose over the α anomer based on a KM value comparison (31). In an attempt to refine the β anomer of HPOP-GlcN (as 100% occupancy for all ligand atoms in the A-chain), negative electron density (contoured at −3.6σ) was observed by the C1 hydroxyl O-atom (in the equatorial position) and positive electron density (contoured at +4.5σ) was found below the C1 O-atom (in the axial position) from an almost final |Fo| − |Fc| map (data not shown). By refining the α anomer of HPOP-GlcN (C1 hydroxyl O-atom in the axial position) instead of the β anomer, the negative and positive electron density peaks are completely absent from the final |Fo| − |Fc| map surrounding the same O-atom.</p><p>Hydroxyls of the glucose moiety from all four glucosamine analogues interact the same way, in that hydrogen bonds are donated to E236, E207, and D131 (interaction with the C4 and C6 hydroxyls); additionally, the C3 hydroxyl of the glucose moiety accepts a hydrogen bond from the side chain NH2 group (Nδ) of N130 (Figure 5a). From the linker component of each inhibitor that stems from the glucose moiety C2-atom, a hydrogen bond is donated from the NH group (of the peptide bond) to E207. The benzene moiety of BENZ-GlcN makes van der Waals contact with P94, M295, P103, S210, and F337 (residue of the other subunit) (Figure 5b). The phenyl tail group of CBZ-GlcN and the phenol moiety tail group of HPOP-GlcN adopt an orientation for favorable van der Waals interactions with F337 and M334 side chains (residues from the other subunit of the protein dimer (Figure 5, panels c and d)) along with P103 and N105 side chains in the active site cavity. The aromatic groups of these inhibitors fit in the outer part of the active site hydrophobic pocket and engage in a π-stacking interaction with the phenyl group of F337, which may give rise to the enhanced inhibitor binding strength compared to BENZ-GlcN that lacks this particular π-stacking interaction. DBT-GlcN has a larger aromatic tail group than CBZ-GlcN or HPOP-GlcN, but the 1,1-dioxobenzo[b]thiophene moiety also fits in the outer part of the active site hydrophobic pocket (Figure 5e) and there are some notable similarities and differences. The similarities include tail group orientation to maximize van der Waals contact with residues P103 and N105 (in the active site region) and residues F337 and M334 of the other subunit. The phenyl group of F337 and the aromatic 1,1-dioxobenzo[b]thiophene group on DBT-GlcN are similarly positioned in a π-stacking orientation. Finally, the side chains of F337 and M334 are moved away from their regular positions as observed in all of the other complexes in order to help situate the bulky 1,1-dioxobenzo[b]thiophene group. The key difference to the TcGlcK-CBZ-GlcN complex includes the bending region angle. The 1,1-dioxobenzo[b]thiophene group is much more bulky compared to the phenyl group on CBZ-GlcN; and, DBT-GlcN is positioned at the interface of the small and large domains. In order for the N105 and P103 side chains to be in ideal van der Waals contact distance (i.e., 3.4 – 3.6 Å), the small domain positions itself in more of a semi-open conformation by 7.7° (average bending angle from the A- and B- chains) from the fully closed conformation structure (PDB entry 2Q2R) as determined by the program DynDom (44). However, the TcGlcK-CBZ-GlcN complex only has an average bending angle of 4.1° from the fully closed conformation, revealing an even smaller semi-open conformation of the enzyme.</p><!><p>The coupled hexokinase – glucose 6-phosphate dehydrogenase enzymatic colorimetric assay that produces NADPH (45) was utilized to obtain catalytic activity measurements and inhibition constants (Ki) for TcGlcK and HsHxKIV. All kinetic measurements are with respect to substrates D-glucose (used at variable concentrations) and ATP (used at a constant concentration). For catalytic activity at room temperature (22 °C) and pH 7.6, we report for His6-TcGlcK a KM of 0.948 ± 0.295 mM for D-glucose, which is essentially identical to the KM of 1.00 mM observed by Cáceres et al. (28) at room temperature and pH 8.5. However, with respect to His6-HsHxKIV, we obtain a 7-fold higher KM of 41.8 ± 8.7 mM for D-glucose compared to the measurement of 6.0 mM exhibited by the non-His-tagged wt-HsHxKIV (46). Our construct for His6-HsHxKIV is based off of a design used in X-ray crystallography experiments to promote protein crystal growth (34) and results in an enzyme having 11 residues truncated from the N-terminus (with an N-terminal 6x His-tag included) along with 8 residues truncated from the C-terminus. This difference in the KM values for D-glucose is likely resultant from the truncations. Table 2 reveals a comparison of our kinase kinetic parameters (KM, kcat, and kcat/KM) to reported literature values. Inhibition constants for glucosamine analogues are provided in Table 3 for both TcGlcK and HsHxKIV. Representative Lineweaver-Burk plots for all inhibitors tested on both kinases are found in the Supplementary Information section. The glucosamine analogues are revealed as competitive inhibitors through X-ray crystallography because they all bind in the active site; however, the analogues showed mixed-type inhibition from their corresponding Lineweaver-Burk plots. We suspect that the mixed-type inhibition may be a consequence of the glucosamine analogue inhibitor preventing the fully closed conformation of the glucokinase being examined. We were not able to measure the Ki for HsHxKIV and DBT-GlcN because DBT-GlcN at concentrations needed for inhibition caused an interference with the colorimetric assay by absorbing light at 340 nm. The strongest TcGlcK inhibitor known to date is CBZ-GlcN, which has a Ki of 0.71 ± 0.05 μM. HPOP-GlcN was observed as a weaker inhibitor by 1.8-fold to TcGlcK compared to CBZ-GlcN, but that is not surprising since HPOP-GlcN binds as the less-preferred α anomer (31). The trend for inhibitor affinity of TcGlcK is as follows: CBZ-GlcN > HPOP-GlcN > DBT-GlcN > BENZ-GlcN. All of the analogues tested had higher inhibitor affinity toward TcGlcK compared to HsHxKIV. The most selective inhibitors of the series were CBZ-GlcN and HPOP-GlcN with selectivity ratios of 245 and 186, respectively, and the least selective inhibitor was BENZ-GlcN with a selectivity value of 12 (Table 3).</p><!><p>The glucosamine analogues are generally in compliance with the Lipinski's Rule of Five (Ro5) (47, 48), from their physiochemical parameters (Table 4). BENZ-GlcN, CBZ-GlcN, and DBT-GlcN each had a Lipinski score of 4, whereas HPOP-GlcN had a score of 3. The criteria evaluated for the Ro5 included the following: (1) molecular weight (MW) ≤ 500 g/mol, (2) CLogP ≤ 5, (3) number of H-bond donors ≤ 5, and (4) number of H-bond acceptors ≤ 10. In addition, the inhibitors were also in agreement with other key physiochemical parameters that are common amongst many successful drugs and were previously described by Luzina and Popov (49), as follows: (1) polar surface area (PSA) < 140 Å2, (2) molar refractivity (MR) in the range of 40 – 130 cm3/mol, and (3) the number of atoms in the molecule (including hydrogen atoms) in the range of 20 – 70 atoms. All of the inhibitors follow these criteria except for DBT-GlcN, which has a slightly higher PSA of 162.62 Å2.</p><p>From an assessment of ligand accessible surface area (ASA) in the active site of TcGlcK and the Ki value trend of the glucosamine analogue inhibitors was that as ligand ASA contact increases from 243 – 365 Å2 (Table 5), CBZ-GlcN had not required the upper limit of the ASA to maintain its relative strong inhibition. This suggests that even though van der Waals contact and glucose-moiety binding interactions served key roles in the inhibitor's binding strength, the π-stacking interaction of the phenyl tail group with the phenyl group of F337 made another key interaction. This is surprising because DBT-GlcN makes similar interactions as CBZ-GlcN and DBT-GlcN has a much higher ligand ASA contact in the TcGlcK active site (365 Å2). It was expected that DBT-GlcN would have stronger inhibition because of more van der Waals contact in the region (i.e., 365 Å2 vs. 275 Å2, see Table 5). The 6-fold weaker inhibition of DBT-GlcN appears to be explained by the larger bending region angle of 7.7° vs. 4.1° (Table 5). When the two domains are more open from the closed conformation, more solvent access in the active site is evident and will give rise to weaker inhibition. This may occur by the destabilization of the hydrophobic pocket, less optimal hydrogen bonding geometries (especially with glucose-moiety binding residues, N130 and D131), and due to an increased dielectric constant within the active site.</p><p>CBZ-GlcN has a reasonably low molecular weight of 313 g/mol and is an efficient lead compound with a ligand efficiency (LE) of 0.38 kcal/mol for TcGlcK inhibition. The compound therefore has the possibility of further optimization to be a better inhibitor. Table 5 shows LEs for the entire set of glucosamine analogue compounds tested in this study. HPOP-GlcN had the second highest LE of 0.35 kcal/mol. DBT-GlcN was not as effective, having an LE of 110 cal/mol less than CBZ-GlcN, primarily due to its MW of 401 g/mol and slightly higher Ki value. The LE of BENZ-GlcN was also as low as DBT-GlcN (even though it has a lower MW of 283 g/mol) with a value of 0.30 kcal/mol, but this is explained by the fact that its benzene ring does not engage in the key π-stacking interaction as the other three inhibitors. Although BENZ-GlcN has a lower LE with respect to the other analogues, it has the possibility for optimization because of its lower MW. The benzene ring of BENZ-GlcN simply adds to the non-hydrogen atoms (N) and thereby lowers the efficiency term.</p><p>In order to advance glucosamine analogues for the purpose of crafting stronger TcGlcK inhibitors, the following parameters should be maintained in the inhibitor design: (1) the sugar moiety should have the glucose stereochemistry and exhibit the β anomer, (2) the inhibitor should maintain a bending region angle in TcGlcK to be less than 5.0° from the closed conformation (as demonstrated by an X-ray crystal structure of a TcGlcK complex), (3) the ligand ASA in the active site should be ≥ 275 Å2 to help maximize van der Waals interactions, (4) a linker having an aromatic tail group, as in the case of CBZ-GlcN, should be installed, (5) the aromatic tail group should not be too bulky (a phenyl group appears to be a good size), and (6) maintain a Lipinski score of 4 for the Ro5. Development of glucosamine analogues for the purpose of medicinal quality compounds against T. cruzi amastigotes requires a focus on the compound BENZ-GlcN, as follows: (1) maintain the set of guidelines for strong TcGlcK inhibitors (see above) except for guideline number 5 and (2) maintain a short linker length (as it is in the case of BENZ-GlcN). By maintaining these parameters and performing structure-activity relationship (SAR) studies, the best inhibitor optimization possibilities will be anticipated.</p><!><p>The in vitro treatment of T. cruzi amastigotes in mammalian NIH-3T3 fibroblasts using glucosamine analogues, such as BENZ-GlcN, CBZ-GlcN, HPOP-GlcN, or DBT-GlcN reveals a trend that does not correlate entirely well with the TcGlcK inhibitor affinity trend. The in vitro IC50 values were measured 4 days after incubation using a glucose-rich DMEM medium. The strongest TcGlcK inhibitor in the study, CBZ-GlcN (Ki = 0.71 μM), was the second most inhibitory for in vitro culture with an IC50 of 48.73 ± 0.69 μM (results not shown). On the other hand, BENZ-GlcN, which was the least potent of the compound series as a TcGlcK inhibitor (Ki = 32 μM) had the highest inhibitory effect against T. cruzi cells with an IC50 of 16.08 ± 0.16 μM (Figure 6a). Benznidazole has a T. cruzi IC50 value of 1.12 ± 0.095 μM (Figure 6b) that compares similarly to the reported values of 1.43 μM (50) and 1.5 μM (51). Amphotericin B has an IC50 value of 0.48 ± 0.0032 μM (results not shown). BENZ-GlcN has a 14-fold weaker inhibition effect and a 34-fold weaker inhibition effect compared to benznidazole and amphotericin B, respectively.</p><p>The least effective inhibitors against parasites were HPOP-GlcN and DBT-GlcN with IC50 values > 50 μM (results are not shown because the highest concentration of inhibitors tested was 50 μM and changes in the percent activity at 50 μM were not significant). The observations for BENZ-GlcN are suggestive that BENZ-GlcN is either a promiscuous inhibitor that binds to more than one T. cruzi target and leads to a T. cruzi cytotoxic effect or that this inhibitor prevents parasite invasion into mammalian host cells. In either case, this is a surprising result because Willson and colleagues determined that for highly similar analogues of BENZ-GlcN, the compounds (intended to bind to the drug-target TbHxK) were incapable of producing significant biological activity through in vitro cultures of T. brucei using a human physiological concentration of glucose (5 mM) in the growth medium with a 3-day incubation period of parasites (22).</p><p>There are various possibilities for why BENZ-GlcN has a good inhibition effect against T. cruzi parasites in comparison to T. brucei parasites. One possibility is that BENZ-GlcN is very effective at inhibiting T. cruzi parasite invasion into host cells. Another possibility is that BENZ-GlcN can be imported into the glycosomes of T. cruzi amastigotes through a compatible glucose transporter having strong affinity and thereby acting to kill the parasites. This hypothesis is supported by the findings of Tetaud et al. (53), in which they observed for N-acetyl-D-glucosamine (GlcNAc) (an analogue of BENZ-GlcN), the T. cruzi epimastigote form expresses a hexose transporter that is 71 times more sensitive for GlcNAc compared to the hexose transporter that is expressed in T. brucei (bloodstream form). In these trypanosomal parasites, inhibition constants for GlcNAc against radiolabeled D-glucose uptake were found to be 0.156 mM and 11.11 mM, respectively. BENZ-GlcN is also modestly potent as a glucokinase inhibitor and it most likely binds strongly to TcHxK. Since TcHxK is found at higher concentrations in T. cruzi cells with respect to TcGlcK (as μg of enzyme per mg of glycosome) (28), if TcHxK is being inhibited by BENZ-GlcN in addition to TcGlcK, glycolysis would be expected to be severely impacted. Possible causes for the poorer inhibition effects observed by HPOP-GlcN and DBT-GlcN may be related to a lack of inhibition of T. cruzi invasion into mammalian host cells, T. cruzi glucose transporters not having sufficient affinity for these glucosamine analogue inhibitors, and/or these inhibitors may be very selective to only TcGlcK and will not act as inhibitors to TcHxK, if they are indeed imported by a glucose transporter. The data is also suggestive that since the DMEM medium has a very high glucose concentration, the glucose most likely is not a major factor for enzyme competition (with respect to the inhibitors) because we observe modestly potent inhibition to sub-micromolar inhibition affinity with BENZ-GlcN and CBZ-GlcN, respectively.</p><!><p>Crystal structures of TcGlcK complexed with glucosamine analogues, such as CBZ-GlcN or DBT-GlcN, reveal that in addition to their glucose moieties binding in the glucose binding site, the aromatic moieties of these compounds bind in a newly identified hydrophobic cavity found at the intersubunit region adjacent to the glucose binding site. The HsHxKIV structure lacks this particular hydrophobic pocket because the enzyme is a monomer. Furthermore, the region is quite distinct between the two enzymes. For example, by superimposing the monomers of TcGlcK (PDB entry 2Q2R) and HsHxKIV (PDB entry 4MLE) (54), a large 14-residue loop (L165 – G178) is found in the small domain of HsHxKIV, which corresponds to a smaller 8-residue loop region (G102 – I110) in TcGlcK (also part of the small domain), see Figure 4. These loops exhibit different conformations based on their flexibility (Figure 7a). In TcGlcK, the G102 – I110 loop includes residues P103 and F104 for which van der Waals contact is made with some of the glucosamine inhibitors (see Figure 5). With this loop being shorter with respect to the L165 – G178 loop of HsHxKIV, steric hindrance is avoided from the aromatic moieties of the glucosamine inhibitors bound in the active site. The L165 – G178 loop of HsHxKIV interacts weakly with a second loop (S281 – Q287) from its large domain by making only van der Waals interactions. However, the corresponding loop segment (K230 – N234) found in the large domain of TcGlcK is highly different in its orientation, by being completely absent from the intersubunit region by the glucose binding site (Figure 7a). Since the S281 – Q287 loop of HsHxKIV is oriented adjacent to the glucose binding site; side chains from residues N283, Q286, and Q287 have the potential to block this region from the tail groups of glucosamine analogues (especially for inhibitors with longer linker groups than the compounds studied herein) unless the small and large domains of the monomer can break into the open conformation (or a semi-open conformation) to allow binding. The side chains from T168 and N166 on HsHxKIV (found on the large L165 – G178 loop) may also prevent glucosamine analogue tail groups from being able to bind in this region. However, this prediction will need to be confirmed by having X-ray crystal structures of glucosamine analogue complexes of HsHxKIV. The reason behind why the TcGlcK K230 – N234 loop is oriented in a completely different manner from the case of HsHxKIV (and from the cases pertaining to the H. sapiens hexokinase isozymes) relates to a β-turn that is present in HsHxKIV (G261 – G264) but absent in TcGlcK (Figure7a). The β-turn (G261 – G264) of HsHxKIV causes the α7 helix to be positioned ideally in order to permit its S281 – Q287 loop to meet at the region adjacent to the glucose binding site. With a missing β-turn in TcGlcK at this segment that is shown by a multiple sequence alignment (Figure 7b) the TcGlcK K230 – N234 loop cannot be positioned at the outer part of the active site (intersubunit region). All of the HsHxK isozymes have this β-turn and have a similar structural comparison at this region (Figure 7c).</p><p>Despite the lack of a crystal structure for TcHxK, secondary structure prediction from primary structure was generated for this enzyme (Figure 7b), to reveal the secondary structure elements in this region. The secondary structure and primary structure both reveal clues for the nature of the outer part of the active site in TcHxK. This prediction shows that TcHxK lacks the β-turn, but it has a 10-residue loop for the corresponding 5-residue TcGlcK K230 – N234 loop. This result appears to suggest that the TcHxK 10-residue loop segment will not interfere with the intersubunit region adjacent to the glucose binding site and our glucosamine inhibitors have the potential to bind to this enzyme. Finally, until the TcHxK structure is solved, there will be the unanswered question as to whether this region has involvement in an interface with another subunit, as is the case of dimeric TcGlcK.</p><!><p>The work described in this study reveals four glucosamine analogue inhibitor complexes of TcGlcK. X-ray crystallography has revealed an important hydrophobic pocket adjacent to the glucose binding site of TcGlcK (at the interface of the protein dimer), but is absent in HsHxK isoenzymes. This hydrophobic pocket contributes significantly to TcGlcK glucosamine analogue inhibitor affinity and selectivity, particularly when the inhibitor tail groups are aromatic and form a π-stacking interaction with residue F337 (of the opposite subunit of the dimer). We propose that the glucosamine analogues of this study will have difficulty binding in HsHxKIV because of residue P153 along with other residue side chains (T168, N166, Q286, and K56) that contribute to steric clash effects. However, these types of steric clashes were not evident in TcGlcK based on having crystal structures of glucosamine analogue complexes. Enzyme inhibition constants were determined for both TcGlcK and HsHxKIV, and from this glucosamine analogue series, CBZ-GlcN was determined to be the strongest glucokinase inhibitor known to date (Ki = 0.71 μM for TcGlcK). The inhibitors also had good selectivity of inhibition over HsHxKIV. Additionally, inhibitors CBZ-GlcN and BENZ-GlcN showed effective in vitro biological activity against T. cruzi amastigotes co-cultured in mammalian NIH-3T3 fibroblasts.</p><p>The trend for TcGlcK inhibitor affinity did not correlate well with the in vitro biological assay growth inhibition results for the small number of compounds tested. This loss in correlation does not permit a firm assignment of TcGlcK as a chemically-validated drug target, one of the objectives we pursued in this study. The biological effectiveness of CBZ-GlcN and BENZ-GlcN may be a consequence of enzyme inhibitor promiscuity; specifically, being able to bind to both TcGlcK and TcHxK that severely diminishes the needed glycolytic and/or PPP metabolic flux. Therefore, we are left with the unsolved challenge of drug-target validation. It is possible, however, to carry out genetic drug-target validation for TcHxK and TcGlcK given the unfortunate case that genetic validation by RNAi in T. cruzi cannot be performed because the organism lacks a functional RNAi pathway (29, 30). Drug-target identification by a method involving in vitro reduced susceptibility analysis by candidate compounds (in this case, the TcGlcK glucosamine analogues) in T. cruzi isolates followed by genome sequencing of the resistant parasites could be implemented (57, 58). However, this method is usually employed in those instances where selective pressure exerted by decades of extensive use of parasite growth inhibitors (on human populations) has resulted in the evolution of resistant strains carrying genetic polymorphisms in the inhibitor target genes (e.g., Plasmodium falciparum has drug resistance in certain strains developed after a ten-year period of usage of chloroquine, quinine, mefloquine, and artesunate (58)). This scenario is less likely to have occurred in the case of TcGlcK glucosamine analogues developed in this study, since natural populations of T. cruzi have never been exposed to these candidate compounds before. The best case for genetic drug-target validation will be until the challenges of RNAi are overcome in T. cruzi.</p>

PubMed Author Manuscript

Integrating hydrogen production with anodic selective oxidation of sulfides over a CoFe layered double hydroxide electrode

Replacing the sluggish oxygen evolution reaction (OER) with oxidation reactions for the synthesis of complex pharmaceutical molecules coupled with enhanced hydrogen evolution reaction (HER) is highly attractive, but it is rarely explored. Here, we report an electrochemical protocol for selective oxidation of sulfides to sulfoxides over a CoFe layered double hydroxide (CoFe-LDH) anode in an aqueous-MeCN electrolyte, coupled with 2-fold promoted cathodic H 2 productivity. This protocol displays high activity (85-96% yields), catalyst stability (10 cycles), and generality (12 examples) in selective sulfide oxidation.We demonstrate its applicability in the synthesis of four important pharmaceutical related sulfoxide compounds with scalability (up to 1.79 g). X-ray spectroscopy investigations reveal that the CoFe-LDH material evolved into amorphous CoFe-oxyhydroxide under catalytic conditions. This work may pave the way towards sustainable organic synthesis of valuable pharmaceuticals coupled with H 2 production.

integrating_hydrogen_production_with_anodic_selective_oxidation_of_sulfides_over_a_cofe_layered_doub

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

<p>Electrochemical water splitting is considered to be a promising hydrogen production approach to deal with the increasing global energy demand and environmental problems associated with fossil fuel utilization. 1 Water electrolysis involves two halfreactions, including H 2 and O 2 evolution reactions (HER and OER). 2 The overall reaction rate is oen restricted by the anodic OER because of its more sluggish kinetics. 2c As a result, much higher overpotential is needed for the OER to match the H 2 production rate, thereby undermining the overall energy conversion efficiency. 3 Although advanced non-noble-metal OER electrocatalysts have been developed, accomplishing an OER efficiency comparable with that of the HER still remains a challenge. 4 In addition, the economic value of O 2 is far inferior to that of H 2 . Recently, electrooxidation of low-cost organic agents has emerged as an alternative strategy to replace the OER, achieving lower overpotential for total water splitting and at the same time producing high-value chemicals. 5 Electrooxidation has been exploited for converting biomass-derived platform chemicals, e.g., ethanol, glycerol, and 5-hydroxymethyfurfural (HMF). 6 Despite the advanced concept of co-production of valuable chemicals and H 2 , only a handful of simple molecules were studied. 6f,g The anodic synthesis of functional and complex organic molecules, especially pharmaceuticals with medicinal signicance, has rarely been explored.</p><p>In the family of pharmaceuticals, organosulfur compounds play essential roles in many biological processes linked with human disease therapy. 7 For example, sulforaphane offers efficient chemoprotection against various cancers (e.g. prostate, lung, breast, and colon cancer). 7g Direct oxidation of their sulde precursors is a straightforward and atom-economical synthetic method, and thus has been widely applied for the synthesis of sulfoxides. 8 However, current synthetic methods oen require homogeneous catalysts (e.g. Mn-and Fe-based catalysts) and strong oxidizing agents such as H 2 O 2 . 9 Additional oxidants such as peroxy acids, prevalent iodine reagents or oxone are also needed. 10 In 2012, He and co-workers reported a catalyst-free protocol for the selective oxidation of suldes with an inorganic oxidant oxone in ethanol. 11 However, the utilization of oxone may result in purication issues and environmental contamination. More importantly, over-oxidation of sulfoxide to sulfone or co-oxidation of other functional groups in the sulfoxide molecule would occur and generate medically useless products. Recently, Xu and co-workers reported an efficient heterogeneous polyoxovanadate catalyst for sulde oxidation; nevertheless the reaction was assisted by a strong oxidant (tert-butylhydroperoxide (TBHP)) at elevated temperatures. 12 Jiang's group reported a photocatalysis strategy to achieve sulde selective oxidation driven by visible light under mild conditions, but the use of homogeneous catalysts (UO 2 (OAc) 2 -$2H 2 O) would make separation difficult. 12 Therefore, the selective oxidation of suldes using heterogeneous catalysts under ambient conditions remains challenging.</p><p>Electrochemical oxidation by using heterogeneous electrocatalysts shows promise to achieve selective sulde oxidation under mild conditions while avoiding separation issues. Recently, an environmentally benign electrochemical oxidation of suldes was reported by Laudadio and colleagues, who evaluated the electrocatalysis with continuous reactors that shows great promise for practical applications. 13 Non-noble metal-based layered double hydroxides (LDHs), a class of inorganic layered materials with unique 2D structures in which diand tri-valence cations are dispersed within the layers at the atomic level, 14 have emerged as efficient electrocatalysts for the OER 15 and anodic organic agents for electrooxidation, for example converting HMF to value-added 2,5-furandicarboxylic acid (FDCA). 16 However, there is no report of developing LDHs as the electrocatalyst for sulde selective oxidation with the goal of synthesis of sulfoxide-containing pharmaceuticals.</p><p>In this regard, we report an electrochemical approach for selective oxidation of suldes to replace the OER over CoFe layered double hydroxides supported on carbon cloth (CoFe-LDH/CC) as the anode in an aqueous-MeCN electrolyte, coupled with 2-fold promoted cathodic H 2 productivity (Fig. 1a). A variety of aryl, heteroaryl and alkyl suldes at the CoFe-LDH/ CC anode can be selectively converted into the corresponding sulfoxides with 85-96% yields under ambient conditions (Fig. 1b). Importantly, this method was successfully extended to the synthesis of complex pharmaceutical compounds with sulfoxide moieties from their sulde precursors in good yields, including ricobendazole (78%), omeprazole (70%), sulindac (63%) and amino acid methionine (89%). Moreover, the catalyst CoFe-LDH/CC was used in the gram-scale synthesis of diphenyl sulfoxide with 83% yield (up to 1.63 g) and amino acid methionine with 86% yield (up to 1.79 g). Preliminary investigations suggest that the in situ formed amorphous metal oxyhydroxide acts as the active species for selective oxidation of suldes to sulfoxides via a radical process.</p><!><p>Initially, CoFe-LDH/CC was fabricated by a facile electrodeposition method (see the ESI † for details). 17 As illustrated in Fig. 2a, Co and Fe nitrate precursors were converted to CoFe hydroxide by electrolysis with the formation of a LDH nanoarray grown on a carbon cloth cathode. As shown in the scanning electron microscope (SEM) and transmission electron microscope (TEM) images (Fig. 2b and c, ESI, Fig. S1a †), the crosslinked LDH nanosheets are homogeneously grown on CC with an average thickness of 9.4 nm (ESI, Fig. S1b and c †). Meanwhile, the atomic Co/Fe ratio is 3.01 as determined by inductively coupled plasma-atomic emission spectrometry (ICP-AES, ESI, Table S1 †), which is consistent with the stoichiometry of the introduced values. The energy dispersive spectrometry (EDS) mapping technique reveals the uniform dispersion of Co and Fe species in the nanosheet (Fig. 2d). A high-resolution TEM image (HRTEM, Fig. 2e) shows that the interplanar spacing is approximately 0.26 nm, which is assigned to the (012) plane of the CoFe-LDH phase. 18 The characteristic ( 003), ( 006), (012), and (110) planes of the LDH structure were also observed in the selected area electron diffraction (SAED) pattern of the nanosheet (Fig. 2f). Consistently, these diffractions were observed in the X-ray diffraction (XRD) pattern of the CoFe-LDH/CC (Fig. 2g).</p><p>The selective electrooxidation of diphenyl sulde (1a) was chosen as a model reaction to evaluate the activity of CoFe-LDH/ CC and to establish the optimal reaction conditions. In the reaction, diphenyl sulfoxide (2a) was the desired product, and diphenyl sulfone (3a) was the over-oxidized byproduct. To investigate the effect of the applied voltage, the potential was increased from 2.0 to 3.5 V vs. Ag/AgCl, resulting in the yield of over-oxidized byproduct 3a increasing from trace to 30%. This result suggests that the relatively low potential was a key factor for inhibiting the formation of over-oxidized byproducts, in agreement with the previous literature. 13 Aer extensive optimization of the electrolyte, solvent, cathode and atmosphere (ESI, Table S2 †), we found that upon using MeCN/H 2 O (1 : 1 v/v) as the mixed solvent and nBu 4 NH 2 PO 4 as the electrolyte, the desired product 2a was obtained in 85% yield and 85% faradaic efficiency (FE) in 2 h at room temperature (entry 1 of Table 1), which represents a high performance compared to existing homo-and heterogeneous catalysts (ESI, Table S3 †). Specically, H 2 O was initially tested as the solvent for its wide use in the OER and anodic oxidations, but moderate conversion and yield of 2a were obtained, accompanied by overoxidation (entry 2, Table 1). Replacing it by MeCN or other organic solvents showed even lower activities (entry 3 in Table 1, entries 1-4 in the ESI, Table S2 †). We found that using a mixed solvent of MeCN and H 2 O could furnish signicantly higher efficacy, which is likely due to MeCN showing higher conductivity and capacity to dissolve the electrolyte, substrate, and reactants compared with other employed solvents, which may lead to catalytic enhancement (entries 5-6 in the ESI, Table S2 †). 19 The use of electrolytes other than nBu 4 NH 2 PO 4 showed lower activity (entries 7-9 in the ESI, Table S2 †), which is due to the promotion effect of nBu 4 NH 2 PO 4 for the oxidation reaction. 12b In addition, the yield of 2a didn't show obvious change when the reaction atmosphere was varied from air to N 2 (entry 4 in Table 1), indicating that H 2 O in the electrolyte could likely be the source of oxygen in 2a. It should be noted that under our reaction conditions in air, it remains possible that oxygen in the product can be from the air. A blank experiment using CC as the anode showed much lower conversion for sulde oxidation, suggesting CoFe-LDH/CC was essential for catalytic activity of sulde oxidation (entry 5 in Table 1). Finally, no reaction occurred in the absence of electric current, demonstrating the nature of electrocatalysis (entry 6 in Table 1). Pt ions were not observed in the electrolyte aer electrooxidation by using an Inductive Coupled Plasma Emission Spectrometer (ICP), indicating that Pt was not leached into the electrolyte. According to the conversion of 1a and total yields of products 2a (or with 3a), there remained other byproducts while their structures can't be recognized as no other signals were observed by crude NMR analysis.</p><p>In the following catalytic studies, the catalytic reactions were examined over the as-synthesized CoFe-LDH/CC under the optimized conditions. Fig. 3a shows the linear sweep voltammetry (LSV) curves in the presence and absence of 1a over the CoFe-LDH/CC anode. Aer adding 1a, the current density signicantly increases, and the required potential for gaining current density of 5 mA cm À2 dramatically decreased from 1.90 V (without 1a) to 1.39 V (with 1a). The desired product 2a increased with the time, and to our delight, the overoxidized product (sulfone 3a) was rarely observed (<5% yield, Fig. 3b). More importantly, 2a can be obtained on a gram scale with 83% yield (up to 1.63 g) (for details, see Experimental procedures in the ESI †). We then assembled a two-electrode setup in which CoFe-LDH/CC was used as the anode for 1a oxidation and Pt as the cathode for the HER. Impressively, it shows two-fold higher activity for H 2 production compared with the one without 1a that follows the OER process (Fig. 3c), indicative of enhanced energy efficiency. To a certain extent, the organic solvent MeCN has some slight effect on the faradaic efficiency of H 2 (90%), lower than the faradaic efficiency of 96% in aqueous electrolyte. Aer ten-cycle runs, no apparent decrease of activity was observed (Fig. 3d), indicative of promising catalytic stability.</p><p>The structural stability of the used CoFe-LDH/CC anode was also demonstrated (ESI, Fig. S2 †). Subsequently, the generality of the electrocatalytic selective oxidation using CoFe-LDH/CC was studied by using 12 examples (Fig. 4). In general, the suldes and their derivatives with functional groups including -H, -F, -Cl and -Br worked very well to produce sulfoxides (2a-2g) under the optimized conditions. The reaction efficiency was not affected by steric hindrance (2d, 2f and 2g). A multi-substituted sulfoxide (2h) was also produced efficiently. Aliphatic and benzyl groups could be tolerated, and target products 2i and 2j were furnished in 91% and 86% yields, respectively. Notably, nitrogen-containing hetero-aromatics can be selectively oxidized to form the corresponding sulfoxide (2k) without N-oxide formation. Finally, alkyl sulfoxide (2l) was well tolerated under our reaction conditions, given that it is oen sensitive to oxygenation conditions. 13 The late-stage oxygenation of complex molecules to construct sulfoxides is oen accompanied by the existence of various functional groups, and maintaining these functional groups is very important but remains challenging. By using our selective method, we expect that complex pharmaceuticals can be precisely modied at the last stage, in which sulde is oxidized to sulfoxide while the side reactions on other functional groups are avoided. As a proof-of-concept, we focused on four important drug molecules with sulfoxide moieties (Fig. 5). Specically, the corresponding suldes were synthesized rst by the reported procedures (for Experimental procedures see the ESI †), which was followed by electrochemical oxidation to the targeted sulfoxides by using CoFe-LDH/CC as the anode under the optimized conditions. The catalytic results show that, ricobendazole 20 (2m), an effective drug for a potential anticancer agent, was obtained by selective oxidation in 78% yield. Omeprazole 20 (2n), a well-known drug for treating gastroesophageal reux, was also obtained in 70% yield. Sulindac 20 (2o), another anticancer reagent, was generated with 63% yield even in the presence of oxygen-sensitive sequential conjugated alkenes. Amino acid methionine (2p), 20 a biologically relevant compound, could be efficiently synthesized in good yield. Moreover, this electrochemical protocol could be used in the gram-scale synthesis of sulfoxides (2p) with 1.79 g and high yield of 86% was still obtained. Therefore, this protocol potentially enables access to electrochemical synthesis of valuable pharmaceutical related sulfoxides without external oxidants. In addition, the CoFe-LDH/CC electrocatalyst potentially overcomes disadvantages, such as catalyst reusability, which are oen encountered by using homogeneous catalysts in conventional methods.</p><p>In order to understand the reaction mechanism, a set of control experiments were carried out (Fig. 6). When the standard reaction was performed in the presence of 2,2,6,6tetramethyl-1-piperidinyloxy (TEMPO) or butylated hydroxytoluene (BHT) as the radical scavengers, 21 the reaction was completely inhibited (Fig. 6a). According to a study on oxygenation of suldes, 12b,21 the sulde radical (Ph 2 Sc) or persulfoxide radical (Ph 2 SOOc) was generally present in the process. The EPR experiments were designed to detect radical intermediates by adding the radical trapping agent DMPO (5,5-dimethyl-1pyrroline N-oxide). No obvious signals could be observed in the absence of 1a (Fig. 6b, black line). Under the standard conditions, a typical signal of the DMPO-radical adduct was determined by the EPR during reaction (Fig. 6b, red line). We speculated that it might be a sulde radical (Ph 2 Sc) or persulfoxide radical (Ph 2 SOOc) generated in the reaction with the DMPO adduct. 22 These ndings suggest that the electrooxidation reaction over CoFe-LDH/CC may involve a radical process. To conrm the source of the oxygen atom in the formed product, oxidation of 1a was carried out in MeCN-H 2 18 O electrolyte, and the GC mass spectrum analysis shows that one 18 O atom was introduced into 2a (Fig. 6c). This result suggests that water could likely supply oxygen for the sulfoxide oxidation under our reaction conditions.</p><p>To gain an insight into the active species in this catalytic system, the structural evolution of the CoFe-LDH/CC anode was in-depth investigated using XRD, TEM, SAED, X-ray photoelectron spectroscopy (XPS), and X-ray absorption ne structure spectroscopy (XAFS). 23 As shown in Fig. 7a, the XRD pattern of the used CoFe-LDH/CC (1 st recycled CoFe-LDH/CC) shows only two broad diffraction peaks at 24.5 and 44.0 , which are attributed to carbon cloth, indicating that CoFe-LDH undergoes dramatic reconstruction to form an amorphous structure under reaction conditions. Consistently, the SAED further conrmed the amorphous nature of the used CoFe-LDH (ESI, Fig. S3 †).</p><p>We then investigated the structure of CoFe-LDH before and aer reaction using XPS and XAFS techniques. As shown in Fig. 7b, the XPS signal of Co2p 3/2 in the starting material shows a typical satellite peak at 787.7 eV and a peak at 783.0 eV attributed to Co II in LDHs. Aer the reaction, the intensity of the peak of Co 2+ decreased with the appearance of a new peak at 780.1 eV which could be attributed to Co 3+ , suggesting the oxidation of Co 2+ to Co 3+ in the electrooxidation process. 24 Concomitantly, the deprotonation of hydroxyl on CoFe-LDH occurred, giving rise to a peak at 529.1 eV in the O1s XPS spectrum of the used CoFe-LDH that can be attributed to the typical M-O-M species in metal oxyhydroxides (Fig. 7c). 24,25 Meanwhile, the Fe2p XPS spectra reveal a stable Fe III state in the fresh and the used CoFe-LDH (ESI, Fig. S4 †). In addition, X-ray absorption near-edge structure (XANES) analysis in Fig. 7d shows that the Co K edge prole of the used CoFe-LDH/CC moved to a higher energy and next to that of the crystalline CoFe-oxyhydroxide reference (c-CoFeOOH, obtained from CoFe-LDH/CC by anodization in 1 M KOH, ESI, Fig. S5 †), indicating that some Co II ions in starting CoFe-LDH/CC were oxidized to Co III during the reaction, which is consistent with the XPS data. In addition, the extended XAFS spectra reveal that the length of the rst coordination shell C-O (2.03 Å) and the second coordination shell Co-Co(Fe) (3.13 Å) in the initial CoFe-LDH was decreased to 1.88 Å and 2.83 Å, respectively (Fig. 7e), which are identical to those of c-CoFeOOH, reecting its metal oxyhydroxide nature. This structural transformation during electrooxidation was induced by oxidation of Co II ions to Co III . 26 Notably, the intensities of both Co-O and Co-Co(Fe) in the used CoFe-LDH are lower than those of fresh CoFe-LDH and c-CoFeOOH in the spectra of wavelet transformed EXAFS (ESI, Fig. S6 †), owing to its amorphous structure. Moreover, the EXAFS analysis show that the as-formed oxyhydroxide structure was stable aer four cycle reactions (Fig. 7f). We can conclude that the starting CoFe-LDH/CC material evolved into the corresponding amorphous oxyhydroxide during electrooxidation as the real active species. In addition, SCN À , oen adopted to poison metal sites, 27 was used as an indicator for active sites. As shown in Fig. S7, † a remarkable decrease of 2a yield was observed by adding SCN À , thus indicating that the metals in CoFe-LDH contribute to the catalytic activity. As an electrochemical reaction takes place at the interface between the catalyst and electrolyte, we evaluated the stability of CoFe-LDH for 7 h using chronoamperometry (CA) at 2.0 V versus the Ag/ AgCl electrode (Fig. S8 †). A current densityof $11 mA cm À2 was retained, indicating the stability of the catalyst over long reaction times.</p><p>On the basis of our results and the previous literature, 6d,28 a reaction mechanism of this catalytic system is proposed in Fig. 8. In the neutral aqueous-organic electrolyte (MeCN as the organic phase), water molecules undergo Volmer and Heyrovsky Because of the modulated electronic structure, the redox behavior of Co cations could probably be manipulated by Fe incorporation, leading to the active phases in sulde electrooxidation with higher activity, which is in agreement with recent reports. 29 This amorphous oxyhydroxide might act as the active phase for selective sulde oxidation using water as the oxygen source.</p><!><p>In summary, we demonstrate an efficient electrochemical method for selective electrooxidation of suldes with promoted HER over a CoFe-LDH/CC electrode. A variety of sulde model compounds were selectively electrooxidized to sulfoxides in high yields (85-96%) at the anode under ambient conditions, showing high catalytic performance compared to the existing homo-and heterogeneous catalysts. Importantly, this method can be applied to the synthesis of four complex pharmaceuticals with sulfoxide moieties from their sulde precursors in good yields, including ricobendazole (78%), omeprazole (70%), sulindac (63%) and amino acid methionine (89%). Moreover, CoFe-LDH/CC was used in the gram-scale synthesis of diphenyl sulfoxide with 83% yield (up to 1.63 g) and amino acid methionine with 86% yield (up to 1.79 g). In addition, the CoFe-LDH/ CC electrocatalyst could be reused, maintaining the yield for more than ten cycles. Mechanistic studies indicate that the reaction pathway proceeds through a radical process, and the in situ formed CoFe-oxyhydroxide may serve as the active species for the sulde oxidation. The efficiency of this reaction system may pave the way for the electrochemical synthesis of valuable organic molecules by using heterogeneous catalysts without external oxidants under ambient conditions.</p><!><p>There are no conicts to declare.</p>

Royal Society of Chemistry (RSC)

Discovery of Phenylalanine Derivatives as Potent HIV-1 Capsid Inhibitors from Click Chemistry-based Compound Library

The HIV-1 capsid (CA) protein plays essential roles in both early and late stages of HIV-1 replication and is considered an important, clinically unexploited therapeutic target. As such, small drug-like molecules that inhibit this critical HIV-1 protein have become a priority for several groups. Therefore, in this study we explore small molecule targeting of the CA protein, and in particular a very attractive inter-protomer pocket. We report the design, parallel synthesis, and anti-HIV-1 activity evaluation of a series of novel phenylalanine derivatives as HIV-1 CA protein inhibitors synthesized via Cu(I)-catalyzed alkyne-azide 1,3-dipolar cycloaddition (CuAAC) reaction. We demonstrate robust inhibitory activity over a range of potencies against the HIV-1 NL4-3 reference strain. In particular, compound 13m exhibited the greatest potency and lowest toxicity within this new series with an EC50 value of 4.33 \xce\xbcM and CC50 value of >57.74 \xce\xbcM (SI > 13.33). These values are very similar to the lead compound PF-74 (EC50= 5.95 \xce\xbcM, CC50 > 70.50 \xce\xbcM, SI > 11.85) in our assay, despite significant structural difference. Furthermore, we demonstrate via surface plasmon resonance (SPR) binding assays that 13m interacts robustly with recombinant HIV-1 CA and exhibits antiviral activity in both the early and late stages of HIV-1 replication. Overall, the novel parallel synthesis and structure-activity relationships (SARs) identified within this study set the foundation for further rational optimization and discovery of CA-targeting compounds with improved potency.

discovery_of_phenylalanine_derivatives_as_potent_hiv-1_capsid_inhibitors_from_click_chemistry-based_

Introduction<!>Chemistry<!>Anti-HIV activity evaluation in MT-4 cells<!>Binding to HIV-1 CA protein<!>Determination of the action stage of 13m<!>Molecular Dynamics (MD) simulation<!>Conclusions<!>Chemistry<!>Tert-butyl(S)-(1-((4-methoxyphenyl)(methyl)amino)-1-oxo-3-phenylpropan-2-yl)carba mate (2)<!>(S)-2-amino-N-(4-methoxyphenyl)-N-methyl-3-phenylpropanamide (3)<!>(S)-N-(1-((4-methoxyphenyl)(methyl)amino)-1-oxo-3-phenylpropan-2-yl)propiolamide (4)<!>General procedure for the synthesis of azide substituents (6 and 10)<!>General procedure for the synthesis of azide substituents (8)<!>General procedure for the synthesis of azide substituents (11 and 12)<!>General procedure for the synthesis of target compounds (13a-13t)<!>(S)-N-(1-((4-methoxyphenyl)(methyl)amino)-1-oxo-3-phenylpropan-2-yl)-1-(2-methyl benzyl)-1H-1,2,3-triazole-4-carboxamide (13a).<!>(S)-N-(1-((4-methoxyphenyl)(methyl)amino)-1-oxo-3-phenylpropan-2-yl)-1-(3-methyl benzyl)-1H-1,2,3-triazole-4-carboxamide (13b).<!>(S)-N-(1-((4-methoxyphenyl)(methyl)amino)-1-oxo-3-phenylpropan-2-yl)-1-(4-methyl benzyl)-1H-1,2,3-triazole-4-carboxamide (13c).<!>(S)-1-(2-chlorobenzyl)-N-(1-((4-methoxyphenyl)(methyl)amino)-1-oxo-3-phenylpropa n-2-yl)-1H-1,2,3-triazole-4-carboxamide (13d).<!>(S)-1-(3-chlorobenzyl)-N-(1-((4-methoxyphenyl)(methyl)amino)-1-oxo-3-phenylpropa n-2-yl)-1H-1,2,3-triazole-4-carboxamide (13e).<!>(S)-1-(4-chlorobenzyl)-N-(1-((4-methoxyphenyl)(methyl)amino)-1-oxo-3-phenylpropa n-2-yl)-1H-1,2,3-triazole-4-carboxamide (13f).<!>(S)-1-(2-cyanobenzyl)-N-(1-((4-methoxyphenyl)(methyl)amino)-1-oxo-3-phenylpropan-2-yl)-1H-1,2,3-triazole-4-carboxamide (13g).<!>(S)-1-(3-cyanobenzyl)-N-(1-((4-methoxyphenyl)(methyl)amino)-1-oxo-3-phenylpropan-2-yl)-1H-1,2,3-triazole-4-carboxamide (13h).<!>(S)-1-(4-cyanobenzyl)-N-(1-((4-methoxyphenyl)(methyl)amino)-1-oxo-3-phenylpropan-2-yl)-1H-1,2,3-triazole-4-carboxamide (13i).<!>(S)-N-(1-((4-methoxyphenyl)(methyl)amino)-1-oxo-3-phenylpropan-2-yl)-1-(2-nitrobe nzyl)-1H-1,2,3-triazole-4-carboxamide (13j).<!>(S)-N-(1-((4-methoxyphenyl)(methyl)amino)-1-oxo-3-phenylpropan-2-yl)-1-(3-nitrobe nzyl)-1H-1,2,3-triazole-4-carboxamide (13k).<!>(S)-N-(1-((4-methoxyphenyl)(methyl)amino)-1-oxo-3-phenylpropan-2-yl)-1-(4-nitrobe nzyl)-1H-1,2,3-triazole-4-carboxamide (13l).<!>(S)-N-(1-((4-methoxyphenyl)(methyl)amino)-1-oxo-3-phenylpropan-2-yl)-1-(naphthal en-2-ylmethyl)-1H-1,2,3-triazole-4-carboxamide (13m).<!>(S)-N-(1-((4-methoxyphenyl)(methyl)amino)-1-oxo-3-phenylpropan-2-yl)-1-(naphthal en-l-ylmethyl)-1H-1,2,3-triazole-4-carboxamide (13n).<!>(S)-1-(benzo[d][1,3]dioxol-5-yl)-N-(1-((4-methoxyphenyl)(methyl)amino)-1-oxo-3-phenylpropan-2-yl)-1H-1,2,3-triazole-4-carboxamide (13o).<!>Methyl(S)-3-(4-((1-((4-methoxyphenyl)(methyl)amino)-1-oxo-3-phenylpropan-2-yl)carbamoyl)-1H-1,2,3-triazol-1-yl)benzoate (13p).<!>(S)-N-(1-((4-methoxyphenyl)(methyl)amino)-1-oxo-3-phenylpropan-2-yl)-1-((phenylth io)methyl)-1H-1,2,3-triazole-4-carboxamide (13q).<!>N-((S)-1-((4-methoxyphenyl)(methyl)amino)-1-oxo-3-phenylpropan-2-yl)-1-((phenylsu lfinyl)methyl)-1H-1,2,3-triazole-4-carboxamide (13r).<!>(S)-N-(1-((4-methoxyphenyl)(methyl)amino)-1-oxo-3-phenylpropan-2-yl)-1-((phenylsu lfonyl)methyl)-1H-1,2,3-triazole-4-carboxamide (13s).<!>1-(2-(hydroxymethyl)-5-(5-methyl-2,4-dioxo-3,4-dihydropyrimidin-1(2H)-yl)tetrahydrofuran-3-yl)-N-((S)-1-((4-methoxyphenyl)(methyl)amino)-1-oxo-3-phenylpropan-2-yl)-1H-1,2,3-triazole-4-carboxamide (13t).<!>Assessment of Inhibitory Activity on HIV-1 Replication in MT-4 Cells<!>Cytotoxicity Assay[37]<!>Binding to CA Proteins Analysis via Surface Plasmon Resonance (SPR)<!>Initial Structure Preparation<!>Molecular Dynamics Simulation Production<!>Clustering<!>Cells<!>Production of pseudotyped viruses<!>Single-round infection assay<!>Viral late-stage infection assay

<p>Acquired immune deficiency syndrome (AIDS), primarily caused by human immunodeficiency virus type 1 (HIV-1), remains a global problem. Although considerable progress has been made with regards to improving the quality and length of life of patients living with HIV with the application of combinatorial antiretroviral therapy (cART) regimen, current therapies still suffer from adverse side effects and the ever-present problems of natural and emergent drug resistance[1, 2]. Accordingly, there is a continuing need for new anti-HIV drugs, especially those acting on novel targets with new scaffolds and mechanisms of action.</p><p>The HIV-1 capsid (CA) protein plays crucial roles in both early (uncoating, nuclear entry, reverse transcription, integration, etc.) and late stages (maturation and assembly) of the HIV-1 life cycle [3, 4]. In the early stage, the fusion of the virus with target cell membranes triggers the disassembly or uncoating of CA protein, which is tightly associated with the reverse transcription and synthesis of viral cDNA [5, 6]. Although viral uncoating is not completely understood, the correct spatiotemporal disassembly of the capsid during the initial stage of infection is necessary for the completion of reverse transcription and, hence, infectivity. Moreover, in the late stage of HIV-1 replication, CA protein assembly and maturation are essential for the formation and release of infectious viral particles. Extensive mutational analysis has been performed on CA; analysis that has revealed an extreme genetic fragility, with most changes to the capsid being detrimental for the above processes and the fitness of the virus [7, 8]. Therefore, the HIV-1 CA protein has emerged as an attractive antiviral target for the development of small-molecular inhibitors [9-12].</p><p>To date, a series of structurally diverse small-molecular compounds have been identified as HIV-1 CA inhibitors [13-20]. Among them, PF-74 (Fig.1a) is the most widely studied small molecule targeting the HIV-1 CA protein (Fig.1a). PF-74 was developed by Pfizer, USA, and has been demonstrated to inhibit the replication of the majority of isolates at low to sub-micromolar concentrations [11, 13, 21-23]. However, the relatively low inhibitory activity and poor drug-like qualities, most critically extremely poor metabolic stability [13] indicate the urgent needs for further structural optimization of PF-74. Recently, the binding mode of PF-74 in the inter-protomer binding pocket in the CA protein hexamer has been described by several groups in crystallographic studies using native and disulphide stabilized constructs [24-26]. The availability of these structures greatly facilitates the structure-based design of novel compounds based upon the PF-74 scaffold. PF-74 consists of a phenylalanine core (red part, Fig.1a), an indole substituent (pink part, Fig.1a) and a linker between them (blue part, Fig.1a). The interprotomer pocket in which PF-74 binds is defined by four α-helices (H3, H4, H5, and H7) in the NTD interface of one protomer; the phenylalanine core forms interactions with Asn57, with Ile73, Ala105, Tyr130 and Thr107 (Fig.lb). The methylindole ring of PF-74 is oriented towards the CTD interface of the adjacent subunit within the same hexametric ring and interacts with Arg173, Gln63, and Lys182 (Fig.1b) [24-26]. Of particular note, the plasticity of the CTD-NTD interface provides additional space for further modification of the indole substituent and linker of PF-74, as exemplified by compounds I and II with improved potency (Fig.1). Interestingly, in these compounds, the methoxy moiety was frequently observed in the aniline substituent of phenylalanine core (Fig.1) [27, 28].</p><p>Given the attractiveness of the CA protein as a target, and the wealth of structural and mechanistic information available, we chose PF-74 to focus our optimization efforts upon. Specifically, in this study, we chose to explore modification of methylindole moiety as this region has been demonstrated to be quite tolerant to changes [13]. However, unlike previous studies, we demonstrate the successful implementation of a parallel synthesis strategy using the facile Cu(I)-catalyzed alkyne-azide 1,3-dipolar cycloaddition (CuAAC) reaction (Fig.1) [29-32]. This strategy, installs a 1,2,3-triazole unit as a linker in the newly designed compounds, further diversifying the compounds and, most likely, improving solubility (Fig. 1). We chose not to modulate the phenylalanine core region of PF-74 and its reported analogues but chose to include the methoxy-bearing aniline substituent present within the high potency analogues (PF-74).</p><p>Therefore, in this novel study, we report the parallel synthesis and antiviral activity of 20 newly designed phenylalanine derivatives. Furthermore, we demonstrate via surface plasmon resonance (SPR) binding assays that representative compounds from this study interact robustly with recombinant HIV-1 CA and exhibit antiviral activity in both the early and late stages of HIV-1 replication. Overall, the novel parallel synthesis and structure-activity relationships (SARs) identified within this study set the foundation for further rational optimization and discovery of CA-targeting compounds with improved potency.</p><!><p>The straightforward synthetic route towards the target compounds was depicted in Scheme 1. (Tert-butoxycarbonyl)-L-phenylalanine (1) was selected as the starting material. It was reacted with 4-methoxy-N-methylaniline, followed by removing the Boc protection to produce the free amine (3). Then, 3 was treated with propiolic acid to yield the key intermediate 4. Meanwhile, the substituted benzyl bromide (5) or chloride (9) were converted to azide 6 or 10 by nucleophilic substitution with NaN3, respectively. Further oxidation of 10 in the presence of 3-chloroperbenzoic acid generated compounds 11 and 12. Substituted phenylboric acids 7 were reacted with NaN3 by the catalytic CuSO4.5H2O to provide the corresponding azides 8. Finally, the target products (13) were obtained by CuAAC reaction between the azide substituents (6, 8, 11 or 12) and key intermediate 4. The newly synthesized compounds were characterized by MS and NMR spectra.</p><!><p>Having demonstrated the utility of our new parallel synthesis scheme, we next sought to test the newly synthesized compounds for in vitro anti-HIV-1 activity in a multicycle assay using fully infectious HIV-1 NL4-3 virus and MT-4 target cells. PF-74 was included as an in-line control to allow for direct comparison with the new compounds. Finally, the toxicity of PF-74 and the new compounds towards the MT-4 cells was also assessed. Table 1 shows the anti-HIV potency (EC50, as measured by a luciferase gene expression assay [33]), cytotoxicity (CC50) as well as selectivity index (SI, the ratio of CC50/EC50) for each of the compounds and control.</p><p>As shown in the Table 1, it can be observed that some of newly synthesized compounds exhibited from moderate to excellent activity against HIV-1 NL4-3 virus with EC50 values ranging from 14.93 μM to 4.33 μM and SI values between 1.34 and 13.33. Among them, 13m (EC50 = 4.33 ± 0.83 μM, CC50> 57.74 μM), 13p (EC50 = 6.91 ± 2.43 μM, CC50 >19.47 μM) and 13r (EC50 = 6.65 ± 1.47 μM, CC50> 19.32 μM) turned out to be the most potent HIV-1 inhibitors, which were equipotent with PF-74 (EC50 = 5.95 ± 1.32 μM, CC50 > 70.50 μM). In particular, 13m also exhibited an equivalent SI value (13.33) comparable to PF-74 (SI: 11.85).</p><p>The exploratory SARs analysis was derived from the results. Firstly, our attention was focused on the SAR of the substituted benzyls (13a-13n). When making a comparison between 13m and 13n with a terminal β- and α-substituted naphthalene group in their structures, respectively, 13m is far more active than 13n in inhibiting HIV-1 NL4-3, indicating that β-substituted naphthalene might form a better interaction with the CTD of CA protein. Similarly, variation of the position of identical substituent in the benzyl (from the ortho-position to the para-position) resulted in different inhibitory activity (13a (2-Me) > 13b (3-Me) and 13c (4-Me); 13i (4-CN) ≈ 13h (3-CN) > 13g (2-CN); 13j (2-NO2) > 13k (3-NO2) and 131 (4-NO2)), with the exception of compounds that were endowed with chlorinated benzyl (13d/13e/13f). These results indicated that benzyl with different positions of the same substituent would have a great influence on the inhibitory activity against HIV-1 NL4-3 virus. In addition, it was noteworthy that different substituents at the same position demonstrated significant effects on the activity results, as suggested by comparison of 13a/13d/13g/13j, 13b/13e/13h/13k and 13c/13f/13i/13l.</p><p>Next, we turned our attention to the SAR of synthesized compounds with substituted benzenes (13o/13p). Detailed comparison of the activities of 13o with 13p suggested that one more flexible substituent at the benzene (13p) might strengthen the interaction with the CTD binding site compared to substituent bearing with rigid construction (13o). Further optimization of phenylalanine derivatives was focused on incorporating a group including S atom or its oxidation groups to the linker. Comparing the anti-HIV-1 (NL4-3) activity between 13q, 13r and 13s reveals that had certain influence on the antiviral activity (Table 1). Among compounds above, 13r exhibited the best inhibitory activity with the EC50 value of 6.65 ± 1.47 μM, which is comparable to the reference agent PF-74.</p><p>Finally, the commercially available zidovudine unit was introduced as one CTD substituent, resulting in the compound 13t with moderate activity toward HIV-1 NL4-3 (EC50 = 10.45 ±3.08 μM).</p><p>On the whole, preliminary SAR studies of these newly designed compounds revealed that the anti-HIV-1(NL4-3) activities of phenylalanine derivatives were not only sensitive to types of CTD substituents but also affected by different linkage positions. The biological evaluation results and the SAR analysis described above will be beneficial to further design of CA protein inhibitors targeting the conformationally dynamic CTD-NTD interface.</p><!><p>We next sought to demonstrate that the newly synthesized and designed compounds retained target specificity to the HIV-1 CA protein. We chose to perform this analysis on three of our highest potency compounds, 13m, 13p, and 13r. The successful use of surface plasmon resonance (SPR) to quantify the binding interactions of CA-targeting small molecules has been previously reported [13-16]. Therefore, we chose this methodology for our studies.</p><p>Moreover, to ascertain whether or not there was any oligomeric preference of the compounds, we chose three different available CA protein constructs; the CA NTD, monomeric CA, and hexameric CA. Again, PF-74 was utilized as the reference compound. Figure 2 shows the results of this analysis.</p><p>As depicted in Fig. 2A, compounds 13m, 13p, 13r, and PF-74 all bind to the three different CA protein constructs (NTD, monomer and hexamer). PF-74 interacts the tightest overall, with equilibrium dissociation constant (KD) values for the NTD, monomer, and hexamer, of 3.6 μM, 2.7 μM and 64.3 nM, respectively. These values are consistent with the previous affinity studies reported in the literature [13, 24, 26] and demonstrate that PF-74 has a preference for binding to the hexameric form of the CA protein. This result, again, is consistent with the structural information regarding the binding mode of PF-74 in which the molecule makes contact with two adjacent protomers in the CA hexamer. In contrast to the oligomeric preference shown by PF-74, compounds 13m, 13p, and 13r, display comparable affinities to all of the CA protein constructs used, with the exception of 13r that appears to prefer the monomeric form of CA. Moreover, compounds 13m, 13p, and 13r have a markedly different kinetic signatures to PF-74, having rapid on- and off-rates, compared to the rapid on- but relatively slow off-rate of the parental compound. Taken together, this analysis demonstrates that the novel 1,2,3-triazole-containing phenylalanine derivatives retain the target specificity of the parental PF-74 and can be classified as HIV-1 CA inhibitors. Moreover, the use of SPR to determine not only an affinity but a kinetic signature for the compounds may provide a rapid method of screening for higher potency analogues. A recent study has demonstrated that the antiviral potency of a small series of HIV-1 entry inhibitors can be best correlated with the off-rate parameter of their interaction with their protein target [34]. As such, next-generation analogues in this series could be prioritized via a rapid off-rate screening strategy.</p><!><p>The observation that all of the new compounds interact with both monomeric and hexameric forms of the HIV-1 CA protein has implications for the stage at which they exert there antiviral effect. PF-74 binds to both CA monomer and hexamer and inhibits HIV-1 replication in the early, infective stages, and the late, assembly stages of the virus lifecycle [11]. We, therefore, sought to see if this dual-stage mode of action was true for our new compounds and chose to focus on 13m as the most potent representative of our derivatives. To achieve this, we utilized the modular nature of the single round infection assay, in which singly-round infective HIV-1 particles are generated recombinantly in HEK293T cells, and then used to infect U87.CD4.CCR5 cells: this allows for the separation of early and late stages of the replication cycle [35].</p><p>Effects on post-integration and assembly events were identified by producing Env pseudotyped viruses in the absence or presence of various concentrations of 13m. HIV-1B41 Env pseudotyped virus (which encodes for firefly luciferase as a reporter gene) were diluted ten-fold and used to infect U87.CD4.CCR5 target cells. Compound-induced late-stage effects were demonstrated by a decrease in infectivity in the target cells, compared to the control virus that was generated without the compound present. Compound-induced early-stage effects were observed by using virus produced in the absence of compound to infect U87.CD4.CCR5 target cells in the absence or presence of various concentrations of 13m. As can be seen in Table 2, 13m displays effects in both the post-integration, assembly and pre-integration, infective stages of the HIV-1 life-cycle. The control compound PF-74 also exhibits this dual-stage inhibition profile. These preliminary mechanism-of-action studies set the stage for more rigorous studies with these and higher potency compounds.</p><!><p>To better understand the activity results and find the potential binding modes of this series of molecules, the representative compound 13m was analyzed by MD simulation to find its binding to the active site of CA HIV-1 monomer. Figure 3A shows the root mean square deviation (RMSD) of the protein amino acid residues for all heavy atoms during the 1μs MD simulation. It is clear from the figure that the protein structure has deviated from the starting X-ray structure 5HGL. Furthermore, the figure shows that the protein structure could exist in different conformational ensembles due to the different RMSD values. To further investigate the deviation of the protein from the starting X-ray structure, the root mean squares fluctuation (RMSF) of residues was calculated and plotted in Figure 3B. This figure shows that most amino acid residues have deviated from the starting structure, which also indicates the presence of the protein in other conformational forms. The presence of the protein structure in different conformational forms could be accompanied with different binding modes of 13m. Accordingly, RMSD of 13m heavy atoms was calculated in reference to the first frame of the MD simulation and plotted in Figure 3C. Indeed, it is obvious from the figure that 13m is present in different structural clusters, which potentially indicate binding to the active site in different modes.</p><p>According to the RMSD and RMSF results, the entire trajectory has been clustered based on 13m (no fit), after alignment of protein residues, to see the binding modes of 13m in the active site. The clustering procedure yielded nine different structural clusters with two most populated. Figure 4 shows representative structures of the two most populated clusters with their corresponding expanded views for 13m binding to the active site. According to the clustering results, it is clear that CA HIV-1 monomer exists in two different conformational ensembles with RMSD between the representative structures of 6.4 Ǻ. This high RMSD value indicates that they are totally different. Inspection of the expanded views of the first and second clusters in the figure shows that in the most populated cluster 13m binds in similar way to PF-74 structure, where the core scaffold is oriented to the inside of the active site and the substituent is oriented to the outside of the active site. On the other hand, in the second most populated cluster, the core of 13m is oriented to the outside of the active site and the substituent is oriented to the inside of the active site, which is in contrast to the binding of PF-74. According to these results, the binding of 13m in two different binding modes could potentially induce the two different conformations of CA HIV-1 monomer. Furthermore, binding of 13m in two different binding modes could enhance its activity due to the increased chance of binding to the active site.</p><p>Interaction of 13m with CA HIV-1 monomer fixed it to the open conformation as shown by Jacques et. al in 5HGL X-ray structure[21], Figures 5 A and C show that GLN50 (in both clusters) is kept far from PRO1, HIS12, THR48 and ASN51 triad, which is important to form hydrogen bond to induce the close conformation. This could fix the hexamer to one conformation and disables its function. In the first conformational cluster 13m binds in similar mode of PF-74 but with different amino acid residues. Where, the phenyl ring of the core region could form hydrophobic interaction with LYS70 similar to PF-74. Furthermore, 13m forms hydrophobic interactions with MET66 and LEU56. It also forms aliphatic hydrogen boning with ASN74 through its methoxy group in the core region, and hydrogen bond with ASN57 at the substituent region. 13m in the other binding mode, which is different from the binding mode of PF-74, forms hydrophobic interaction with LYS70 through its substituent region in contrast to the first binding mode. Also, it forms hydrophobic interactions with MET66 and ALA64 through its core region. Interaction of 13m with CA HIV-1 monomer in different binding modes involves more amino acids, which increases the possibility of inhibiting this protein and could accounts for its high inhibitory activity.</p><!><p>In brief, taking the most studied HIV CA inhibitor PF-74 as lead compound, we designed and expeditiously synthesized a series of 1,2,3-triazole-containing phenylalanine derivatives via CuAAC reaction. Among them, 13m exhibited the best anti-HIV-1 activity (EC50 = 4.33 μM, SI > 13.33), being similar to the lead PF-74 (EC50 = 5.95 μM, SI > 11.85). Direct binding studies using SPR demonstrated that this series of phenylalanine derivatives interact with the HIV-1 CA protein, irrespective of its oligomeric status. In-line with this observation, we have demonstrated the compound 13m inhibits the replication of HIV-1 in both the early and late stages. Finally, through molecular dynamics simulation, we can conclude that 13m potentially has two different binding modes to HIV-1 CA monomer, which has implication for the precise manner of CA protein inhibition in each of the discrete stages of replication. We envisioned that the conformationally dynamic CTD-NTD interface still have ample space for further modification to form potential interaction with nearby hotpot residues. Ongoing studies regarding the further development of these novel capsid inhibitors from our lab will be reported in due course.</p><!><p>All melting points of the compounds were determined on a micro melting point apparatus and were uncorrected. 1H-NMR and 13C-NMR spectra were obtained via a Bruker Avance-400 NMR-spectrometer in DMSO or CDCl3 using TMS as internal reference. Chemical shifts were expressed in δ units (ppm) and J values were presented in hertz (Hz). Related mass spectra dates were determined by a LC Autosampler Device: Standard G1313A instrument. TLC was performed on Silica Gel GF254 for TLC and spots were visualized by irradiation with UV light (λ = 254 nm). Meanwhile, flash column chromatography was performed on column packed with Silica Gel 60 (200-300 mesh). Rotary evaporator under reduced pressure condition was used to concentrate the reaction solutions. Solvents were of reagent grade and purified with standard methods when necessary.</p><!><p>A solution of (tert-butoxycarbonyl)-L-phenylalanine (1, 8.75 mmol, 2.3 g) in 15 mL dichloromethane was added PyBop (10.9 mmol, 5.7 g) at 0°C, and the mixture stirred for 0.5 h. Subsequently, DIEA (21.87 mmol, 3.61 mL) and 4-methoxy-N-methylaniline (7.29 mmol, 1.0 g) were added to the mixture and then stirred at room temperature for another 8-9 h (monitored by TLC). The resulting mixture was evaporated under reduced pressure and the residue was initially washed by 1N HCl and extracted with ethyl acetate (3 × 20 mL). Then, the combined organic layer was washed with saturated sodium bicarbonate (3 × 50 mL), dried over anhydrous Na2SO4, filtered, and concentrated under reduced pressure to afford corresponding crude intermediate 2 as yellow oil with a yield of 94%. 1H NMR (400 MHz, DMSO-d6δ 7.21 (d, J = 8.4 Hz, 3H), 7.15 (d, J = 7.1 Hz, 2H), 7.02 (d, J = 8.4 Hz, 2H), 6.85 – 6.75 (m, 2H), 4.15 (q, J = 5.4 Hz, 1H), 3.80 (s, 3H), 3.12 (s, 3H), 2.81–2.54 (m, 2H), 1.30 (s, 9H). 13C NMR (100 MHz, DMSO) δ 172.21, 158.98, 155.74, 138.53, 136.13, 129.28, 128.47, 126.70, 115.21, 78.33, 55.94, 53.53, 37.86, 37.10, 28.65. ESI-MS: m/z 385.4 (M + 1)+, C22H28N2O4 (384.2).</p><!><p>Trifluoroacetate (34.2 mmol, 5.0 eq) was added dropwise to intermediate 2 (2.63 g, 6.84 mmol, 1.0 eq) in 30 mL dichloromethane and stirred at room temperature for 6-7 h. Then, the resulting mixture solution was alkalized to pH = 9 with saturated sodium bicarbonate solution, and then extracted with dichloromethane (3 × 20 mL). The combined organic layer was dried over anhydrous Na2SO4, filtered, and concentrated under reduced pressure to yield the crude product, which was purified by flash column chromatography to afford compound 3 as yellow oil with the yield of 84%. 1H NMR (400 MHz, DMSO-d4) δ 7.34–7.15 (m, 3H), 6.90 (m, 6H), 3.77 (s, 3H), 3.35 (t, J = 6.9 Hz, 1H), 3.06 (s, 3H), 2.76 (dd, J = 12.9, 6.8 Hz, 1H), 2.46 (dd, J = 12.9, 7.1 Hz, 1H). 13C NMR (100 MHz, DMSO) δ 174.96, 158.76, 139.02, 136.36, 129.51, 128.93, 128.46, 126.54, 115.03, 55.83, 53.38, 42.25, 37.45. ESI-MS: m/z 285.3 (M + 1)+, C17>H20N2O2 (284.1).</p><!><p>Propiolic acid (4.22 mmol, 0.3 g) and HATU (5.28 mmol, 2.0 g) were mixed in dichloromethane and stirred in an ice bath for 1 h. Then, the intermediate 3 (6.33 mmol, 1.80 g) and DIEA (7.03 mmol, 1.16 mL) were added to the above solution slowly at 0°C. The reaction system was then stirred at room temperature for additional 12 h. The solvent was removed under reduced pressure and then 1N HCl (30 mL) was added, extracted with ethyl acetate (3 × 30 mL). The combined organic layer was washed with saturated sodium bicarbonate (3 × 50 mL). The resulting organic layer was washed with saturated salt water, dried over anhydrous Na2SO4, filtered, and concentrated under reduced pressure to afford corresponding crude product, which was purified by flash column chromatography to yield compound 4 as yellow oil with the yield of 38%. 1H NMR (400 MHz, DMSO-d6) δ 9.16 (d, J = 7.8 Hz, 1H), 7.17 (td, J = 7.0, 4.5 Hz, 5H), 7.00 (d, J = 8.9 Hz, 2H), 6.85 – 6.82 (m, 2H), 4.45 (ddd, J = 9.7, 7.7, 4.5 Hz, 1H), 3.79 (s, 3H), 3.12 (s, 3H), 2.91 – 2.71 (m, 2H), 2.69 (s, 1H). 13C NMR (100 MHz, DMSO) δ 170.90, 159.10, 151.91, 137.93, 135.87, 129.19, 128.63, 126.96, 115.17, 78.37, 76.80, 55.91, 52.53, 37.85, 36.86. ESI-MS: m/z 337.6 (M + 1)+, C20H20N2O3 (336.1).</p><!><p>Benzyl bromide (5, 1 g, 1 eq) and sodium azide (NaN3, 1.5 eq) were dissolved in 10 mL N, N-dimethylformamide (DMF). The mixture was added 20 mL water after stirred at room temperature for 12-13 h, and then extracted by ethyl acetate (3 × 20 mL). The organic phase was combined, dried over anhydrous Na2SO4, filtered, and concentrated to yield the crude azide substituent 6, which was used for the next step without further purification. The abovementioned method was applied to the substituent (azidomethyl)(phenyl)sulfane (10, C7H7N3S). 1H NMR (400 MHz, DMSO-d6) δ 7.52–7.47 (m, 2H), 7.42–7.35 (m, 2H), 7.33 – 7.27 (m, 1H), 4.87 (s, 2H).</p><!><p>Different substituted phenylboronic acid compounds (7, 1 g, 1 eq), sodium azide (NaN3, 1.5 eq) and copper sulfate pentahydrate (CuSO4·5H2, 0.1 eq) were mixed in the 10 mL methanol and then stirred at room temperature for 8-9 h. Subsequently, the reaction solvent was removed under reduced pressure and 20 mL of water was added. Aqueous solution was extracted by ethyl acetate (3 × 20 mL), the combined phase was dried over anhydrous Na2SO4, filtered, and concentrated to afford the crude azide substituents 8 which was used for the next step directly.</p><!><p>Under ice cooling, after a solution of 3-chloroperbenzoic acid (mCPBA, 580 mg, 75%, 1.26 mmol) in CH2Cl2 (1.0 mL) was added dropwise to a solution of the intermediate (azidomethyl)(phenyl)sulfane (10, 168 μL, 1.17 mmol) in dry CH2Cl2 (10 mL), the mixture was stirred at room temperature for 15 h. Then saturated sodium bicarbonate solution was poured into the reaction mixture and extracted with ethyl acetate (3 χ 50 mL). The combined organic phase was dried over anhydrous sodium sulphate and then evaporated. The residue was purified using flash chromatography with n-hexane/ethylacetate to afford azide substituent ((azidomethyl)sulfinyl)benzene (12, C7H7N3O2S). 1H NMR (400 MHz, Chloroform-d) δ 8.00 – 7.94 (m, 2H), 7.77 – 7.70 (m, 1H), 7.63 (dd, J = 8.4, 7.1 Hz, 2H), 4.31 (s, 2H, CH2).</p><p>This method was also applied to synthesize substituent ((azidomethyl)sulfinyl)benzene (11, C7H7N3OS). Compared to the synthesis of azide substituent ((azidomethyl)sulfinyl)benzene (12), the only difference of ((azidomethyl)sulfinyl)benzene (11) synthesis is that the reaction would be finished after reacting with mCPBA at 0°C merely for 1 h. 1H NMR (400 MHz, Chloroform-d) δ 7.64–7.56 (m, 2H), 7.52–7.35 (m, 3H), 4.21 (d, J = 12.0 Hz, 1H), 3.97 (d, J = 11.9 Hz, 1H)</p><!><p>The key intermediate 4 (1.0 eq), azide substituents (6, 8, 10, 11 and 12, 1.1 eq), ascorbic acid sodium (0.6 eq) and CuSO4.5H2O (0.3 eq) were dissolved in the solution of tetrahydrofuran/water (v:v = 1:1). The resulting mixture was stirred at 30-60°C for 4-6 h. Then the reaction mixture was extracted with ethyl acetate (3 × 10 mL), and the combined organic phase was washed with saturated salt water (3 × 10 mL), dried over anhydrous MgSO4, filtered, and concentrated under reduced pressure to give the corresponding crude target product, which was purified by flash column chromatography to afford product 13a-13t. Yield: 30%-80%.</p><!><p>White solid, yield: 70%. mp: 60-62°C. 1H NMR (400 MHz, DMSO-d6) δ 8.49 (s, 1H), 8.35 (d, J = 8.0 Hz, 1H), 7.31 – 7.12 (m, 8H), 7.08 (d, J = 7.5 Hz, 1H), 7.00 (d, J = 8.4 Hz, 2H), 6.92 – 6.82 (m, 2H), 5.66 (s, 2H), 4.66 (q, J = 7.4 Hz, 1H), 3.80 (s, 3H), 3.12 (s, 3H), 2.81 (m, 2H), 2.30 (s, 3H). 13C NMR (100 MHz, DMSO-d6) δ 171.28, 159.56, 159.07, 142.56, 137.98, 136.76, 135.93, 134.20, 130.94, 129.35, 129.24, 129.12, 128.93, 128.60, 127.04, 126.89, 126.79, 115.19, 55.92, 51.76, 51.68, 37.88, 37.27, 19.09. ESI-MS: m/z 484.5 (M + 1)+, 506.4 (M + 23)+, C28H29N5O3 (483.2).</p><!><p>Yellow solid, yield: 30%. mp: 58-60°C. 1H NMR (400 MHz, DMSO-d6) δ 8.59 (s, 1H), 8.34 (d, J = 8.0 Hz, 1H), 7.23 – 7.12 (m, 8H), 7.01 (d, J = 8.2 Hz, 3H), 6.88 (d, J = 6.8 Hz, 2H), 5.60 (s, 2H), 4.67 (q, J = 6.0 Hz, 1H), 3.81 (s, 3H), 3.12 (d, J = 4.9 Hz, 3H), 2.89 (d, J = 7.1 Hz, 2H), 2.29 (s, 3H). 13C NMR (100 MHz, DMSO-d6) δ 171.28, 159.56, 159.08, 142.57, 137.97, 136.77, 135.93, 134.19, 130.94, 129.35, 129.23, 129.13, 128.93, 128.60, 127.03, 126.89, 126.79, 115.19, 55.91, 51.75, 51.69, 37.88, 37.31, 19.09. ESI-MS: m/z 484.5 (M + 1)+, 506.4 (M + 23)+, C28H29N5O3 (483.2).</p><!><p>Yellow solid, yield: 55%. mp: 63-65°C. 1H NMR (400 MHz, DMSO-d6) δ 8.56 (s, 1H), 8.32 (d, J = 8.0 Hz, 1H), 7.36 – 7.08 (m, 9H), 7.00 (d, J = 8.4 Hz, 2H), 6.92 – 6.83 (m, 2H), 5.58 (s, 2H), 4.66 (q, J = 7.3 Hz, 1H), 3.80 (s, 3H), 3.12 (s, 3H), 2.88 (d, J = 7.1 Hz, 2H), 2.28 (s, 3H). 13C NMR (100 MHz, DMSO-d6) δ 171.27, 159.58, 159.07, 142.67, 138.12, 137.98, 135.92, 133.09, 129.80, 129.34, 129.24, 128.60, 128.47, 126.90, 115.19, 55.91, 53.34, 51.73, 37.88, 37.27, 21.17. ESI-MS: m/z 484.5 (M + 1)+, 506.4 (M + 23)+, C28H29N5O3 (483.2).</p><!><p>Yellow solid, yield: 60%. mp: 162-164°C. 1H NMR (400 MHz, DMSO-d6) δ 8.56 (s, 1H), 8.38 (d, J = 8.0 Hz, 1H), 7.54 (dd, J = 7.6, 1.7 Hz, 1H), 7.40 (pd, J = 7.5, 1.7 Hz, 2H), 7.29 – 7.11 (m, 6H), 7.01 (d, J = 8.5 Hz, 2H), 6.94 – 6.75 (m, 2H), 5.76 (s, 2H), 4.67 (q, J = 7.3 Hz, 1H), 3.80 (s, 3H), 3.12 (s, 3H), 2.89 (d, J = 7.0 Hz, 2H). 13C NMR (100 MHz, DMSO-d6) δ 171.28, 159.52, 159.08, 142.50, 137.99, 135.93, 133.30, 133.12, 131.07, 130.87, 130.15, 129.35, 129.24, 128.60, 128.26, 127.45, 126.89, 115.19, 55.92, 51.78, 51.41, 37.88, 37.25. ESI-MS: m/z 504.3 (M + 1)+, 526.4 (M + 23)+, C27H26ClN5O3 (503.1).</p><!><p>White solid, yield: 65%. mp: 288-290°C. 1H NMR (400 MHz, DMSO-d6) δ 8.64 (s, 1H), 8.36 (d, J = 8.0 Hz, 1H), 7.42 (dd, J = 6.0, 2.9 Hz, 3H), 7.37 – 7.25 (m, 1H), 7.23 – 7.10 (m, 5H), 7.01 (d, J = 8.4 Hz, 2H), 6.93 – 6.80 (m, 2H), 5.67 (s, 2H), 4.67 (q, J = 7.3 Hz, 1H), 3.80 (s, 3H), (s, 3H), 2.89 (d, J = 7.0 Hz, 2H). 13C NMR (100 MHz, DMSO-d6) δ 171.26, 159.54, 159.07, 142.77, 138.44, 138.00, 135.92, 133.78, 131.24, 129.35, 129.25, 128.75, 128.60, 128.37, 127.27, 127.18, 126.89, 115.19, 55.92, 52.75, 51.78, 37.88, 37.24. ESI-MS: m/z 504.3 (M + 1)+, 526.3 (M + 23)+, C27H26ClN5O3 (503.1).</p><!><p>White solid, yield: 50%. mp: 172-174°C. 1H NMR (400 MHz, DMSO-d6) δ 8.61 (s, 1H), 8.34 (d, J = 8.1 Hz, 1H), 7.51 – 7.42 (m, 2H), 7.35 (d, J = 8.3 Hz, 2H), 7.26 – 7.09 (m, 5H), 7.01 (d, J = 8.8 Hz, 2H), 6.93 – 6.81 (m, 2H), 5.65 (s, 2H), 4.67 (q, J = 7.3 Hz, 1H), 3.80 (s, 3H), 3.12 (s, 3H), 2.89 (d, J = 7.0 Hz, 2H). 13C NMR (100 MHz, DMSO-d6) δ 171.26, 159.54, 159.08, 142.75, 137.98, 135.93, 135.08, 133.47, 130.40, 129.34, 129.28, 129.24, 128.60, 127.16, 126.89, 115.20, 55.92, 52.73, 51.75, 37.88, 37.26. ESI-MS: m/z 504.3 (M + 1)+, 526.3 (M + 23)+, C27H26ClN5O3 (503.1).</p><!><p>Light yellow solid, yield: 58%. mp: 168-170°C. 1H NMR (400 MHz, DMSO-d6) δ 8.62 (s, 1H), 8.39 (d, J = 8.0 Hz, 1H), 7.93 (dd, J = 7.7, 1.3 Hz, 1H), 7.77 – 7.70 (m, 1H), 7.61 – 7.55 (m, 1H), 7.38 (d, J = Hz, 1H), 7.17 (dt, J = 13.7, 7.5 Hz, 5H), 7.01 (d, J = 8.5 Hz, 2H), 6.91 – 6.84 (m, 2H), 5.87 (s, 2H), 4.67 (q, J = 7.2 Hz, 1H), 3.81 (s, 3H), 3.13 (s, 3H), 2.90 (d, J = 6.9 Hz, 2H). 13C NMR (100 MHz, DMSO-d6) δ 171.25, 159.47, 159.08, 142.63, 138.84, 137.99, 135.92, 134.36, 133.91, 129.94, 129.79, 129.35, 129.25, 128.60, 127.65, 126.89, 117.38, 115.20, 111.66, 55.93, 51.79, 51.76, 37.89, 37.24. ESI-MS: m/z 495.4 (M + 1)+, 512.5 (M + 18)+, 517.4 (M + 23)+, C28H26N6O3 (494.2).</p><!><p>White solid, yield: 62%. mp: 168-170°C. 1H NMR (400 MHz, DMSO-d6) δ 8.65 (s, 1H), 8.35 (d, J = 8.0 Hz, 1H), – 7.81 (m, 2H), 7.62 (dt, J = 15.3, 7.8 Hz, 2H), 7.17 (dt, J = 13.8, 7.6 Hz, 5H), 7.01 (d, J = 8.5 Hz, 2H), 6.93 – 6.82 (m, 2H), 5.72 (s, 2H), 4.67 (q, J = 7.3 Hz, 1H), 3.81 (s, 3H), 3.12 (s, 3H), 2.89 (d, J = 7.0 Hz, 2H). 13C NMR (100 MHz, DMSO-d6) δ 171.24, 159.50, 159.08, 142.82, 137.97, 137.57, 135.93, 133.45, 132.59, 132.25, 130.60, 129.35, 129.23, 128.60, 127.37, 126.89, 118.85, 115.20, 112.22, 55.92, 52.60, 51.73, 37.88, 37.31. ESI-MS: m/z 495.4 (M + 1)+, 517.5 (M + 23)+, C28H26N6O3 (494.2).</p><!><p>White solid, yield: 45%. mp: 190-192°C. 1H NMR (400 MHz, DMSO-d6) δ 8.65 (s, 1H), 8.37 (d, J = 8.0 Hz, 1H), 7.91 – 7.81 (m, 2H), 7.47 (d, J = 8.0 Hz, 2H), 7.17 (dt, J = 13.7, 7.6 Hz, 5H), 7.07 – 6.96 (m, 2H), 6.94 – 6.82 (m, 2H), 5.77 (s, 2H), 4.68 (q, J = 7.2 Hz, 1H), 3.81 (s, 3H), 3.13 (s, 3H), 2.90 (d, J = 7.0 Hz, 2H). 13C NMR (100 MHz, DMSO-d6) δ 171.26, 159.53, 159.08, 142.82, 141.51, 138.01, 135.92, 133.25, 129.34, 129.25, 129.20, 128.61, 127.54, 126.89, 118.97, 115.20, 111.51, 55.92, 52.90, 51.80, 37.88, 37.20. ESI-MS: m/z 495.4 (M + 1)+, 517.5 (M + 23)+, C28H26N6O3 (494.2).</p><!><p>Yellow solid, yield: 60%. mp: 63-65°C. 1H NMR (400 MHz, DMSO-d6 δ 8.57 (s, 1H), 8.41 (d, J = 8.0 Hz, 1H), 8.16 (d, J = 8.0 Hz, 1H), 7.76 (t, J = 7.6 Hz, 1H), 7.66 (t, J = 7.8 Hz, 1H), 7.18 (dt, J = 13.2, 6.8 Hz, 5H), 7.09 (d, J = 7.7 Hz, 1H), 7.01 (d, J = 8.4 Hz, 2H), 6.94 – 6.83 (m, 2H), 6.02 (s, 2H), 4.68 (q, J = 7.2 Hz, 1H), 3.81 (s, 3H), 3.13 (s, 3H), 2.91 (d, J = 6.9 Hz, 2H). 13C NMR (100 MHz, DMSO-d6) δ 171.28, 159.51, 159.09, 148.01, 142.64, 138.00, 135.94, 134.93, 130.82, 130.69, 130.26, 129.35, 129.25, 128.61, 127.89, 126.90, 125.63, 115.20, 55.92, 51.80, 50.80, 37.88, 37.25. ESI-MS: m/z 515.5 (M + 1)+, 537.4 (M + 23)+, C27H26N6O5 (514.2).</p><!><p>Light yellow solid, yield: 67%. mp: 61-63 °C. 1H NMR (400 MHz, DMSO-d6) δ 8.69 (s, 1H), 8.37 (d, J = 8.0 Hz, 1H), 8.25 (t, J = 2.0 Hz, 1H), 8.22 (d, J = 8.3 Hz, 1H), 7.78 (d, J = 7.7 Hz, 1H), 7.69 (t, J = 7.9 Hz, 1H), 7.18 (dd, J = 18.4, 7.9 Hz, 5H), 7.01 (d, J = 8.4 Hz, 2H), 6.92 – 6.83 (m, 2H), 5.83 (s, 2H), 4.67 (q, J = 7.2 Hz, 1H), 3.80 (s, 3H), 3.12 (s, 3H), 2.89 (d, J = 6.9 Hz, 2H). 13C NMR (100 MHz, DMSO-d6) δ 171.24, 159.50, 159.08, 148.37, 142.83, 138.11, 137.98, 135.93, 135.23, 130.94, 129.34, 129.24, 128.59, 127.44, 126.89, 123.73, 123.38, 115.20, 55.92, 52.51, 51.77, 37.88, 37.25. ESI-MS: m/z 515.5 (M + 1)+, 537.4 (M + 23)+, C27H26N6O5 (514.2).</p><!><p>Light yellow solid, yield: 57%. mp: 65-67°C. 1H NMR (400 MHz, DMSO-d6) δ 8.67 (s, 1H), 8.38 (d, J = 8.0 Hz, 1H), 8.29 – 8.20 (m, 2H), 7.54 (d, J = 8.6 Hz, 2H), 7.17 (dt, J = 13.8, 7.7 Hz, 5H), 7.01 (d, J = 8.8 Hz, 2H), 6.93 – 6.84 (m, 2H), 5.84 (s, 2H), 4.68 (q, J = 12 Hz, 1H), 3.81 (s, 3H), 3.13 (s, 3H), 2.90 (d, J = 7.0 Hz, 2H). 13C NMR (100 MHz, DMSO-d6) δ 171.25, 159.51, 159.09, 147.77, 143.44, 142.85, 137.99, 135.93, 129.53, 129.34, 129.24, 128.61, 127.60, 126.89, 124.42, 115.21, 55.93, 52.64, 51.78, 37.88, 37.25. ESI-MS: m/z 515.4 (M + 1)+, C27H26N6O5 (514.2).</p><!><p>White solid, yield: 65%. mp: 72-74°C. 1H NMR (400 MHz, DMSO-d6) δ 8.66 (s, 1H), 8.34 (d, J = 8.0 Hz, 1H), 7.93 (dd, J = 8.9, 4.9 Hz, 3H), 7.87 (s, 1H), 7.54 (dd, J = 6.5, 3.1 Hz, 2H), 7.49 – 7.41 (m, 1H), 7.24 – 7.12 (m, 5H), 7.00 (d, J = 8.4 Hz, 2H), 6.94 – 6.82 (m, 2H), 5.82 (s, 2H), 4.67 (q, J = 7.4 Hz, 1H), 3.80 (s, 3H), 3.12 (s, 3H), 2.88 (d, J = 7.0 Hz, 2H). 13C NMR (100 MHz, DMSO-d6) δ 171.26, 159.57, 159.08, 142.74, 137.97, 135.93, 133.57, 133.21, 133.01, 129.35, 129.23, 129.05, 128.60, 128.31, 128.09, 127.46, 127.19, 127.07, 126.98, 126.88, 126.11, 115.20, 55.91, 53.73, 51.73, 37.88, 37.31. ESI-MS: m/z 520.4 (M + 1)+, 542.4 (M + 23)+, C31H29N5O3 (519.2).</p><!><p>White solid, yield: 68%. mp: 78-80°C. NMR (400 MHz, DMSO-d6) δ 8.54 (s, 1H), 8.33 (d, J = 8.0 Hz, 1H), (d, J = 8.1 Hz, 1H), 7.99 (t, J = 8.5 Hz, 2H), 7.57 (ddt, J = 20.7, 14.9, 7.3 Hz, 3H), 7.43 (d, J = 7.1 Hz, 1H), 7.24 – 7.09 (m, 5H), 6.99 (d, J = 8.4 Hz, 2H), 6.91 – 6.80 (m, 2H), 6.15 (s, 2H), 4.64 (q, J = 7.6 Hz, 1H), 3.79 (s, 3H), 3.10 (s, 3H), 2.87 (d, J = 5.3 Hz, 2H). 13C NMR (100 MHz, DMSO-d6) δ 171.25, 159.50, 159.06, 142.56, 137.95, 135.91, 133.84, 131.54, 130.99, 129.67, 129.33, 129.21, 128.58, 127.85, 127.35, 127.12, 126.88, 126.69, 126.05, 123.59, 115.17, 55.91, 51.72, 51.54, 37.86, 37.29. ESI-MS: m/z 520.4 (M + 1)+, 542.4 (M + 23)+, C31H29N5O3 (519.2).</p><!><p>White solid, yield: 65%. mp: 136-138°C. 1H NMR (400 MHz, DMSO-d6 δ 9.10 (s, 1H), 8.42 (d, J = 8.0 Hz, 1H), 7.53 (d, J = 2.2 Hz, 1H), 7.41 (dd, J = 8.4, 2.2 Hz, 1H), 7.25 – 7.10 (m, 6H), 7.06 – 7.00 (m, 2H), 6.95 – 6.89 (m, 2H), 6.16 (s, 2H), 4.72 (q, J = 7.2 Hz, 1H), 3.81 (s, 3H), 3.14 (s, 3H), 2.94 (d, J = 6.9 Hz, 2H). 13C NMR (100 MHz, DMSO-d6) δ 171.20, 159.36, 159.10, 148.64, 148.29, 143.20, 137.94, 135.94, 131.05, 129.40, 129.26, 128.62, 126.92, 125.40, 115.22, 114.89, 109.11, 102.71, 55.93, 51.79, 37.92, 37.34. ESI-MS: m/z 500.3 (M + 1)+, 522.4 (M + 23)+, C27H25N5O5 (499.1).</p><!><p>White solid, yield: 80%. mp: 170-172°C. 1H NMR (400 MHz, DMSO-d6) δ 9.37 (s, 1H), 8.52 (d, J = 8.0 Hz, 1H), 8.50 – 8.43 (m, 1H), 8.27 – 8.20 (m, 1H), 8.09 (dt, J = 7.8, 1.3 Hz, 1H), 7.78 (t, J = 8.0 Hz, 1H), 7.29 – 7.13 (m, 5H), 7.02 (d, J = 8.4 Hz, 2H), 6.99 – 6.88 (m, 2H), 4.73 (q, J = 7.2 Hz, 1H), 3.92 (s, 3H), 3.82 (s, 3H), 3.15 (s, 3H), 2.95 (d, J = 6.9 Hz, 2H). 13C NMR (100 MHz, DMSO-d6) δ 171.19, 165.67, 159.23, 159.11, 143.57, 137.96, 136.98, 136.65, 135.94, 131.09, 130.00, 129.40, 129.26, 128.63, 126.92, 125.59, 125.49, 121.25, 115.22, 55.93, 53.07, 51.86, 37.93, 37.34. ESI-MS: m/z 514.3 (M + 1)+, 536.4 (M + 23)+, C28H27N5O5 (513.2).</p><!><p>Yellow solid, yield: 32%. mp: 64-66°C. 1H NMR (400 MHz, DMSO-d6) δ 8.43 (s, 1H), 8.38 (d, J = 8.0 Hz, 1H), 7.42 (d, J = 7.3 Hz, 2H), 7.34 (q, J = 8.2, 7.5 Hz, 3H), 7.22 (d, J = 7.9 Hz, 2H), 7.15 (d, J = 6.4 Hz, 3H), 7.01 (d, J = 8.4 Hz, 2H), 6.86 (d, J = 6.7 Hz, 2H), 5.99 (s, 2H), 4.64 (q, J = 7.4 Hz, 1H), 3.81 (s, 3H), 3.13 (s, 3H), 2.88 (d, J = 6.9Hz, 2H). 13C NMR (100 MHz, DMSO-d6) δ 171.25, 159.37, 159.08, 142.64, 138.02, 135.93, 132.43, 131.29, 129.83, 129.32, 129.26, 128.59, 128.41, 126.89, 126.55, 115.20, 55.93, 52.57, 51.88, 37.89, 37.09. ESI-MS: m/z 502.3 (M + 1)+, 524.4 (M + 23)+, C27H27N5O3S (501.1).</p><!><p>White solid, yield: 45%. mp: 77-79°C. 1H NMR (400 MHz, DMSO-d6) δ 8.46 (t, J = 9.6 Hz, 1H), 8.34 (d, J = 25.1 Hz, 1H), 7.56 (d, J = 12.2 Hz, 5H), 7.37 – 7.10 (m, 5H), 7.03 (d, J = 8.4 Hz, 2H), 6.87 (d, J = 6.8 Hz, 2H), 6.00 (d, J = 13.4 Hz, 1H), 5.77 (dd, J = 13.4, 3.6 Hz, 1H), 4.66 (q, J = 7.4 Hz, 1H), 3.82 (s, 3H), 3.14 (s, 3H), 2.90 (d, J = 6.8 Hz, 2H). 13C NMR (100 MHz, DMSO-d6 δ 171.31, 159.26, 159.10, 142.07, 142.01, 140.50, 140.42, 138.08, 135.95, 132.17, 129.74, 129.71, 129.32, 128.59, 128.28, 128.13, 126.89, 124.89, 124.84, 115.22, 68.55, 55.94, 51.97, 37.91, 37.03. ESI-MS: m/z 518.4 (M + 1)+, 540.4 (M + 23)+, C27H27N5O4S (517.1).</p><!><p>Light greeen solid, yield: 56%. mp: 68-70°C. 1H NMR (400 MHz, DMSO-d6) δ 8.57 (d, J = 7.9 Hz, 1H), 8.50 (s, 1H), 7.85 – 7.59 (m, 5H), 7.34 – 7.12 (m, 5H), 7.09 – 6.99 (m, 2H), 6.91 – 6.80 (m, 2H), 6.40 (s, 2H), 4.65 (q, J = 7.2 Hz, 1H), 3.81 (s, 3H), 3.14 (s, 3H), 2.90 (d, J = 6.6 Hz, 2H). 13C NMR (100 MHz, DMSO-d6) δ 171.27, 159.11, 159.07, 142.60, 138.06, 136.27, 135.94, 135.47, 130.08, 129.31, 128.99, 128.60, 128.50, 126.90, 115.22, 67.46, 55.94, 52.05, 37.90, 37.02. ESI-MS: m/z 534.3 (M + 1)+, 556.3 (M + 23)+, C27H27N5O5S (533.1).</p><!><p>White solid, yield: 50%. mp: 143-145°C. 1H NMR (400 MHz, DMSO-d6) δ 11.37 (s, 1H), 8.73 (s, 1H), 8.35 (d, J = 8.1 Hz, 1H), 7.89 – 7.76 (m, 1H), 7.18 (dt, J = 15.1, 8.1 Hz, 5H), 7.02 (d, J = 8.4 Hz, 2H), 6.95 – 6.80 (m, 2H), 6.42 (t, J = 6.6 Hz, 1H), 5.43 (dt, J = 8.6, 5.5 Hz, 1H), 5.28 (t, J = 5.2 Hz, 1H), 4.70 (q, J = 7.3 Hz, 1H), 4.23 (q, J = 4.0 Hz, 1H), 3.81 (s, 3H), 3.74 – 3.54 (m, 2H), 3.13 (s, 3H), 2.91 (d, J = 7.0 Hz, 2H), 2.83 – 2.59 (m, 2H), 1.81 (s, 3H). 13C NMR (100 MHz, DMSO-d6) δ 171.25, 164.19, 159.52, 159.09, 150.89, 142.69, 137.98, 136.72, 135.92, 129.35, 129.25, 128.62, 126.90, 126.69, 115.22, 110.11, 84.73, 84.29, 61.09, 60.06, 55.93, 51.73, 37.90, 37.60, 37.29, 12.73. ESI-MS: m/z 604.5 (M + 1)+, 621.7 (M + 18)+, 626.5 (M + 23)+, C30H33N7O7 (603.2).</p><!><p>Inhibitory activity of compounds against HIV-1 infection in MT-4 Cells was measured as the reduction in luciferase gene expression after multiple rounds of virus infection of the cells similar to that described previously[36]. Briefly, 200 TCID50 of virus (NL4-3) was used to infect MT-4 cells in the presence of various concentrations of compounds. Two days after infection, the culture medium was removed from each well and 100 μL of Bright Glo reagent (Promega, Luis Obispo, CA) was added to the cells for measurement of luminescence using a Victor 2 luminometer. The effective concentration (EC50) against HIV-1 strains was defined as the concentration that caused a 50% reduction of luciferase activity (Relative Light Units) compared to virus control wells.</p><!><p>A CytoTox-Glo cytotoxicity assay (Promega) was used to determine the cytotoxicity of the synthesized compounds. Parallel to the antiviral assays, MT-4 cells were cultured in the presence of various concentrations of the compounds for 1 day. The percent of viable cells was determined by following the protocol provided by the manufacturer. The 50% cytotoxic concentration (CC50) was defined as the concentration that caused a 50% reduction in cell viability.</p><!><p>All binding assays were performed on a ProteOn XPR36 SPR Protein Interaction Array System (Bio-Rad Laboratories, Hercules, CA). The instrument temperature was set at 25°C for all kinetic analyses. ProteOn GLH sensor chips were preconditioned with two short pulses each (10 seconds) of 50 mM NaOH, 100 mM HCl, and 0.5% sodium dodecyl sulfide. Then the system was equilibrated with PBS-T buffer (20 mM sodium phosphate, 150 mM NaCl, and 0.005% polysorbate 20, pH 7.4). The surface of a GLH sensorchip was activated with a 1:100 dilution of a 1:1 mixture of 1-ethyl-3-(3-dimethylaminopropyl) carbodiimide hydrochloride (0.2 M) and sulfo-N-hydroxysuccinimide (0.05 M). Immediately after chip activation, the HIV-1 NL4-3 capsid protein constructs, purified as in Xu el al. [16], were prepared at a concentration of 100 μg/ml in 10 mM sodium acetate, pH 5.0 and injected across ligand flow channels for 5 min at a flow rate of 30 μl/min. Then, after unreacted protein had been washed out, excess active ester groups on the sensor surface were capped by a 5 minute injection of 1 M ethanolamine HCl (pH 8.0) at a flow rate of 5 μl/min. A reference surface was similarly created by immobilizing a non-specific protein (IgG b12 anti HIV-1 gp120; was obtained through the NIH AIDS Reagent Program, Division of AIDS, NIAID, NIH: Anti-HIV-1 gp120 Monoclonal (IgG1 b12) from Dr. Dennis Burton and Carlos Barbas) and was used as a background to correct non-specific binding.</p><p>To prepare a compound for direct binding analysis, compound stock solutions, along with 100% DMSO, and totaling 30μl was made to a final volume of 1 ml by addition of sample preparation buffer (PBS, pH 7.4). Preparation of analyte in this manner ensured that the concentration of DMSO was matched with that of running buffer with 3% DMSO. Serial dilutions were then prepared in the running buffer (PBS, 3% DMSO, 0.005% polysorbate 20, pH 7.4) and injected at a flow rate of 100 μl/min, for a 1 minute association phase, followed by up to a 5 minutes dissociation phase using the "one shot kinetics" capability of the Proteon instrument[38]. Data were analyzed using the ProteOn Manager Software version 3.0 (Bio-Rad). The responses from the reference flow cell were subtracted to account for the nonspecific binding and injection artifacts. The equilibrium dissociation constant (KD) for the interactions, and derived from a minimum of three experiments, were calculated in ProteOn Manager Version 3.1.0.6 (Bio-Rad, Hercules, CA), using the equilibrium analysis function.</p><!><p>The initial X-ray structure of hexameric HIV-1 CA was downloaded from the protein databank (PDB code 5HGL with a resolution of 3.1 Å) [21]. Only one monomer of the hexameric structure was used for the entire MD study. This structure, which is used in the binding assay, has four mutated amino acids (M185A, E45C, A14C and W184A). 5HGL structure misses amino acids Ala88, Gly89, Pro90, Ile91, Ala92, Pro93, Gly94, Gln95, Lys182, Asn183, Ala184, and Ala185. Accordingly, we used HIV-1 CA X-ray structure 3GV2 to extract the missed amino acids with their corresponding tertiary structure and add them to 5HGL structure after their alignment using discovery studio software [39]. Then, the whole system was energy minimized to remove any strains due to the added amino acid residues, which is then used for further study.</p><p>13m structure was sketched by discovery studio in the S-configuration, and no ionizable groups were detected. Then, its conformers were generated by OMEGA module of OPENEYE Scientific Software Inc. using default parameters [40, 41]. 13p conformers were docked into the active site of the modeled 5HGL structure using OPENEYE Scientific Software Inc. module OEDOCKING 3.0.1 with chemgauss4 scoring function [42-45]. The docking procedure resulted in only one docked conformer to the active site of the X-ray structure. We used this structure, i.e the docked 13m to the modified 5HGL, in the subsequent molecular dynamics simulation.</p><!><p>Atomic point charges for 13m were derived from AM1-BCC charge model with ANTECHAMBER module of AMBER14 [46]. Coordinate and topology parameters were prepared using ff14sb force field for HIV-1 CA monomer and GAFF force field for 13m. The whole system was solvated in explicit water TIP3PBOX octahedral solvent box model with 9Å cut, and the system was neutralized by Na+ ions. Water was minimized for 10000 cycles using steepest descent and then conjugate gradient algorithms. The whole system was then minimized for 5000 cycles using steepest descent followed by conjugate gradient algorithms. Then, water was equilibrated for 20 ps at constant volume and periodic boundaries with weak strength restraints on the whole system through the equilibrium stage, a force constant of 10 Å as position restraint. The whole system was then equilibrated for 40 ps using constant pressure periodic boundaries with no restraints. The minimized and equilibrated structure was used in the molecular dynamics simulation for 1 μs at constant temperature (300 K) and constant pressure (1 atm). Non-bonded forces were calculated at a Cutoff distance of 10 Å. H Mass Repartition was used to shift the mass of all hydrogen atoms of solute to 3.024 Da [47]. This allowed us to use an integration time step of 4 fs during MD [47]. SHAKE bond length constraint involving hydrogen atoms was turned on.</p><!><p>All frames were imaged and then water molecules and Na+ ions were stripped off. All frames were aligned against the first frame of the MD production using protein residues only. They were then clustered by DBSCAN algorithm [48] implemented in CPPTRAJ of AMBER14 on 13m using minimum points of 3 and epsilon of 2.5 with no frame orientation (no fit), which clustered all frames according to 13m to explore the binding site of 5HGL.</p><!><p>Human embryonic kidney 293T cells (a gift from Dr. Irwin Chaiken, Drexel University, Philadelphia, PA) were cultured in Dulbecco's Modified Eagle's Medium (DMEM), 10% Fetal Bovine Serum (FBS), 100 U/ml penicillin, 100 μg/ml streptomycin and 2mM L-glutamine. Human astroglioma U87 cells stably expressing CD4/CCR5 (obtained from Professor Hongkui Deng, Peking University, and Prof. Dan Littman, New York University, USA, through the AIDS Research and Reference Reagent Program, Division of AIDS, NIAID, NIH)[35, 49, 50] were cultured in DMEM supplemented with 10% FBS, 100 U/ml penicillin, 100 μg/ml streptomycin and 2 mM L-glutamine, 300μg/ml G418 (Thermo Scientific, Waltham, MA) and 1 μg/ml Puromycin (Thermo Scientific). All cells were incubated, unless otherwise stated, at 37°C in a humidified in a 5% CO2 air environment chamber.</p><!><p>A dual transfection of two plasmids (3:4 ratio of the viral backbone vector to the viral envelope vector) in HEK293T cells (0.8×106 cells/well) produced single-round infectious HIV-1B41 pseudotyped luciferase-reporter viruses1. The viral backbone vector is an Env-deficient HIV-1 pNL4-3-LucR+E- plasmid that carries the luciferase-reporter gene [51]. The viral envelope vector is a plasmid expressing the HIV-1 gp160B41 envelope [52, 53]. A calcium phosphate transfection (ProFection Mammalian Transfection System, Promega, Madison, WI) was used to co-transfect these two plasmids. Following a 5-hour incubation, the media containing the transfection reagents and the DNA was removed, the cells were washed with DMEM, and fresh culture media was added. 72 hours post-transfection, the pseudovirus containing supernatants were clarified, filtered, and stored at −80°C.</p><!><p>The details of the single-round HIV-1 infection assay for detecting viral infectivity have been published previously [51, 54, 55]. Briefly, U87.CD4.CCR5 (1.2 × 104 cells/well) target cells were seeded in 96-well luminometer-compatible tissue culture plates (Greiner bio-one). After a 24-hour incubation at 37°C, the compound or a DMSO vehicle control (Sigma) was mixed with pseudotyped virus and the mixture was added to the target cells. After a 48-hour 37°C incubation, the media was removed from each well and the cells were lysed by adding 50 μl/well of luciferase lysis buffer (Promega). Following one freeze-thaw cycle, 50ul/well of luciferase assay substrate (Promega) was added and a GloMax 96 microplate luminometer (Promega) was used to measure the luciferase activity of each well. Luciferase activity of virus produced with compound were normalized to the luciferase activity of virus produced from DMSO vehicle control treated cells. The compound-induced effects are manifested as a decrease in normalized luciferase activity which indicates a decrease in infectivity in the target cells when the compound was present.</p><!><p>Single-round infectious envelope-pseudotyped luciferase-reporter viruses were produced from 293T cells[35] in the presence of the compound (from 100 μM to 0.01 μM with the same DMSO concentration) or DMSO vehicle control (Sigma) and incubated for 72 hours at 37°C. The resulting pseudovirus-containing supernatants were clarified, filtered, and underwent one freeze-thaw cycle before being diluted ten-fold and used to infect U87.CD4.CCR5 target cells. Target cells with pseudotyped viruses were incubated for 48 hours at 37°C. Following the 48-hour incubation, the media was removed from each well and the cells were lysed by adding 50 μl/well of luciferase lysis buffer (Promega). Following one freeze-thaw cycle, 50ul/well of luciferase assay substrate (Promega) was added and a GloMax 96 microplate luminometer (Promega) was used to measure the luciferase activity of each well. Luciferase activity of virus produced with compound were normalized to the luciferase activity of virus produced from DMSO vehicle control treated cells. The compound-induced effects are manifested as a decrease in normalized luciferase activity which indicates a decrease in infectivity in the target cells when the compound was present.</p>

PubMed Author Manuscript

A Redox Auxiliary Strategy for Pyrrolidine Synthesis via Photocatalytic [3+2] Cycloaddition

Cycloaddition reactions can be used to efficiently assemble pyrrolidine rings that are significant in a variety of chemical and biological applications. We have developed a method for the formal cycloaddition of cyclopropyl ketones with hydrazones that utlizes photoredox catalysis to enable the synthesis of a range of structurally diverse pyrrolidine rings. The key insight enabling the scope of photoredox [3+2] cycloadditions to be expanded to C=N electrophiles was the use of a redox auxiliary strategy that allowed for photoreductive activation of the cyclopropyl ketone without the need for an exogenous tertiary amine co-reductant. These conditions prevent the deleterious reductive ring-opening of the cyclopropyl substrates, enabling a range of less-reactive coupling partners to participate in this cycloaddition.

a_redox_auxiliary_strategy_for_pyrrolidine_synthesis_via_photocatalytic_[3+2]_cycloaddition

Introduction<!>Results and Discussion<!>Conclusions<!>General procedure for [3 + 2] cycloadditions with hydrazones<!>Ethyl 5-(1-methyMH-imidazole-2-carbonyl)-1-(2-oxooxazolidin-3-yl)-3-phenylpyrrolidine-2-carboxylate (13)<!>Ethyl 3-(4-methoxyphenyl)-5-(1 -methyl-1H-imidazole-2-carbonyl)-1-(2-oxooxazolidin-3-yl)pyrrolidine-2-carboxylate (14)<!>Ethyl 3-(3-methoxyphenyl)-5-(1-methyl-1H-imidazole-2-carbonyl)-1-(2-oxooxazolidin-3-yl)pyrrolidine-2-carboxylate (15)<!>Ethyl 5-(1-methyl-1H-imidazole-2-carbonyl)-1-(2-oxooxazolidin-3-yl)-3-(4-(trifluoromethyl)phenyl)pyrrolidine-2-carboxylate (16)<!>Ethyl 3-(3,5-bis(trifluoromethyl)phenyl)-5-(1-methyl-1H-imidazole-2-carbonyl)-1-(2-oxooxazolidin-3-yl)pyrrolidine-2-carboxylate (17)<!>Ethyl 3-(4-bromophenyl)-5-(1 -methyl-1H-imidazole-2-carbonyl)-1 -(2-oxooxazolidin-3-yl)pyrrolidine-2-carboxylate (18)<!>Ethyl 5-(1-methyl-1H-imidazole-2-carbonyl)-1-(2-oxooxazolidin-3-yl)-3-(o-tolyl)pyrrolidine-2-carboxylate (19)<!>Ethyl 5-(1-methyl-1H-imidazole-2-carbonyl)-1-(2-oxooxazolidin-3-yl)-3-(pyridin-3-yl)pyrrolidine-2-carboxylate (20)<!>Ethyl 3-(furan-3-yl)-5-(1-methyl-1H-imidazole-2-carbonyl)-1-(2-oxooxazolidin-3-yl)pyrrolidine-2-carboxylate (21)<!>Diethyl 5-(1-methyMH-imidazole-2-carbonyl)-1-(2-oxooxazolidin-3-yl)-3-phenylpyrrolidine-2,4-dicarboxylate (22)<!>Diethyl 5-(1-methyMH-imidazole-2-carbonyl)-1-(2-oxooxazolidin-3-yl)pyrrolidine-2,3-dicarboxylate (23)<!>Ethyl 1-(2-oxooxazoMdin-3-yl)-3-phenyl-5-picolinoylpyrroMdine-2-carboxylate (24)<!>3-(2-(1-Methyl-1H-imidazole-2-carbonyl)-4-phenylpyrrolidin-1-yl)oxazolidin-2-one (25)<!>Ethyl 2-methyl-5-(1-methyl-1H-imidazole-2-carbonyl)-1-(2-oxooxazolidin-3-yl)-3-phenylpyrrolidine-2-carboxylate (26)<!>1-(Benzyloxy)-5-methyl-4-phenylpyrrolidin-2-yl)(1-methyl-1H-imidazol-2-yl)methanone (27)

<p>Pyrrolidines feature prominently in wide variety of bioactive natural products and pharmaceutical compounds.1 The importance of these ring systems has motivated the development of a wide variety of strategies for their construction.2 Cycloadditions have been particularly valued for their capacity to assemble structurally complex, densely functionalized pyrrolidines with high efficiency (Figure 1). The most well-developed cycloaddition methods for pyrrolidine synthesis are reactions of azomethine ylides and related 1,3-dipoles. 3 Somewhat less well-developed but nevertheless widely utilized are formal [3 + 2] cycloadditions between imines and donor-acceptor cyclopropanes.4</p><p>We speculated that photoredox catalysis might offer an alternate disconnection to complement these existing cycloaddition methodologies. Our laboratory recently demonstrated that diverse cyclopentanes are accessible via formal [3 + 2] cycloadditions between aryl cyclopropyl ketones and alkenes. 5 These methods exploit a dual Lewis acid–photoredox catalyst system that results in the photoreductive generation of a ring-opened distonic radical anion as the key reactive intermediate. We reasoned that if these intermediates could be intercepted by imines as reaction partners instead of alkenes, the resulting method would be an attractive new route to the production of five-membered nitrogen heterocycles. However, substituting highly polarized C=X bonds for olefins in reaction methods is rarely a straightforward process, and exploratory experiments demonstrated that the methods we had developed for the [3+2] synthesis of cyclopropanes did not smoothly translate to the synthesis of pyrrolidines. This objective therefore requires a modified approach. The results of our exploratory experiments towards this goal are reported herein.</p><!><p>Our initial investigations screened several combinations of cyclopropyl ketones (e.g., 1) and imine coupling partners (2) using conditions we had previously reported for the analogous photoredox [3 + 2] synthesis of cyclopentane rings. In those cases where new pyrrolidine products were observed, the yields were relatively low, and they were accompanied by the production of ring-opened compounds arising from the reductive cleavage of the cyclopropyl ketone (Scheme 1). To account for this observation, we conducted a more thorough analysis of the intermediates involved in this reaction. Visible light promoted excitation of Ru(bpy)32+ affords an excited state that undergoes reductive quenching by i-Pr2NEt. The resulting Ru(I) complex is a stronger reductant than Ru*(bpy)32+. We have speculated that the Ru(I) complex is the species that reduces the Lewis-acid-bound ketone to afford the activated distonic radical anion intermediate 4, which is the key intermediate leading to productive cycloaddition. However, i-Pr2NEt itself is also a competent terminal reductant in photoredox reactions.6 If the productive reaction between 4 and its reaction partner is relatively slow, we speculated that overreduction of 4, either by i-Pr2NEt or a related species derived from its oxidation7 could produce ring-opened product 5.</p><p>The role of i-Pr2NEt in our original reaction design was to quench Ru(bpy)32+ to produce a longer-lived, more strongly reducing Ru(I) complex. The desired [3 + 2] cycloaddition, however, is an isohypsic transformation overall, and thus a terminal reductant is not required for a balanced reaction scheme. Moreover, tertiary amines can be ligands for Lewis acidic metal centers, including the lanthanoid Lewis acids that have proven to be optimal in our studies. The formation of these Lewis acid-base complexes can be deleterious to the overall rate of dual Lewis acid-photoredox catalytic reactions.5b This analysis suggested that the reactivity of this system might be substantially improved by developing conditions that could operate without the need for an exogenous amine as a reductive quencher.</p><p>We reasoned that the need for the amine would be circumvented by designing a new reaction system with two key features: (1) a substrate with a more positive reduction potential than phenyl cyclopropyl ketones, and (2) a photocatalyst with an intrinsically more negative excited state reduction potential than Ru(bpy)32+.8 These changes would enable the design plan outlined in Scheme 2. Visible light photoexcitation of a strongly reducing photocatalyst to its triplet state would trigger one-electron reduction of a Lewis acid-activated cyclopropyl ketone (A). Reversible ring-opening of the ketyl radical (A·−) would afford distonic radical anion B·−. In the absence of a competitive reductant, we propose that the lifetime of this intermediate will be sufficiently long to undergo productive [3 + 2] cycloaddition with less-reactive imine reaction partners without deleterious overreduction. The resulting product radical anion (C·−) could result in the formation of product C by either participating in chain-propagating reduction of another equivalent of the Lewis acid bound cyclopropyl ketone substrate or by regenerating the photocatalyst by back electron transfer to its oxidized state. Key to the realization of this design plan would be the identification of a combination of cylcopropyl ketone and photocatalyst that could photogenerate the desired radical anion intermediate without the need for a tertiary amine co-reductant.</p><p>To satisfy the requirement for a more easily reduced substrate, we considered the use of a "redox auxiliary," which our laboratory recently defined as a cleavable moiety that can be temporarily installed to modulate the redox properties of a given substrate.9 Our initial demonstration of the utility of this concept was in the design of a photocatalytic [2 + 2] enone cycloaddition,9a and we hypothesized that this strategy might also prove to be enabling in the context of a [3 + 2] cycloaddition as well. In particular, we surmised that cyclopropyl heteroaryl ketones bearing additional coordination sites would exhibit a stronger binding affinity for a Lewis acid co-catalyst and thus exhibit a more pronounced change in reduction potential when coordinated. To test this hypothesis, we performed a series of cyclic voltammetry (CV) studies for three model arylcyclopropanes with and without added Lewis acid (Table 1). In all cases, an irreversible reduction was observed in the absence of Lewis acid. Approximate Ep/2 values were determined using the method described by Nicewicz.10 The most electron-rich ketone (2-imidazolyl) proved to be the hardest to reduce, and the most electron-deficient (2-pyridyl) the easiest. When these CV studies were repeated in the presence 1 equiv of Sc(OTf)3, new features were observed in each voltammogram, the apparent Ep/2 potentials of the heteroaromatic ketones shifted to significantly more positive reduction potentials than the phenyl auxiliary. This effect is consistent with the greater propensity of the chelating heteroaryl ketones to coordinate strongly to a Lewis acid catalyst. Encouragingly, the Lewis acid-shifted potentials of the heteroaryl ketones were both within the range of the excited state reduction potential of many heteroleptic and homoleptic iridium photocatalysts.</p><p>We therefore hypothesized that the direct, amine-free photoredox activation of aryl cyclopropyl ketones would be significantly facilitated by the use of a heteroaryl redox auxiliary group. To test this premise, we examined these ketones in a model [3 + 2] cycloaddition with styrene using Ir(ppy)2(dtbbpy)PF6 (E0(PC+/PC*) = −0.98 V vs. SCE) as photocatalyst and Sc(OTf)3 as a Lewis acid. While the parent phenyl cyclopropyl ketone was completely unreactive under these conditions (Table 1, entry 1), consistent with the electrochemical data, the 2-imidazolyl and 2-pyridyl cyclopropyl ketones afforded good yields of the [3 + 2] cycloadduct in the absence of reductive quencher (entries 2 and 3). The 2-imidazolyl ketone provided the highest yield, and we elected to continue our investigations using this auxiliary group.</p><p>We next turned our attention to identifying an optimal photocatalyst for the [3 + 2] cycloaddition reaction between cyclopropyl imidazolyl ketones and imines. Because the cyclopropyl ketone substrate would need to interact directly with the short-lived excited state of the photocatalyst, we predicted that its identity would have a significant effect on the rate of the cycloaddition. Thus, we prepared a series of photocatalysts spanning a range of excited-state redox potentials 11 and assessed their ability to promote a model [3 + 2] cycloaddition between imidazolyl ketone 9 and gyloxalate oxime ether 10 in the absence of an amine quencher (Table 2). As expected, there is a relationship between the excited state reduction potential of the photocatalyst and the rate of the reaction: in general, more strongly reducing photocatalysts provided higher yields. Interestingly, this was not true of the Ir photocatalysts with the most negative excited state reduction potentials (Ir(3-tBuppy)3 and Ir(ppy)3), which resulted in significantly diminished yields (entries 10 and 11). These two strongly reducing photocatalysts have a correspondingly much less positive ground-state reduction potential. Because the reaction mechanism proposed in Scheme 2 requires a reduction step to regenerate the active photocatalyst, this observation would be consistent with a change in ratedetermining step where catalyst turnover is slowed due to increased stabilization of the Ir(IV) oxidation state. Thus the optimal photocatalyst that emerged from this screen was not the most electron-rich photocatalyst but instead the homoleptic iridum complex bearing 4-CF3ppy ligands (entry 8).</p><p>Having identified an appropriate combination of redox auxiliary and photocatalyst that enables photoreductive direct activation of cyclopropyl ketones, we conducted a routine investigation of reaction parameters to optimize the yield of the pyrrolidine cycloadduct. The effect of key variables on the reaction are summarized in Table 3. Hydrazone 12 was a more efficient coupling partner than the analogous oxime ether, though a decrease in diastereoselectivity was observed. Using Yb(OTf)3 as a Lewis acid instead of Sc(OTf)3 provided an increase in diastereoselectivity (entry 6), which was further improved by performing the reaction in THF instead of CH2Cl2 and by increasing the loading of Yb(OTf)3 (Entries 4 and 5). Further improvements in rate and selectivity were observed upon the addition of a desiccant (MgSO4) and using a more intense light source (entries 2 and 3). Finally, there was an unexpected increase in d.r. at higher loadings of hydrazone to provide cycloadduct 13 in 85% yield and 5:1 d.r. after 8 h (entry 1).12 Control studies verified the requirement of all reaction components for productive reactivity.</p><p>Figure 2 summarizes experiments probing the scope of this process under optimized conditions. The scope with respect to the β-aryl substituent was relatively general. Electron-rich aryl rings were well tolerated in good yields and modest diastereoselectivities (14 and 15). Very electron-deficient aryl rings were also well tolerated in good yields and good to excellent diastereoselectivities (16 and 17). A potentially reactive aryl bromide was well tolerated under these conditions (18), and placing steric bulk in the ortho position of the aromatic ring was also well tolerated (19). Both electron-deficient and electron-rich heteroaryl-substituted cyclopropanes were viable substrates (20 and 21), though Lewis basic moieties such as pyridine rings that could compete for the Lewis acid co-catalyst result in reduced yields. Trisubstituted cyclopropanes exhibited excellent regioselectivity and good diastereoselectivities under these conditions (22), although these reactions were relatively sluggish. A more electron-deficient cyclopropane derivative with an ester substituent was also a viable substrate in this transformation (23). This result highlights the generality that can be accessed by exploiting radical intermediates: both electron-withdrawing and electron-donating substituents are well tolerated. As suggested by our earlier experiments while exploring redox auxiliaries, a 2-pyridyl cyclopropyl ketone also shows productive reactivity in this transformation (24). The substrate scope for imine derivatives proved to be significantly more limited. A formaldehyde-derived hydrazone was an excellent substrate under these conditions (25). A ketimine derived from ethyl pyruvate showed productive reactivity to generate a quaternary stereocenter (26), although the rate of this reaction was significantly diminished. Acetaldehyde-derived hydrazones produced intractable mixtures of products; however, the benzyl oxime ether derived from acetaldehyde was an excellent coupling partner in this transformation and afforded modest yields and high diastereoselectivities (27).</p><p>Finally, we investigated conditions that might be able to convert the cycloadducts to simpler pyrrolidines. Although several methods for the cleavage of hydrazines and hydrazides via the reduction of N–N bonds have been developed,13 the reductive scission of fully substituted hydrazines is an unsolved problem. An extensive screen of reported conditions revealed that the imidazolyl ketone was sensitive to these strongly reducing conditions. Alternatively, several reports have described the removal C-acylimidazolyl moieties via N-methylation of the imidazole followed by displacement of the resulting N-heterocyclic carbene to afford acids, esters, and amides.14 However, the nucleophilic hydrazine moiety proved to undergo competitive alkylation and complicated this cleavage strategy. Ultimately, we were unable to identify a set of reaction conditions that could controllably cleave either of these functional groups in synthetically useful yields. Future work to develop this study into a practical method for pyrrolidine synthesis, therefore, will focus on the investigation of conditions to engage alternate radicalophilic imine reaction partners that might result in more readily cleaved N-subtitutents. These studies will be facilitated by the central insight of the studies described above, which is that the use of a problematic tertiary amine co-reductant in photoredox reactions can be obviated by rational tuning of the relative electrochemical properties of the substrate and photocatalyst.</p><!><p>We have determined that tertiary amine co-reductants that are commonly utilized to facilitate electron transfer in photoredox catalysis can have unintended deleterious side effects. In particular, they can interfere with redox-neutral cycloaddition reactions by promoting undesired overreduction processes. We have shown that these issues can be circumvented by using a photocatalyst with a more negative excited state reduction potential in conjunction with a modified substrate that is more susceptible to one-electron reduction. We have shown that this strategy enables the photoreductively triggered formal [3+2] cycloaddition of aryl cyclopropyl ketones with relatively unreactive hydrazone radicalophiles. We anticipate that the implementation of these insights to other photoredox processes could have a similar beneficial impact on the scope of reactions in this increasingly important class of synthetically useful transformations.</p><!><p>A 25 mL Schlenk tube was charged with MgSO4 (100 wt%) and flame-dried under vacuum. A separate flame-dried vial was charged with cyclopropane (1 equiv), hydrazone (2 equiv), Yb(OTf)3 (1 equiv), Ir(4-CF3-ppy)3 (0.01 equiv) and THF (0.1 M). The vial was briefly sonicated until homogenous, the contents were transferred to the cooled Schlenk tube, and the Schlenk tube was sealed. The reaction mixture was degassed by three freeze-pump-thaw cycles and then backfilled with nitrogen. The reaction vessel was placed 15 cm away from a 34 W blue LED (Kessil) and irradiated for the indicated time. After this time, the reaction mixture was diluted with EtOAc and extracted with H2O. The aqueous phase was extracted once with EtOAc, and the combined organics were washed with 10% aq. K2CO3 and brine. The organic layer was then dried over Na2SO4, filtered, and concentrated under reduced pressure. The cycloadduct was then purified by column chromatography.</p><!><p>Reaction was carried out according to the general procedure with (1-methyl-1H-imidazol-2-yl)(2-phenylcyclopropyl)methanone (91.0 mg, 0.4 mmol), (E)-ethyl 2-((2-oxooxazolidin-3-yl)imino)acetate (149.0 mg, 0.8 mmol), Yb(OTf)3 (248.0 mg, 0.4 mmol), Ir(4-CF3-ppy)3 (3.4 mg, 0.004 mmol), and THF (4 mL). The reaction was complete after 8 hours giving the crude product as a yellow oil (5:1 d.r.). Product was purified twice by column chromatography (3:1:1 hexanes/EtOH/NEt3 then 2:3 acetone/pentanes) to give the pure product as a white solid (140 mg, 0.34 mmol, 85 % yield). Relative stereochemistry determined by single crystal X-ray crystallography.12 1H NMR (400 MHz, Chloroform-d) δ 7.31 – 7.14 (m, 5H), 7.14 (s, 1H), 7.04 (s, 1H), 5.69 (dd, J = 9.5, 8.4 Hz, 1H), 4.89 (d, J = 9.2 Hz, 1H), 4.25 (dtd, J = 21.1,8.6, 7.0 Hz, 2H), 4.13 – 3.97 (m, 2H), 4.05 (s, 3H), 3.91 (td, J = 8.9, 7.5 Hz, 1H), 3.74 (dq, J = 9.3, 6.4, 5.7 Hz, 1H), 3.66 (dq, J = 10.7, 7.1, 6.4 Hz, 1H), 3.10 (dt, J = 13.0, 8.5 Hz, 1H), 2.20 (ddd, J = 13.0, 9.6, 7.3 Hz, 1H), 0.79 (t, J = 7.1 Hz, 3H). 13C NMR (101 MHz, Chloroform-d) δ 189.2, 170.7, 155.5, 142.0, 140.9, 129.6, 128.4, 128.1, 127.0, 126.9, 67.7, 66.4, 62.3, 60.3, 46.4, 44.6, 36.1, 35.8, 13.6. M.p. 151–154.5 °C. HRMS (ESI) calculated for [C21H25N4O5]+ {M+H+} requires 413.1820, found 413.1820.</p><!><p>Reaction was carried out with a modified version of the general procedure with (2-(4-methoxyphenyl)cyclopropyl)(1-methyl-1H-imidazol-2-yl)methanone (104.8 mg, 0.41 mmol), (E)-ethyl 2-((2-oxooxazolidin-3-yl)imino)acetate (149.0 mg, 0.8 mmol), Yb(OTf)3 (76.0 mg, 0.123 mmol, 0.3 equiv.), Ir(4-CF3-ppy)3 (3.4 mg, 0.004 mmol), and THF (4 mL). The reaction was complete after 8 h giving the crude product as a yellow oil (2:1 d.r.). Product was purified twice by column chromatography (3:1:1 hexanes/EtOH/NEt3 then 2:3 acetone/pentanes) to give the pure product as a white solid (109 mg, 0.25 mmol, 60% yield). Major Diastereomer: 1H NMR (500 MHz, Chloroform-d) δ 7.18 (d, J = 8.7 Hz, 2H), 7.13 (d, J = 0.9 Hz, 1H), 7.07 (d, J = 0.8 Hz, 1H), 6.83 (d, J = 8.7 Hz, 2H), 5.58 (dd, J = 9.1, 1.7 Hz, 1H), 4.50 (d, J = 8.0 Hz, 1H), 4.28 – 4.14 (m, 4H), 4.08 (dt, J = 11.6, 7.8 Hz, 1H), 4.01 (s, 3H), 3.85 – 3.76 (m, 4H), 3.70 (dq, J = 10.7, 7.2 Hz, 1H), 3.65 – 3.59 (m, 1H), 2.67 (td, J = 12.1,9.1 Hz, 1H), 2.54 (ddd, J = 12.5, 7.6, 1.8 Hz, 1H), 0.85 (t, J = 7.2 Hz, 3H). 13C NMR (126 MHz, Chloroform-d) δ 188.9, 172.3, 158.8, 157.7, 143.6, 129.8, 129.2, 129.0, 127.7, 113.6, 69.0, 61.7, 60.5, 59.7, 55.3, 43.8, 41.0, 36.0, 29.5, 13.7. M.p. 186.1–184.3 °C. HRMS (ESI) calculated for [C22H27N4O6]+{M+H+} requires 443.1925, found 443.1925.</p><!><p>Reaction was carried out with the general procedure with (2-(3-methoxyphenyl)cyclopropyl)(1-methyl-1H-imidazol-2-yl)methanone (103.2 mg, 0.40 mmol), (E)-ethyl 2-((2-oxooxazolidin-3-yl)imino)acetate (149.0 mg, 0.8 mmol), Yb(OTf)3 (248.0 mg, 0.40 mmol), Ir(4-CF3-ppy)3 (3.4 mg, 0.004 mmol), and THF (4 mL). The reaction was complete after 8 hours giving the crude product as a yellow oil (3.5:1 d.r.). Product was purified twice by column chromatography (3:1:1 hexanes/EtOH/NEt3 then 2:3 acetone/pentanes) to give the pure product as a colorless oil (150.2 mg, 0.34 mmol, 84 % yield). 1H NMR (500 MHz, Chloroform-d) δ 7.16 – 7.10 (m, 2H), 7.03 (d, J = 0.9 Hz, 1H), 6.93 (dd, J = 2.5, 1.7 Hz, 1H), 6.85 (dt, J = 7.6, 1.3 Hz, 1H), 6.72 (ddd, J = 8.2, 2.6, 1.0 Hz, 1H), 5.66 (t, J = 8.9 Hz, 1H), 4.89 (d, J = 8.9 Hz, 1H), 4.32 – 4.18 (m, 2H), 4.10 (ddd, J = 9.4, 8.3, 7.2 Hz, 1H), 4.06 – 3.98 (m, 4H), 3.87 (td, J = 8.9, 7.0 Hz, 1H), 3.81 – 3.69 (m, 5H), 3.11 (dt, J = 13.1,8.7 Hz, 1H), 2.18 (ddd, J = 13.1, 9.1, 6.9 Hz, 1H), 0.83 (t, J = 7.1 Hz, 3H). 13C NMR (126 MHz, Chloroform-d) δ 189.3, 170.6, 159.4, 155.5, 142.6, 142.0, 129.6, 129.0, 127.0, 120.7, 113.5, 112.9, 67.6, 66.4, 62.3, 60.3, 55.2, 46.4, 44.7, 36.0, 35.8, 13.7. HRMS (ESI) calculated for [C22H27N4O6]+ {M+H+} requires 443.1925, found 443.1925.</p><!><p>Reaction was carried out with the general procedure with (1-methyl-1H-imidazol-2-yl)(2-(4-(trifluoromethyl)phenyl)cyclopropyl)methanone (121.3 mg, 0.41 mmol), (E)-ethyl 2-((2-oxooxazolidin-3-yl)imino)acetate (149.0 mg, 0.8 mmol), Yb(OTf)3 (248.0 mg, 0.40 mmol), Ir(4-CF3-ppy)3 (3.4 mg, 0.004 mmol), and THF (4 mL). The reaction was complete after 12 hours giving the crude product as a yellow oil (5:1 d.r.). Product was purified twice by column chromatography (3:1:1 hexanes/EtOH/NEt3 then 2:3 acetone/pentanes) to give the pure product as a white solid (168.1 mg, 0.35 mmol, 85 % yield). 1H NMR (500 MHz, Chloroform-d) δ 7.51 (d, J = 8.1 Hz, 2H), 7.45 (d, J = 8.2 Hz, 2H), 7.14 (d, J = 0.9 Hz, 1H), 7.05 (d, J = 0.8 Hz, 1H), 5.70 (t, J = 9.0 Hz, 1H), 4.96 (d, J = 8.9 Hz, 1H), 4.32 – 4.19 (m, 2H), 4.11 (ddd, J = 9.5, 8.3, 7.1 Hz, 1H), 4.05 (s, 3H), 4.06 – 3.96 (m, 1H), 3.94 (td, J = 8.9, 6.5 Hz, 1H), 3.81 – 3.67 (m, 2H), 3.15 (dt, J = 13.2, 8.9 Hz, 1H), 2.16 (ddd, J = 13.2, 9.0, 6.6 Hz, 1H), 0.80 (t, J = 7.1 Hz, 3H). 13C NMR (126 MHz, Chloroform-d) δ 188.9, 170.2, 155.5, 145.4, 141.9, 129.8, 129.2 (q, J = 32.3 Hz), 128.8, 127.2, 125.1 (q, J = 3.7 Hz), 124.2 (q, J = 272.0), 67.4, 66.1,62.4, 60.5, 46.4, 44.2, 36.1, 35.5, 13.6. 19F NMR (377 MHz, Chloroform-d) δ −62.6. M.p. 175.1-177.6 °C. HRMS (ESI) calculated for [C22H24F3N4O5]+ {M+H+} requires 481.1693, found 481.1694.</p><!><p>Reaction was carried out with the general procedure with (2-(3,5-bis(trifluoromethyl)phenyl)cyclopropyl)(1-methyl-1H-imidazol-2-yl)methanone (147.1 mg, 0.41 mmol), (E)-ethyl 2-((2-oxooxazolidin-3-yl)imino)acetate (149.0 mg, 0.8 mmol), Yb(OTf)3 (248.0 mg, 0.40 mmol), Ir(4-CF3-ppy)3 (3.4 mg, 0.004 mmol), and THF (4 mL). The reaction was complete after 14 hours giving the crude product as a yellow oil (9:1 d.r.). Product was purified twice by column chromatography (3:1:1 hexanes/EtOH/NEt3 then 2:3 acetone/pentanes) to give the pure product as a white solid (174.0 mg, 0.32 mmol, 78 % yield). 1H NMR (500 MHz, Chloroform-d) δ 7.85 (d, J = 1.5 Hz, 2H), 7.74 – 7.69 (m, 1H), 7.15 (d, J = 0.9 Hz, 1H), 7.05 (d, J = 0.9 Hz, 1H), 5.68 (dd, J = 9.5, 8.1 Hz, 1H), 5.00 (d, J = 8.3 Hz, 1H), 4.33 – 4.19 (m, 2H), 4.09 (ddd, J = 9.4, 8.4, 7.1 Hz, 1H), 4.04 (s, 3H), 4.02 – 3.92 (m, 2H), 3.84 – 3.70 (m, 2H), 3.21 (ddd, J = 13.4, 9.5, 8.7 Hz, 1H), 2.17 (ddd, J = 13.8, 8.2, 5.7 Hz, 1H), 0.84 (t, J = 7.1 Hz, 3H). 13C NMR (126 MHz, Chloroform-d) δ 188.6, 169.7, 155.6, 143.7, 141.8, 131.4 (q, J = 33.2 Hz), 129.8, 128.9 (q, J = 3.8 Hz), 127.2, 123.3 (q, J = 272.8 Hz), 120.9 (sept, J = 7.6 Hz), 67.3, 65.8, 62.4, 60.7, 46.5, 44.3, 36.0, 35.0, 13.5. 19F NMR (377 MHz, Chloroform-d) δ −62.9. M.p. 159.8 – 161.4 °C. HRMS (ESI) calculated for [C23H23F6N4O5]+ {M+H+} requires 549.1567, found 549.1570.</p><!><p>Reaction was carried out with the general procedure with (2-(4-bromophenyl)cyclopropyl)(1-methyl-1H-imidazol-2-yl)methanone (123.4 mg, 0.40 mmol), (E)-ethyl 2-((2-oxooxazolidin-3-yl)imino)acetate (149.0 mg, 0.8 mmol), Yb(OTf)3 (248.0 mg, 0.40 mmol), Ir(4-CF3-ppy)3 (3.4 mg, 0.004 mmol), and THF (4 mL). The reaction was complete after 8 hours giving the crude product as a yellow oil (5:1 d.r.). Product was purified twice by column chromatography (3:1:1 hexanes/EtOH/NEt3 then 2:3 acetone/pentanes) to give the pure product as a white solid (159 mg, 0.32 mmol, 85 % yield). 1H NMR (500 MHz, Chloroform-d) δ 7.37 (d, J = 8.5 Hz, 2H), 7.19 (d, J = 8.5 Hz, 2H), 7.14 (d, J = 0.9 Hz, 1H), 7.04 (d, J = 0.9 Hz, 1H), 5.67 (t, J = 9.0 Hz, 1H), 4.91 (d, J = 8.9 Hz, 1H), 4.31 – 4.19 (m, 2H), 4.9 (ddd, J = 9.4, 8.3, 7.1 Hz, 1H), 4.04 (s, 3H), 4.00 (ddd, J = 9.3, 8.3, 6.8 Hz, 1H), 3.84 (td, J = 9.0, 6.8 Hz, 1H), 3.81 – 3.69 (m, 2H), 3.11 (dt, J = 13.2, 8.8 Hz, 1H), 2.12 (ddd, J = 13.2, 9.1,6.7 Hz, 1H), 0.87 (t, J = 7.1 Hz, 3H). 13C NMR (126 MHz, Chloroform-d) δ 189.0, 170.3, 155.5, 141.9, 140.2, 131.2, 130.2, 129.7, 127.1, 120.8, 67.4, 66.1,62.3, 60.5, 46.4, 43.9, 36.0, 35.7, 13.7. M.p. 169.0–172.3 °C. HRSM (ESI) calculated for [C21H24BrN4O5]+ {M+H+} requires 491.0925, found 491.0925.</p><!><p>Reaction was carried out with the general procedure with (1-methyl-1H-imidazol-2-yl)(2-(o-tolyl)cyclopropyl)methanone (96.0 mg, 0.40 mmol), (E)-ethyl 2-((2-oxooxazolidin-3-yl)imino)acetate (149.0 mg, 0.8 mmol), Yb(OTf)3 (248.0 mg, 0.40 mmol), Ir(4-CF3-ppy)3 (3.4 mg, 0.004 mmol), and THF (4 mL). The reaction was complete after 12 hours giving the crude product as a yellow oil (4:1 d.r.). Product was purified by column chromatography (3:1:1 hexanes/EtOH/NEt3) and recrystallization (acetone/pentanes liquid/liquid diffusion) to give the pure product as a white crystalline solid (128.0 mg, 0.30 mmol, 75 % yield). 1H NMR (500 MHz, Chloroform-d) δ 7.33 – 7.30 (m, 1H), 7.14 (d, J = 0.9 Hz, 1H), 7.12 – 7.05 (m, 3H), 7.04 (d, J = 0.9 Hz, 1H), 5.69 (dd, J = 10.2, 7.4 Hz, 1H), 4.82 (d, J = 9.7 Hz, 1H), 4.35 – 4.18 (m, 3H), 4.09 – 3.95 (m, 5H), 3.69 (dq, J = 10.7, 7.1 Hz, 1H), 3.58 (dq, J =10.7, 7.1 Hz, 1H), 2.98 (dt, J = 12.5, 7.6 Hz, 1H), 2.38 (s, 3H), 2.32 (dt, J = 12.6, 9.8 Hz, 1H), 0.74 (t, J = 7.1 Hz, 3H). 13C NMR (126 MHz, Chloroform-d) δ 189.21, 171.0, 155.7, 142.1, 138.3, 136.7, 129.8, 129.6, 127.0, 127.0, 126.7, 126.0, 67.0, 66.9, 62.2, 60.3, 46.5, 40.5, 36.1, 35.2, 19.9, 13.5. M.p. 156.9–158.6 °C. HRMS (ESI) calculated for [C22H27N4O5]+ {M+H+} requires 427.1976, found 427.1978.</p><!><p>Reaction was carried out with a modified version of the general procedure with (1-methyl-1H-imidazol-2-yl)(2-(pyridin-3-yl)cyclopropyl)methanone (91.0 mg, 0.40 mmol), (E)-ethyl 2-((2-oxooxazolidin-3-yl)imino)acetate (149.0 mg, 0.8 mmol), Yb(OTf)3 (149.0 mg, 0.20 mmol), Ir(4-CF3-ppy)3 (3.4 mg, 0.004 mmol), and THF (4 mL). The reaction was complete after 8 hours giving the crude product as a yellow oil (5:1 d.r.). Product was purified twice by column chromatography (3:1:1 hexanes/EtOH/NEt3 then 3:1 acetone/pentanes) to give the pure product as a colorless oil in a mixture of inseparable diastereomers (62.8 mg, 0.15 mmol, 38 % yield, 6.5:1:0.5 dr). 1H NMR (500 MHz, Chloroform-d) δ 8.47 (dd, J = 2.4, 0.8 Hz, 1H), 8.44 (dd, J = 4.8, 1.6 Hz, 1H), 7.79 (dt, J = 8.0, 1.9 Hz, 1H), 7.21 (ddd, J = 8.0, 4.7, 0.8 Hz, 1H), 7.15 (d, J = 0.9 Hz, 1H), 7.05 (d, J = 0.9 Hz, 1H), 5.70 (t, J = 9.0 Hz, 1H), 4.98 (d, J = 8.6 Hz, 1H), 4.31 – 4.19 (m, 2H), 4.15 – 4.08 (m, 1H), 4.05 (s, 3H), 3.99 (ddd, J = 9.2, 8.3, 6.9 Hz, 1H), 3.86 (td, J = 8.8, 6.0 Hz, 1H), 3.82 – 3.70 (m, 2H), 3.17 (dt, J = 13.4, 9.1 Hz, 1H), 2.13 (ddd, J = 13.3, 8.7, 6.0 Hz, 1H), 0.85 (t, J = 7.2 Hz, 3H). 13C NMR (126 MHz, Chloroform-d) δ 188.9, 170.0, 155.5, 149.9, 148.5, 141.9, 136.8, 135.8, 129.8, 127.2, 123.3, 67.3, 65.8, 62.4, 60.6, 46.4, 41.7, 36.1, 35.3, 13.7. HRMS (ESI) calculated for [C20H24N5O5]+ {M+H+} requires 414.1772, found 414.1772.</p><!><p>Reaction was carried out with the general procedure with (2-(furan-3-yl)cyclopropyl)(1-methyl-1H-imidazol-2-yl)methanone (86.0 mg, 0.40 mmol), (E)-ethyl 2-((2-oxooxazolidin-3-yl)imino)acetate (149.0 mg, 0.8 mmol), Yb(OTf)3 (248.0 mg, 0.40 mmol), Ir(4-CF3-ppy)3 (3.4 mg, 0.004 mmol), and THF (4 mL). The reaction was complete after 8 hours giving the crude product as a yellow oil (4:1 d.r.). Product was purified twice by column chromatography (3:1:1 hexanes/Et0H/NEt3 then 2:3 acetone/pentanes) to give the product as a clear oil in a 4:1 mixture of two inseparable diastereomers (113.0 mg, 0.28 mmol, 70 % yield). Product decomposes at room temperature. Major Diastereomer: 1H NMR (500 MHz, Chloroform-d) δ 7.31 – 7.26 (m, 2H), 7.13 (d, J = 0.9 Hz, 1H), 7.05 (d, J = 0.9 Hz, 1H), 6.39 (dd, J = 2.0, 0.9 Hz, 1H), 5.58 (t, J = 8.8 Hz, 1H), 4.76 (d, J = 8.3 Hz, 1H), 4.31 – 4.13 (m, 3H), 4.4 (s, 3H), 3.93 (q, J = 7.4 Hz, 2H), 3.81 (td, J = 8.2, 6.5 Hz, 1H), 3.04 (ddd, J = 13.0, 9.2, 8.1 Hz, 1H), 2.11 (ddd, J = 13.0, 8.4, 6.6 Hz, 1H), 1.03 (t, J = 7.2 Hz, 3H). 13C NMR (126 MHz, Chloroform-d) δ 189.20, 170.71, 155.60, 143.10, 142.52, 140.00, 129.59, 127.02, 124.19, 110.76, 67.35, 66.26, 62.26, 60.46, 46.38, 36.06, 35.72, 35.34, 13.82. HRMS (ESI) calculated for [Ci9H23N4O6]+ {M+H+} requires 403.1612, found 403.1609. Minor diastereomer: 1H NMR (500 MHz, Chloroform-d) δ 7.41 – 7.40 (m, 1H), 7.37 – 7.35 (m, 1H), 7.14 (d, J = 0.9 Hz, 1H), 7.07 (d, J = 0.9 Hz, 1H), 6.58 (dd, J = 2.0, 0.9 Hz, 1H), 5.65 (dd, J = 8.5, 5.8 Hz, 1H), 4.35 (d, J = 7.0 Hz, 1H), 4.24 – 4.10 (m, 3H), 4.09 – 4.00 (m, 1H), 4.00 (s, 3H), 3.98 – 3.93 (m, 1H), 3.81 – 3.74 (m, 1H), 3.57 (dt, J = 9.2, 7.1 Hz, 1H), 2.77 (dt, J = 12.9, 8.8 Hz, 1H), 2.23 (ddd, J = 13.0, 7.2, 5.8 Hz, 1H), 1.25 (t, J = 7.1 Hz, 3H). 13C NMR (126 MHz, Chloroform-d) δ 189.4, 171.2, 156.9, 142.9, 141.8, 139.4, 129.3, 127.5, 125.8, 110.0, 68.2, 64.9, 61.7, 61.3, 43.8, 36.4, 36.1,33.3, 14.1.</p><!><p>Reaction was carried out with the general procedure with ethyl 2-(1-methyl-1H-imidazole-2-carbonyl)-3-phenylcyclopropanecarboxylate (119.0 mg, 0.40 mmol), (E)-ethyl 2-((2-oxooxazolidin-3-yl)imino)acetate (149.0 mg, 0.8 mmol), Yb(OTf)3 (248.0 mg, 0.40 mmol), Ir(4-CF3-ppy)3 (3.4 mg, 0.004 mmol), and THF (4 mL). The reaction was stopped after 24 hours giving the crude product as a yellow oil (7:1 d.r.). Product was purified twice by column chromatography (3:1:1 hexanes/EtOH/NEt3 then 2:3 acetone/pentanes) to give the product as colorless oil (73.6 mg, 0.15 mmol, 38 % yield). 1H NMR (500 MHz, Chloroform-d) δ 7.40 – 7.36 (m, 2H), 7.30 – 7.25 (m, 3H), 7.23 – 7.18 (m, 1H), 7.17 (s, 1H), 7.08 (s, 1H), 5.96 (d, J = 8.7 Hz, 1H), 5.05 (d, J = 9.2 Hz, 1H), 4.25 – 4.17 (m, 1H), 4.14 – 4.00 (m, 8H), 3.78 – 3.63 (m, 2H), 3.63 – 3.53 (m, 2H), 1.09 (t, J = 7.1 Hz, 3H), 0.79 (t, J = 7.1 Hz, 3H). 13C NMR (126 MHz, Chloroform-d) δ 187.5, 171.5, 169.6, 155.4, 142.7, 140.0, 130.0, 128.7, 128.3, 127.6, 127.3, 67.5, 66.2, 62.0, 61.2, 60.5, 52.0, 47.3, 36.0, 29.7, 13.9, 13.56. HRMS (ESI) calculated for [C24H29N4O7]+ {M+H+} requires 485.2031, found 485.2032.</p><!><p>Reaction was carried out with a modified version of the general procedure with ethyl 2-(1-methyl-1H-imidazole-2-carbonyl)cyclopropanecarboxylate (89.0 mg, 0.40 mmol), (E)-ethyl 2-((2-oxooxazolidin-3-yl)imino)acetate (372.0 mg, 2.0 mmol), Sc(OTf)3 (98.0 mg, 0.20 mmol), Ir(4-CF3-ppy)3 (3.4 mg, 0.004 mmol), and MeCN (4 mL). The reaction was complete after 20 hours giving the crude product as a yellow oil (1.2:1 d.r.). Product was purified twice by column chromatography (3:1:1 hexanes/EtOH/NEt3 then 2:3 acetone/pentanes) to give the pure product in a 1.2:1 ratio of separable diastereomers (83.0 mg, 0.20 mmol, 51 % yield). 1H NMR (500 MHz, Chloroform-d) δ 7.12 (d, J = 0.9 Hz, 1H), 7.06 (d, J = 0.9 Hz, 1H), 5.53 (dd, J = 9.3, 2.3 Hz, 1H), 4.46 (d, J = 7.8 Hz, 1H), 4.23 – 4.09 (m, 7H), 3.98 (s, 3H), 3.67 – 3.59 (m, 2H), 2.69 (ddd, J = 12.8, 11.0, 9.3 Hz, 1H), 2.48 (ddd, J = 12.8, 8.4, 2.3 Hz, 1H), 1.33 – 1.19 (m, 6H). 13C NMR (126 MHz, Chloroform-d) δ 188.70, 171.14, 171.05, 157.14, 143.14, 129.37, 127.69, 64.24, 61.70, 61.24, 61.02, 60.00, 43.86, 41.93, 35.99, 27.87, 14.10, 14.05. HRMS (ESI) calculated for [C18H25N407]+ {M+H+} requires 409.1718, found 409.1719. Minor diastereomer: Colorless oil. 1H NMR (500 MHz, Chloroform-d) δ 7.14 (d, J = 0.9 Hz, 1H), 7.04 (d, J = 0.9 Hz, 1H), 5.44 (dd, J = 8.7, 7.9 Hz, 1H), 4.55 (d, J = 8.3 Hz, 1H), 4.28 (ddd, J = 9.4, 8.6, 6.1 Hz, 1H), 4.25 – 4.15 (m, 3H), 4.15 – 4.08 (m, 2H), 4.02 – 3.94 (m, 4H), 3.79 (ddd, J = 9.2, 8.3, 6.1 Hz, 1H), 3.64 (q, J = 8.2 Hz, 1H), 2.93 (dt, J = 12.8, 7.8 Hz, 1H), 2.42 (dt, J = 12.8, 8.6 Hz, 1H), 1.27 (t, J = 7.1 Hz, 3H), 1.22 (t, J = 7.1 Hz, 3H). 13C NMR (126 MHz, Chloroform-d) δ 188.6, 171.0, 170.8, 155.7, 142.0, 129.6, 127.0, 67.2, 65.4, 62.2, 60.9, 60.9, 46.5, 44.5, 36.0, 31.2, 14.1, 14.1. HRMS (ESI) calculated for [C18H25N4O7]+ {M+H+} requires 409.1718, 409.1718.</p><!><p>Reaction was carried out with a modified version of the general procedure with (2-phenylcyclopropyl)(pyridin-2-yl)methanone (89.4 mg, 0.40 mmol), (E)-ethyl 2-((2-oxooxazolidin-3-yl)imino)acetate (224.0 mg, 1.20 mmol), Sc(OTf)3 (99.0 mg, 0.20 mmol, 0.5 equiv.), Ir(4-CF3-ppy)3 (3.4 mg, 0.004 mmol), and THF (4 mL). The reaction was complete after 8 hours giving the crude product as a yellow oil (20:1 d.r.). Product was purified twice by column chromatography (3:1:1 hexanes/EtOH/NEt3 then 1:3 acetone/pentanes) to give the pure product as a colorless oil in a 20:1 mixture of two inseparable diastereomers (65.6 mg, 0.16 mmol, 40 % yield). Product decomposes slowly at room temperature. 1H NMR (500 MHz, Chloroform-d) δ 8.65 (ddd, J = 4.8, 1.7, 0.9 Hz, 1H), 8.08 (d, J = 7.9 Hz, 1H), 7.84 (td, J = 7.7, 1.7 Hz, 1H), 7.46 (ddd, J = 7.5, 4.7, 1.2 Hz, 1H), 7.32 – 7.26 (m, 2H), 7.23 (t, J = 7.6 Hz, 2H), 7.20 – 7.12 (m, 1H), 5.89 (t, J = 9.1 Hz, 1H), 4.96 (d, J = 9.1 Hz, 1H), 4.32 −4.20 (m, 2H), 4.19 – 4.11 (m, 2H), 3.89 (td, J = 9.0, 6.7 Hz, 1H), 3.79 – 3.65 (m, 2H), 3.18 (dt, J = 13.1,8.9 Hz, 1H), 2.06 (ddd, J = 13.1,9.5, 6.8 Hz, 1H), 0.81 (t, J = 7.1 Hz, 3H). 13C NMR (126 MHz, Chloroform-d) δ 198.2, 170.5, 155.6, 152.5, 149.1, 141.2, 136.8, 128.4, 128.1, 127.4, 126.9, 122.3, 67.4, 66.0, 62.5, 60.3, 46.5, 44.5, 35.5, 13.6. HRMS (ESI) calculated for [C22H24N3O5]+ {M+H+} requires 410.1711, found 410.1707.</p><!><p>Reaction was carried out with the general procedure with (1-methyl-1H-imidazol-2-yl)(2-phenylcyclopropyl)methanone (91.0 mg, 0.40 mmol), 3-(methyleneamino)oxazolidin-2-one (91.0 mg, 0.8 mmol), Yb(OTf)3 (248.0 mg, 0.40 mmol), Ir(4-CF3-ppy)3 (3.4 mg, 0.004 mmol), and THF (4 mL). The reaction was complete after 12 hours giving the crude product as a yellow oil (5:1 d.r.). Product was purified twice by column chromatography (3:1:1 hexanes/Et0H/NEt3 and 9:1 EtOAc/pentanes) and to give the product in a 13:1 mixture of diastereomers as a white solid (95.0 mg, 0.28 mmol, 70 % yield). 1H NMR (500 MHz, Chloroform-d) δ 7.33 – 7.27 (m, 4H), 7.23 – 7.17 (m, 1H), 7.14 (d, J = 0.9 Hz, 1H), 7.04 (d, J = 0.9 Hz, 1H), 5.59 (dd, J = 10.6, 6.7 Hz, 1H), 4.33 – 4.20 (m, 2H), 4.03 (s, 3H), 3.89 – 3.72 (m, 4H), 3.48 (dd, J = 9.0, 7.4 Hz, 1H), 2.92 (dt, J = 12.0, 6.8 Hz, 1H), 2.08 (dt, J = 12.6, 10.3 Hz, 1H). 13C NMR (126 MHz, Chloroform-d) δ 190.1, 155.9, 142.8, 142.4, 129.5, 128.5, 127.3, 127.1, 126.6, 67.6, 61.6, 58.3, 44.7, 42.5, 38.2, 36.1. HRMS (ESI) calculated for [C18H21N4O3]+ {M+H+} requires 341.1608, found 341.1606.</p><!><p>Reaction was carried out with the general procedure with (1-methyl-1H-imidazol-2-yl)(2-(pyridin-3-yl)cyclopropyl)methanone (91.0 mg, 0.40 mmol), ethyl 2-((2-oxooxazolidin-3-yl)imino)propanoate (160.0 mg, 0.8 mmol), Yb(0Tf)3 (248.0 mg, 0.40 mmol), Ir(4-CF3-ppy)3 (3.4 mg, 0.004 mmol), and THF (4 mL). The reaction was quenched after 72 hours giving the crude product as a yellow oil (>10:1 d.r.). Product was purified twice by column chromatography (8:1:1 hexanes/EtOH/NEt3 then 2:1 acetone/pentanes) to give the product as a colorless oil with minor impurities (52.9 mg, 0.12 mmol, 31 % yield, >10:1 dr; 65 % RSM). 1H NMR (500 MHz, Chloroform-d) δ 7.26 – 7.17 (m, 5H), 7.13 (d, J = 0.9 Hz, 1H), 7.02 (d, J = 0.9 Hz, 1H), 6.0 (dd, J = 9.7, 7.6 Hz, 1H), 4.30 – 4.24 (m, 2H), 4.10 (q, J = 8.9 Hz, 1H), 4.4 (s, 3H), 3.89 – 3.82 (m, 1H), 3.80 – 3.73 (m, 1H), 3.72 – 3.64 (m, 1H), 3.52 (dd, J = 10.5, 7.3 Hz, 1H), 2.96 (dt, J = 12.4, 7.4 Hz, 1H), 2.48 (q, J = 10.4 Hz, 1H), 1.64 (s, 3H), 0.88 (t, J = 7.1 Hz, 3H). 13C NMR (126 MHz, Chloroform-d) δ 189.9, 173.2, 157.0, 142.0, 139.0, 129.5, 128.3, 128.1, 127.2, 126.8, 73.7, 64.4, 62.0, 60.5, 54.0, 47.3, 36.1, 33.4, 20.2, 13.6. HRMS (ESI) calculated for [C22H26N5O5]+ {M+H+} requires 427.1976, found 427.1976.</p><!><p>Reaction was carried out with a modified version of the general procedure with (1-methyl-1H-imidazol-2-yl)(2-phenylcyclopropyl)methanone (91.0 mg, 0.40 mmol), (E)-acetaldehyde O-benzyl oxime (298.0 mg, 2.0 mmol), Sc(OTf)3 (98.0 mg, 0.20 mmol), Ir(4-CF3-ppy)3 (3.4 mg, 0.004 mmol), and CH2Cl2 (4 mL). The reaction of irradiated with a 8 W blue LED strip. The reaction was stopped after 72 hours giving the crude product as a yellow oil (10:1 d.r.). Product was purified by column chromatography (1:4 acetone/pentanes) to give the product as a colorless oil as an inseparable mixture of two diastereomers (90.2 mg, 0.24 mmol, 60 % yield, 10:1 dr). 1H NMR (500 MHz, Chloroform-d) δ 7.54 (d, J = 7.0 Hz, 2H), 7.36 (t, J = 7.4 Hz, 2H), 7.33 – 7.19 (m, 4H), 7.6 – 7.00 (m, 3H), 4.51 (d, J = 10.2 Hz, 1H), 4.25 (d, J = 10.2 Hz, 1H), 4.16 – 4.04 (m, 2H), 4.01 (s, 3H), 3.42 (dq, J = 9.3, 6.1 Hz, 1H), 2.33 – 2.20 (m, 2H), 1.33 (d, J = 6.1 Hz, 3H). 13C NMR (126 MHz, Chloroform-d) δ 193.0, 142.9, 141.8, 137.2, 129.4, 128.8, 128.4, 128.2, 128.1, 127.7, 127.5, 127.4, 77.3, 71.0, 65.3, 48.2, 36.3, 34.3, 18.3. HRMS (ESI) calculated for [C23H26N3Oe]+ {M+H+} requires 376.2020, found 376.2016.</p>

PubMed Author Manuscript

Roles of Arginine and Lysine Residues in the Translocation of a Cell-Penetrating Peptide from 13C, 31P and 19F Solid-State NMR

Cell-penetrating peptides (CPPs) are small cationic peptides that cross the cell membrane while carrying macromolecular cargoes. We use solid-state NMR to investigate the structure and lipid interaction of two cationic residues, Arg10 and Lys13, in the CPP penetratin. 13C chemical shifts indicate that Arg10 adopts a rigid \xce\xb2-strand conformation in the liquid-crystalline state of anionic lipid membranes. This behavior contrasts with all other residues observed so far in this peptide, which adopt a dynamic \xce\xb2-turn conformation with coil-like chemical shifts at physiological temperature. Low-temperature 13C-31P distances between the peptide and the lipid phosphates indicate that both the Arg10 guanidinium C\xce\xb6 and the Lys13 C\xce\xb5 lie in close proximity to the lipid 31P (4.0 - 4.2 \xc3\x85), proving the existence of charge-charge interaction for both Arg10 and Lys13 in the gel-phase membrane. However, since lysine substitution in CPPs are known to reduce their translocation ability, we propose that low temperature stabilizes both lysine and arginine interactions with the phosphates, whereas at high temperature the lysine-phosphate interaction is much weaker than the arginine-phosphate interaction. This is supported by the unusually high rigidity of the Arg10 sidechain and its \xce\xb2-strand conformation at high temperature. The latter is proposed to be important for ion pair formation by allowing close approach of the lipid headgroups to guanidinium sidechains. 19F and 13C spin diffusion experiments indicate that penetratin is oligomerized into \xce\xb2-sheets in gel-phase membranes. These solid-state NMR data indicate that guanidinium-phosphate interactions exist in penetratin, and guanidinium groups play a stronger structural role than ammonium groups in the lipid-assisted translocation of CPPs across liquid-crystalline cell membranes.

roles_of_arginine_and_lysine_residues_in_the_translocation_of_a_cell-penetrating_peptide_from_13c,_3

<!>Materials and Methods<!>Solid-state NMR experiments<!>Arg10 conformation and dynamics in penetratin<!>13C-31P distances between penetratin and lipid headgroups<!>Oligomeric structure of penetratin in the lipid membrane<!>Interaction of charged sidechains in penetratin with lipid phosphates

<p>Cell-penetrating peptides (CPP) are arginine- and lysine-rich cationic peptides that can readily enter cells not only by themselves but also carrying other macromolecular cargos (1-3). Thus they are promising drug-delivery molecules. Many studies have established that the intracellular entry of CPPs is related to their strong affinity to lipid bilayers (4). The lipid membrane can be the plasma membrane of the cell or the endosomal membrane from which CPPs must escape after endocytosis (5). The fundamental biophysical question of interest is how these highly cationic peptides cross the hydrophobic part of the lipid bilayer against the free energy barrier, and doing so without causing permanent damage to the membrane, in contrast to another family of cationic membrane peptides, antimicrobial peptides (AMPs).</p><p>Several models have been proposed to explain the membrane translocation of CPPs. The inverse micelle model proposes that transient inverse micelles form in the membrane to trap the peptides from the outer leaflet and subsequently release them to the inner leaflet (6, 7). However, this model is inconsistent with the lipid 31P spectra (8), and the large rearrangement of lipids is difficult to achieve energetically. The electroporation model (9) posits that at low concentrations CPPs bind only to the outer leaflet of the bilayer, thus causing a transmembrane electric field. Above a threshold peptide concentration, the membrane is permeabilized in a electroporationlike manner, which creates transient defects that enable the peptides to distribute to both leaflets, thus relieving the membrane curvature stress (9-11). The third model posits that the guanidinium ions in these arginine-rich peptides associate with the lipid phosphate groups to neutralize the arginine residues and thus allow the peptides to cross the membrane without a high energy penalty. This model is supported by phase transfer experiments of oligoarginnines (12) and by molecular dynamics simulation of the HIV-1 Tat peptide, which showed transient association of arginine residues with the phosphate groups on both sides of the bilayer (13).</p><p>We recently investigated the depth of insertion and conformation of a CPP, penetratin, using solid-state 13C and 31P NMR. Penetratin is the first discovered CPP and is derived from the third helix (residues 43-58) of the Drosophilia Antennapedia homeodomain (14). Using Mn2+ paramagnetic relaxation enhancement (PRE) experiments, we showed that penetratin is bound to both leaflets of the lipid bilayer at both low and high concentrations (peptide-lipid molar ratios of 1:40 and 1:15) (15, 16). This data indicates that the electroporation model is unlikely for penetratin. No 31P peaks at the isotropic frequency or the hexagonal-phase frequency were observed, thus ruling out the inverse micelle model. In addition, we found that penetratin undergoes an interesting conformational change, as manifested by 13C chemical shifts, from a β-sheet structure in the gel-phase membrane to a coil-like conformation in the liquid-crystalline membrane (17). The coil-like conformation at high temperature has non-negligible residual order parameters of 0.23 - 0.52 (17), indicating that the peptide remains structured. We hypothesized that the high-temperature conformation is a β-turn that undergoes uniaxial rotation around the bilayer normal.</p><p>Given the above experimental evidence against the electroporation model and the inverse micelle model, we now test the validity of the guanidinium-phosphate complexation model for the membrane translocation of penetratin. For this purpose we measured 13C-31P distances between several peptide sidechains and lipid 31P atoms. As shown before for an arginine-rich antimicrobial peptide, strong associations with the lipid phosphates manifest as short 13C-31P distances (18, 19). We show here that the cationic Arg10 in penetratin indeed exhibits shorter distances to phosphate groups than hydrophobic residues. However, another cationic residue, Lys13, also exhibits short 13C-31P distances, despite the fact that Lys mutants of CPPs have much weaker translocation activities. We show that the answer to this puzzle lies not in the low-temperature structure and distances of the two residues, but in their high-temperature dynamic structures, which differ significantly. And it is the structure in the liquid-crystalline membrane that accounts for the distinct roles of Arg and Lys in CPP entry into the cell. Finally, we investigate the oligomeric structure of penetratin in gel-phase membranes using 19F and 13C spin diffusion NMR.</p><!><p>All lipids were purchased from Avanti Polar Lipids (Alabaster, AL) and used without further purification. Penetratin (RQIKI WFQNR RMKW KK), which contains three arginines and four lysines, was synthesized using standard Fmoc solid-phase peptide synthesis methods (20). Uniformly 13C, 15N-labeled arginine was purchased from Cambridge Isotope Laboratory and incorporated into Arg10 in the peptide. Ile3 and Lys13 were labeled in two other peptide samples as described before (17). 4-19F-Phe7 labeled penetratin was used for 19F experiments to determine the oligomeric structure. All peptide samples were purified by HPLC to > 95% purity.</p><p>Hydrated membrane samples were prepared using an aqueous-phase mixing method. Lipids were first codissolved in chloroform at the desired molar ratios and dried under a stream of N2 gas. After lyophilization in cyclohexane overnight, the dry lipid powder was suspended in water and freeze-thawed several times before the peptide solution was added. The solution was incubated overnight to facilitate binding, then centrifuged at 55,000 rpm for 3 hours to obtain a hydrated membrane pellet. For the Arg10 experiments, a hydrated DMPC/DMPG (8:7) membrane with a peptide : lipid molar ratio of 1:15 was used in most 2D correlation and distance measurements. The molar ratio was chosen to balance the positive charges of the peptide (+7) by the negatively charged PG lipids (-1). This DMPC/DMPG sample is supplemented by a hydrated POPC/POPG (8:7) sample for the conformation study, and by a trehalose-protected dry POPE/POPG (8:7) sample for the 13C-31P REDOR experiment. For distance measurements of other residues, several trehalose-protected dry DMPC/DMPG membranes were used to ensure that both the peptide and the lipid headgroups are completely immobilized at low temperature (21). Trehalose is known to protect the lamellar structure of the lipid bilayer in the absence of water (22). Below we refer to the non-trehalose containing membrane samples as hydrated samples to distinguish from the dry trehalose-containing samples.</p><!><p>All experiments were carried out on a Bruker DSX-400 (9.4 Tesla) spectrometer (Karlsruhe, Germany) at a resonance frequency of 100.7 MHz for 13C, 376.8 MHz for 19F and 162.1 MHz for 31P. Magic-angle-spinning (MAS) probes tuned to 1H/13C/31P and 1H/19F/X and equipped with 4 mm spinning modules were used for all experiments. Low temperature was achieved using a Kinetics Thermal System XR air-jet sample cooler (Stone Ridge, NY). The temperature of the sample was read from a thermocouple placed near the rotor and was not further calibrated. Typical 90° pulse lengths are 3.5-5.0 μs. 1H decoupling fields of 50-80 kHz were used. 13C, 31P and 19F chemical shifts were referenced externally to the α-Gly 13C' resonance at 176.49 ppm on the TMS scale, the hydroxyapatite 31P signal at 2.73 ppm and the Teflon 19F signal at -122 ppm, respectively.</p><p>13C cross polarization (CP) MAS experiments were conducted with a contact time of 0.5-1.0 ms at a typical Hartman-Hahn field strength of 50 kHz. For variable-temperature experiments, samples were stabilized for at least 20 min at each temperature before data acquisition. 2D 13C-13C dipolar assisted rotational resonance (DARR) experiments (23) were conducted under 5 kHz MAS and with a mixing time of 20 ms and 30 ms. A 1H-driven spin diffusion experiment with a longer mixing time of 50 ms was used to detect inter-residue cross peaks of penetratin in DMPC/DMPG membranes.</p><p>13C-1H and 15N-1H dipolar couplings were measured using DIPSHIFT (24, 25) experiments at 303 K under 3.401 and 3.000 kHz MAS, respectively. The MREV-8 sequence was used for 1H homonuclear decoupling (26), with an 105° 1H pulse length of 4.0 μs. In the C-H DIPSHIFT experiment, the 13C-13C dipolar coupling is removed by MAS while the 13C-13C scalar coupling has no effect on the t1-dependent intensity modulation due to the constant time nature of the evolution period. The normalized t1 intensities were fitted using a home-written Fortran program. The best-fit values were divided by the theoretical scaling factor of 0.47 for the MREV-8 sequence. For the doubled N-H DIPSHIFT experiment, the fit value was further divided by 2 to obtain the true couplings. The ratio between the true coupling and the rigid limit value gives the order parameter SXH. The rigid-limit coupling used was 22.7 kHz for C-H and 10.6 kHz for N-H dipolar couplings. Simulations took into account the difference between the XH and XH2 spin systems.</p><p>Frequency-selective rotational-echo double-resonance (REDOR) experiments were used to measure distances between peptide 13C and lipid 31P (27, 28). The experiments were conducted at ~230 K under 4 kHz MAS. A rotor-synchronized soft 13C Gaussian 180° pulse of 1000 μs was applied in the middle of the REDOR period to suppress the 13C-13C J-coupling between the on-resonance 13C and its directly bonded 13C spins. 31P 180° pulses of 9 μs were applied every half rotor period. The DMPC/DMPG sample were used to measure the Arg10 sidechain and CO distances to 31P, whereas the POPE/POPG sample was used to measure the Cα -31P distance, as the DMPC Cγ peak (54 ppm) overlaps with the Arg10 Cα signal.</p><p>A double-quantum (DQ) selective REDOR experiment (Figure 5a) was designed to measure the 13C-31P distance of Ile3 sidechains, whose signals overlap extensively with the lipid 13C peaks. An SPC-5 pulse train (29) was used to create DQ coherence of the labeled 13C sites and suppress the natural abundance lipid 13C signals. The efficiency of the DQ-REDOR experiment is about 20% of the single-quantum selective REDOR experiment.</p><p>For the 13CO-31P REDOR experiment, the DQ-REDOR experiment was not used due to the low sensitivity of the CO signal. Thus, the lipid natural-abundance contribution to the 13CO signal was corrected using the equation (S/S0)observed = 0.79(S/S0)peptide + 0.21(S/S0)lipid, where the weight fractions were obtained from the peptide-lipid molar ratio. At the low temperature used for the REDOR experiments, the lipid and peptide CO groups have very similar CP efficiencies, thus the natural abundance correction is relatively accurate.</p><p>All 13C-31P REDOR data were fit by two-spin simulations. As we showed before, for distances shorter than 5 Å, two-spin simulations are sufficient. For distances larger than 7 Å, the two-spin simulation only slightly overestimates the distances compared to the vertical distance from 13C to the 31P plane obtained from a multi-spin simulation (18).</p><p>The oligomeric structure of penetratin was determined using the 19F CODEX experiment (30, 31). The experiments were conducted at 233 K on a trehalose-protected DMPC/DMPG sample to freeze potential motion of the peptide. Two experiments were conducted for each mixing time: an exchange experiment (S) with the desired mixing time (τm) and a short z-filter (τz), and a reference experiment (S0) with interchanged τm and τz. The normalized intensity, S/S0, was measured as a function of the mixing time until it reached a plateau. The inverse of the equilibrium S/S0 value gives the minimum oligomeric number. Error bars were propagated from the signal-to-noise ratios of the isotropic peak and its sidebands. The τm-dependent CODEX curve was simulated as described before (32) to extract intermolecular distances.</p><!><p>In the present study, we focus on Arg10, one of the three arginine residues in penetratin, to understand whether cationic residues in general and arginine residues in particular play a special role in the membrane translocation of the peptide. We first investigate the conformation of Arg10. We recently reported the reversible conformational change of many penetratin residues in the lipid bilayer between a β-turn state at high temperature and a β-strand state at low temperature. This conformational change was manifested as chemical shift changes and was observed at Ile3, Ile5, Gln8, Asn9 and Lys13. The chemical shift change is independent of the membrane composition (POPC/POPG and DMPC/DMPG), anionic lipid fraction (PC/PG = 8:7 and 3:1), and peptide concentration (P/L = 1:15 and 1:30) (17).</p><p>Figure 1a shows the 2D 13C-13C correlation spectra of Arg10-labeled penetratin in DMPC/DMPG bilayers at 303 K and 234 K. Most intra-residue cross peaks are seen, and show no frequency differences between high and low temperatures. Thus, in contrast to all other residues examined, Arg10 does not have temperature-induced conformational change. The difference of the experimental isotropic chemical shifts from the random coil values reflects the secondary structure of the protein (33). Based on the 13CO, 13Cα and 13Cβ 13C isotropic shifts (Supporting Information Table S1), we find Arg10 adopts β-strand conformation at both high and low temperatures.</p><p>One-dimensional 13C CP-MAS spectra scanned between 303 K and 233 K (Figure 1b) confirm the lack of chemical shift changes in DMPC/DMPG bilayers. Moreover, the 1D spectra show that the penetratin 13C lines are broader at low temperature. This phenomenon is common to many membrane peptides (34, 35), and can be attributed to conformational distribution of the peptide in the lipid membrane, which is averaged at high temperature but frozen in at low temperature. For the sidechain Cδ signal, the largest line broadening is observed between 288 K and 263 K, below which the lines sharpen again. This is a definitive signature of intermediatetimescale motion, which means that at 303 K the Arg10 sidechain undergoes fast torsional motion. The lack of exchange broadening for the Cα and CO signals indicate that the Arg10 backbone is already in the slow motional limit at 303 K.</p><p>We also measured the Arg10 chemical shifts in POPC/POPG (8:7) membranes (Supporting Information Figure S1) and similarly found only β-strand chemical shifts in a wide temperature range. Figure 2 plots the 13C secondary chemical shifts of Arg10 at two temperatures in two different lipid membranes. The temperature-independent β-strand conformation of Arg10 differs from all other residues examined so far in penetratin (17).</p><p>The β-strand conformation is usually more rigid than a coil or turn conformation due to hydrogen bond constraints, and thus should have order parameters close to 1 (17, 36). To verify this, we measured the C-H dipolar couplings of various Arg10 segments in DMPC/DMPG bilayers. Figure 3a shows the 13C-1H DIPSHIFT curves of Cα and Cδ at 303 K. The backbone Cα-Hα dipolar coupling is 20.9 kHz, corresponding to an SCH of 0.92, which translates to a small motional amplitude of 13° (37). This order parameter fits into the SCH range of 0.89-0.94 measured for the other five residues when in the β-strand conformation. In comparison, the sidechain Cδ has a SCH of 0.42. While this value is much lower than the backbone due to the many torsional motions of the sidechain, it is actually larger than all other measured sidechain order parameters, which range from 0.23 to 0.37 in the β-sheet conformation (17). For comparison, the Lys13 sidechain Cε was previously found to have an SCH order parameter of 0.33 (17). We also measured the N-H dipolar coupling of the guanidinium Nη group, and found an N-H dipolar coupling of 3.2 ± 0.5 kHz (Figure 3b). This translates to an SNH of 0.30±0.05, which is significant considering this segment is six bonds away from the backbone Cα.</p><!><p>The main goal of the present study is to determine if the cationic residues in penetratin interact strongly with the negatively charged lipid phosphates. 13C-31P distance measurements can provide this information site-specifically. Figure 4 shows the 13C-31P REDOR data of Arg10, Lys13, and Ile3 sidechains in DMPC/DMPG membranes. The experiments were carried out at 233 K where the 31P chemical shift span is 195 - 198 ppm, corresponding to fully immobilized headgroups. The arginine Cζ and lysine Cεgroups directly neighbor the cationic amines and have well resolved chemical shifts of 157.0 ppm and 40.2 ppm, respectively, thus they are ideal reporters of the interaction of these sidechain ends with the lipid headgroups. Arg10 Cζ exhibits significant 13C-31P REDOR dephasing of S/S0 = 0.35 by 10 ms (Figure 4a), indicating relatively short distances to 31P. The time dependence of the REDOR intensities cannot be fit to a single distance due to the presence of a kink around 12 ms. Instead, a combination of a long distance of 6.0 Å and a short distance of 4.2 Å at a 1:1 ratio is found by a least-squares analysis to fit the data best (RMSD = 0.029). The short distance of 4.2 Å can only be satisfied if the guanidinium N-H groups are within hydrogen bonding distance with the O-P groups (18). Similarly, Lys13 Cε exhibits significant dephasing and the REDOR intensities are best fit by two distances of 4.0 Å and 5.5 Å (1:1) (RMSD = 0.036). Again, the short distance supports hydrogen bonding with the lipid phosphate groups. Details of the two-distance best fit for Arg10 and Lys13 are shown in Supporting Information Figure S2. Single-distance fitting of the Arg10 REDOR data indicate that the longer distance must be greater than 5.5 Å while the shorter distance must be smaller than 4.8 Å (Figure 4a). Further, the 13C-31P distances cannot be shorter than 3.6 Å due to steric constraints. Thus, two-distance REDOR curves were calculated using short distances of 3.6 - 4.8 Å with an increment of 0.2 Å and long distances of 5.4 - 7.2 Å with an increment of 0.3 Å, and the two contributions were averaged at a 1:1 ratio. The Lys13 data was analyzed similarly. The simulations indicate that the experimental uncertainties for these 13C-31P REDOR data are about ±0.2 Å for distances shorter than 5.5 Å and ±0.4 Å for distances longer than 5.5 Å.</p><p>Are the short 13C-31P distances of Arg10 and Lys13 specific to the cationic sidechains or are they also true for hydrophobic residues in penetratin? To answer this question, we measured the 13Cγ2-31P and 13Cδ±-31P distances of the neutral hydrophobic residue Ile3. Its Cδ is three bonds away from Cα, which is similarly separated from the backbone as lysine Cε. To remove the lipid natural abundance 13C signals that overlap with the Ile Cγ2 and Cδ1 peaks between 8.9 and 19.0 ppm, we designed a DQ selective REDOR experiment, whose pulse sequence is shown in Figure 5a. The DQ-selected spectra of Ile3 Cγ2 and Cδ1 signals are shown in Figure 5b (middle and bottom spectra). The REDOR dephasing of Ile3 is shown in Figure 4c. Much less REDOR decay is observed, with S/S0 values of ~0.80 at 16 ms. The data is best fit to a distance of 6.9 Å for both Cγ2 and Cδ1, which is 2.7-2.9 Å longer than the Arg10 Cζ and Lys13 Cε. Thus, the short 13C-31P distances are specific to arginine and lysine sidechains instead of being true for all sidechains.</p><p>We also measured the 13C-31P distances of Arg10 backbone Cα and CO, which are 6.8 Å and 7.8 Å, respectively (Figure 6). These values fall into the range of 6.9-8.2 Å previously measured for other residues (15). The 13CO data was corrected for the lipid natural abundance signals, whose systematic uncertainty is much smaller than the random noise of the data. All 13C-31P distances are summarized in Table 1.</p><!><p>The 19F CODEX experiment was used to determine the oligomeric number and intermolecular distances of penetratin in gel-phase membranes. Figure 7a shows the normalized exchange intensities of 4-19F-Phe7 penetratin in trehalose-protected DMPC/DMPG bilayers. The CODEX intensities decay to an equilibrium value of 0.35 by 2.5 s, indicating three-spin clusters detectable by the 19F distance ruler. To fit the decay trajectory quantitatively, we first assumed an equilateral triangle geometry for the three 19F spins (Figure 7b). The best-fit possible under this assumption gives an internuclear distance of 9.0 Å for each side of the triangle; however, the fit curve (dashed line) does not capture the fast initial decay of the experimental data. To better fit the bi-exponential nature of the data, we then used a triangular geometry with one short distance of much less than 9 Å and two distances longer than or comparable to 9 Å. Modeling of penetratin as a trimer of antiparallel β-strands (see below) yielded one distance of 6.0 Å and two distances of about 10 Å (Figure 7c), which were found to give excellent fit to the experimental data.</p><p>To further constrain the intermolecular packing of penetratin in the lipid membrane, we measured a 2D 1H-driven 13C spin diffusion spectrum with a mixing time of 50 ms. Figure 8 shows the 2D spectrum of Ile5, Gln8, Lys13-labeled penetratin in DMPC/DMPG bilayers at 249 K. Two inter-residue cross peaks were observed: I5α-K13α and Q8δ-I5α. Since the peptide adopts a β-strand conformation at this temperature, the intramolecular distances are ~27 Å and 11 Å for I5α-K13α and Q8δ-I5α, respectively, which are too long to be observed by 13C spin diffusion NMR. Thus, these cross peaks must result from intermolecular contacts, which are most likely less than 6 Å for the 50 ms mixing time used.</p><p>Using standard geometries for β-sheets, where inter-strand hydrogen bonds have RN-O distances of 2.8 - 3.4 Å and backbone torsion angles are ϕ = -139°, ψ = 135°, we built a β-sheet model for penetratin at low temperature that is consistent with the 19F and 13C spin diffusion data (38, 39). Three penetratin β-strands are arranged as a trimer, with the middle strand antiparallel to the two outer strands and shifted by one residue (Figure 9a). This arrangement gives inter-strand I5α-K13α distances of 4.4 - 5.8 Å, consistent with the 2D 13C spectrum. The three Phe7 rings point to the same side of the β-sheet, giving 19F-19F distances of 6.0 Å, 9.7 Å and 10 Å (Figure 9b). Short Q8δ-I5α distances cannot be satisfied within the same β-sheet, but requires two β-sheets stacked in parallel, with an inter-sheet distance of ~10 Å. This gives a Q8δ-I5α distance of 5.7 Å (Figure 9a), where the Q8 χ1 angle is -177°, which is the dominant rotamer of Gln in the β-sheet conformation (40).</p><!><p>The main finding of the current study is that an arginine and a lysine sidechain in penetratin both form close contacts with the lipid phosphates at low temperature. The Arg10 Cζ-P distance of 4.2 Å and Lys13 Cε-P distance of 4.0 Å both indicate the formation of N-H…O-P hydrogen bonds. Figure 10 shows the sidechain conformations of Arg10 and Lys13 and the spatial arrangements with a phosphate group that satisfy the distances measured here. The N-O distances in both cases must be less than 3.0 Å to satisfy the experimental Cζ and Cε distances to 31P.</p><p>The short distances of the Arg10 sidechain to the lipid 31P indicates that guanidinium-phosphate complexation occurs not only in antimicrobial peptides but also in cell-penetrating peptides, even though they differ in whether they cause permanent membrane damage. The similarity of lipid-peptide charge-charge attraction and hydrogen-bond formation suggests that penetratin, like some AMPs, also uses this interaction as the main mechanism for its function, which is crossing the lipid membrane. The fundamental driving force for the complex formation is the reduction in the free energy when a neutral species crosses the bilayer. The complexation entails that the peptide drags some lipid headgroups into the hydrophobic region of the membrane, thus causing membrane disorder. However, since no isotropic signal was observed in the 31P spectra of penetratin-containing POPC/POPG (8:7) membranes (15), the disorder is probably transient and not observable on the NMR timescale or under NMR experimental conditions. The 13C-31P distances must be measured at low temperature in the gel-phase membrane in order to freeze molecular motions that would average the dipolar couplings. At physiological temperature where motion is abundant and the penetratin structure is neither a canonical α-helix nor a β-strand (17), whether the 13C-31P distances remain short is not possible to determine, but can be inferred from the sidechain dynamics of the residues (see below).</p><p>More interestingly, we found that Arg10 and Lys13 sidechains both establish short distances to 31P at low temperature. This is at first puzzling, since it is well documented that CPP analogs where arginine residues were replaced by lysine have much weaker translocation abilities (41, 42). For penetratin, which contains three arginines and four lysines, cellular uptake efficiencies have been compared among the wild-type peptide, the all-arginine analog, and the all-lysine analog. The efficiency was found to be the highest for the all-arginine analog and the lowest for the all-lysine analog (43). Since both arginine and lysine bear a positive charge at neutral pH, the higher activity of arginine-rich peptides has been suggested to be due to more diffuse charge distribution of the guanidinium group, or the ability of guanidinium ions to form multiple hydrogen bonds with oxyanions in a spatially directed manner (44).</p><p>We propose that the similar 13C-31P distances of Arg10 and Lys13 is only true at low temperature at which the REDOR experiments are carried out. Low temperature stabilizes charge-charge interactions, and masks different lipid interactions between Arg and Lys at physiological temperature. Indeed, there is good evidence for a much weaker interaction of the lysine sidechain with phosphates at ambient temperature. First, whereas the Arg10 Nη-Hη order parameters could be measured, attempts to determine the Lys13 Nζ order parameter failed due to unstable 1H-15N cross polarization, which in itself indicates extensive dynamics of the amino group. Second, Arg10 Cδ and Nη have C-H and N-H order parameters of 0.42 and 0.30±0.05 at 303 K, which are relatively large considering that they are three and six bonds away from the backbone Cα. Further, the similar order parameters indicate that the guanidinium moiety as a whole is relatively rigid at physiological temperature, which should facilitate its complexation with the phosphate groups. In comparison, the Lys13 Nζ- Hζ order parameter, although not directly measurable, can be estimated as the product of the Cε SCH of 0.33 with an additional scaling factor of 0.33 due to the three-site jumps of the amino group. Thus, the maximum Lys13 Nζ-Hζ order parameter should be only 0.10, which is much smaller than the Arg10 Nη order parameter of 0.30. With this small order parameter, the Lys13 amino group is unlikely to form any long-lasting hydrogen bonds with lipid phosphates.</p><p>The fact that Arg10 adopts a β-sheet backbone conformation that is independent of the temperature or membrane composition, in contrast to all other residues seen so far in penetratin, further supports the unique interaction of arginine with lipid phosphates. Residues Ile3, Ile5, Gln8, Asn9, and Lys13 all exhibit coil-like chemical shifts at high temperature, which were assigned to a β-turn conformation (17). These data, together with the Arg10 chemical shifts measured here, suggest that the β-turn connect a short stretch of β-strand encompassing Arg10. β-turn residues separated by short β-strands are present in various naturally occurring proteins. For example, elastins have recurring (VPGVV)n sequences where the central PG residues adopt a β-turn conformation whereas the flanking Val residues have the β-strand conformation (45).</p><p>For penetratin, between the β-turn Asn9 and Lys13, there are two arginine residues and one Met (RRM). It is very likely that the conformational propensity of Arg10 is not unique to this residue but is also true for Arg11, because the peptide backbone may be forced into an extended structure in order to allow the lipid headgroups to approach the charged guanidinium moieties to form the guanidinium-phosphate complex. In other words, guanidinium-phosphate interactions may be the cause of the persistent β-sheet conformation of Arg10 at high and low temperatures. As a corollary, the fact that Lys13 adopts the β-turn instead of β-strand conformation at high temperature is yet another piece of evidence that the ammonium group has much weaker interactions with the phosphates in the liquid-crystalline membrane.</p><p>The β-sheet oligomeric structure of penetratin in the gel-phase membrane is energetically favorable. The establishment of intermolecular C=O … H-N hydrogen bonds reduces the free-energy cost of inserting the peptide into the membrane (46). As discussed above, the extended conformation (Figure 9) may facilitate the close approach of lipid phosphate groups with the cationic sidechains, thus allowing both arginine and lysine to interact with the phosphates and establish short 13C-31P distances (Figure 10).</p><p>In conclusion, we have shown that strong guanidinium-phosphate interactions exist in the cell-penetrating peptide penetratin, similar to antimicrobial peptides. Moreover, by considering not only low-temperature 13C-31P distances of Arg10 and Lys13, but also high-temperature order parameters of the two sidechains and the unique high-temperature β-strand conformation of Arg10, we deduce that the arginine sidechain interacts more strongly with lipid phosphates than the lysine sidechain at physiological temperature. Therefore, charge- and hydrogen-bond-stabilized guanidinium-phosphate interaction is not only responsible for membrane translocation of this cationic peptide, but also influences the conformation of the peptide.</p>

PubMed Author Manuscript

DUAL WAVELENGTH PHOTOACTIVATION OF cAMP AND cGMP-DEPENDENT PROTEIN KINASE SIGNALING PATHWAYS

The spatial and temporal organization of biological systems offers a level of complexity that is challenging to probe with conventional reagents. Photoactivatable (caged) compounds represent one strategy by which spatiotemporal organizational complexities can be addressed. However, since the vast majority of caged species are triggered by UV light, it is not feasible to orthogonally control two or more spatiotemporal elements of the phenomenon under investigation. For example, the cGMP- and cAMP-dependent protein kinases are highly homologous enzymes, separated in time and space, which mediate the phosphorylation of both distinct and common protein substrates. However, current technology is unable to discriminate, in a temporally or spatially selective fashion, between these enzymes and/or the pathways they influence. We describe herein the intracellular triggering of a cGMP-mediated pathway with 360 nm light and the corresponding cAMP-mediated pathway with 440 nm light. Dual wavelength photoactivation was assessed in A10 cells by monitoring the phosphorylation of vasodilator-stimulated phosphoprotein (VASP), a known substrate for both the cAMP- and cGMP-dependent protein kinases. Illumination at 440 nm elicits a cAMP-dependent phosphorylation of VASP at Ser157 whereas 360 nm exposure triggers the phosphorylation of both Ser157 and Ser239. This is the first example of wavelength-distinct activation of two separate nodes of a common signaling pathway.

dual_wavelength_photoactivation_of_camp_and_cgmp-dependent_protein_kinase_signaling_pathways

<!>RESULTS AND DISCUSSION<!>Selection of Photosensitive Moieties<!>Caged cGMP-Independent PKG<!>Caged 8-Substituted cAMP<!>PKA- and PKG-Mediated Phosphorylation of VASP<!>Light-Triggered VASP Phosphorylation with Caged Analogs of cAMP and PKG<!>Materials<!>Assessment of PKG and PKA activity<!>Preparation of caged Ile63Ser PKG<!>Photolysis<!>Microscopy<!>Cell culture<!>Immunofluorescence

<p>Biological behavior is inherently dynamic, be it at the cellular, tissue, organ, or organismal level.(1) The response of living organisms and their component parts to environmental stimuli often reflects the spatiotemporal elements associated with that stimulus. For example, metastasis is driven by the directed migration of tumor cells toward spatially focused chemoattractant gradients.(2) However, spatiotemporal control of biological behavior is not limited to just environmental stimuli. As a result of evolution's efficiency, individual proteins, and the biochemical pathways they inhabit, are often used for more than one purpose. The activation of a specific protein can have a multitude of biological consequences yet, within the appropriate spatiotemporal context, elicits only a single response. Cytochrome c, as a key member of the mitochondrial electron transport chain, is essential for life in eukaryotes. However, upon release from the mitochondrion, it serves as a purveyor of death.(3) Only a micron separates life from death, a fact which highlights the control exerted by the spatiotemporal context under which the biological event occurs. Unfortunately, conventional probes of cellular biochemistry, such as inhibitors, activators, or sensors, typically lack the temporal or spatial resolution to adequately address biochemically-driven behaviors, especially those that transpire during a short time frame or occur within a restricted spatial environment.</p><p>A wide variety of photoactivatable (caged) compounds and biomolecules have been described since their first introduction in the late 1970s.(4-9) These species are biologically inactive until exposed to high intensity light, most commonly at the UV/visible boundary (~360 nm).(7-15) This special property allows bioactive species to be loaded into a biological system in an inert form. Subsequent exposure to light, delivered as a high photon flux with a sharp focus, can furnish a degree of spatiotemporal control not feasible through conventional means. However, a number of challenges remain before this technology can be utilized to address some of the pressing issues that lie at the frontiers of biological research. For example, the degree of spatial resolution is constrained by both the diffraction limit of light as well as by the rapid diffusion of caged and photoactivated molecules into and out of the illuminated region of interest. In addition, the investigator is typically forced to take, as an article of faith, that uncaging has been successful. These issues have recently been addressed by attaching caged compounds to localization sequences (spatial control)(16) and by linking the photoactivation process to a fluorescence increase (confirmation of uncaging)(16-18). Perhaps most challenging, however, is the complexity of signaling pathways and the limited impact that a single agent may have in helping to elucidate the dynamic properties associated with these intracellular processes. A potentially powerful advance would be the ability to selectively activate two or more bioreagents, thereby providing separate spatiotemporal control at two or more nodes (e.g. activation and inhibition) of a signaling pathway.</p><p>We describe herein the use of wavelength-distinguishable photolabile moeities to create triggers for two closely related signaling pathways, namely those mediated by the cAMP- and cGMP-dependent protein kinases (PKA and PKG, respectively) (Fig. 1). PKA and PKG possess 70% sequence homology in their catalytic domains(19) and exhibit the same consensus sequence preferences with respect to peptide-based substrates. However, their downstream effects range from overlapping to distinct.(20-22) For example, both enzymes have been implicated in cell motility through their direct action on vasodilator-stimulated phosphoprotein (VASP), a member of the Ena/VASP family of proteins.(23) VASP is regulated by phosphorylation at Ser157, Ser239, and Thr278, which promotes changes in protein-protein interactions of the actin network.(23) The latter interactions, in turn, regulate cell motility.(24) However, the PKA- and PKG-mediated phosphorylation of VASP, particularly in cells, is the subject of considerable controversy.(23, 25-28) Although it has been reported that PKA acts preferentially at Ser157, and PKG at Ser239,(29) the kinetics of phosphorylation appears to be a function of many factors including levels of cyclic nucleotides, the intracellular concentration of PKA and PKG, and cell type.(23, 25-28) A potentially confounding issue is that pathway activation is dependent upon a myriad of inputs and/or is subject to deactivation by endogenous regulatory mechanisms. We've addressed these issues in the context of wavelength-selective triggering of PKA and PKG action on VASP. First, since the activation of these pathways can be initiated by cyclic nucleotides or by enzymes themselves, we've explored both possibilities by generating a light activatable cyclic nucleotide and using it in combination with a light-triggerable protein kinase. Second, we've bypassed the assortment of intracellular regulatory mechanisms by using caged species that, upon photolysis, are constitutively activated and thus resistant to up- or down-regulation by endogenous enzymes or other factors. These strategies furnish the means to trigger, in a wavelength-selective fashion, the cAMP- and cGMP-dependent protein kinase pathways.</p><!><p>There is overwhelming evidence that signaling pathways, and the cellular behavior that they mediate, are profoundly influenced by the spatial and/or temporal context in which they operate. Indeed, cells exhibit a remarkable ability to process two or more spatiotemporal inputs and subsequently render a decision concerning the behavioral course of action.(1) Unfortunately, conventional reagents are limited in their ability to probe pathway performance with a high degree of spatial or temporal resolution, particularly with respect to multiple inputs. One attractive strategy is the use of two or more photosensitive reagents that can be activated with light in a selective fashion. This may be achieved with reagents that are distinguished by pronounced differences in their photolytic quantum yields, in their sensitivity to two-photon (versus conventional) photolysis, or on the basis of their wavelength sensitivity. For any given pair of reagents, a minimum requirement is simply that one of the two reagents is activatable without disturbing the other. Indeed, Ellis-Davies and his colleagues recently employed two-photon illumination to discriminate between a dinitroindolinyl caged glutamate and that of an aminocoumarin-caged γ-aminobutyric acid.(30) Desired receptor activation was achieved via the proper control of laser power, light wavelength, and caged compound concentrations. An alternative strategy, namely the use of photolysis wavelength alone to distinguish between differentially caged species, enjoys the potential advantage of ready application across a broad biological platform such as subcellular, cellular, and multicellular-based settings, including analyses that range from single cells (e.g. microscopy) to large cell populations (e.g. western blot and flow cytometry readouts). In addition, a variety of sources can be used to deliver light of desired wavelengths to the sample, from relatively inexpensive systems (Hg or Xe lamps coupled with appropriate filters) to higher end lasers. The photolabile nitrobenzyl moiety (near UV), in conjunction with far UV(31-33) or visible light-sensitive caging groups(34-38), has paved the way for wavelength-selective photoactivatable pairs, such as protecting groups recently reported for thiols.(38) We employed the latter to create wavelength selective triggers for the highly homologous PKA and PKG pathways.</p><!><p>Dual wavelength photoactivation of multiple biological molecules requires the use of photosensitive groups that can be cleaved at separate wavelengths. For this reason the ortho-nitrobenzyl moiety was employed in combination with an amino-substituted coumarin derivative(36, 37) to cage PKG and 8-Br-cAMP, respectively (Fig. 1). The coumarin absorbs at wavelengths longer than 450 nm, whereas the nitrobenzyl moiety displays no significant absorbance beyond 400 nm (Fig. 2). A "white light" source (Hg arc lamp) was employed, in combination with a 440 ± 10 nM bandpass filter, to selectively photolyze the coumarin-caged cAMP (vide infra). By contrast, a UV bandpass filter (300 nm – 390 nm window that is centered at 360 nm) was used to activate the nitrobenzyl-caged PKG (Fig. 2 and vide infra).</p><!><p>The caged reagents developed in this study are endowed with the following key property: upon photoactivation, the active species is impervious to either up or down regulation by intracellular processes. Specifically, we employed a cAMP derivative that is not prone to inactivation by phosphodiesterase-mediated hydrolysis. Furthermore, we constructed a PKG mutant that is not dependent upon cGMP for activity. Consequently, in both cases, the corresponding caged analogs, upon photolysis, produce species that are instantaneously active and remain so throughout the duration of the experiment.</p><p>Single site-directed mutagenesis was carried out using a baculovirus expression vector to generate a constitutively active, cGMP-independent, Ile63Ser PKG.(39) This mutation was confirmed by sequence analysis. The mutant enzyme was expressed in Sf9 insect cells using a baculoviral expression system as previously described.(40) The PKG mutant was purified on a 8-(2-aminoethyl)aminoadenosine-3′, 5′-cyclic monophosphate (8-AEA-cAMP) agarose resin with typical yields of 2-3 mg per L of harvested Sf9 cells, at greater than 95% purity (Supplemental Fig. 1). Activation of wild-type and Ile63Ser PKG was assessed in the absence and presence of cGMP, thereby confirming that the mutant enzyme is constitutively active (specific activity of 1.44 ± 0.05 μmol/min/mg without and 1.53 ± 0.04 μmol/min/mg with cGMP) but that wild-type PKG activity is dependent on cGMP (0.16 ± 0.02 μmol/min/mg without and 1.00 ± 0.10 μmol/min/mg with cGMP) (Supplemental Fig. 2).</p><p>Preparation of a caged Ile63Ser PKG required modification of one or more residues at or near the active site so that catalytic activity is compromised. Previous studies have revealed that modification of Cys518 inactivates PKG.(41) Therefore, we employed a thiol reactive reagent to generate a caged PKG whose activity can be restored by 360 nm photolysis (Fig. 3a). We do note that PKG contains 11 cysteine residues per monomer, of which eight are accessible without denaturation.(42) Ellman's titration revealed that nitrobenzylbromide modifies 5 - 6 cysteine residues on Ile63SerPKG (Supplemental Fig. 3). Restoration of PKG activity is in proportion to illumination time (Fig. 3a and Supplemental Fig. 4a), thereby confirming that the process is light dependent. Complete activation requires 20 min of photolysis. This relatively long timeframe is a consequence of the experimental setup, namely benchtop photolysis of a macroscopic sample contained within eppendorf tube. By contrast, photolysis under a microscope transpires through a narrow beam that delivers an intense photon flux to a highly focused region (e.g. a single cell), resulting in <sec photolysis times.(14) Caged Ile63Ser PKG displays a specific activity of 0.06 ± 0.05 μmol/min/mg, which is 4.6% of the activity displayed by Ile63Ser PKG. 360 nm photolysis of the caged enzyme generates a 15.7-fold increase of enzymatic activity (0.94 ± 0.04 μmol/min/mg). By contrast, no increase in enzymatic activity was observed upon exposure of the caged enzyme to 440 nm light. The relative activity values (15.7-fold) displayed by pre- and post-photolyzed caged PKG compare favorably to those of unstimulated and cGMP-stimulated wild-type PKG (i.e. 6 - 10-fold).(43) Photoactivation of caged PKG was examined at a variety of wavelengths (Supplemental Fig. 5). As expected, regeneration of PKG activity is most efficient at 360 nm, presumably due to the large overlap with the absorbance spectrum of the nitrobenzyl moiety. Nevertheless, there is significant activation of the caged enzyme at 400 nm, 405 nm and 410 nm. Although photolysis of the nitrobenzyl moiety likely proceeds with a low quantum yield at these wavelengths,(44) the intense H-line emission (405 nm) of the Hg arc lamp light source may be responsible for the photocleavage observed in the 400 - 410 nm range. However, illumination at 420 or 440 nm (Hg G line) fails to activate caged PKG.</p><!><p>8-substituted-cAMP derivatives are hydrolyzed very slowly by phosphodiesterases relative to cAMP itself.(45) Consequently, we utilized a caged derivative of 8-Br-cAMP so that, upon photolysis, a comparatively long-lived active analog of cAMP is unleashed. The 7-[bis(carboxymethyl)amino]coumarin derivative of 8-Br-cAMP was prepared as previously reported.(36, 37) The PKA holoenzyme (0.08 ± 0.01 μmol/min/mg) displays a 2-fold increase (0.16 ± 0.02 μmol/min/mg) in activity upon exposure to 8-Br-cAMP (10 μM). All commercially available sources of PKA holoenzyme tested have a high level of residual activity and all display a similar modest response to cAMP. The coumarin-caged 8-Br-cAMP was unable to stimulate PKA holoenzyme (0.08 ± 0.01 μmol/min/mg) until photolyzed at either 360 nm (0.15 ± 0.01 μmol/min/mg) or at 440 nm (0.16 ± 0.01 μmol/min/mg) (Fig. 3b and Supplemental Fig. 4b).</p><!><p>PKA and PKG regulate VASP activity by phosphorylating residues Ser157 and Ser239, but the pattern and kinetics of phosphorylation by these two enzymes remains controversial.(23, 25-28) We investigated the selectivity of PKA- and PKG-catalyzed phosphorylation at these two sites in A10 cells. Cells were serum starved and then stimulated with either 100 μM 8-Br-cAMP or 8-Br-cGMP. VASP phosphorylation was monitored by western blot (lysates) and by immunofluorescence (fixed cells) using pSer157-selective and pSer239-selective antibodies. VASP is only phosphorylated at Ser157 by cAMP stimulation but both Ser157 and Ser239 are phosphorylated when the cells are stimulated with cGMP (Supplemental Figs. 6a and 7a-c). Lysates from serum-starved cells were also incubated with increasing concentrations of constitutively active PKA or PKG (Supplemental Fig. 6b). The phosphorylation pattern is similar to that observed with the cyclic nucleotide treated cells, namely PKA fails to phosphorylate Ser239 except at very high concentrations (5.3 μM) and PKG phosphorylates both residues even at the lowest concentration tested (26 nM). The distinct VASP phosphorylation patterns induced by these cyclic nucleotide-dependent protein kinases furnished the means to monitor wavelength-selective activation of PKA and PKG.</p><!><p>A10 cells were loaded with the long wavelength (440 nm) caged analog of cell permeable 8-Br-cAMP and the short wavelength (360 nm) caged derivative of PKG, which was microinjected. VASP phosphorylation was monitored in the absence and presence of photolysis (360 nm or 440 nm). VASP phosphorylation at Ser157 and Ser239 was observed in cells containing caged PKG, but only upon illumination at 360 nm (Fig. 4a-c). Microinjection of mouse IgG, caged PKG in the absence of photolysis, or caged PKG exposed to 440 nm, all fail to induce VASP phosphorylation. Illumination of noninjected cells at 360 nm or 440 nm likewise has no effect on phosphorylation status. It should be noted that in all phosphoSer157 VASP immunofluorescence experiments there appears to be low levels of phosphorylation even prior to stimulation. This is a consequence of the antibody used in these experiments, which appears to display some affinity for nonphosphorylated VASP as assessed by western blot analysis (data not shown). In addition, the total fluorescence reading for each experiment varied due to storage of the antibody over time. Consequently, 8-Br-cAMP and 8-Br-cGMP were always run as controls for all experiments. We note as an aside that the released photolyzed caging byproducts (in particular the potentially electrophilic o-nitrosobenzaldehyde) do not appear to have deleterious consequences in terms of VASP phosphorylation. These results are consistent with the general absence of observed harmful consequences of these species on cellular integrity.(46)</p><p>Cell permeable caged 8-Br-cAMP was simply incubated with A10 cells for 30 min at 37 °C prior to photolysis at either 360 nm or 440 nm. Both photolytic conditions induce phosphorylation of the PKA-specific site Ser157 (Fig. 5a-c). This selective phosphorylation is consistent with the results observed with 8-Br-cAMP stimulation. In the absence of photolysis, however, caged 8-Br-cAMP fails to trigger VASP phosphorylation, demonstrating that intracellular PKA was not activated. Finally, dual wavelength photoactivation of PKA and PKG signaling pathways in A10 cells was carried out by loading cells with both the long wavelength sensitive caged 8-Br-cAMP and the UV sensitive caged PKG. Photolysis at 440 nm furnishes selective activation of the PKA pathway, namely phosphorylation of VASP at Ser157, whereas both the PKA and PKG pathways are activated at 360 nm, as demonstrated by the phosphorylation of Ser157 and Ser239 (Fig. 6a-c).</p><p>The caged reagents employed in this study were designed with several features in mind. First, these species, upon photoactivation, are constitutively active and thus impervious to potentially interfering up or down regulation by the endogenous biochemistry of the cell. A previously described cGMP-independent PKG mutant(39) and the phosphodiesterase-resistant 8-Br-cAMP(45) were used as the biochemical triggers to activate the PKG and PKA pathways, respectively. Second, selective photolysis was achieved by employing two photosensitive moieties, only one of which is removed by wavelengths greater than 420 nm. Specifically, the nitrobenzyl-caged PKG is only photolyzed at wavelengths shorter than 410 nm, whereas the coumarin-caged cyclic nucleotide is sensitive to wavelengths up to and including 440 nm. Third, caged enzymes, upon photolysis, enjoy "specificity of action", but lack cell permeability, whereas caged small molecules (e.g. cAMP) can be rendered cell permeable, but often display off target effects. We chose one example from each category to demonstrate the utility of wavelength-selective perturbation of cell signaling. Fourth, photolysis was achieved using a white light source in conjunction with edge or band filters, all of which are inexpensive and commercially available. Finally, we note that both the coumarin and nitrobenzyl moieties absorb and suffer photolysis at 360 nm. Consequently, orthogonal activation of two separate phenomena requires that long wavelength (440 nm) light serve as the initial trigger and short wavelength (360 nm) as the concluding trigger. However, the coumarin and nitrobenzyl moieties can be interchangeably used as caging moieties on bioactive species, thereby providing the investigator with the triggering sequence option of his or her choice.</p><!><p>2-Nitrobenzyl bromide, 8-Br-cAMP, 8-Br-cGMP and PKA holoenzyme were purchased from Sigma. GST tagged PKGIα was purchased from Invitrogen. All primary antibodies and HRP secondary antibodies were from Santa Cruz Technologies. Secondary antibodies used for immunofluorescence were from Invitrogen. Cell culture media and solutions were from Invitrogen. Competent DH5α cells were from Stratagene, Sf9 insect cells were from Invitrogen, and A10 cells were from the tissue culture facility at the University of North Carolina. All other chemicals were from Fisher or Sigma unless otherwise noted.</p><!><p>The in vitro activities of PKG and the PKA holoenzyme were determined using a coupled enzyme assay and either Leu-Arg-Arg-Arg-Arg-Phe-Ser-amide or Leu-Arg-Arg-Ala-Ser-Leu-Gly substrates, both of which were synthesized by standard Fmoc solid phase peptide synthesis.(47) Briefly, phosphorylation of the peptide was coupled to pyruvate kinase and lactate dehydrogenase resulting in the oxidation of NADH. Formation of the latter was monitored at 340 nm. Pyruvate kinase and lactate dehydrogenase were maintained as non-rate limiting enzymes. The activities of PKG and PKA were determined in 50 mM Tris pH 7.5, 100 mM NaCl, 10 mM MgCl2, 10% glycerol, 1 mM peptide, 1 mM ATP, 1mM DTT, 1 mM phosphoenol pyruvate, and 0.2 mM NADH ± 10 μM 8-Br-cAMP or 8-Br-cGMP.</p><!><p>Ile63Ser PKG was extensively dialyzed against 50 mM Tris pH 7.5, 100 mM NaCl, 10 mM MgCl2 and 10% glycerol to remove all trace amounts of reducing reagents prior to modification with nitrobenzyl bromide. The latter was conducted with 6 μM Ile63Ser PKG and 1 mM nitrobenzyl bromide maintained at 4°C for 4 days. The reaction was quenched with 10 mM DTT and then dialyzed against buffer to remove the nitrobenzyl-DTT adduct.</p><!><p>Photolysis was carried out using an Oriel 200 W Hg arc lamp (model 68700) equipped with a beam bending filter. A UV bandpass colored glass filter (Newport, FSQ-UG1), 400 nm ± 10 nM bandpass filter (Newport, 10BPF10-400), 405 nm ± 10 nM bandpass filter (Newport, 10BPF10-405), 410 nm ± 10 nM bandpass filter (Newport, 10BPF10-410), 420 nm ± 10 nM bandpass filter (Newport, 10BPF10-420) and a 440 nm ± 10 nM bandpass filter (Newport, 10BPF10-440) were used for wavelength-selective photolysis. For in vitro assays, caged-PKG or caged-cAMP were photolyzed in an eppendorf tube and irradiated for up to 40 min on ice and then added to the coupled assay. A10 cells were photolyzed using the same filters in MatTek gridded glass bottom dishes for 15 min and then allowed to recover for 1 h at 37 °C with 5% CO2.</p><!><p>All fluorescent microscopy imaging was performed with an inverted Olympus IX81 microscope equipped with a Hamamatsu C8484 camera, 60X oil immersion Plan S-Apo objective and FITC, TxRed and Cy5.5 filter cubes (Semrock). Microinjection was conducted with an Eppendorf FemtoJet and Injectman NI 2 system attached to the microscope set at 100 hPa injection pressure for 0.7 sec. Metamorph or Image J software were employed for imaging analysis and overlays.</p><!><p>A10 rat aortic smooth muscle cells were passaged by treatment with 0.5% trypsin + 0.53 mM EDTA before reaching confluence and maintained in DMEM containing 20% fetal bovine serum (FBS) supplemented with gentamycin and kanamycin at 37 °C in a 5% CO2 incubator. Two days prior to microscopy experiments, cells were plated into either MatTek 6-well glass bottom dishes for cell permeable studies or MatTek 50 mm gridded glass bottom dishes for microinjection studies. The day prior to the microscopy experiments, all cells were serum starved in DMEM containing 2% FBS with gentamycin and kanamycin at 37 °C in a 5% CO2 incubator. On the day of the microscopy experiments, the media was replaced with Leibovitz's L-15 without phenol red supplemented with 2% FBS.</p><!><p>Serum starved A10 cells were loaded with 100 μM coumarin 8-Br-cAMP for 30 min at 37 °C and then washed 3 times with L-15 media or were microinjected with 10 mg/ml purified mouse IgG (sigma) ± 50 μM PKG constructs. Photolysis of cells was carried out as described above and cells were allowed to recover for 1 h at 37 °C with 5% CO2 before fixing. Non-injected control cells were stimulated with 100 μM 8-Br-cAMP and/or 8-Br-cGMP for 1 h prior to fixation. All cells were briefly washed with PBS and fixed with 4% paraformaldehyde (Electron Microscopy Sciences) for 15 min at room temperature. Cells were then washed with PBS, permeablized with 0.5% triton on ice for 20 min, washed with PBS and blocked with 5% donkey serum for 3 h. Fixed cells were incubated with 1:50 dilution of the primary antibodies in donkey serum overnight at 4 °C, washed and incubated with 1:500 secondary antibodies in donkey serum at room temperature for 2 h. After incubation with antibodies, cells were washed with PBS and imaged on the microscope with the appropriate filters. Primary antibodies (1) for controls: 1:125 normal goat IgG + 1:125 normal rabbit IgG as controls, (2) for phospho239 VASP and VASP staining: 1:50 goat anti-phospho239VASP + 1:50 rabbit anti-VASP and (3) for phospho239VASP and phospho157VASP staining: 1:50 goat anti-phospho239VASP + 1:50 rabbit anti-phospho157VASP. Secondary antibodies (1) for studies with cell permeable reagents 1:500 donkey anti-goat Alexa 568 + 1:500 donkey anti-rabbit Alexa 488 and (2) for microinjection studies 1:500 donkey anti-goat Alexa 568 + 1:500 donkey anti-rabbit Alexa 488 + 1:500 donkey anti-mouse Alexa 680.</p>

PubMed Author Manuscript

Local Solvent Acidities in \xce\xb2-Cyclodextrin Complexes with PRODAN Derivatives

The local solvent acidities (SA scale) of six 6-carbonyl-2-aminonaphthalene derivatives as \xce\xb2-cyclodextrin complexes in water are determined through fluorescence quenching. The local polarities (ETN scale) are determined through the shift of the emission center-of-mass. The apparent SA values reflect the solvent structure surrounding the guest\xe2\x80\x99s carbonyl group, whereas the apparent ETN values reveal the net polarity of the entire guest molecule. Comparison of these values affords greater insight into the structures of the host-guest complexes. Derivatives 1 and 5 show unusually large acidities indicative of highly exposed carbonyl groups. The remaining compounds give emission intensities pointing to shielded carbonyl groups. In this study PRODAN and its derivatives are functioning as dual channel sensors of their local environment.

local_solvent_acidities_in_\xce\xb2-cyclodextrin_complexes_with_prodan_derivatives

Introduction<!>Experimental<!>Results and Discussion<!>Conclusions

<p>Small fluorescent molecules possessing charge-transfer emissive states are particularly useful as polarity probes. Among these fluorophores, PRODAN (6-propionyl-2-dimethyaminonaphthalene, 1, Figure 1)1 finds considerable use as a sensor, especially for lipid bilayers.2-6 The Stokes shift of PRODAN is affected by specific H-bonding interactions in addition to the general effects of the solvent's polarity.7-9 In fact, in correlating the solvatochromic shifts, using Catalán's solvent parameters, the coefficient for the polarity term (SPP) is only twice as large as the solvent acidity (SA) term.10, 11 Although the amino group in PRODAN could potentially accept a H-bond, the carbonyl group is thought to be the dominant H-bond acceptor site for PRODAN's excited state.12 The dependence of the emission maximum on both polarity and H-bonding complicates conclusions about the probe's immediate environment.3, 11</p><p>Recently we have shown that emission intensities of 1 - 6 (Figure 1) in a series of hydroxylic solvents correlate strongly with the solvent's H-bond donating abilities.13 Specifically, the logarithm of the degree of quenching in the solvent is linearly related to Catalán's solvent acidity parameter for that solvent. The magnitude of the quenching is particularly pronounced with the carbonyl-twisted derivatives 3, 4 and 6. Because the Stokes shifts are a measure of the local polarity, we suggested that these compounds could act as dual channel chemosensors. In this paper we use PRODAN and the five derivatives as sensors of the local environment of a simple binding site; namely, the interior of a β-cyclodextrin (CD) cavity (Figure 1).14, 15 This seven-unit glucose host molecule is known to form a 1:1 complex with PRODAN in aqueous solution.16-18 Molecular modeling results suggest that PRODAN is oriented with its long axis parallel to the C7 axis of β-CD with the dimethylamino group extending outward from the secondary face. The emission of the PRODAN/β-CD complex is blue-shifted by 20 nm and enhanced several fold in intensity. The solvatochromic shift has been attributed to the lower polarity of the cyclodextrin cavity relative to water, while the intensity increase has been thought to arise from the more rigid environment within the cavity that reduces the pathways for non-radiative relaxation back to the ground state. Such an increased rigidity is essentially an increase in the local viscosity. It is known that a large increase in viscosity can slow solvent relaxation around PRODAN to the point that it will exhibit dual emission from the locally-excited and intramolecular charge-transfer (ICT) states.19 Smaller changes in viscosity have no effect on the solvatochromism.20 Viscosity effects on emission intensity are common with fluorophores equipped with molecular rotors, especially those with twisted ICT states.21 While an emissive TICT state has been proposed for PRODAN, 2, 22, 23 we have shown through the behavior of constrained derivatives that its excited state is planar, and therefore cannot be considered a molecular rotor.24-26 In this paper the emission position and intensity for the βCD-bound PRODAN derivatives are attributed to changes in the H-bonding interactions with the surrounding water molecules.</p><!><p>Compounds 1 - 6 were prepared previously and sublimed under vacuum before use.24, 25, 27 β-Cyclodextrin was a gift from Amaizo and was recrystallized twice from water and dried in vacuo at 100°C before use. Solvents used for photophysical characterization were spectrophotometric grade. Fluorescence emission data were collected using a fiber optic system with a 300W light source monochromated to 365 nm with a bandpass of 5 nm and a high sensitivity Ocean Optics Maya CCD detector. Absorption spectra were obtained from the same fiber optic system with a miniature deuterium/tungsten light source. Binding isotherm data were generated from the emission spectra of a series of aqueous solutions in which the fluorophore concentration was held constant and the β-CD concentration was varied. The following procedure was typical: 5 μL of a stock solution of fluorophore (ca. 5 mg/10 mL MeOH) and a variable volume (0 to 500 μL) of a stock solution of β-CD (ca. 500 mg/ 25 mL H2O) were diluted to 5 mL with H2O and stirred for 4 hours before recording the emission spectrum. Reference solutions of the same concentration of fluorophore in MeOH and EtOH were also prepared. The relative molar absorptivities of the fluorophore in H2O, MeOH and EtOH and of the fluorophore/β-CDcomplex in H2O were determined by the method of standard additions. The electronic noise was subtracted from the raw emission data, and the abscissa scale was converted to wavenumbers before subsequent mathematical treatment. Emission intensity values were determined through numerical integration of the intensity vs. cm−1 data, I(ν~), between 26300 cm−1 (380 nm) and 13300 cm−1 (750 nm). The emission center of mass (ν~CM) was determined from equation 1. (1)ν~CM=∫I(ν~)⋅ν~dν~∫I(ν~)dν~ Plots of the integrated intensity, Iint, vs. [β-CD] were fit to equation 2 using non-linear least squares. Here IL is the limiting intensity of the plot and hence the intensity of the fluorophore/β-CD complex, while fH2o is the emission factor ratio between the fluorophore solvated in water and complexed with β-CD. This factor accounts for differences in the absorption at the excitation wavelength (365 nm) and in the relative quantum yields (equation 3). (2)Iint=IL(fH2O+K[β−CD])1+K[β−CD] (3)fH2O=εH2O⋅ΦH2Oεcomplex⋅Φcomplex The integrated emission intensities of 1 - 6 in MeOH and EtOH give the emission factor ratios fMeoH and fEtoH (equation 4) after correcting for the different indices of refraction (η). (4)fROH=IROHIL⋅ηROH2ηwater2=εROH⋅ΦROHεcomplex⋅ΦcomplexR=Me,Et Rearranging equation 4 gives the ratios of the relative quantum yields of the complex vs. water, methanol and ethanol, respectively (equation 5) (5)ΦcomplexΦROH=fROH−1εROHεcomplexR=H,Me,Et The apparent solvent acidities in the complexes are determined through equation 6. The slope m is determined using the two points on the plot of −log(ΦROH) vs. SA where the intensities of the fluorophore in the reference solvents bracket that of the complex (equation 7) (6)SAapp=SAROH−log(Φcomplex∕ΦROH)m (7)m=log(εROH∕IROHηROH2)−log(εR′OH∕IR′OHηR′OH2)SAROH−SAR′OH</p><!><p>In this investigation PRODAN and derivatives function as sensors of the solvent acidity of hydroxylic media. In particular, the quenching of 1 - 6 is related to the H-bonding ability of the media, and it is expressed by an apparent SA value on Catalán's SA scale.13 The fluorophores sense their local solvent acidity as they are bound to β-CD. Fluorescence titrations of 1 - 6 with β-CD gave hyperbolic binding isotherms characteristic of 1:1 complexes. Figure 2 shows a typical plot with 6 and β-CD together with the intensities of the same concentrations of 6 in ethanol and methanol. The limiting intensity (IL) of the titration corresponds to the emission of the fluorophore in the β-CD complex, and it is determined by fitting the binding isotherm to equation 2. This fitting function has the binding constant (K) and an emission factor ratio ( fH2o, equation 3) as parameters in addition to IL. In reality, only two parameters are needed to fit the binding isotherm. The emission factor ratio could have been expressed in terms of the limiting intensity (fH2o = (Io/IL)*(εcomplex/εwater) because the intensity of the free fluorophore (Io, without added β-CD) and the molar absorptivites of the bound and free fluorophore are determined experimentally. However, this approach would propagate the use of the Io value in the best-fit analysis. Because all of the PRODAN derivatives suffer the greatest quenching in water, the Io value has the largest relative uncertainty. Indeed, for the twisted PRODAN derivatives, 3, 4 and 6, the free fluorophore is quenched by an order of magnitude or more compared to the bound fluorophore. Also, aggregation of the more hydrophobic derivatives, especially 2, 3 and 4, make the measurement of Io problematic.13 Because of the problems with relying only on fH2o for determining the relative quantum yield of the complex, the emission factor ratios for the fluorophore in ethanol and methanol were also determined as references. The fluorescence intensity of the bound fluorophore typically lies between the intensities in these solvents. In all cases the relative quantum yield of the complex was calculated through interpolation using the fROH values that bracket that of the complex.</p><p>The relative quantum yield of the fluorophore in the β-CD complex is translated to an apparent solvent acidity through equation 6. This equation is derived from our previous work that showed an empirical linear relationship between the logarithm of the relative quenching and the solvent acidity parameter of Catalán. The slope of the line segment (m, equation 7) between the bracketing reference solvents is used in interpolating the apparent solvent acidity in the complex. The solvent acidity values for the six complexes are collected in Table 1 together with the respective binding constants. The range of apparent acidity values is large considering the structure similarity of the fluorophores. The lowest value (0.42) is nearly that of ethanol (0.40), and the highest value (0.91) is closer to that of water (1.062) than methanol (0.605). The binding affinities correlate poorly with the apparent solvent acidities.</p><p>The apparent solvent acidities offer added insight into the structure of the host-guest complex. In particular, these values reveal the nature of the H-bonding environment in the immediate proximity of the carbonyl group. A small value indicates a restricted water structure, while a large value indicates a less-perturbed water structure. A restricted water structure implies that the carbonyl group is relatively hidden from the bulk water. Both of the t-butyl derivatives, 3 and 4, and the seven-membered ring derivative, 6, give SAapp values reflecting a relatively less-water-accessible carbonyl group. On the other hand, PRODAN and the six-membered ring derivative, 5, have exposed carbonyl groups according to the SA values. The large difference between the SAapp values for 5 and 6 is surprising since their structures differ only by a methylene group.</p><p>Complexation of the fluorophore with β-CD not only enhances the emission intensity, but it also blue-shifts the emission maximum. Such a change results from both H-bonding interactions and polarity effects with the solvent. Reichardt's ETN solvent polarity parameter reflects both of these terms, and not surprisingly, the emission maximum for PRODAN shows a strong linear correlation with ETN.28-30 Figure 3 and Table 2 show that derivatives 2 - 6 behave similarly. The line segments for 3 and 4 are truncated because the emission data in water are blue-shifted due to aggregation and are not included in the plots. These plots show good linear fits and a narrow range of slope values.</p><p>The fluorescence titrations of 1 - 6 with β-CD gave limiting solvatochromic shifts corresponding to the emission of the β-CD complex. Using the linear fits from Figure 3, these shifts are converted to apparent ETN values, and they are collected in Table 3. The range of the values is rather narrow: less than 0.10. The values indicate polarities that are closer to MeOH (0.762) than to H2O (1.00) or EtOH (0.654). They are consistent with other experimental determinations of the polarity of the β-CD cavity.31 For example, the fluorescence lifetime of the 2-naphthol/ β-CD complex (7.2 ns) lies between the lifetimes for 2-naphthol in methanol (5.9 ns) and ethanol (8.9 ns), but is significantly greater than the lifetime in water (4.8 ns).32</p><p>The comparison between the apparent ETN and SA values sheds light on the nature of the fluorophore/β-CD complex. While the ETN scale and the SA scale are close in magnitude, they do not allow for direct comparison. The latter can be made by converting the apparent values to a percentage of the range between the values for ethanol and water. The results of these conversions are also shown in Table 3 along with the differences between the percentage values. Interpretation of these results requires understanding the factors that determine the apparent ETN and SA values. The ETN values depend in part on polarity effects experienced by the entire molecule, whereas SA values only depend on H-bonding with the carbonyl group. If the rescaled ETN and SA values are close, then the polarity effects on the molecule and the H-bonding effects on the carbonyl are similar. However, Table 3 shows that values are different suggesting that the carbonyl groups do not experience the same net environment as do the molecules. This behavior is consistent with axial inclusion of the naphthalene groups resulting in one end being buried in the less polar cavity while the other end is exposed to bulk water. The high acidities compared to the net polarities for PRODAN and 5 are consistent with inclusion complexes where the carbonyl groups, not the dimethyl amino groups, extend out from the secondary face. Derivatives 2 and 3 containing a piperidine ring give complexes that experience a relatively polar environment overall but show poor H-bonding with the carbonyl, especially with 3. These data suggest that the piperidine ring inhibits penetration into the β-CD cavity and that it is exposed to the solvent while the carbonyl group is hidden despite the fact that the piperidine group is more hydrophobic than the carbonyl groups, especially in the case of 2. The tert-butyl group in 3 and 4 and the seven-membered ring in 6 give rise to strongly and deeply included carbonyl groups as shown by the large binding constants and the relatively low apparent polarities and acidities.</p><!><p>The dual sensor capability of PRODAN and derivatives towards H-bonding interactions and solvent polarity has been demonstrated for guest-host complexes with β-cyclodextrin. Integrated emission intensities report on the H-bonding donating ability of the solvent surrounding the carbonyl group, whereas emission center-of-mass values are indicative of the polarity felt by the entire guest molecule. Together these values offer greater insight into the average structures of the bound complexes.</p>

PubMed Author Manuscript

Biomimetic hydrogen-bonding cascade for chemical activation: telling a nucleophile from a base

Hydrogen bonding-assisted polarization is an effective strategy to promote bond-making and bondbreaking chemical reactions. Taking inspiration from the catalytic triad of serine protease active sites, we have devised a conformationally well-defined benzimidazole platform that can be systematically functionalized to install multiple hydrogen bonding donor (HBD) and acceptor (HBA) pairs in a serial fashion. We found that an increasing number of interdependent and mutually reinforcing HBD-HBA contacts facilitate the bond-forming reaction of a fluorescence-quenching aldehyde group with the cyanide ion, while suppressing the undesired Brønsted acid-base reaction. The most advanced system, evolved through iterative rule-finding studies, reacts rapidly and selectively with CN À to produce a large (>180-fold) enhancement in the fluorescence intensity at l max ¼ 450 nm.

biomimetic_hydrogen-bonding_cascade_for_chemical_activation:_telling_a_nucleophile_from_a_base

Introduction<!>Design principles<!>Synthesis of a minimalist prototype<!>Spectroscopic studies and response to the cyanide ion<!>Structural evolution of the hydrogen bonding network<!>Biomimetic hydrogen bonding network<!>Hydrogen bonding network: effects on the reactivity and solution structure<!>Telling a nucleophile from a base<!>Selectivity toward cyanide and reaction stoichiometry<!>Kinetics studies: rapid detection of cyanide under ambient conditions<!>Conclusions<!>Conflicts of interest

<p>Hydrogen bonding is a versatile functional motif for chemical structure design. An elaborate arrangement of hydrogen bonding donor (HBD) and acceptor (HBA) units has been exploited for molecular recognition, 1-5 signaling, [6][7][8] selfassembly, [9][10][11][12] and chemical activation. [13][14][15] Beyond the paradigm of a simple HBD-HBA pair (Fig. 1a), a serial array of multiple HBD-HBA pairs can also be constructed in a cascade fashion (Fig. 1b). [16][17][18] With an amphoteric proton mediator in the middle, electronic polarization at one end automatically increases the donor/acceptor ability of the other end of the extended network. [19][20][21][22][23] In such a cooperative arrangement of paired HBD-HBA units, the induced polarization at the mediator site makes the hydrogen bonds shorter and stronger, when compared to a single HBD-HBA contact. 19 A prominent example of the cascade hydrogen bonding in action is a class of enzymes having a catalytic triad, such as serine protease, canonical esterase, and lactonase. [23][24][25][26][27] At the active site of these enzymes, the carboxylate group of aspartate, imidazole group of histidine, and hydroxyl group of serine constitute an extended hydrogen bonding array (Fig. 2a).</p><p>In biological systems, the amphoteric imidazole group functions as a mediator of the cascade hydrogen bonding (Fig. 1b). The hydrogen bonding between aspartate and histidine polarizes the amine N-H group to enhance the basicity of the imine nitrogen, which translates to a stronger N/H-O hydrogen bond between histidine and serine. An unusually downeld-shied proton resonance of the imidazole N-H proton at ca. 15 ppm reects tight hydrogen bonding, which polarizes the serine O-H group to enhance its nucleophilicity toward electrophilic substrates. 23 For the enzymes, it would be challenging to control the strength of hydrogen bonds by using limited types of functional groups offered by amino acid residues. As such, the construction of a hydrogen bonding array, in which the whole is more than the sum of its parts, is an effective strategy to make the best out of what is naturally available.</p><!><p>Taking inspiration from the chemical activation strategy of such a catalytic triad, we have designed a benzimidazole-based synthetic surrogate 1 (Fig. 2b). In the case of enzymes, the three-dimensional polypeptide scaffold enables the precise positioning and alignment of HBD-HBA units. Instead of using peptides, we employed the rigid p-conjugated backbone of benzimidazole as a minimalist articial platform to build a biomimetic hydrogen bonding network. Through facile functionalization at the 2-, 4-, and 7-positions of the benzimidazole ring (Fig. 3), various HBD and HBA units can be installed around the imidazole core. Additionally, the rich photophysics of the 2-arylbenzimidazole motif offers excellent opportunities to develop these biomimetic small molecules into uorescent probes. 28,29 In this paper, we report the chemistry of a latent uorophore 1 (Fig. 2 and 3) and its application for the detection of the toxic cyanide ion. From the simplest molecular prototype 2, a systematic increase in the number of HBD-HBA units helps polarize the aldehyde carbonyl group (Fig. 3b). The enhanced electrophilicity of the aldehyde moiety of 1 effectively promotes the selective covalent capture of the cyanide ion (Fig. 3c), and allows rapid uorescence turn-on detection. More importantly, comparative studies on 1 and its simpler "site models" 2-4 established that multiple interconnected HBD-HBA pairs help distinguish a nucleophile from a Brønsted base.</p><!><p>A single HBD-HBA pair was rst installed onto the benzimidazole platform to prepare 2 (Scheme 1). To maximize the effects of chemical transformation on the emissive properties, it would be ideal if the functional group that is activated by hydrogen bonding (Fig. 1c) functions also as a uorescence quencher. For such a purpose, an aldehyde group was employed as an HBA unit. Through fast intersystem crossing, an aldehyde group facilitates the quenching of a p-conjugated uorophore to which it is directly attached. [30][31][32][33] The chemical structure of 2 (Fig. 3b) satises this minimum requirement of this design concept. As outlined in Scheme 1, the synthesis of 2 involved oxidative condensation of diamine 5 and aldehyde, and subsequent oxidation of the primary alcohol group. Except for the reduction of the thiadiazole precursor, all reactions proceeded in moderate to good yields (> 75%). The poor isolation yield of the intermediate 5 might be due to the instability of the electron-rich diamine functionality. The synthetic modularity in the imidazole ring construction aided structural diversication (vide infra).</p><!><p>Compound 2 responds to the cyanide anion by large changes in both UV-vis absorption and uorescence emission spectra. As shown in Fig. 4a, the addition of cyanide to a solution of 2 in DMSO led to a decrease in the absorption at l max ¼ 345 nm with the development of a new band at l max ¼ 390 nm. While 2 is only weakly emissive due to the efficient uorescence quenching by the aldehyde group, the intrinsic blue emission of benzimidazole (l max,em ¼ 465 nm) was restored upon addition of cyanide (Fig. 4b).</p><p>In the control experiments with other chemical species, however, we realized that a large enhancement in the emission intensity also occurs with the addition of a Brønsted base, such as Et 3 N (Fig. 4b). The excitation spectrum of 2, obtained in the presence of Et 3 N, revealed that absorption at l z 400 nm is responsible for the emission at l max,em ¼ 465 nm (Fig. S1a †). This nding implicates that the loss of hydrogen bonding by deprotonation of the acidic benzimidazole proton could also contribute to the uorescence enhancement, presumably by suppressing the intersystem crossing of aldehyde as the main quenching pathway.</p><p>In other words, both addition and deprotonation reactions could take place in the reaction of 2 with cyanide, but the uorescence responses are essentially indistinguishable.</p><p>To delineate the nature of the chemical reaction that is responsible for the uorescence enhancement (Fig. 4b), we carried out detailed solution 1 H NMR spectroscopic studies. At room temperature (T ¼ 298 K), the 1 H NMR spectrum of 2 in DMSO-d 6 (Fig. 5a, top) shows a pair of broadened resonances associated with benzimidazole N-H and aldehyde C-H protons. In stark contrast, the 1 H NMR spectrum of a solution of 2 in CDCl 3 (Fig. S2 †) shows a well-resolved spectral pattern that is consistent with the chemical structure of 2, thus ruling out the involvement of impurities in the complicated 1 H NMR spectrum obtained in DMSO-d 6 .</p><p>We suspect that the intramolecular N-H/O bond of 2 does not provide sufficient thermodynamic bias toward one tautomer over the other in DMSO-d 6 (Fig. 5b and Scheme 2). Indeed, the two distinct resonances of the benzimidazole N-H protons at d ¼ 13.31 and 13.04 ppm converge into a single averaged feature at d ¼ 12.93 ppm as the temperature is increased (Fig. 5b, coalescence at T c ¼ 343 K). From eqn (1), the corresponding rate constant (k c ) can be determined using the Dn value (108 Hz) at the slow exchange limit. 34 By applying the Eyring equation in eqn ( 2), an activation energy of DG ‡ ¼ 16.4 kcal mol À1 was determined, which is typical for the energy barrier for the tautomerization. 35,36 In this process, the resonances of the aldehyde protons at d ¼ 10.81 and 10.34 ppm, each associated with the two tautomers, also converge into a single resonance at d ¼ 10.54 ppm.</p><p>The addition of cyanide to 2 in DMSO-d 6 produced multiple products, presumably from the reactions involving nucleophilic attack as well as simple deprotonation, as deduced from the 1 H NMR spectrum (Fig. 5a, bottom). We tentatively concluded that the weak and solvent-exposed hydrogen bond of 2 is prone to tautomerization, and subjected to multiple reaction pathways with cyanide functioning as either a nucleophilic Lewis base or a simple Brønsted base.</p><!><p>To suppress the undesired acid-base chemistry observed for 2 (Fig. 4 and 5), additional functional groups were introduced to construct tighter hydrogen bonding. We postulated that a stronger bond polarization (Fig. 1) of cascade hydrogen bonding should shi the tautomer equilibrium to make the hydrogen-bonded form dominant even in a polar solvent environment (Scheme 3, bottom), thereby enhancing the electrophilicity of the aldehyde group toward covalent capture of the cyanide anion. As summarized in Scheme 3, probes 3 and 4 were readily prepared from the common diamine intermediate 5 (Scheme 1). For 3, a hydroxyl group is installed as the HBD at the 2-aryl ring extending from the benzimidazole core. Similarly to the catalytic triad (Fig. 2a), the O-H/N hydrogen bonding between the hydroxyl group and the imine nitrogen atom of benzimidazole is expected to push the tautomer equilibrium to reinforce the N-H/O hydrogen bonding between the aldehyde and amine moieties of the benzimidazole core. Furthermore, the acidic phenolic O-H (pK a $10) 37 could also function as a sacri-cial proton donor in the acid-base reaction to keep the N-H group intact. Even when the acid-base reaction takes place, the deprotonation reaction should occur preferentially at the hydroxyl group, so that the benzimidazole uorophore would be less perturbed.</p><p>As a further structural elaboration, probe 4 has an additional hydrogen bonding acceptor, the -OMOM group. Here, the hydrogen bonding between the ether oxygen atom of -OMOM and the imidazole N-H group is anticipated to help planarize the p-system and strengthen the bifurcated hydrogen bonding. It makes intuitive chemical sense that the multiple HBD-HBA pairs within 4 should work cooperatively in the same direction (Fig. 1b) to preferentially stabilize the desired tautomer which benets from stronger hydrogen bonding (Scheme 3, bottom).</p><p>Indeed, single-crystal X-ray crystallography conrmed the presence of an extensive hydrogen bonding network within 4. As shown in Fig. 6, the X-ray structure of 4 revealed three hydrogen bonds (dashed lines) around the imidazole core with short N imidazole /O phenol (2.568(3) A) and N imidazole /O ether (2.634(4) A) distances and essentially a coplanar arrangement of the extended p-system (torsional angle for N1-C1-C9-C10 ¼ 0.65 ; C2-C3-C8-O4 ¼ 1.92 ).</p><p>Cascade hydrogen bonds: effects on the reactivity and solution structure Both 3 and 4 exhibited dramatic uorescence enhancement upon treatment with the cyanide anion (Fig. 7). Unlike 2, however, they remained silent toward Et 3 N, thus differentiating the Brønsted base from the Lewis base. When comparison is made with 2, slightly blue-shied but almost superimposable emission spectra (l max,em ¼ 445 nm) were observed for the cyanide reaction products of 3 and 4.</p><p>We anticipated that multiple HBD-HBA pairs within 3 and 4 would shi the tautomer equilibrium to benet from the additive and reinforcing dipole alignment (Scheme 3, bottom). The results from 1 H NMR spectroscopic studies, however, were rather inconclusive. As shown in Fig. 8, the coexistence of two tautomers was still observed at r.t. for both 3 and 4. While the broadened proton resonances of 3 resemble those of 2 (Fig. 5a, top), the two tautomers of 4 appear as sharp and well-resolved spectral patterns in ca. 1 : 1 ratio (Fig. 8a and b, top).</p><p>Similar to the case of 2, a mixture of products were obtained when cyanide was added to 3 (Fig. 8a, bottom), presumably reecting the tautomer equilibrium. The sharp and wellresolved 1 H NMR spectrum of 4 suggests much slower interconversion between the two tautomers (Scheme 3, bottom), which is consistent with the more extensive hydrogen bonding array in 4 than in its simpler analogues 2 and 3. Nevertheless, the coexistence of the two tautomers for 4 still resulted in the formation of multiple products upon reaction with CN À (Fig. 8b, bottom). Apparently, the presence of three hydrogen bonds (Fig. 6) is still insufficient for a complete conversion of 4 to a single product.</p><!><p>As described above, studies on 2-4 suggested the need for a stronger and more tightly regulated hydrogen bonding network to control both the tautomer equilibrium and the reactivity toward the nucleophile. We thus decided to install an additional HBD-HBA motif onto 4 to prepare 1 (Fig. 3b). Within a six-membered ring setting, the highly polarized N-H bond of an amide moiety is ideally suited to make a good HBD-HBA pair with the imidazole imine-N atom, as predicted by the energyminimized DFT (B3LYP-D3/6-31G(d,p)) model shown in Fig. 2b. In line with the schematic diagram shown in Fig. 1, comparative DFT studies on 1 and the simple HBD-HBA pair 2 predict a large difference in the molecular dipole moment of 6.4413 D (for 1) vs. 0.5614 D (for 2). The molecular electrostatic potential (MEP) maps of 1 and 2 (Fig. S4 †) also show a larger polarization of electron density across cascading dipoles.</p><p>To functionalize the 4-position of the benzimidazole core, the overall synthetic scheme needed to be modied to introduce a bromo substituent at the early stage (Scheme 4). From the bromo-functionalized diamine, sequential oxidative condensation and oxidation reactions afforded 6 in a high yield (72% for two steps). By a cross-coupling reaction onto this bromo position, various functional groups could be installed. An aliphatic amide group was chosen in our molecular design to suppress direct electronic conjugation with the benzimidazole core, thereby minimizing perturbation of the photophysical properties. A palladium-catalyzed Suzuki-Miyaura crosscoupling of 6 with a-(acetylamino)benzylboronic ester furnished the target compound 1. The low isolation yield (ca. 17%) in this nal step is due to the purication procedure involving repetitive recrystallization.</p><p>The single-crystal X-ray structure of 1 shown in Fig. 9 conrmed the presence of multiple hydrogen bonds. The interatomic distances of hydrogen bonds, 2.5431(1)-2.9007(1) Å, are similar to those of 4 (Fig. 6). The essentially co-planar arrangement of the p-conjugated backbone of 1 (torsional angle for N1-C1-C9-C10 ¼ 4.17 ; C2-C3-C8-O4 ¼ 2.03 ) further validates the functional role of the hydrogen bonding network as a conformational lock. Unlike the DFT computational model (Fig. 2b), the amide N-H group of 1 is twisted away from the benzimidazole-N atom. A close inspection of the crystal packing diagram revealed an extensive intermolecular N amide -H/O carbonyl hydrogen bonding network between adjacent molecules, which is reinforced further by p-p stacking and C-H/p contacts (Fig. S5 †). Apparently, such intermolecular interactions in the condensed phase prevail over the inherent propensity of 1 to make the intramolecular N amide -H/N imidazole hydrogen bond as a discrete molecular species (Fig. 2b).</p><!><p>As shown in Fig. 10a, the addition of the cyanide anion to a solution of 1 elicited a rapid and dramatic (> 180-fold) enhancement in the emission intensity at l max,em ¼ 450 nm, whereas no spectral change was observed with the Brønsted base Et 3 N. A large spectral change was also observed in the electronic excitation upon the addition of cyanide (Fig. 10b) with a color change to yellow. In contrast, only a slight increase in the absorption at l z 400 nm region was observed with Et 3 N, with the rest of the spectrum remaining essentially superimposable (Fig. 10b). This visually discernible colorimetric change and uorescence turn-on allow naked-eye detection of the cyanide anion (Fig. 10, inset pictures).</p><p>To investigate the solution structure of 1, 1 H NMR spectroscopic studies were carried out. The 1 H NMR spectrum of 1 (4.0 mM) in DMSO-d 6 measured at T ¼ 293 K indicates the dominance of one prevailing tautomer (> 90%, based on the peak integration values; Fig. 11a, top), which remains essentially invariant even with increasing the temperature up to T ¼ 363 K (Fig. 11b). The 2D ROESY NMR spectrum of 1 (4.0 mM) obtained in DMSO-d 6 at r.t. revealed prominent ROE signals between (i) benzimidazole N-H and aldehyde C-H, and (ii) benzimidazole N-H and methylene protons of the -OMOM group (Fig. S6 †). These ROE correlations provide compelling evidence for the dominant tautomeric form of 1, as predicted by DFT computational studies (Fig. 2b). The gradual up-eld shis of the phenolic O-H (d ¼ 13.46 to 13.30 ppm) and amide N-H protons (d ¼ 8.99 to 8.68 ppm) with increasing temperature (Fig. 11b) also suggest their involvement in intramolecular hydrogen bonding. 38,39 The high conformational stability of 1 was established further by concentration-dependent 1 H NMR studies. Within the concentration range of 1.5-4.0 mM, no noticeable change was observed in the 1 H NMR spectrum of 1 in DMSO-d 6 , implying that 1 remains folded in solution against intermolecular hydrogen bonding (Fig. S7 †). In stark contrast, compound 3 having relatively weak hydrogen bonds undergoes signicant broadening of benzimidazole aromatic proton resonances with increasing sample concentration (Fig. S8 †). Furthermore, the addition of a small amount of H 2 O (2 mL) to a solution of 3 in DMSO-d 6 (2.0 mM, 500 mL) sharpened the resonances of these aromatic protons by rapid proton exchange with the N-H group (Fig. S9 †). No spectral change was observed for 1 under the same conditions (Fig. S10 †).</p><p>With the hydrogen-bonded tautomeric form prevailing for 1 in solution, the addition of the cyanide anion resulted in a clean and complete conversion to the cyanohydrin adduct (Fig. 11a, bottom). To better interpret the 1 H NMR spectrum of the reaction product, we carried out 2D-COSY NMR studies. As shown in</p><!><p>To compare the ability of 1-4 to distinguish a nucleophile (i.e. CN À ) from a Brønsted base (i.e. Et 3 N), each of the probe molecules was treated with either the cyanide anion (25 equiv.) or Et 3 N (25 equiv.). The emission intensity was measured at l em ¼ 450 nm, and the ratio I CN /I Et 3 N was calculated for each molecule. With an increasing number of HBD-HBA units installed around the same uorogenic benzimidazole core (Fig. 3b), a systematic increase in the I CN /I Et N ratio was observed along the series 2 / 3 / 4 / 1 (Fig. 13). Stronger and networked hydrogen bonds seem to promote the bond-forming reaction while effectively suppressing the undesired acid-base reaction. Comparative 1 H NMR studies (Fig. 5, 8, and 11) establish that such hydrogen bonds can also shi the solution equilibrium toward the more reactive tautomer to the cyanide anion.</p><!><p>To test the selectivity of the probe 1, aqueous solution samples of 12 different anions, including CN À , F À , Cl À , Br À , I À , N 3 À , SCN À , OAc À , NO 3 À , ClO 4 À , PF 6 À , and OH À (25 equiv., delivered as sodium salts except for KPF 6 and KOH), were added to 1 (0.100 mM) in a DMSO-H 2 O (99 : 1, v/v) mixed-solvent system, and the emission spectra were recorded under identical conditions. As summarized in Fig. 14a and b, the uorescence turn-on response was observed exclusively for the cyanide anion.</p><p>A large enhancement in the emission intensity was also observed at l em ¼ 450 nm when the cyanide anion was subsequently added to the mixture of 1 and other anions except OH À (Fig. 14c). We suspect that deprotonation by a strong base could disrupt the cascade hydrogen bonding, thus diminishing the response of 1 toward subsequently added CN À . The reaction stoichiometry between 1 and cyanide was determined by Job plot analysis using UV-vis spectroscopy. A sharp maximum at the mole fraction of 0.5 provides compelling evidence for the formation of a 1 : 1 adduct (Fig. 14d), which is also consistent with the results from 1 H NMR (Fig. 11a and 12) and HPLC-MS studies (Fig. S11 †).</p><!><p>The rate constant for the bimolecular reaction between 1 and cyanide was determined by recording time-dependent UV-vis absorption spectra (Fig. S12 †). A large enhancement in the absorption at l ¼ 400 nm brought by the addition of the cyanide anion helped track the progress of the reaction over time (Fig. 10b). The activation of the electrophilic aldehyde group by the hydrogen bonding array led to a rapid chemical transformation. Even at a low temperature (T ¼ 15 C), the reaction of 1 with cyanide in DMSO-MeCN (1 : 1, v/v) was completed within < 2 seconds (Fig. S12a †). Under the pseudo-rst-order kinetic reaction conditions, the kinetic trace was tted to obtain pseudo-rst-order rate constants, k 0 (¼k 2 [CN À ] 0 ); the secondorder rate constant k 2 was estimated from the linear relationship between k 0 and [CN À ] 0 (Fig. S12b †). While the precise determination of rate constants was hampered by the fast reaction rate, the calculated k 2 value (1.3 Â 10 3 M À1 s À1 ) is among that of the fastest-responding cyanide probes that operate by a covalent capture strategy. 32,[43][44][45][46][47][48][49][50] With 20 equiv. of cyanide anion, the reaction half-life t 1/2 is as short as 0.53 s.</p><!><p>As a synthetic mimic of a biological hydrogen bonding triad, a T-shaped p-conjugated platform was structurally elaborated. In our molecular design, the amphoteric benzimidazole core reinforces bond-polarizing HBD-HBA networks to activate an electrophilic aldehyde group for the covalent capture of the toxic CN À anion. We found that a systematic increase in the number of hydrogen bonds allows the molecules to distinguish bond-making nucleophiles from proton-abstracting Brønsted bases. The most advanced molecular probe 1 has four hydrogen bonds around the uorogenic benzimidazole core, and detects the cyanide ion by a rapid and selective turn-on response. Efforts are currently underway in our laboratory to expand the scope of this non-covalent design strategy to other types of chemical transformations of relevance to target-specic signal transduction.</p><!><p>There are no conicts to declare.</p>

Royal Society of Chemistry (RSC)

Role of TAPP1 and TAPP2 adaptor binding to PtdIns(3,4)P2 in regulating insulin sensitivity defined by knock-in analysis

Insulin sensitivity is critically dependent on the activity of PI3K (phosphoinositide 3-kinase) and generation of the PtdIns(3,4,5)P3 second messenger. PtdIns(3,4,5)P3 can be broken down to PtdIns(3,4)P2 through the action of the SHIPs (Src-homology-2-domain-containing inositol phosphatases). As PtdIns(3,4)P2 levels peak after those of PtdIns(3,4,5)P3, it has been proposed that PtdIns(3,4)P2 controls a negative-feedback loop that down-regulates the insulin and PI3K network. Previously, we identified two related adaptor proteins termed TAPP [tandem PH (pleckstrin homology)-domain-containing protein] 1 and TAPP2 that specifically bind to PtdIns(3,4)P2 through their C-terminal PH domain. To determine whether TAPP1 and TAPP2 play a role in regulating insulin sensitivity, we generated knock-in mice that express normal endogenous levels of mutant TAPP1 and TAPP2 that are incapable of binding PtdIns(3,4)P2. These homozygous TAPP1R211L/R211LTAPP2R218L/R218L double knock-in mice are viable and exhibit significantly enhanced activation of Akt, a key downstream mediator of insulin signalling. Consistent with increased PI3K and Akt activity, the double knock-in mice display enhanced whole body insulin sensitivity and disposal of glucose uptake into muscle tissues. We also generated wild-type and double TAPP1R211L/R211LTAPP2R218L/R218L knock-in embryonic fibroblasts and found that insulin triggered enhanced production of PtdIns(3,4,5)P3 and Akt activity in the double knock-in fibroblasts. These observations provide the first genetic evidence to support the notion that binding of TAPP1 and TAPP2 adaptors to PtdIns(3,4)P2 function as negative regulators of the insulin and PI3K signalling pathways.

role_of_tapp1_and_tapp2_adaptor_binding_to_ptdins(3,4)p2_in_regulating_insulin_sensitivity_defined_b

INTRODUCTION<!>Materials and general buffers<!>Generation and genotyping of TAPP1R211L/R211L and TAPP2R218L/R218L knock-in mice<!>Animals<!>Blood glucose and plasma insulin measurement<!>PtdIns(3,4)P2\xe2\x80\x93agarose pull-down<!>Antibodies<!>Preparation of tissue lysates, immunoblotting and Akt kinase assay<!>Generation and stimulation of MEFs (mouse embryonic fibroblasts)<!>Hyperinsulinaemic\xe2\x80\x93euglycaemic clamp studies<!>Measurement of PtdIns(3,4,5)P3 levels<!>Generation and analysis of TAPP1R211L/R211L and TAPP2R218L/R218L knock-in mice<!>Increased activation of Akt in TAPP1R211L/R211LTAPP2R218L/R218L double knock-in mice<!>Enhanced whole body insulin sensitivity in TAPP1R211L/R211LTAPP2R218L/R218L double knock-in mice<!>Increased levels of PtdIns(3,4,5)P3 and Akt activation in TAPP1R211L/R211LTAPP2R218L/R218L knock-in fibroblasts<!>DISCUSSION

<p>The PI3K (phosphoinositide 3-kinase) pathway plays a central role in regulating most cellular responses to insulin [1]. This is emphasized by the findings that marked insulin resistance is invariably induced by inhibiting the PI3K pathway in genetically modified animal models, or by administering inhibitors or by employing high-fat diets that suppress this signalling network [2,3]. There is also ample evidence that insufficient PI3K pathway activation is a hallmark of Type 2 diabetes in humans [2]. We now have a good understanding of how activation of the PI3K pathway controls insulin signalling responses. Once activated by insulin, PI3K phosphorylates the D3 hydroxy group of PtdIns(4,5)P2 to generate PtdIns(3,4,5)P3, the key second messenger of the insulin signalling pathway. PtdIns(3,4,5)P3 triggers cellular responses to insulin by recruiting the Akt protein kinases that possess a PtdIns(3,4,5)P3-binding PH (pleckstrin homology) domain to the plasma membrane. This induces a conformational change in Akt, permitting its activation by the upstream PDK1 (phosphoinositide-dependent kinase 1) and mTORC (mammalian target of rapamycin complex) 2 protein kinases [4]. Once activated, Akt phosphorylates numerous substrates to regulate responses to insulin [5], including phosphorylating AS160 (Akt substrate of 160 kDa) to stimulate glucose uptake [6] and GSK3 (glycogen synthase kinase 3) to promote glycogen synthesis [7].</p><p>The mechanisms that regulate the inactivation of the insulin signalling pathway also play a vital role in regulating overall insulin actions. For example, overactivation of the mTORC1 pathway as a result of obesity induces insulin resistance by stimulating phosphorylation and degradation of IRS (insulin receptor substrate) adaptor proteins required for activation of PI3K by insulin [8]. Furthermore, knock out of PTP (protein tyrosine phosphatase) 1B, which inactivates the insulin receptor, markedly sensitizes mice to insulin and protects animals to developing insulin resistance when fed on a high-fat diet [9]. The lipid phosphatases that break down PtdIns(3,4,5)P3 also play vital roles in controlling sensitivity to insulin. For example, a 50% reduction in the expression of the PTEN (phosphatase and tensin homologue deleted on chromosome 10), which converts PtdIns(3,4,5)P3 into PtdIns(4,5)P2, induces insulin sensitization [10] and is sufficient to reverse marked insulin resistance in a mouse knock-in model in which the ability of PDK1 to interact with PtdIns(3,4,5)P3 has been ablated [11]. PtdIns(3,4,5)P3 can also be converted into PtdIns(3,4)P2 by SHIP (Src-homology-2-domain-containing inositol phosphatase) 1 and SHIP2 [12,13]. Mice lacking SHIP2 display enhanced activation of Akt following insulin administration and are highly resistant to weight gain when placed on a high-fat diet [14]. There has been much discussion of whether PtdIns(3,4)P2 functions as a signalling molecule in its own right, since insulin and other agonists that activate the PI3K pathway significantly increase its levels [15]. Detailed analysis confirms that the majority of PtdIns(3,4)P2 originates from PtdIns(3,4,5)P3 and, consistent with this, levels of PtdIns(3,4)P2 peak after those of PtdIns(3,4,5)P3 [16].</p><p>To date, numerous PH-domain-containing proteins have been identified that interact with PtdIns(3,4,5)P3, specifically GRP1 (guanine-nucleotide-releasing protein 1), Btk and ARNO [ARF (ADP-ribosylation factor) nucleotide-binding-site opener], or bind both PtdIns(3,4,5)P3 and PtdIns(3,4)P2 with similar affinity, such as Akt, PDK1 and DAPP1 (dual adaptor for phosphotyrosine and 3-phosphoinositides 1) [15,17]. However, to our knowledge, the only proteins to have been identified that specifically interact with PtdIns(3,4)P2 with high affinity are TAPP (tandem PH-domain-containing protein) 1 and TAPP2, related adaptor proteins consisting of two sequential PH domains in which the C-terminal PH domain binds PtdIns(3,4)P2. The N-terminal PH domain does not interact with any lipid tested [18]. The structure of the C-terminal PH domain of TAPP1 suggests that several conserved basic residues interacted with the 3- and the 4-phosphate groups of PtdIns(3,4)P2. Mutation of one of these residues, Arg211 on TAPP1 or the equivalent Arg218 residue in TAPP2, completely prevented interaction of these proteins with PtdIns(3,4)P2 [18]. Binding of TAPP1 to PtdIns(3,4,5)P3 is inhibited by steric hindrance of an alanine residue located close to the position in which the 5′-phosphate would be expected to reside [19]. Interestingly, the equivalent residue in PtdIns(3,4,5)P3-binding PH domains is frequently glycine. Mutation of the alanine residue to glycine in TAPP1 resulted in it being capable of interacting with PtdIns(3,4,5)P3 and PtdIns(3,4)P2 with similar affinity [19]. Evidence also suggests that TAPP1 binds PtdIns(3,4)P2 selectively in vivo, as TAPP1 relocated from the cytosol to the plasma membrane of cells following stimulation with agonists that induced PtdIns(3,4)P2, but not with those that induced mainly PtdIns(3,4,5)P3 [20]. Similarly, in B-cells, both TAPP1 and TAPP2 translocated to the plasma membrane in response to antigen stimulation and this correlated with the formation of PtdIns(3,4)P2 rather than production of PtdIns(3,4,5)P3 [21,22].</p><p>The biological functions of TAPP1 and TAPP2 are not well characterized. Apart from the PH domains, the only other known functional region is a C-terminal PDZ-binding motif that interacts with several PDZ-binding proteins, including PTPN13 (known previously as PTPL1 or FAP-1) [23] as well as the scaffolding proteins MUPP1 (multiple PDZ-domain-containing protein 1) [20], syntrophin [24] and utrophin [25]. One hypothesis was that TAPP1 and TAPP2, by specifically recognizing PtdIns(3,4)P2, could function to recruit signalling molecules or complexes to the plasma membrane that down-regulate the PI3K signalling pathways [23]. As a first step to exploring this idea, we generated knock-in mice expressing point mutants of TAPP1 and TAPP2 unable to interact with PtdIns(3,4)P2. The resulting TAPP1R211L/R211LTAPP2R218L/R218L double knock-in mice are viable and displayed significantly enhanced whole body insulin sensitivity in a hyperinsulinaemic–euglycaemic clamp study. We have also demonstrated that TAPP1R211L/R211L TAPP2R218L/R218L double knock-in fibroblast cells display enhanced PtdIns(3,4,5)P3 levels and Akt activation in response to insulin. Our results suggest that enhanced insulin sensitivity is mediated by increased Akt activation, thereby stimulating glucose uptake in heart and skeletal muscle. TAPP1R211L/R211LTAPP2R218L/R218L knock-in mice represent the first mouse model for these proteins and support the notion that TAPP1 and TAPP2 operate as negative regulators of the PI3K signalling pathway.</p><!><p>Protein G–Sepharose and [γ-32P]ATP were purchased from GE Healthcare. Human insulin (Actrapid) from Novo-Nordisk was obtained from Ninewells Pharmacy, Dundee. PI-103 PI3K inhibitor [26] was synthesized by Natalia Shpiro (MRC Protein Phosphorylation Unit, Dundee). Lysis buffer consisted of 50 mM Tris/HCl (pH 7.5), 1 mM EDTA, 1 mM EGTA, 0.3% CHAPS, 1 mM sodium orthovanadate, 50 mM sodium fluoride, 5 mM sodium pyrophosphate, 0.27 Msucrose, 0.1%2-mercaptoethanol and Complete™ protease inhibitor cocktail (Roche).</p><!><p>TaconicArtemis generated the TAPP1R211L/R211L and TAPP2R218L/R218L knock-in mice described in Figure 1. The knock-in mice were generated and maintained on an inbred C57BL/6J background. Genotyping was performed by PCR using genomic DNA isolated from ear biopsies. For TAPP1 knock-in mice, Primer 1 (5′-CCTCATTCAGAGTATGCAGC-3′) and Primer 2 (5′-CTACCTTCAAGGAAGTGTTCC-3′), and for TAPP2 knock-in mice, Primer 1 (5′-GGATGTTCTCAAGACTCACG-3′) and Primer 2 (5′-CATCATGGGAGTAAGAGTAGG-3′) were used to detect the wild-type and knock-in alleles. The PCR programme consisted of 5 min at 95°C, then 35 cycles of 30 s at 95°C, 30 s at 60°C and 30 s at 72°C, and 5min at 72°C. DNA sequencing was performed by DNA Sequencing & Services (MRC Protein Phosphorylation Unit, College of Life Sciences, University of Dundee; http://www.dnaseq.co.uk) using Applied Biosystems Big-Dye version 3.1 chemistry on an Applied Biosystems model 3730 automated capillary DNA sequencer.</p><!><p>Mice were maintained under specific pathogen-free conditions and all procedures were carried out in accordance with the regulations set by the University of Dundee and the U.K. Home Office.</p><!><p>Blood glucose levels were determined using the Ascensia Breeze 2 blood glucose monitoring system (Bayer) following tail incision. For plasma insulin measurement, blood was collected from mice following tail incision and using sodium-heparinized capillary tubes (Hawksley). The blood was centrifuged at 3000 g for 15 min, and the supernatant was collected. Plasma insulin levels were determined using a rat/mouse insulin ELISA kit from Millipore (No EZRMI-13K) according to the instructions of the manufacturer. Rat insulin ranging from 0.2 to 10 ng/ml was used as a standard.</p><!><p>Mouse tissues were homogenized in 10 mM Hepes (pH 7.4), 150 mM NaCl, 0.25% Nonidet P40, 1 mM benzamidine, 0.1 mM PMSF and 2 mM sodium orthovanadate. Tissue lysates were centrifuged at 18000 g for 15 min at 4°C, and the supernatant was snap-frozen and stored at −80°C. Then, 1 mg of tissue lysate was incubated with 25 μl of PtdIns(3,4)P2 PIP beads (Echelon) for 1.5 h at 4°C. Beads were washed three times with 10 mM Hepes (pH 7.4), 150 mM NaCl and 0.25% Nonidet P40 and proteins were eluted with SDS/PAGE sample buffer (250 mM Tris/HCl, pH 6.8, 5% SDS, 5% 2-mercaptoethanol and 32.5% glycerol) at 95°C for 5 min. Eluted proteins were analysed by immunoblotting.</p><!><p>The following antibodies were raised in sheep and affinity-purified on the appropriate antigen. The total antibody used for immunoprecipitation and immunoblotting of Akt1 (S742B, third bleed) was raised against full-length His–Akt1. The anti-TAPP1 antibody (S022C, second bleed) was raised against residues 252–356 of mouse GST (glutathione transferase)–TAPP1. The anti-TAPP2 antibody (S392A, first bleed) was raised against residues 396–415 of mouse GST–TAPP2. The anti-PRAS40 (proline-rich Akt substrate of 40 kDa) antibody (S115B, first bleed) was raised against the sequence DLPRPRLNTSDFQKLKRKY corresponding to residues 238–256 of human PRAS40. An antibody that recognizes PRAS40 phosphorylated at Thr246 (S114B, second bleed) was raised against the peptide CRPRLNpTSDFQK. The anti-FOXO1 (forkhead box O1) antibody (S457, third bleed) was raised against GST–FOXO1 comprising residues 2–655 of human FOXO1. The antibodies against Akt pThr308 (#9275), Akt pSer473 (#9271), FOXO1 pThr24 (#9464) and GSK3α/β pSer21/pSer9 (#9331) were purchased from Cell Signaling Technology. The anti-GSK3α/β antibody (#44–610) was purchased from Biosource. The anti-IRS1 antibody (06–248) was from Millipore and the antibody against IRS1 pTyr612 (44–816G) was purchased from Invitrogen. The anti-GAPDH (glyceraldehyde-3-phosphate dehydrogenase) antibody (ab8245) was purchased from Abcam. Detection of immune complexes was performed using either fluorophore-conjugated secondary antibodies (Molecular Probes) followed by visualization using an Odyssey® LI-COR imaging system or by HRP (horseradish peroxidase)-conjugated secondary antibodies (Pierce) and an enhanced chemiluminescence reagent.</p><!><p>Following a 5 h fast, a bolus of insulin (1 m-unit/g of body weight) was intravenously injected through the inferior vena cava to mice that had been anaesthetized by pentobarbital (86 μg/g of body weight, intraperitoneally injected). After 20 min, tissues (heart, liver, gastrocnemius muscle) were extracted, frozen in liquid nitrogen and stored at −80°C. Tissues were homogenized on ice in a 10-fold mass excess of ice-cold lysis buffer using a Kinematica Polytron. Tissue lysates were centrifuged at 18 000 g for 15 min at 4°C, and the supernatant was snap-frozen and stored at −80°C. Lysates (20 μg) were analysed by immunoblotting using the antibodies, as indicated in the Figures. The activity of Akt1 was assessed either by immunoblotting of tissue lysate (20 μg) using phosphospecific antibodies or by kinase activity assays. Briefly, Akt1 was immunoprecipitated from 1 mg of tissue lysate, and kinase activity was measured using the Crosstide peptide (GRPRTSSFAEG) as described previously [27]. Values of Akt activity are given as means±S.E.M. for the number of mice indicated in the Figure legends, and significance was determined by unpaired two-tailed Student's t tests.</p><!><p>MEFs isolated from mouse embryos at E13.5 (embryonic day 13.5) were generated as described previously [28] and immortalized by continuous passaging. Cells were cultured in DMEM (Dulbecco's modified Eagle's medium) containing 10% serum (Sigma), 2 mM L-glutamine, 50 units/ml penicillin G and 50 μg/ml streptomycin (Life Technologies). Cells were serum-starved in DMEM with L-glutamine, penicillin and streptomycin for 16 h before stimulation. Cells were stimulated with insulin (10 nM) for 15 or 30 min. Cells were subsequently lysed in lysis buffer and the lysates were centrifuged at 18 000 g for 15 min at 4°C. The supernatant was snap-frozen and stored at −80°C. Lysates (10 μg) were analysed by immunoblotting using the antibodies, as indicated in the Figures.</p><!><p>Male mice at 6 months of age were used for a hyperinsulinaemic–euglycaemic clamp studies which were performed at Vanderbilt-NIH Mouse Metabolic Phenotyping Center, Nashville, TN, U.S.A. (http://www.mc.vanderbilt.edu/MMPC/), as described previously in detail [29]. Catheters were implanted in a carotid artery and a jugular vein of mice for sampling and infusions respectively 5 days before study [30]. Insulin clamps were performed on 5 h fasted mice [29]. [3–3H]glucose (2.4 μCi) was primed and continuously infused for a 90-min equilibration period (0.04 μCi/min) and a 2 h clamp period (0.12 μCi/min). Baseline blood or plasma parameters were determined as the mean of values obtained in blood samples collected at −15 and −5 min. At zero time, insulin infusion (4 m-units·kg−1·min−1) was started and continued for 165 min. Blood glucose was clamped at 150–160 mg/dl using a variable GIR (glucose infusion rate). Mice received heparinized saline-washed erythrocytes from donors at 5 μl/min to prevent a fall of haematocrit. Insulin clamps were validated by assessment of blood glucose over time. Blood glucose was monitored every 10 min, and the GIR was adjusted as needed. Blood was taken at 80–120 min for the determination of [3–3H]glucose. Clamp insulin was determined at 100 and 120 min. At 120 min, 13 μCi of [14C]2DG (2-[14C]deoxyglucose) was administered as an intravenous bolus. Blood was taken at 122, 125, 130 and 135 min for the determination of [14C]2DG. After the last sample, mice were anaesthetized and tissues were collected.</p><p>Plasma insulin was determined by ELISA (Millipore). Non-esterified fatty acids were assayed enzymatically using the Wako Diagnostics. Radioactivity of [3–3H]glucose, [14C]2DG, and [14C]2DG 6-phosphate in plasma and tissue samples were determined by liquid-scintillation counting [30]. Whole body glucose appearance (Ra) and disappearance (Rd) rates were determined using non-steady-state equations [31]; EndoRa was determined by subtracting the GIR from total Ra. Glucose uptake was calculated as described previously [32].</p><p>The area under the curve was calculated using GraphPad Prism software. Values are given as means±S.E.M. for the number of mice indicated in the Figure legends. Significance was determined by unpaired two-tailed Student's t tests.</p><!><p>PtdIns(3,4,5)P3 levels were measured employing a TR-FRET (time-resolved fluorescence resonance energy transfer) displacement assay that was described previously [33]. Briefly, wild-type or knock-in MEFs cultured on 10-cm-diameter dishes were deprived of serum overnight and then left untreated in the absence or presence of 1 μM PI-103 or stimulated with insulin (10 nM) for the indicated times. Cells were lysed by incubation with 1.5 ml of ice-cold 0.5 M trichloroacetic acid. Cells were then pelleted by centrifugation at 13 000 g for 1 min, and non-charged lipids were extracted by washing cell pellets with 1 ml of a neutral solvent (methanol/chloroform, 2:1 v/v) for 20 min. PtdIns(3,4,5)P3 was extracted by washing with 0.5 ml of an acidic solvent (methanol/chloroform/12 M HCl, 80:40:1, by vol.) for 20 min. The acidic extract containing PtdIns(3,4,5)P3 was phase-split by the addition of 0.18 ml of chloroform and 0.3 ml of 0.1M HCl. After centrifugation at 12 000 g for 1 min, the lower organic phase containing PtdIns(3,4,5)P3, was collected and dried under vacuum. The lipid pellet was resuspended by sonication (15 s at 40 W in a cup sonicator bath) in 0.06 ml of TR-FRET assay buffer consisting of 50 mM Tris/HCl (pH 7.4), 0.15 M NaCl, 0.1 mM EDTA, 0.1 mM EGTA, 2mM dithiothreitol and 1.2% (w/v) sodium cholate. Duplicate 25 μl samples were assayed for PtdIns(3,4,5)P3 content using the TR-FRET displacement assay described previously [33]. The estimated PtdIns(3,4,5)P3 levels were normalized to cell density by estimating protein levels in the extracted cell pellets. This was achieved by recovering cells from the upper phase of the phase-split material by the addition of 1 ml of acetone (to remove residual chloroform) and centrifugation. The air-dried pellets were then dissolved overnight by gentle shaking in 0.2 M NaOH and 1% (w/v) SDS at 50°C. Protein in the dissolved pellets was estimated by Micro BCA (bicinchoninic acid) protein assay kit (Thermo Scientific) according to the manufacturer's instructions.</p><!><p>To study the physiological role of TAPP1 and TAPP2, we generated knock-in mice in which the critical arginine residues required for PtdIns(3,4)P2 binding (Arg211 of TAPP1 and Arg218 of TAPP2 [18,19]) were mutated to leucine in order to abolish the ability of these adaptors to interact with phosphoinositides. The strategy to generate and genotype the TAPP1R211L/R211L and TAPP2R218L/R218L knock-in mice is summarized in Figure 1. The knock-in mice were generated and maintained on an inbred C57BL/6J background.</p><p>Single TAPP1R211L/R211L or TAPP2R218/R218L or double TAPP1R211L/R211LTAPP2R218L/R218L were viable and born at the expected Mendelian frequency (Table 1). The strategy used to breed wild-type TAPP1+/+TAPP2+/+ control and experimental TAPP1R211L/R211LTAPP2R218L/R218L double knock-in mice is outlined in Table 1. The TAPP1 and TAPP2 proteins were expressed at the same levels in all tissues examined of wild-type TAPP1+/+TAPP2+/+ and homozygous TAPP1R211L/R211LTAPP2R218L/R218L double knock-in mice, establishing that mutation does not affect protein stability (Figure 2A). TAPP1 (Figure 2B) and TAPP2 (Figure 2C) proteins derived from wild-type TAPP1+/+TAPP2+/+ tissues, but not those from TAPP1R211L/R211LTAPP2R218L/R218L double knock-in tissues, interacted with agarose resin conjugated to PtdIns(3,4)P2, confirming that the knock-in mutation ablated the ability of TAPP1 and TAPP2 to interact with PtdIns(3,4)P2. The TAPP1R211L/R211LTAPP2R218L/R218L double knock-in mice compared with wild-type TAPP1+/+TAPP2+/+ mice displayed normal body weight (Figure 2D) as well as fed and fasted levels of blood glucose (Figure 2E) and plasma insulin (Figure 2F). We have maintained the double knock-in mice over a 14-month period and no discernable overt phenotype was noted.</p><!><p>To investigate how inhibiting the ability of TAPP1 and TAPP2 to interact with PtdIns(3,4)P2 affects the insulin signalling pathway, we injected mice deprived of food for 5 h with insulin (1 m-unit/g of body weight) and analysed phosphorylation and activation of Akt in various insulin-responsive tissues, including liver (Figure 3A), heart (Figure 3B) and gastrocnemius skeletal muscle (Figure 3C). We observed that there was no difference in the basal Akt activity between wild-type TAPP1+/+TAPP2+/+ and TAPP1R211L/R211LTAPP2R218L/R218L double knock-in mice, whereas insulin-stimulated Akt activation was significantly enhanced in tissues derived from insulin-injected TAPP1R211L/R211LTAPP2R218L/R218L double knock-in compared with wild-type TAPP1+/+TAPP2+/+ animals (Figure 3). Increased activation of Akt in TAPP1R211L/R211LTAPP2R218L/R218L double knock-in mice was accompanied by enhanced phosphorylation of Akt at Thr308 (PDK1 site) and Ser473 (mTORC2 site). Quantitative immunoblotting analysis confirmed that insulin stimulation of double knock-in mice induced a statistically significant increase in Thr308 and Ser473 phosphorylation in heart and gastrocnemius tissues (see Supplementary Figure S1 at http://www.BiochemJ.org/bj/434/bj4340265add.htm). Despite enhancement of Akt activity, this was not translated into a significant increase in the phosphorylation of Akt substrates examined (PRAS40, GSK3α/β and FOXO1). Moreover, no marked changes were observed in IRS1 phosphorylation at Tyr612 (Figures 4A–4C), a key-binding site for PI3K, which is phosphorylated by the insulin receptor [34]. This is considered further in the Discussion.</p><!><p>To investigate whether increased insulin-induced activation of Akt in the TAPP1R211L/R211LTAPP2R218L/R218L double knock-in mice enhanced insulin sensitivity in the most physiological context, we performed a hyperinsulinaemic–euglycaemic clamp, a method that is considered the gold standard for assessing insulin actions on whole body glucose homeostasis in vivo [35], on TAPP1R211L/R211LTAPP2R218L/R218L double knock-in mice and wild-type TAPP1+/+TAPP2+/+ control mice. These studies were undertaken at the Vanderbilt NIH Mouse Metabolic Phenotyping Center and we employed the clamp study in conscious and unrestrained animals [29]. Blood glucose levels were maintained by a variable glucose infusion with a constant insulin infusion rate of 4m-units·kg−1·min−1 (see Supplementary Figure S2A at http://www.BiochemJ.org/bj/434/bj4340265add.htm). Interestingly, the key index of whole body insulin sensitivity, the GIR required to maintain euglycaemia, was increased at all time points in TAPP1R211L/R211LTAPP2R218L/R218L double knock-in mice compared with wild-type TAPP1+/+TAPP2+/+ control mice (Figure 4A). Calculation of the area under the curve during establishment of the clamp between 80 and 120 min revealed significantly higher (25%) GIR in TAPP1R211L/R211LTAPP2R218L/R218L double knock-in mice compared with wild-type TAPP1+/+TAPP2+/+ control mice. The steady-state whole body glucose disappearance rate (Rd) during the clamp was also significantly higher (23%) in TAPP1R211L/R211LTAPP2R218L/R218L double knock-in mice in comparison with wild-type TAPP1+/+TAPP2+/+ mice (Figure 4B).</p><p>These results demonstrate that TAPP1R211L/R211LTAPP2R218L/R218L double knock-in mice display significantly enhanced whole body insulin sensitivity measured in conscious unrestrained animals in vivo. Endogenous glucose appearance (EndoRa) is the sum of hepatic glycogenolysis and gluconeogenesis in the postabsorptive state. In the basal state, EndoRa was comparable in TAPP1R211L/R211LTAPP2R218L/R218L double knock-in mice and wild-type TAPP1+/+TAPP2+/+ mice. EndoRa was suppressed equally in both genotypes during insulin clamps (Figure 4C). EndoRa therefore did not account for the enhanced insulin sensitivity observed in TAPP1R211L/R211LTAPP2R218L/R218L double knock-in mice. The optimal insulin dose used to examine liver is much lower than the optimal insulin dose used to examine muscle. The insulin dose used was chosen specifically to amplify an insulin effect on glucose flux in cardiac and skeletal muscle. EndoRa was maximally suppressed and therefore an enhanced effect at the liver could not be elucidated. Considering the elevated Akt activation in TAPP1R211L/R211LTAPP2R218L/R218L double knock-in mice, it is very possible that an enhanced insulin suppression of EndoRa would have been observed at a lower insulin dose (<2.0 m-units·kg−1·min−1). In addition, non-esterified fatty acid concentrations were comparable in wild-type TAPP1+/+TAPP2+/+ and TAPP1R211L/R211LTAPP2R218L/R218L double knock-in mice, indicating that insulin's effect on lipolysis was similar between the two genotypes (see Supplementary Figure S2B). Glucose uptake measured by infusion of [14C]2DG during the steady-state phase of the clamp revealed comparable glucose uptake into adipose and diaphragm of wild-type TAPP1+/+TAPP2+/+ and TAPP1R211L/R211LTAPP2R218L/R218L double knock-in mice (Figure 4D). Interestingly, significant increases in rates of glucose uptake in gastrocnemius muscle (77% increase) and heart (75% increase) were detected in TAPP1R211L/R211LTAPP2R218L/R218L double knock-in mice in comparison with wild-type TAPP1+/+TAPP2+/+ mice in which we also found increased activation of Akt (Figure 3). As is often the case in insulin-sensitive mice [30], the basal and clamp insulin levels were reduced. This is important to consider, since the increased insulin action was observed in the TAPP1R211L/R211LTAPP2R218L/R218L double knock-in mice despite reduced insulin concentrations (see Supplementary Figure S2C). These results indicate that the enhanced whole body insulin sensitivity observed in TAPP1R211L/R211LTAPP2R218L/R218L double knock-in mice is likely to be mediated through increased glucose uptake into skeletal muscle and heart tissues.</p><!><p>We generated immortalized MEFs derived from wild-type TAPP1+/+TAPP2+/+ and TAPP1R211L/R211LTAPP2R218L/R218L double knock-in E13.5 stage embryos. These cells were stimulated with 10 nM insulin for various time points and PtdIns(3,4,5)P3 levels were measured employing a TR-FRET assay described previously [33]. This revealed similar levels of PtdIns(3,4,5)P3 in non-stimulated wild-type and or double knock-in cells (Figure 5A). There was also no marked difference in PtdIns(3,4,5)P3 levels between cells treated with the PI3K inhibitor PI-103. However, following insulin stimulation, at all time points examined, we observed a 1.5–2-fold increase in PtdIns(3,4,5)P3 levels in the TAPP1R211L/R211LTAPP2R218L/R218L cells compared with TAPP1+/+TAPP2+/+ fibroblasts (Figure 5A). We also observed that Akt was phosphorylated and activated to a significantly greater extent in TAPP1R211L/R211LTAPP2R218L/R218L double knock-in fibroblasts compared with wild-type TAPP1+/+TAPP2+/+ cells (Figure 5B). Moreover, unlike in mouse tissues, we observed that the TAPP1R211L/R211LTAPP2R218L/R218L double knock-in fibroblasts stimulated with insulin displayed a modest increase in the phosphorylation of PRAS40 and GSK3α/β compared with wild-type TAPP1+/+TAPP2+/+ cells (Figure 5B).</p><!><p>The key finding of the present study is that the TAPP1R211L/R211LTAPP2R218L/R218L double knock-in mice displayed significantly enhanced whole body insulin sensitivity. This provides the first genetic evidence establishing that the TAPP1 and TAPP2 adaptor proteins regulate insulin sensitivity by binding to PtdIns(3,4)P2. Our data from in vivo clamp studies suggest that insulin sensitivity results from increased glucose disposal into heart and skeletal muscle. This is supported by the 1.5–2- fold increase in activity of Akt which is well known to control glucose transport in response to insulin [36]. The increased activation of Akt observed in TAPP1R211L/R211LTAPP2R218L/R218L knock-in mice is likely to result from enhanced activation of PI3K that would account for the increased phosphorylation of Akt at Thr308 (PDK1 T-loop site) and Ser473 (mTORC2 site, hydrophobic motif) (Figure 3). This increased Akt phosphorylation and activation is also observed in fibroblasts derived from TAPP1R211L/R211LTAPP2R218L/R218L double knock-in embryos (Figure 5B). Overall, these findings support the notion that binding of TAPP1 and TAPP2 to PtdIns(3,4)P2 results in down-regulation of PI3K and the insulin signalling pathway.</p><p>It should be noted that the increased activation of Akt observed in TAPP1 and TAPP2 double knock-in mice did not lead to a marked increase in the phosphorylation of Akt substrates that we have analysed (PRAS40, GSK3 and FOXO1) in various insulin-responsive tissues (Figure 3). Similar results have been obtained in other studies where marked alterations in insulin sensitivity were correlated with changes in Akt activity, which were not reflected by monitoring phosphorylation of PRAS40, GSK3 or FOXO1. For example, knock-in mutation of the PDK1 PH domain to prevent interaction with PtdIns(3,4,5)P3 in mice results in a ~2-fold inhibition of Akt, resulting in marked insulin resistance, without significantly affecting the phosphorylation of Akt substrates [11,37]. This lack of effect on phosphorylation of Akt substrates is likely to be a result of the inherent spare capacity and amplification of signalling pathways. These data suggest that in vivo insulin sensitivity can be intricately correlated with Akt activation and even 1.5–2-fold changes in Akt activity, which are relatively modest, are sufficient to induce profound changes in insulin sensitivity. More work needs to be undertaken to identify the key substrates that Akt phosphorylates to regulate insulin sensitivity and how modest changes in Akt activity influence this process. It is also likely that PtdIns(3,4,5)P3 will stimulate other Akt-independent pathways that modulate overall insulin sensitivity.</p><p>It makes sense to employ PtdIns(3,4)P2 as a signal to down-regulate the PI3K pathway, as the levels of this 3-phosphoinositide peak later than those of PtdIns(3,4,5)P3. PtdIns(3,4)P2 would serve to down-regulate PI3K and signal the need for decreased formation of PtdIns(3,4,5)P3 production. Functionally, this would accelerate the return of insulin action to the pre-fed state. Further evidence supporting the notion that PtdIns(3,4)P2 acts as a negative regulator of the PI3K pathway is provided by analysis of SHIP2-knockout mice. Mice lacking SHIP2 that converts PtdIns(3,4,5)P3 into PtdIns(3,4)P2 displayed similar increased Akt activation in response to insulin in liver and muscle, as we have found in the TAPP1R211L/R211LTAPP2R218L/R218L double knock-in mice [14]. These results have been interpreted to result from increased levels of PtdIns(3,4,5)P3 in the SHIP2-knockout mice leading to increased activation of Akt. However, it would be interesting to explore whether the increased activation of Akt results instead from diminished recruitment of the TAPP1 and TAPP2 adaptor proteins to the plasma membrane as a result of decreased levels of PtdIns(3,4)P2 being produced in SHIP2-knockout mice. SHIP1 and SHIP2 have also been implicated in having roles in the negative regulation of the PI3K pathway in immune cells [38]. The activity of SHIP1 and SHIP2 is exquisitely controlled by interaction with growth factor receptors and by phosphorylation of tyrosine residues [39]. Presumably, being able to intricately regulate the activity of the SHIPs is vital, as they play dual roles in directly controlling absolute levels of PtdIns(3,4,5)P3, as well as acting as gatekeepers of PtdIns(3,4)P2 production that could in turn influence PI3K signalling networks via recruitment of the TAPP1 and TAPP2 adaptors.</p><p>An important question for future research is to define the mechanism by which recruitment of TAPP1 and TAPP2 to plasma membrane down-regulates the PI3K pathway. To date, TAPP1 and TAPP2 have been reported to interact via their C-terminal residues with at least three PDZ-domain-binding proteins, including PTPN13 (known previously as PTPL1 or FAP-1) [23], as well as the scaffolding proteins MUPP1 [20], syntrophin [24] and utrophin [25]. PTPN13 [40], MUPP1 [41], syntrophin and utrophin [42] are known to interact with a large number of binding partners involved in regulating numerous signalling pathways, so it may not be straightforward to deconvolute which of these is involved in regulating insulin sensitivity. It is also possible that TAPP1 and TAPP2 bind to other regulators of the PI3K pathway that have not been identified. PTPN13 is an attractive candidate to be a mediator of PI3K pathway signalling, as the three-dimensional structure of PTPN13 closely resembles the structure of PTP1B, one of the physiological PTPs that acts on the insulin receptor [43]. PTPN13 contains a positively charged pocket located near the catalytic site, reminiscent of the second phosphotyrosine-binding site in PTP1B, which is required to dephosphorylate peptides containing two adjacent phosphotyrosine residues as occurs, for example, in the activated insulin receptor [44]. Consistent with this, PTPN13, like PTP1B, interacted with and dephosphorylates a diphosphorylated insulin receptor peptide much more efficiently than monophosphorylated peptides. This indicates that PTPN13 may down-regulate the PI3K pathway by dephosphorylating the insulin or potentially other growth factor receptors that contain tandem phosphotyrosine residues [43].</p><p>To address the role of PTPN13 in regulating insulin sensitivity, we have generated catalytically inactive PTPN13C2374A/C2374A knock-in mice (S. Wullschleger and D.R. Alessi, unpublished work). These mice are viable, display no obvious phenotype and did not display marked insulin sensitization in the initial studies that we have undertaken (S. Wullschleger and D.R. Alessi, unpublished work). This suggests that PTPN13 may not be rate-limiting in the mechanism by which TAPP1 and TAPP2 control insulin sensitivity. In future work, it will be important to study which proteins are associated with TAPP1 and TAPP2 in insulin-responsive tissues and evaluate whether these are involved in down-regulating insulin signalling when recruited to the plasma membrane.</p><p>In conclusion, the TAPP1R211L/R211LTAPP2R218L/R218L double knock-in mice represent the first mouse model for these adaptor proteins and support the notion that TAPP1 and TAPP2 operate as negative regulators of the PI3K signalling pathway. As TAPP1 and TAPP2 are expressed in all tissues examined, these adaptors may have roles to play in modulating PI3K activity and PtdIns(3,4,5)P3 levels in systems beyond controlling insulin sensitivity. In future work, it will be important to define the mechanism by which TAPP1 and TAPP2 induce down-regulation of PI3K by binding to PtdIns(3,4)P2 and establish whether this system plays more general roles in other biological systems, such as in B-cells and T-cells where TAPP1 and TAPP2 are highly expressed. These results also indicate that if compounds could be developed that inhibit binding of TAPP1 and TAPP2 to PtdIns(3,4)P2, these could be deployed to improve insulin sensitivity in insulin-resistant diabetic patients. Interestingly, a recent study has shown that it is possible to design small molecules that bind to the 3-phosphoinositide-binding sites of PH domains [45] and it would be interesting to employ a similar approach to develop compounds that prevent TAPP1/TAPP2 binding to PtdIns(3,4)P2. It will also be important to identify the proteins that interact with TAPP1 and TAPP2 to control insulin sensitivity, as these might also represent new therapeutic targets for the treatment of insulin resistance.</p>

PubMed Author Manuscript

Exploring the Thickness-Dependence of the Properties of Layered Gallium Sulfide

Group III layered monochalcogenide gallium sulfide, GaS, is one of the latest additions to the two-dimensional (2D) materials family, and of particular interest for visible-UV optoelectronic applications due to its wide bandgap energy in the range 2.35–3.05 eV going from bulk to monolayer. Interestingly, when going to the few-layer regime, changes in the electronic structure occur, resulting in a change in the properties of the material. Therefore, a systematic study on the thickness dependence of the different properties of GaS is needed. Here, we analyze mechanically exfoliated GaS layers transferred to glass substrates. Specifically, we report the dependence of the Raman spectra, photoluminescence, optical transmittance, resistivity, and work function on the thickness of GaS. Those findings can be used as guidance in designing devices based on GaS.

exploring_the_thickness-dependence_of_the_properties_of_layered_gallium_sulfide

Introduction<!><!>Introduction<!>Sample Fabrication<!>Number of Layers and Thickness by Scanning Electron Microscopy<!>Structure and Photoluminescence by Raman Spectroscopy<!>Work Function and Morphology by Kelvin Probe and Atomic Force Microscopy<!>Valence Band Analysis by X-Ray Photoelectron Spectroscopy<!>Electrical Resistivity and Optical Transmittance Measurements<!>Colorimetry for Thickness Determination<!><!>Raman and Photoluminescence Spectra<!><!>Raman and Photoluminescence Spectra<!>Optical and Electrical Properties<!><!>Optical and Electrical Properties<!>Thickness Dependence of the Work Function<!><!>Thickness Dependence of the Work Function<!><!>Conclusion

<p>Low-dimensional layered semiconductors are receiving increasing interest due to the possibility to tailor their light–matter interaction by varying their properties with the number of layers, especially in the few-layer regime. In this context, semiconducting gallium monochalcogenides such as gallium sulfide, selenide, telluride, GaX (X = S, Se, Te) are one of the latest additions to the two-dimensional (2D) materials family, and of particular interest for visible-UV optoelectronic applications due to their wide energy bandgap (Lu et al., 2020). Specifically, gallium monosulfide, GaS, in the bulk form, has an indirect bandgap of 2.35 eV corresponding to the electronic transition from Γ→M points in the band structure (Chen et al., 2015), whereas the bandgap of GaS monolayer has been calculated to be above 3 eV, with different values reported in the range 3.1–3.3 eV (Zhuang and Hennig, 2013; Jung et al., 2015). A direct bandgap (Γ→Γ transition) in the range 2.8–3.0 eV has also been reported for bulk GaS (Kepinska et al., 2001; Ho and Lin, 2006) and at 3.88 eV for the monolayer (Jung et al., 2015). Because of those bandgap values, GaS has potential to exhibit photoluminescence (PL) in the green-blue spectral region and to be exploited as a UV photodetector (Chen et al., 2019; Lu et al., 2020).</p><p>GaS crystallizes in a highly anisotropic layered structure of increasing interest due to its non-toxicity, high chemical and thermal stability, and resistance to oxidation. Specifically, the basal surface of the layered structure, shown in Figure 1, is extremely inert to chemisorption of contaminants as the sticking coefficient for contaminants on GaS has been reported to be undetectably small (Williams and McEvoy, 1972); consequently, contaminants are only loosely bound on GaS basal surface, and, hence, can easily be removed by heating in vacuum or exposure to an electron beam (Williams et al., 1972).</p><!><p>(A) Side and (B) top view of the crystalline structure of 2H-GaS. (C) Images of the yellow bulk GaS crystal and of the mechanical exfoliated layers on glass substrate, showing the reduction of the yellow color by reducing the number of layers. (D) SEM images of the exfoliated GaS showing its layered structure down to monolayer. The inset in (D) is for the multilayer border where the piling up of the various layers can be clearly seen and layers can be counted.</p><!><p>Within the GaS layer, there is strong covalent binding, with two covalently bonded gallium, Ga, atoms between two layers of sulfur, S, atoms (see Figure 1A). Conversely, the inter-layers binding is of the Van der Waals type, classifying GaS as a 2D Van der Waals semiconductor. The unit cell of GaS is hexagonal, with the 2H-phase, β-GaS, crystal structure being the most energetically favorable polytype with lattice constants a = b = 3.587 Å, c = 15.492 Å. The interlayer separation of ∼0.75 nm is shown in Figure 1. Interestingly, the formation energy of a monolayer of GaS has been calculated to be 0.06 eV/atom, which is even lower than that of MoS2 (0.08 eV/atom) (Zhuang and Hennig, 2013), indicating that monolayer GaS can be obtained by mechanical exfoliation, as shown in Figures 1C,D.</p><p>In this work, we report a survey on the dependence of the main structural, optical, and electrical properties of mechanically exfoliated GaS on thickness, from bulk down to monolayer. This knowledge is relevant for designing devices and applications exploiting different thicknesses of GaS.</p><p>As Raman spectroscopy is the prime non-destructive characterization technique for layered materials, we report the Raman and PL data acquired at the same spot as a function of thickness from bulk GaS to monolayer. Furthermore, one of the most critical parameters in the design of novel electronic devices based on semiconducting layered materials is the work function (WF). This parameter is relevant for designing and understanding the band alignment at metal–semiconductor interfaces and in semiconducting heterostructures for photodetectors or phototransistors (Liu et al., 2021). For layered materials at the ultrathin regime, the WF is expected to critically depend on the number of layers of the material. For instance, it has been demonstrated that, for the prototypical transition metal dichalcogenide MoS2, the WF increases monotonically with the increase in the number of layers (Li et al., 2013; Choi et al., 2014). To the best of our knowledge, such study has not yet been performed on GaS; hence, we report the dependence of the GaS WF as a function of the GaS thickness as obtained by Kelvin probe force microscopy (KPFM). Moreover, x-ray photoelectron spectroscopy analysis of the valence band (VB) has been used to analyze the position of the valence band maximum (VBM) with respect to the Fermi level (EF) and corroborate the KPFM data for profiling bands as a function of thickness. To further guide the design of optoelectronic devices, we provide values of the resistivity and transmittance of GaS as a function of thickness in dark and under visible and UV illumination.</p><!><p>Few-layer GaS samples were obtained by mechanical exfoliation from commercially available bulk crystals purchased from 2D Semiconductors and HQ Graphene and transferred to glass and (285 nm) SiO2/Si substrate by the thermal tape method. Several exfoliations were executed in order to obtain samples of decreasing thickness as inferred by the disappearing of the yellow color characteristics of GaS bulk crystal, as shown in Figure 1. Glass substrates were cleaned by diluted H2O:H2O2 for 1 h at room temperature, followed by a water rinse to obtain an -OH terminated surface to improve adhesion with GaS.</p><!><p>Scanning electron microscopy (SEM) was carried out for the morphological characterization of the samples with a Zeiss Supra 40 FEG SEM equipped with a Gemini field emission gun. Analyses were carried out at an extraction voltage of 3 kV and a 30-µm aperture.</p><!><p>Raman spectroscopy (LabRam Horiba) was performed using a ×100 microscope objective (NA = 0.9) and exciting wavelengths of 633 and 473 nm. For the Raman measurements, performed with an excitation wavelength of 633 nm, the exfoliated flakes were deposited on glass substrates to avoid misinterpreting and overlapping of one of the GaS Raman modes with that at 303 cm−1 of conventionally used SiO2/Si substrates. The 473-nm laser was used to excite the GaS above the bandgap and also acquire PL spectra.</p><!><p>The WF of GaS flakes with different thicknesses was measured by Kelvin probe electrical force microscopy (KPFM) using the Autoprobe CP (Thermomicroscope) through the measurement of the local variation of the surface potential (SP). The sample topography and SP were recorded in a single-pass mode using gold-coated Si tips (their frequency is ∼80 Hz) in non-contact mode. The oscillating potential, V ac, applied to the tip is 5 V at a frequency ω of 13 kHz. The samples were electrically connected to the ground of the microscope (the sample stage).</p><p>For the Kelvin probe force microscopy experiments, the flakes were deposited on a reference Au/Si substrate, as the WF of gold at 4.75 eV (as measured by us on the same equipment and corroborated by x-ray photoelectron spectroscopy measurements) was used as reference. All measurements were collected in air at room temperature.</p><!><p>For profiling the bands' energy levels, we determined the position of the VBM with respect to the Fermi level by x-ray photoelectron spectroscopy (XPS) using a Scanning XPS Microprobe (PHI 5000 Versa Probe II, Physical Electronics) equipped with a monochromatic Al Kα x-ray source (1,486.6 eV), with a spot size of 200 µm. Survey (0–1,200 eV) and high-resolution spectra (C 1s, O 1s, S2p, S2s, Ga2p3, Ga3d, and valence band region) were recorded in FAT mode at a pass energy of 117.40 and 29.35 eV, respectively. Spectra were acquired at a take-off angle of 45° with respect to the sample surface. Surface charging was compensated using a dual beam charge neutralization system, and the hydrocarbon component of C1s spectrum was used as internal standard for charging correction, and it was fixed at 285 eV.</p><!><p>Electrical current–voltage, I–V, measurements were performed by the Keithley617 Programmable Electrometer. The voltage source has been used in conjunction with the electrometer section, to apply to the samples voltages from −2 to +2 V, where GaS has ohmic behavior. Contacts were made using silver. Current was measured in the dark whereas photoresponse was investigated in the visible range under a 100 mW·cm−2 AM1.5 spectrum lamp and in the UV range using a 405-nm laser of 250 mW cm−2 as source.</p><p>UV–Vis transmittance spectra were measured on the same glass samples with a Perkin Elmer Lambda 900 spectrometer.</p><!><p>Although mechanical exfoliation has become a widely used technique to achieve 2D layers, one of its main drawbacks is the difficulty in obtaining large area samples with homogeneous number of layers (as it can also be inferred by Figure 1D), and it generally results in randomly distributed flakes of different thickness. Consequently, there is a need for non-destructive, reliable, effective, and fast methods for inferring thickness. Interestingly, because of the contrast in the optical properties between GaS flake and its substrate, a full gamut of colors allows one to identify the thickness of mechanically exfoliated GaS transferred onto substrates, as shown in Figure 2. Optical microscopy methods relying on colorimetry can provide an effective solution to this problem. By calculating the reflectance of a system consisting of a GaS layer of variable thickness on an infinite substrate using a Fresnel laws-based model, and then its conversion to color coordinates using color matching functions, it is possible to predict the apparent color of a GaS flake of a specific thickness on a given substrate. Following this procedure, we have developed a methodology (Gutierrez et al., 2021) and a code (Gutiérrez, 2021) that generates color rulers for the quick assessment of the thickness of GaS flakes on various substrates. As an example, Figure 2 shows the color evolution of GaS as a function of thickness on glass and on 285-nm SiO2/Si substrates as seen under an optical microscope when running a Raman measurement. By comparing the color of the flakes appearing under the microscope of the Raman system with those color rulers, it is possible to infer the thickness of the GaS flakes.</p><!><p>Color evolution of GaS flakes as a function of their thickness on (A) glass and on (B) 285-nm SiO2/Si substrates as determined according to the free-available code developed (Gutiérrez, 2021). The last image on the right is an example of an optical micrograph of a non-homogeneous exfoliated flake, with the different colors corresponding to different number of layers.</p><!><p>Figure 3 shows the thickness dependence of the Raman spectra of few-layer GaS samples. The Raman spectrum of bulk β-GaS (space group P63/mmc and point group D4 6h), as well as of thick layers, is characterized by six modes at 22.8 cm−1 ( E2g2 ), 74.7 cm−1 ( E1g1 ), 189 cm−1 ( A1g1 ), 291.8 cm−1 ( E1g2 ), 295.8 cm−1 ( E2g1 ), and 360.9 cm−1 ( A1g2 ). The most intense and investigated peaks are the A1g1 , A1g2 , and E2g1 (the latter often including the contribution of the nearby E1g2 ), whose vibrational modes are sketched in Figure 3A. Conversely, GaS monolayer (space group P-6m2 and point group D1 3h) shows the E1g2 barely distinguishable as shown in Figure 3B. For a more accurate analysis, line-shape analysis for each Raman mode was performed by using one Lorentzian component. Central Raman shifts (ω) and the full width at half maximum FWHM (Γ) for each of the spectra recorded are shown in Table 1. No significant variation in the peaks position can be observed for all modes; similarly, the FWHM variation from thick layers to monolayer is within 1 cm−1. This negligible dependence of Raman modes on thickness is mainly due to weak inter-layer interactions, and it is consistent with a previous work where it is reported a red-shift in the A1g1 mode of only 1.4 cm−1 when going from the monolayer (187.6 ± 0.3 cm−1) to a 38-nm-thick layer (189.0 ± 0.1 cm−1), whereas a constant position within the uncertainty was reported for A1g2 (Alencar et al., 2020). Similarly, the broadening of all Raman modes is within 1 cm−1 going from monolayer to bulk. It is worth mentioning that for high-crystalline quality GaS, the 1:1 ratio between A1g1 and A1g2 is preserved in the range of thickness from bulk to monolayer, indicating that the stacking order of the layers is preserved during exfoliation.</p><!><p>Typical Raman spectra of (A) thick GaS and (B) few-layer GaS raging from the monolayer (ML) to six layers transferred onto glass substrates (acquired using a 633-nm laser source). (C) PL spectra of same samples with different thickness acquired irradiating the sample with a 473-nm laser. The blue intense spectrum is for a >300-nm-thick sample; the black line is for the 1–6 L samples. Intermediate states are for layers with a thickness of ≈300 nm.</p><p>Raman shifts (ω) and FWHM (Γ) for the three modes A1g1 , E2g1 , and A1g2 measured for GaS layers with thickness ranging from monolayer to 120 nm. The 633-nm laser excitation was used.</p><!><p>Literature has given little attention to the PL of GaS. Figure 3C shows the PL spectra for samples of different thickness using a continuous-wave excitation from a 473-nm laser source. Noteworthy, the monolayer GaS as well as all the layered samples up to approximately 300 nm do not show any PL. A sharp PL peak at approximately 2.5 eV starts to be seen for thicknesses above 300 nm and increases with the increase in thickness, as shown in Figure 3C. This can be explained considering two main factors: 1) GaS is an indirect bandgap semiconductor requiring both photons and phonons for radiative recombination and defect-assisted recombination plays an important role, as due to the requirement of phonon momentum conservation, the radiative recombination on the indirect transition will be inefficient and sensitive to traps (Leonhardt et al., 2020). For monolayer and few layers, if the trap density is mainly localized at the substrate/GaS interface, traps result in the observed quenching of PL for the few-layer regime, as the substrate-interaction traps act as recombination centers in the bandgap. This is supported by literature (Shin et al., 2016), reporting that the surface roughness of the underlying substrate can result in inhomogeneous strain that leads to bandgap modifications in thin transition metal dichalcogenides causing the appearance of hole traps. The origin of these traps, however, is still under investigation. 2) By increasing the thickness, both the direct and indirect bandgaps are affected by the interlayer interactions along the c-axis and the appearance of new radiative recombination paths. This condition leads to the appearance of a distinct PL peak.</p><!><p>Figure 4 shows typical optical transmittance spectra measured on 30–40 nm and ≈5-µm-thick GaS. Simulated optical transmittance spectra are also shown as reference for different GaS thicknesses from 1 Ml to 5 μm. The simulations were performed using the Transfer Matrix Method (Born et al., 1999) with the assumption of a flat interface multilayer GaS/glass system. The optical constants of GaS used in the simulations were experimentally measured by spectroscopic ellipsometry on a bulk crystal c-axis oriented. Interestingly, for a highly oriented defect-free GaS, the calculated spectra indicate a transmittance of 99.9% for the monolayer, 94.75% for approximately 10 layers, and 82.3% for approximately 20 nm (i.e., 20 layers), approaching the bulk 80% transmittance for a thickness higher than 20 nm. The measured optical transmittance line shape for the 30–40 nm GaS is in good agreement with the simulations performed for GaS layers with thickness in the range 10–50 nm. These spectra are characterized by a pronounced dip at ≈3.9 eV consistent with the interband critical point GS2 in the dielectric function as reported by Schlüter et al. (1976) and Isik et al. (2013). In the low-energy range, the lower measured transmittance as compared with simulation can be associated with phenomena not considered in the model such as surface roughness, scattering, as well as inhomogeneities in the sample. In the case of the ≈5-µm layer, the main difference between the measured and simulated spectra is in the onset and slope of the transmittance around the energy bandgap and the damped interference system. In this thick case, these differences can likely be attributed to polycrystallinity and defects or doping of the sample. This is supported by comparing the optical transmittance obtained by the model and that measured in a single crystal by Nakamura et al. (2021), in which the spectra also show a sharp step around the energy bandgap. These results provide evidence about the high density of defects introduced in the layers by the mechanical exfoliation process, which are also potentially the radiative traps causing the PL in Figure 3C for the thick GaS layers.</p><!><p>(A) Calculated and measured transmittance as a function of the GaS thickness. (B) Thickness dependence of the resistivity of GaS layers under dark, visible-, and UV-light irradiation. For comparison, the yellow shadowed area indicates the range of the dark resistivity values reported in literature for GaS (Manfredotti et al., 1976; Mancini et al., 1983; Micocci et al., 1990; Szałajko and Nowak, 2007).</p><!><p>Furthermore, GaS has been demonstrated to be exploitable in blue-UV photodetectors (Chen et al., 2019; Lu et al., 2020). A critical parameter to be considered in the photodetector design is the resistivity of the active layer under light irradiation. Therefore, the photoresistivity of layered GaS with thickness ranging from ≈40 to 1900 nm transferred onto glass was also investigated. The calculated resistivity as a function of the GaS thickness with and without light irradiation is shown in Figure 4B.</p><p>In the dark state, GaS shows a typical resistivity of ≈107 Ω·cm independently of the thickness. This value is consistent with those reported by other authors (Manfredotti et al., 1976; Mancini et al., 1983; Micocci et al., 1990; Szałajko and Nowak, 2007). When illuminated, the GaS resistivity decreases for all the investigated thicknesses, consistently with an increase of the light absorption depth and generation of photocurrent in the GaS layers. Furthermore, the resistivity is lower when samples are illuminated by UV light than visible light. The lower resistivity values obtained under UV indicates a larger amount of free carriers generated due to the fact that all the excitation photons have an energy (405 nm/3.05 eV) above the GaS bandgap, making the electron–hole pair generation more efficient than in the case of the visible light source, for which part of the emission spectrum contains photons below the GaS bandgap unable to generate photocarriers.</p><!><p>Figure 5 illustrates the use of KPFM to determine the WF of GaS layers with different thickness, as it appears from the optical image (Figure 5B) of the sample, where the different colors correspond to flakes of different thickness, which have been mapped by atomic force topographies. The exfoliated GaS was deposited on a gold substrate, as shown in Figure 5A. A typical topography and corresponding surface potential SP map, 15 × 15 μm, are shown in Figures 5C,D, respectively. The difference between the SP of GaS flake, SPGaS, and the SP of the gold, SPAu, quantifies the difference in their Fermi levels according to the following relation: ΔSP=SPGaS−SPAu=WFtip−WFGaS−WFtip+WFAu=WFAu−WFGas (1) where WFtip is the tip work function, WFAu is the gold work function, and WFGaS is the work function of the GaS flake.</p><!><p>(A) Picture of GaS flakes on gold; (B) optical image of a 750 μm × 750 μm area of GaS flakes (C,D) examples of 15 μm × 15 µm topographic and SP maps for a 26-nm-thick GaS flake. (E) Representative topographical (black line) and SP (red line) profiles for a GaS flake with 26-nm thickness against gold. (F) Thickness dependence of the difference in surface potential (SP) between gold and GaS. (G) Thickness dependence of the GaS work function (WF).</p><!><p>Figure 5E shows representative topographic and SP profiles obtained for a GaS flake with 26-nm thickness against gold. Specifically, the SP of the GaS flake is higher than the Au SP, i.e., SPGaS > SPAu, corresponding to a WF of the GaS flake 250 meV lower than that of the gold reference, which we previously calibrated to be 4.75 eV (Giangregorio et al., 2015).</p><p>By plotting the difference between the surface potential of the GaS flakes with a known thickness, SPGaS, and the surface potential of gold, SPAu, and the corresponding WFGaS, as shown in Figures 5F,G, we found that GaS layers always have a SP higher than Au independently of their thickness. Moreover, the WFGaS increases with the decrease in the GaS thickness from 3.55 ± 0.10 eV for GaS bulk to 4.70 ± 0.05 eV for GaS monolayer.</p><p>In order to explain this trend of the WF, we consider that 1) going from bulk to monolayer, the WF becomes more and more sensitive to the chemical and physical conditions at its surface, and 2) it depends on crystalline orientations, surface contamination, and surface roughness, which induce stress fluctuations affecting the Fermi level as well as the electrostatic potential in the vicinity of surface. It is worth pointing out that adsorption of atoms or molecules on GaS changes the surface dipole layer and hence the WF; e.g., electronegative adsorbates (e.g., O, C, and S) increase the WF. Oxygen adsorption involves localized orbital overlap and charge transfer between the adsorbate and surface atoms. The GaS bulk WF value is small compared with the oxygen ionization energy (13.618 eV), causing electron transfer from the GaS layer. Consequently, the WF increases as the negative pole of the adsorded oxygen molecule points toward the vacuum, so the surface space charge or surface dipole presents an electrostatic field that causes an increase in the WF. The effect of adsorbates becomes more important with the decrease in the number of layers down to monolayer. The presence of those C and O adsorbates was confirmed by x-ray photoelectron spectroscopy (XPS).</p><p>These measurements of WF are useful to profile the bands and Fermi level variations as a function of GaS thickness. For monolayer GaS, we consider as reference the values calculated by Zhuang and Hennig (2013) of the bandgap Eg = 3.19 eV, of the valence band maximum VBM = −6.77 eV and of the conduction band minimum CBM = −3.58 eV, as shown in Figure 6. In profiling the bands, the WF values were complemented with experimental values of the difference of the VBM with respect to Fermi level derived from the XPS valence band analysis. In the case of the few-layer GaS, we profiled the measured values of the WF and EF of 4.70 ± 0.05 eV and of the VBM position with respect to Fermi level of 0.25 ± 0.05 eV. Assuming an energy bandgap similar to that of GaS monolayer, the VBM and CBM were set at −4.95 ± 0.10 and −1.76 ± 0.10 eV, respectively. For comparison, the values of VBM, CBM, and EF reported by Carey et al. (2017) for few-layer GaS are also shown. In the case of bulk GaS, the measured WF = 3.55 ± 0.10 eV, the difference of 0.65 ± 0.15 eV for the VBM with respect to EF, and the energy bandgap of 2.35 eV led to VBM and CBM at −4.2 ± 0.25 and −1.85 ± 0.25 eV, respectively. From these bands and Fermi level profiling, it can be inferred that the analyzed GaS samples are p-type semiconductors. This is consistent with other studies on GaS crystals grown by the Bridgman method (Lieth et al., 1969). As an example, for GaS monolayer, it has been reported that it becomes p-type under the gallium-poor and sulfur-rich conditions (Chen et al., 2015). The adsorbates mentioned above as well as the interface traps mentioned in Figure 3C could also contribute to the p-type doping of exfoliated GaS layers. The detailed identification of those radiative traps and p-type doping defects is in progress.</p><!><p>Profiling of the valence band maximum (VBM), conduction band minimum (CBM), and Fermi level with respect to the vacuum level for GaS monolayer, few layers, and bulk considering theoretical (Zhuang and Hennig, 2013; Carey et al., 2017) and experimental values.</p><!><p>GaS layers of different thickness have been exfoliated and transferred to glass substrates. Different properties, such as structural properties from Raman spectra, PL, optical transmittance, resistivity, and WF have been investigated as a function of the number of layers. The Raman spectra measured in layers with thickness ranging from the monolayer to 120 nm show no significant variation in the peak position and broadening, whereas their intensity is proportional to sample polarizability and, hence, increases with thickness. A model based on a planar stack of layers is able to reproduce the line shape of the optical transmittance spectra for few layers and micron-thick GaS layers. Phenomena of surface roughness, inhomogeneities, defects, or unintentional doping clearly decrease the transmittance. GaS dark resistivity is in the range of ≈107 Ω·cm, independently of the thickness. Under visible and UV illumination, the resistivity decreases, and a pronounced dependence on GaS thickness is found. Finally, the analysis of the WF, using Kelvin probe force microscopy, shows an increase in the WF going from Bulk GaS down to monolayer. Accordingly, GaS bands have been profiled as a function of thickness. Although the study on this new 2D material is in progress, those trends can be useful to design optoelectronic devices based on GaS.</p>

PubMed Open Access

Synthesis and application of a novel bis-1,2,3-triazole ligand containing a 2,2\xe2\x80\x99-bipyrrolidine core

Herein, we describe the synthesis of a novel bis-1,2,3-triazole ligand which contains an internal N-alkylated 2,2\xe2\x80\x99-bipyrrolidine linker. By using simple starting materials, the ligand could be generated in good yield through several synthetic steps. To investigate the potential for the application of this ligand in transition metal catalysis, we generated a bis-Au(I) complex in nearly quantitative yield and examined its reactivity in the context of alkyne hydration. Both alkyl and aryl terminal alkynes could be efficiently converted to their corresponding ketones in nearly quantitative yields with only 1% catalyst loading under mild conditions.

synthesis_and_application_of_a_novel_bis-1,2,3-triazole_ligand_containing_a_2,2\xe2\x80\x99-bipyrrol

Introduction<!>Results and Discussion<!>Conclusions<!>General.<!>Synthesis of ethyl 3-phenylpropiolate (2).<!>Synthesis of ethyl 5-phenyl-1H-1,2,3-triazole-4-carboxylate (3).<!>Synthesis of (5-phenyl-1H-1,2,3-triazol-4-yl)methanol (4).<!>Synthesis of methyl 2-[4-(hydroxymethyl)-5-phenyl-2H-1,2,3-triazol-2-yl]-2-methylpropanoate (5).<!>Synthesis of methyl 2-[4-(bromomethyl)-5-phenyl-2H-1,2,3-triazol-2-yl]-2-methylpropanoate (6).<!>Synthesis of dimethyl 2,2\xe2\x80\x99-{[(2,2\xe2\x80\x99-bipyrrolidine)-1,1\xe2\x80\x99-diylbis(methylene)]bis-(5-phenyl-2H-1,2,3-triazole-4,2-diyl)}bis-(2-methylpropanoate) (ligand 7).<!>Synthesis of complex 8.<!>General procedure for alkyne hydration (Scheme 4).

<p>Over the past century, ligand design has been integral to the realization of new catalytic approaches relying on transition metals. In many cases, subtle changes in ligand structure and electronic character have dramatic influences on the stability and reactivity of the metal to which it is bound. This intimate relationship between structure and reactivity has therefore been an impetus to pursue new ligand systems. Despite the structural complexity inherent to many effective classes of ligand, much of this work relies on the inherent properties of privileged small molecules, which when derivatized in unique ways, form structurally unique ligand scaffolds.</p><p>In the past decade, our group has found success in implementing functionalized 1,2,3-triazoles (TAs) as ligands for late transition metals. Thus far, we have demonstrated that these electron-poor heterocycles have a high affinity to bind to mid to late transition metals such as Rh,1 Pd,2 Fe,3 Ir,4 and Au.5–9 In the case of Au(I), neutral TA ligands can dramatically enhance the thermal stability and chemoselectivity of the Au-cation.10,11 In the absence of neutral secondary ligands or strong sigma-donating spectator ligands, Au-cation often decomposes to a metallic form through external reduction or disproportionation pathways.12 By harnessing the productive partnership between Au(I) and TA, we have broadened the scope of Au-catalysis by generating new catalyst libraries that offer broad electronic and structural range.</p><p>Our insight and background in TA synthesis and its metal coordination has naturally led us to consider new and abstract ligand systems containing this heterocycle. Furthermore, we were particularly interested in the prospect of a bis-TA ligand that could simultaneously accommodate two metal atoms, as described in Scheme 1. In the context of Au-catalysis, generating a new class of bis-Au(I) complexes may provide new avenues to access and productively utilize Au-Au intermediates. Additionally, the 2-coordinate nature of Au(I) would allow us to avoid the formation of chelation complexes. To establish this structural paradigm, we realized it would necessary to identify an adequate TA scaffold and linker. After considering several viable linkers, we chose to investigate the 2,2'-bipyrrolidine structure, as this would provide a mild synthetic platform hinging on N-alkylation, a step easily conceived through an SN2 reaction. Moreover, a TA subunit containing a leaving group would be paramount to execute this design.</p><!><p>As described in the introduction, our first task was to generate a TA subunit that would be poised for amination in the presence of 2,2'-pyrrolidine backbone. This would ultimately involve discerning a TA-synthesis that would easily allow the incorporation of a leaving group. With this in mind, we could easily adopt an adequate alkyne precursor for the formation of the heterocyclic core. As described in Scheme 2, alkyne 2 could be easily generated under typical alkyne-acylation procedures. From alkyne 2, we could efficiently access TA 3 through cycloaddition with NaN3 with good overall yield. The ester on TA 3 was then reduced to the corresponding alcohol using excess LiAlH4, which will be important for the generation of a good leaving group later. Following reduction, N2-alkylation was performed to give TA 5 in high yields and N2 selectivity. The N2 selectivity was confirmed by performing a series of 1D NOE experiments on the product obtained. This step was performed for two reasons, the first being to inhibit unwanted dimerization when the subsequent alkyl bromide is formed, and the second being to specifically investigate N1 or N3 binding with the Au-cation. Compound 5 was then treated with PBr3 at lower temperatures to give the key and final product, TA 6.</p><p>Following the successful synthesis of TA 6, we proceeded with the di-N-alkylation of (2R,2′R)-2,2′-bipyrrolidine as illustrated in Scheme 3. After a brief reaction screening, it was determined that treating the two components with potassium carbonate at room temperature in DCM efficiently provided target 7 in good yield. With the desired compound in hand, we then generated the bis-Au complex 8 by treating the ligand with two equivalents of Ph3PAuCl and two equivalents of AgOTf. Upon the addition of AgOTf, a white precipitate (AgCl) could be seen almost immediately. The complex prepared through this step was characterized using 1H, 13C, and 31P NMR analysis. However, we have not been able to grow an adequate crystal for X-ray analysis. It is important to note that the broad nature of the NMR spectra for the complex suggests some dynamic bonding behavior in solution. This is likely due to the coalescence of a number of different regioisomer signals at room temperature. Nevertheless, the dramatic change in 1H and 13C NMR spectra are strongly indicative of some significant level of bonding interaction. Additionally, the difference in chemical shift (δ 25.8 ppm) observed in the 31P NMR signal relative to pure Ph3PAuCl (δ 34.3 ppm) and Ph3PAuOTf (δ 28.8 ppm) is a compelling result that strongly supports complex formation.10 At this point though, it is difficult to determine any specific binding motif in solution. It is also important to note here that the complex generated when using only one equivalent of Ph3PAuCl was quite unstable and clearly started to decompose upon work up and solvent removal.</p><p>Following the synthesis of complex 8, we next wanted to establish some catalytic activity. To accomplish this, we investigated terminal alkyne hydration under relatively mild conditions. At room temperature, very little alkyne hydration could be seen after 24 hours. However, as shown in Scheme 4, when the reaction temperature was raised to 50 °C both alkyl and aryl terminal alkynes could undergo hydration in almost quantitative yield. Interestingly, only the free ligand could be observed at the end of the reaction, which suggests that the Au-catalyst is decomposing upon reaction completion. This is perhaps due to a decomposition pathway resulting from competitive coordination from water or the resulting ketone product. Despite the decomposition of the catalyst observed upon complete conversion of the starting material, we were surprised to see such high efficiency given the presence of a tertiary amine within the ligand. In some cases, amines and thiols lead to Au(I)-catalyst deactivation due to their high nucleophilicity.13 However, the bulkier tertiary amine and competitive coordination of other functional groups in the ligand is likely to inhibit deactivation to any great extent.</p><!><p>As reported here, we have been able to synthesize a new bis-TA ligand which appeared to bind effectively to Au(I). Based on NMR data, the ligand is interacting with Au(I) in a very dynamic fashion. This catalyst exhibits good thermal stability and high efficiency in the hydration reaction of terminal aryl and alkyl alkynes, which proves that the tertiary amines within the ligand do not lead to Au(I) deactivation. In summary, we believe these results may give way to new ligand-metal paradigms that will offer insight to new methodological advances.</p><!><p>All reactions were carried out under an atmosphere of nitrogen using oven or flame dried glassware and standard syringe/septa techniques. Unless noted, all commercial reagents and solvents were used without further purification. Flash column chromatography was performed on 230–430 mesh silica gel. Analytical thin layer chromatography was performed with pre-coated, glass-baked plates (250μ) and visualized by fluorescence or charring with potassium permanganate stain. Melting point were recorded on Mel-Temp. 1H NMR, 13C NMR and 31P NMR spectra were recorded on an Agilent 400 MHz spectrometer. Chemical shifts for starting materials and products were reported relative to tetramethylsilane (0.00 ppm) or CD3OD (3.31 ppm) for 1H NMR data, CDCl3 (77.0 ppm) or CD3OD (49.9) for 13C NMR and H3PO4/D2O for 31P NMR data. Data are presented as follows: chemical shift (ppm), multiplicity (s = singlet, d = doublet, t = triplet, dd = doublet of doublets, m = multiplet, br = broad), coupling constant J (Hz) and integration. ESI-MS spectra were collected using a Thermo Scientific Orbitrap Q Extractive Plus (Bremen, Germany) in the positive ion mode. The samples were infused with a flow rate of 10 μL/min and sprayed at a high voltage of 5 kV.</p><!><p>To a nitrogen flushed round bottom flask with a solution of phenylacetylene (2.04 g, 20 mmol) in distilled THF (20 mL) at –78 °C was added n-BuLi (8.4 mL, 21 mmol, 2.5 M in hexane) dropwise. The solution was stirred at this temperature for approximately 1 hour. After this time, ethyl chloroformate (2.3 mL, 24 mmol, neat) was added at −78 °C. The solution stirred at this temperature and was monitored by TLC. Complete conversion could be observed after approximately 2.5 hours. Upon completion, the reaction was quenched through the addition of a saturated solution of NH4Cl at room temperature. The organic layer was extracted with ethyl acetate (3 × 20 mL) and dried using sodium sulfate. This solution was filtered through a plug of cotton and concentrated by rotary evaporation. The crude reaction product was then purified using column chromatography (gradient from 20:1 to 5:1 hexanes: ethyl acetate) to give 2.8 g (80% yield) of the alkyne as a clear oil. The 1H and 13C NMR spectra for this internal alkyne matched identically to the many previous reported syntheses.14, 15, 16</p><!><p>To a gently stirred solution of alkyne 2 (1.75 g, 10 mmol) in DMSO (20 mL, 0.5 M) was added NaN3 (1.9 g, 30 mmol) in four portions over 20 minutes. Once the NaN3 was completely dissolved, the unsealed reaction was heated to 80 °C for 8 hours. Upon completion, distilled water was slowly added to the reaction followed by slow and incremental addition of 1.0 M HCl until a pH of 1 was reached. The solution was then extracted using DCM (3 × 30 mL). The organic layer was washed with brine and dried over sodium sulfate. Following filtration and concentration of the organics by rotary evaporation, the crude product was purified via recrystallization (5:1 Hexanes: DCM) to give triazole 3 (1.7 g, 80%) as an off-white powder. mp 92–93 °C. 1H-NMR (400 MHz; CDCl3): δ 7.79 (m, 2H), 7.40 (m, 3H), 4.36 (q, J 6.9 Hz, 2H), 1.28 (t, J 7.2 Hz, 3H); 13C-NMR (101 MHz; CDCl3): δ 161.2, 129.7, 129.2, 128.2, 61.6, 13.9. HRMS Calculated for C11H12N3O2 [M+H]+ : 218.0924, Found: 218.0930.</p><!><p>17 A solution of 3 (1.5 g, 7 mmol) in THF (23 mL, 0.3 M) was cooled to 0 °C in an ice bath. LiAlH4 (380 mg, 10 mmol) was then added to the solution in four portions over 20 minutes. The reaction mixture was then warmed to room temperature and stirred for 30 minutes. Upon completion, as confirmed by TLC, the reaction was then cooled back down to 0 °C and quenched through dropwise addition of a saturated ammonium cloride solution. The reaction mixture was then acidified by addition of 1.0 M HCl. This was then extracted using DCM (3 × 30 mL). The organics were then dried over sodium sulfate, filtered and concentrated via rotary evaporation. The crude reaction material was then purified using trituration and recrystallization (approximately 3:1 Hexanes: DCM) to give triazole 4 (1.0 g, 82 %) as a white solid. m.p. 133–134 °C. 1H-NMR (400 MHz; CD3OD): δ 7.59–7.57 (m, 2H), 7.38–7.34 (t, J 7.5 Hz, 2H), 7.26–7.24 (m, 1H), 4.65 (s, 2H); 13C-NMR (101 MHz; CD3OD): δ 143.2, 140.7, 132.3, 128.7, 127.1, 126.9, 54.6. Calculated for C9H10 N3O [M+H]+ : 176.0818, Found: 176.0821.</p><!><p>A solution containing alcohol 4 (875 mg, 5 mmol), methyl α-bromoisobutyrate (1.8 g, 10 mmol) and potassium carbonate (1.37 g, 10 mmol) in DMF (10 mL, 0.5 M) was heated to 60 °C for 6 hours. The reaction was cooled to room temperature and water (20 mL) was added. This solution was then extracted with diethyl ether (3 × 20 mL) and dried over sodium sulfate. This mixture was filtered and concentrated using rotary evaporation. The crude product was then purified using column chromatography (gradient from 10:1 to 3:1 hexanes: ethyl acetate) to give 1.20 g (85 % yield) of triazole 5 as a viscous colorless oil. 1H-NMR (400 MHz; CDCl3): δ 7.83–7.80 (m, 2H), 7.45–7.41 (m, 2H), 7.36 (dd, J 8.6, 6.1 Hz, 1H), 4.86 (s, 2H), 3.69 (s, 3H), 1.97 (s, 6H); 13C-NMR (101 MHz; CDCl3): δ 172.4, 145.5, 143.8, 130.4, 128.7, 128.3, 127.5, 67.8, 56.2, 52.9, 25.3. Calculated for C14H17N3NaO3 [M+Na]+ : 298.1162, Found: 298.1164.</p><!><p>A solution of ester 5 (1.10 g, 4 mmol) in DCM (10 mL, 0.4 M) was cooled to 0 °C under a stream of nitrogen gas. PBr3 (1.7 g, 6.4 mmol) was then added to the solution dropwise. The reaction was then stirred at this temperature and monitored by TLC until completion. Upon completion, a saturated sodium bicarbonate solution was added to the reaction at 0 °C. Once gas evolution was no longer apparent, water (10 mL) was added to the crude reaction mixture. This solution was then extracted using DCM (3 × 20 mL). The organic extracts were then dried over sodium sulfate, filtered and concentrated. The crude reaction mixture was then purified using column chromatography (gradient from 12:1 to 6:1 hexanes: ethyl acetate) to give 835 mg (62% yield) of triazole 6 as a white solid. Mp 90–92 °C. 1H-NMR (400 MHz; CDCl3): δ 7.80 (dd, J 8.2, 1.1 Hz, 2H), 7.48–7.44 (m, 2H), 7.40–7.38 (m, 1H), 4.66 (s, 2H), 3.70 (s, 3H), 1.98 (s, 6H); 13C-NMR (101 MHz; CDCl3): δ 172.1, 145.6, 141.0, 130.0, 128.8, 128.6, 127.6, 68.2, 53.0, 25.3, 22.6. Calculated for C14H17BrN3O2 [M+H]+ : 338.0499, Found: 338.0499.</p><!><p>A solution of 6 (674 mg, 2 mmol), (2R,2′R)-2,2′-bipyrrolidine (265 mg, 1.9 mmol) and potassium carbonate (410 mg, 3 mmol) in DCM (7 mL, 0.3 M) was stirred at room temperature. The reaction was monitored for completion using TLC. Upon completion, water (10 mL) was added and extractions using DCM (3× 15 mL) were performed. The organic layers were dried over sodium sulfate, filtered and concentrated. The crude reaction mixture was then purified using column chromatography (gradient from 5:1 to 1:1 hexanes ethyl acetate) to give 1.05 g (85% yield) of ligand 7 as a light yellow solid. mp 64–65 °C, 1H-NMR (400 MHz; CDCl3): δ 7.94 (d, J 7.3 Hz, 4H), 7.37 (t, J 7.3 Hz, 4H), 7.31 (d, J 7.0 Hz, 2H), 3.97 (d, J 12.8 Hz, 2H), 3.66 (s, 6H), 3.51 (d, J 12.8 Hz, 2H), 2.83 (t, J 7.6 Hz, 2H), 2.76 (t, J 6.5 Hz, 2H), 2.28 (td, J 9.6, 6.8 Hz, 2H), 1.92 (2s, J 3.1 Hz, 12H), 1.77 (m, 4H), 1.63 (m, 4H); 13C-NMR (101 MHz; CDCl3): δ 172.5, 146.0, 142.8, 131.2, 128.3, 128.1, 127.9, 67.5, 64.8, 54.7, 52.8, 49.6, 26.1, 25.4, 25.2, 23.9. Calculated for C36H47N8O4 [M+H]+ : 655.3715, Found: 655.3726.</p><!><p>Ligand 7 (82 mg, 0.125 mmol) and Ph3PAuCl (123 mg, 0.25 mmol) were dissolved in DCM (625 μL, 0.2 M) at room temperature. To this solution was added AgOTf (64 mg, 0.25 mmol). AgCl immediately precipitated out of solution as a white solid. This solution stirred for 2 hours followed by gravity filtration through two pipettes filled halfway with celite. The filtrate was then collected and concentrated to give a light yellow solid. Recrystallization was then performed to give 190 mg (95% yield) of complex 8 as an off-white solid. 1H-NMR (400 MHz; CDCl3): δ 7.56–7.34 (m, 38H), 4.49 (m, 2H), 3.67 (s, 7H), 3.39 (s, 1H), 2.91 (s, 1H), 1.90 (m, 22H); 13C-NMR (101 MHz; CDCl3): δ 172.1, 134.1, 133.9, 133.8, 132.4, 129.6, 129.5, 129.4, 129.2, 129.1, 128.1, 127.8, 68.7, 53.1, 25.2; 31P-NMR (162 MHz; CDCl3): δ 25.8.</p><!><p>In an NMR tube, alkyne (phenylacetylene: 24 mg, 0.24 mmol; 1-hexyne: 20 mg, 0.24 mmol) complex 8 (5 mg, 0.0024 mmol) and p-xylene (25 mg, 0.24 mmol, internal standard) were dissolved in deuterated methanol (0.6 mL, 0.4 M). The reaction was then heated to 50 oC and monitored every 10 minutes by 1H NMR until completion.</p>

PubMed Author Manuscript

SIGNALING PATHWAYS REGULATING THE BLOOD-TESTIS BARRIER

Throughout mammalian spermatogenesis, preleptotene/leptotene spermatocytes traverse the blood-testis barrier during stages VIII-XI of the seminiferous epithelial cycle while trapped within a dynamic intermediate compartment that is sealed at north and south poles by tight junctions, basal ectoplasmic specializations, desmosomes and gap junctions. In order for spermatocytes to gain entry into the adluminal compartment of the seminiferous epithelium for continued development, \xe2\x80\x98old\xe2\x80\x99 junctions present above migrating spermatocytes disassemble, while \xe2\x80\x98new\xe2\x80\x99 junctions assemble simultaneously below these germ cells. In this way, the integrity of the blood-testis barrier and the homeostasis of the seminiferous epithelium can remain intact during spermatogenesis. Previous studies have shown an array of cellular events, including protein internalization and cytoskeletal remodeling, to underline blood-testis barrier restructuring, whereas other studies have reported BTB dysfunction to associate with activation of the p38 mitogen-activated protein kinase pathway. Herein, we discuss the signaling pathways and mechanisms involved in blood-testis barrier restructuring in the mammalian testis.

signaling_pathways_regulating_the_blood-testis_barrier

1. Introduction<!>2. Functions<!>Transcriptional regulation<!>Endocytosis, recycling and degradation<!>Cytoskeletal remodeling<!>3. Cascades<!>Cytokines, testosterone and receptors<!>Integrins, laminins and focal adhesion proteins<!>5. Associated pathologies and concluding remarks

<p>Throughout spermatogenesis in the seminiferous epithelium of the mammalian testis, germ cells develop into spermatozoa through a series of cell divisions and morphological changes after which they are released from the epithelium (i.e., spermiation) (O'Donnell et al., 2011, de Kretser and Kerr, 1988, Kerr et al., 2006). During this time, developing germ cells are anchored to and supported by Sertoli cells, 'nurse-like' epithelial cells that give rise to the basic structure of the seminiferous epithelium. One crucial function of Sertoli cells is the formation of the blood-testis barrier (BTB), which separates the seminiferous epithelium into two distinct compartments, adluminal and basal. The BTB is constituted by different junction types, including tight junctions (TJs), basal ectoplasmic specializations (ESs; a testis-specific anchoring junction), desmosomes and gap junctions (GJs) (Mruk and Cheng, 2004, Cheng and Mruk, 2002). Unlike other mammalian blood-tissue barriers, the BTB is unique because junctions are not organized into discrete domains; instead, they are intermixed, co-existing and co-functioning. This unparalleled organization of junctions allows for strict coordination across the different structures and makes the BTB one of the tightest blood-tissue barriers. While the BTB is known to protect post-meiotic germ cells from the systemic circulation (i.e., immune system), it also restructures cyclically to allow preleptotene/leptotene spermatocytes entry into the adluminal compartment for further development. In rats, for instance, spermatocytes traverse the BTB during stages VII-XI of the seminiferous epithelial cycle, which is comprised of 14 stages in total (Russell et al., 1990, Hess and Renato de Franca, 2008, Clermont and Perey, 1957). Previous studies have demonstrated that 'old' junctions situated above migrating spermatocytes disassemble, while 'new' junctions assemble simultaneously below these germ cells (Figure 1). This creates a brief scenario where a migrating spermatocyte can be microscopically observed as trapped in between two barriers, thereby creating an intermediate compartment. This highly coordinated process is mediated by an elaborate signaling network that involves the participation of an array of molecules from the extracellular milieu. Among these, autocrine and paracrine factors (e.g., cytokines and testosterone) secreted locally by testicular cells under the regulation of the hypothalamus and the pituitary gland are the best studied regulators of BTB function (Xia et al., 2005, Li et al., 2009b). In other studies, we have shown that biologically-active laminin fragments generated during spermiation, which occurs at late stage VIII of the seminiferous epithelial cycle, also facilitate BTB restructuring via a local autocrine axis (Yan and Cheng, 2006, Yan et al., 2008b, Su et al., 2012) (Figure 1). Herein, we briefly discuss our current understanding of how extracellular factors activate common signaling pathways within Sertoli cells to ultimately bring about BTB restructuring, a prerequisite for germ cell movement.</p><!><p>Cytokines are secretory proteins capable of eliciting a signal transduction cascade upon binding to cell surface receptors, resulting in pleiotropic effects. Several cytokines are known to be produced by Sertoli and/or germ cells, including tumor necrosis factor (TNF)-α, members of the transforming growth factor (TGF) superfamily (e.g., TGF-β2 and -β3, activins and inhibins), interleukins (IL; e.g., IL-1α and IL-6) and interferons (Li et al., 2009b, Xia et al., 2005). On the other hand, testosterone, a steroid synthesized by Leydig cells residing in the interstitium, is critical for spermatogenesis (Walker, 2011, Verhoeven et al., 2010). Because testosterone is produced locally, its level is approximately 100-fold higher in the testis than in the systemic circulation (O'Donnell et al., 2006). In general, BTB function is disrupted by cytokines, but enhanced by testosterone. It is also worth noting that many of the reported effects of cytokines and testosterone on BTB function are derived from Sertoli cells cultured at high density on Matrigel™-coated substrata. This allows Sertoli cells to polarize and to assemble junctions that structurally and functionally mimic the BTB in vivo (Mruk and Cheng, 2004).</p><!><p>The binding of cytokines to their receptors results in the recruitment of adaptor proteins to the ligand/receptor complex, which in turn activates a cascade of signaling events that ultimately affect gene transcription and cell function. For instance, TGF-βs are best known to regulate transcription via SMAD-mediated pathways (Kang et al., 2009), whereas TNF-α and IL-1α are both capable of activating nuclear factor-κB and promoting the transcription of proinflammatory molecules (Rickert et al., 2011, O'Neill, 2008). Likewise, testosterone can elicit genomic effects via classical testosterone signaling upon binding to androgen receptors in Sertoli cells (Walker, 2010, Walker, 2011). However, these classical genomic pathways fail to account for some of the effects of cytokines and testosterone in Sertoli cells. For example, TNF-α is known to perturb Sertoli cell TJ integrity, resulting from a down-regulation in occludin mRNA and protein levels (Siu et al., 2003, Lydka et al., 2012) (Figure 1). Nevertheless, BTB function is controlled in part through transcriptional regulation, which is triggered by cytokines and testosterone to elicit changes in the steady-state levels of BTB constituent proteins (Lui and Cheng, 2007). In the following sections, we discuss recent studies that point to non-genomic effects which are equally important in the regulation of BTB dynamics during spermatogenesis.</p><!><p>Biochemical studies have shown the kinetics of internalization of integral membrane proteins (e.g., occludin and N-cadherin) in Sertoli cells to be accelerated by both testosterone and cytokines, including TNF-α, TGF-β2, -β3 and IL-1α (Xia et al., 2009, Yan et al., 2008a, Lie et al., 2011) (Figure 1). Under the influence of testosterone or cytokines, these proteins are internalized within early endosomes via a clathrin-mediated mechanism, resulting in junction restructuring (Xia et al., 2009). These findings are also supported by an increase in the association of occludin with caveolin-1 and Rab-11, regulators of endocytosis/transcytosis and recycling, respectively (Su et al., 2010), suggesting that a caveolin-mediated mechanism may be involved in BTB restructuring. Furthermore, testosterone facilitates recycling of integral membrane proteins back to the plasma membrane, thus supporting assembly of the 'new' BTB below migrating spermatocytes (Yan et al., 2008b) (Figure 1). In contrast, TGF-β2 and -β3 are known to perturb TJ integrity, which is the result of protein degradation as shown by an increase in occludin association with the late endosome marker Rab9 as well as by an induction in the level of the ubiquitin-conjugating enzyme E2 J1 (Su et al., 2010). Unlike TGF-β, IL-1α decelerates the kinetics of occludin degradation (Lie et al., 2011), possibly promoting subsequent recycling when induced by other factors such as testosterone. In light of the opposed effects of testosterone and cytokines on BTB function, the events of protein endocytosis, recycling and degradation require strict coordination to facilitate junction restructuring during germ cell movement.</p><!><p>At the BTB, TJs, basal ESs and GJs link to filamentous (F-) actin. The basal ES is most notably characterized by hexagonally-packed actin bundles sandwiched in between the plasma membrane and the endoplasmic reticulum (Mruk and Cheng, 2004, Vogl et al., 2008). While this unique arrangement of F-actin stabilizes ES-mediated adhesion, the cyclic restructuring of the BTB is thought to be facilitated by the remodeling of these actin bundles into a highly branched and less organized network (Figure 1). These changes in F-actin organization are mediated by key actin-regulatory proteins under the control of cytokines as shown by recent studies. For instance, IL-1α induces the mislocalization of epidermal growth factor receptor pathway substrate 8 (Eps8; an actin-capping and actin bundling protein) away from the Sertoli cell surface, thereby destabilizing actin bundles, cell junctions and BTB integrity (Lie et al., 2011, Lie et al., 2009). Concomitantly, IL-1α also increases the steady-state level of actin-related protein (Arp) 3 (Lie et al., 2011), a component of the Arp2/3 complex, which is an actin nucleation machinery capable of forming nascent branches on actin filaments (Goley and Welch, 2006). The effects of cytokines on F-actin dynamics are not restricted to Il-1α; the association between Arp3 and drebrin E, an actin-binding protein, is induced by TNF-α and TGF-β3, suggesting that an increase in Arp2/3 recruitment by drebrin E may facilitate actin branching (Li et al., 2011). Furthermore, TGF-β3 also increases the active GTP-bound form of Cdc42 (Wong et al., 2010), a Rho GTPase capable of activating Wiskott-Aldrich syndrome protein (WASP), which in turn is required for the activation of the Arp2/3 complex (Goley and Welch, 2006). Collectively, these cytokines elicit similar effects in favoring the formation of a branched actin network, which facilitates the internalization of integral membrane proteins and the restructuring of the BTB.</p><!><p>Despite our growing knowledge on the multitude of effects elicited by different autocrine and paracrine factors, it is still poorly understood how signals are propagated and how they are coordinated within Sertoli cells to bring about BTB restructuring. One of the best-studied intracellular signaling pathways to associate with BTB disassembly is the p38 mitogen-activated protein kinase (MAPK) cascade (Cheng et al., 2011, Wong and Cheng, 2005) (Figure 2). On the other hand, signaling pathways that associate with BTB assembly are much more difficult to study because a multitude of cellular events are known to take place during spermatogenesis. At the very least, classical and non-classical testosterone signaling pathways are involved in BTB assembly and integrity (Walker, 2010, Walker, 2011, Verhoeven et al., 2010).</p><p>p38 MAPKs are a class of serine/threonine protein kinases that are activated in response to a wide range of extracellular stimuli, especially those generated under stress and inflammatory conditions (Kumar et al., 2003). Under the three-tiered kinase cascade followed by MAPK pathways in general, p38 MAPKs are phosphorylated and activated by several MAPK kinases (MKKs; e.g., MKK3, 4 and 6), which in turn are activated by MAPK kinase kinases (MEKKs; e.g., MEKK1-4 and TGF-β-activated kinase 1) (Cuadrado and Nebreda, 2010) (Figure 2). Upstream of this cascade, signals from the cell surface are relayed by activators such as small G proteins, including Rac and Cdc42 (Keshet and Seger, 2010). In the testis, the p38 MAPK pathway is clearly implicated in BTB disassembly. The loss of barrier function induced by TGF-β3 and TNF-α, as well as by the environmental toxicant cadmium, associates with an increase in p38 phosphorylation (Lui et al., 2003a, Li et al., 2006, Lui et al., 2003b). Bisphenol A, another environmental toxicant, also triggers the phosphorylation of extracellular signal-regulated kinases (ERK) 1/2 in Sertoli cells (Li et al., 2009c, Li et al., 2009a). Equally important, TGF-β3-induced TJ disruption is blocked by a specific p38 inhibitor SB202190 (Lui et al., 2003a). Within this signaling pathway, MEKK2 likely functions as a p38 activator given its increase in mRNA level after TGF-β3 administration (Lui et al., 2003a). Further upstream, activation of the p38 MAPK pathway in Sertoli cells is Cdc42-dependent since p38 phosphorylation is induced by overexpression of Cdc42, but not overexpression of its dominant negative mutant T17N (Wong et al., 2010). Collectively, these results illustrate that BTB disassembly induced by TGF-β3 and TNF-α is mediated via the activation of the p38 MAPK pathway. While small GTPases that activate the p38 pathway are generally not considered to be part of this cascade, they are likely to promote BTB disruption through additional effectors. For instance, both Cdc42 and Rac are activators of nucleation promoting factors, namely WASP and WASP family verprolin-homologous protein (WAVE), respectively, which in turn promote Arp2/3-mediated actin nucleation during junction restructuring (Goley and Welch, 2006).</p><!><p>Cytokines such as TNF-α, TGF-βs and IL-1α are synthesized as precursor proteins of ~30 kDa in Sertoli and/or germ cells, which are subsequently cleaved into mature forms of ~17 kDa. Despite numerous studies illustrating their effects on BTB dynamics, these cytokines are largely dispensable for spermatogenesis because knockout models are either perinatally lethal or viable and fertile (Xia et al., 2005). On the other hand, testosterone is indispensable for male fertility; in the absence of testosterone or functional androgen receptors, spermatogenesis cannot go to completion as a result of (i) compromised BTB integrity, (ii) inability of germ cells to develop beyond meiosis, and (iii) spermiation failure (Verhoeven et al., 2010, Walker, 2010, Walker, 2011). Since germ cells do not express androgen receptors, Sertoli cells are the main transducers of testosterone signals within the seminiferous epithelium that are essential for both germ cell development and BTB integrity. This is illustrated by the up-regulation in Sertoli cell androgen receptor expression during stages VI-VIII of the seminiferous epithelial cycle, which coincides with BTB restructuring and spermiation. As discussed previously, testosterone signaling is known to be mediated via both classical and non-classical pathways (Walker, 2010, Walker, 2011). In the classical pathway, testosterone binds to androgen receptors present within the Sertoli cell cytoplasm. Upon ligand binding, heat shock proteins release androgen receptors so that they may enter the nucleus to initiate transcriptional regulation. More recent studies in Sertoli cells have described the existence of at least two non-classical pathways, which involve the influx of Ca2+ ions and the stimulation of epidermal growth factor receptor via Src activation (Walker, 2010). As described previously, testosterone-induced endocytosis and recycling of BTB proteins is likely mediated via a non-classical pathway as these effects are rapidly manifested (Yan et al., 2008b), but additional studies are needed to better understand the mechanisms at play.</p><!><p>Integrins are cell surface receptors comprised of transmembrane heterodimers that mediate cell-cell and cell-matrix adhesion by binding ligands such as laminins. Besides having roles in adhesion, integrins are also important signal transducers between cells and the external environment via outside-in, as well as inside-out signaling (Margadant et al., 2011). In the seminiferous epithelium of the mammalian testis, the integrin/laminin complex localizes to two adhesion sites, namely the (i) apical ES at the Sertoli cell-elongating/elongated spermatid interface and the (ii) hemidesmosome at the Sertoli cell-basement membrane interface (Yan and Cheng, 2006, Yan et al., 2008b) (Figure 1). Recent studies show that integrins and laminins mediate crosstalk among the apical ES, hemidesmosome and BTB, forming a local autocrine axis to facilitate barrier restructuring. Prior to spermiation at late stage VIII, laminin at the apical ES is proteolytically cleaved to generate biologically-active fragments, which are capable of disrupting BTB integrity by decreasing the steady-state level of occludin (Yan et al., 2008a, Su et al., 2012). Furthermore, these laminin fragments also down-regulate the level of β1-integrin at the hemidesmosome, which also indirectly affects BTB integrity (Yan et al., 2008a) (Figure 1). These data support the contention that BTB restructuring and spermiation, which occur at opposite ends of the seminiferous epithelium, are strictly coordinated. At this point, the precise biological actions of laminin fragments are not completely understood, and additional studies are needed.</p><!><p>Herein, we have summarized the current state of knowledge as it relates to the regulation of the blood-testis barrier in mammals. Since BTB integrity is critical for spermatogenesis and fertility, additional studies are needed to better understand how this unique ultrastructure is regulated. This is important because this information may help to identify new targets for non-hormonal contraception in the future. BTB dysfunction may also associate with unexplained cases of infertility, and additional studies along these lines are needed as well.</p>

PubMed Author Manuscript

Multiscale Macromolecular Simulation: Role of Evolving Ensembles

Multiscale analysis provides an algorithm for the efficient simulation of macromolecular assemblies. This algorithm involves the coevolution of a quasiequilibrium probability density of atomic configurations and the Langevin dynamics of spatial coarse-grained variables denoted order parameters (OPs) characterizing nanoscale system features. In practice, implementation of the probability density involves the generation of constant OP ensembles of atomic configurations. Such ensembles are used to construct thermal forces and diffusion factors that mediate the stochastic OP dynamics. Generation of all-atom ensembles at every Langevin timestep is computationally expensive. Here, multiscale computation for macromolecular systems is made more efficient by a method that self-consistently folds in ensembles of all-atom configurations constructed in an earlier step, history, of the Langevin evolution. This procedure accounts for the temporal evolution of these ensembles, accurately providing thermal forces and diffusions. It is shown that efficiency and accuracy of the OP-based simulations is increased via the integration of this historical information. Accuracy improves with the square root of the number of historical timesteps included in the calculation. As a result, CPU usage can be decreased by a factor of 3-8 without loss of accuracy. The algorithm is implemented into our existing force-field based multiscale simulation platform and demonstrated via the structural dynamics of viral capsomers.

multiscale_macromolecular_simulation:_role_of_evolving_ensembles

I. Introduction<!>II. Methodology<!>A. Order Parameters<!>B. Deductive Multiscale Approach<!>C. Role of Dynamical Ensembles<!>III. Results and Discussion<!>A. Relationship Between Ensemble Size and Langevin Timestepping: Implications for the history method from MD Results<!>B. Applicability of the Multiscale Approach and Correlation Between Ensembles<!>C. Ensemble Size and Langevin Timestepping: A History Enhanced Multiscale Simulation Analysis<!>D. Dependence of Simulation Parameters on System Characteristics: Demonstration via Stable and Unstable HPV Pentamer Constructs<!>IV. Conclusion

<p>A focus of interest in theoretical and computational nanosciences is to predict the behavior of macromolecular assemblies such as viruses using their N -atom description and Newton's equations of motion.1 Molecular dynamics (MD) has been widely used to achieve such simulations. However, the simulation time for nanometer scale assemblies has been limited to tens or sometimes few hundred nanoseconds.2, 3 While an advantage of N-atom approaches is that, given an inter-atomic force field, they offer the possibility of calibration-free modeling, they are limited by the system size, simulation timestep and hardware requirements. Recently, billion atom MD simulations have been accomplished.4–6 However, these simulations neglect one or more of Coulomb interactions, bonded forces, and rapidly fluctuating hydrogen atoms. All the latter are central to biomolecular structure and dynamics. Thus, all-atom simulation of large macromolecular assemblies remains a computational challenge.7, 8</p><p>Standard MD packages include CHARMM,9 GROMACS,10 and NAMD.11 Interest in large systems has stimulated the development of MD algorithms that take advantage of computational efficiencies enabled by parallel and graphical processor unit implementations.12, 13 A variety of coarse-grained approaches including bead, shape-based,14, 15 rigid region decomposition,16 and symmetry constrained17–19 models, as well as principal component analysis20, 21 and normal mode analysis guided approaches21 have been introduced to reduce the computational burden of large system simulations, but they do so at the expense of losing atomic scale resolution.</p><p>We have undertaken a deductive multiscale approach that folds the physics underlying the existence of slowly evolving variables into the computations for large systems.22–25 These variables, denoted space warping order parameters (OPs),26, 27 describe coherent, overall structural changes of the system. Furthermore, mathematical reformulation of the underlying molecular physics simultaneously captures high frequency atomic fluctuations and evolves the coarse-grained state. More precisely, we start with the N-atom Liouville equation and obtain Langevin equations for stochastic OP dynamics.28 Since specifying the coarse-grained variables leaves great uncertainty in the detailed all-atom state, quasi-equilibrium ensemble of all-atom configurations consistent with the instantaneous state of the OPs is generated. This ensemble is used with Monte Carlo (MC) integration to construct factors (forces and diffusions) in the Langevin equations needed to advance the OPs to the next timestep. Such an approach yields a rigorous way to transfer information between variables on different space-time scales, avoiding the need to make and calibrate phenomenological expressions for evolving the state of OPs. This scheme has been implemented as the Deductive Multiscale Simulator (DMS) software system (denoted SimNanoWorld in previous publications).22, 29 DMS is used to capture polyalanine folding from a linear to a globular state,27 Ostwald's ripening in nanocomposites,30 nucleation/front-propagation and disassembly pathways involving the structure and stability of virus capsids,31, 32 counter-ion induced transition in viral RNA and stability of RNA-protein complexes over a range of salinity and temperatures.26 Result from DMS simulations are comparable to those from conventional MD (notably NAMD), but the former is faster and more statistically significant as it is derived from evolving ensembles of all-atom configurations.29 The objective here is to further accelerate these calculations while maintaining accuracy and all-atom resolution.</p><p>The Langevin model of Brownian motion has been extensively used to describe the dynamics of particles in a heat bath under conditions near equilibrium.33 Several MC techniques have been used to numerically integrate ordinary and general Langevin-type equations.34 Their applicability depends on the magnitude of Langevin timesteps relative to that of velocity autocorrelation functions decay. Similarly, there are MC schemes based on the independent single-variate velocity and displacement distribution functions.35 Other extensively used numerical integrations schemes include ones proposed by Gunsteren and Berendsen,35 Brooks-Brunger-Karplus, and the Langevin impulse integrator.36 The accuracy of these schemes ranges from first to second order. These schemes present a general numerical procedure for integrating Langevin equations. However, they are not meant to explicitly address the coevolution of slow structural variables with an ensemble of rapidly fluctuating ones, as is the case here.</p><p>The purpose of this study is not to discover a new Langevin integrator. Rather, we introduce a procedure that effectively enhances the size of aforementioned quasi-equilibrium ensembles by using configurations from earlier times (i.e., as ones move back in "history") and simultaneously accounts for the dynamical OP and noise characteristics. In this way, statistical error for the numerical integration of factors in the Langevin equations is reduced. As a result, one can perform a multiscale simulation which still provides all-atom detail, but now with less stringent limitations on the Langevin timestep for OP evolution. With this, enhanced numerical efficiency of multiscale simulations is achieved without compromising with accuracy. Computational cost of Langevin OP simulations is mediated by the characteristic OP time, size of the all-atom ensemble for MC integration, and number of historical ensembles considered in extended MC sampling. Contributions from these three factors are evaluated to obtain a range of parameters that imply optimal simulation efficiency.</p><p>In the following, we review our OP based multiscale methodology and extend discussions on the use of dynamical ensembles to enhance the sample size for MC computation of thermal average forces (Sect. II). This algorithm is numerically demonstrated for all-atom simulations of Human Papillomavirus16 (HPV16) capsomers (Sect. III). Interplay between several numerical parameters is studied to identify those providing simulation accuracy as well as efficiency. Simulations with such parameters are used to investigate contrasting long-time behavior of different capsomer constructs. Conclusions are drawn in Sect. IV.</p><!><p>In this section various components of our deductive multiscale approach are discussed. A central element of our multiscale analysis is the construction of OPs for describing the coarse-grained features of a macromolecular assembly. An OP mediated model captures the separation in timescales between the coherent (slow) and non-coherent (fast) degrees of freedom. In effect, OPs filter out the high frequency atomistic fluctuations from the low frequency coherent modes. This property of OPs enables them to serve as the basis of a multiscale approach for simulating the dynamics of macromolecular systems. Here, our methodology is outlined and discussion is extended on a dynamical ensemble enhancement scheme to accurately compute factors in the Langevin equation for OP dynamics.</p><!><p>Consider a macromolecular assembly described via the positions of its N constituent atoms labeled i = 1,⋯ N. Let the i-th atom in the system be moved from its original position r⇀i0 via (1)r⇀i=∑kΦ⇀kUki+σ⇀i, where the Φ⇀kandUki≡Uk(r⇀i0) are k-th OP and basis function respectively. For example, the use of k have been taken to be products of Legendre polynomials in the X, Y, Z Cartesian directions, i.e., Uk(r⇀i0)=Uk1(Xi0)Uk2(Yi0)Uk3(Zi0).23 r⇀i0 is the reference position of atom i which, through the OPs and the Eq. (1) is deformed into the instantaneous position r⃑i. Since we seek a dimensionality reduction, the number of Φ⃑ḵ is much less than the number N of atoms. Given a finite truncation of the k sum in Eq. (1), there will be some residual displacement (denoted σ⃑i) for each atom in addition to the coherent deformation generated by the k sum.</p><p>An explicit expression for the Φ⃑k is obtained by minimizing the mass-weighted square residual with respect to the Φ⃑k.23 One obtains (2)Φ⇀k=∑i=1NmiUkir⇀iμk;μk=∑i=1NmiUki2, where mi is the mass of atom i. Inclusion of mi in developing Eq. (2) gives Φ⃑k the character of a generalized center-of-mass. For example, if Uki is independent of i then Φ⃑k is proportional to the center-of-mass of the assembly. A subset of OPs defined in this way constitutes a strain tensor accounting for compression-extension-rotation, while others describe more complex deformations such as tapering, twisting, bending and their various combinations.26, 27 The µk serve as effective masses associated with each OP, implying the spatial scale they capture. The masses primarily decrease with increasing complexity of Uki.26 Thus, OPs with higher k probe smaller regions in space. In summary, a model based on this set of OPs simultaneously probes structure over a diverse range of spatial scales via different orders in k.</p><!><p>Eq. (2) implies that for a given set of atomic positions the corresponding OPs are uniquely defined. However, the converse is not true, i.e., there exist multiple all-atom configurations consistent with a given set of Φ⃑k. Thus, an OP-based theory of macromolecular assemblies is statistical in character since specifying the coarse-grained variables leaves great uncertainty in the detailed all-atom state. To address this issue, the theory should provide an algorithm for evolving the coarse-grained variables and another for coevolving the probability of the detailed all-atom states. This conceptual framework has been shown to yield stochastic equations for the propagation of the slow OPs and those for constructing the coevolving ensemble of all-atom configurations.</p><p>The description adapted starts with the probability density ρ of the N atomic positions and momenta Γ. However, this formulation masks the underlying hierarchical organization of a macromolecular assembly. To address this, here, ρ is hypothesized to depend on Γ both directly, and via a set of OPs, indirectly. This "unfolding" of the N-atom probability density makes the multiple dependencies of ρ on Γ and time t explicit. With this ansatz, a perturbation analysis of the Liouville equation yields sets of coupled Langevin equations for the OPs (3)dΦ⇀kdt=∑k'D⇀⇀kk'f⇀k'+ξ⇀k,28 where the diffusivity factors D⇀⇀kk are related to the correlation function of OP momenta Π⃑k via (4)D⇀⇀kk'=1μkμk'∫−∞0dt〈Π⇀k'(t)Π⇀k〉.28 Π⃑k is the value of the OP momentum for a given N-atom configuration, Π⃑k'(t) is advanced in time through Newtonian mechanics, and the 〈⋯〉 implies thermal average over configurations. Variance of noise ξ⃑k is bound by D⇀⇀kk. The thermal average force f⃑k is given by (5)fkα=−∂F∂Φkα;α=X,Y,Z22 for OP-constrained Helmholtz free-energy F, where (6)F=−1βln⁡Q(Φ,β), Q(Φ, β) = ∫ dΓ* Δ(Φ − Φ*)e−βH* is the partition function constructed from configurations consistent with the set of Φ⃑k (denoted Φ collectively). Eq. (3) implies overall structural dynamics through evolution of the OPs. It has been implemented as the DMS nanosystem simulator for the case of a single system29 and more recently for a set of interacting subsystems.37</p><p>A commonly used approach for treating far-from-equilibrium systems involves projection operators.35, 38–40 It is very general in the sense that no approximations are made in arriving at an equation for the reduced probability of a subset of variables (OPs in our case). However, this kinetic equation requires construction of a memory function, which usually can only be constructed using extensive MD simulations or experimental data. This is numerically expensive for N-atom problems except when the memory functions have short relaxation times.38 In our analysis, the OPs of interest are much slower than the characteristic rate of atomistic fluctuations, and therefore the relaxation times are typically short relative to characteristic times of OP dynamics.28 Under these conditions, our multiscale approach leads to the same set of Langevin equations as those from projection operators. However, the multiscale approach is more direct; we do not start with the projection operators and eventually resort to perturbation methods for constructing memory functions. Rather we make an ansatz that the N-atom probability density has multiple (initially unspecified) space-time dependencies, and analyze the resulting Liouville equation.25</p><p>While several coarse-grained modeling approaches account for large-scale processes, important all-atom features of an assembly can be lost.41 However, processes like the interaction of an antibody with a viral capsid can depend sensitively on atomic structure.39, 40 To capture such details, in DMS, an ensemble of all-atom configurations consistent with the42, 43 instantaneous OPs description is constructed. To accomplish this, residuals σ⃑i are constructed by changing those Φ⃑k that do not contribute to the k-sum (Eq. (1)). By definition, OPs with higher k probe smaller regions in space. Consequently, they account for small-scale incoherent displacement of each atom in addition to coherent deformations generated by the other, lower k, OPs. Short MD (NAMD) runs are performed starting with configurations from this residual-generated ensemble to arrive at an enriched ensemble that is consistent with a given set of OPs (Φ). This procedure for generating ensembles is called hybrid sampling. Further details are provided elsewhere.22</p><p>Given an all-atom structure at time t=0, a set of space warping OPs is constructed via Eq. (2). Then, an ensemble of all-atom configurations consistent with this set of OPs is generated via the aforementioned hybrid sampling scheme. This ensemble is then employed to compute factors (thermal average forces f⃑k and diffusion coefficients D⇀⇀kk') that mediate the Langevin Φ⃑k dynamics. The f⃑k are expressed in terms of atomic forces F⃑i via (7)f⇀k=〈f⇀km〉;f⇀km=∑i=1NUkiF⇀i.28</p><p>The atomic forces F⃑i computed for each member of an OP-constrained ensemble of atomic configurations are used to calculate the macroscopic force (or OP forces) f⇀km. MC integration averaging of f⇀km over the ensemble is carried out to obtain the thermal average force f⃑k. Short MD runs (~1 ps) are performed on configurations from this ensemble to calculate the OP velocity correlation functions needed to construct the D⇀⇀kk' (Eq. (4)). Using these f⃑k and D⇀⇀kk', the OPs are evolved in time via the Langevin equation. The evolved OPs are used to generate a new ensemble of atomic configurations and the cycle repeats. Thus, OPs constrain the ensemble of atomic states (Eqs. (1)–(2)), while the latter determine the diffusion factors (Eq. (4)) and thermal average forces (Eq. (7)) that control OP evolution (Eq. (3)). With this, the two way transfer of structural information that couples microscopic motions to large-scale structural dynamics is captured. Also, accounting for the dynamically changing ensemble of atomistic configurations consistent with the evolving set of OPs provides statistical significance to DMS predictions.</p><!><p>A factor limiting the efficiency of the above algorithm is the need to generate a sufficiently rich all-atom ensemble at every OP timestep. If the ensemble is too small, statistical errors in the thermal forces and diffusion factors can misdirect the evolution and therefore limit the size of the Langevin timestep Δt needed to advance the system from a given time t to t + Δt. Thus, to increase the Δt, larger ensembles are required. This would then increase the computational expenditure of the multiscale simulation. Here, we address this issue by constructing ensembles from timesteps in the history prior to t to effectively enhance the ensemble needed to proceed to t + Δt. This is accomplished in a manner that accounts for the coevolution of the all-atom ensemble with the OPs as follows.</p><p>Let t be the present time and th be a time NhΔt in the past (th = t − NhΔt). Consider the integration of Eq. (3) from a time th to t + Δt. The result is (8)Φ⇀k(t+Δt)=Φ⇀k(th)+∫tht+Δtdt'(∑k'D⇀⇀kk'f⇀k'+ξ⇀k).</p><p>Investigation of the D⇀⇀kk' (Eq. (4)) shows that the diagonal factors dominate when the Uki are orthogonalized.28 With this Eq. (8) becomes, (9)Φ⇀k(t+Δt)=Φ⇀k(th)+∫tht+Δtdt'(D⇀⇀kkf⇀k+ξ⇀k).</p><p>In a simple lowest order method Nh is taken to be 0. Here, we take Nh > 0 to fold historical ensemble information into the timestepping algorithm as follows. Breaking the integration interval into 1 Nh + 1 segments of length Δt yields (10)Φ⇀k(t+Δt)=Φ⇀k(th)+∑j=0Nh∫th+jΔtth+(j+1)Δtdt'(D⇀⇀kkf⇀k+ξ⇀k).</p><p>Since the thermal average forces depend on the OPs, which to a good approximation change slowly, the first term inside the integral can be approximated via a simple rectangle rule. With this the thermal average force contribution becomes (11)∑j=0Nh∫th+jΔtth+(j+1)Δtdt'(D⇀⇀kkf⇀k)=Δt∑j=0Nh(D⇀⇀kkf⇀k)t=th+jΔt, Such discretization applies provided Δt<∑j=0Nh(fkα)t=th+jΔt/∑j=0Nh(fkα')t=th+jΔt;α=X,Y,Z.44 Since the force ξ fluctuates rapidly around zero, integration of the stochastic term is taken to follow the Ito formula.45 When ξ is white noise, one obtains (12)∑j=0Nh∫th+jΔtth+(j+1)Δtdt'(ξ⇀k)=Δt1/2∑j=0Nh(ξ¯k)t=th+jΔt, where 〈ξ⇀k〉=0and12∫−∞0dt〈ξ⇀k(t)ξ⇀k(0)〉=D⇀⇀kk. Thus, the fact that the ensemble is changing over history as manifest in the diffusion and forces is accounted for. With this, the discretrization algorithm becomes (13)Φ⇀k(t+Δt)=Φ⇀k(th)+Δt∑j=0Nh(D⇀⇀kkf⇀k)t=th+jΔt+Δt1/2∑j=0Nh(ξ⇀k)t=th+jΔt.</p><p>This is the basis of our history enhanced multiscale algorithm that is implemented in DMS via the workflow of Fig. 1. To arrive at Eq. (13) it has been assumed that the Langevin timestep is constant. This framework can be easily generalized to address adaptive timestepping. Furthermore, we demonstrate the method using a simple Langevin integrator. In a follow-on work, this workflow will be implemented to higher-order numerical integration schemes.</p><p>An error analysis of the above approach is now considered. To simplify this analysis, and as observed for the demonstration system here, the diffusion factors are approximately constant in the interval th to t. With this, Eq. (13) becomes (14)Φ⇀k(t+Δt)=Φ⇀k(th)+ΔtD⇀⇀kk∑j=0Nh(f⇀k)t=th+jΔt+Δt1/2∑j=0Nh(ξ⇀k)t=th+jΔt.</p><p>Using the definition of Δt, the second term in Eq. (14) takes the form (t−th)D⇀⇀kkf⇀eff, where (15)f⇀eff=1Nh∑j=0Nh(f⇀k)t=th+jΔt.</p><p>Dividing f⃑k into f⇀k∞ and g⃑k, where f⇀k∞ is calculated from a very large ensemble the sum in Eq. (15) becomes (16)f⇀eff=1Nh∑j=0Nh(f⇀k∞+g⇀k)t=th+jΔt.</p><p>The sum of the f⇀k∞ is of O(Nh) as the f⇀k∞ change coherently on the timescale of OP evolution. In contrast, the g⃑k contribution is a sum of random factors of fluctuating sign and zero mean since g⃑k represents the MC error associated a finite ensemble. Therefore, the latter is of the order O(Nh1/2). The magnitude of the g – terms divided by that of the f∞ – terms is thus of O(Nh−1/2). Next, let Ne be the number of all-atom configurations in the ensemble used to calculate thermal average forces at a given timestep. The MC integration error is O(Ne−1/2).34 Thus, assuming the ensembles at each timestep are statistically independent (shown below), the error in the history enhanced ensemble is O((NeNh)−1/2). Since ensemble errors in the thermal average forces misdirect the OP evolution, the O(Ne−1/2) error implies a limit on the Langevin timestep which is improved by a factor of Nh1/2 when history enhancement is used (i.e., upper bound on the value of Δt in Eq. (13) increases by Nh1/2 with increase in the number of history terms (Nh) in the associated fk-summation). Overall accuracy of the history enhancement method also reflects the limit on (a) Δt due to the characteristic time of OP evolution and (b) Nh due to the need for periodic refreshment of reference structure, r⇀i0, (denoted re-referencing) for "on-the-fly" OP definition during a DMS run. The interplay of these factors is investigated in the next section in the context of obtaining optimal simulation parameters for numerical efficiency.</p><!><p>Here, DMS implementation of the history enhanced Langevin algorithm (Sect. II) is demonstrated via all-atom simulations of HPV16 capsomers in 0.3M NaCl solution. The T=1 L1 HPV16 Virus-Like Particle (VLP) contains 12 pentamers joined by "attacking arms" that stabilize it via strong hydrophobic interactions.46 Each pentamer is composed of five L1 protein monomers. A C-terminal of the L1 protein consists of four helical regions h2, h3, h4 and h5 that maintain intra- and inter-pentameric connectivity. While h2, h3 and h5 are responsible for L1 protein folding and pentameric stability, h4 maintains inter-pentamer organization and, thereby, overall T=1 structure.46 It has been experimentally shown that h4-truncated L1 proteins successfully form stable pentamers but fail to organize into a T=1 VLP, while h2,h3,h4 truncation prevents stable pentamer formation.46 We simulate the expansion and consequent stability of a h4-truncated pentamer when it is isolated from the rest of the VLP. Simulations presented include 24 5ns DMS runs with different ensemble sizes (Ne) and number of history timesteps (Nh); a 5ns MD run for benchmarking results of these DMS runs; and 30–40ns DMS runs of complete, h4-truncated and h2,h3,h4-truncated pentamer showing contrasting long-time behaviors of respective structures. These systems contain 3–4×104 atoms. Further details of conditions used for these simulations are provided in Table I.</p><p>In the following, first, data from NAMD simulations are used to introduce the history enhanced MC scheme for computing thermal average forces. Effect of correlations between dynamical ensembles on the MC error analysis is discussed. Then, the history enhanced approach is used with DMS to understand the effects of Langevin timestep (Δt), ensemble size (Ne) and number of historical ensembles (Nh) on simulation accuracy and performance. An optimal set of simulation parameters (Δt, Ne and Nh) is obtained and used to simulate long-time behaviors of three pentamer constructs.</p><!><p>The magnitude of Langevin timestepping Δt depends explicitly on the spatio-temporal scale of motion the associated OPs capture. For example, OPs capturing the overall motion of a macromolecular assembly change much more slowly than ones describing changes in individual macromolecule. Consequently, Δt for simulating a macromolecule is much less than that for the entire assembly.</p><p>Consider the case of an isolated HPV pentamer. Its degrees of freedom are constrained inside a T=1 VLP. When all other pentamers in the capsid are removed instantaneously, this pentamer initially expands and then stabilizes to a new state. A 5ns MD simulation is performed that captures an early phase of this expansion. Here, this trajectory is used to investigate the rate of OP dynamics and compute an optimal timestep (Δt) for their Langevin evolution. Under friction dominated conditions (Eq. (3)), the rate of change of OPs is directly correlated to the thermal average forces.22 These forces, in turn, are statistical in nature and require sufficient averaging to accurately predict OP dynamics. For example, if the ensemble is too small, statistical errors in the thermal forces and diffusion factors can misdirect the evolution and therefore limit the size of the Langevin timestep Δt. Therefore, the optimal value of Δt for accurately capturing OP evolution should reflect the dynamical nature of thermal average forces, which in turn depend on size of the ensemble used. To understand this effect, structures are chosen every 20ps form the 5ns MD trajectory, constant OP ensembles are generated via hybrid sampling (Sect. II-B), and thermal average forces are computed using ensembles of different sizes (Ne) keeping Nh=0 (Fig. 2(a)).</p><p>At a given point in time, consider generation of all-atom ensembles with Ne varying over a range of values from 100 to 3200. While atomic forces in these ensembles show no clear trend (Fig. S1 in Supporting Information) (even though the underlying structures are dramatically different), OP forces (f⇀km, Eq. (7)) constructed from the same ensembles are peaked about a given value (Fig. 2(a)–(b)). Such peaks suggest strong thermal average forces; this induces coherence in large-scale dynamics.22 In the present context, positively peaked OP forces results in positive thermal average forces. Langevin evolution using these forces increases magnitude of the corresponding OP, thereby implying overall expansion of the pentamer. This suggests that the thermal average forces are an effective measure of coherence in subtle trends of the inter-atomic forces manifested in our OP-constrained ensembles. Their construction enables self-consistent transfer of structure and dynamics information from the atomic to the larger-scales.</p><p>Well resolved peaks in the distribution of OP forces suggest that many of the atomistic configurations in the ensemble contribute to similar OP forces. Such configurations are restricted to those consistent with instantaneous OP values. Together, these imply a modest size atomistic ensemble is sufficient for computing thermal average forces (f⇀k=〈f⇀km〉). However, generating these modest ensembles is still computationally expensive. Thus, practical sampling limits impose statistical errors in thermal average forces. As ensemble size decreases, the g-term in Eq. (16) increases. These g contributions randomly shift the numerically computed average forces around their correct values (Fig. 2(a)). To preserve accuracy of the predicted OP dynamics using such incomplete forces, the Langevin timestep should be reduced. This is because an erroneous thermal average when applied throughout a large timestep can take the system far away from its correct evolution pathway. But, over a sequence of small timesteps, these random errors tend to cancel (See below, Sect. III-C). However, this implies loss of multiscale simulation efficiency.</p><p>Here, we introduce a procedure that uses Nh ensembles from timesteps in the history prior to a given time t to effectively enhance the ensemble needed to proceed to t + Δt. Consequently, a slowly evolving ensemble is accounted for as a collection of small, less compete ones each of which captures some of the instantaneous influence of the evolving OP. With this history algorithm, statistical error in the MC integration of thermal average forces (resulting from lack of complete ensembles) is reduced and therefore numerical restrictions of Δt decreases limiting it to those values inherent in the accurate OP dynamics.</p><p>A rough estimate of maximum allowed Δt value can be obtained using the forces in Fig. 2(a) and their ratio with their derivatives, i.e., Δt<(fkαfkα') when Nh=0. This is shown as a function of ensemble size in Fig. 2(b). As ensemble sizes increases from 100 to 800 the required time step increases as O(Ne1/2), as expected from the statistical arguments of Sect. II. Larger ensemble size removes numerical noise making the force more coherent. Consequently, the numerical timestep increases and reaches a limit implied by the characteristic timescale of OP evolution. For the present example, approximately a timestep Δt of ~60ps is achievable using a sample size Ne of 800 or more. However, generating such ensembles is computationally expensive for the macromolecular assemblies of interest. To address this, a composite ensemble of size 3200 is obtained by sampling 400 structures over 8 timesteps (i.e., using Nh=8 and Ne=400) in the history of a given time t. The population distribution of the history enhanced OP forces is in agreement with those from a large ensemble at a given time (Fig. 2(c)). Furthermore, the history enhanced thermal average forces are in agreement with those computed from Ne=3200, Nh=0 ensembles for the entire 5ns trajectory (Fig. 2(c)). Using the history enhanced thermal average forces implies that the characteristic Langevin timestep is increased from 20 to 60ps (Fig. 2(b) (green point)), i.e., Δt(fkαfkα')→20ps when Ne=400 and Nh=0, but when Nh=8 keeping Ne fixed ∑j=0Nh(fkα)t=th+jΔt/∑j=0Nh(fkα')t=th+jΔt→60ps. Thus, the limiting OP timestep of 60ps obtained using Ne=3200 is recovered via Ne of 400 enhanced over 8 history steps. Other combinations of Nh and Ne that reproduce similar results are shown in the Supporting Information (Fig. S2). This analysis is valid only if all-atom ensembles used in the history enhancement are mutually uncorrelated, as shown in the following.</p><!><p>The OP velocity autocorrelation function provides a criterion for the applicability of the present multiscale approach. If the reduced description is complete, i.e., the set of OPs considered do not couple strongly with other slow variables, then the correlation functions decay on a time scale much shorter than the characteristic time(s) of OP evolution. However, if some slow modes are not included in the set of OPs, then these correlation functions can decay on timescales comparable to those of OP dynamics. This is because the missing slow modes, now expressed through the all-atom dynamics, couple with the adopted set of OPs. The present approach fails under such conditions. For example, putting the lower limit of integral in Eq. (4) to −∞ is not a good approximation and the decay may not be exponential; rather it may be extremely slow so that the diffusion factor diverges. Consequently, atomistic ensembles required to capture such long-time tail behavior in correlation functions are much larger than those for capturing a rapid decay. In Fig. S3 it is seen that the velocity autocorrelation function decays on a scale of >10ps when an OP Φk with k1=1, k2=0 and k3=0 (Eqs. (1)–(2)), implying extension-compression along the X-axis is missing, but on a scale of ~1ps when it is included. Here, such situations are avoided via an automated procedure of understanding the completeness of the reduced description and adding the OPs when needed (discussed briefly in Sect. SI1 of Supporting information).26 Adapting this strategy ensures that the OP velocity autocorrelation functions decay on timescales orders of magnitude shorter than those characterizing coherent OP dynamics, and thus the present multiscale approach applies. Consequently, the history enhanced multiscale method allows one to use larger timesteps (i.e., 10ps or more versus <1ps, to account for the long-time tails, for the demonstration problem).</p><p>The present OP velocity autocorrelation functions decay on the ps timescale. Thus, multiple 1ps NAMD trajectories are used to compute OP velocities, correlations within which are ensemble averaged to construct the diffusion factors (Eq. (4)). Adapting this procedure is computationally practical as the timescale of decay is orders of magnitude shorter than that characterizing coherent OP dynamics. This procedure could be further optimized in two ways to make autocorrelation functions decay even faster. (a) OP force autocorrelation functions can be used to construct friction coefficients between OPs.35 This matrix of friction factors can then be inverted to obtain the diffusion matrix D⇀⇀kk'. Since the OP force autocorrelation functions decay faster than those for the OP velocities (Fig. S4), shorter MD runs are required to obtain the statistics for computing these functions. Thus, one saves computational time. (b) Since the OPs are constructed using orthonormalized polynomial basis functions, they mix overall rotational with extension-compression modes. This Coriolis-type coupling can be minimized to facilitate greater separation in scales between the slow and fast degrees of freedom, thereby possibly allowing more rapid decay of the autocorrelation functions. One way of achieving this separation is to cast the OPs in terms of Eckart internal, rather than Cartesian, coordinates.47 Within this framework, there is no coupling between vibrational and rotational degrees of freedom at equilibrium. Related techniques involving a translating and rotating internal coordinate system are found to resolve molecular vibrations using only normal modes; translation and rotation are treated as vibrational motions with zero frequency.47–49 In a similar way, use of the Eckart frame could result in correlation functions with shorter decay times, and therefore greater computational efficiency. In a related ongoing work, the idea of constructing OPs to capture deformations with respect to an evolving reference structure (not a fixed one r⇀i0 in Eq. (2)) is exploited. This dynamical reference configuration makes the associated OPs slower, thereby increasing the timescale separation with atomic fluctuations and reducing the decay time.</p><p>In any practical computation, the ensembles created are incomplete. Correlation of information between these ensembles must be considered in evaluating the history method. If the ensembles of all-atom configurations constructed at consecutive Langevin timesteps are very similar then using both adds no additional information to the net history enhanced (two-timestep) ensemble. Thus, for ensemble enhancement to be beneficial, the incomplete all-atom ensembles from consecutive timesteps should be uncorrelated. This is seen to be the case for the present example (Fig. 3) where ensembles characterizing OPs at discrete intervals of time are independent. Since ensembles involved in the MC integration are uncorrelated, the convergence of the thermal average forces is expected to be ∝(NeNh)1/2. In this way, a large slowly varying ensemble can be accounted for via multiple atomistically uncorrelated smaller ones, each of which captures some of the instantaneous influence of the evolving OP. Here, the correlation coefficient between all-atom ensembles at two different Langevin timesteps is defined in terms of the covariance of the atomic forces obtained at these time points divided by the product of their standard deviations.44</p><p>When sufficiently large Boltzmann ensembles are constructed at every Langevin timestep, thermal average forces between consecutive Langevin timesteps should be correlated (Fig. S5), i.e., these OP forces depend on OPs which change only slightly from one Langevin step to another. However, finite sampling size introduces random ensemble error "noise" and reduces the correlation between the OP forces (Fig. S6). This also underlies the shifting distribution of OP force peaks as observed in Fig. 2(a). To address this ensemble noise we integrate historical information as follows.</p><!><p>The history algorithm is implemented in DMS as per the workflow of Fig. 1. The first few loops in this workflow is executed considering Nh = 0. During these steps, the use of small but computationally practicable ensembles limit Δt to small values. As the number of loops exceeds Nh, the Langevin equation is integrated using the history method (Eq. (13)). Thus, the effective ensemble size and hence Δt is increased without loss of simulation accuracy.</p><p>Multiscale simulation results using different timesteps and sample sizes (Changing Nh for a given Ne) are compared to those from the 5ns MD trajectory of the L1-pentamer expansion as discussed above. Limits on these parameters as obtained from the MD analysis (Sect. III-A) are Δt ~60ps and Ne > 800 when Nh=0 (Fig. 2(b)). DMS trajectories generated using ensemble of size 200 (Ne=200, Nh =0) to 1600 (Ne =200, Nh =8) are successful in reproducing NAMD results when Δt is 20ps (Fig. 4). This implies, statistical errors in incomplete thermal average forces when applied over a sequence of small timesteps cancels out, thereby reproducing results similar to those using more complete forces and larger timesteps. The result is also consistent with the fact that using ensembles of 200 or more configurations suffice to attain forces that imply Δt of 20ps (Fig. 2(b)). As the timestep increases, only runs with larger history ensemble sizes reproduce the MD derived OP trajectory. With smaller ensembles, statistical errors in the thermal forces and diffusion factors misdirect the OP evolution when the Langevin timestep is large. This is reflected in artificial overshoot and undershoot observed in the OP evolution (Fig. 4). Structurally, such behavior of OPs creates an abrupt increase in the amplitude as well as frequency of large scale motions, thereby evolving the system far away from its free-energy minimizing pathway. The timestep of 60ps is achieved using an ensemble of 1600 configurations (Ne=200 and Nh=8). This step-size is the maximum that can be used respecting the limit Δt<∑j=0Nh(fkα)t=th+jΔt/∑j=0Nh(fkα')t=th+jΔt (Fig. 2(b)), and thereby reflects the characteristic timescale of OP evolution as imposed by the thermal regime of motion. Overshoots and undershoots in OP dynamics also appear as the Langevin timestep exceeds the characteristic time of OP evolution. However, in such cases, further increase in ensemble size (even an infinite ensemble) does not suffice to restore the OP dynamics since error from the numerical integration algorithm with Δt<∑j=0∞(fkα)t=th+jΔt/∑j=0∞(fkα')t=th+jΔt is unacceptably large for the rectangular scheme used here (Eq. (13)).</p><p>There is a limit on the number of history timesteps, Nh, to be included in the Langevin integration that arises from our OP construction procedure (Eqs. (1)–(2)). These OPs are defined to describe the large scale dynamics as deformations of a fixed reference structure r⇀i0. However, the reference structure occasionally must be changed to accurately capture a structural transition. This is implied by the evolution of an appropriate reference structure on a timescale much greater than that characteristic of the OPs (Fig. S7(a)). For example, while the OPs change every Langevin timestep, typically the reference structure, and therefore the polynomials U change every 30 timesteps or more (Figs. S7(b)). Every time the reference structure changes, OPs are redefined in terms of the new reference. The history integration should not include time points beyond this re-referencing limit, as the very definition of OPs, OP forces and velocities are different across such reference structure transitions. Thus, for the case studied here Nh<= 30.</p><p>In summary, to capture the expansion of a h4-truncated pentamer the ensemble size Ne without the history enhancement must be greater than 800. It has been shown, using a maximum of 8 history steps and minimum of 200 structures per Langevin timestep, result of the equivalent non-history ensemble can be recovered. Consequently, in comparison with a non-history (Nh = 0) calculation with small Ne (200) and short Δt (e.g., 20ps), the history simulation with same Ne enables larger Δt (e.g., 60ps), and is therefore 3 times faster. Alternatively, if Ne is increased to 800 or 1600 for improving Δt keeping Nh = 0, then again the history formulation provides a speed-up of 4–8 over the non-history calculation as the considerable computational time for constructing sufficiently large ensembles is reduced. DMS without history enhancement is an order of magnitude faster than conventional MD.29 With history the speed is enhanced by 3–8 folds. However, DMS is more appropriately comparable with ensemble MD as an ensemble of size, e.g., 1600 is constructed every Langevin timestep. Furthermore, this efficiency is size dependent and has been shown to increase as system size increases.28, 29</p><!><p>The history enhanced Langevin scheme is used to simulate three HPV pentamer systems (1) complete, (2) h4-truncated and (3) h2, h3, h4-truncated pentamers. While the first two structures initially expand and then remain stable, the third has weak intra-pentameric connections and thereby expands more extensively (Figs. 5(a)–(c) and S8). As a result for the third system, the stress on selected monomers is reduced by ~50% (Fig. 5(d)). Variations in large scale motion are reflected in the timescale of OP evolution characterizing the three systems. For example, the complete and h4-truncated pentamers remain bound (both in simulation and in experiments), and gradually approach an expanded equilibrium state. This transition is simulated using a Langevin timestep of 60ps. In contrast, the h2, h3, h4-truncated pentamer is unstable. It demonstrates significant large scale motions, however does not immediately split into monomers as there are secondary hydrophobic interactions that support a transient long-lived state. The diffusion coefficients for system 3 are greater than those for 1 and 2 (Fig. 5(e)). This reflects the higher level of fluctuations is system 3. Consequently, the Langevin timestep required for numerical stability is reduced to 40ps. Furthermore, a larger number of OPs are required, reflecting the importance of shorter scale fluctuations. The increased number of OPs decreases the number of residual degrees of freedom and therefore the required ensemble size decreases also. Nonetheless, the computational efficiency is still improved via history ensemble enhancement, but the net advantage is somewhat diminished.</p><!><p>The simulation of many-atom systems like supramolecular assemblies can be greatly accelerated using multiscale techniques. However, these techniques require the construction of ensembles of all-atom configurations in order to compute diffusions and thermal average forces for advancing the coarse-grained variables (OPs in our case). As the approach yields the co-evolution of the OPs with the quazi-equillibrium distribution of all-atom configurations, it is effectively an ensemble MD method and thereby achieves statistically significant results. However, constructing such ensembles increases the computational burden, resulting in loss of efficiency. This difficulty stems from the use of ensembles which do not represent the space of all-atom configurations adequately. As a result, the coarse-grained state of the system will be somewhat misdirected in a given Langevin timestep, limiting the timestep size required to maintain the accuracy. It was shown here that this difficulty can be overcome using ensemble from previous timesteps. This requires that the latter are integrated into the computation in a manner which respects the changing nature of the ensembles over the past time period. Thus, ensemble from the evolution history cannot simply be added into a larger ensemble attributed to only one time.</p><p>In the history method presented here, it was shown that an acceleration of multiscale simulation of a factor of 3–8 over simulations which ignore history can be attained. The size of the allowed timestep increases with the number of timepoints included in the integration over history. The relation between Langevin timestep, ensemble size and re-referencing needed to sustain numerical accuracy and efficiency was established. The above points were demonstrated using three viral capsomers with different C-terminal truncations (thereby structural stability). While efficiency of the computations has some limits that depend on system detail, we expect the history enhanced approach appears to have broad applicability.</p>

PubMed Author Manuscript

Enzymatic Amplification of DNA/RNA Hybrid Molecular Beacon Signaling in Nucleic Acid Detection

A rapid assay operable under isothermal or non-isothermal conditions is described wherein the sensitivity of a typical molecular beacon (MB) system is improved by utilizing thermostable RNase H to enzymatically cleave an MB comprised of a DNA stem and RNA loop (R/D-MB). Upon hybridization of the R/D-MB to target DNA, there was a modest increase in fluorescence intensity (~5.7x above background) due to an opening of the probe and concomitant reduction in the F\xc3\xb6rster resonance energy transfer efficiency. Addition of thermostable RNase H resulted in the cleavage of the RNA loop which eliminated energy transfer. The cleavage step also released bound target DNA, enabling it to bind to another R/D-MB probe and rendering the approach a cyclic amplification scheme. Full processing of R/D-MBs maximized the fluorescence signal to the fullest extent possible (12.9x above background), resulting in a ~2\xe2\x80\x932.8 fold increase in the signal-to-noise ratio observed isothermally at 50 \xc2\xb0C following the addition of RNase H. The probe was also used to monitor real-time PCR reactions by measuring enhancement of donor fluorescence upon R/D-MB binding to amplified pUC19 template dilutions. Hence, the R/D-MB-RNase H scheme can be applied to a broad range of nucleic acid amplification methods.

enzymatic_amplification_of_dna/rna_hybrid_molecular_beacon_signaling_in_nucleic_acid_detection

INTRODUCTION<!>Materials<!>R/D-MB Probe Design<!><!>R/D-MB Probe Design<!>Fluorescence Measurements<!>RNase H Cleavage Reactions and Quantifying the S:N Ratio<!>R/D-MB/RNase H Concentration Dependence Experiments<!>Real-Time PCR Reactions<!>Thermal Denaturation Profiles<!>Isothermal Temperature Dependence of R/D-MB Performance<!>ssDNA Target Concentration Dependence of R/D-MB Performance<!>Real-Time PCR Experiments<!>DISCUSSION

<p>Since their advent in 1996 [1–2], molecular beacons (MB) have become a dominant method of biomolecular recognition [3] due to their high selectivity, sensitivity, and stability. MB's are used to quantitatively detect DNA and RNA through the use of Förster Resonance Energy Transfer (FRET) [4], a distance-dependent nonradiative energy transfer phenomenon that can occur between a "donor" and "acceptor" fluorophore that exhibit spectral overlap. FRET is typically employed in the MB design by taking a stem-loop DNA oligomer and attaching a fluorescent donor and a nonfluorescent acceptor (the "quencher") to the 3′ and 5′ ends, respectively. The resulting tight proximity between the FRET pair inherent in the stem configuration produces a highly efficient FRET and concomitantly low donor fluorescence background (Fclose) in the absence of probe targets [5]. When target oligomers recognize and bind to the complimentary loop region of the MB to form a double strand, the stem is forced apart which parts the fluorophore from the quencher and leads to a reduction in FRET and concomitant increase in donor fluorescence intensity (Fopen) [5]. Using this mechanism, MB's are able to have a high signal-to-noise (S:N) ratio when applied in immobilized platforms and especially solution based assays. Design improvements leading to higher S:N ratios have led MBs to surpass their originally envisioned application in monitoring PCR reactions [1–2]. As detailed by Wang et al. [5], they are now used in protein assays, enzyme monitoring, cellular mRNA detection, biosensors, molecular computing, and as aptamers.</p><p>Even though the MB's particular use of FRET has produced S:N ratios that are superior to what is seen from alternatives, the sensitivity of conventional MBs is nonetheless inadequate for detecting oligomeric targets found in low abundance. This limits their use to applications involving highly abundant or stimulated gene products [5]; it is thus desirable to improve their sensitivity as dictated by the S:N ratio calculated from [6]: (1)Fopen-FbufferFclose-Fbuffer where Fopen is the donor fluorescence when MB is bound to its target, Fclose is the donor fluorescence in the absence of target, and Fbuffer is the background fluorescence in the absence of probe. It is seen from Eqn. 1 that the S:N ratio of a particular MB design may be attenuated by: 1) incomplete initial quenching in the absence of targets, which increases Fclose; 2) incomplete abolishment of FRET once a target is bound, which decreases Fopen; and 3) a lack of the type of processive cycling seen in the Catacleave probe [7–8] or the SNP assay of the Liu lab [34] which would enable one target molecule to react with multiple MBs and reduce the concentration of targets needed to achieve a particular Fopen. Most attempts at improving the S:N ratio have focused on improving the MB design to either reduce Fclose by enhancing the initial quenching efficiency or increase Fopen by enhancing the performance of the donor [5]. It is evident from the literature that targeting Fclose has proven more successful to date based on fold enhancements of the S:N ratio.</p><p>The quenching efficiency seen in MB designs has been improved from a basal range of 85–97% [5] to higher values using a variety of design enhancements. Gold surfaces [9–10] and gold nanoparticles [11–12] have been employed as quenchers paired with fluorescent dyes like Rhodamine 6G and Fluorescein (FAM) [11]. Specifically, the average quenching efficiency for a Gold/Rhodamine 6G and Gold/Fluorescein pairing was found to be 99.5% and 98.7% with corresponding S:N ratios of 182 and 76, respectively [11]. Superquenching has also proven effective at reducing Fclose and has involved the use of a multiple-quencher assembly paired with one fluorophore. A seminal study by Yang et al. showed that using a three quencher assembly instead of a single quencher significantly increased the S:N ratio of their MB design from 14 to 320 [13]. Besides the improvement of quenching efficiency, another strategy that has proven successful at reducing Fclose is the use of a chemiluminescent detection approach with a greatly reduced background signal [14–15]. Chemiluminescent MBs have been able to achieve lower detection limits in the femto to attomolar range, which is ~4 orders of magnitude more sensitive than typical fluorescent MB probes [14]. Thus focusing design modifications on improving Fclose has led to several orders of magnitude improvement in the MB S:N ratio.</p><p>A strategy along different lines to improving the S:N ratio of the MB has been the enhancement of Fopen by increasing the maximal fluorescence intensity of a given fluorophore using techniques such as multiple fluorophore based systems [16–19], wavelength-shifting MBs [20], conjugated polymers [21], and quantum dot MB systems [22]. In particular, quantum dots and conjugated polymers have been shown to produce a much higher fluorescence intensity than that of a single small molecule organic fluorophore. The maximal fluorescence intensity of a single quantum dot can have the same magnitude as the combined emission from 20 individual organic fluorophores [22]. Due to their high quantum efficiency, conjugated polymers, such as the water-soluble polyelectrolyte Poly (phenylene ethylene) (PPE), have also been shown to be ≥ 6 times brighter than conventional organic dyes such as Cy3, Fluorescein, tetramethylrhodamine, and Alexa Fluor 488 [5]. Under the experimental conditions tested by Tan et al., a single PPE chain was shown to be 20 and 6 times brighter than a Cy3 and Alexa Fluor 488, respectively [5]. However, it is evident that attempts to improve Fopen by employing enhanced donors have only increased the S:N by one order of magnitude.</p><p>Another Fopen-oriented improvement that could be made to general MB design is to ensure that once the MB has recognized and bound to its target, FRET is abolished by maximizing the distance between the fluorophore and quencher. According to Förster theory, FRET efficiency is inversely proportional to the sixth power of the distance between the centers of the fluorophore and quencher [23]. A sufficient increase in distance would drive the FRET efficiency to zero and ensure the maximum possible donor emission regardless of the fluorophore chosen. Current MB technologies can only separate the fluorophore/quencher pairing by a distance dictated by the number of bases in the complimentary loop region. If a bound loop could be selectively digested while leaving the target intact, this would not only maximize separation of the donor from the quencher, but also liberate the previously bound target and allow it to bind to another free probe. This would create an amplification cycle wherein each target single-stranded DNA (ssDNA) participates in multiple cleavage events until the entire MB population is processed [27]. A generalized implementation of this approach could be to render the entire loop of a MB as a ribonucleotide sequence complementary to target and use commercially available Thermostable RNase HI to selectively digest the loop.</p><p>Technically, previous studies performed by the Liu lab [24–25,34] have employed CpRNase HII [24–25] and TthRNase HII [34] to selectively nick MBs containing a single ribonucleotide in their the loop sequence (DNA-rN1-DNA MBs) when bound to target DNA with a perfectly matched sequence specifically nicked the MB loop portion of the target•MB helix, allowing the liberated target to bind to another free and unprocessed probe. This amounted to an ingenious assay for detecting single-nucleotide polymorphisms (SNPs) that takes advantage of CPT to amplify the signal when an SNP is present. Though the Liu lab cast the technique as originally envisioned with CpRNase HII as amenable to general oligonucleotide detection [24], there are some limitations with the approach that make it less effective for this purpose. The CpRNase HII based assay must be run isothermally at mesophilic temperatures and cannot be combined with approaches requiring thermal cycling, such as PCR or double-stranded DNA detection. Later, the addition of TthRNase HII [34] to the method made it compatible with thermal cycling, but the assay was not readily compatible with PCR due to enzyme-mediated inhibition of amplification, which is a serious drawback. Furthermore, the technique requires expression and purification of nickases that are not commercially available. Also, multiple studies have shown that the binding efficiency and kinetics that control RNase H mediated cleavage is better with four or more ribonucleotide bases as opposed to three or less [28–29].</p><p>We rationalized that a more generalized approach to marrying MBs and CPT for trace oligonucleotide detection would be to rely on the faster kinetics produced by the combination of a full RNA loop perfectly matched to target and Thermostable RNase HI, which is commercially available and requires at least 4 consecutive ribonucleotides for binding and digestion [8]. Thus the technique would require relying completely on stable DNA/RNA hybridization for target recognition and concomitant melting of the stem, which to our knowledge has never been done. Thus in this study, we developed and tested a generalized design for a hybrid MB compatible with CPT and PCR that contains a DNA stem and a long 21-bp RNA loop that is processed with Hybridase™ Thermostable RNase HI in nuclease free buffer. We showed through a melting curve analysis that the ssDNA target was recognized by and annealed to the R/D-MB loop to form a double-stranded helix in which the stem was broken apart and an initial fluorescence increase was obtained (Fopen-1). After adding thermostable RNase HI enzyme, the RNA loop strand of the R/D-MB•target complex was specifically cleaved, producing an enhanced donor fluorescence signal (Fopen-2) and leaving behind an intact, unbound ssDNA target as expected [26–27]. Testing a constant probe concentration against varying concentrations of ssDNA target in the presence of RNase H demonstrated accumulation of cleaved R/D-MB probes until a maximum Fopen was achieved. Finally, application of the R/D-MB probe in monitoring a PCR was demonstrated. The combined experiments show that complete reliance on RNA to DNA hybridization works very well in MBs, producing a broadly applicable scheme that has superior S:N ratios and kinetics when combined with thermostable RNase HI.</p><!><p>Hybridase™ thermostable RNase H was from Epicentre Technologies and RNase H reaction buffer (75 mM KCl, 50 mM Tris-HCl @ pH 8.3, 3 mM MgCl2, and 10 mM dithiothreitol (DTT)) was from New England Biolabs. All PCR related agents were purchased from Invitrogen, including Native Taq DNA polymerase, 10 mM dNTP mixture, 10 μM forward (5′-TGTGGAATTGTGAGCGGATAAC-3′) and reverse (5′-CCTCTTCGCTATTACGCCAG-3′) primers, and a 16 reaction GeneTailor™ Site-Directed Mutagenesis System. Plasmid purification was done using the Qiagen Plasmid Mini Purification Kit. Stock (100 μM) ssDNA target and R/D-MB solutions were synthesized by Sigma and Biosearch Technologies, respectively.</p><!><p>The R/D-MB probe used in this study was designed using the software Beacon Designer from Premier Biosoft International (http://www.premierbiosoft.com). The sequences of the R/D-MB and ssDNA target were designed as followed:</p><!><p>Molecular Beacon Probe: FAM-5′-CTCGCG-CCGCAGAUUAUAGUGGACGCC-CGCGAG-3′- BHQ1</p><p>Target ssDNA Sequence: AAAAA-GGCGTCGACTATAATCTGCGG-AAAAA</p><!><p>The DNA stem region of the R/D-MB probe is underlined and the RNA loop region is italicized. Attached at the 5′ end of the DNA stem is the donor fluorophore, Fluorescein, while attached at the 3′ end is a Black Hole Quencher 1. The complimentary ssDNA target sequence is shown in plain typeface and the two end-flanking regions are italicized. A melting temperature (Tm) of 72 °C was predicted for the design, as well as an annealing temperature (Ta) of 64 °C for half-maximal annealing between the RNA loop region and ssDNA target. A StepOne™ Real-Time PCR system was set to monitor and record fluorescence as a function of temperature using ramps from 25 to 95 °C at 0.3 degrees per second.</p><!><p>All fluorescence measurements involving melting curve analysis, target recognition and binding, RNase H cleavage, and monitoring of real-time PCR reactions were performed using a StepOne™ Real-Time PCR system from Applied Biosystems. All real-time fluorescence data was monitored and recorded using StepOne Software v2. Recorded fluorescence intensities were means calculated by averaging fluorescence intensity values from three replicates per sample type.</p><!><p>RNase H cleavage reactions were carried out in a 1x RNase H reaction buffer with 50 nM of R/D-MB probe, 0.1–200 nM of ssDNA target, and 2.5 U of thermostable RNase HI in a total reaction volume of 20μL. R/D-MB probe and target ssDNA were first incubated at 95 °C for 2 min, followed by the addition of RNase H and monitoring of fluorescence intensity over a 100 min timeframe at a set reaction temperature using the StepOne™ Real-Time PCR system. Fluorescence excitation light was passed through an optical filter suitable for FAM.</p><p>To determine the sensitivity of the assay at a set reaction temperature, samples were carefully prepared that could be used to quantify the S:N ratio as given in Equation 1. For each experimental run, a control sample was made containing only 1x RNase H buffer made up in nuclease free water (Bioexpress) and used to measure Fbuffer. Fclose was measured from samples containing only R/D-MB probe in an RNase H buffer and nuclease free water. Fopen was measured by creating two different kinds of samples, each in 1x RNase H buffer: R/D-MB probe + target ssDNA and R/D-MB probe + target ssDNA + RNase H. Two different sets of S:N ratios were calculated, one representing Fopen-1 (R/D-MB + target ssDNA), and the other Fopen-2 (R/D-MB + target ssDNA + RNase H) as shown in Fig. 1.</p><!><p>Different concentrations of ssDNA target were incubated with 50 nM R/D-MB probe in the presence or absence 2.5 U of thermostable RNase HI in 1x RNase H reaction buffer made up in nuclease free water for a total sample volume of 20μL. The 50 nM R/D-MB probe concentration was selected after running several preliminary experiments to determine the lowest probe concentration that could be used to achieve a strong fluorescence signal with a constant ssDNA target concentration (data not shown). S:N ratios at each target ssDNA concentration tested was calculated according to Eqn. 1 by comparing the fluorescence intensities of samples containing either R/D-MB + non-target ssDNA, R/D-MB + target ssDNA, and R/D-MB + target ssDNA + RNase H. All fluorescence measurements were recorded using the 48 well-plate StepOne™ Real-Time PCR system over a 100 min timeframe.</p><!><p>Prior to running any PCR reactions, site-directed mutagenesis was used (via the GeneTailor™ Site-Directed Mutagenesis kit) to insert the ssDNA target sequence complimentary to the RNA loop of the R/D-MB into a pUC19 plasmid. After the sequence insertion, the plasmid was transformed into DH5α™-T1R E. coli competent cells, and plasmid purification was carried out using a Qiagen Plasmid Mini Purification Kit. A 1% agarose gel was run to show that the target sequence (239 bp) was indeed inserted into the plasmid (data not shown).</p><p>Real-time PCR reactions were carried out in 1x RNase H buffer using 2.5 U of Native Taq DNA polymerase, 200 nM R/D-MB probe, 250 nM of dNTP mixture, 200 nM of ssDNA primer (reverse and forward), 2.5 U of thermostable RNase H, and 1 μL of diluted pUC19 template in a total reaction volume of 50μL. The PCR protocol followed consisted of 1 cycle of 95 °C for 3 min, followed by a 40-cycle amplification scheme wherein each cycle consisted of a 95 °C incubation step for 30 sec, followed by a 50 °C step for 45 s, and a 72 °C step for 30 s. The protocol was completed with a final extension step at 72 °C for 8 min. Real-time data was collected during the 50 °C annealing stage. Triplicates were run for each serial dilution of template and a "no template" control was used to test for the presence of contamination.</p><!><p>To verify the integrity of the R/D-MB probe design, the melting temperature of the R/D-MB probe stem and the temperature for half-maximal annealing between the RNA loop region of the R/D-MB and the complimentary ssDNA target sequence were measured and compared with predicted values. Melting curves were determined by measuring the fluorescence of the 200 nM R/D-MB as a function of temperature in the absence and presence of 200 nM ssDNA target and are shown in Fig. 2. Comparison between the thermal denaturation profiles of the R/D-MB + target or R/D-MB probe and the values predicted for Ta or Tm, respectively, confirmed that our design produced a properly functioning R/D-MB having a Ta of approximately 64 °C and Tm of 72 °C.</p><!><p>Having established that the R/D-MB probe showed the expected thermally dependent behavior, the next experimental step in testing the proposed scheme (Fig. 1) was to confirm the enhancements of Fopen expected from nuclease-mediated loop cleavage and cyclic amplification. Hybridase™ thermostable RNase HI was chosen for this purpose as it is able to digest the RNA portion of a RNA-DNA hybrid, without affecting ssDNA or unhybridized RNA, over a broad temperature range that extends as high as 95 °C. Furthermore, its optimal activity is 3-fold higher than that of E. coli RNase H. Though thermostable RNase HI is optimally active above 65°C, since the Tm and Ta of the R/D-MB probe is 72 °C and 64 °C, respectively, reaction temperatures lower than Ta were chosen for isothermal characterization. Thus to determine the optimum reaction temperature which would maximize the S:N ratio as stated in Eqn. 1, isothermal RNase H cleavage reactions were carried out at 45, 50, 55, and 60 °C. No temperatures below 45 °C were tested, as using temperatures ≥45 °C ensured the prevention of non-specific binding.</p><p>Shown in Fig. 3A are the results from a representative set of isothermal experiments from which we determined the RNase-H-dependent enhancements of the S:N ratio seen at a specific temperature (in this case, 50 °C). The fluorescence background (Fclose) was first measured from samples containing R/D-MB only or R/D-MB plus non-target ssDNA, both of which gave fluorescence intensities ~7× higher than control samples of buffer alone. In the presence of ssDNA target, donor fluorescence intensity increased by another ~5.7× to a level representing Fopen-1. When thermostable RNase H was added to a R/D-MB + target ssDNA sample, an additional ~2.2× increase in fluorescence intensity (or a ~12.9× increase over the intensity of samples containing R/D-MB + non-target) was observed that corresponds to Fopen-2. The fact that the trace for the test sample containing R/D-MB probe + target + RNase H reached saturation suggests that all of the R/D-MB probes were processed by the cyclical binding of ssDNA target and RNA-loop cleavage by RNase H. Hence at 50 °C with the concentrations tested, cleavage of all R/D-MB probes was nearly complete within 10 min after the reaction had started.</p><p>Final fluorescence intensities were determined according to this scheme at the isothermal reaction temperatures of 45, 50, 55, and 60 °C, and then these intensities were used to calculate the ratios of S:N according to Eqn. 1 (Fig. 3B). Two ratios were computed, one for Fopen-1 and the other for Fopen-2. Over the temperatures tested, the highest S:N ratios were obtained at 45°C (S:N = 19.9 and 9.7 for Fopen-2 and Fopen-1, respectively), followed by 50°C and 55°C (S:N = ~14–15 and ~6.4), and lastly 60 °C (S:N = 7.9 and 3.5). The reason for the higher S:N ratios at 45°C compared to the next highest temperature at 50 °C (~1.4× difference) was that at lower temperatures and in the absence of ssDNA targets, there is a higher probability that an individual R/D-MB within the whole population of probes will be found with an annealed stem (see red trace in Fig. 2). This would result in a higher overall quenching efficiency averaged over the probe population and thus a lower Fclose. Comparing the S:N ratios based on Fopen-2 with the S:N ratios based on Fopen-1, it was evident that the addition of RNase H increased the overall S:N ratio of the assay by ~2x at each temperature setting. Based on these results, the optimal isothermal temperature setting for our R/D-MB + RNase H reaction system is likely 45 °C due to the lower Fclose value in comparison to the other settings. However, subsequent experiments were conducted at 50 °C because it was expected that the higher temperature would be more effective when applying the R/D-MB in PCR applications. Nevertheless, these results demonstrated our rationale that the combination of the full RNA loop of the R/D-MB with thermostable RNase HI cleavage reactions could provide an Fopen oriented enhancement to the S:N ratio of a conventional MB system, and that this enhancement is applicable over a broad temperature range and obtainable over a useful time scale.</p><!><p>Having characterized the temperature dependence of the R/D-MB at one set of concentrations, it was important to examine the dependence of probe performance on the concentration of target ssDNA. Representative traces of donor fluorescence intensity versus time (min) are plotted in Fig. 4A. Fig. 4B shows the corresponding S:N ratios based on Fopen-1 (determined from samples with R/D-MB + target) or Fopen-2 (determined from samples with R/D-MB + target + RNase H) at a 100 min incubation time. Both Figs. 4A and 4B illustrate the effects of the relative concentration of target ssDNA on probe performance. At target ssDNA concentrations of 20 nM and greater, all R/D-MB probes were mostly processed by RNase H enzymatic activity within a 40 min timeframe as indicated by the fluorescence amplitude reaching saturation at ~30,000 arbitrary units. When the target ssDNA was approximately equimolar or in excess to the probe (i.e. ≥40 nM) in samples containing RNase H, saturation was reached within 20 min. The 200 nM target ssDNA condition reached saturation within less than 1 min, suggesting that the kinetics of RNase H binding and cleavage are very rapid at the 50 nM substrate level. Thus at lower concentrations of ssDNA target, the rate of the cleavage reaction may be limited by the target-to-probe binding equilibrium and lower effective substrate (i.e. target•probe) concentrations. However, even at a concentration 500× more dilute than that of R/D-MD, a detectable change in fluorescence intensity was observed for 0.1 nM ssDNA target in the presence of RNase H. It was noted that signals from the lowest concentrations of ssDNA target tested (i.e. 0.1 nM and 1 nM) in the presence of RNase H did not reach saturation within the 100 min timeframe, but it was evident that saturation would have been reached given a longer observation time.</p><p>Depicted in Fig. 4B is the [target ssDNA] dependence after a 100 min sample incubation time of RNase H mediated enhancement of the S:N ratio. Note that the x-axis was deliberately left unscaled because [probe•target], the true substrate of RNase H, was not determined. Enhancement was quantified by comparing the S:N ratio based on Fopen-1 (from samples incubated without RNase H) with the S:N ratio based on Fopen-2 (from samples matched in concentration of ssDNA target and containing RNase H). Though the concentration of the true substrate of RNase H (i.e. probe•target) was not determined, the S:N ratio based on Fopen-2 increased in relation to increasing target ssDNA concentration until 5 nM ssDNA target was reached. At data points ≥5 nM the S:N ratio based on Fopen-2 saturated at ~20, suggesting that this was the maximum ratio possible at 100 min of reaction time due to the cleavage-induced abolishment of FRET within every probe in the sample population. Similarly, the S:N ratio based on Fopen-1 steadily increased with higher target ssDNA concentration. One can see that the increase in Fopen-1 started to decrease relative to the increase in [ssDNA] starting at the 100 nM data point. An RNase H mediated enhancement of the S:N ratio by ≥2× of was observed when [target ssDNA] ≤ 20 nM or when the molar ratio of target to probe was less than 2:5. This is due to the fact that as more R/D-MB was bound by ssDNA target, the more Fopen-1 approached the Fopen-2 determined at the same target ssDNA concentration. This made sense, as the relatively longer 21-bp RNA loop should facilitate good separation between donor fluorophore and quencher when the RNA loop of a R/D-MB is hybridized with a ssDNA target. Nevertheless, the fact that the saturation point associated with Fopen-2 was ~1.2× the highest point associated with the maximum Fopen-1 suggests that the signal is appreciably enhanced by the complete abolishment of FRET. Thus RNase-H-mediated enhancement of S:N is expected to improve the detection of target ssDNA, and especially at concentrations much lower (≤ 10−1×) than that of the R/D-MB due to cyclic processing.</p><p>Since the prior conclusions about S:N ratios were drawn from data tabulated at 100 minutes; this begs the question of whether the trends observed would still hold true at shorter incubations times. Fig. 4C provides the answer by tabulating S:N ratios as a function of [target ssDNA] based on Fopen-2 at incubation times of 10, 20, 30, 45, and 60 minutes. For comparison, S:N ratios based on Fopen-1 were tabulated at 10 and 60 min. The data is depicted as a series of plots for clarity. As in Fig. 4B, the x-axis was deliberately left unscaled because [probe•target] was not determined. It can be seen that at 10 min, RNase H mediated cleavage of R/D-MBs enhanced the S:N ratio at [target ssDNA] ≥ 1 nM, but at [target ssDNA] < 1 nM actually reduced the S:N ratio. This was likely due to the fact that at [target ssDNA] = 0.1 nM the concentration of probe•target must logically be ≤ 0.1 nM, resulting in a slow rate of cleavage. However, at 20 min the S:N ratio was enhanced at every [target ssDNA] tested, and at 30 min enhancement was generally even greater. For lower target ssDNA concentrations (i.e. 1 and 5 nM), RNase H mediated enhancement was as high as ~2.8×. It is also seen that as incubation times increase going from 10 min and up to 60 min, the S:N decreases with [target ssDNA] ≥ 20 nm. This can be explained with the slight increase in fluorescence signal associated with the R/D-MB only sample (Fclose) (Fig 4A) with time. This was also observed with the R/D-MB + non-target sample. This slight increase in Fclose with time would increase the denominator for the S:N calculation as shown in Eqn. 1, and thus would decrease the S:N with time. This is also why S:N ratios as high as ~30 are observed at shorter incubation times, vs. the maximum of ~20 seen at 100 min in Fig. 4B. The phenomenon was likely due to the choice of 50 °C [5] as an assay temperature, as our traces at 45 °C did not show any increase in fluorescence with time (data not shown). Overall, the Fopen-oriented enhancements to the S:N ratio offered by the R/D-MB + RNase H design approach requires a trade-off between better signal and longer incubation times, especially at lower concentrations of target ssDNA.</p><p>Since it was evident that this trade-off stems from the rate of the concentration dependence of the rate of the RNase H cleavage reaction, mono-exponentials were fit to our data on the [target ssDNA] dependence of R/D-MB fluorescence intensity in the presence of RNase H. Nonlinear regression of the monoexponentials to the data produced excellent fits (R2 ≥ 0.997), implying that even though [probe•target] was unknown, [probe•target] increased as [target ssDNA] was increased, as would be expected. Fig. 4D shows approximated first-order rate constants at the target ssDNA concentrations tested, and it is seen that the rate constants increased as the target ssDNA concentration increased. Unfortunately, it was not possible to fit the data sets to the Michaelis Menten model because [probe•target] was unknown. Future experiments could be conducted to determine the equilibrium concentration of probe•target based on protocols from Behlke et al. [30–31], and Yang el al [32]. However, the data presented in Fig. 4D is still useful as a "standard curve" for predicting the original concentration of target ssDNA based on the rate of fluorescence increase. Overall, our experiments characterizing the ssDNA target concentration dependence of R/D-MB + RNase H performance further demonstrated our rationale that the full RNA loop approach offers a broadly applicable Fopen-oriented enhancement of S:N ratio.</p><!><p>It was important to test the R/D-MB in the context of a PCR reaction, since real-time PCR has been accepted as the standard method for nucleic acid detection and quantification [33]. Presently, sequence-specific probes used in real-time PCR applications include Taqman, MBs, and scorpion primers. Of these nucleic acid detection probes, real-time PCR studies done by Wang et al. [33] have shown the MBs produced the highest S:N ratios and are the most sensitive out of the common nucleic acid probes used. However, one serious drawback of MBs during PCR is that only one MB probe can hybridize to a single amplified DNA sequence during a given PCR cycle. This limits the sensitivity of the scheme for nucleic acid detection because the progressive accumulation of fluorescence signal is completely dependent on amplification of target ssDNA [8]; but in a CPT (cyclic probe technology) approach, the cyclic nature of target binding, probe cleavage, and target release facilitates the accumulation of fluorescence intensity based on the activity of RNase H. Thus we rationalized that the R/D-MB approach could be applied as an application of CPT in PCR.</p><p>To test this rationale and establish the performance of the R/D-MB + thermostable RNase HI system during a typical PCR, we performed real-time PCR experiments with a pUC19 plasmid containing a 239-bp insert that served as the DNA template in the presence of 200 nM R/D-MB and 2.5 U thermostabe RNase H in a 50μL PCR buffer solution. Since PCR is not an isothermal process, it is necessary to introduce a detection phase within the amplification cycle. This was done by including a 45 sec R/D-MB probe cleavage step at 50°C [8]. This step serves as a precaution to allow the RNA loop of the R/D-MB to hybridize with its ssDNA target, while also giving appropriate time for RNase H to cleave its substrate consisting of probe•target. Using an initial template concentration of 155 pM, serial dilutions were made ranging from a dilution factor of 10× to 105× more dilute. The results plotted in Fig. 5A shows the normalized fluorescence intensities based on RNase H mediated cleavage of the different concentrations of diluted template DNA that were tested versus PCR cycle number. The threshold cycle Ct,, which is defined as the PCR cycle at which the initial rise in exponential fluorescence begins for a given template dilution, was identified as the point when the level of fluorescence started to rise above the normalized value of 0.3. Ct values for each template dilution were determined accordingly and plotted against the log of the template dilution as shown in Fig. 5B. The expected linear relationship between template dilution and threshold cycle was obtained, and from this plot it is a simple matter to determine the initial concentration of an unknown sample once the Ct value is known. Thus the R/D-MB + RNase H scheme performs robustly over the thermal cycling approach that PCR represents, and thus could likely be applied to other assays based on thermal cycling, such as the detection of double stranded DNA.</p><!><p>From the standpoint of Eqn. 1, our study has introduced a novel Fopen oriented approach to enhancing the S:N ratio of a conventional MB system by including a full RNA loop in the MB design that is complimentary to a desired target ssDNA sequence and is intended to be cleaved by Hybridase™ thermostable RNase HI. This combines the inherently low Fclose of an MB, due to tight localization between the fluorophore and quencher, with all the advantages of CPT to surpass some of the Fopen oriented weaknesses affecting conventional MB systems. It was demonstrated in this study that complete reliance on RNA-DNA hybridization chemistry worked in well in the MB scheme (Fig. 2) at greater than mesophilic temperatures (Fig. 3). With adequate incubation time, fluorophore-quencher FRET could be completely abolished (Fig. 4B) and all the R/D-MB probes could be processed (Figs. 4B, 4C). The accumulation of fluorescence signal in the system presented was thus a direct result of multiple cycles of R/D-MB hybridization to ssDNA target followed by thermostable RNase HI mediated cleavage of the RNA loop of the bound probe. As with any CPT assay, this method allows just a single ssDNA target to bind to multiple R/D-MBs until all the R/D-MBs are cleaved by RNase H. Most significantly, this Fopen oriented approach to improving the sensitivity of a conventional MB offers amplification that is fully compatible with PCR, and is additional to and should be compatible with many of the other previously reported Fopen- and Fclose-oriented approaches to enhancing conventional MB sensitivity.</p><p>In comparison to one of its closest relatives found in the literature, the CataCleave probe [7–8], the R/D-MB + thermostable RNase H system offers several advantages due to its nature as a MB. For example, hybridization of the CataCleave probe to its target sequence does not lead to any change in donor fluorescence emission, meaning that detection requires that RNase H mediated cleavage is applied. As demonstrated in Fig. 4, the R/D-MB-RNase H system can be applied as a conventional MB without addition of RNase H if desired, with good performance still obtained. It may also be seen that RNase H mediated enhancement of the S:N ratios seen in our study is greater than that obtained by Harvey et al. with the CataCleave approach. Even with 200 nM CataCleave probe at the highest [target ssDNA] tested, Harvey et al. obtained an S:N ratio of ~6, whereas S:N ratios as high as ~30 were seen in our study (Fig. 4C). Presumably, this is due to a higher initial fluorescence background (Fclose) associated with the CataCleave probe than compared to the R/D-MB. In the CataCleave probe design the fluorophore/quencher pairing are separated by four ribonucleotides, which results in a greater distance and concomitantly reduced FRET than that facilitated by the stem configuration of a conventional MB probe.</p><p>As mentioned in the introduction, other studies by Liu et al. [24–25,34] incorporated single ribonucleotides into the loop portion of their MB design with the intention that RNase H nickases would nick the hybridized probes in what amounts to a very useful assay for the detection of single nucleotide polymorphisms (SNPs) [34]. One major difference between the approach presented in this study and that of the Liu lab is that the SNP probe relies principally on DNA-DNA hybridization for target recognition and stem melting. Though the SNP probe scheme was presented as potentially generally applicable to trace oligonucleotide detection, initial studies based on the use of CpRNase HII [24–25] produced designs only applicable to isothermal and mesophilic temperature settings. The S:N ratios seen in those studies were also lower than was demonstrated here; for example, a maximum S:N ratio of ~14 may be calculated based on the background fluorescence from probe without target present vs. the fluorescence intensity observed when cleavage of 1000 nM probe, facilitated by 32 nM target, saturates the system [24]. This maximum S:N ratio of ~14, which is less than half of what was observed in our study with much less probe consumed, is what would be expected if the SNP probe was applied generally. Also, due to the lower temperature settings required by the assay, it was likely more susceptible to non-specific binding. Though the assay was improved in a later study by using TthRNase HII from T. thermophilus instead of CpRNase H11 [34], it was still incompatible with PCR apparently due to RNase H mediated inhibition of the amplification reaction. As can be seen from Fig. 5, our full-RNA loop probe is able to function very well in the PCR environment.</p><p>Another expectation for our design was that the use of a full RNA loop and thermostable RNase HI would lead to a more rapid assay time. We rationalized that using an entire loop comprised of RNA would ensure maximal enzyme binding and cleavage efficiency. From our traces on R/D-MB + ssDNA samples (Fig. 3 and 4), probe-target hybridization appears to be rapid, suggesting that the rate limiting factor is the concentration of probe•target. Compared to the TthRNase HII based SNP assay from the Liu lab [34], our assay is approximately twice as fast. One can see in Fig. 4C from our study that in our assay a S:N of 20 was achieved for 0.1 probe:target within 20 minutes. Using Fig. 2 from Liu et al. 2010 to obtain an estimate of Fclose, Fig. 3A shows that it took 40 minutes for the SNP assay to achieve a S:N ratio of 20. Hence the full RNA loop R/D-MB + thermostable RNaseHI assay presented in this study is indeed able to achieve higher S:N ratios more rapidly.</p><p>In summary, our R/D-MB + thermostable RNase H system amplifies the fluorescence signal under both isothermal and non-isothermal conditions as illustrated by our basic RNase H cleavage and real-time PCR experiments. This allows our method to be compatible with a variety of experimental protocols. As demonstrated by our isothermal experiments, the S:N ratio of a typical MB system was enhanced by as high as ~2.8× by thermostable RNase H mediated cleavage, and S:N ratios as high as ~30 (with a 10 min incubation) were observed. It was shown that the approach is useful over a variety of time scales. It was also demonstrated that the assay works under thermal cycling conditions, and a PCR was successfully monitored using the full RNA loop R/D-MB + thermostable RNase H scheme. Since the approach constitutes an MB design, it is compatible with all previously reported design enhancements reported for improving MB performance, e.g. the use of superquenchers, better fluorophores, etc. For example, if the ~22.9 fold reduction in Fclose seen in Yang et al. [13] was obtained with our system, one could expect the maximum Fopen,1 from Fig. 4B to be increased from ~16.5 to ~377, and the maximum Fopen,2 to increase from ~20 to ~457. Thus other MB enhancements could drastically improve the S:N ratio of our design. Another compatible enhancement would be to create a longer stem that is part of the target-binding sequence [31], which could produce higher hybridization rates while also improving selectivity. Likewise, using the R/D-MB + thermostable RNase HI approach presented in this study is expected to enhance the performance of most existing MB designs.</p>

PubMed Author Manuscript

Spectroscopic investigation of defect-state emission in CdSe quantum dots

CdSe quantum dots are the most studied Cd-based quantum dots with their high quantum yield, high photostability, narrow emission band, and easy synthesis procedure. They are frequently used to develop light emitting diode (LED) due to their unique photophysical properties; however, their narrow emission band causes a challenge to design white LEDs because white light emission requires emission in multiple wavelengths with broad emission bands. Here in this study, we developed CdSe quantum dots with a narrow band-edge emission band and broad defect-state emission band through a modified two-phase synthesis method. Our results revealed that defect-state emission is directly linked to the surface of quantum dots and can be excited through exciting surfactant around the quantum dot. The effect of surfactant on emission properties of CdSe quantum dots diminished upon growing a shell around CdSe quantum dots; as a result, surface-dependent defect-state emission cannot be observed in gradient heterogeneous alloyed CdSxSe1-x quantum dots.

spectroscopic_investigation_of_defect-state_emission_in_cdse_quantum_dots

1. Introduction<!>2. Materials and methods<!>2.1. Synthesis of cadmium myristate<!>2.2.Synthesis of sodium hydrogen selenide<!>2.3. Synthesis of CdSe quantum dots with high amount of defect states<!>2.4. Synthesis of CdSxSe1-x quantum dots<!>2.5. Optical characterization of CdSe and CdSxSe1-x quantum dots<!>3.1. Synthesis of CdSe and CdSxSe1-x quantum dots<!><!>3.2. Absorption and emission properties of CdSe quantum dots<!>3.3. Origin of defect-state emission in CdSe quantum dots<!>4. Conclusion

<p>Cadmium chalcogenide (Cd-chal) quantum dots (QDs) are the earliest quantum dots which have been synthesized and explored since the beginning of 1990s [1,2]. They are semiconductor materials with fluorescence properties and possess unique photophysical and structural characteristics such as high quantum yield, high photostability, single narrow emission band, wide absorption bands, high molar extinction coefficient, small size (2–10 nm), semiconductor nature, and modifiable surface [1–4]. With their unique properties, Cd-Chal QDs have been widely used in many different technologies such as solar cells, LEDs, biotechnology, military, and medicine [5–11]. Since they have excellent photophysical properties, they are frequently used in LED and solar cell applications [8,9] and even high-tech brands, such as Samsung, have adapted QDs into their monitor systems [12,13].</p><p>However, one of these excellent properties of QDs, i.e. having a narrow emission band, can be considered a disadvantage in white light emitting diode (wLED) design. In order to overcome this disadvantage, wLEDs with multiple quantum dot layers, which emit light in different colors, have been produced [14]. However, another unique feature of QDs that complicates design of multilayered QD films is their high molar extinction coefficient [1,2]. With high molar extinction coefficient, green light emissive QDs strongly absorb light in blue light spectrum [1,2] which are the two main colors in wLED design [14].</p><p>To prevent self-absorption of QDs in multilayered QD films, multicolor light emitting QDs with single composition have been proposed [3,15–21]. Two general ways to design multicolor light emitting QDs have been proposed: 1) doping binary or ternary QDs with an extra element [3,15–19], 2) creating surface defects [20,21]. Both of these methods cause formation of defect states and thus formation of defect-state emission in QDs' emission spectrum. In a recent study, Ünlü et al. doped CdSe quantum dots with Te through a modified two-phase synthesis method and obtained controllable defect-state emission due to Te doping [3]. In a different study, Yuan et al. synthesized nitrogen-doped carbon quantum dots to achieve white light emission through a single-component system [15]. Yang et al. synthesized dual emissive Ag:InP/ZnS quantum dots through a hot injection method and achieved high quantum yield around 75% [16]. Zhang et al. developed Mn-doped InP/ZnS quantum dots via a growth-doping method for wLED design [17]. Budak et al. synthesized boron- and nitrogen-doped graphene QDs through a hydrothermal synthesis method and observed a controllable dual emission [18]. Again, Zhang et al. synthesized Cu:InP/ZnS through a growth-doping method and achieved dual emissive QDs with quantum yield around 75% [19]. Samuel et al. achieved to create surface-dependent emission in CdSe quantum dots through a surface reconstruction method and as a result increased the quantum yield of QDs by 7% [20]. Moreover, Samuel et al. explored the potential of CdSe QDs with surface defect emission for wLED design [21]. However, so far, defect states on CdSe QDs have never been created through two-phase synthesis method, the method which takes place at ambient conditions and results in high quantum yield [3,4,22–24].</p><p>In this study, we synthesized CdSe quantum dots with high defect-state emission through a modified two-phase synthesis method. The synthesis was carried under ambient pressure with an inert gas atmosphere at mild temperature. The CdSe QDs synthesized in this work possessed 2 emission bands; one emission band belonged to band-edge emission and the other belonged to defect-state emission. The defect-state emission disappeared as the CdSe QDs were covered with CdS layer. The defect-state emission could be triggered through excitation of surfactant of CdSe quantum dot. This study is the first to examine defect-state emission in CdSe QDs which are synthesized through a two-phase synthesis method and correlates the excitation of defect-state emission directly to the surfactant, tri octylphosphine oxide.</p><!><p>All chemicals that were used in this work were of the highest purity and were purchased from the Sigma-Aldrich Co. They were used without further purification.</p><!><p>Cadmium myristate was synthesized through a general procedure reported previously in the literature [3,4]. Firstly, 1.28 g (10 mmoles) of cadmium oxide (CdO) was mixed with 4.56 g (20 mmoles) of myristic acid and then the mixture was heated at 210 °C for 15 min under inert gas (nitrogen) atmosphere. Formation of cadmium myristate (CdMA) was identified as brownish CdO–myristic acid mixture became a colorless liquid and bubbling out of O2(g), which is the by-product, stopped. CdMA was purified through recrystallization in toluene, and obtained CdMA solid was dried and stored at 4 °C for further experiments.</p><!><p>Sodium hydrogen selenide (NaHSe) was synthesized through a general procedure reported previously in the literature [3,4]. Firstly, 10 mg of Se was put in a 100-mL reaction flask, sealed with a plastic plug and nitrogen gas was purged into a sealed reaction flask to remove oxygen completely. Then, 15 mg of sodium borohydride (NaBH4) was dissolved in 10 mL of ultrapure water and bubbled with nitrogen gas. NaBH4 solution was then mixed with Se under an inert gas atmosphere. The mixture was heated to 60 °C to speed up the reaction. The reaction was stopped as Se solid completely disappeared in solution and bubbling out of H2(g), which is the by-product, finished. NaHSe was prepared freshly for each experiment.</p><!><p>The CdSe quantum dots were obtained through a modified two-phase synthesis method [3,4]. Firstly, 0.1 g of CdMA and 1.5 g of trioctyl phosphineoxide (TOPO) were mixed in 45 mL of toluene at 80 °C until they were completely dissolved under an inert gas atmosphere. After CdMA and TOPO were dissolved, reaction temperature was set to 100 °C and 42 mL of ultrapure water was added to the system and the two-phase reaction mixture was kept stirring for 5 min under an inert gas atmosphere to stabilize the temperature of the reaction system. Then, 3 mL of NaHSe solution (3 mg NaHSe in total) was added to the reaction system and CdSe quantum dot formation began. Aliquots were taken from toluene to monitor the emission properties of the quantum dots by using fluorescence spectrophotometer. The reaction was stopped by cooling the solution to the room temperature at the 3rd h after NaHSe injection. The quantum dots were purified through precipitation by addition of excess methanol to toluene. This synthesis procedure was repeated to check repeatability.</p><!><p>CdSxSe1-x quantum dots were synthesized through a general procedure proposed in the literature before. In short, the same procedure used in the synthesis of CdSe quantum dots was performed with a slight modification; 60 mg of thiourea dissolved in 42 mL of ultrapure water was added together with 3 mL of NaHSe to reaction system in order to start growth of CdSxSe1-x.</p><!><p>Optical properties of CdSe and CdSxSe1-x quantum dots were checked using an ultraviolet–visible (UV–Vis) spectrophotometer and a fluorescence spectrophotometer. Absorption spectra of quantum dots were collected using a Scinco Neosys-2000 double-beam UV–Vis spectrophotometer. Emission and excitation spectra of nanocrystals were collected using a Varian Cary Eclipse fluorescence spectrofluorometer. Emission spectra of CdSe and CdSxSe1-x quantum dots were recorded by using excitation wavelength (λexc) 350 nm. Each sample was dissolved in 10 mL of toluene after purification and diluted by 10 factor before characterization in order to prevent errors resulting from self-absorption. The same samples were used to collect the excitation spectrum. Excitation spectra of CdSe quantum dots were collected at two wavelengths; 450 nm and 580 nm. Excitation spectra of CdSxSe1-x quantum dots were collected at 525 nm.</p><!><p>Both CdSe and CdSxSe1-x quantum dots were synthesized through a modified two-phase approach [3,4,25]. In traditional two-phase synthesis approach, Cd precursor is dissolved in the nonpolar phase (toluene) together with surfactant where chalcogen precursors are dissolved in the polar (aqueous) phase. The quantum dots form at the interface of the two phases and dissolve in the nonpolar phase. Compared to traditional solvothermal synthesis methods, the two-phase synthesis method offers more control on size of the quantum dots since the reaction time is lengthened out considerably. Generally, in the two-phase synthesis method, synthesis of CdSe quantum dots takes place in a sealed hydrothermal reaction vessel in order to minimize formation of defect states [22,23,26]. In 2013, Ünlü et al. proposed a modified two-phase approach to synthesize CdSxSe1-x quantum dots with heterogeneous–gradient structure in a 3-necked reaction vessel under nitrogen atmosphere without sealing and creating pressurized synthesis conditions [4]. By creating a heterogeneous–gradient structure, Ünlü et al. managed to synthesize core-shell like alloyed CdSxSe1-x quantum dots with very high quantum yield (up to 90%) by using oleic acid or TOPO as surfactant [4]. As a consequence, CdSxSe1-x quantum dots had a single very narrow emission band with almost no defect-state emission [4]. However, in Ünlü et al.'s study, it should be noted that using oleic acid instead of TOPO as surfactant brought out two important consequences: 1) As oleic acid was used as surfactant, the reaction slowed down dramatically, 2) The quantum yield of the CdSxSe1-x quantum dots dropped to 50% as TOPO was used as a surfactant [4]. Both of these outcomes showed that using TOPO as a surfactant yields an increase in formation of defect states [4]. In the current study, CdSe quantum dots were synthesized in a 3-necked reaction vessel without sealing and creating pressurized synthesis conditions by using TOPO as the surfactant. By doing so, CdSe quantum dots were formed with a high amount of defect state, which could be observed in the emission spectrum of CdSe quantum dots (Figures 1 and 2). The defect-state emission disappeared in the emission spectrum of CdSxSe1-x quantum dots (Figure 3) as a result of core-shell like gradient alloyed structure.</p><!><p>Absorption (Black) and emission spectra (Red, λexc= 350 nm) of CdSe QDs. The intensity of absorption spectrum is normalized to 1 with respect to the intensity at 380 nm. The intensity of emission spectrum is normalized to 1 with respect to the intensity at 460 nm.</p><p>Emission spectrum of CdSe QDs with defect states at 1 h (black) and at 3 h (red). The intensity of each spectrum was normalized to 1 with respect to the intensity of the first peak.</p><p>Emission spectrum of CdSe (Red) and CdSxSe1-x (Black) QDs with λexc=350 nm. The intensity of each spectrum was normalized to 1 with respect to the intensity of the first peak.</p><!><p>Absorption spectrum of CdSe quantum dots displayed typical features of CdSe quantum dots (Figure 1) [3,4,25]. CdSe absorption spectrum had 2 shoulders at 390 nm and 420 nm (Figure 1) due to formation of CdSe quantum dots, a typical feature of the absorbance spectrum of CdSe quantum dots [3,4,25]. As another typical feature of quantum dot absorption, CdSe absorption spectrum possessed an intensive band below 300 nm (Figure 1) [3,4,25]. Emission spectrum of CdSe quantum dots had two intense bands; one narrow band with a peak at 450 nm (band-edge emission) and one broad band with a peak at 580 nm (defect-state emission) (Figure 1).</p><p>The peak position of the band-edge emission band shifted towards the low energy region with increase in time, which is due to increase in size of quantum dots (Figure 2). The peak point of defect-state emission shifted towards the low energy region, too. However, the relative intensity of defect-state emission significantly decreased with increase in reaction time (Figure 2). As these results were compared with results of a previous work published by Ünlü et al., which was about controlling defect-state emission in Te-doped CdSe, the outcomes were completely different [3]. In the aforementioned work, CdSe quantum dots were doped with Te atoms in the presence of oleic acid as surfactant and defect-state emission was observed as a result [3]. The peak point of defect-state emission in Te-doped CdSe quantum dots was steady, did not shift by reaction time, and became more intense with increase in reaction time and with increase of Te ratio in synthesis medium [3], which showed that Te doping is the main reason behind formation of defect states in Te-doped CdSe quantum dots. As the results of current study were compared with those of Ünlü et al.'s work [3], it was observed that the origin of defect-state emission in TOPO-capped CdSe quantum dots was completely different from that of oleic-acid-capped Te-doped CdSe quantum dots.</p><p>In order to control the synthesis conditions of CdSe quantum dots, a parallel synthesis was performed with the same conditions to synthesize CdSxSe1-x quantum dots. The emission spectrum of CdSxSe1-x quantum dots had a single, narrow emission band with a peak point at 525 nm (Figure 3). The complete disappearance of defect-state emission was a result of core-shell like structure of CdSxSe1-x quantum dots [4]. As the rapidly formed CdSe-rich inner structure was covered with a slow growing CdS outer shell, the defect-state emission disappeared (Figure 3.). This result indicated that as the defected core is surrounded by an extra inorganic slow-growing shell, the amount of surface-based defects decreased; as a result, defect-state emission could no longer be observed (Figure 3).</p><!><p>In order to understand the difference between characteristics of defect-state emission and band-edge emission, photoluminescence excitation (PLE) spectra of CdSe quantum dots at 455 nm and 580 nm were collected (Figure 4). PLE spectra of CdSe quantum dots at 455 nm and 580 nm had both similarities and differences. Both spectra had 3 peaks at 420 nm, 390 nm, and 330 nm. The peaks at 390 nm and 420 nm were also observed in the absorbance spectrum of CdSe quantum dots, which shows that both band-edge emission and defect-state emission can be excited at these wavelengths. The peak at 330 nm in PLE spectra of CdSe quantum dots at 455 nm and 580 nm revealed that both band-edge and defect-state emission could be excited with similar wavelengths, because all of these excitation bands belonged to the same structure, the CdSe quantum dots. The peak at 290 nm in PLE spectrum of CdSe quantum dots at 580 nm is due to harmonic wave of 580 nm emission and did not occur due to emission properties of CdSe quantum dots, but occurred due to emission wavelength which was chosen by user [3,4,25]. However, the peak at 260 nm in the PLE spectrum of CdSe quantum dots at 580 nm completely disappeared in the PLE spectrum of CdSe quantum dots at 455 nm (Figure 4). In the absorbance spectrum of CdSe quantum dots, the peak at 260 nm was very intense, as was observed in different types of quantum dots in the literature as a common feature [3,4,25]. However, this peak does not contribute to fluorescence of quantum dots and is attributed to nonradiative electron transitions and background materials such as surfactants [3,4,25]. According to the results shown in Figure 4, the band-edge emission could not be excited by 260 nm; however, the defect-state emission could be excited. This result indicated that the defect-state emission in CdSe quantum dots was derived from the nanocrystal structure, but defect-state emission could also be obtained through excitation of surfactant which covered the inorganic CdSe core. The defect states on the surface of CdSe quantum dot are the main factor that clarifies the defect-state emission of the CdSe quantum dots.</p><p>To understand the effect of surface defects on defect-state emission, CdSxSe1-x quantum dots were synthesized. As was discussed before, alloyed CdSxSe1-x quantum dots have heterogeneous–gradient structure with CdSe rich inner core covered with CdS rich outer shell [4]. The slow growth rate of CdS-rich outer shell minimizes the amount of surface defects; as a result, quantum dots with higher quantum yield and narrower band edge emission were obtained [4]. As the PLE spectra of CdSxSe1-x quantum dots and CdSe quantum dots were compared, it was observed that there was no excitation peak in PLE spectrum of CdSxSe1-x quantum dots at emission peak (525 nm) which was located in between band-edge emission (455 nm) and defect-state emission (580 nm) peaks of CdSe quantum dots (Figure 5). This result revealed that as the amount of defects on the surface of CdSe quantum dots were minimized, the defect-state emission disappeared.</p><!><p>CdSe quantum dots are the most studied quantum dots with their unique optical properties. However, most of the research on CdSe quantum dots focuses on band-edge emission properties of CdSe quantum dots. Here in this paper, we managed to synthesize CdSe quantum dots with two intense emission bands, band-edge emission and defect-state emission, through two-phase synthesis methods. Due to the relatively slow growth rate in the two-phase synthesis method, we were able to control the relative intensity of defect-state emission in CdSe quantum dots. Emission and PLE spectra of CdSe quantum dots revealed that the defect-state emission of CdSe quantum dots can be excited separately from band-edge emission and the origin of defect-state emission in TOPO-capped CdSe quantum dots solely depends on the defects on the surface of CdSe quantum dots.</p>

PubMed Open Access

Selective Hydrogenation of 5-Hydroxymethylfurfural to 2,5-Dimethylfuran Over Popcorn-Like Nitrogen-Doped Carbon-Confined CuCo Bimetallic Catalyst

A new type of biomass-based liquid fuel, 2,5-dimethylfuran (DMF), has attracted significant attention owing to its unique physical properties and carbon neutrality. It can be obtained from the hydrogenation of 5-hydroxymethylfurfural (HMF), an important biomass platform compound. In this study, we developed a nitrogen-doped carbon-confined CuCo bimetallic catalyst with a popcorn-like structure for the selective hydrogenation of HMF with high efficiency and adequate stability. Under optimized conditions, 100% HMF conversion and 93.7% DMF selectivity were achieved. The structure of the catalyst was characterized using XRD, XPS, SEM, and TEM. It was observed that carbon spheres, which were covered by nitrogen-doped carbon nanotubes, uniformly formed, while metal particles were confined in the nitrogen-doped carbon nanotubes. The popcorn-like structure exhibited a larger surface area and provided more contact sites, while the confined metal particles were the main active sites. The synergistic effect between Cu and Co was beneficial for DMF selectivity.

selective_hydrogenation_of_5-hydroxymethylfurfural_to_2,5-dimethylfuran_over_popcorn-like_nitrogen-d

Introduction<!>Chemicals<!>Preparation of the Catalyst<!>Characterization<!>Test of Catalytic Activity<!>Structural Analysis<!><!>Structural Analysis<!><!>Structural Analysis<!><!>Hydrogenolysis of 5-Hydroxymethylfurfural to 2,5-Dimethylfuran<!><!>Hydrogenolysis of 5-Hydroxymethylfurfural to 2,5-Dimethylfuran<!><!>Hydrogenolysis of 5-Hydroxymethylfurfural to 2,5-Dimethylfuran<!><!>Hydrogenolysis of 5-Hydroxymethylfurfural to 2,5-Dimethylfuran<!><!>Reaction Pathway<!><!>Reaction Pathway<!><!>Reaction Pathway<!>Conclusion<!>Data Availability Statement<!>Author Contributions<!>Funding<!>Conflict of Interest<!>Publisher’s Note<!>Supplementary Material<!>

<p>With the increasing consumption of fossil fuels, the problems of resource depletion and the environmental pollution caused by fossil fuels have become increasingly serious. Therefore, there is an urgent need to develop sustainable and clean energy sources. Among them, biomass energy has attracted the attention of researchers as it can realize the carbon cycle and effectively reduce environmental pollution. Notably, 2,5-dimethylfuran (DMF) is considered a high-quality biomass-based liquid fuel owing to its unique physical properties (Tian et al., 2010; Alamillo et al., 2012; Wang H et al., 2019), such as high energy density, high boiling point, and low solvency in water (compared to first-generation bioethanol), and it can be produced by the selective hydrogenation of 5-hydroxymethylfurfural (HMF). HMF is an important biomass-based platform compound obtained from the hydrolysis of cellulose (Hu et al., 2008; Guo et al., 2020). However, because of its abundant reactive functional groups, side reactions and overhydrogenation easily occur during the hydrogenation of HMF to DMF (Zhang et al., 2017; Mishra et al., 2020; Wang et al., 2021). Developing highly efficient catalysts for the selective hydrogenation of HMF to DMF remains a challenge.</p><p>The current catalysts for the hydrogenation of HMF can be divided into two types: noble and non-noble metal catalysts. The most important feature of precious metals such as Pd, Pt, Ru, and Rh, is that they can achieve a high yield under mild conditions. Wang et al. (2018) obtained >90% DMF yields using Pt-Co bimetallic catalysts, while Zu et al. (2014) obtained 93.4% DMF yields using Ru/Co3O4 catalysts, and Zhang J et al. (2019) obtained 89.7% DMF yields using PdCl2 at room temperature. However, noble metal catalysts are limited by their scarcity and high cost. Therefore, there is an urgent need to develop low-cost, non-noble metal catalysts. Commonly used non-noble metals are Co, Cu, Ni, and Fe. Yang et al. (2019) obtained a 94.1% DMF yield using a synthetic Co/rGO catalyst at 200°C. Akmaz et al. (2019) prepared a Mn/Co bimetallic catalyst and obtained a DMF yield of 91.8%. Zhang Z et al. (2019) prepared a Co–CoOx catalyst and achieved a DMF yield of 83.3% after 12 h at 170°C. Generally, copper and cobalt are favorable for hydrogenation of C=O, C–O bond, bimetallic catalysts are more active than monometallic catalysts (Guo et al., 2021; Zhao et al., 2022). Although non-noble metal catalysts reduce the cost, they also introduce the disadvantages of requiring harsh reaction conditions and easy deactivation. Thus, the development of an efficient and stable non-noble metal catalyst is crucial. In addition, nitrogen-doped carbon materials have been widely used as catalyst supporter which can anchor and stabilize metal nanoparticles and promote electron transfer to improve the performance of catalysts (Wang et al., 2020; Guo et al., 2021). Based on the aforementioned information, in this study, a nitrogen-doped carbon-confined copper-cobalt bimetallic catalyst was synthesized using a two-step solvothermal-reducing calcination method. It can be observed that the catalyst has a popcorn-like structure, and the surface is evenly covered by carbon nanotubes, which provides more surface area for the contact between the catalyst and the substrate. It has been used in the selective hydrogenation of HMF to DMF and has achieved adequate results. The main paragraph text follows directly on here.</p><!><p>Co (NO3)2•6H2O, 2-methylimidazole, 5-hydroxymethylfurfural (97.0%), 5-methylfurfural, 2-butanol, urea, and melamine were procured from Shanghai Macklin Biochemical Co. Cu (NO3)2•3H2O, methanol, tetrahydrofuran (THF), and isopropanol (IPA) were procured from Guangzhou Chemical Reagent Factory. Glucose was procured from Tianjin Zhiyuan Chemical Reagent Co. Ltd.</p><!><p>Copper-cobalt bimetallic catalysts were prepared using a solvothermal-reducing calcination method with 2-methylimidazole as the nitrogen sources and both 2-methylimidazole and glucose as the carbon sources. 3.69 g 2-methylimidazole, 0.5 g glucose, cobalt nitrate hexahydrate and copper nitrate trihydrate with different Cu/Co molar ratios (the amount of Co remains constant) were dissolved in 60 ml of methanol and stirred for 10 min to facilitate dissolution. This solution was then transferred to a 100 ml Teflon reactor and maintained at 120°C for 12 h. After cooling to room temperature, the solution was filtered five times with methanol and transferred to a vacuum drying oven, where it was allowed to stay overnight at 80°C. Subsequently, the obtained brown precursor powder was calcined in a tube furnace under a reducing H2 (5%)/N2 atmosphere at a rate of 2°C/min to 440°C for 8 h and then at a rate of 2°C/min to 900°C for 2 h to obtain the target Cu–Co bimetallic catalyst. The prepared catalyst was named xCuCo–IG (where x represents the molar ratio of copper to cobalt, I represents the nitrogen source 2-methylimidazole, G represents the carbon source glucose). The monometallic catalysts were named Cu-IG and Co-IG, respectively.</p><p>For comparison, catalysts with melamine or urea as the nitrogen source were prepared and named 2CuCo–MG and 2CuCo–UG, respectively. A catalyst without a nitrogen source (2CuCo–G) and a catalyst without glucose (2CuCo–I) were also prepared. The catalyst calcined in a N2 atmosphere was named 2CuCo–IG (N2).</p><!><p>The X-ray diffraction (XRD) patterns of the powder samples were recorded using a BRUKER D8 ADVANCE diffractometer. X-ray photoelectron spectroscopy (XPS) data was measured using a Thermo ESCALAB 250Xi spectrometer. The specific surface area and porosity of the samples were obtained using a Micrometrics ASAP2460. The morphology of each sample was investigated using field emission scanning electron microscopy (SEM, SU8020), and the element mapping was performed using energy dispersive spectrometer (EDS). Transmission electron microscopy (TEM) and high-resolution transmission electron microscopy (HRTEM) images were obtained using a FEI Tecnai G2 F30. The Raman spectral profile was obtained using a Renishaw inVia at an excitation wavelength of 532 nm.</p><!><p>The catalytic performance in the hydrogenation of HMF was investigated in a stainless-steel autoclave. First, 0.2 g of the substrate (HMF), 0.02 g of the catalyst, and 0.05 g of toluene were dissolved in 30 ml of 2-butanol and subsequently poured into the reactor. After installation, hydrogen was purged at least eight times to remove air and charged with H2 at the corresponding pressure; the agitation speed was modulated to 400 rpm, followed by an increase in the temperature to the target temperature of 180°C. After the reaction was completed and cooled to room temperature, the reaction products were analyzed using the internal standard curve method to determine their conversion and selectivity with an Agilent 6820 gas chromatograph. The calculations for the conversion of HMF and DMF selectivity were performed as follows: Conversion(HMF)=(1−ntn0)×100%, Selectivity=niConv.HMF×100%, where nt represents the molar amount of HMF after the reaction, n0 represents the molar amount of initial HMF, and ni represents the molar amount of the product after the reaction.</p><!><p>First, the 2CuCo–IG catalyst precursor with a Cu/Co molar ratio of 2:1 was synthesized using a one-pot solvothermal reaction of metal salts, glucose (as the carbon source), and 2-methylimidazolem (as the nitrogen source and carbon source). Precursors mainly showed spherical structures with non-smooth surfaces (Figure 1A). Soon after the calcination under a H2 (5%)/N2 mixture atmosphere, the catalyst exhibited a popcorn-like structure with carbon balls uniformly covered by carbon nanotubes (Figure 1B). TEM and HRTEM images (Figures 1C,D) indicated that carbon balls were covered by carbon nanotubes, and Cu–Co metal particles were confined to the tips of the nanotubes with an average size of 6–8 nm. Figure 1D shows a lattice fringe of 0.209 nm attributable to the plane of Cu (111) (Liu J et al., 2020; Viar et al., 2020), and a lattice distance of 0.203 nm attributable to the plane of metallic Co (111) (Ma et al., 2020). In addition, the SEM and elemental images of Cu, Co, C in the 2CuCo–IG catalyst showed that copper and cobalt were uniformly dispersed throughout the catalyst (Figures 1E–H).</p><!><p>(A) SEM profiles of the precursor, (B) 2CuCo–IG obtained after the calcination of H2 (5%)/N2; (C,D) TEM and HRTEM profiles of 2CuCo-IG; (E–H) the SEM and elemental images of the catalyst 2CuCo–IG.</p><!><p>To investigate the reason for the formation of carbon nanotubes, we prepared catalysts such as 2CuCo–G (no nitrogen source added), 2CuCo–UG (nitrogen source replaced by urea), 2CuCo–MG (nitrogen source replaced by melamine), and 2CuCo–I (no glucose carbon source added). As shown in Supplementary Figures S2A–D, it can be observed that when the nitrogen source was changed, the catalyst structure also changes and no longer develops a popcorn-like structure. When no glucose carbon source was added, no spherical support structure was formed, but the carbon nanotubes still appeared (Supplementary Figure S2D), which indicates that carbon nanotubes are most likely formed during the calcination of 2-methylimidazole. The effect of the calcination atmosphere on the morphology of the catalyst was also investigated (Supplementary Figure S1). The carbon nanotubes on the catalyst calcined under a N2 atmosphere are fewer and significantly finer than those formed under a mixed atmosphere. This means that both the formation and morphologies of carbon nanotubes are highly dependent on the nitrogen source and the calcination atmosphere.</p><p>A series of copper-cobalt bimetallic catalysts was prepared with different Cu/Co ratios, and their XRD patterns are shown in Figure 2. The diffraction peaks at 43.3°, 50.4°, and 74.1° are attributed to metallic copper (PDF #89-4307) (Chen et al., 2020; Rao et al., 2020; Xu et al., 2021), and those at 44.2°, 51.5°, and 75.9° are attributed to metallic cobalt (PDF #89-2838) (Chen et al., 2017; Solanki and Rode, 2019). It can be observed that mono metal catalyst Cu–IG contains only metallic copper diffraction peak, while Co–IG catalyst contains only metallic cobalt diffraction peak. The CuCo bimetallic catalyst contains both metallic copper and metallic cobalt diffraction peaks, with an increase in the ratio of copper to cobalt, the diffraction peaks of both elements exhibited different variation trends. The diffraction peaks of metallic copper continuously increased as the proportion of metallic copper increased. The diffraction peaks of metallic cobalt also increased initially. However, when this ratio was exceeded 2.5, the diffraction peaks of metallic cobalt decreased. This shows that a suitable ratio of copper to cobalt can improve the crystal structure of metallic cobalt.</p><!><p>XRD patterns of the copper–cobalt bimetallic catalysts with different Cu/Co ratios.</p><!><p>The N2 adsorption/desorption curves of the catalysts are shown in Supplementary Figure S3. The existence of hysteresis loops indicated that the prepared catalysts were typical mesoporous materials. The specific surface area and pore size data are listed in Supplementary Table S1. Among the catalysts with different Cu/Co ratios, the 2CuCo–IG catalyst had the largest specific surface area, and the large pore volume and the adequate pore diameter improved contact, significantly improving the utilization of the catalyst.</p><p>The XPS profiles of the 2CuCo–IG catalyst are shown in Figure 3. Two peaks are visible at 284.8 and 286.0 eV in the high-resolution XPS spectrum of C 1s mainly corresponding to C–C and C–O–C (Xia et al., 2016; Rao et al., 2021). The peak near 289.0 eV is attributed to C–N in the catalyst (Figure 3A) (Liu L et al., 2020; Zhu K et al., 2020). Two peaks were visible at 398.6 and 401.2 eV in the high-resolution XPS spectrum of N 1s corresponding to pyridine N and graphite N (Figure 3B) (Hu et al., 2019). The elemental nitrogen content was determined to be 2.5% using XPS, including 15.0% pyridine nitrogen and 85.0% graphite nitrogen. The high-resolution XPS spectrum of Co 2p shows the characteristic peak of Co0 at 778.3 eV (Liu M et al., 2020), while the peak at 780.3 eV corresponds to CoOx (Figure 3C) (Zhu J et al., 2020). The diffraction peaks of Cu 2p1/2 and Cu 2p3/2 in the high-resolution XPS spectrum of Cu 2p were present at 952.3 and 932.3 eV; this indicates that the valence state of Cu is in the metallic phase (Figure 3D) (Kang et al., 2016), which is consistent with XRD results. The two peaks at 1350 and 1600 cm−1 in the Raman spectrum (Supplementary Figure S4) are the D and G bands, respectively (ID/IG = 0.9), indicating a relatively high degree of graphitization of carbon. It should be noted that high contents of graphitic carbon are conducive to the catalytic hydrogenation reaction (Gong et al., 2019).</p><!><p>XPS spectra of the catalyst 2CuCo–IG. (A) High-resolution spectra of C 1s, (B) N 1s, (C) Co 2p, and (D) Cu 2p.</p><!><p>The effects of the different carbon/nitrogen sources and calcination atmospheres on the catalyst activity were also investigated (Figure 4). Both the catalyst without the addition of a glucose carbon source (2CuCo–I) and that without the addition of 2-methylimidazole as a nitrogen source (2CuCo–G) exhibited very low DMF selectivity. When the nitrogen source was changed to urea or melamine, the selectivity of DMF was only 35.4 and 12.2%, respectively. This implies that both carbon and nitrogen sources are indispensable, and that the type of nitrogen source also has a significant impact on DMF selectivity. The 2CuCo–IG (N2) calcined in a N2 atmosphere showed 82.3% DMF selectivity which is lower than that of the 2CuCo–IG calcined in a H2 (5%)/N2 atmosphere. This may be because the reducing atmosphere is beneficial for formation and dispersion of metallic CuCo particles. As showed in Supplementary Figure S1, more nitrogen doped carbon nanotubes in which CuCo bimetal particles were confined were formed in the 2CuCo–IG catalyst because of lower depletion of carbon under reducing atmosphere than inert atmosphere. The calcination atmosphere is also benefit for the thorough reduction of Cu although the carbon also can partially reduction of Cu. On other hand, The ratio of Co0/CoOx in the 2CuCo–IG is higher than that in 2CuCo–IG (N2) although both of 2CuCo–IG and 2CuCo–IG (N2) catalyst contain metallic Co and oxidation state Co (Figure 3 and Supplementary Figure S5).</p><!><p>Effect of the different carbon/nitrogen sources and calcination atmospheres on catalyst performance. Reaction conditions: 0.2 g of HMF, 0.04 g of the catalyst, 30 ml of 2-Butanol, 2 MPa H2, 4 h, and an agitation speed of 400 rpm.</p><!><p>The effects of the different Cu/Co ratios on the catalyst activity are shown in Figure 5. HMF conversion was 79.8% and DMF selectivity was only 5.3% when a mono-copper metal was used as the catalyst, whereas 100% HMF conversion and 66.0% DMF selectivity were achieved when a mono-cobalt metal was used as the catalyst. All the CuCo bimetallic catalyst showed higher conversion and DMF selectivity than monometallic catalysts. The best proportion was obtained when the copper-to-cobalt ratio was 2, and a 93.7% DMF yield was achieved. By comparison, we speculated that there will be a synergistic effect between Cu and Co which played an important role in improving the catalytic performance.</p><!><p>Effects of different copper-cobalt ratios on HMF conversion and the DMF yield. Reaction conditions: 0.2 g of HMF, 0.04 g of the catalyst, 30 ml of 2-Butanol, 2 MPa H2, 4 h, and an agitation speed of 400 rpm.</p><!><p>The effects of the different reaction conditions on the catalytic activity of the selective hydrogenation of HMF to DMF were investigated (Figure 6). The selectivity of DMF increased when the temperature gradually increased, and the highest selectivity (93.7%) was obtained at 180°C. Further increasing the temperature will result in the deep hydrogenation of DMF in the C=C bond of the furan ring to form 2,5-Dimethyloxolane (DMTHF), resulting in a decrease in DMF selectivity (Figure 6A). The effect of the reaction time is comparable to that of the temperature (Figure 6B), and it shows the highest selectivity at 4 h. For the effect of the pressure factor, DMF selectivity rapidly increased when the pressure increased from 1 to 2 MPa; however, excessive pressure can also lead to overhydrogenation (Figure 6C). The effect of the reaction solvent on the activity is shown in Figure 6D. The use of different solvents caused dramatic changes in the activity level, which indicated that the solvent was present in the reaction. The effect of the solvent will be discussed in the following section. Overall, the catalyst showed the best performance when 2-butanol was used as the solvent.</p><!><p>Catalytic performance of HMF hydrogenolysis and distribution of products on the 2CuCo–IG catalyst over different reaction conditions. (A) 4 h, 2.0 MPa H2, 30 ml of 2-Butanol; (B) 180°C, 2.0 MPa H2, 30 ml of 2-Butanol; (C) 4 h, 180°C, 30 ml of 2-Butanol; (D) 4 h, 180°C, 2.0 MPa H2, 30 ml of 2-Butanol.</p><!><p>The cycling performance of the catalyst is shown in Figure 7. Since the catalyst itself is magnetic, it can be recovered easily after the reaction. After five cycles, no distinct decrease was observed in either HMF conversion or DMF selectivity. This indicates that the catalyst is highly stable and adequately reusable.</p><!><p>Catalyst cycling performance test. Reaction conditions: 0.2 g of HMF, 0.04 g of the catalyst, 30 ml of 2-Butanol, 2 MPa H2, 4 h, and an agitation speed of 400 rpm.</p><!><p>The hydrogenation of HMF to DMF is divided into two main pathways: routes I and II (Figure 8) (Nakagawa et al., 2013; Qian et al., 2015; Wang Q et al., 2019). Route I is the hydrogenation and dehydration of HMF to produce MF, by further hydrogenation to produce MFA and finally DMF. Route II is the hydrogenation of HMF to produce DHMF, followed by hydrogenation and dehydration to produce MFA and finally DMF. DMF could also be transformed into DMTHF by the C=C hydrogenation in the furan ring. The results of the time optimization experiment (Figure 6B) showed that the intermediate product that appeared at 1 h was DHMF, indicating that the reaction process was route II.</p><!><p>Reaction pathways of the selective hydrogenation of HMF to DMF.</p><!><p>To further study the effect of the solvent on product selectivity and the role of Cu/Co at each step of the hydrogenation reaction, a series of hydrogenation reactions was designed using MF with an aldehyde group and a furan ring and MFA with a hydroxyl group and a furan ring as substrates (Tables 1, 2). When methanol was used as the solvent, the selectivity of DMF was lower than that when 2-butanol was used, in both the hydrogenation of MF and MFA. Furthermore, when methanol was used, more etherification occurred, and more condensation products were formed. This may be due to the steric effects of 2-butanol that suppressed etherification and condensation reactions and resulted in the target product DMF.</p><!><p>Effects of the different reaction conditions on MF conversion and products.</p><p>Reaction conditions: 0.2 g of MF, 0.02 g of the catalyst, 140°C, 2 h, 2 MPa corresponding gas, 30 ml of the corresponding solvent, and an agitation speed of 400 rpm.</p><p>Effects of the different reaction conditions on MFA conversion and products.</p><p>Reaction conditions: 0.2 g of MFA, 0.02 g of the catalyst, 140°C, 2 h, 2 MPa corresponding gas, 30 ml of the corresponding solvent, and an agitation speed of 400 rpm.</p><!><p>The effects of the different Cu/Co ratios on the reaction process were investigated (Tables 1, 2). A series of hydrogenation reactions was designed using MF with an aldehyde group and a furan ring and MFA with a hydroxyl group and a furan ring as substrates. In the one and four lines of Table 1, it was observed that the higher the copper content, the higher the MF conversion and MFA selectivity in the hydrogenation reaction of MF, which indicates that metallic copper is active for the hydrogenation of C=O bonds. Meanwhile, the higher the cobalt content, the higher the MFA conversion and DMF selectivity in the hydrogenation reaction of MFA, which indicates that Co/CoOx is more active for the hydrogenation of C–O bonds. The synergistic effect of copper and cobalt promotes the whole hydrogenation process.</p><!><p>In summary, a popcorn-like nitrogen-doped carbon-confined CuCo bimetallic catalyst was prepared using a two-step solvothermal-reducing calcination method. The 2CuCo–IG catalyst performed well in the HMF selective hydrogenation to DMF with an HMF conversion of 100% and a DMF yield of 93.7%. The popcorn-like structure provided more active sites and electrons, and the confinement effect of nitrogen-doped carbon nanotubes and the synergistic effect of copper and cobalt were the main reasons for the high catalytic efficiency. The 2-butanol solvent not only provided hydrogen but also reduced the unwanted reactions of etherification and condensation using steric effects during the reaction. Meanwhile, the catalyst exhibited adequate recycling performance; thus, it can be reused.</p><!><p>The original contributions presented in the study are included in the article/Supplementary Material, further inquiries can be directed to the corresponding authors.</p><!><p>Conceptualization and methodology: JZ and ZL. Validation and investigation: PH and WT. Supervision: JZ and ZL. Writing-original draft: PH. Writing-review and editing, JZ, QW, JL, and ZL. All authors approved the submitted version.</p><!><p>This work was supported by the National Natural Science Foundation of China (Grant numbers 22005070, 22078077 and 21676060) and the Natural Science Foundation of Guangdong Province (Grant number 2021A1515010078), and the Scientific and Technological Plan of Guangdong Province, China (Grant number 2019B090905007). The work is also supported by Qingyuan Huayuan Institute of Science and Technology Collaborative Innovation Co. Ltd.</p><!><p>This study received funding from Qingyuan Huayuan Institute of Science and Technology Collaborative Innovation Co. Ltd. The funder was not involved in the study design, collection, analysis, interpretation of data, the writing of this article or the decision to submit it for publication.</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><!><p>The Supplementary Material for this article can be found online at: https://www.frontiersin.org/articles/10.3389/fchem.2022.882670/full#supplementary-material</p><!><p>Click here for additional data file.</p>

PubMed Open Access

Data-driven discovery of cardiolipin-selective small molecules by computational active learning

Subtle variations in the lipid composition of mitochondrial membranes can have a profound impact on mitochondrial function. The inner mitochondrial membrane contains the phospholipid cardiolipin, which has been demonstrated to act as a biomarker for a number of diverse pathologies. Small molecule dyes capable of selectively partitioning into cardiolipin membranes enable visualization and quantification of the cardiolipin content. Here we present a data-driven approach that combines a deep learning-enabled active learning workflow with coarse-grained molecular dynamics simulations and alchemical free energy calculations to discover small organic compounds able to selectively permeate cardiolipincontaining membranes. By employing transferable coarse-grained models we efficiently navigate the allatom design space corresponding to small organic molecules with molecular weight less than z500 Da.After direct simulation of only 0.42% of our coarse-grained search space we identify molecules with considerably increased levels of cardiolipin selectivity compared to a widely used cardiolipin probe 10-N-nonyl acridine orange. Our accumulated simulation data enables us to derive interpretable design rules linking coarse-grained structure to cardiolipin selectivity. The findings are corroborated by fluorescence anisotropy measurements of two compounds conforming to our defined design rules. Our findings highlight the potential of coarse-grained representations and multiscale modelling for materials discovery and design.

data-driven_discovery_of_cardiolipin-selective_small_molecules_by_computational_active_learning

Introduction<!>Overview<!>Coarse-grained molecular design space<!>Coarse-grained molecular dynamics (CGMD) simulations<!>Alchemical free energy calculations<!>Chemical space embedding<!>Active learning<!>Inference of design rules<!>Functional group analysis<!>Active learning identies highly cardiolipin-selective compounds<!>Data-driven discovery of chemical design rules<!>All-atom backmapping<!>Validating simulations against experimental measurements<!>Conclusions

<p>Mitochondria are double-membrane-bound organelles found in most eukaryotic cells (Fig. 1a). Their principal function is to generate the cell's supply of adenosine triphosphate (ATP), the main source of chemical energy used to drive biochemical reactions. 1 Through their key role in bioenergetics, mitochondria play an essential part in controlling cell proliferation, and they are involved in cell signaling and the activation of apoptosis. [2][3][4] The inner mitochondrial (IMM) membrane partitions the intermembrane space from the matrix and contains the membrane-bound ATP synthase proteins responsible for ATP generation. The composition of this membrane includes approximately 20% cardiolipin (CL) (Fig. 1b), a phospholipid comprising four acyl chains that is exclusively found in energy-generating membranes. 1,5 Abnormalities in CL composition of inner mitochondrial membranes are linked to pathologies including Barth syndrome, Tangier disease, heart failure, and neurodegeneration. 5,6 The CL content of a membrane can be experimentally assayed using molecular stains to visualize and quantify the presence of CL. A primary challenge in the molecular design of cardiolipin-selective small molecules 7 is the high degree of structural similarity between CL and other common phospholipids. The most chemically similar phospholipid is phosphatidylglycerol (PG) (Fig. 1c), a precursor of which CL is essentially a dimer. [8][9][10] Apart from its role in the CL synthesis pathway, PG is prevalent in bacterial membranes and a minor constituent of eukaryotic membranes. 11 There are two principal differences in the CL and PG headgroups: CL possesses two chemically distinct phosphatidyl moieties whereas PG has only one, and PG possesses two hydroxyl groups while CL retains only one. The dianionic CL structure thereby stands out as the main difference to the monoanionic PG. A number of selective probes have been described for the purpose of detecting CL, [12][13][14][15][16][17] including the uorescent dye 10-N-nonyl acridine orange (NAO) (Fig. 1d), 12,13 but the degree to which these probes are optimally selective for CL relative to PG is not clear. 13,[18][19][20][21][22][23] Engineering selectivity into a molecular probe requires manipulation of the structure and physicochemical properties to stabilize favorable interactions with CL relative to other phospholipids, in particular, its chemically similar precursor PG. Due to the structural similarity of the two lipids, the design process must exploit the subtle differences in binding affinity mediated by the existence of two phosphate groups due to the four-acyl chain structure of CL relative to its two-acyl chain PG competitor. This motivates a design strategy targeting the lipid headgroups as the site of these distinguishing characteristics. The challenges posed by this subtle structural distinction are further compounded by the absence of a clear binding site-as might be anticipated in, for example, protein-ligand interactions-and the anticipated importance of inter-molecular, multi-body interactions between phospholipid chains within the membrane.</p><p>The objective of this work is to engage the design challenge posed by the subtle chemical differences between phospholipids and the lack of a dened binding site in lipid structures. To this end, we use a combination of a high-throughput virtual screening approach and data-driven active learning. The virtual screening is based on coarse-grained molecular dynamics simulations and alchemical free energy calculations, and the active learning employs deep representational learning using neural networks, Gaussian process regression surrogate models, and Bayesian optimization. As our objective function, we choose to maximize the difference DDG in partitioning free energies between a CL and a PG membrane environment to maximize the thermodynamic preference of the molecule for CL relative to PG. By performing our screening using coarsegrained modeling we can substantially reduce the size of our search space without sacricing coverage of chemical space. Because transferable coarse-grained models rely on a nite set of interaction types, many molecules map to the same coarsegrained representation. 25 The accuracy and transferability of our coarse-grained model enable us to tractably and efficiently engage the space of $10 60 small organic molecules with molecular weight up to 500 Da. 26 The multiscale-based many-toone mapping reduces to only 124 327 unique coarse-grained topologies that maintain the pertinent physicochemical properties. Within this reduced chemical space, we conduct seven rounds of iterative computational screening and surrogate model building linking molecular structure to thermodynamic properties. We discover 242 compounds with up to 184% superior predicted thermodynamic selectivity for CL membranes relative to NAO.</p><p>The discovery of high performing compounds aer sampling only 0.42% of the CG molecular design space exploits the power of the active learning protocol to efficiently guide sampling towards the most promising regions of chemical space to explore. Our molecular discovery platform also identies human-interpretable design principles that furnish both novel molecular designs and promote new understanding of the subtle physicochemical molecular properties that lead to high CL selectivity. Specically, post hoc analysis of the simulation data accumulated over the course of our active learning process enables us to identify chemical motifs that determine CL selectivity by building a sparsity-enforced structure-property regression model. We validate our computational model by performing experimental uorescence anisotropy measurements of CL selectivity for two molecules selected according to our learned design rules and observe good concordance with our computational predictions. Our computational discovery platform is generically transferable to other molecular engineering applications by modular substitution of the selectivity assay for other addressable properties of interest.</p><!><p>We develop and deploy a computational screening active learning cycle for the data-driven discovery of small molecule probes highly selective to CL membranes (Fig. 2). In a nutshell, we (i) conduct coarse-grained molecular dynamics (CGMD) simulations and alchemical free energy calculations to measure the thermodynamic preference DDG of a particular candidate molecule for a CL membrane relative to a PG membrane, (ii) build supervised regression models over a learned lowdimensional latent space to predict the performance of new compounds that have not yet been simulated, and (iii) apply Bayesian optimization to the trained regression models to identify the next most promising compounds to submit for computational screening. Aer running the active learning cycle for several rounds, we analyze the learned models in order to extract the physicochemical design rules underpinning the observed performance of the simulated candidates. We provide below the methodological details of each component of this computational workow. A more comprehensive description of the theoretical bases and numerical implementations of these approaches is provided in the ESI. †</p><!><p>The molecular design space for our computational screening is formed from organic molecules with molecular weights less than or equal to 500 Da. We impose this upper weight threshold in order to promote high mobility and diffusion into and through membranes. 7 No atom types are excluded from our design space. The number of molecules satisfying these criteria is $10 60 , 26 motivating the use of approximations and simplications to efficiently screen this subset of chemical space. To this end, we employ the coarse-grained (CG) Martini 2 molecular model, that both greatly reduces the cost of our molecular simulations and signicantly reduces the size of chemical compound space by grouping molecules into a smaller number of CG representations. [27][28][29] The Martini model was parameterized against thermodynamic data to generate 14 neutral and four charged CG bead types (14 + 4 bead-type model) representing most physicochemical interactions relevant in biomolecular settings. This building-block approach allows the rapid generation of new representations without requiring individual reparameterization and provides a good balance between chemical accuracy and computational efficiency. The Martini CG force eld has been widely used in the study of membrane organization and dynamics, [30][31][32][33] drugmembrane permeability, [34][35][36][37][38] and membrane-protein interactions. [39][40][41][42][43] More recently it has been shown that even coarser models, going as low as ve bead types, represent the underlying physical properties comparably well while facilitating more thorough coverage of chemical compound space by reducing the combinatorial complexity. 25 We created a 5 + 1 bead-type CG model through extending the ve bead-type 5 + 0 reduced Martini model 25 by one charged bead-type to fully represent the candidate small molecule probes. The additional charged bead-type represents single positive or negative charges (Q0AE). We offset this increase in complexity by removing the two non-polar bead types representing only hydrogen bond donor or acceptor properties (T3d, T3a), as the reduced Martini model already contains a nonpolar hydrogen bond donor-and acceptor bead-type (T3) that represents both interaction types simultaneously. The complete 5 + 1 model comprises the set of ve neutral bead types ordered by descending polarity and one charged bead type {T1, T2, T3, T4, T5, Q0AE} as illustrated in Fig. S1 in the ESI. † This reduced model allowed us to explore the chemical compound space more efficiently compared to the 14 + 4 interaction type scheme applied in Martini. 27 The combinatorial candidate space of compounds less than 500 Da was dened by constructing all plausible molecular graphs containing ve or fewer 5 + 1 beads, resulting in 124 327 candidate compounds. embedding of the discrete molecular design space encompassing all coarse-grained molecular candidates. Gaussian process regression (GPR) surrogate models are fit using all accumulated simulation data to predict the CL selectivity DDG of all untested compounds within the design space. (c) These surrogate model predictions are then interfaced with a Bayesian optimization platform to select the next most promising compounds for computational simulation. This process iteratively continues until multiple consecutive active learning rounds fail to identify new top-performing compounds. (d) Using our accumulated simulation data we extract design rules linking the inclusion/omission of chemical functional groups to the degree of CL selectivity DDG by building of interpretable linear models using graph representational learning. (e) The design rules are used to select candidate compounds that are subjected to fluorescence anisotropy measurements as well as CG free energy calculations to validate our findings. Structures in this panel are drawn with ChemSketch, 24 the icons were obtained from https://Flaticon.com.</p><!><p>We constructed CL and PG membranes comprising 98 and 118 lipid molecules, respectively, using the CHARMM-GUI Martini maker. 44 The membranes were solvated in 3287 water particles for CL and 1754 water particles for PG. A sufficient number of sodium ions (Na + ) to maintain charge neutrality were added: two Na + ions per CL headgroup since the double negative charged CL model was chosen, [45][46][47] and one Na + ion per PG headgroup. The CL and PG phospholipid membranes were represented using the Martini 2 force eld, 27 water using the rened parameters for polarizable water, 29 and ions using the polarizable ion model. 28 Coarse grained molecular dynamics (CGMD) simulations were conducted using GROMACS 2018.6 (ref. 48) implementing the standard Martini run parameters introduced for GPU acceleration. 49 Full details of the force eld and run parameters are reported in the ESI. †</p><!><p>We evaluate CG candidate molecules based on their thermodynamic affinity for a CL membrane environment relative to a PG membrane. It is our goal to maximize the affinity of the probe for CL and simultaneously minimize its affinity for PG in order to maximize CL selectivity. It is computationally intractable to consider all possible competing phospholipid environments in our molecular screen, so we adopt PG as the negative design target as the most chemically similar phospholipid to CL and therefore the most challenging target for negative design. We quantify selectivity by computing the relative partitioning free energy DDG for each candidate molecule between the transfer free energy from the water phase to the interface region of a PG membrane ðDG PG W/I Þ and the same transfer for a CL membrane ðDG CL W/I Þ;</p><p>We focus on the water-membrane interface since it is the water-facing lipid headgroups that are the site of the primary chemical difference between CL and PG, and therefore the region of interest for engineering target selectivity.</p><p>All free energies are calculated using CGMD alchemical free energy calculations, 50 using the MBAR method 51 with tools provided by the pymbar 52 package. Full details of the calculation procedure are reported in the ESI. † Calculation of DDG for a single candidate molecule requires approximately $24 GPU h for an uncharged molecule and $48 GPU h for a charged molecule on a single NVIDIA Tesla V100 GPU card. Exhaustive simulation of all 124 327 candidate molecules would therefore require $4.8 M GPU h conservatively assuming an average simulation time per molecule of $36 GPU h. To reduce the overall computational cost, we adopted a three-step hierarchy for the calculations that enables early exit for unviable candidate molecules. Only candidates that meet our minimum requirements of easily partitioning into and aligning with the interface region of the PG membrane are subjected to the most computationally expensive calculations in the CL membrane.</p><p>2.4.1. Assessment of interfacial preference in PG. In the rst step, we perform alchemical transformations to compute the free-energy difference DG PG of transferring the candidate molecule from a vacuum reference state to the interface of the PG membrane, (Fig. 2a and ESI Fig. S2(1) †). The cumulative probability of the positions of the compound along the membrane normal, z, is evaluated over the trajectory to determine whether a candidate molecule inserted into the membrane will preferentially position itself at the interface (ESI, Fig. S7(1) †). If it spontaneously moves astray toward the hydrophobic membrane midplane or water phase, we declare the candidate "non-interfacial." For computational efficiency, we assume that this determination can be made by considering only PG membranes so we do not repeat this calculation in CL. We eliminate non-interfacial candidates from further consideration since they do not preferentially reside near the phospholipid headgroups where CL and PG are most chemically differentiated.</p><p>2.4.2. Calculation of PG transfer free energies. For candidate structures deemed interfacial, the second step consists of alchemical free-energy calculations in bulk liquids: water (DG W ) and octane (DG O ). The two environments are used to evaluate the transfer free energies from vacuum into the water phase and the membrane midplane, respectively (Fig. 2a and ESI Fig. S2 and S7(2) †). In line with Menichetti et al., 36 2.4.3. Calculation of CL transfer free energies. If the previous bulk simulations conrm the interfacial nature of the compound, we perform a nal set of alchemical transformations. We compute the transfer free energy of the compound from vacuum to the interface of the CL membrane DG CL (ESI, Fig. S2 and S7(3) †). We then use this value to compute the CL transfer free energies DG CL W/I ¼ DG CL À DG W from the water phase to the interface and DG CL O/I ¼ DG CL À DG O from the midplane to the interface. We now have all the ingredients to quantify the compound's thermodynamic preference for the CL membrane relative to its PG counterpart, DDG ¼ DG CL W/I À DG PG W/I : Larger negative values of the transfer free-energy difference DDG are indicative of stronger thermodynamic selectivity for CL relative to PG.</p><!><p>Each of the 124 327 CG molecular candidates in the design space is represented as a molecular graph composed of up to ve beads selected from the possible six bead types and different charge states of the 5 + 1 model {T1, T2, T3, T4, T5, Q0AE}. 25 The nodes of the graph represent the identity of each bead in the molecule and the edges capture the connectivity of the beads within the CG topology. Performing molecular design and optimization directly over the discrete molecular design space has been demonstrated using kernel-based methods. [53][54][55][56][57][58][59] Here, we choose to project candidate molecules into a learned, smooth, low-dimensional, and continuous embedding. This approach renders our design space amenable to the construction of low-dimensional and robust surrogate models and the use of off-the-shelf Bayesian optimization algorithms. 60,61 We learn an appropriate latent space embedding in a data-driven fashion by training a regularized autoencoder (RAE), 62 a deterministic adaptation of the variational autoencoder (VAE) architecture, 63 over the corpus of 124 327 CG molecular graphs (Fig. 2b). The encoder-decoder architecture is composed of a message passing neural network encoder and a permutationinvariant graph decoder. [64][65][66][67][68] The network accepts graphstructured molecular representations where the nodes are featurized with one-hot representations of the bead type and the charge state. The edge between two nodes relies on the corresponding Lennard-Jones 6-12 interaction parameter. It is the objective of the RAE encoder to learn a smooth and continuous latent representation containing the salient information about the input graphs from which the decoder can accurately reconstruct (i.e., auto-encode) the same graph. We achieve good reconstruction performance employing a 16-dimensional bottleneck layer between the encoder and decoder that denes our latent space dimensionality and representation, and within which we dene measures of proximity between candidates and construct our surrogate models and perform global optimization via active learning. Although RAEs are generative models that when trained on open data sets are in principle capable of producing never-before-seen reconstructions, our model operates within a fully enumerated and xed molecular design space such that we do not require this generative functionality. Rather, upon completing training, the decoder is discarded and the encoder alone is purposed for dimensionality reduction mapping our 124 327 candidate design space into a continuous embedding for downstream application within our active learning workow. RAE models were constructed and trained using PyTorch. 69 The RAE was trained only once over all 124 327 candidate molecules prior to commencing active learning. Training required $36 GPU h on a single NVIDIA Tesla V100 GPU card. Full details of the network architecture and training are provided in the ESI. †</p><!><p>The primary goal of our active learning cycle is to efficiently discover CG compounds within the 124 327-member candidate space with high selectivity for CL membranes (i.e., large negative DDG). The high computational cost to measure the selectivity of each candidate precludes an exhaustive traversal of these candidates and we must employ a more computationally efficient search strategy. We achieve this by iterating between targeted CGMD simulations of promising candidates designed using a surrogate model relating molecular structure to CL selectivity DDG and the training and interrogation of these surrogate models constructed over the 16-dimensional RAE latent space. To ensure broad initial coverage of the molecular design space, we seed the rst round of active learning by conducting CGMD simulations of 100 compounds residing closest to the centroids of a 100-cluster k-means partitioning of the latent space. 70 We then trained a Gaussian process regression (GPR) 71 surrogate model with a Gaussian kernel to learn a mapping from the 16-dimensional latent space coordinates to the calculated DDG. The trained model was then applied to predict the DDG values for all remaining 124, 327 À 100 ¼ 124, 227 compounds in the design space along with the predicted model uncertainties. The predictions of the surrogate GPR model were then interfaced with a Bayesian Optimization (BO) 72 framework using the Expected Improvement (EI) 72,73 acquisition function and kriging believer batched sampling 74 to identify a 60-molecule batch of candidates with the most promising (i.e., lowest) values of DDG (Fig. 2c). Importantly, the EI acquisition function accounts for both the value of the GPR predictions and its estimated uncertainties, enabling the BO-directed active learning search to direct inquiry towards both regions of design space with favorable predictions (exploitation) and underexplored regions with high uncertainties (exploration). This batch of 60 compounds is then subjected to CGMD free-energy calculations to evaluate their associated DDG values and the cycle repeats. Each round of active learning involves updating the training dataset based on all accumulated simulation data, retting the GPR to predict out-of-training sample DDG selectivity, and BO selection of the most promising compounds to simulate next (Fig. 2a-c). Convergence of the active learning cycle is monitored by tracking the distribution of DDG values measured in each round, and the process is terminated when multiple consecutive rounds fail to identify new best performing candidates. As detailed below, we compute DDG values for a total of N ¼ 439 interfacial molecular candidates.</p><!><p>Aer completing the active learning screen, we analyzed the collected library of DDG values to extract human-interpretable design rules linking the presence or absence of particular CG structural motifs to the calculated values of the transfer free energy (Fig. 2d). We constructed these interpretable models by decomposing each CG topology as a distribution of relative subgraph frequencies and then performed sparse LASSO regression to rank the most impactful subgraph motifs. We begin by enumerating k ¼ 1608 topologically unique subgraphs with 1-5 CG beads that are contained within the N ¼ 439 interfacial compounds. Each CG topology is then featurized as a length-k vector of subgraph frequencies, reweighted to account for the over counting of smaller subgraphs necessarily contained within larger subgraphs, and normalized to unit length. 75 Assembling these featurizations into a normalized frequency matrix F ˛IR (N¼439)Â(k¼1608) we adopt a simple and interpretable linear model for predicting transfer free energy based on these relative subgraph frequencies,</p><p>DDG mean is the arithmetic mean of all N ¼ 439 DDG values and the regression coefficients q ˛IR 1608 assign weights to the different subgraph frequencies. The coefficients q are learned by minimizing the LASSO regression loss between the predicted and the calculated free energies, DDG predicted , DDG ref , respectively. The resulting expression yields</p><p>where a is the L 1 regularization weight promoting model sparsity. As a / N the L 1 regularization penalty in eqn (3) dominates corresponding to the null model where q ¼ 0, but as a / 0 nonzero elements progressively accumulate within q corresponding to the participation of more subgraph motifs into the regression model. This L 1 regularization serves to prevent overtting by identifying and retaining only a small number of the most generalizable features represented in our training dataset. The optimal value for a is selected using crossvalidation (ESI, Fig. S19 †) and the learned nonzero weights in q can be interpreted as the most critical subgraph motifs for transfer free-energy prediction. Further, the linear nature of eqn (2) admits a simple interpretation to the sign of the learned weights: large negative weights q k < 0 correspond to subgraph motifs predictive of good cardiolipin selectivity, while large positive weights q k > 0 correspond to subgraph motifs predictive of poor cardiolipin selectivity. Analyzing a rank-ordering of the largest negative/positive weights provides a means for automatic selection and discovery of design rules identifying subgraph motifs most impactful for DDG inference. Lastly, our trained model allows us to extrapolatively predict CL selectivity of unseen and arbitrarily large CG topologies via a decomposition into learned contributions of their constituent 1-5 bead subgraphs.</p><!><p>The CG model integrates out atomic-level information to only keep essential physicochemical properties. The ability to backmap from coarse-grained to all-atom resolution would offer chemical insight as to compounds of interest. As such, we aim to identify the diversity of all-atom structures given CG graphs of interest. This problem is more easily addressed by working from high-to low resolution: we consider a large set of molecules and coarse-grain them all. In this work we relied on the Generated DataBase (GDB). 76 We only considered molecules mapping to a single CG bead (1481 molecules of a size up to six heavy atoms). 35 On top of their original neutral form, we used the Calculator Plugin of Marvin 77 by ChemAxon to estimate their protonation state at pH 7. In addition, we account for aromatic groups by further considering around 22 000 small veor sixmember cyclic compounds. The CG representations of cyclic molecules were used to evaluate how heteroatoms or substituents present in cyclic hydrocarbons affected the choice of bead types in the coarse-graining process. To automatically identify functional groups, bead types were linked to chemical information using an algorithm described by Ertl 78 and implemented in RDKit. 79 We extended the Ertl algorithm to recognize and name the most common functional groups found in bioactive molecules 80 as well as additional chemical structures repeatedly encountered in our GDB-derived dataset. More detail of this analysis is provided in the ESI. † Probabilities of functional groups mapping to a CG bead type were inferred through observation frequencies. The probabilities between CG bead type and functional group offer a practical link to translate CG design rules into chemical structures.</p><p>3 Experimental methods . Liposome solutions with a total lipid concentration of 300 mM, 0.6 mM of DPH, and 3 mM of the indicated substance, were prepared as follows: the lipids (dissolved in chloroform) were mixed in the described ratio with DPH (dissolved in chloroform) and, if needed, the small molecule candidate compound (dissolved in methanol). The solvents were removed under a gentle nitrogen stream and the lipids were desiccated overnight under vacuum. The next day, the lipids were dissolved in 10 mM 4-(2-hydroxyethyl)-1-piperazineethanesulfonic acid (HEPES)-buffer (pH 7.4, 150 mM NaCl) via vortexing 2Â for 1 min, followed by 5 freeze-thaw circles. 250 mL of each sample was measured at 25 C. Each lipid composition was measured at least three times using fresh lipid preparations. Anisotropy was measured using a Fluoro-Max-4 uorescence spectrometer (Horiba (Bensheim, Germany)). Single point anisotropy was measured with 5 nm bandwidth, an excitation wavelength of 350 nm and an emission wavelength of 452 nm. Each sample was measured ve times and the results were averaged. Samples in the absence and presence of the small molecule candidates were measured leading to the anisotropy difference,</p><p>where A s is the measured uorescence anisotropy of a lipid membrane in presence of the small molecule, while A B is the anisotropy measured for an empty membrane. Experimental errors are given as standard error of the mean calculated from the three repetitions with fresh liposomes for the individual measurements, and calculated based on error propagation for the difference in anisotropy.</p><!><p>We perform active-learning directed CG alchemical free energy calculations within an RAE-learned chemical space embedding to discover compounds with high thermodynamic selectivity for CL membranes. We conduct seven rounds of active learning during which we consider 520 molecules (100 seeded and 60 per round) within the search space of 124 327 possible CG candidates. The majority (439 of 520) of these molecules were observed to be interfacial and for which we performed the complete alchemical free-energy calculations to determine DDG. The remaining 81 compounds failed to partition to the membrane interface, were deemed non-interfacial, and the free energy calculations terminated early as a resource-saving strategy.</p><p>The intent of the active learning search was to efficiently navigate chemical space to identify protable regions of our learned latent space embedding densely populated with highperforming candidates by balancing exploration-compounds predicted with high uncertainty, and exploitation-compounds predicted with high performance. The performance of the active learning search is presented in Fig. 3a, which illustrates the distribution of DDG values within each round, and Fig. 3b, which illustrates the corresponding location within the latent space of the selected candidates. The distribution of DDG values tends towards more negative values over the course of the search, indicating that the active learning strategy is successfully discovering high-performing candidate molecules. Particular rounds tend to perform more exploration, such as Rounds 6 and 7, reected by more variance in the sampled DDG distribution, broader latent space sampling, and a higher proportion of selected non-interfacial compounds. Other rounds, such as Rounds 4 and 5, tend to be more exploitative, reected by more localized latent space sampling and DDG distributions with more weight in the negative tail. While these trends in sampling help provide some intuition for the progress of the active learning process, our use of the expected improvement acquisition function within our Bayesian optimization selection procedure means that each round naturally balances exploration and exploitation to select the next most effective compounds to simulate in order to maximize the chances of discovering high-performing candidates within the design space.</p><p>We assessed convergence of the active learning by monitoring the performance of the best observed candidates aer each round. Throughout the rst four rounds we observe an approximately linear decrease in the overall best observed DDG, with the following three rounds yielding no overall improvement but nonetheless identifying relatively high performing candidates (Fig. 4a). The stagnant improvement over the last three rounds motivates us to terminate sampling aer Round 7. Over the course of the seven rounds of active learning we observed a 22.2% improvement in the best calculated DDG aer simulating a total of only 520 molecules, corresponding to a mere 0.42% of the 124 327 possible candidates. We present in Fig. 4b the 12 molecules with the best (i.e., most negative) DDG identied over the course of our search. Importantly, the best performing candidate possesses a DDG ¼ À3.27 kcal mol À1 corresponding to a 184% improvement over the NAO uorescent dye molecule which possess a thermodynamic selectivity of only DDG ¼ À1.15 kcal mol À1 calculated under our CG model. A full accounting of the measured DDG values for all 520 candidates is provided in the Data availability statement. 81 We can approximately quantify the savings afforded by our active learning process in discovering our best performing molecule by comparing against a baseline of naïve random sampling. A simple statistical analysis reveals that performing random selection of candidates within our 124 327 molecule candidate space and still considering 60 molecules per round, we would expect to happen upon our best-performing candidate with a 25% chance of success aer $518 rounds and with a 50% chance of success aer $1036 rounds. Although chemical intuition and prior experience could be used to guide the search, the bias and preconceptions that this introduces risks missing non-intuitive but high-performing candidates, and this baseline random search comparison nonetheless effectively highlights the value of our data-driven approach for guided molecular discovery.</p><!><p>Having completed seven rounds of active learning and calculating DDG values for N ¼ 439 compounds, we proceed to use graph representational learning to discover design rules correlating the presence/absence of chemical motifs to CL selectivity. In the interest of simplicity and interpretability, we adopt a simple linear model given in eqn (2) for predicting the calculated transfer free energy of a molecule via a featurization based on its decomposition into topologically unique structures with 1-5 CG beads. This model is trained using the LASSO regression algorithm to promote sparsity where select elements from the learned model weights are set precisely to zero q k ¼ 0, and the nonzero weights kq k k > 0 correspond to a small number of the most generalizable features retained by the model. Based on the linear structure of eqn (2) the sign of learned nonzero weights q k indicates whether the corresponding subgraph motifs contribute to more favorable DDG values leading to better CL selectivity if q k < 0, or unfavorable DDG values if q k > 0. Furthermore, as the subgraph representations F n,k in eqn (2) are normalized and therefore unitless, the coefficient weights q k carry the same units as DDG of kcal mol À1 . The magnitude of these coefficient weights q k can therefore be interpreted as the extent in kcal mol À1 that the representation of specic subgraph motifs improve, if q k < 0, or degrade, if q k > 0, upon the average CL selectivity DDG mean . Analyzing the prevailing structural features and characteristics contained in the largest magnitude model weights serves as a data-driven approach for unveiling critical determinants of CL selectivity.</p><p>In Fig. 5 we present a rank ordering of the largest magnitude nonzero model weights q k partitioned by the sign of the weight, sgn(q k ). These largest magnitude weights can be interpreted as corresponding to CG motifs with the highest predictive capacity and therefore the most inuential for determining CL selectivity. We note that while this analysis pertains to the CG space our active learning search is performed in, the following Sec. 4.3 As the weights q k have units of kcal mol À1 these weights can be interpreted as the magnitude of the influence of the corresponding subgraph motif on CL selectivity.</p><p>provides further analysis linking CG beads and structures to atomistic functional groups. This current analysis reveals the importance of subgraph motifs containing positive net charges (Q0+), apolar (T5) bead types, and weakly polar T3 bead with both hydrogen bond donor-and acceptor properties in promoting favorable CL selectivity (i.e., negative values of DDG). Contrariwise, negative charges (Q0À) and highly polar beads (T1, T2) tend to impair CL selectivity. These observations can be intuitively rationalized, given the two phosphate groups of CL that are predominantly negatively charged in the physiological pH range, 46,82 and the close proximity of the hydrophobic lipid tails to the interface region caused by the unique shape of the CL headgroup. The presence of apolar regions in the candidate structures therefore facilitates easy insertion into the Electrostatic interactions have been described as primary interaction modes of small molecules with the CL headgroup and the important role of hydrophobic interactions has also been recognized. [12][13][14]83,84 The small CL headgroup containing a single hydroxy group also suggests that hydrogen bonding may represent an important interaction type for targeting CL, and this correlates well with the frequent appearance of the T3 beads in the identied chemical motifs. Taken together, this analysis allows us to extract the following design rules for highly CL-selective small molecules from the active learning results:</p><p>(1) At least one, ideally two sites that will carry a positive charge at physiological pH (pH z 7.3).</p><p>(2) Hydrophobic areas in the molecule that induce alignment with or insertion into the lipid bilayer.</p><p>(3) Functional groups able to form hydrogen bonds with the CL headgroup.</p><!><p>We adopted the 5 + 1 CG model in order to efficiently screen chemical space at the cost of integrating atomic representations into coarse-grained beads. The top performing candidates identied within our active learning search are therefore represented as bead graphs as opposed to chemical structures. In order to approximately recover these lost degrees of freedom, we perform a backmapping analysis to identify functional groups with physical and chemical properties consistent with each CG bead. Due to the information removed by the coarse-graining procedure, there are multiple all-atom structures consistent with each CG representation, i.e., the mapping is many-to-one. The most prevalent bead to all-atom functional group mappings resulting from our analysis are shown in Fig. 6 and 7. The mapping probabilities for cycle-containing groups are presented in Fig. S25 in the ESI. † While a rigorous strategy for the backmapping of any CG molecule to all possible atomistic counterparts is beyond the scope of the present work, we illustrate in Fig. 8 the diversity of all-atom structures represented by a single coarse-grained molecule by presenting the range of GDB-derived small molecules whose coarse-grained mapping corresponds to a T3-T5 dimer as one of the chemical motifs correlated with increased CL selectivity (see Sec. 2.7). We identify a total of 2157 GDB-derived small molecules consistent with a T3-T5 dimer and render the chemical structures of seven of these in an attempt to convey the chemical diversity contained within this single coarse-grained dimer. We intentionally adopted a coarsegrained model for our screening procedure in order to reduce the size of chemical space and accelerate our simulations. This allowed us to efficiently identify CG representations of molecules predicted to have high CL selectivity, but results in enormous redundancy in the number of all-atom structures consistent with a single CG representation. Naturally the CG model is not able to rank compounds that map to the same CG representation. As such, CG modeling acts as a funnel to efficiently lter out uninteresting compounds. The results yield a small set of top performing CG structures, which can be further backmapped to an atomistic resolution. The set of Fig. 6 Five functional groups most frequently mapped to the uncharged beads of the 5 + 1 reduced Martini model. 25,35 Dark colors represent higher mapping probabilities. Fig. 7 Predominant functional groups mapped to the charged bead type Q0AE within our 5 + 1 reduced Martini model. 25,35 (a) Functional groups with apK a # 7 are most likely to be negatively charged under physiological conditions (pH z 7.3) and are mapped to the Q0À bead. (b) Those with bpK a $ 7 are most likely to be positively charged and are mapped to the Q0+ bead. consistent all-atom representations can be ltered to further design other desirable characteristics, such as functional groups that endow uorescence or other optical responses desirable for imaging agents. They also naturally lend themselves to validation and applications, through all-atom simulations or experiments. In the next section we pursue the latter strategy to validate our computational predictions using uorescence-anisotropy experiments. In future work, we would also like to pursue all-atom calculations of the top-performing candidates identied in our CG calculations but currently face challenges associated with the absence of validated all-atom force eld parameters and exceedingly high computational complexity of all-atom free-energy calculations in membrane systems. 85</p><!><p>We test the validity of our CG predictions of thermodynamic selectivity and of the learned design rules by comparing calculated DDG transfer free energies against experimental measurements of CL selectivity for two molecules: quinaldine red (4-[(E)-2-(1-ethylquinolin-1-ium-2-yl)ethenyl]-N,N-dimethylaniline) and benzothiazolium (3-butyl-2-methyl-1,3benzothiazol-3-ium) (Fig. 9). We selected these molecules according to our learned design rules, they were therefore predicted to preferentially partition into CL membranes. These molecules both possess at least one ionizable group that will likely be positively charged at pH z 7, hydrophobic character, and aromatic ring systems and partial charges that may participate in hydrogen bond formation. We experimentally measured the interaction of these two molecules with model membranes consisting of 100% 1,2-dioleoyl-sn-glycerol-3phosphocholine (PC), 90% PC/10% 1,2-dioleoyl-sn-glycerol-3phosphoglycerol (PG), and 90% PC/10% CL. We hypothesized that permeation of the molecules into the lipid membranes would affect lipid packing and order. This can be monitored using an experimental uorescence anisotropy assay to report the differential uorescence anisotropy DA ¼ A s À A B of a lipid membrane in presence of the candidate molecule A s and in its absence A B . Our hypothesis is that the two selected molecules should lead to a moderate increase in anisotropy DA if they incorporate into a membrane, and to a larger increase if they not only incorporate, but specically interact with a lipid species in the membrane. The measurements of the pure PC membrane were added to differentiate between simple incorporation and specic interactions, since the latter were not to be expected with this particular phospholipid. We present the results of these measurements in Fig. 9a. A negative control experiment conrms no change in DA in the absence of any molecular candidate. The addition of benzothiazolium led to slight increases in DA (i.e., decrease in membrane uidity) for all three model membranes, but this increase was similar in all cases indicating no preferential interaction that would result in a differential response. Quinaldine red led to more signicant increases in DA for all three membranes and a rank ordering in the magnitude of the response of PC/CL > PC/PG > PC, indicating a preferential partitioning into the CL-containing model membrane. These results indicate that our design rules successfully identied quinaldine red as a CL-selective molecule, whereas benzothiazolium was a false positive. Our learned design rules are a heuristic tool by which to identify candidate molecules, so we next sought to subject these two molecules to CG free energy calculations to determine if the DG and DDG values recapitulated the experimental trends.</p><p>We constructed CG representations of quinaldine red and benzothiazolium under the 5 + 1 model and subjected these to alchemical free energy calculations. Our experimental measurements of DA for NAO were clouded by interferences between the uorescence properties of the compound and the dye used for the anisotropy measurements (DPH). We therefore do not report the results in this work. We nevertheless also performed these calculations for NAO as a baseline comparison.</p><p>As illustrated in Fig. 9b, all three compounds showed favorable partitioning from bulk water and octane into the interface regions of both the PG and the CL membranes (i.e., DG W/I , DG O/I < 0). These results are consistent with the experimental observations of elevated DA for all three model membranes that is indicative of spontaneous thermodynamic partitioning of the molecules into the membranes. As demonstrated in Fig. 9c, the calculated relative partitioning free energies DDG follow the same trends as the experimental DA trends. In the case of benzothiazolium, the DDG ¼ À0.82 kcal mol À1 is close to thermal energy k B T z 0.6 kcal mol À1 at room temperature, showing moderate CL selectivity. This is in reasonable agreement with the lack of difference in anisotropy change observed in the experiments. In contrast, the DDG ¼ À1.67 kcal mol À1 of quinaldine red is more than twice as large as that for benzothiazolium and nearly three times larger than the thermal energy scale. This is consistent with the differential response in the experimental DA measurements showing signicantly elevated selectivity of quinaldine red for the CL membrane environment. Finally, we observe that the DDG ¼ À1.15 kcal mol À1 calculated for NAO is superior to that for benzothiazolium but inferior to that for quinaldine red, suggesting that the latter may offer superior CL selectivity.</p><!><p>The phospholipid cardiolipin (CL) is found exclusively within energy transducing membranes and constitutes approximately 20% of the composition of the inner mitochondrial membrane in eukaryotes. 1,5 It plays a key metabolic role while having demonstrated capability to also behave as a biomarker for detecting a number of diverse pathologies. A standing challenge in developing reliable CL-based diagnostics lies in the lack of molecules that display selectivity in binding to CL compared to other phospholipid membranes. With the number of drug-like molecules being on the order of $10 60 , a major challenge is the efficient exploration and ltration of this enormous molecular design space. Using transferable CG models to systematically reduce the size of chemical compound space we mapped this all-atom space into a more manageable 124 327-member CG design space. Nevertheless, a naïve Edisonian trial-and-error or random sampling within this design space remains prohibitively expensive. This motivated us to employ techniques in Bayesian optimization and deep representational learning to perform a directed data-driven search over the CG space of small organic molecules in an effort to minimize the data complexity, develop surrogate models to exploit the subtle structural differences in the CL and PG lipids to wield control over CL selectivity and efficiently discover the best performing candidates within an iterative active learning loop. Aer seven rounds of active learning and simulation of only 0.42% of the molecular design space, we identied candidate molecules with 184% better selectivity than the uorescent dye NAO that is commonly used as a selective CL stain.</p><p>Analyzing our accumulated CG simulation data with interpretable linear models enabled automatic discovery of chemical motifs predictive of good CL selectivity. We were able to derive design rules correlating the presence/omission of functional groups to CL selectivity by probabilistically mapping each CG bead to possible all-atom functional groups and relating these to the automatically identied CG motifs. CL selectivity is linked to positively charged groups, hydrophobicity, and the ability to form hydrogen bonds. In contrast, negatively charged groups and overall polarity led to the molecule partitioning into the bulk water, and dominant apolarity led to a tendency of the candidate molecules to insert into the membrane midplane regions.</p><p>We tested our computational models against experimental measurements of CL selectivity for two candidate molecules selected using our learned design rules. The trends in our calculated transfer free energies and relative partitioning free energy differences are in good accord with experimental measurements providing support and validation of our computational approach.</p><p>In future work, we plan to conduct experimental testing of other candidate molecules identied by our learned design rules and CG screen and conduct all-atom free energy calculations with a second higher-resolution virtual screen. This work adopted thermodynamic selectivity as the sole optimization objective for virtual screening without regard to the optical activity of the candidate molecules. Since the ultimate goal is to discover high-selectivity CL dyes, in subsequent work we plan to employ computational ltrations and/or multi-objective optimization strategies to discover molecules that are both highly selective for CL and possess either inherent uorescent activity or can support the addition of uorescent tags without compromising their thermodynamic behavior. Finally, we note that the coarse-grained chemical space embeddings and active learning search are generically transferable to other molecular design applications by replacing the thermodynamic selectivity prediction for a computational or experimental assay of the property of interest.</p>

Royal Society of Chemistry (RSC)

Secreted Frizzled-Related Protein-2 (sFRP2) Augments Canonical Wnt3a-induced Signaling

Secreted Frizzled-Related Proteins (sFRP) are involved in embryonic development as well as pathological conditions including bone and myocardial disorders and cancer. Because of their sequence homology with the Wnt-binding domain of Frizzled, they have generally been considered antagonists of canonical Wnt signaling. However, additional activities of various sFRPs including both synergism and mimicry of Wnt signaling as well as functions other than modulation of Wnt signaling have been reported. Using human embryonic kidney cells (HEK293A), we found that sFRP2 enhanced Wnt3a-dependent phosphorylation of LRP6 as well as both cytosolic \xce\xb2-catenin levels and its nuclear translocation. While addition of recombinant sFRP2 had no activity by itself, Top/Fop luciferase reporter assays showed a dose-dependent increase of Wnt3a-mediated transcriptional activity. sFRP2 enhancement of Wnt3a signaling was abolished by treatment with the Wnt antagonist, Dickkopf-1 (DKK1). Wnt-signaling pathway qPCR arrays showed that sFRP2 enhanced the Wnt3a-mediated transcriptional up-regulation of several genes regulated by Wnt3a including its antagonists, DKK1 and Naked cuticle-1 homolog (NKD1). These results support sFRP2\xe2\x80\x99s role as an enhancer of Wnt/\xce\xb2-catenin signaling, a result with biological impact for both normal development and diverse pathologies such as tumorigenesis.

secreted_frizzled-related_protein-2_(sfrp2)_augments_canonical_wnt3a-induced_signaling

Introduction<!>Cell culture<!>Antibodies and recombinant proteins<!>Western blot<!>Transfection and reporter assays<!>RNA Extraction, cDNA synthesis, and PCR array analysis<!>Statistical analysis<!>sFRP2 enhances the Wnt3a-induced \xce\xb2-catenin stabilization and its nuclear translocation<!>sFRP2 enhances Wnt3a/\xce\xb2-catenin transcriptional reporter activity<!>sFRP2 enhances LRP6 phosphorylation<!>Dickkopf-1 (DKK1) antagonizes both Wnt3a/sFRP2-induced LRP6 phosphorylation and transcriptional activity<!>sFRP2 enhances expression of four genes known to be regulated by Wnt3a signaling<!>sFRP2 enhances Wnt3a-transcriptional activity in HSG cells but not in L-cells<!>Discussion<!>

<p>Canonical Wnt/β-catenin signaling is a highly conserved pathway that is essential for cell-fate, patterning during development, and is often involved in human diseases such as cancer [1; 2]. The stabilization and nuclear translocation of β-catenin are two hallmarks of canonical Wnt signaling [1; 2]. In the absence of Wnt, β-catenin binds to scaffold proteins, Axin and adenomatosis polyposis coli (APC), and undergoes phosphorylation thereby triggering its proteasome-dependent degradation. Signaling starts with Wnt proteins binding to two cell surface receptors, a member of the Frizzled (Fz) serpentine receptor family [3; 4] and a single-pass transmembrane receptor, LRP6 (low density lipoprotein receptor-related protein-6), or closely the related, LRP5 [5; 6]. When Wnt binds to its receptors, β-catenin phosphorylation is suppressed leading to its accumulation in the cytoplasm and translocation to the nucleus where it interacts with members of the T-cell factor/lymphoid enhancer factor (TCF/LEF) family of transcription factors, consequently initiating target gene transcription [7]. Wnt/β-catenin signaling is modulated by an array of secreted molecules, including Wnt inhibitory signaling factor-1 (WIF1), Cerebrus, Sclerostin, Dickkopf-1 (DKK1), and secreted Frizzled-related proteins (sFRP). Sclerostin and DKK1 antagonize canonical signaling by binding to LRP5/6, whereas WIF1, Cerebrus, and sFRP2 are reported to interact directly with Wnt proteins [8; 9].</p><p>The sFRPs constitute a family of 5 proteins in mammals: Frzb (sFRP3), sFRP1, SFRP2, sFRP4, and sFRP5 [10]. They are expressed in adults and during embryonic development in dynamic as well as spatially restricted manners. Furthermore, their expression is altered in various bone pathologies [11], retinal degradation [12], hypophosphatemic diseases [13], and myocardial disorders [14; 15] as well as in different types of cancer [9]. Because of their sequence homology with the Wnt-binding domain of the Fz receptors, sFRPs have been considered antagonists of canonical Wnt signaling by binding to Wnt proteins and preventing signal transduction [16; 17]. However, sFRPs have recently been reported to synergize or mimic Wnt activities by direct interaction with Fz receptors [18; 19; 20], by antagonizing each other's action [21], enhancing extracellular transport of Wnt proteins [22], or by playing roles other than directly controlling Wnt signaling pathways [23].</p><p>In this study, we used two cell lines of epithelial tissue origin (HEK293 and human salivary gland intercalated duct cell line, HSG) to show that sFRP2 enhances Wnt3a's ability to induce: 1) phosphorylation of LRP5/6; 2) β-catenin stabilization and its nuclear translocation; and 3) changes in expression of known Wnt-mediated genes. Furthermore, we show that a specific inhibitor of Wnt canonical signaling, DKK1, negates all sFRP2 effects on Wnt3a signaling.</p><!><p>The murine L-cells M(TK−) (ATCC, Manassas, VA) and the human HEK293A cells (Invitrogen, Carlsbad, CA) were propagated in Dulbecco's Modified Eagle's Medium (DMEM) (Gibco-BRL, Gaithersburg, MD) containing 10% fetal bovine serum (FBS; Atlanta Biologicals, Lawrenceville, GA). HSG cells (established from an irradiated human salivary gland as described [24]) were grown in DMEM/F12 medium (Gibco-BRL) containing 10% FBS. Media were supplemented with L-glutamine (22 mM), penicillin (100 IU/mL), and streptomycin (100 μg/mL). For conditioned media (CM), L-cells were grown to subconfluence, replenished with fresh complete culture medium that was then collected 48 h later, centrifuged at 700 × g for 10 minutes, and stored in aliquots at −80 °C.</p><!><p>Mouse monoclonal anti-β-catenin antibody was from BD Transduction Laboratories (Cat. #610154, Franklin Lakes, NJ). Rabbit anti-phospho-LRP6 (Ser1490) antibody was from Cell Signaling (Cat. #2568, Beverly, NJ). Mouse monoclonal anti-β-actin antibody was from Sigma-Aldrich (Cat. #A1978, St. Louis, MO) and mouse monoclonal anti-TATA-binding protein (TBP) was from Abcam (Cat. #ab61411, Cambridge, UK), IRDye 800 goat anti-mouse IgG and IRDye 680 goat anti-rabbit IgG second antibodies were from LI-COR Biosciences (Lincoln, NE). Carrier-free recombinant mouse sFRP2, recombinant human Wnt3a, and recombinant human DKK1 were purchased from R & D Systems (Minneapolis, MN).</p><!><p>1×106 cells/well were seeded into 6-well plates. After overnight culture, cells were treated with Wnt3a alone or in combinations of sFRP2 and DKK1 for 2 h at 37 °C. For total cellular protein, cells were lysed with 100 μl 2x Laemmli Sample Buffer. Alternatively, cytosolic and nuclear fractions were prepared using the NE-PER Nuclear and Cytoplasmatic Extraction Kit (Pierce, Rockford, IL). Equal volumes of extracts were electrophoresed on 4–12 % Bis-Tris Nu-PAGE gels (Invitrogen) and transferred to Immobilon-FL membrane (Millipore, Billerica, MA). Blots were blocked for 1h in Odyssey Blocking Buffer (LI-COR Biosciences), followed by incubation with the first antibody overnight at 4°C and then with IRDye 680 goat anti-rabbit IgG or IRDye 800 goat anti-mouse for 1h at room temperature (RT). Blots were analyzed using the LI-COR Odyssey infrared imaging system (LI-COR Biosciences).</p><!><p>1.5×104 HEK293A cells/well were plated into 96-well plates. After overnight culture, the cells were transfected with 50 ng of the Super 8x TopFlash plasmid or the control plasmid Super 8x FopFlash (Addgene Inc., Cambridge, MA) and 1 ng of the pRL-SV40 vector (Promega, Madison, WI) in Opti-MEM medium using 0.25 μl Lipofectamine 2000 per well according to manufacturer's protocol (Invitrogen). After 6 h, cells were switched to serum-containing medium. Eighteen hours later, cells were treated with Wnt3a and/or sFRP2. For inhibition experiments, cells were incubated with DKK1 for 30 min prior to treatment with Wnt3a and sFRP2. After 24-h treatment, cells were lysed and the luciferase activity was determined using the Dual Luciferase kit from Promega. All reporter assays were performed in triplicate and each experiment was repeated at least three times. The relative firefly luciferase activity was normalized with its respective Renilla luciferase activity.</p><!><p>Using a Trizol/RNeasy hybrid protocol, total RNA was isolated from cells treated for 2 h with Wnt3a alone or in combination with sFRP2. This protocol takes the RNA extracted with Trizol (Invitrogen) through an additional column-based clean-up RNeasy step (Qiagen, Germantown, MD) following instructions for both kits. The first-strand cDNA was synthesized from 1 μg of total RNA (quantified by NanoDrop, Wilmington, DE) using the RT2 First Strand Kit (SABiociences, Frederick, MD) (including a DNase pretreatment to remove any genomic DNA). cDNA reactions were mixed with the SYBR Green qPCR Master Mix and transferred into 96-well RT2 Profiler PCR Array for human or mouse WNT-Signaling Pathway (SABiosciences), containing primers for 84 genes related to Wnt-mediated signal transduction as well as housekeeping genes. The two-step real-time PCR reaction was performed using the MyiQ5 instrument (Bio-Rad, Hercules, CA) according to SABioscience's MyiQ5-specific instructions.</p><!><p>Threshold cycle values obtained from RT2 Profiler PCR Arrays (three experiments) were analyzed using the PCR Array Data Analysis Web Portal from SABioscienecs. All other statistical analyses were performed using GraphPad software (San Diego, CA). Multiple comparisons were performed with a one-way analysis of variance (ANOVA) followed by Newman-Keuls test comparing all pairs of columns. Results are expressed as mean ± SEM.</p><!><p>The stabilization of β-catenin and its nuclear translocation are two hallmarks of canonical Wnt-signaling. HEK293A cells were treated for 2 h with purified recombinant Wnt3a plus/minus recombinant sFRP2 and the total β-catenin levels in cell lysates were visualized by immunodetection on Western blots. As shown in Fig. 1A, incubation with 3 nM Wnt3a alone increased the total β-catenin level in HEK293A cells, indicating that Wnt3a was able to activate the canonical pathway in these cells. The addition of 6 nM recombinant sFRP2 with the 3 nM Wnt3a increased the total β-catenin accumulated. Furthermore, using nuclear protein-enriched cell fractions, the induction of overall β-catenin protein level upon Wnt3a/sFRP2 treatment was also associated with greater β-catenin nuclear translocation (Fig. 1B). Treatment with 6 nM sFRP2 alone did not effect either β-catenin accumulation or its relative nuclear translocation (data not shown).</p><!><p>Nuclear β-catenin associates with transcription factors of the T-cell factor/lymphoid enhancing factor family (TCF/LEF), leading to transcriptional activation of Wnt-targeted genes. Therefore, to further characterize the sFRP2 effects on the Wnt3a/β-catenin signaling, HEK293A cells were transfected with either the reporter plasmid, Super 8x TopFlash (containing 8 TCF/LEF-binding sites driving the transcription of the luciferase enzyme), or the control plasmid, Super 8x FopFlash (containing inactive TCF/LEF-binding sites). After a recovery period, the TCF/LEF transcriptional activity was measured upon treatment with 3 nM Wnt3a alone or in combination with increasing doses of sFRP2 (3 nM, 6 nM, and 12 nM). As expected, Wnt3a alone increased the transcription of the luciferase reporter gene (Fig. 1C) by 13.7-fold (±1.9, n=5). As seen in the same figure, addition of 3–12 nM sFRP2 further enhanced this transcriptional activity an additional 6 to 10-fold in a dose-dependent manner (e.g., 132-fold ±14.6 over control for 3 nM Wnt3a and 12 nM sFRP2; n=5). Neither treatment with Wnt3a alone nor together with sFRP2 changed the luciferase activity of the control, FopFlash plasmid (Fig. 1C).</p><!><p>Wnt3a mediates its effects through engagement with two distinct cell surface receptors, Frizzled (Fz) and LRP6/5, resulting in the phosphorylation of the latter [5]. Therefore, phosphorylation-dependent LRP6 antibody was used on Western blots to visualize its phosphorylation levels upon treatment with 3 nM Wnt3a ± 6 nM sFRP2. As shown in Fig. 1D, treatment with Wnt3a induced the expected phosphorylation of LRP6 and simultaneous treatment with both Wnt3a and sFRP2 further augmented the LRP6 phosphorylation in HEK293A cells.</p><!><p>DKK1 is known to antagonize the Wnt signaling by binding LRP6 thereby preventing the Fz/LRP6 complex formation and subsequent LRP6 phosphorylation [8]. To gain more insight into the mechanisms of the sFRP2 action on the Wnt3a/β-catenin signaling pathway, HEK293 cells were first incubated with DKK1 and then Wnt3a plus/minus sFRP2. Next, the levels of LRP6 phosphorylation and TCF/LEF transcriptional activities were determined. Recombinant DKK1 blocked both the Wnt3a and the Wnt3a/sFRP2-enhanced phosphorylation of LRP6 (Fig. 2A). DKK1 pretreatment also abolished TCF/LEF transcriptional activity induced by both Wnt3a and Wnt3a/sFRP2 (Fig. 2B). Together these data indicate that sFRP2 augments canonical Wnt3a/β-catenin signaling at the level of its cell surface receptor.</p><!><p>To show that the sFRP2 enhancement of Wnt3a signaling had downstream biological effects, the expression of genes known to be related to Wnt signaling was studied using Wnt-signaling Pathway qPCR arrays. Treatment with Wnt3a plus sFRP2 enhanced mRNA expression of human WNT11 (4.0-fold; p= 0.03) and T gene (4.2-fold; p=0.002) over control. Interestingly, mRNA expression of two Wnt signaling inhibitors, DKK1 and NKD1, was also significantly elevated by treatment with the Wnt3a/sFRP2 combination (4.5-fold; p=0.003 and 5.7-fold; p=0.008, respectively) (Fig. 3).</p><!><p>Next, we asked whether sFRP2 had the same Wnt3a signal-enhancing effect on cells of a different origin. Since HEK293 cell line has recently been suggested to be of neuronal lineage [25], we tested a human salivary gland cell line, HSG, of epithelial origin and a murine L-cells of fibroblastic origin (commonly used in studies of Wnt signaling). As shown in Fig. 4A, Wnt3a alone increased the TopFlash luciferase activity by 2.4-fold (±0.4; n=6) in HSG cells and by 11.9-fold (±1.9; n=3) fold in L-cells. The luciferase activity increased significantly in a dose-dependent manner up to 13.0-fold (±2.7; n=6) fold in HSG cells with increasing concentrations of sFRP2. The addition of two molar excess of sFRP2 to Wnt3a-treated HSG cells also increased phosphorylation of the LRP6 receptor as well as the β-catenin accumulation suggesting the same pathways are enhanced in these epithelial cells (data not shown). Surprisingly, co-treatment with Wnt3a and recombinant sFRP2 protein did not significantly change the luciferase activity in L-cells (Fig. 4B).</p><p>Analysis of mouse Wnt-signaling Pathway qPCR arrays revealed a low expression of sFRP2 in L-cells (data not shown), suggesting that the lack of enhancement by the exogenously added sFRP2 in L-cells may have been due to endogenous expression of sFRP2 or other enhancing factor by the L-cells themselves. To test this hypothesis, HEK293A cells were transfected with the reporter construct, Super 8x TopFlash, and later exposed to increasing concentrations of L-cell-conditioned media (12.5%, 25%, and 50%; v/v) with or without the presence of 3 nM Wnt3a. As shown in Fig. 4C, co-treatment with Wnt3a and L-cell CM resulted in a dose-dependent increase of luciferase reporter gene activity when compared with Wnt3a treatment alone. Treatment with 50% CM alone did not increase luciferase activity (remaining at the level observed for DMEM medium control) showing that L-cells did not produce significant amounts of Wnt3a. The increased transcriptional activity upon treatment with 50% CM plus Wnt3a was comparable with that observed upon treatment with Wnt3a plus recombinant sFRP2 and was completely inhibited by DKK1 (Fig. 4C).</p><!><p>Currently, sFRP proteins appear to represent the largest family of Wnt modulators. These proteins were first predicted to be antagonists of canonical Wnt signaling based on their sequence homology with the ligand-binding domain of Frizzled receptors and were expected to exert their inhibitory effects by sequestering Wnt proteins from their receptors [16; 17]. However, some recent reports suggest that the function of sFRPs is more complex [23]. In this study, we provide several lines of evidence that sFRP2 enhances the Wnt3a-mediated canonical signaling in the context of HEK293A and HSG cells. Moreover, our results indicate that sFRP2 acts at the ligand/receptor level because recombinant sFRP2 substantially enhanced the Wnt3a-mediated phosphorylation of the LRP6 co-receptor and was efficiently blocked by the antagonist, DKK1. Furthermore, exogenously added DKK1 blocked the activation of the TCF/LEF reporter gene. Finally, co-treatment with Wnt3a and sFRP2 significantly increased expression of several known downstream target genes of β-catenin/TCF/LEF, including two Wnt antagonists DKK1 [26] and NKD1 [27], the transcription factor T [28], as well as WNT11 [29]. Our data support a recently published report showing that different sFRPs, including sFRP2, may co-function with Wnt3a to enhance osteoblastic differentiation and its transcriptional activity [30].</p><p>The precise molecular mechanism responsible for the sFRP2-mediated augmentation of the Wnt3a-mediated canonical signaling, however, remains unclear. At this time, sFRPs are proposed to act not only by directly binding to Wnt proteins but also directly with Frizzled proteins [18; 19]. More recently, it was shown that sFRP2 stabilizes β-catenin via interaction with Frizzled receptors without functionally interfering with Wnt3a ligand in a model of intestinal epithelium [20]. Although, we did not observe any effects on β-catenin signaling by sFRP2 alone, our results do not enable us to conclude whether sFRP2 binds Wnt3a or a cell/tissue specific Frizzled receptor. The latter model provides the possibility that the ability of sFRP2 to modulate Wnt3a signaling may depend on its ability to bind a specific Frizzled receptor. This suggests that the ability of sFRP2 to modulate Wnt3a may vary with the expression of the Frizzled receptors and, thus, depend on molecular and cellular context. For example, GeneChip assays performed on NIH3T3 cells and the rat pheochromocytoma cell line, PC12, revealed changes in expression in 355 and 129 genes, respectively, after Wnt3a-stimulation. However, only two Wnt-regulated genes were shared by both cell lines [31]. Consistent with this notion, that sFRP2 function might occur in a tissue-specific manner, we did not observe any enhancing effects of sFRP2 on Wnt3a signaling in L-cells of fibroblastic origin, which is agreement with the study by Lee et al. [32] showing that sFRP2-transfected fibroblasts do not counteract the dermomyotome-inducing activity of Wnt3a in presometic mesoderm explants. However, other studies have observed that sFRP2 is capable of antagonizing Wnt3a-activity in L-cells [33; 34]. One possible explanation for this difference might be that many earlier studies relied on the use of conditioned media that may have included bioactive proteins of unknown type and concentrations, whereas recent efforts, including ours, are based on the use of currently available purified proteins.</p><p>Exogenously added sFRP2 did not enhance the Wnt3a-mediated signaling in L-cells, perhaps reflecting the cell specific expression of different Frizzled receptors. However, since HEK293A cells treated with L-cell-conditioned media (which according to the Wnt-Signaling Pathway qPCR Array analyses express sFRP2 message) showed an enhanced transcriptional activity, one can speculate that cultured L-cells produce endogenous sFRP2 (or other factor) which induced the Wnt3a signaling in these cells, thereby making them insensitive to exogenous sFRP2.</p><p>In conclusion, our results further establish sFRP2 as a potential enhancer of canonical Wnt signaling and therefore also support its possible role as a paracrine/autocrine positive modulator of this key cellular signal transduction pathway and, therefore, may have biological relevance for both development and various diseases.</p><!><p>sFRP2 enhances: (A)Wnt3a-mediated β-catenin accumulation, (B) β-catenin nuclear translocation, (C) TCF/LEF transcriptional activity, and (D) LRP6 phosphorylation in HEK293A. Cells were treated with Wnt3a ± indicated sFRP2 for 2 h. The levels of total cellular and nuclear β-catenin as well as phosphorylated LRP6 were detected on Western blots. Antibodies against β-actin and TATA-binding protein (TBS) were used as loading controls. Shown are representative results of three independent experiments. For luciferase assays, cells were co-transfected with Renilla transfection-control plasmids plus either TopFlash (□) or FopFlash (■) plasmids. After transfection, cells were treated with Wnt3a ± sFRP2 for 24 h before extraction for luciferase assays. Shown are means ± SEM from at least 3 experiments each performed in triplicate wells. ***P<0.0001 compared with control by using of ANOVA and the Newman-Keuls test.</p><p>DKK1 blocks sFRP2-enhanced Wnt3a LRP6 phosphorylation (A) and TCF/LEF transcriptional activity (B). HEK293A cells were treated with Wnt3a alone or with indicated sFRP2 for 2 h. For inhibition studies, cells were pre-treated with 6 nM DKK1 for 30 min. The levels of LRP6 phosphorylation were analyzed by Western blot with equal loading verified by detection of cellular β-actin. For luciferase assay, cells were co-transfected with TOPflash and Renilla plasmids for 6 h and then treated with Wnt3a with/without sFRP2 at concentrations indicated for 24 h. Shown are mean ± SEM of at least 3 independent experiments, each performed in triplicate wells. ***P<0.0001compared with control by using of ANOVA and the Newman-Keuls test.</p><p>sFRP2 augments Wnt3a-induced gene expression in HEK293A cells. Cells were treated with 3 nM Wnt3a alone or in combination with 6 nM sFRP2 for 6 h. RNA was isolated and analyzed using Wnt Signaling Pathway RT2 Profile PCR Arrays. Shown are mean ± SEM of 3 independent experiments.</p><p>sFRP2 enhances Wnt3a-mediated transcriptional activation in HSG cells (A) but not in L-cells (B). Cells were co-transfected with TopFlash (□) or FopFlash (■) construct and Renilla (transfection control) plasmids and 24 h later treated with Wnt3a alone or together with indicated sFRP2 for 24h. (C) After transfection with TopFlash and Renilla plasmids, HEK293A cells were treated with L-cell-conditioned medium plus/minus 3 nM Wnt3a. For inhibition experiments, cells were pretreated with 6 nM DKK1 for 30 minutes. For controls, cells were treated with Wnt3a and sFRP2 or with 50 % CM alone. After 24 h, cell lysates were analyzed for luciferase activity. Shown are mean ± SEM of at least three experiments, each performed in triplicate wells. ***P<0.0001; **P<0.001; *P<0.01 compared with control by using of ANOVA and the Newman-Keuls test.</p>

PubMed Author Manuscript

Evaluation of a biocoagulant from devilfish invasive species for the removal of contaminants in ceramic industry wastewater

This study evaluated the effectiveness of a biocoagulant produced from the devilfish invasive species and its combination with two chemical coagulants (aluminum sulfate and ferric sulfate) to remove turbidity, chemical oxygen demand, and total suspended solids in ceramic industry wastewater using a combined experimental design of Mixture-Process. This design optimized the coagulation process and evaluated the effects and interactions between mixture components and coagulant doses. An analysis of variance was used to analyze the experimental data obtained in the study, and the response surface plots by response type (turbidity, chemical oxygen demand, and total suspended solids) were obtained. Results showed that the coagulation treatment could be technically and economically feasible since efficiencies of turbidity, chemical oxygen demand, and total suspended solids removal of 74, 79, and 94% could be achieved using an optimal coagulant dose of 800 mg/L with a mixture of 35% biocoagulant and 65% ferric sulfate. Analysis of variance results showed that the models are significant, and the lack of fit is not required according to the probability value (p value), which were < 0.0001, and > 0.05, respectively. Hence, the experimental data were fitted to a combined reduced special cubic x linear model. These results support the use of devilfish meal as a biocoagulant, being more feasible in dual systems when mixed with ferric sulfate.

evaluation_of_a_biocoagulant_from_devilfish_invasive_species_for_the_removal_of_contaminants_in_cera

<!>Material and methods<!>TGA, and heavy metals content for devilfish meal (biocoagulant).<!>Conclusions

<p>The introduction of invasive species has been identified as the most significant cause of global change 1,2 because it is the second most common cause of species extinction (after habitat destruction) 3 . This activity causes ecological impacts that can extend throughout the food chain, altering the functioning of the elements that make up an ecosystem and deteriorating ecosystem and environmental services throughout the world 1,4,5 . Similarly, it has been reported that these species have significant socioeconomic and health impacts that must be considered together with the environmental ones [6][7][8] .</p><p>The challenges for controlling and managing invasive species are approached from different points of view. One of these points is how to promote their management 9,10 , establishing policies that integrate the effects (cost-benefit) of the introduction, control, management, or eradication of these species on ecosystems. Therefore, one alternative is their sustainable use, considering causing their death to improve the control of biological invasions according to its type, providing some use or application to reduce their category as invasive species 2,[11][12][13] .</p><p>Devilfish is an invasive species of the Loricariidae family native to South America of the Amazon basin, but it has been introduced in several countries, as in Mexico 14,15 . Devilfish has become an environmental problem in the country, at least in the last 20 years 16 . Its presence affects other fish species since they destroy their eggs and compete for food 17 . One of the most significant impacts is the negative impact that they generate on the economy and welfare of fishers who sell Nile tilapia since have reduced its population by 83%, generating a collapse of commercial fishing that has affected the state of health, unemployment, and the emigration of fishers 14,18 . Due to the environmental and socioeconomic impacts that this species generates, fishers cause their death to better control these biological invasions with the purpose of their well-being or economy is not harmed. However, when these species are killed, they are not put to any use and are only generated as waste 11 .</p><p>Devilfish biomass has been studied to produce adsorbent materials 19 and bioanodes of microbial fuel cells 20,21 as low-cost alternatives for water and wastewater treatment. Likewise, its meal has been studied as a partial replacement of commercial fish meal in fish feed formulations [22][23][24] since it is a good source of proteins, minerals, and fats 25 . The collagen is the principal constituent of the intramuscular connective tissue of a fish, exerting a vital function in the texture of their meat 26 and has been mainly used as the ideal coagulation material 27 . Even though the devilfish meat has a high-quality protein content to be consumed with nutritional benefits, it has been reported the possible health risks in the direct consumption of devilfish meat, or indirect to be used the devilfish meal as food for fishes due to exposure to heavy metals such as Cadmium, Chromium, Manganese, and Lead 16 . In this sense, one alternative to explore the use of the devilfish biomass is producing a biocoagulant applied in wastewater treatment.</p><p>Biocoagulants have been studied due to their capability to improve the removal efficiency of contaminants in the water 28 . These materials are based on natural polymers like cellulose, mucilage, and collagen, and they are produced from the animal base, plant base, microorganisms, and bacteria 28,29 . Biocoagulants used in wastewater treatment are mainly based on plant bases. Some of them are banana pith 30 , Jatropha curcas 31 , and Moringa oleifera 32 . The use of animal biomass has been scarcely reported for this purpose, but it attracts more attention when biocoagulants are produced from invasive species.</p><p>The ceramic industry wastewater requires a primary treatment based on coagulation/flocculation/sedimentation processes since it contains inorganic matter associated with sand and clays. Hanife 33 reported the 95% of Chemical Oxygen Demand (COD) removal in the ceramic industry wastewater using a dose of 3.3 g/L of alum at pH 5. It was better than ferric chloride. However, it has been reported that chemical coagulants (iron salts and aluminum salts) have contributed to toxicological issues and environmental effects caused by the constant consumption and large doses during the treatment 28 . Therefore, this treatment based on natural materials, as the use of biocoagulants, has been a topic of interest in the investigation due to their positive effects when their use reduces or avoids the possible toxicity for human health and biota caused by using chemical agents in the wastewater treatment 28 . Biocoagulants produced from Jatropha curcas, Moringa oleifera, Dolichos lablab, and Cicer arietinum seeds have been proved in the treatment of synthetic water with kaolin and clay materials, reporting efficiencies of turbidity removal > 88% 31,34 .</p><p>In this context, this work aims to evaluate the effectiveness of a biocoagulant from the devilfish invasive species to remove contaminants of the ceramic industry wastewater, studying it at different concentrations and using only the biocoagulant or its combination with chemical coagulants (aluminum sulfate and ferric sulfate). This study is based on an experimental design that allowed to find the optimal operation conditions (optimal doses and proportions of mixture) to treat mentioned wastewater. This work presents an integrated solution to two problems, the first one the exploitation of the invasive species devilfish as biomass helpful in controlling its population in ecosystems that it does not belong and the second one the production of a biocoagulant that can be used as primary treatment in wastewaters, reducing the use de chemical agents.</p><!><p>Feedstock. This research produced a biocoagulant from devilfish meat, and its effectiveness was compared with two chemical agents used as conventional coagulants. The non-live devilfish samples were provided by the Fishermen's Society of Tenosique (Tabasco, Mexico). Four fishes weighing between 200 g and 1 kg (giving 1.7 kg) were used, obtaining approximately 0.5 kg of meat.</p><p>Ceramic industry wastewater was used as a case study to evaluate the effectiveness of the biocoagulant. The samples were collected at the wastewater discharge point of a ceramic industry (located on latitude 22°55′23″ N and longitude 102°40′40″ W). This wastewater was selected because the stages required for its treatment are at the primary level, using sedimentation-coagulation-flocculation processes since it contains inorganic matter associated with sand and clays.</p><p>The chemical agents used as conventional coagulants were aluminum sulfate and ferric sulfate (reagent grade, Mca. Meyer). Furthermore, a commercial flocculant (anionic polymer) was used for all the tests. This flocculant was prepared at 1 mg/L and was provided by a Mexican company.</p><p>Preparation, thermogravimetric analysis, and heavy metal content of the biocoagulant. The biocoagulant was prepared from a devilfish meal. The devilfish samples of different sizes were pooled together. Their visceral organs were eliminated, and then they were cleaned and rinsed with water (first with tap water and second with distilled water), and their meat was cut into smaller portions (approximately 50-100 g). The meat portions were boiled at 100 °C for 30 min to separate water and oil. Then, the liquid was removed using a cloth strainer 35 , and they were stored at 70 °C for 24 h to eliminate any further moisture. After this, they were ground into powder using a meat grinder (commercial blender), sifted through a 500 μm mesh, and kept in a sterile place 23 . The yield was 182 g of biocoagulant (devilfish meal) per kg wet weight devilfish meat.</p><p>The thermal decomposition of the devilfish meal was studied by a thermogravimetric analysis technique (TA Instruments TGA Q500). Approximately 18.94 mg of sample was analyzed in a nitrogen atmosphere and a range of temperatures from ambient to 600 °C. A heating rate of 5 °C/min was used. Likewise, the devilfish meal's www.nature.com/scientificreports/ X initial and X final were the turbidity, COD, and TSS values before and after the coagulation process, respectively. The results were expressed as the mean of two measurements. Furthermore, the pH was analyzed to identify a possible variation of this parameter when the jar test is performed.</p><p>Statistical analysis and optimization. An analysis of variance ANOVA was performed for each response (turbidity, COD, and TSS) using Design-Expert® Version 12 Software (Trial version). The significance of the model and the lack of fit were established at a p value < 0.05. 3D surface plots and final model equations in terms of the mixture's components and the numeric factor were presented by response type. Finally, an optimization analysis was performed under the conditions of the mixture's components (range 0-1) and the doses (range 200-800 mg/L) to obtain the highest removal percentage of the three quality parameters using the obtained final model equations.</p><!><p>TGA is a technique used to determine the composition of a material and predict its thermal stability at temperatures up to 1000 °C. It measures the amount and change rate of the mass loss of a material as a function of temperature or time in a controlled atmosphere 39 . Figure 1 shows the thermal degradation curve's TGA thermogram (solid line) and its first derivative (dotted line) for devilfish meal, where three different regions related to the most pronounced mass losses are observed. The principal mass loss is presented in three different thermal events with temperature ranges given by 40-175 °C, 175-350 °C, and 350-600 °C, respectively.</p><p>The first thermal event (40-175 °C) corresponds to the loss of the physisorbed and chemical water in the devilfish meal 40 , representing a mass loss of 7.7%. The most representative weight loss (49.4%) occurred in the second thermal event (175-350 °C). At this point, carbohydrates, and high molecular weight polysaccharides such as proteins, lipids, and other organic compounds are breaking down since this weight loss is in the range of temperatures where the degradation of starch and protein occurs 39 . These losses are also related to the combustion of collagen 40 , which is the interest component since it was thought to take advantage of this property so that the devilfish meal could act as a biocoagulant. Finally, the devilfish meal had a weight loss of 42.9% in the third thermal event (350-600 °C). This loss indicates that devilfish meal contains minerals and coupled salts such as calcium and phosphorus compounds 39,40 . The total mass loss of the devilfish meal was obtained at 600 °C.</p><p>The concentrations of cadmium, mercury, and lead of the devilfish meal were determined by ICP-MS. The results were 0.003 ± 0.01 mg/kg, 0.051 ± 0.04 mg/kg, and 1.211 ± 0.08 mg/kg, respectively. The cadmium and mercury concentrations are within the reference value (0.5 mg/kg) stipulated by the Health Official Mexican Standards NOM-242-SSA1-2009 36 . However, the lead concentration is 1.2 times higher than the reference value (1.0 mg/kg) established by the mentioned standard 36 . Although some authors have proposed its use as a source of proteins of high commercial value in food production and as a replacement for commercial fish meal [22][23][24] . Other authors suggest that the consumption of devilfish meat by humans and the use of devilfish meal as feed for fishes must be evaluated due to the possible health risk by the exposition to heavy metals bioaccumulated in these species 16 , such is the case of the lead content found in the devilfish meal produced in this work.</p><p>One of the reasons for the high lead content found in the devilfish meal could be the contamination that affects the water quality. Agriculture, mining, and food processing are the main industrial activities in Mexico, and their wastewater may increase the level of coastal contamination to be discharged without previous treatment. Some studies have reported high cadmium and lead concentrations in commercially important aquatic organisms of the coastal zone of Mexico 41,42 . If devilfish has contact with this contaminated water, it is exposed to the bioaccumulation of heavy metals in its body. On the other hand, devilfish is characterized by being a predatory species 17 , and some authors indicate that especially predatory marine species bioaccumulate heavy metals over www.nature.com/scientificreports/ a long lifetime 43 . So, due to the possibility of representing health risks of the consumption of devilfish meat and the consumers' resistance to integrating it into their diet, an alternative is its use to produce a biocoagulant since its lead content does not represent a health risk during wastewater treatment. The World Health Organization (WHO) establishes a quality reference value of 0.01 mg/L of lead in drinking water 44 , and assuming that the biocoagulant is used at the maximum concentration proposed in this work (800 mg/L), a lead concentration of 0.0001 mg/L would be introduced, being a value 110 times lower than the reference value.</p><p>Water physicochemical characterization before coagulation treatment. Ceramic tile effluent contains significant concentrations of fine suspended particles of clay minerals and dyestuffs or glazes, resulting in colored wastewater and high concentrations of turbidity and suspended solids 45 . Such effluents may severely negatively impact the receiving water quality when discharged untreated. Physicochemical characterization of the samples of ceramic industry wastewater before the coagulation treatment is shown in Table 2.</p><p>The temperature of ceramic industry wastewater (Table 2) complies with the recommended limit (< 25 °C) by the WHO since microorganisms can proliferate at temperatures above 25 °C and may increase problems related to odor, color, taste, and corrosion 44 . The electrical conductivity indicates the salt content or dissolved ionic components in water 46 . Studies of freshwaters suggest that streams supporting well-mixed fisheries must range between 150 and 500 µS/cm. Outside this range could indicate that the water is unsuitable for certain aquatic organisms 47 . The value found in the ceramic industry wastewater is close to or slightly above the range (Table 2). Therefore, it could cause an environmental risk to direct discharge in water bodies. The pH of the ceramic industry wastewater is alkaline (Table 2) and is within the quality range (6.5-8.5) established by the WHO for any purpose 44 .</p><p>The WHO establishes a quality value for turbidity in water < 5 NTU 44 . The turbidity value obtained in this study (Table 2) exceeded ~ 200 times more than the quality parameter. Excessive turbidity in water may increase treatment costs due to problems caused in the flocculation, filtration, and disinfection processes. Likewise, it also makes the sight of the receiving water bodies, where the effluent is being discharged, unpleasant for full-contact recreation 48 . Therefore, ceramic industry wastewater requires a treatment based on coagulation-flocculation processes since it would be difficult to remove the turbidity at parameters < 5 NTU using only a sedimentation process.</p><p>The COD measures the susceptibility to oxidation of the organic and inorganic materials present in water bodies and effluents from wastewater treatment plants. The Mexican water quality guidelines specify COD limit concentrations for the wastewater discharge between 60 and 210 mg/L 49 . The United States Environmental Protection Agency and the German Environment Agency established COD discharge limits of 125 mg/L and 150 mg/L for urban wastewater treatment plants 50 . The COD value of the ceramic industry wastewater (Table 2) is 1.5 times higher than the mentioned upper limit, so it would be expected that the COD-causing substances would be removed by coagulation treatment.</p><p>Water quality guidelines for suspended solids concentrations in freshwater systems have been suggested. Although, some authors note that the reference values are based on the concept of turbidity, being a surrogate measure of the concentration of solids. Hence, they are not often valid 51 . Canadian Council of Ministers of the Environment (CCME) suggests that suspended solids concentrations should not increase by more than 25 mg/L from background levels for any short-term exposure (maximum 1 day) and more than 5 mg/L from background levels for any long-term exposure (between 1 and 13 days) 52 . The TSS value of the ceramic industry wastewater (Table 2) is within both mentioned limits. Therefore, this parameter does not cause a water quality deterioration. However, it was considered necessary due to its relationship with turbidity. Some authors have reported that turbidity is influenced by the shape and particle size of suspended solids, the presence of phytoplankton, and dissolved humic and mineral substances. Hence, a high turbidity reading can be recorded without requiring a high TSS concentration 51 . This condition could be an issue of the low values obtained in this study, linked to the fact that very fine particles are responsible for turbidity and were not retained during the filtration.</p><p>Water physicochemical characterization after coagulation treatment. The physicochemical characterization of the samples of ceramic industry wastewater after the coagulation treatment is shown in Table 3.</p><p>Data recorded in Table 3 revealed that the pH of the wastewater samples slightly changed after the coagulation treatment. The more remarkable changes (< − 1.5) in pH occur when only ferric sulfate or aluminum sulfate is www.nature.com/scientificreports/ used or in the mixture of both chemical coagulants. These results indicate that the biocoagulant is not affecting pH values during the wastewater treatment. Chemical coagulation is a process highly dependent on pH changes. Therefore, the specific operational pH needs to be analyzed for each coagulant type 53 . The optimal pH range for ferric salts is > 5, and aluminum salts range from 6.5 to 7.2 54 . The pH of the wastewater samples was not adjusted to the optimal condition of each coagulant used since the intention was to test the effect of the biocoagulant and the chemical agents without adjusting the pH of the wastewater samples. Some authors have reported that the biocoagulants have the advantage of not being sensitive to pH 55 . This action would facilitate an escalation of the treatment without acidifying or basifying the water to treat. However, the ceramic industry wastewater with a pH above 8 (Table 2) could affect the performance of the chemical agents (ferric sulfate and aluminum sulfate) used in the jar tests, and these chemical agents could acidify the pH to a value out of the recommended limit to discharge treated water. The removal percentages of turbidity, COD, and TSS in Table 3 were taken as the responses in the experimental design. Their analysis and discussion are presented in "ANOVA results and 3D response surface plots".</p><p>ANOVA results and 3D response surface plots. Table 4 shows the ANOVA results obtained for each water quality parameter (turbidity, COD, and TSS) and the final equations in terms of components and the factor (doses) by response type. It can be noted that the models are statistically significant (p value ≤ 0.05), and the lack of fit is not statistically significant (p value ≥ 0.05), so the equations are adequate to describe the observed data. These data were fitted to a combined reduced special cubic x linear model. The special cubic model includes the linear part, double and triple interactions, which are referred to the mixture's proportions as variables, and the linear model includes the effect of the used doses as a numeric factor. The model was reduced to terms that are only significant. Hence, it can be observed that the significant effects were the linear mixture of the three components (coagulants) and some double interactions between the coagulants and the doses for the case of the www.nature.com/scientificreports/ turbidity as the response. The linear mixture of the three components (coagulants) and some double and triple interactions between the coagulants and the doses were the significant effects for COD and TSS responses. The analysis of variance ANOVA in detail for turbidity, COD, and TSS are shown in Tables S1, S2, and S3 (Supplementary materials), respectively. The 3D response surface plots for the removal of turbidity, COD, and TSS in ceramic industry wastewater are shown in Figs. 2, 3, and 4, respectively; being the responses analyzed for three doses: low (200 mg/L), medium (500 mg/L), and high (800 mg/L).</p><p>Considering the turbidity as a quality parameter, if the coagulation treatment was performed with a dose of 200 mg/L, a turbidity removal efficiency of 65.5% is achieved using a mixture of 80% biocoagulant and 20% aluminum sulfate (Fig. 2a). However, using a dose of 500 mg/L or 800 mg/L of ferric sulfate alone, efficiencies of 67.8% and 79.4% are achieved (Fig. 2b,c). Therefore, ferric sulfate has a more significant effect in removing turbidity from ceramic industry wastewater under conditions of doses higher than 500 mg/L. On the other hand, a mixture of 1:4 of aluminum sulfate and biocoagulant could significantly influence the turbidity removal, although not representing one of the best efficiencies (< 65%). Hence, an analysis of the effect on COD and TSS was also performed.</p><p>Regarding COD, if the coagulation treatment was performed using a dose of 200 and 500 mg/L, COD removal efficiencies of 75.1% and 77.4% are achieved with a mixture of 37% biocoagulant and 63% ferric sulfate (Fig. 3a,b). On the other hand, using a dose of 500 mg/L, achieving a COD removal efficiency of 79.5% with a mixture of 42% aluminum sulfate and 58% ferric sulfate is possible. Increasing the doses to 800 mg/L and maintaining the same mixture proportions of both sulfates (Fig. 3c), an efficiency approximated 89% is achieved. Furthermore, using a mixture of 31% biocoagulant and 69% ferric sulfate, an efficiency of 79.9% is achieved. So, a mixture of biocoagulant and ferric sulfate or both sulfates have a more significant effect in removing COD from ceramic industry wastewater.</p><p>Regarding TSS, if the coagulation treatment was performed using a dose of 200 mg/L, a TSS removal efficiency of 98.1% is achieved with aluminum sulfate alone (Fig. 4a). However, it is possible to achieve an efficiency of 94.5% with a mixture of 41% biocoagulant and 59% ferric sulfate. If the dose was increased to 500 mg/L using only aluminum sulfate, an efficiency of 98.5% is achieved. Likewise, with the same proportion of mixture of the biocoagulant (41%) and ferric sulfate (59%), a removal efficiency of 94.7% is obtained (Fig. 4b). Therefore, a dose increase would not be worthwhile when the improvement in efficiency is not so significant. These conditions vary with a coagulant dose of 800 mg/L, where a TSS removal of approximately 100% is achieved with a mixture of 22% biocoagulant and 78% aluminum sulfate or 41% biocoagulant and 59% ferric sulfate (Fig. 4c). Likewise, it is appreciated that increasing the dose of aluminum sulfate has more effect in removing TSS. However, ferric sulfate has a better effect, even when is mixed with the biocoagulant. The optimal doses and the proportions of the mixture of the coagulants will depend on the interest in improving the water quality parameters. So, an optimization analysis was performed considering the maximum removal percentages obtained for the three quality parameters together. The obtained better condition was when the coagulation treatment was performed applying a dose of 800 mg/L with a mixture of 35% biocoagulant and 65% ferric sulfate, achieving removal efficiencies of 74.2%, 79.8%, and 94.3% for turbidity, COD, and TSS, respectively. Increasing the doses to 800 mg/L may not be feasible in terms of costs. However, it should be appreciated that reducing the dose could affect removal efficiencies. A trial was performed in the laboratory under these treatment conditions and with the same methodology, obtaining removal efficiencies of 73.1%, 77.2%, and 94.6% for turbidity, COD, and TSS, respectively. Therefore, an error of less than 3.3% is obtained in the model's prediction.</p><p>Discussion on the effectiveness of the biocoagulant. A biocoagulant can coagulant very fine particles and organic and inorganic matter suspended in water 55 by promoting adsorption, polymer bridging, and charge neutralization mechanisms during the coagulation process 53 . Chitin and polysaccharides are the main characteristics of the shellfish to be used in coagulation processes 56 . The high protein content in devilfish meals (used as biocoagulant) could favor the formation of flocs since its effectiveness could be linked to the charge neutralization through the interaction of two particles with oppositely charged ions and to bridging mechanisms by the forming particle-polymer particle complexes during the adsorption of particles onto polymer chains. Other mechanisms such as electrostatic patch and sweeping, adsorption, complexation, chelation, entrapment, and precipitation may also contribute to the formation of flocs 56 . A comparison between the effectiveness of different biocoagulants applied to the water, and wastewater treatment with the results obtained in this study is shown in Table 5.</p><p>Different biocoagulants have been used for industrial wastewater, natural and synthetic water treatment. Biocoagulants produced from seeds of Jatropha curcas, Moringa oleifera, and Cicer arietinum have efficiently removed the turbidity in water with clay and kaolin (Table 5). However, the biocoagulant produced from devilfish meal is more feasible in dual systems when mixed with ferric sulfate. Therefore, its application could represent an alternative for reducing the use of chemical coagulants, the decrease in sludge generation during the treatment, and the treatment cost as aid coagulant 59 . Based on Table 5, the use of aid coagulants as Cassia obtusifolia seed gum and Ocimum basilicum (basil) have reported improvements in removal efficiency of contaminants in different wastewater types 57,58 , such as the case of this study where the use of the biocoagulant from devilfish improves the removal efficiency of turbidity, COD, and TSS in ceramic industry wastewater. The prices of biocoagulants must be competitive with those of chemical coagulants. The drinking water and wastewater treatment costs using biocoagulants range USD 0.0025-2 and USD 0.015-19.5 per cubic meter of treated water. In contrast, chemical coagulants' water treatment costs are approximately USD 1.50 and USD 0.15-1.80, respectively 53 . Although some biocoagulants show a higher cost than using chemical coagulants, the cost could be reduced when the feedstock comes from waste, and the cost of the resultant sludge handling is also included.</p><p>The aluminum coagulants have the disadvantage of producing a less dense floc than iron coagulants. However, iron coagulants have the disadvantage of increasing up 40% the weight of sludge compared to aluminum coagulants 54 . A high dose (800 mg/L) of ferric sulfate is required to remove efficiently turbidity, COD, and TSS of the ceramic industry wastewater. Hence, one aspect that must be considered is the generation of more sludge. However, when the ferric sulfate is mixed with the biocoagulant, the amount of sludge could be reduced. The quantification of sludge generated was not estimated during the jar tests, so it is suggested to be evaluated in the future. Furthermore, applying the coagulation treatment in the ceramic industry wastewater requires adequate disposal of the generated sludge in landfills. Although, this waste could also be valorized as fill material in the construction or as raw material in the sanitary ware production 60 .</p><p>Yonge 61 reported USD 0.20-0.30/kg costs for ferric sulfate coagulant. Salas-Razo et al. 62 estimated a supply market to the price of USD 0.07/kg of devilfish and a cost of production of silage acid of devilfish in USD 0.14/ www.nature.com/scientificreports/ kg. This last price could be associated with the production cost of devilfish meals. Although the production cost of the produced biocoagulant was not calculated in this work, and it could be subjective, the values shown reflect that the biocoagulant from devilfish meal could represent a lower-cost alternative material than conventional coagulants. Finally, applying dual systems reduce the treatment cost and the use of chemical coagulants. Based on the optimization results obtained in this study, the treatment cost in the coagulation process could reduce when 280 mg/L of biocoagulant and 520 mg/L of ferric sulfate instead of 800 mg/L of ferric sulfate are used in the ceramic industry wastewater. Hence, the biocoagulant from devilfish could represent a biodegradable and lowcost material in wastewater treatment.</p><!><p>This study produced a biocoagulant from devilfish meal, an invasive species that cause socioeconomic and ecological impacts, and its meal could have heavy metals putting human health at risk if was consumed. The effectiveness of the biocoagulant and two chemical coagulants (ferric sulfate and aluminum sulfate) was evaluated for the removal of turbidity, COD, and TSS from ceramic industry wastewater using a combined experimental design of Mixture-Process. This design optimized the operation conditions of the coagulation processes found that removal efficiencies of 74%, 79%, and 94% for turbidity, COD, and TSS, respectively, are achieved using an optimal coagulant dose of 800 mg/L with a mixture of 35% biocoagulant and 65% ferric sulfate. ANOVA results showed that the models are significant (p value < 0.0001), and the lack of fit is not required (p value > 0.05). Hence, the experimental data were fitted to a combined reduced special cubic x linear model, having the model prediction an error of 3.3% concerning observed results. These results support the use of devilfish meal as a biocoagulant, being more feasible in dual systems when mixed with ferric sulfate.</p>

Scientific Reports - Nature

Distinctive properties of Arabidopsis SUMO paralogues support the in vivo predominant role of AtSUMO1/2 isoforms

Protein modification by SUMO (small ubiquitin-related modifier) has emerged as an essential regulatory mechanism in eukaryotes. Even though the molecular mechanisms of SUMO conjugation/deconjugation are conserved, the number of SUMO machinery components and their degree of conservation are specific to each organism. In the present paper, we show data contributing to the notion that the four expressed Arabidopsis SUMO paralogues, AtSUMO1, 2, 3 and 5, have functionally diverged to a higher extent than their human orthologues. We have explored the degree of conservation of these paralogues and found that the surfaces involved in E1-activating enzyme recognition, and E2-conjugating enzyme and SIM (SUMO-interacting motif) non-covalent interactions are well conserved in AtSUMO1/2 isoforms, whereas AtSUMO3 shows a lower degree of conservation, and AtSUMO5 is the most divergent isoform. These differences are functionally relevant, since AtSUMO3 and 5 are deficient in establishing E2 non-covalent interactions, which has not been reported for any naturally occurring SUMO orthologue. In addition, AtSUMO3 is less efficiently conjugated than AtSUMO1/2, and AtSUMO5 shows the lowest conjugation level. A mutagenesis analysis revealed that decreases in conjugation rate and thioester-bond formation are the result of the non-conserved residues involved in E1-activating enzyme recognition that are present in AtSUMO3 and 5. The results of the present study support a role for the E1-activating enzyme in SUMO paralogue discrimination, providing a new mechanism to favour conjugation of the essential AtSUMO1/2 paralogues.

distinctive_properties_of_arabidopsis_sumo_paralogues_support_the_in_vivo_predominant_role_of_atsumo

INTRODUCTION<!>Cloning, expression and protein purification<!>Yeast two-hybrid experiments<!>In vitro pull-down assay<!>Polyclonal anti-AtSUMO1 antibody production<!>In vitro SUMO conjugation, polymeric chain formation and E1-thioester assays<!>SUMO-conjugation rate quantification<!>Bioinformatics<!>Accession numbers<!>Conservation of E1-, E2- and SIM-interacting residues among AtSUMO paralogues<!>E2 non-covalent interactions with SUMO isoforms<!>Identification and validation of AtCAT3 as a bona fide SUMO target in vitro<!>AtSUMO1 Asp63 is necessary for efficient polySUMO chain formation<!>SUMO isoforms display a distinct conjugation rate in vitro<!>Non-conserved residues involved in the E1 interaction are responsible for differences in the SUMO paralogue conjugation rate<!>DISCUSSION

<p>In plants, regulation of protein activity by SUMO (small ubiquitin-related modifier) attachment is a post-translational modification that has been shown to be essential during seed development and to have a major role in abiotic and biotic stress responses [1]. A common property between plants and animals is that the SUMOylation system appears to be a target for pathogenic effectors [2-5], as well as the accumulation of SUMO conjugates in response to heat and oxidative stresses [6,7]. But there are also biological processes specific to plants in which SUMO has a relevant role, such as flowering [8], phosphate starvation [9], drought responses [10] and the abscisic acid signalling pathway [11,12], a hormone that mediates plant responses to environmental stresses and a key regulator of plant growth and development.</p><p>SUMO is covalently attached to target proteins by the sequential action of E1-activating, E2-conjugating and E3-ligase enzymes [13]. SUMO activation is mediated by a heterodimeric enzyme consisting of a large subunit, SAE (SUMO-activating enzyme) 2, and a small subunit, SAE1. The SAE2 subunit contains the adenylation, catalytic cysteine, UFD (ubiquitin fold) and C-terminal functional domains [14]. The adenylation domain is responsible for SUMO recognition and adenylation of the C-terminus of SUMO. After adenylation, the catalytic cysteine thiol group attacks the SUMO C-terminal adenylate resulting in the formation of a thioester bond between the E1 and SUMO, in a mechanism that involves a rotation of the cysteine domain [15]. At this stage, SUMO can be transferred to the E2-conjugating enzyme. SUMO-charged E2 is competent to transfer SUMO to the target lysine residue in the substrate, although this reaction is facilitated by E3-ligase enzymes in vitro and in vivo [16,17].</p><p>During conjugation, SUMO molecules establish non-covalent interactions with the E1-activating and the E2-conjugating enzymes. Structural studies have identified eleven residues on the HsSUMO (human SUMO) surface that establish contacts with the E1-activating enzyme and which, presumably, are responsible for conferring modifier specificity [14]. SUMO can be attached to the target protein as a monomer or polymer, and polySUMO chains have been shown to act as signals to promote ubiquitination of the SUMO-modified substrate targeting it for proteasomal degradation [18]. SUMO chain growth is dependent on the presence of a SUMOylation consensus site at the SUMO N-terminal tail, and polymerization is facilitated by non-covalent interactions between SUMO and the E2-conjugating enzyme [19,20]. In vitro, the chain length is modulated by the relative abundance of HsSUMO2/3, which can build SUMO chains, compared with HsSUMO1, which does not have a SUMOylation consensus site and that could function as a chain terminator [21].</p><p>The consequences of covalent SUMO attachment to target proteins are very variable and include regulation of subcellular localization, protein activity and stability, and protein–protein interactions. At the molecular level, the SUMOylation outcome is achieved through the interaction with specific effectors that contain a SIM (SUMO-interaction motif). Most SIMs consist of a hydrophobic core of three to four aliphatic residues flanked by acidic residues [22,23]. Structural and functional studies determined that a hydrophobic groove surrounded by basic residues in SUMO is crucial for SIM interaction. Interestingly, the relative position between the hydrophobic groove and the basic residues differ among HsSUMO isoforms and this could confer SUMO–SIM interaction specificity [24,25]. During conjugation, SUMO paralogue selection can be mediated by SIM-dependent recruitment of targets to SUMO thioester-charged E2 and/or SUMO-modified E2 [26,27]. Moreover, SIMs have been identified in E3 ligases and shown to regulate ligase activity and localization [28–30].</p><p>In plants, much less is known about the molecular mechanisms that regulate SUMOylation, and the complexity of the SUMOylation components is apparently higher when compared with other organisms. In Arabidopsis, expression has been detected for the four SUMO paralogues AtSUMO1, AtSUMO2, AtSUMO3 and AtSUMO5. Among them AtSUMO1 and 2 are the most closely related isoforms. Previous studies have shown that Arabidopsis SUMO paralogues do not serve as equivalent substrates of AtSUMO proteases, referred to as ULP (ubiquitin-like protein-specific protease). The four proteases AtULP1a, c, d and AtESD4 (Arabidopsis early is short days 4) displayed similar peptidase and isopeptidase activities towards AtSUMO1 and AtSUMO2 isoforms, although none of them showed a significant activity towards AtSUMO5. Only AtULP1a exhibited a poor peptidase activity towards AtSUMO3 and no isopeptidase activity at all [31–33]. In addition, in non-quantitative assays, only AtSUMO1, 2 and 3 were shown to be conjugated to the yeast substrate PCNA (proliferating-cell nuclear antigen), and the capacity to form polymeric chains was displayed exclusively by AtSUMO1 and AtSUMO2 [32]. At first glance, it seemed that the situation of the AtSUMO1 and 2 isoforms resembles that of the human SUMO2 and 3 isoforms according to their capacity to form polymeric chains, as opposite to the HsSUMO1 and the AtSUMO3 isoforms that are conjugated as monomers. But, in contrast, homology studies failed to cluster human and Arabidopsis isoforms according to their ability or inability to polymerize [11,32]. In addition, AtSUMO3 and 5 are not capable of complementing the lethal double-mutant atsumo1atsumo2 plants [34], whereas SUMO1-knockout mice are viable. These results suggest that mammalian SUMO1 and SUMO2/3 have partially redundant functions [35,36], in contrast with AtSUMO paralogues that seem to have developed more divergent functions [34,37].</p><p>To gain new insights into the complex Arabidopsis SUMOylation system, we have assessed whether AtSUMO1, 2, 3 and 5 have distinct molecular properties that might influence their in vivo conjugation and biological function. We have found that Arabidopsis SUMO isoforms have heterogeneous properties at different molecular levels. AtSUMO1 and 2 were competent to interact non-covalently with their cognate E2-conjugating enzyme AtSCE1 (Arabidopsis SUMO-conjugating enzyme 1), whereas AtSUMO3 and AtSUMO5 did not conserve this property. Mutagenesis analysis revealed that the single residue Asp63, conserved in the AtSUMO1/2 surface but not in AtSUMO3 and AtSUMO5, is essential for non-covalent interactions with E2, and that it is necessary for polySUMO chain formation. It is even more significant the fact that SUMO isoforms differed in their conjugation rate, AtSUMO1/2 being the most efficiently conjugated paralogues, AtSUMO3 was less efficiently conjugated and AtSUMO5 showed the lowest conjugation level. A mutagenesis analysis showed that the lower conjugation rates of AtSUMO3 and AtSUMO5 were related to changes in the SUMO residues involved in the E1 interaction, which also affected thioester-bond formation. These results suggest that the first step in the SUMO-conjugation cascade would have a regulatory role in SUMO paralogue discrimination. Overall, the results of the present study suggest that AtSUMO1/2 might be the most efficiently conjugated SUMO isoforms in vivo, and we postulate that this could constitute a molecular mechanism to assure conjugation of the essential AtSUMO1/2 paralogues compared with the non-essential AtSUMO3 and 5.</p><!><p>AtSUMO isoforms, AtSAE2, AtSAE1a/b and AtCAT3 (Arabidopsis catalase isoform 3) were amplified by PCR from cDNA obtained from 2-week-old plants grown on MS (Murashige and Skoog) plates under LD (light/dark) at 22°C (Superscript® III reverse transcriptase from Invitrogen and Pfu DNA polymerase from Stratagene). AtSAE1a/b were cloned into pET15b (Novagen) to encode a native polypeptide, and AtSAE2, AtSCE1 [11], AtSUMO1-(1–93), AtSUMO2-(1–92), AtSUMO3-(1–93) and AtSUMO5-(1–103) were cloned into pET28a (Novagen) to encode an N-terminal His6-fusion protein. AtCAT3Ct-(419–472) (Ct is C-terminal tail) was cloned into pGEX-6P1 (Amersham) to encode an N-terminal GST (glutathione transferase)-fusion protein. Plasmids were transformed individually, or co-transformed in the case of AtSAE2- and AtSAE1a/b-containing plasmids, into Escherichia coli strain BL21 Codon Plus RIL (Stratagene). Cultures (1–4 litres) were incubated at 37°C until they reached an A600 of 0.6–0.8, and protein expression was induced by adding 0.1 mM IPTG (isopropyl β-D-thiogalactopyranoside) for 4 h at 30°C. Cells were harvested and resuspended in 50 mM Tris/HCl (pH 8.0), 20% (w/v) sucrose, 350 mM NaCl, 20 mM imidazole, 0.1% Nonidet P40, 1 mM PMSF, 1 mM 2-mercaptoethanol, 1 μg/ml leupeptin, 1 μg/ml pepstatin and 50 μg/ml DNAse. Protein extracts were prepared by sonication [five pulses of 1 min and 2–3 power setting (Sonifier 250, Branson)] and clarified by centrifugation (18000 g for 30 min at 4°C). Purification via IMAC–Sepharose resin (GE Healthcare) or glutathione–Sepharose (GE Healthcare) was performed according to the manufacturer's instructions. SDS/PAGE analysis of the purified proteins is shown in Supplementary Figure S1 (at http://www.BiochemJ.org/bj/436/bj4360581add.htm).</p><!><p>Yeast expression constructs pGBKT7:AtSCE1/AtUBC10 and pGADT7:AtSUMO1-(1–93)/AtUBI have been generated previously [11]. AtSUMO3-(1–93) and AtSUMO5-(1–103) were cloned into pGADT7 AD (Clontech) to encode an N-terminal GAL4-activation domain fusion protein. AtSUMO1 and AtSUMO3 mutant alleles were generated by QuikChange® site-directed mutagenesis (Stratagene). Plasmids, as indicated, were co-transformed into the yeast strain HF7c using the lithium acetate method as described in the Clontech Yeast Protocols Handbook. Transformed yeast culture was plated on to permissive SD (synthetic dextrose) medium complemented with histidine. A single clone per transformation was selected, disaggregated by vortex agitation in SD medium without amino acids, and serial dilutions were performed (1, 1:8, 1:32 and 1:64). Aliquots (5 μl) of each dilution were sowed on non-selective (SD medium complemented with histidine) or selective (SD medium not complemented with histidine) plates. After incubation for 2 days at 30°C, protein interactions were analysed using histidine auxotrophy as a selective marker.</p><!><p>His–AtSCE1 (100 μM) and AtSUMO1/D63N (25 μM) were incubated in 40 μl of binding buffer [20 mM Tris/HCl (pH 8.0), 50 mM NaCl and 20 mM imidazole] for 5 h at 4°C. Next, 10 μl of Ni2+-IMAC–Sepharose resin was added to the binding mixture, and the mixture was incubated for 30 min at 4°C. The binding mixture was transferred to micro bio-spin chromatography columns (Bio-Rad) and the resin was washed four times with 20 μl of binding buffer. The proteins bound to the resin were eluted with 20 μl of binding buffer containing 300 mM imidazole. The input (0.8 μl) and eluate (3.5 μl) fractions were separated by SDS/PAGE and either stained with Coomassie Fluor Orange (Molecular Probes, C-33250) or subjected to immunoblot analysis with anti-SUMO1 antibodies, as indicated.</p><!><p>Polyclonal antibodies were raised against the purified N-terminal His6-fusion protein of AtSUMO1-(1–93). Purified protein (1 mg) was resolved by SDS/PAGE, the gel was stained with Coomasie Blue and the gel slice containing His–AtSUMO1 was used to immunize rabbits (Cocalico Biological). A 1:1000 dilution of the serum produced was used in the immunoblot analyses.</p><!><p>In conjugation assays, we used the C-terminal tail of the AtCAT3 (residues 419–472) fused to GST, GST–AtCAT3Ct. Reactions were carried out at the indicated temperatures in 25 μl reaction mixture volumes containing 1 mM ATP, 50 mM NaCl, 20 mM Hepes (pH 7.5), 0.1% Tween 20, 5 mM MgCl2, 0.1 mM DTT (dithiothreitol), 2 μM SUMO, 0.5 μM AtSAE2/AtSAE1a, 0.5 μM AtSCE1 and 5 μM GST–AtCAT3Ct. After the specified incubation time, reactions were stopped by the addition of protein-loading buffer, boiled for 5 min and 10 μl aliquots were resolved by SDS/PAGE. Polymeric chain formation reactions were performed at 37°C in the same reaction buffer as SUMO-conjugation assays and in the presence of 100 μM AtSUMO1, 1 μM AtSAE2/AtSAE1a and 10 μM AtSCE1. Reaction products were detected by immunoblot analysis with anti-GST polyclonal anti-bodies (Sigma, G7781) or with anti-AtSUMO1 polyclonal antibodies, as indicated. E1-thioester assays were performed at 30°C in 50 μl reaction mixture volumes containing 1 mM ATP, 50 mM NaCl, 20 mM Hepes (pH 7.5), 0.1% Tween 20, 5 mM MgCl2, 0.1 mM DTT, 10 μM SUMO and 5 μM AtSAE2/AtSAE1a. At the indicated time points, 15 μl aliquots were removed and analysed by SDS/PAGE followed by Coomasie Fluor Orange staining according to manufacturer's protocol (Molecular Probes, C-33250). As a thioester-bond formation control, an aliquot of each reaction was treated with 100 mM DTT before loading on to polyacrylamide gels.</p><!><p>Reaction products were detected using ECL (enhanced chemiluminescence) and Western blot detection reagents (GE Healthcare), and the signal was acquired with the LAS-3000 imaging system and quantified with Multi Gauge V3.0 (Fujifilm). Signals were normalized against known amounts of GST included in each blot. When data are represented by relative units, SUMO-conjugation or E1-thioester rates are referred to the average value calculated using all the rates obtained in each independent experiment.</p><!><p>Sequence alignments were performed using the protein multiple alignment software MUSCLE [38] and alignments were edited with GeneDoc software (www.psc.edu/biomed/genedoc). Protein structure models were generated using the SWISS-MODEL workspace [40] on automated mode. AtSUMO1/3/5 and AtSCE1 models were generated using PDB code 2PE6 [2.40 Å (1 Å = 0.1 nm)] or PDB code 2IY1 (2.46 Å) as templates. Models were assembled and images were generated using PyMOL (http://www.pymol.org).</p><!><p>The assigned accession numbers for the genes studied are as follows: At4g26840 (AtSUMO1), At5g55160 (AtSUMO2), At5g55170 (AtSUMO3), At2g32765 (AtSUMO5), At2g21470 (AtSAE2), At4g24940 (AtSAE1a), At5g50580 (AtSAE1b), At3g57870 (AtSCE1), At4g02890 (AtUBI), At5g53300 (AtUBC10) and At1g20620 (AtCAT3).</p><!><p>The overall degree of homology among SUMO paralogues ranges from 83% sequence identity between AtSUMO1 and AtSUMO2, to 42% and 30% sequence identity between AtSUMO1 and AtSUMO3 or AtSUMO5 respectively. Interestingly, these differences are also present in the degree of conservation found between residues involved in E1 and E2 non-covalent interactions. AtSUMO1 and AtSUMO2 have identical amino acid residues at the positions involved in non-covalent interactions with the E1-activating and E2-conjugating enzymes. In contrast, AtSUMO3 and AtSUMO5 have more divergent interacting surfaces. AtSUMO3 has 55% and 75% degree of conservation of E1- and E2-interacting residues respectively, whereas AtSUMO5 is the most divergent isoform, showing a degree of conservation of 36% and 56% for E1- and E2-interacting residues (Figures 1A and 1B, and Supplementary Figure S2 at http://www.BiochemJ.org/bj/436/bj4360581add.htm). Previous structural studies identified 11 residues in the HsSUMO1 surface involved in E1 interactions [14]. Taken as reference positions in AtSUMO1, the four residues, Gln25, Gly27, Gly92 and Gly93, are identical in all Arabidopsis paralogues. Among the others, two are divergent only in the AtSUMO5 isoform, Arg66 and Asp85, and five are not conserved in either AtSUMO3 or AtSUMO5, Asn56, Met87, His89, Gln90 and Thr91 (Figures 1A and 5A). All of them are identical between AtSUMO1 and 2. With regard to the residues involved in E2 non-covalent interactions, those residues establishing lateral chain contacts with the E2-conjugating enzyme are identical across AtSUMO1/2, HsSUMO2/3 and yeast ySmt3, except for the residue Met87 in AtSUMO1, which is highly variable (Figures 1B and 2A). The deduced consensus motif considering the most frequent residues would be Asp63/Glu63, Glu79, Asp82 and Asp85/Glu85 (the residue numbering is for AtSUMO1). When we analysed AtSUMO3, it is remarkable that the acidic Asp63/Glu63 residue in the consensus sequence is substituted with a polar asparagine residue. This change is also present in the most divergent SUMO isoform AtSUMO5 that, in addition, has substitutions in Asp77/Glu77 and Asp85/Glu85 by histidine and cysteine respectively (Figures 1B and 2A).</p><p>Another important interacting surface on SUMO involves the second β-sheet and the downstream α-helix, which form a hydrophobic groove flanked by basic residues that accommodates SIMs [42]. Structural studies have determined that aliphatic and aromatic residues constitute this hydrophobic groove, HsSUMO1 Ile34, His35, Phe36, Val38, Leu47 and Tyr51 [23]. Whereas functional studies identified residues required for the role of HsSUMO2 in transcriptional inhibition, which include the hydrophobic Val30 and Ile34 (equivalent to Ile34 and Val38 in HsSUMO1), the polar Thr38 (Thr42 in HsSUMO1), and the four basic residues Lys33, Lys35, Lys42 and Arg50 (Lys37, Lys39, Lys46 and Arg54 in HsSUMO1) [24]. Among the basic residues, HsSUMO1 Lys39 has been proposed to interact with phosphorylated residues located next to the hydrophobic core, which is the essential component of SIMs [22]. The previous functional amino acids identified in HsSUMO1 and HsSUMO2 are conserved in AtSUMO1/2, suggesting that these paralogues will share the molecular basis for the SIM interaction as their human orthologues. In contrast, major changes are present in AtSUMO3 and AtSUMO5. The hydrophobic Val30 and Ile34 residues shown to be necessary for transcriptional repression in HsSUMO2 are substituted by an acidic residue in AtSUMO5 (Asp40) and a polar residue (Asn34) in AtSUMO3 respectively. In addition, the polar Thr38 in HsSUMO2, which is also necessary for transcriptional repression, is substituted by the hydrophobic alanine residue in both AtSUMO3 and AtSUMO5 (positions 38 and 48 respectively). Finally, the basic Lys39 in HsSUMO1 proposed to interact with phosphorylated residues in the target is substituted by an uncharged glycine residue in AtSUMO5 (Figure 1C). Considering that no SIM-containing targets have been identified in Arabidopsis, we have focused on the functional analysis of E1 and E2 non-covalent interacting residues in Arabidopsis SUMO paralogues, according to their role in conjugation.</p><!><p>In order to assess the effect of changes in residues involved in SUMO–E2 non-covalent interactions, we performed yeast two-hybrid assays. In these experiments, the capacity of the yeast strain HF7c to grow in the absence of histidine was used as a marker for the interaction between proteins. Previous studies have shown that AtSUMO1 and 2 were capable of interacting with AtSCE1 in similar assays [11]. We found that histidine auxotrophy was restored only when AtSCE1a was co-transformed with AtSUMO1, but not with AtSUMO3 or AtSUMO5. In these experiments, AtSUMO1 was used as a positive control and Arabidopsis ubiquitin was used as a negative control. When AtSCE1 was replaced by the ubiquitin-conjugating enzyme AtUBC10, we observed that histidine auxotrophy was conferred only when ubiquitin was co-expressed, consistent with the specificity of the system (Figure 2B). The results of the present study demonstrate that AtSUMO3/5 are not competent to interact with AtSCE1, suggesting the existence of a change in the SUMO-interacting surface that might be common to both paralogues. As described above, the central aspartate residue in AtSUMO1 and 2 (Asp63 and Asp62) is replaced by an asparagine residue at the equivalent position in AtSUMO3 and 5 (Asn63 and Asn73) (Figures 1B and 2A). To test the role of this divergent residue, we generated the mutants AtSUMO1/D63N and AtSUMO3/D63N and assessed their capacity to interact with AtSCE1. None of these mutant isoforms interacted with AtSCE1 in yeast two-hybrid assays (Figure 2C), indicating that the presence of an aspartate residue at position 63 is essential, but not sufficient, for SUMO–E2 non-covalent interactions. To further evaluate the essential role of Asp63 in non-covalent interactions with E2, we performed in vitro pull-down assays. In these assays, His–AtSCE1 was incubated in the presence of AtSUMO1 or AtSUMO1 D63N. After binding to a Ni2+ -charged resin, His–AtSCE1 was eluted with imidazole and we observed that AtSUMO1 was co-eluted with His–AtSCE1. In contrast, a small amount of AtSUMO1/D63N was present in the elution fraction to the same extent as in the negative control, where His–AtSCE1 was omitted.</p><!><p>In order to reconstitute a complete Arabidopsis SUMOylation system that allowed us the biochemical characterization of the SUMO paralogues, we aimed to identify an endogenous SUMO target. Since SUMOylation is involved in oxidative stress responses, we analysed whether oxidative stress scavengers could be SUMO targets. We found that AtCAT3 contained a SUMOylation consensus site at its C-terminal domain (Supplementary Figure S3A at http://www.BiochemJ.org/bj/436/bj4360581add.htm). Previous studies have determined that SUMO-conjugation sites are located in an extended structure on the surface of the target protein in order to be accessible to the SUMOylation machinery [43]. To determine the position of the putative AtCAT3 SUMOylation site on the quaternary structure, we performed the AtCAT3 structure prediction using the Exiguobacterium oxidotolerans catalase structure as a template (PDB code 2J2M). The model generated model indicated that the putative SUMO acceptor lysine residue Lys423 in AtCAT3 is fully exposed at the protein surface (Supplementary Figures S3B and S3C). Validation of AtCAT3 as a SUMO substrate was performed by in vitro SUMOylation reactions containing the AtCAT3 C-terminal domain, which includes the predicted SUMOylation site (GST–AtCAT3Ct; Supplementary Figure S3D), in the presence of the reconstituted Arabidopsis SUMOylation system, AtSAE2, AtSAE1a, AtSCE1 and AtSUMO1. As a result, we detected SUMO conjugation to AtCAT3Ct in an ATP-dependent manner. In addition, the mutant AtCAT3Ct/K423R was unable to accept SUMO (Supplementary Figure S3D). These results validate AtCAT3 as a SUMO target and identify Lys423 as the acceptor site for SUMO modification. The advantage of using AtCAT3Ct as a substrate for in vitro reactions as opposed to other targets described in the literature is that it does not require the presence of an E3 ligase in order to be modified, which simplifies the biochemical analysis of SUMO conjugation.</p><!><p>Since AtSUMO1/D63N prevented AtSCE1 non-covalent interactions, we tested whether this mutation affected polySUMO chain formation. In vitro polySUMO chain-formation assays were performed in the presence of AtSAE2/AtSAE1a, AtSCE1, and the native or mutated SUMO form. Under these conditions, the native AtSUMO1 isoform efficiently built polymeric chains and it was also conjugated to AtSCE1, on which polySUMO chains were also formed. When Asp63 was replaced by an asparagine residue, a reduction in polySUMO chain formation was observed. This defect was more evident from the second conjugation cycle, independently of whether polySUMO chains in the strict sense or polySUMO chains built on AtSCE1 were analysed (Figures 3A and 3B). These results indicate that the mechanism to build polySUMO chains is conserved in Arabidopsis and that, presumably, the naturally occurring E2-non-interacting SUMO isoforms AtSUMO3 and AtSUMO5 will not interfere with polyAtSUMO1/2 chain formation in vivo. Consistent with a main role for SUMO–E2 non-covalent interactions in polySUMO chain formation, the mutant AtSUMO1/D63N was conjugated to the target AtCAT3Ct with the same efficiency as the native AtSUMO1 (Figure 3C).</p><!><p>Since AtSUMO isoforms show differences in their capacity to establish non-covalent interactions with their cognate E2-conjugating enzyme, we explored the possibility that they could also differ in their conjugation rates. First, we chose a short incubation time, 10 min, in order to compare the first conjugation cycle. In this way, we avoided conjugation rate underestimation of the isoforms competent for polySUMO chain formation. We were also interested in analysing the effect of the incubation temperature on conjugation rate, which could be biologically relevant since SUMO conjugates accumulate massively upon heat stress. Reaction mixtures were incubated at 22, 37, 42 and 48°C, and we observed that the highest reaction rate was achieved at 42°C (Figure 4A). Under these experimental conditions, we did not observe conjugation of AtSUMO5. All other isoforms were conjugated to AtCAT3 at different rates. In general, AtSUMO1 and 2 were better conjugated than AtSUMO3, the highest difference being observed at 42°C. At this temperature, the AtSUMO1 and AtSUMO2 conjugation rate is 2.4- and 3.2-fold higher than AtSUMO3 respectively (Figure 4B). Next, we performed a time-course analysis that allowed us to detect AtSUMO5 conjugation after 60 min incubation at 37°C. At 42°C, AtSUMO5 conjugation was very weak, and differences in the conjugation level between AtSUMO1 and AtSUMO5, as well as between AtSUMO1 and AtSUMO3, were more pronounced than at 37°C (Figure 4C).</p><!><p>As has been described above, SUMO residues involved in E1 recognition are not evenly conserved between AtSUMO isoforms. In order to evaluate the effect of these changes in conjugation rate, we focused on those positions that were not conserved only in the less conjugated isoform AtSUMO5, which corresponds to AtSUMO1 Arg66 and Asp85. We also focused on AtSUMO1 His89 which is not conserved either in AtSUMO3 or AtSUMO5, and whose equivalent position in Nedd8 has previously been reported to confer modifier specificity [44]. We mutated residues Arg66, Asp85 and His89 of AtSUMO1 to those present in AtSUMO3 and AtSUMO5 at the equivalent positions, and analysed their conjugation rates (Figures 5B and 5C). At 42°C, the most dramatic effect was observed in D85C followed by H89A and R66E mutants, which showed 24%, 40% and 50% of the native AtSUMO1 rate respectively. The H89E substitution had a smaller effect and showed 90% of the native AtSUMO1 activity. Similar to what we observed during the conjugation analysis of SUMO isoforms, at 37°C differences in conjugation levels were smaller, although their behaviour was similar, and the D85C mutation was the most affected. These mutants, with the exception of the H89E substitution, also showed SUMO–E1 thioester-bond formation defects, and the D85C mutation was the most affected. These results suggest that the conjugation level reduction of these mutants was the result of E1-interaction defects.</p><!><p>The molecular consequences of protein modification by SUMO will be dependent on the specific SUMO paralogue that is attached to the target, and major effort has been put into elucidating the mechanisms involved in SUMO paralogue specificity. Among these mechanisms, ULP-mediated deconjugation and SIM-mediated conjugation have been proposed to facilitate SUMO paralogue selection [45]. In the present study we have described results supporting a role of the E1-activating enzyme in SUMO paralogue discrimination in Arabidopsis. The analysis of the molecular properties of the four AtSUMO isoforms expressed indicates that they have diverged to a higher degree than their human orthologues, and that the essential AtSUMO1 and AtSUMO2 are the most functionally conserved isoforms.</p><p>Non-covalent interactions between SUMO and its cognate E2-conjugating enzyme is an intrinsic property of the system. Initial studies showed that this property is conserved in AtSUMO1 and 2 [11] but, surprisingly, our results showed that AtSUMO3 and 5 do not retain the capacity to interact with their cognate E2, AtSCE1. This situation is unique to the Arabidopsis system, since all Hs-SUMO isoforms and yeast Smt3 interact efficiently with their cognate E2-conjugating enzyme. We have also identified AtSUMO1 Asp63 as an essential residue for establishing E2 non-covalent interactions. But the fact that AtSUMO3 Asn63 substitution with aspartate did not confer competence for this interaction suggests that the inability of AtSUMO3/5 to interact with AtSCE1 could be the result of two types of amino acid changes: the loss of essential residues and the appearance of residues that would be detrimental for this interaction. SUMO–E2 non-covalent interactions have been proposed to be involved in SUMO chain formation of human SUMO2 and 3 isoforms, which also contain a SUMO attachment site in their N-terminal tail [19,20]. In this model, HsSUMO1 would function as a chain terminator, since it does not comprise an acceptor lysine residue for another SUMO molecule, whereas it retains the capacity to interact non-covalently with E2 enzyme. In Arabidopsis, a SUMOylation consensus site is only present in those paralogues capable of interacting non-covalently with AtSCE1, which are AtSUMO1/2, correlating with their ability to form polymeric chains. On the other hand, considering their inability to interact with AtSCE1, it is not clear whether AtSUMO3/5 would have a role as polyAtSUMO1/2 chain terminators, as has been proposed for the HsSUMO1, suggesting that other molecular mechanisms might regulate polySUMO chain length in Arabidopsis. Supporting this, recent proteomic studies have failed to identify AtSUMO3/5 peptides in purified AtSUMO1 conjugates [46].</p><p>Another remaining question to be addressed was whether AtSUMO paralogues also differed in their conjugation levels. In addition, since SUMO conjugates accumulate dramatically upon heat stress, we were interested in studying the effect of the temperature in conjugation reactions. In previously reported assays, experiments were designed in such a manner that non-quantitative results were obtained (in most cases the incubation time ranged from a few hours to overnight) [32,47]. The results of the present study, using a quantitative SUMOylation assay, showed that AtSUMO1 and 2 isoforms were more efficiently conjugated in comparison with AtSUMO3, whereas AtSUMO5 showed the lowest conjugation level. For the SUMO isoforms tested, with the exception of AtSUMO5, the conjugation rate increased with temperature, and it was striking to observe that the highest activity occurred at 42°C. Even though this temperature is higher than standard environmental conditions, it highlights the robustness of the SUMO-conjugation system. At the same time it suggests that the massive and rapid SUMO conjugation observed in plants upon heat-shock treatments could be mediated, at least in part, by the increasing activity of the conjugation system with temperature. In addition, this effect is more pronounced in the case of AtSUMO1 and AtSUMO2 conjugation, which are the isoforms that are greatly conjugated under heat stress.</p><p>Differences in conjugation efficiency among SUMO paralogues are also specific to the Arabidopsis SUMOylation system, since all three human SUMO isoforms have been shown to form E1- and E2-thioester bonds and be conjugated to the substrate RanGAP with the same efficiency [14], suggesting that, in this case, SUMO paralogue selection will be dependent on availability or SIM-mediated interactions. The conservation analysis of the 11 SUMO residues involved in the E1 adenylation domain interaction [14] showed that five and seven residues are not conserved in AtSUMO3 and AtSUMO5 respectively, suggesting that adenylation could be deficient for these paralogues. A mutagenesis analysis revealed that some of these substitutions were responsible for a lower conjugation rate, although the effect was dependent on the nature of the substitution. The most deleterious was the D85C substitution present in AtSUMO5, supporting the previous result indicating that AtSUMO5 was the less-conjugated isoform. Interestingly, this position in HsSUMO1, Glu89, was previously shown to be crucial for E1-thioester formation [19], supporting our results pointing to a critical role of this residue in SUMO–E1 recognition. As to AtSUMO3, the only substitution tested occurs at a position previously proposed to contribute to modifier discrimination [44] and, when introduced in AtSUMO1, it reduced the conjugation rate to a value equivalent to that of AtSUMO3, suggesting that this substitution has a major contribution in AtSUMO3 conjugation-rate reduction. Conjugation defects in these mutants were more pronounced when reaction mixtures were incubated at higher temperatures, similar to what we observed when comparing SUMO paralogues. Moreover, the reduction in conjugation correlated with a reduction in E1-thioester formation, supporting a function of the residues tested in SUMO recognition by the E1. To our knowledge, this natural occurring SUMO paralogue discrimination by the E1-activating enzyme has not been reported previously in any other system.</p><p>Among the different studies that have addressed characterization of the Arabidopsis SUMOylation system, in vitro AtSUMO5 conjugation has only been detected to a mammalian substrate, RanGAP, in the presence of the mammalian conjugation system [31]. It is well established that SUMO conjugation is highly regulated by specific protein–protein interactions, and using heterologous systems to test SUMO conjugation might give results that could not be relevant in vivo. This could explain why AtSUMO5 conjugation was facilitated by the SUMO mammalian system, but when we have used the Arabidopsis system, which presumably is more selective, AtSUMO5 was very poorly conjugated. Similarly, a previous report showed that HsSUMO1 was able to interact with the Arabidopsis ubiquitin-conjugating enzyme AtUBC10 in yeast two-hybrid assays. In contrast, in the same assays, neither AtSUMO1 nor AtSUMO2 interacted with AtUBC10, suggesting that selective interactions within the SUMO pathway are more permissive when heterologous SUMOylation components are used [11], and highlighting the importance of using homologous systems in biochemical studies.</p><p>In vivo, different evidence points to a predominant role for AtSUMO1/2 paralogues. Early studies have shown that endogenous AtSUMO1 and 2 [7,11] and AtSUMO3 [7] were conjugated in planta, although AtSUMO3 conjugate levels were lower than for AtSUMO1/2. Instead, endogenous AtSUMO5 conjugation has not been observed. In addition, according to the Genevestigator database, atsumo3 and atsumo5 mRNA levels are on average 10-fold lower than atsumo1/2. Furthermore, none of the SUMO-specific proteases identified are competent to process AtSUMO5, and only AtULP1a displays an inefficient peptidase activity towards AtSUMO3, suggesting that, even if expressed at lower levels, it would have to be determined which fraction is present in its mature/conjugable form [31,32]. Finally, the fact that double-mutant atsumo1atsumo2 plants are not viable [34] suggests a biological specialization and, regardless of AtSUMO3 and 5 in vivo function, it seems clear that they cannot compensate for the AtSUMO1/2 loss. A recent study suggests that AtSUMO3 function might differ from AtSUMO1 and AtSUMO2 in flowering and salicylic-acid-dependent responses, although homozygous atsumo3 mutant plants showed normal plant development [37]. Interestingly, a more recent report described the identification of proteins that are specifically conjugated by AtSUMO3 and not AtSUMO1, although the molecular mechanism of this specificity remains elusive [48]. Surprisingly, almost half of the substrates analysed were exclusively modified by AtSUMO3, in contrast with the low levels of AtSUMO3 conjugates previously detected in crude plant extracts.</p><p>Overall, different molecular mechanisms seem to converge in order to assure a proper conjugation of the essential Arabidopsis SUMO1/2 isoforms compared with the non-essential AtSUMO3/5. These mechanisms comprise regulation of expression levels, maturation and release from targets, and AtSUMO1/2 are the highest expressed isoforms and the most efficient substrates of the characterized endogenous proteases [31,32], suggesting that most of the endogenous pool of mature SUMO will comprise these two isoforms. In the present study, we provide evidence for the existence of a preferential conjugation of AtSUMO1/2 compared with AtSUMO3/5, which is determined by a role of the E1-activating enzyme in SUMO paralogue discrimination.</p>

PubMed Author Manuscript

Ab initio Study of Anchoring Groups for CuGaO2 Delafossite-Based p-Type Dye Sensitized Solar Cells

Here we report the first theoretical characterization of the interface between the CuGaO2 delafossite oxide and the carboxylic (–COOH) and phosphonic acid (–PO3H2) anchoring groups. The promising use of delafossites as effective alternative to nickel oxide in p-type DSSC is still limited by practical difficulties in sensitizing the delafossite surface. Thus, this work provides atomistic insights on the structure and energetics of all the possible interactions between the anchoring functional groups and the CuGaO2 surface species, including the effects of the Mg doping and of the solvent medium. Our results highlight the presence of a strong selectivity toward the monodentate binding mode on surface Ga atoms for both the carboxylic and phosphonic acid groups. Since the binding modes have a strong influence on the hole injection thermodynamics, these findings have direct implications for further development of delafossite based p-type DSSCs.

ab_initio_study_of_anchoring_groups_for_cugao2_delafossite-based_p-type_dye_sensitized_solar_cells

Introduction<!>Computational details and structural models<!><!>Computational details and structural models<!>Results and discussion<!><!>Results and discussion<!><!>Results and discussion<!><!>Results and discussion<!><!>Results and discussion<!><!>Conclusions<!>Data Availability<!>Author Contributions<!>Conflict of Interest Statement

<p>The increasing world energy demands have boosted research toward the development of technologies that can exploit renewable sources in an efficient way (Hagfeldt et al., 2010). Being inexhaustible and relatively well spread all over the globe, solar energy has the highest potential to satisfy the present and future global energy needs. The photovoltaic (PV) field has rapidly evolved from traditional and commonly used silicon panels based on semiconductor p/n junctions to competitive thin film technologies relying on less abundant elements (Parida et al., 2011). Branching off from these devices, dye-sensitized solar cells (DSSC) arose 20 years ago, initially as cost-effective alternatives to solid-state photovoltaic solar cells. Now that the cost competitiveness with Si is less relevant, DSSCs are still in the spotlight, due to their lightweight, transparency, flexibility and best performance at diffuse and low-intensity light, which make them suitable for portable and indoor applications (Benesperi et al., 2018).</p><p>The most studied DSSCs are photoanodes where the electric current arises from the electron injection from the LUMO of a photoexcited dye to the conduction band of the n-type semiconductor (typically TiO2 or ZnO) where such dye is anchored. To date, traditional n-type DSSCs with metallic counterelectrodes have reached ~14% in photo–conversion efficiencies (PCE), while more recently the closely related hybrid organic/inorganic perovskite-based set-ups have exceeded the ~20% PCE (Yang et al., 2017). In order to overcome the intrinsic efficiency drawbacks of first generation DSSCs with only one chromophore, tandem cells have been proposed (He et al., 2000; Green, 2003; Nakasa et al., 2005; Nattestad et al., 2010). In tandem cells, a n-DSSC is coupled to a photocathode made of a p-type semiconductor that is sensitized by a dye with a red-shifted adsorption (with respect to dye at the photoanode), so that a larger portion of the solar spectrum can be harvested. Besides the potential higher photocurrent and PCEs for PV applications, tandem cells can be devised as well as photoelectrochemical devices able to perform photoinduced chemical reactions of great relevance in the energy conversion scenario (e.g., water splitting) (Prévot and Sivula, 2013). The actual application of tandem cells is however hampered for the limited efficiencies of the p-type DSSCs, which reach efficiencies of ~2%, still far below the ~14% of their n-type counterparts (Odobel et al., 2010).</p><p>A main culprit behind such poor performances has been ascribed to NiO, which is the most used p-type semiconductor (SC) in p-DSSCs due to its low cost and easy manipulation (Morandeira et al., 2005; Mori et al., 2008; Odobel et al., 2012). Unfortunately, NiO presents substantial intrinsic drawbacks: low electrical conductivity, low hole mobility and high valence band edge potential with respect to the most common I−/I3- electrolyte, which results in a too low photocathode open circuit potential (VOC) (Odobel and Pellegrin, 2013). The development of new efficient p-DSSC as alternative to NiO is thus of primary relevance.</p><p>Copper delafossites with formula CuMO2 (M = Al, Ga, Cr…) (Sullivan et al., 2016) have emerged among the few materials that can indeed outperform NiO as p-SC. Delafossites have wide optical bandgap and low valence band edge (Marquardt et al., 2006; Yu et al., 2012; Kumar et al., 2013). Nattestad et al. reported one of the first consistent comparisons between NiO and CuAlO2 in p-DSSC, obtaining VOC values of 218 and 333 mV, respectively (Nattestad et al., 2011). Later, experimental and theoretical investigations confirmed the lower VBedge position of CuAlO2 compared with NiO (Yu et al., 2014; Das et al., 2015; Schiavo et al., 2018). CuGaO2 and CuCrO2 were also reported as possible alternative to NiO, with lower band edge positions and higher transparency (Yu et al., 2012). With respect to NiO-based p-DSSC, with the new Ga and Cr delafossite oxides the measured increase in VOC was about 160 and 110 mV, respectively (Renaud et al., 2012; Powar et al., 2014). Several works show that the efficiency of these materials in p-DSSCs can be further improved by doping them with a divalent cation (e.g., Mg) at the M site (Scanlon and Watson, 2011; Jiang et al., 2013). This, in fact, enhances the p-type conductivity and has a positive effect on the morphology of the nanoparticles, which can expose a higher surface area resulting in a better light harvesting (Renaud et al., 2014).</p><p>However, despite the positive effect on VOC, the replacement of NiO with CuMO2 does not always result in an overall significant increase of photocurrent and/or PCE. While the VOC is determined by the p-SC VBedge vs. the electrolyte redox couple reduction potential, the overall cell efficiency strongly depends also on the interfacial electronic processes that occur between the semiconductor and the sensitizer. First, the lower the position of the valence band, the smaller is the driving force for hole injection to a given dye. For this reason, typical "p-type dyes" used for NiO may not be suitable for delafossites. For example, Renaud et al. tested the C343 prototype coumarin dye on CuGaO2 and did not measure any photocurrent (Renaud et al., 2012). P1, PMI-6T-TPA, or PMI-NDI dyes are used to sensitize CuGaO2, CuAlO2, and CuCrO2 delafossites, delivering small but non-zero current (Yu et al., 2012, 2014).</p><p>Another critic issue related to the delafossite-dye interface is the limited light-harvesting arising from poor dye coverage (Renaud et al., 2014). When synthesized via conventional solid-state reaction, CuMO2 tend to form large-size (>1 μn) anisotropic plate-like particles that densely stack along the basal planes, resulting in a limited surface area available for dye sensitization (Yu et al., 2014). With focused but less straightforward synthetic methods (e.g., following the hydrothermal route) it is possible to obtain smaller particles, close to the ideal 20–40 nm (i.e., the typical TiO2 nanoparticle size in n-DSSC) (Xiong et al., 2013; Yu et al., 2014). Nevertheless, these alternative synthetic routes are challenging, also depending on the element at the M site, and advances toward increasing dye coverage are highly desirable, even for particles with non-ideal shape or morphology.</p><p>In this context, an open issue is whether the dye-anchoring groups that are used for binding the dye to the rocksalt NiO surfaces are also good for a stable and irreversible binding to the Cu-based delafossite oxide most exposed surfaces. To address this specific issue related to the dye-electrode interface in delafossite-based p-DSSCs, we present here a first-principles study on two anchoring groups on CuGaO2 (001) surface: carboxylic acid (–COOH), which is the most common anchoring group used in both n- and p-type DSSCs, and phosphonic acid (–PO3H2), which is one of the anchoring groups explored in NiO-based p-type DSSCs in search for higher efficiencies (Pellegrin et al., 2011; Klein et al., 2018) and guarantees more stability in aqueous environments where COOH tends to desorb from oxide surfaces (De Angelis et al., 2011; Galliano et al., 2017).</p><p>First-principles calculations are a valuable tool for obtaining an atomistic insight on the structures, chemical interactions, and electronic features at the dye-electrode interfaces, thus providing a better understanding of the undergoing elementary processes. While electronic, adsorption and excited properties and level alignment of dyes anchored on n-type semiconductors (mainly TiO2) have been studied extensively using density functional theory (DFT) and time-dependent DFT (Martsinovich and Troisi, 2011; Labat et al., 2012; Adamo and Jaquemin, 2013; Bai et al., 2014), p-type semiconductor/dye interfaces have been investigated only more recently, and, in particular, considering NiO as semiconductor. An early work by Preat et al. focuses on computing the electronic excitation and electron injection processes of P1 and derived dyes using as model for NiO a single Ni atom capped by a hydroxyl group to ensure the overall neutrality of the system (Preat et al., 2011). Regarding periodic-slab and spin-polarized DFT calculations, taking into account also the intrinsic magnetism of NiO, only few recent computational studies have addressed the interaction of dye/anchoring groups with the most stable surface of NiO, i.e., the (001) (Muñoz-García and Pavone, 2015; Kontkanen et al., 2016; Wykes et al., 2016; Carella et al., 2018). Wykes et al. (2016) have studied formic and benzoic acid (HCOOH and PhCOOH) as prototypes of the most commonly used carboxyl-based anchoring groups, as well as mono and dimer phenyl-silane [PhSi(OH)3] anchoring candidates, due to the known capacity of alkoxysilanes to form Si-O-metal bonds. Besides different adsorption strengths, they find that such anchoring groups induce different shifts of the NiO VB and CB edges. A previous work by our group on the coumarin dye C343 (which features –COOH as anchoring group) and its analogous with –PO3H2 adsorbed on NiO (001) has shown that not only the nature of the binding group but also the binding modes largely affect the thermodynamic driving force for hole injection, due to the interface dipole generated by the H released to the surface in bi- and tri-dentate cases (Muñoz-García and Pavone, 2015). Piccinin et al. recently reported ab initio molecular dynamics simulations on the coumarin dye with –PO3H2 anchoring group on NiO (001) including explicit water molecules, showing also that the hole injection driving force (NiO-VB/Dye-HOMO level alignment) is very sensitive of the specific anchoring modes and their dynamics in a protic solvent (Piccinin et al., 2017). These works highlight the key role played by the anchoring ligand adsorbed on the surface on the overall performance of the DSSC.</p><p>While a few theoretical works have addressed the study of the electronic structure of bulk Cu-based delafossites for DSSC applications (Gillen and Robertson, 2011; Schiavo et al., 2018), the features of the dye-delafossite interface are essentially unexplored. In this work, we perform state-of-the-art DFT-based periodic calculations on anchoring groups –COOH and –PO3H2 on delafossite CuGaO2, which has been proposed as the most convenient for DSSC applications thanks to its wide band gap and easier p-type doping with respect to other delafossites such as CuAlO2 or CuCrO2 (Schiavo et al., 2018). After evaluating the change in the VB edge position upon Mg doping, we report adsorption energies and relevant structural and electronic features of the different anchoring groups/modes on the surface, taking into account the different combination of surface metal sites (Cu or Ga) that can be involved in the dye anchoring. Finally, we present a comparison of our results on CuGaO2 with those on NiO.</p><!><p>We performed periodic spin-polarized density functional theory (DFT) calculations with projector-augmented wave (PAW) potentials (Blöchl, 1994; Kresse and Joubert, 1999) and plane waves (PW) basis set by using the Vienna Ab Initio Simulation Package (VASP, version 5.4.1) (Kresse and Hafner, 1993, 1994; Kresse and Furthmüller, 1996a,b). The generalized-gradient approximation of Perdew Burke and Ernzherof (PBE) has been exploited for the exchange-correlation density functional (Perdew et al., 1996, 1997). To describe the strong-correlated nature of Cu d electrons we applied the rotationally invariant DFT+U approach of Dudarev (Dudarev et al., 1998) as implemented in VASP (Rohrbach et al., 2003). As in previous works on Cu(I)-containing oxides, we applied an average U-J effective value of 6 eV on Cu d electrons (Isseroff and Carter, 2012; D'Arienzo et al., 2017; Schiavo et al., 2018). SCF energy convergence threshold was set to 10−5 eV, and for minimum-energy structural optimizations the total forces on each atom were set to be all below 0.05 eV·Å−1. Regarding numerical parameters, with a kinetic energy cut-off of 750 eV for the PW and a k-point Monkhorst-Pack (Monkhorst and Pack, 1976) sampling grid of 6 × 6 × 2 for bulk CuGaO2 (see below) we obtained convergence of total energies within 5 meV per formula unit. K-point sampling for the slabs used have been scaled accordingly.</p><p>Although delafossites can present different polytypes (Marquardt et al., 2006), here we have considered the 3R phase, which is the only one detected with X-ray diffraction in CuGaO2 (Renaud et al., 2014). The hexagonal 3R polytype of CuGaO2 structure has the Ga(III) ions accommodated in a distorted octahedral cavity in the oxygen sublattice, generating layers of Ga(III)O6 units sharing their edges. These layers are separated by Cu(I) ions, linked in a linear geometry with two oxygens in two different layers (Figure 1A). For simplicity, we have built an orthorhombic supercell with aortho = ahexa, bortho = √3 bhexa, and cortho = chexa (Figure 1B). For surface calculations, we considered the (011) surface (012 in the hexagonal system), which has been identified among the preferred facets of delafossite crystals (Das et al., 2015; Alkhayatt et al., 2016) and, in particular, has the lowest surface energy in CuGaO2 (Schiavo et al., 2018). We cleaved such surface from orthorhombic bulk CuGaO2 with the theoretically determined equilibrium lattice constants (a = b = 2.98 Å, c = 17.64 Å), which deviate <3% from the experimental values (a = b = 2.97 Å, c = 17.17Å) (Ishiguro et al., 1981; Köhler and Jansen, 1986; Crottaz et al., 1996) For electronic structure calculations, the (011) surface was represented as slab with a (2 × 1) periodicity in the xy plane, five tri-atom layers of thickness and 10.5 Å of vacuum, in order to avoid images interaction along the z direction (Figure 1C). For Mg-doped CuGaO2, we have placed one Mg substituting one Ga atom on the central atom layer (Figure 1D), as it has been proven that the Mg substitution occurs at the Ga site (Jiang et al., 2013). This delivers a Mg-doped slab with 3.3% content of Mg (all % signs throughout the text are intended as atom %). An additional slab with (1 × 1) lateral periodicity has been used to study Mg-doped CuGaO2 surface with 6.7% of dopant, by substituting with Mg a single Ga atom in the central slab layer. Both in the undoped and Mg-doped cases, atomic positions of all the atoms have been allowed to relax.</p><!><p>Hexagonal (A) and orthorhombic (B) unit cells of bulk CuGaO2 with calculated cell parameters. Undoped (C) and Mg-doped (D) (011) orthorhombic surface. Color code: Cu (blue), Ga (green), O (red), and Mg (orange).</p><!><p>The relaxed CuGaO2 surface slab was used to compute the workfunction and vacuum energy level in order to determine the VB edge absolute position (i.e., with respect to the NHE) using the approach proposed by Toroker et al. (2011):</p><p>Where BGC is the band-gap center of the slab from our PBE+U calculations and Eg is the eigenvalue gap of CuGaO2 bulk calculated at the HSE level of theory, i.e., 2.1 eV (Schiavo et al., 2018), which is very close to the value of 2.2 eV reported by Iozzi et al. (2015). Evac is the vacuum energy evaluated from the electrostatic potential along the direction normal to the surface plane. For the Mg-doped system, where the BGC is ill-defined due to the typical shoulder of empty valence band states above the Fermi level that appears in the density of states (DOS) in p-type semiconductors, we have calculated the VBedge from the VBedge of un-doped CuGaO2 applying a shift computed from the difference of the workfunctions of the two systems (WFMg:CuGaO2-WFCuGaO2).</p><p>To survey the adsorption modes and the corresponding adsorption energies/properties of –COOH and –PO3H2 anchoring groups on CuGaO2 (001), we have chosen as capping substituent the simple methyl group (–CH3). This is a safe approximation for studying anchoring properties since, as we show below, they depend very little on the dye skeleton.</p><p>CH3-COOH and CH3-PO3H2 molecules have been placed in one side of CuGaO2 (001) slab with 2 × 1 of lateral periodicity and 3 tri-atom layers of thickness. Adsorption energy vary of <50 meV by increasing the lateral periodicity to 3 × 1 and of <25 meV by increasing the slab thickness to 5 tri-layers. Vacuum was increased to 15 Å to prevent interaction between the slab periodic images. Since molecules are adsorbed onto only one side of the slab, dipole corrections have been applied to avoid long-range polarization from the periodic images along the z direction (Neugebauer and Scheffler, 1992). We have considered all the possible anchoring modes for each group, taking also into account the different surface sites where the anchoring can be bound. Final geometries of these structures have been obtained relaxing all atomic coordinate of the molecule and of the two uppermost CuGaO2 tri-atom layers and by freezing those of the bottommost layers.</p><p>After relaxation, we have calculated adsorption energies as follows:</p><p>from the energies of the total system (Eslab+CH3−X), the relaxed pristine slab (Eslab) and the isolated molecule (ECH3−X), with X = CO2H or PO3H2. Using the relaxed geometries from gas-phase calculations, we have computed also the adsorption energies in acetonitrile, a common solvent used in p-DSSCs, by performing single-point energy calculations with the implicit solvent scheme implemented in VASP (Mathew et al., 2014) with relative dielectric constant ε = 35.688.</p><p>For comparison with NiO, we have studied CH3COOH and CH3PO3H2 molecules anchored on NiO (001) following an analogous procedure. Structural and computational details for NiO (001) surface slab (kinetic energy cut-off, k-point sampling, U-J values, slab thickness, lateral periodicity) are exactly those exploited in a recent previous work on the C343/NiO (001) interface (Muñoz-García and Pavone, 2015).</p><!><p>In p-type DSSCs, a first key parameter to assess the suitability of a p-semiconductor is the absolute position of its VB edge with respect to the electrolyte redox potential: the difference between these two values determines the open circuit potential VOC of the cell (Qin et al., 2008; Preat et al., 2011). For Mg doped CuGaO2, Renaud et al. (2014) have reported that Mg doping has a positive effect on the photoelectrode by increasing the specific surface area (SSA) of CuGaO2, but only CuGaO2 nanoparticles with low concentration of Mg (1%) deliver higher efficiencies than pristine CuGaO2. Higher concentrations of Mg (up to 5%), which increases further the SSA, do not lead to higher photocurrent. This drawback has been ascribed to the formation of structural defects or electronic features that promote undesired electron-hole recombination processes. Regarding the VOC with respect to the tris(4,4′-bis-tert-butyl-2,2′-bipyridine)cobalt(II/III) redox couple, the same authors (Renaud et al., 2012) measured a decrease in VOC values at increasing Mg contents. To understand this behavior, we computed VB edge positions of Mg-containing CuGaO2 in comparison to that of pristine CuGaO2 so to dissect to what extent Mg doping affects the VOC without any other structural defects. We have evaluated two different Mg concentrations, 3.3 and 6.7%, by substituting one Ga from the central tri-layer of the slab in the 5-tri-layer slabs with (2 × 1) and (1 × 1) periodicity, respectively. Figure 2 shows the resulting band edge potentials vs. NHE of CuGaO2, Mg-doped CuGaO2 and NiO together with common electrolytes employed in dye sensitized solar cells. All the reduction potentials of the redox mediators are measured in an acetonitrile solution, we refer the interested reader to the experimental references (Boschloo and Hagfeldt, 2009; Feldt et al., 2010; Cong et al., 2016) for the exact composition of the electrolyte.</p><!><p>Calculated absolute positions of the valence band edges for, from left to right: Mg:CuGaO2 with two different dopant concentrations (6.7 and 3.3%), undoped CuGaO2 and NiO. Dashed lines represent the redox potentials of the three electrolytes taken from experimental studies: I3-/I− (Boschloo and Hagfeldt, 2009), [Co(bpy)3]2+/3+ (Cong et al., 2016), and [Cu(bpye)2]+/2+ (Feldt et al., 2010). For completeness open circuit voltage calculated with respect to those electrolytes are reported.</p><!><p>The VB edge of CuGaO2 with 3.3% Mg is of 29 meV higher in energy with respect to the undoped case, and it rises up of another 4 meV when increasing Mg concentration from 3.3 to 6.7%. These numbers are of the same order of magnitude of the experimental values reported by Renaud et al. where a decrease of VOC of 30 meV for 1% Mg with respect to the undoped case and a further decrease of 10 meV when doping is increased up to 5% Mg (Renaud et al., 2014). Such little increase of the VB edge by Mg doping is still very low in comparison how low VB edge is in CuGaO2 with respect to NiO, which makes both CuGaO2 and Mg-doped CuGaO2 suitable for p-DSSCs not only with I3-/I− electrolyte (Boschloo and Hagfeldt, 2009) but also with other redox couples that are not compatible with NiO, such as cobalt(II/III) tris(2,2′-bipyridine) [Co(bpy)3]2+/3+ and copper(I/II) bis(1,1-bis(2-pyridyl)ethane) [Cu(bpye)2]+/2+ (Feldt et al., 2010; Cong et al., 2016). It is important to note that these results provide only a qualitative picture of the energy level alignment between the electrode and the electrolyte redox couple, many features are missing in our model (solvent, band bending at the interface, effects of ionic species, etc.). However, the computed trend is consistent with the VOC values observed in experiments and this agreement provides a good assessment on the quality of our surface slab model.</p><p>Understanding how the dye anchoring groups interact with the delafossite surface is of primary relevance: high adsorption strengths are needed to provide high dye coverages and, thus, high photoconversion efficiency. Moreover, in the case of H-containing anchoring groups, the individuation of preferred anchoring modes can help to explain and predict the emerging efficiencies, as different anchoring modes of the same anchoring group can deliver different band alignments between the dye HOMO and the p-semiconductor VB edge (Muñoz-García and Pavone, 2015; Zhang and Cole, 2015; Adineh et al., 2016).</p><p>We have surveyed all possible anchoring modes for both –COOH and –PO3H2 functional groups on CuGaO2 (011), taking into account the different possible surface adsorption sites and combinations of them. Thus, we have considered monodentate binding on either Ga or Cu surface atoms (M-Ga and M-Cu, respectively), bidentate binding on Ga/Ga, Cu/Cu or Ga/Cu surface atom pairs (B-Ga-Ga, B-Cu-Cu, and B-Ga-Cu) and, for CH3-PO3H2 we have considered also the two possible tridentate binding modes, on two Ga atoms plus one Cu atom or on two Cu atoms plus one Ga atom (T-Ga-Ga-Cu or T-Cu-Cu-Ga). For M cases, we have added the sub index "H" when the OH group forms an H bond to an oxygen surface atom upon relaxation. In B and T cases, the H atoms have been attached to surface oxygen atoms, as far as possible from the anchoring group to avoid artificial overstabilization and distortion of the structures due to H-bond formation between the released H and the O atoms from the anchoring molecules.</p><p>Relaxed geometries of CH3COOH on CuGaO2 (011) in monodentate (M) and bidentate (B) binding modes are shown in Figure 3 and corresponding selected structural parameters together with binding energies (Eads) are listed in Table 1. xyz files of the intermediates structures are reported in Supplementary Information. These geometries and Eads are compared to those of CH3COOH on NiO (001).</p><!><p>Optimized structures of CH3COOH on CuGaO2 (011) 2 × 1 × 3 L slab (right) and on NiO (001) (left). Labels according to the anchoring mode [monodentate (M) and bidentate (B)] and to the surface atoms involved in the adsorption process. Sub index "H" indicates H bonding (dashed gray line) between the OH group and surface oxygen atom upon relaxation in M cases. Color legend: Ni (gray), Cu (blue), Ga (green), O (red), C (brown), and H (light pink).</p><p>Selected structural parameters for relaxed CH3COOH anchored on CuGaO2 (011) and on NiO (001) together with those of the isolated molecule calculated at the DFT-PBE+U level of theory in vacuum.</p><p>Labels of anchoring modes as in Figure 3. O1/O2 correspond to carbonyl/OH oxygen atoms in isolated CH3COOH, respectively. Os identifies a superficial oxygen atom. Adsorption energies (Eads) in eV are collected in the last column. Bond lengths in the adsorbate (dX-C) and distance between the surface and the adsorbate (dCu-X, dGa-X, dNi-X, dOs-X) with X = O1, O2 or H.</p><!><p>Adsorbed CH3-COOH in M binding present similar structural features in CuGaO2 (011) and NiO (001), with the molecule not suffering any significant distortion from the isolated minimum. The main difference resides in the different M-carbonyl oxygen distance to the electrode surface (dO1−Xin Table 1), which is much shorter for bonding to a surface Ga (2.03 Å) than to a surface Cu (2.38 Å). The Ga-O1 distance in CuGaO2 is also very similar to the Ni-O1 distance in NiO (2.05 Å) and both values are similar to typical Ga-Olattice and Ni-Olattice in parent solids (2.00 and 2.11 Å in CuGaO2 and NiO, respectively). Contrarily, Cu-O1 distance is significantly longer than Cu-Olattice distances in CuGaO2 (1.87 Å). In CuGaO2, our calculations predict slightly longer H bonds between the OH and surface oxygen atoms (dH−Osin Table 1) in CuGaO2 with respect to NiO. In bidentate modes, the two oxygen atoms of CH3COOH molecule become equivalent and, accordingly, we obtain two equal M-O distances for B-Ga-Ga and B-Cu-Cu cases of CuGaO2 and B-Ni-Ni of NiO. Differently for the M case, here the Cu-O1 distances are much smaller and similar to those of CuGaO2 bulk. In the mixed B-Ga-Cu case, each Ga-O and Cu-O distance is equal to those in B-Ga-Ga and B-Cu-Cu. In spite of the different surface patterns of CuGaO2 (011) and NiO (001), M-M distances are very similar in both solids, i.e., d(Ni-Ni)NiO = 2.99 Å and d(Ga-Ga)CuGaO2 = d(Cu-Cu)CuGaO2 = 2.98 Å, hence the similarities in the binding geometries. Only a slight tilting of the C-C bond with respect to the z axis is predicted for B-Ga-Ga and in B-Cu-Cu in CuGaO2, having the CuGaO2 (011) surface a non-planar pattern.</p><p>In spite of these structural analogies between CuGaO2 and NiO, the binding energies (Eads in Table 1) of the different anchoring modes of CH3COOH to the two surfaces present substantial differences. In NiO, Eads of both M and B modes are negative and close, being Eads(B) 0.07 eV more stable than Eads(M). We must note that the binding energies reported here for CH3-COOH (as well as those indicated below for CH3-PO3H2) on NiO are similar to those considering the full C343 dye (Muñoz-García and Pavone, 2015) within 0.1 eV, and deliver the same result of B being more stable than M [Eads(M)-Eads(B) = 0.12 eV for C343]. Thus, we can consider the CH3- capping group as a good approximation for studying adsorption properties of different anchoring groups on the electrode surfaces. Such negative and similar Eads values can be translated into expecting NiO covered by the dye in the two anchoring modes in approximately equal ratio in experimental conditions. In CuGaO2, instead, there is a strong preference for the monodentate binding because all B modes surveyed present a positive (i.e., not favorable) Eads. In particular, M binding is stronger at the Ga site than at the Cu site, with Eads values similar to those for NiO. This is consistent with the trend of transition metal-oxo complex bond dissociation energies (BDE) (Luo, 2007) or Ni-O and Ga-O (−366 and −374 KJ/mol), with a smaller value for Cu-O BDE (−287 kJ/mol). Moreover, while surface Ni/Ga atoms are unsaturated species in NiO (001)/CuGaO2 (011), surface Cu retains its bulk-like linear coordination with two oxygen atoms in CuGaO2 (011).</p><p>Analogously, we have explored all possible binding modes of the phosphonic acid group on CuGaO2 (011). Relaxed geometries of CH3PO3H2 on CuGaO2 (011) in monodentate (M), bidentate (B), and tridentate (T) binding modes are shown in Figure 4, while corresponding structural parameters and Eads are listed in Table 2 together with those of CH3PO3H2 on NiO (001).</p><!><p>Optimized structures of CH3PO3H2 on CuGaO2 (011) 2 × 1 × 3 L slab (right) and comparison to NiO (001) (left). Labels according to the anchoring mode [monodentate (M), bidentate (B), and tridentate (T)] and to the surface atoms involved in the adsorption process. Sub index "H" indicates H bonding (dashed gray line) between the OH group and surface oxygen atom upon relaxation in M cases. Color legend: Ni (gray), Cu (blue), Ga (green), O (red), C (brown), H (light pink), and P (lavender).</p><p>Selected structural parameters for relaxed CH3PO3H2 anchored on CuGaO2 (011) in comparison to those on NiO (001) and of the isolated molecule calculated at the DFT(PBE)+U level of theory in vacuum.</p><p>Labels of anchoring modes as in Figure 4. O1 and O2/O3 correspond to phosphoryl and OH/OH oxygen atoms in isolated CH3PO3H2, respectively. Os identifies a superficial oxygen atom. Values reported with an asterisk (*) correspond to the average value with a maximum standard deviation of 0.01 Å. Adsorption energies (Eads) in eV are collected in the last column. Bond lengths in the adsorbate (dX-C) and distance between the surface and the adsorbate (dCu-X, dGa-X, dNi-X, dOs-X) with X = O1, O2 or H.</p><p>) correspond to the average value with a maximum standard deviation of 0.01 Å. Adsorption energies (Eads) in eV are collected in the last column.</p><!><p>In this case, only the monodentate adsorption modes on surface Ga atoms are favored. In particular, we find two stable M geometries, one without any H-bonding to the surface (M-Ga) and one where both the two OH groups form H-bonds with the delafossite surface O atoms (M-GaH). Such H bonds stabilize the latter by 0.92 eV with respect to M-Ga. We explored different starting geometries with phosphoryl oxygen O1 linked to surface Cu but all evolve into either M-Ga or M-GaH upon relaxation. As in NiO, the –PO3H2 group in M binding (with H bonding) is more strongly bound than –COOH. As for –COOH, we find that only M binding of the –PO3H2 group is stable on the delafossite surface, differently from NiO, where M, B, and T modes are all strongly anchored to the surface with Eads lying in a narrow window of energy. In CuGaO2, B modes present positive Eads that go from the few eV of the B-Ga-Ga case to the 1.38 eV in the B-Cu-Cu case. In all the tridentate modes, adsorption energies are very high. So, we can conclude that –PO3H2 anchoring group will bind to CuGaO2 (001) preferentially through a M-GaH-like anchoring mode, with small amounts of M-Ga-like and negligible presence of B-Ga-Ga.</p><p>In order to evaluate to what extent the selectivity toward monodentate binding on CuGaO2 for both anchoring groups described above is affected by Mg-doping and the presence of the solvent, we have also calculated Eads of each anchoring group/mode in the pristine delafossite slab considering acetonitrile as solvent and on the slab containing 3.3% of Mg, both in vacuum and in acetonitrile. We chose acetonitrile as solvent since it is one of the most widely employed solvents in DSSCs and it was also employed as a solvent in the experimental electrolyte compositions we refer to for open circuit voltage determination.</p><p>These three additional sets of Eads, together with those calculated in vacuum on the pristine slabs are gathered in Table 3.</p><!><p>Calculated adsorption energies (Eads) for CH3COOH and CH3PO3H2 on CuGaO2 (001) at the DFT(PBE)+U level of theory in vacuum and acetonitrile (implicit solvent), without and with Mg doping (3.3%).</p><p>Anchoring mode labels as in Figures 4, 5. Negative binding energies are highlighted in bold.</p><!><p>From our calculations, both anchoring groups bind significantly more strongly to the Mg-doped surface than on the undoped surface with Eads that decrease almost uniformly around ~0.5 eV. There are very few exceptions where Eads suffers a negligible increase (max. ~0.05 eV) upon Mg-doping as CH3COOH-B-Cu-Cu and CH3PO3H2-B-Ga-Ga/T-Ga-Ga-Cu. This stabilizing effect can be explained to the higher Lewis acidity of oxide surfaces upon p-type doping, where a hole has been introduced, increasing the ability of such oxide to accept electron-donating species (Metiu et al., 2012). We must note here that, as in bulk CuGaO2 (Schiavo et al., 2018), the presence of Mg does not lead to a significant localized oxidation of any particular cation but to the hole being distributed among all Cu atoms of the system (each Cu losing max. 0.08 e−), with even smaller involvement of Ga. This explains the uniform decrease in Eads in all surface sites and not only on Cu sites. Regarding the solvent, as common trend to all the M and B anchoring modes, the dielectric continuum increases Eads but not to the same extent. Only in the cases of T-modes in CH3PO3H2, it decreases by ~0.5 eV but this decrease is not enough to stabilize the very unfavorable T anchoring modes. From a general perspective, the acetonitrile solvent medium with a mild dielectric constant is expected to weaken the ionic contribution to the bonding between anchoring group O atoms and Ga surface sites. At the same time, for the T modes, the solvent is able to stabilize the dipole moments of the –OH surface species that are formed during the anchoring. In any case, the overall adsorption energy landscape is not qualitatively changed when both the electrolyte solvent and the Mg-dopant are taken into account. Eads values are analogous to those calculated for the pristine slab in vacuum, as the two effects balance each other. Thus, we can state that M binding is likely the preferred anchoring mode for CH3COOH and CH3PO3H2 in operating conditions. This result is of utmost importance for the cell performance since it has been shown that the driving force for hole injection between the dye HOMO and the VB of the semiconductor is maximized in M binding modes where no H has been released to the surface and the interfacial dipole at the electrode surface is minimum (Muñoz-García and Pavone, 2015). Moreover, we can state that Mg doping, besides having an effect on improving p-type conductivity and nanoparticle morphology (Renaud et al., 2014), improves dye coverage by increasing the affinity between the delafossite and dye anchoring groups.</p><p>Figures 5, 6 show the projected density of states of CuGaO2(011) with anchored CH3COOH and CH3PO3H2, respectively, in the most stable binding mode (M-GaH). We have considered all the four systems for which the adsorption energies were computed: pristine/vacuum, Mg-doped/vacuum, pristine/acetonitrile, and Mg-doped/acetonitrile. In all these cases, the molecular states of the anchoring groups are far from the semiconductor VB edge and, hence, they would not contribute to electron/hole transport process. These results are consistent with those recently computed for Ph-COOH and silanes on NiO (Wykes et al., 2016), in line with what is expected for the proper functioning of the solar cell.</p><!><p>Projected density of states (pDOS) calculated at the DFT(PBE)+U level of theory of CH3COOH adsorbed on CuGaO2 (011). Fermi level (dashed gray line) has been set to zero.</p><p>Projected density of states (pDOS) calculated at the DFT(PBE)+U level of theory of CH3PO3H2 adsorbed on CuGaO2 (011). Fermi level (dashed gray line) has been set to zero.</p><!><p>This work reports an ab initio study of CuGaO2 delafossite as alternative to NiO in p-type DSSCs. The semiconductive copper delafossite with chemical formula CuMO2 presents a lower absolute position of the valence band edge than NiO. Thus, besides providing a higher open circuit potential (VOC) when used with traditional I−/I3- than NiO, it enables the use of new and high performing Co- and Cu-based electrolytes. Gallium delafossite CuGaO2 is of particular interest since it can be easily doped with Mg, which enhances the p-type conductivity and the shape and morphology of delafossite nanoparticles. First, we have focused here on studying the change in the valence band absolute position after Mg doping: with ~3 and ~6% dopant contents the VOC of the copper gallium delafossite is mostly unchanged, in agreement with experiments.</p><p>Due to the practical difficulties of sensitizing delafossite oxides, we have addressed the adsorption properties of two anchoring groups widely used for grafting dye molecules on semiconductor surfaces: –COOH and –PO3H2 (carboxylic acid and phosphonic acid, respectively) on the most stable CuGaO2 surface, i.e., the (011). We characterized the interaction between these anchoring groups and the surface in terms of anchoring mode minimum-energy structures and corresponding adsorption energies. We have dissected the effects of Mg doping and of the presence of the solvent on these features, as well as on the electronic structure, which is of key importance to DSSC operation. Contrary to what happens in the case of NiO, on delafossite surface there is a strong selectivity toward monodentate binding modes for both the carboxylic and phosphonic anchoring groups, with a particular affinity toward the Ga surface sites. Since it has been shown that driving force for hole injection from the dye to the semiconductor is jeopardized when protic groups release H to oxide surfaces in bidentate and tridentate modes, our results point out that the combination of –COOH or –PO3H2 anchoring groups with CuGaO2 might deliver much better performance than with NiO. Besides, our calculations show that Mg doping increases the affinity of CuGaO2 surface for both anchoring groups and, thus, the better performances of Mg-containing samples can be also ascribed to a higher sensitization driven by more favorable adsorption energies.</p><p>In conclusion, this work provides a first theoretical characterization of the interface between delafossite oxide and the –COOH or –PO3H2 anchoring groups, thus paving the route to further studies on full dye-sensitized delafossite-based photocathodes in order to help the development of p-type DSSC technologies with first-principles derived rational design guidelines.</p><!><p>The datasets generated for this study are available on request to the corresponding author.</p><!><p>AM-G and MP designed the research. LC, ES, and CB performed the calculations. AM-G, PM, and MP rationalized the results. All authors contributed in writing and revising 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

Reverse Regioselectivity in Reductive Ring Opening of Epoxide Enabled by Zirconocene and Photoredox Catalysis

A ring opening of epoxide with zirconocene and photoredox catalysis has been developed. Compared to the ring opening methods with titanocene, the present protocol exhibited reverse regioselectivity to afford more-substituted alcohols via putative less-stable radicals. The observed regioselectivity could be explained by shifting the transition states to more reactant-like structures by changing the metal center of the metallocene catalyst. ASSOCIATED CONTENT Supporting InformationThe Supporting Information is available free of charge on the ACS Publications website.Experimental procedures, additional experimental results, computational study, and compound characterization (PDF)

reverse_regioselectivity_in_reductive_ring_opening_of_epoxide_enabled_by_zirconocene_and_photoredox_

<p>Epoxide is a common structural motif in bioactive compounds, naturally occurring feedstocks, and various synthetic intermediates. Due to their inherent ring strain, epoxides generally behave as an excellent electrophile. Contrastingly, homolysis of its C-O bond provides a nucleophilic carbon radical leading to a distinct product that is unavailable from polar mechanisms. For such transformations, titanocene (III) (Cp2TiX) has been exclusively exploited over the past 30 years since its introduction by Nugent and RajanBabu. 1,2 In general, an epoxide coordinates to titanocene (III) prepared by reduction of titanocene (IV) with a reducing metal, and the C-O bond cleavage proceeds via single electron transfer (Figure 1A). The resulting carbon radical reacts with a large array of trapping agents such as hydrogen atom donors or olefins to furnish alcohols. So far, considerable research efforts have been directed to broaden this transformation. [3][4][5][6][7] This has enabled remarkable advances in functionalization of the radical generated via epoxide ring opening, even including cross-coupling. 8 Mild reaction conditions have allowed this transformation to contribute to numerous natural product syntheses. [9][10][11] Recent studies showed that the merger of titanocene and photoredox catalysis can circumvent the need for a stoichiometric reducing metal. [12][13][14] When an unsymmetrical epoxide is employed, regioselectivity issues must be addressed. The factors controlling the regioselectivity is a fascinating topic, and detailed studies have been carried out. [15][16][17] In most cases, homolysis predominantly occurs at the C-O bond, which gives a more stable radical (Figure 1B). In particular, thermodynamically stable radicals in benzylic 18 , allylic 19 , and anomeric 20 positions are favorably generated, much like in many other radical reactions. For electronically unbiased epoxides, non-bonding interactions including steric hindrance are considered as a prevailing factor. [15][16][17] However, the princi-ples controlling the regioselectivity still remain to be elucidated, and only few examples exist where the product is derived from a putative less stable radical. 15,21 Considering the Bell-Evans-Polanyi principle, 22,23 activation energy (DETS) decreases if epoxide opening becomes more exothermic (Figure 1C). Likewise, the transition state would shift earlier and become more similar to the starting material. In an earlier transition state, the contribution of stability of the resulting radicals in regioselectivity would diminish according to the Hammond postulate. 24 Inspired by the difference in bond dissociation energy (BDE) between Ti-O (115 kcal/mol) 25 and Zr-O (132 kcal/mol), 25 we became interested in using zirconocene instead of titanocene for the epoxide opening. Its stronger oxophilicity should render the ring opening dramatically exothermic. On this basis, we envisioned that a zirconocene-catalyzed ring opening of epoxide would reverse the conventional regioselectivity expected from relative radical stability. To this end, we set out to investigate a catalytic protocol for the ring opening of epoxides using zirconocene with photocatalysts even though Zr (III) has been scarcely utilized in organic synthesis. [26][27][28][29][30][31][32] Since the reduction of zirconocene typically requires high reducing power (-E1/2 (Cp2ZrCl2) = -1.85 V vs SCE) 33 , we commenced with strongly reducing photocatalyst Ir[(4-MeO)ppy]3 (P1, E1/2 Ir(IV)/Ir(III)* = -1.95 V vs SCE in MeCN). 34 Visible light irradiation of 1A in the presence of zirconocene dichloride, P1, and 1,4-cyclohexadiene (CHD) resulted in no product formation (Table 1, entry 1). A change to Cp2Zr(OTf)2•THF did not furnish either of the alcohols (Table 2, entry 2). Gratifyingly, we found that addition of dimethylthiourea (T1) furnished the desired alcohol 1B in 34% yield along with the isomeric alcohol 1C in 7% yield (Table 1, entry 3). Further screening revealed that the combination of 1methyl-3-phenyl thiourea (T3) and PhCF3 facilitates the reaction with a suppressed amount of 1C (Table 1, entries 4-7). Applying Ir(ppy)3 (P2, E1/2 Ir(IV)/Ir(III)* = -1.88 V vs SCE in MeCN) 34 or Ir[(dFCF3ppy)2(dtbbpy)]PF6 (P3, E1/2 Ir(IV)/Ir(III)* = -1.21 V vs SCE in MeCN) 35 diminished the yield of 1B (Table 1, entries 8 and 9). Although other zirconocenes were not effective, the addition of molecular sieves slightly increased the yield (Table 1, entries 10-12). Control experiments excluding zirconocene, photocatalyst, and light provided essentially no product (Table 1, entries 13-15).</p><p>With the optimized conditions in hand, we next evaluated the substrate scope of this protocol (Scheme 1). First, we examined terminal epoxides with a variety of functional groups. In addition to phenyl (1A), p-methoxyphenyl (1B), alkyl (1C), ester (1D-1F), benzyl and silyl ether (1G and 1H), and chloride (1I) were accommodated to afford the corresponding secondary alcohols with high yield and regioselectivity. Electronically diverse substituents on aromatic rings (1J-1O) were all tolerated. Protected prolinol derivatives (1P-1R), as well as sulfonamide (1S), were found to be suitable substrates. We next evaluated di-and trisubstituted epoxides. The present protocol converted 1,1-disubstituted epoxides 1T and 1U to more-substituted alcohols with retained regioselectivity, whereas ring opening of epoxide 1V afforded 41 % yield of 2V and 3V in total with considerable amount of cyclic product 4V suggesting the addition of a primary radical to the aromatic ring. Spiro epoxides (1W-1Z) uniformly furnished the desired alcohol, although incorporation of bulky adamantane (1AA) considerably reduced regioselectivity. 1,2-Disubstituted epoxides with a variety of ring sizes were well-tolerated (1AB-1AF). Interestingly, ring opening of dibenzyl ether 1AG produced alcohol 2AG along with benzylidene acetal 4AG obtained in the previous titanocene-mediated ring opening via 1,5-hydrogen atom transfer (HAT). 36 In addition to trisubstituted epoxides (1AH-1AJ), natural product-derived epoxides (1AK-1AN) could also be readily opened in good yields. Comparisons to the analogous precedents performed by titanocene highlighted the reverse reactivity in ring opening. As mentioned above, ring opening of 1C provided 2C with high regioselectivity, which was contrasting to the previous study with titanocene. 37 Similarly, the ring opening of 1V afforded tertiary and primary alcohols along with 4V in our conditions, whereas primary alcohol 3V is preferably obtained with titanocene. verted 1AL to 2AL, ring opening of a similar cholesterol (R' = Bz) with titanocene furnished a tertiary radical, and the adjacent hydrogen was abstracted by a cobalt catalyst leading to an allylic alcohol. 38 These contrasting results demonstrated that our catalytic protocol is complementary to conventional methods with titanocene.</p><p>To gain insights into the reaction mechanism, ring opening of cyclopropyl-bearing epoxide 1AO was conducted (Figure 2A). Treatment of this probe with the optimized conditions afforded the allylic alcohol 5AO, suggesting a C-C bond cleavage via a cyclopropylcarbinyl radical. The same product was obtained in Oshima's conditions where Zr(III) was prepared from Cp2Zr(H)Cl. 29 Furthermore, intramolecular radical addition could be achieved with an epoxy ester (1AP; Figure 2B). These observations, as well as formation of 4V and 4AG, are in agreement with the C-O bond cleavage proceeding via a radical pathway. Next, we performed DFT calculations to reveal the energy profile of ring opening and HAT based on a model reaction of 1,1-dimethyl oxirane and metallocene triflates (III) (M = Ti or Zr) (Figure 3A). The reaction proceeds from complex I via the respective transition state (TS-Ia and TS-Ib) to afford the corresponding ring-opened forms (IIa and IIb). The resulting radicals would then undergo HAT with CHD to be converted to alcohols (IIIa and IIIb). The energy profile with primary (purple) and tertiary (aqua) radical with titanocene, and primary (red) and tertiary (blue) radical with zirconocene are shown in Figure 3A. In accord with our design hypothesis, activation free energies of ring opening ΔGTS-I with zirconocene (7.81 and 6.24 kcal/mol, respectively) are remarkably lower than those of titanocene (22.14 and 14.95 kcal/mol, respectively). 39 Similarly, ΔΔGTS-I with zirconocene (1.57 kcal/mol) is much smaller than that of titanocene (7.19 kcal/mol). This significant difference can be accounted for by the consequence of shifting the TS of ring opening earlier, which would reduce the influence of the thermodynamic stability between the resulting primary and tertiary radicals at the TS. Structures of TS-1 provide further insight into shifting the TS of ring opening. The scissile C-O bonds in Zr-TS-I are shorter than that of Ti-TS-I by approximately 0.2 Å, suggesting that Zr-TS-I are more reactant-like. The spin density on the evolving carbon radical in Zr-TS-Ia (0.39) and Zr-TS-Ib (0.29) was smaller than that in Ti-TS-Ia (0.57) and Ti-TS-Ib (0.42). These structural features are consistent with more reactant-like Zr-TS.</p><p>DFT calculations indicated that ring opening with titanocene was endothermic, which is consistent with a previous study. 36 Since titanocene-bounded tertiary radical IIb is more stable than the corresponding primary radical IIa by 6.82 kcal/mol, ring opening under thermodynamic control results in a large abundance of IIb which undergoes HAT to afford IIIb as a major product. In contrast, ring opening with zirconocene was found to be extremely exothermic and practically irreversible. By means of computational studies to provide a rough picture of the energy profile, Zr-TS-Ia was calculated to be stable than Zr-TS-IIb. This discrepancy with our results prompted further investigation into the identity of the zirconocene complex.</p><p>In a possible scenario to explain the regioselectivity, ring opening proceeds via an epoxide-zirconocene-thiourea complex. 40-42 In this case, ΔGTS-Ia might be lower than Δ GTS-Ib, which is opposite to the calculated results with the zirconocene triflate complex. Diffusion-ordered NMR spectroscopy (DOSY) measurements evidently demonstrated that thiourea interacts with zirconocene (Figure 3B). In this experiment, tetramethylthiourea was used to facilitate NMR analysis. A 1:1 mixture of zirconocene and thiourea furnished several new signals in 1 H NMR. The diffusion coefficient for the newly generated signals (D = 1.44 × 10 -9 m 2 /s) was smaller than that of zirconocene (D = 1.55 × 10 -9 m 2 /s) and completely different from that of parent thiourea (D = 3.45 × 10 -9 m 2 /s), suggesting the interaction of zirconocene and thiourea. Thiourea probably interacts with not only Zr(IV) but also Zr(III) species since regioselectivity of the ring opening was influenced by the selected thiourea. Furthermore, Stern-Volmer analysis indicated that quenching of the excited photocatalyst by the zirconocene thiourea complex (see the SI for details) Given these experimental results, a possible mechanism is depicted in Figure 4. The reaction could be initiated by reduction of zirconocene by an excited photocatalyst to furnish Zr(III) and triflate. Coordination between Zr(III) and epoxide could result in C-O bond homolysis to afford a carbon radical, which abstracts hydrogen from 1,4-CHD. The cyclohexadienyl radical undergo oxidation by Ir IV catalyst (E1/2 Ir(IV)/Ir(III) = +0.70 V vs SCE in MeCN). 34 The resulting cation and triflate would work as a Brønsted acid to protonate zirconocene alkoxide to complete the catalytic cycle, with release of the desired alcohol and benzene. 17 In summary, we succeeded in the development of a zirconocene-photoredox cooperative catalysis for the ring opening of epoxide, exhibiting reverse regioselectivity compared to titanocene-mediated reactions. To our knowledge, this is the first example that harnesses a radical intermediate generated by C-O homolysis of epoxide using zirconocene. [43][44][45] Exploring transformations with this new cooperative catalysis and further mechanistic studies are currently underway in our laboratory.</p>

ChemRxiv

Relationship between folate concentration and expression of folate-associated genes in tissue and plasma after intraoperative administration of leucovorin in patients with colorectal cancer

PurposeThe aim of study was to investigate the relationship between folate concentration and expression of folate-associated genes in tumour, mucosa and plasma of patients with colorectal cancer, after intraoperative administration of bolus leucovorin (LV).MethodsEighty patients were randomized into four groups to receive 0, 60, 200, or 500 mg/m2 LV, respectively. Tissue and plasma folate concentrations were assessed by LC–MS/MS. Gene expression of ABCC3/MRP3, FPGS, GGH, MTHFD1L, SLC46A1/PCFT, and SLC19A1/RFC-1 was determined using quantitative PCR.ResultsThe folate concentration in tumour increased with increasing dosage of LV. Half of the patients treated with 60 mg/m2 did not reach a level above the levels of untreated patients. A significant correlation between folate concentration in tumour and mucosa was found in untreated patients, and in the group treated with 60 mg/m2 LV. The 5-MTHF/LV ratio correlated negatively with folate concentration in mucosa, whereas a positive correlation was found in tumour of patients who received 200 or 500 mg/m2 LV. A positive correlation was found between folate concentration and expression of all genes, except MTHFD1L, in patients who received LV. There was a negative correlation between 5-MTHF concentration in plasma of untreated patients and expression of GGH and SLC46A1/PCFT in tumour.ConclusionsThe results indicate the possibility of using the individual plasma 5-MTHF/LV ratio after LV injection as a surrogate marker for tissue folate concentration. Expression of several folate-associated genes is associated with folate concentration in tissue and plasma and may become useful when predicting response to LV treatment.Electronic supplementary materialThe online version of this article (10.1007/s00280-018-3690-9) contains supplementary material, which is available to authorized users.

relationship_between_folate_concentration_and_expression_of_folate-associated_genes_in_tissue_and_pl

Introduction<!>Patients and methods<!>Determination of plasma folate<!>Determination of tissue folate<!>Preparation of RNA and cDNA<!>Real-time quantitative PCR<!>Statistical analyses<!>Results<!><!>Results<!><!>Results<!><!>Results<!><!>Results<!><!>Discussion<!>Conclusions<!>

<p>Colorectal cancer (CRC) is a major cause of cancer death worldwide [1, 2]. The median overall survival from metastatic disease is now reaching almost 30 months in clinical trials, which is due to development of new biological agents in addition an increasing number of patients undergoing surgical resection of localized metastatic disease. A more strategic approach to the delivery of systemic therapy and an expansion in the use of ablative techniques may also contribute to improvement of survival [3]. However, in the clinical setting, the overall survival can be expected to be lower [4].</p><p>The fundament in the treatment of CRC is surgery. However, after primary surgery, treatment with chemotherapy is recommended for advanced tumour stages [5]. The drug 5-fluorouracil (5-FU) is used as a cornerstone of chemotherapy treatment, in the adjuvant, as well as in the palliative setting for CRC [6–8]. This drug is an analogue of uracil, in which the hydrogen at position 5 is replaced by fluorine. 5-FU enters the cell in the same way as uracil and is then converted in several steps intracellularly to the final active metabolite 5-fluoro-2′-deoxyuridine monophosphate (FdUMP). The enzyme thymidylate synthase (TS; EC 2.1.1.45), catalyses the reductive methylation of deoxyuridine monophosphate (dUMP) to deoxythymidine monophosphate (dTMP) using 5,10-methylenetetrahydrofolate (5,10-MeTHF) as the methyl-group donor. Subsequently, dTMP is further phosphorylated to become deoxythymidine triphosphate (dTTP), which is an essential nucleotide needed in DNA synthesis and repair. When FdUMP is replacing dUMP, the TS enzyme is inhibited thereby blocking the conversion of dUMP to dTMP. Since this conversion is the only way for de novo formation of dTMP, the blockage will impair both DNA synthesis and repair [8, 9]. The greatest impact will be on cells having a high proliferation rate, such as cancer cells [10].</p><p>The response rate of colorectal tumours to 5-FU monotherapy is only around 10%. By adding a reduced folate, the tumour response rate can be improved to 21%, as has been shown in a meta-analysis [11]. In clinical practice, patients treated with 5-FU will receive the reduced folate leucovorin (LV) in the form of a stable calcium salt of 5-formyltetrahydrofolic acid (calcium folinate). After injection, LV is transported to the liver where it is mainly metabolized to 5-methyltetrahydrofolate (5-MTHF). This metabolite is then transported to the blood stream to reach other tissues. The maximum plasma concentration of 5-MTHF is reached 30 min after i.v. administration [12]. LV is eliminated by 80–90% through the kidneys.</p><p>The Nordic FLV therapy, which is a regime of bolus 5-FU and LV that was introduced in the 1990s, is still the cornerstone of both adjuvant and palliative treatments for CRC in Nordic countries [13]. Although the Nordic FLV is a well-established regimen, the evidence for a beneficial effect of the LV dosage and administration used is rather limited. Different regimens used in clinical practice worldwide apply LV concentrations that range from 20 to 500 mg/m2 and are often empirically evolved [14].</p><p>Leucovorin needs to be converted in several steps into the active substance 5,10-MeTHF. This metabolite is polyglutamated by the enzyme folylpolyglutamate synthase (FPGS), which increases its cellular retention. 5,10-MeTHF stabilizes the ternary complex between TS and FdUMP, hence, leading to inhibition of the formation of dTMP from dUMP [15]. In order to get a complete inhibition of TS, the active metabolite 5,10-MeTHF must be in excess. Then, a maximal formation of the ternary complex consisting of TS, FdUMP, and 5,10-MeTHF can be achieved [16, 17].</p><p>A recent article published by our group presented data showing a large inter-individual variation of tissue folate concentration in patients with CRC after supplementation with LV at standardized dosage [18]. The concentrations of tetrahydrofolate (THF), 5,10-MeTHF and 5-MTHF in tumour and mucosa were assessed by liquid chromatography electrospray ionization tandem mass spectrometry (LC–MS/MS). The results showed that the 5,10-MeTHF concentration in tumours of many patients given the standardized dose of LV, i.e. 60 mg/m2, did not exceed the values of an untreated control group. Rectal cancer patients in particular required high doses of LV to reach a tumour tissue concentration of 5,10-MeTHF above baseline. Similar results have been described by Houghton et al. [19].</p><p>In the present study, we hypothesize that folate administration has an effect on gene expression that may be clinically relevant. The folate concentration in matching tumour and mucosa tissue obtained from patients with CRC were determined and compared with the concentration in plasma after treatment with increasing doses of LV. The impact of the folate concentration in tissue and plasma on gene expression of six folate pathway genes was analysed.</p><!><p>Eighty patients scheduled for a colorectal resection with a cancer indication were enrolled in the study between January 2011 and January 2012. All patients gave their written informed consent. The pre-operative exclusion criteria were patient inability to understand the study information or inability to provide true informed consent. There were no other exclusion criteria. The patients were pre-operatively randomized into four groups; the first served as control group and received no LV. Groups 2, 3, and 4 received 60, 200, and 500 mg/m2 LV, respectively, administered intravenously as a bolus injection at the initiation of general anaesthesia. The LV was manufactured in the form of calcium folinate (RS-LV) supported by Teva Sweden AB, Helsingborg, Sweden. The surgeon was blinded to the dosage given and all patients were otherwise treated in accordance with normal routines and guidelines.</p><p>During surgery, at the time of removal of the surgical specimen, a research nurse collected fresh tissue samples from both tumour and macroscopically normal-appearing mucosa located 10 cm from the tumour. The biopsies were snap-frozen in liquid nitrogen and stored at − 80 °C until used. Clinical and pathology data regarding diagnosis, tumour differentiation and stage, and pre-operative treatment regimen were retrieved to assess the different groups and enable a better understanding of the factors that might influence treatment responses.</p><!><p>Blood samples were obtained from all patients. The samples were collected at 0, 10, and 30 min. in EDTA vacutainers and immediately centrifuged (4 °C, 2000 g, 10 min). The plasma was stored at − 80 °C until LC–MS/MS analysis. To 1 ml of [(0,2% formic acid in acetonitrile): Methanol (9:1)], 200 µl plasma was added, and mixed well for 10 min. Aminoacetophenone was used as internal standard. Stock aminoacetophenone solution (0.1 mM) was stored at − 80 °C until use. After addition of 20 µl internal standard and 30 µl of water, the samples were centrifuged for 10 min at 21,500g. The supernatant was loaded on a 1 cc Oasis PRIME HLB Cartridge and passed-through. The eluate was collected and evaporated to dryness and reconstituted with 300 µl Mobile phase A before analysis. A stock solution (20 mM) of LV and 5-MTHF was dissolved in extraction buffer (50 mM phosphate buffer (pH 7.0), 1% sodium ascorbate and 0.1% β-mercaptopropanol) and stored at − 80 °C until use. The folate solution was serially diluted in extraction buffer to prepare the calibration curves. A blank plasma sample was used to dilute the standard samples. A mixture of standards and internal standard was extracted as described for the samples. Calibration standards containing ten different concentrations for LV and 5-MTHF were used. The quality (Q) controls low, medium and high were prepared in plasma, and extracted as described for the samples. The extracted ions following MRM transitions were monitored at m/z 460 → 313 for 5-MTHF, m/z 474 → 327 for LV, and 136 → 94 for aminoacetophenone. The mean calibration curves for 5-MTHF (y = 0.1883x + 0.0187, R2) and LV (y = 0.0752x + 0.1836, R2 = 0.9973) were measured on different days (n = 9). Variability was determined by analysing plasma Q-sample (n = 7) at low, medium and high concentration and also between days (n = 9). The relative standard deviation (RSD) for 5-MTHF and LV ranged from 2 to 6.2% within the same day and the variability over 9 days ranged from 4.9 to 10.4%. Sensitivity was assessed by evaluating the limit of detection (LOD) and limit of quantification (LOQ) for the method. The LOD and LOQ were defined as the lowest analyte concentration yielding a signal-to-noise (S/N) ratio of 3 and 10, respectively. The LOD for 5-MTHF and LV was 0.3 and 1.6 pmol/ml, respectively. The LOQ for 5-MTHF and LV was 0.9 was and 5.2 pmol/ml, respectively.</p><!><p>LC–MS/MS was used to measure concentrations of the folate derivatives 5,10-MeTHF, THF, and 5-MTHF, expressed as pmol/g wet-weight (pmol/gww), in tumour tissue and adjacent mucosa [20]. Raltitrexed was used as an internal standard. The sum of the 5,10-MeTHF, THF, and 5-MTHF concentrations was used as a measure of folates in the tissue. The extracted ions following MRM transitions were monitored at m/z 446 → 299 for THF, m/z 458 → 311 for 5,10-MeTHF, m/z 460 → 313 for 5-MTHF, and m/z 459 → 312 for raltitrexed.</p><p>On the day of sample analysis, extraction buffer was prepared containing 50 mM phosphate buffer, pH 7.0, 1% ascorbate, and 0.1% β-mercaptopropanol. The tissue was weighed and placed in an Eppendorf vial and a 10 × volume of extraction buffer was added. Homogenization was performed using a TissueLyser (two disruption steps at 25 Hz for 2.5 min). After a deconjugation step, protein precipitation, centrifugation, and ultrafiltration (30 min at 21,500×g at 20 °C) were performed. The solution at the bottom of the test tube was used for LC–MS/MS analysis. The relative standard deviation (RSD) ranged from 2 to 7% for all analyses, and the variability over 4 days ranged from 3 to 14%. The accuracy of the method was determined by estimating the recovery by adding known amounts of the standard to a sample. The average recoveries were 98, 87, and 93% for THF, 5,10-MeTHF, and 5-MTHF, respectively. The LOD for 5,10-MeTHF, THF and 5-MTHF was 2.1, 1.2 and 0.3 pmol/gww, respectively. The LOQ for 5,10-MeTHF, THF and 5-MTHF was 4.0, 7.0 and 0.9 pmol/gww, respectively. Standard curves for 5,10-MeTHF, THF, and 5-MTHF in tissue have been presented in a previous paper [20].</p><p>The LC–MS/MS analyses were performed on a Waters 2795 LC separation module coupled to a Waters Micromass Quattro Triple-Quadrupole MS system with an electrospray ionization (ESI) source. Folates were detected and quantified using positive electrospray. The separation of folates was performed using an Atlantis dC18 3 µm, 2.1 × 100 mm column (Waters) together with the guard column Atlantis dC18, 3 µm, 2.1 × 10 mm. The mobile phase consisting of eluent A (0.1% of acetic acid in water) and eluent B (0.1% acetic acid in acetonitrile) was used. Calibration graphs were constructed by plotting the peak area ratio of each compound to internal standards against concentration. The standards and samples were processed using the QuanLynx quantitative processing tool in MassLynx (Waters Corp., Milford, MA, USA). A more detailed description of the folate assay development has been published previously [20].</p><!><p>Total RNA was isolated from 10 to 30 mg fresh-frozen tissue using the High Pure RNA Tissue Kit (#12033674001, Roche Diagnostics Scandinavia AB) according to the manufacturer's instructions. cDNA was synthesized using the High Capacity cDNA Reverse Transcription Kit (Applied Biosystems) and run on Gene Amp PCR System 9600 (Perkin Elmer). To optimize each run, the expression level of β-actin was determined in each sample. A second RNA extraction and cDNA synthesis was performed if the concentration was considered to be suboptimal.</p><!><p>Based on recent studies [19], five target genes with putative impact on LV metabolism were chosen for analysis (Supplementary Table 1). These genes are involved in folate transport (ABCC3, SLC19A1/RFC-1 and SLC46A1/PCFT), folate polyglutamation (FPGS and GGH) or folate metabolism (MTHFD1L). The relative gene expression was quantified in tumours and mucosa using real-time quantitative PCR (qPCR). TaqMan Gene Expression Assays (Life Technologies, Stockholm, Sweden) were ordered for each gene from Applied Biosystems at http://www.appliedbiosystems.com. The qPCR was set up in triplicates in 384-well plates using a Nanodrop II (GC Biotech) and was carried out in 5 µl reactions with 1 × TaqMan® Gene Expression Mastermix (Applied Biosystems), 1 × gene-specific TaqMan assay (Applied Biosystems), and 1 µl cDNA. The qPCR was run on a QuantStudioTM 12K Flex Real-Time PCR System (Life Technologies, Stockholm, Sweden) according to a standard protocol. The thresholds and baselines were set manually using the sequence detection systems software (SDS), version 2.4 (Applied Biosystems), and cycle threshold (Ct) values were extracted. Variations between runs were compensated for by normalization against a control sample. There was a linear correlation between the two house-keeping genes ACTB and GAPDH. All Ct values were normalized to a mean value representing both of these genes in order to keep variance to the minimum.</p><!><p>The JMP 11.0/SAS software (SAS Institute Inc. Cary, NC, USA) was used for the statistical analyses. All gene expression calculations were performed using the ΔΔCt method. As the values of the gene expression and folate concentrations were not normally distributed, they were transformed to be logarithmic. Differences between groups were calculated using the Kruskal–Wallis' test, the Pearson's Chi-square test, or the matched-pair analyses (Wilcoxon signed rank test) and data were presented as median and ranges. Values of p ≤ 0.05 were considered significant. No corrections for multiple testing were done.</p><!><p>Three patients were excluded from the study because routine pathology reports revealed a lack of adenocarcinoma tissue; two patients had an obstruction related to diverticulitis, one had a squamous epithelial cancer. Furthermore, during analysis of blood samples, it was discovered that one patient had received a LV dose that was not according to the protocol and, as a consequence, this patient was excluded from the study. Due to surgical complication with extensive bleeding during the operation, it was not possible to remove the tumour of one patient at the primary operation. This resulted in a time span between the LV injection and tissue collection of almost 30 h. Due to the extended time, the tissue values of folates were not reliable and the patient was excluded from the study. Thus, in total, 75 patients were included. Based on clinical diagnosis, 38 patients had colon cancer and 34 had rectal cancer. Three patients had cancer in both rectum and colon synchronously. The median time (min–max) that passed from LV injection to tissue sampling was 170 min (65–285) for patients receiving 60 mg/m2, 165 min (72–457) for patients receiving 200 mg/m2, and 163 min (65–555) for patients receiving 500 mg/m2. The variation was linked to the operating time, which differed depending on the type of surgical procedure. However, there was no significant difference in the time course from LV injection to tissue sampling between treatment groups (p = 0.91). The demographic, clinical and pathological data are shown in Table 1.</p><!><p>Clinicopathological characteristics of the patients with colorectal cancer sub-grouped by leucovorin dose</p><p>RT radiotherapy, CT chemotherapy</p><!><p>The ranges of the tissue folate concentration were very wide, reflecting a huge inter-individual variation (Table 2). The median folate concentration in both tumour and mucosa tissue increased with increasing dosage of LV (Fig. 1a; Table 2). As shown in Fig. 1a, about 50% of patients who received 60 mg/m2 did not reach a folate concentration in tumours above the concentration found in untreated patients. There was a strong and significant correlation between the folate concentration in tumour and mucosa tissue of untreated patients, as well as in patients treated with 60 mg/m2 LV (Fig. 1b), but no correlation was found in patients treated with 200 or 500 mg/m2 LV.</p><!><p>Comparison of median folate concentration and gene expressions levels in tumour and mucosa tissues of patients with CRC sub-grouped by leucovorin dose</p><p>a p by Wilcoxon signed rank test</p><p>bpmol/mg</p><p>a The folate concentration increased with increasing LV dosage in tumour tissue. The horizontal line marks the highest folate concentration found in untreated patients. b A significant correlation was seen between folate concentrations in tumour and mucosa of untreated patients (red dots, no LV, r = 0.63, p = 0.0048) and patients treated with 60 mg/m2 LV (green dots, r = 0.57, p = 0.011), however, the correlation was not significant in patients treated with 200 (blue dots) or 500 (purple dots) mg/m2 LV. The line of fit is shown with the confidence interval (blue-shaded area)</p><!><p>The median concentrations of LV and 5-MTHF in plasma of patients grouped by given LV dose are presented in Table 3. No correlation was found between LV and 5-MTHF in samples obtained from patients who received 60 mg/m2 LV (r = 0.36, p = 0.13). However, there was a strong, positive, correlation between the LV and 5-MTHF concentration in plasma obtained from patients who received 200 or 500 mg/m2 LV (r = 0.60, p = 0.0061 and r = 0.76, p = 0.0001, respectively).</p><!><p>Leucovorin and 5-MTHF levels in plasma of patients with CRC sub-grouped by leucovorin dose</p><p>anmol/ml</p><p>bpmol/ml</p><!><p>When the association between 5-MTHF concentration in plasma at baseline and folate concentration in mucosa was evaluated, a positive correlation was found (r = 0.5, p = 0.035). The maximum plasma concentration achieved for LV was found 10 min. after LV injection whereas the highest 5-MTHF was found after 30 min. Based on these values, the 5-MTHF/LV ratio was calculated for the patients of each group treated with LV (Supplementary Fig. 1). As shown, the ratio between 5-MTHF and LV decreased with higher dose of LV given and a great variation was seen between patients, especially in those who received 60 mg/m2. The 5-MTHF/LV ratio correlated negatively with folate concentration in mucosa (r = − 0.75, p = 0.0009). In patients treated with 200 or 500 mg/m2, however, the ratio correlated positively with folate concentration in tumour tissue (r = 0.51, p = 0.031 and r = 0.52, p = 0.019, respectively).</p><p>Gene expression was analysed in both tumour and mucosa tissues (Table 2). Similar to the folates, the variation in gene expression levels was very high. In untreated patients, the expression of GGH and MTHFD1L was significantly higher in tumour tissue compared to mucosa. After LV injection, a significantly higher gene expression level of GGH, MTHFD1L, SLC19A1/RFC-1, and FPGS was seen in tumour tissue compared to mucosa. However, this was not the case for ABCC3/MRP3 and SLC46A1/PCFTPCFT, which had higher expression levels in mucosa compared to tumour.</p><p>The expression of each gene increased with increasing dose of LV in tumour tissue after adjustment for time passed after LV injection (Fig. 2). However, there was no difference in expression levels according to the LV dose in mucosa (data not shown). The association between folate concentration and expression of each analysed gene in tumour and mucosa was evaluated (Table 4). A significant, negative correlation was seen between folate concentration and GGH and MTHFD1L expression in mucosa of untreated patients, but no correlation was seen in the tumour. In contrast, a significant, positive correlation was found between folate concentration and expression of all genes, except MTHFD1L, in patients who received LV (Table 4). In mucosa, the folate concentration correlated positively with expression of SLC46A1/PCFT.</p><!><p>Mean diamonds showing the difference in expression level of a ABCC3/MRP3 (p = 0.015), b FPGS (p = 0.044), c GGH (p = 0.05), d MTHFD1L (p = 0.18), e SLC19A1/RFC-1 (p = 0.047), and f SLC46A1/PCFT (p = 0.011) in tumour tissue of each group. The p values are based on the difference between patient groups treated with the lowest and highest doses of LV, i.e. 60 and 500 mg/m2. As shown, the expression increased with increasing dose of LV. The horizontal line in the centre of each diamond shows the mean of each group. The top and bottom points of the diamonds show the upper and lower 95% confidence points</p><p>Pairwise correlation of gene expression and folate concentration in tissue obtained from untreated patients and patients receiving 60, 200, or 500 mg/m2 of LV</p><p>a r = correlation coefficient</p><p>b p by Pearson</p><!><p>As shown in Table 5, there was a negative correlation between 5-MTHF concentration in plasma obtained from untreated patients and expression of GGH and SLC46A1/PCFT in tumour tissue, but no correlation was seen for any of the other genes. In mucosa of untreated patients, the 5-MTHF concentration tended to correlate negatively with SLC46A1/PCFT.</p><!><p>Pairwise correlation of gene expression in tissue and 5-MTHF levels in plasma obtained from untreated patients</p><p>a r = correlation coefficient</p><p>b p by Pearson</p><!><p>As previously stated, bolus treatment with 5-FU is given in combination with LV, which significantly enhances the therapeutic effect of 5-FU. FLV-treatment, both in the adjuvant and palliative setting, is usually combined with other chemotherapy drugs such as oxaliplatin or irinotecan, as well as specific antibodies. However, many patients do not respond to the chemotherapy, or will develop drug resistance during treatment. The inter-individual variation in treatment response may relate to differences in drug metabolism [21]. Although 5-FU-based treatment is a cornerstone in almost all CRC regimens, there is no clinically useful biochemical predictive marker that can be used to evaluate treatment effect.</p><p>As mentioned previously, LV has to be metabolized intracellularly to the active metabolite 5,10-MeTHF. The optimal intracellular concentration of folates for effective TS inhibition is presently not known but is expected to depend on the individual TS activity as well as the activity of several folate-associated enzymes. In order to get a complete inhibition of the target enzyme TS, 5,10-MeTHF must be in excess. Consequently, a low 5,10-MeTHF concentration in the tumour tissue is associated with clinically impaired 5-FU activity. High doses of LV are needed to achieve a maximum formation of the ternary complex between TS, FdUMP, and 5,10-MeTHF. LV is a racemic mixture of the natural (S) and unnatural (R) diastereoisomers of 5-formyltetrahydrofolate. In contrast to the natural isomer, which disappears rapidly from plasma, the unnatural form has a much longer half-life [12]. High concentration of the unnatural isomer can interfere with the transport and metabolism of the natural isomer. Both of the isomers can interact with TS, and inhibit its activity and thereby also the effectiveness of 5-FU chemotherapy [22].</p><p>The results of the present study showed that there was a high inter-individual variation of folates in both tumour and mucosa. This finding is in agreement with previous results published by our group [18] as well as by others [23]. Higher concentrations were found in tumour tissue compared to mucosa in all patient groups. The folate concentration increased with increasing LV dosage in both tumour and mucosa tissue. It is noteworthy that for many patients who received a dose of 60 mg/m2 LV, the resulting folate concentration in tumour tissue was as low, or even lower, than the concentration found in untreated patients. This result indicates that some patients may be unable to transport and/or metabolize LV properly. The low tissue folate concentration in some patients may also reflect an initial folate deficiency that requires a higher dosage of LV to achieve satisfying concentrations. Other results of the study points in the same direction; there was a strong and significant correlation between folate concentration in tumour and mucosa obtained from untreated patients as well as in those treated with 60 mg/m2 LV, but not between tumour and mucosa of patients treated with 200 or 500 mg/m2 LV. The lack of correlation in the latter groups may be due to faster saturation of folates in tumour tissue. Differences between patients treated with 60 versus 200 or 500 mg/m2 LV were also reflected in plasma folates and were especially apparent when the 5-MTHF/LV ratio was compared between groups. The 5-MTHF/LV ratio might be useful as a surrogate marker for folate concentration in tumour tissue and, thus, clinically relevant.</p><p>In a previous retrospective study, the expression of 22 folate pathway genes having possible impact on the metabolism of LV was analysed in tumours deriving from 193 patients with stage III CRC treated with bolus adjuvant FLV chemotherapy [24]. The analysed genes were involved in folate transport, polyglutamation and metabolism. The result showed that high expression of two genes involved in folate transport, SLC46A1/PCFT and SLC19A1/RFC-1, correlated positively with longer disease-free survival of the patients. It was hypothesized that poor response to FLV therapy in some patients was linked to low expression of these genes. In the present study, the expression of SLC46A1/PCFT and SLC19A1/RFC-1, as well as four other folate pathway genes (ABCC3/MRP3, FPGS, GGH, and MTHFD1L) was determined.</p><p>The ABCC3/MRP3 gene encodes a protein that belongs to the superfamily of adenosine triphosphate (ATP)-binding cassette (ABC) transporters. ABCC3/MRP3 carries out an outward transport of a variety of molecules, including monoglutamated forms of reduced folates [25]. The enzyme FPGS converts folate monoglutamates to polyglutamates [26]. Polyglutamate forms of 5,10-MeTHF are better stabilizers of the ternary complex with TS and FdUMP. Studies have shown that LV is ineffective if cells are incapable of metabolizing folate to polyglutamate because the ternary complex will dissociate more readily in the absence of these polyglutamates. The enzyme GGH, on the other hand, cleaves folate polyglutamates to monoglutamates [27]. It has been suggested that GGH regulates the intracellular folate concentration [28]. MTHFD1L is another gene that seems to have a critical role in folate cycle maintenance [29]. The gene encodes a mitochondrial enzyme that catalyses the conversion of 10-formylTHF to THF and formate [30]. Studies have shown that an increased level of MTHFD1L may support colorectal cancer growth [31].</p><p>The results of the study showed that tumour tissue of untreated patients had higher expression of FPGS, GGH, MTHFD1L, and SLC19A1/RFC-1 compared with mucosa, whereas expression of ABCC3/MRP3 and SLC46A1/PCFT was higher in mucosa compared with tumour. These results are in agreement with those of our previous study with patients with stage III CRC, where tumour and mucosa tissue were obtained at primary surgery, before onset of adjuvant treatment [24]. Higher expression of ABCC3/MRP3 in mucosa compared with tumour tissue has also been reported by Kobayashi et al. at both the mRNA and protein level [32]. For each of the analysed genes, the expression in tumour increased with increasing dose of LV. Such an increase was not seen in mucosa. After LV injection, the expression of FPGS, GGH, MTHFD1L, and SLC19A1/RFC-1 was significantly higher in tumour tissue compared to the mucosa. However, this was not the case for ABCC3/MRP3 and SLC46A1/PCFT, which consistently had higher expression levels in mucosa, compared to tumour in each of the treatment groups.</p><p>Several previous studies have shown that folate deficiency results in a significant upregulation of folate transport genes such as SLC19A1/RFC-1 and SLC46A1/PCFT [33, 34]. Folate deficiency also results in decreased half-lives of these two genes [35]. In the present study, no correlation between gene expression and folate concentration was found in tumour tissue of untreated patients; however, the gene expression levels of GGH and SLC46A1/PCFT in tumour were negatively correlated with 5-MTHF in plasma (i.e. low folates correlated with high expression). Furthermore, high GGH and MTHFD1L expression was associated with low folate concentration in mucosa of untreated patients. This negative correlation may reflect a folate deficiency. After LV treatment, a significant positive correlation was seen between the folate concentration and expression of all genes, except MTHFD1L, in tumour tissue. However, only SLC46A1/PCFT gene expression correlated with folate concentration in mucosa. Studies have shown that the expression level of folate-associated genes in tumour tissue are different compared to mucosa. This disparity may relate to the folate status but also to differences in the regulation of folate metabolism, as has been suggested by Sadahiro et al. [36].</p><p>There are limitations of the present study. Firstly, the number of patients included in each group was limited, which may have affected the statistical calculations. However, similar results have been obtained previously in a large and unrelated study, at least at the gene expression level, which strengthens the results. Secondly, tissue was only obtained at one occasion for each of the patients. Thus, no baseline values could be compared with values after LV treatment at the individual level. A larger study is now being planned where samples will be obtained from each patient before and after treatment with LV in combination with 5-FU. Thirdly, the time span for sampling of plasma 5-MTHF was too short to detect the peak level of this metabolite. In the planned study, plasma samples will be obtained during a longer time span in an attempt to detect the peak concentration. Fourthly, the time to tissue sampling after LV injection differed between patients, however, as the time to injection was known, it could be adjusted for in the statistical calculations.</p><!><p>In summary, there was a high inter-individual variation in tissue and plasma folate concentration in response to administration of LV among patients. Half of the patients who received 60 mg/m2 LV did not reach a folate concentration in tumour tissue above the level of untreated patients. Thus, a large part of patients with CRC may benefit from higher LV doses than recommended, or treatment with folate forms with different metabolic profiles. The fact that a strong correlation between mucosa and tumour tissue was found suggests that there is a possibility to use the remnant colorectal mucosa after surgery as a surrogate for tumour tissue during chemotherapy treatment. Prediction of the folate concentration in tissue before treatment with LV would be valuable in order to identify patients who would benefit from higher doses than commonly used. The results further indicate that it might be possible to use the individual response, measured as the ratio between 5-MTHF and LV concentrations in plasma after LV injection, as a surrogate marker for the folate concentration in the target tissue. In the CRC metastatic setting, the possibility to analyse a blood sample would be a useful tool for individualization of folate-based treatment. Analysis of folate-associated gene expression in tissue obtained either from primary tumours, or metastatic lesions in a palliative setting, might become useful when choosing the most optimal LV dose needed to maximize tissue concentration of 5,10-MeTHF. However, the results need to be confirmed in extended studies.</p><!><p>ESM1 (DOCX 41 kb)</p>

PubMed Open Access

Synthesis of 10-<i>O</i>-aryl-substituted berberine derivatives by Chan–Evans–Lam coupling and investigation of their DNA-binding properties

Eleven novel 10-O-aryl-substituted berberrubine and berberine derivatives were synthesized by the Cu 2+ -catalyzed Chan-Evans-Lam coupling of berberrubine with arylboronic acids and subsequent 9-O-methylation. The reaction is likely introduced by the Cu 2+ -induced demethylation of berberrubine and subsequent arylation of the resulting 10-oxyanion functionality. Thus, this synthetic route represents the first successful Cu-mediated coupling reaction of berberine substrates. The DNA-binding properties of the 10-O-arylberberine derivatives with duplex and quadruplex DNA were studied by thermal DNA denaturation experiments, spectrometric titrations as well as CD and LD spectroscopy. Fluorimetric DNA melting analysis with different types of quadruplex DNA revealed a moderate stabilization of the telomeric quadruplex-forming oligonucleotide sequence G 3 (TTAG 3 ) 3 . The derivatives showed a moderate affinity towards quadruplex DNA (K b = 5-9 × 10 5 M −1 ) and ct DNA (K b = 3-5 × 10 4 M −1 ) and exhibited a fluorescence light-up effect upon complexation to both DNA forms, with slightly higher intensity in the presence of the quadruplex DNA. Furthermore, the CD-and LD-spectroscopic studies revealed that the title compounds intercalate into ct DNA and bind to G4-DNA by terminal stacking.

synthesis_of_10-<i>o</i>-aryl-substituted_berberine_derivatives_by_chan–evans–lam_coupling_and_inves

Introduction<!>Synthesis<!>Absorption and emission properties<!>DNA-binding properties<!>Spectrometric titrations<!>CD and LD spectroscopy<!>Conclusion

<p>Berberine (1a) is the most prominent member of the protoberberine family, i.e., a group of tetracyclic isoquinolinium alkaloids [1]. Like many members of this family, berberine (1a) is a natural product and may be isolated from different plants such as berberis vulgaris, hydrastis canadensis or coptidis rhizome [2]. Particularly, the latter is established in traditional Chinese medicine as source for anti-inflammatory extracts [2]. Berberine has also been employed in modern medicine because of its antimicrobial [3], antiprotozoal [4], antiviral [5], and anti-inflammatory [6] activity, and it is used in the treatment of tuberculosis [7], diarrhea [8], diabetes [9], cardiovascular diseases [10] or high cholesterol levels [11]. Most interestingly, it has been found that berberine (1a) exhibits a selective cytotoxicity against cancer cells [12,13], which is mainly based on its DNAbinding properties [14]. Since the complexation with DNA leads to significant changes of the fluorescent quantum yields of the bound berberine, it can also be used as fluorescent probe in bioanalytical chemistry [15,16]. Because of these attractive properties of this lead structure, numerous berberine derivatives have been synthesized in order to further improve the biological activity [17][18][19][20][21][22][23][24][25][26]. Among those derivatives, specifically 13-and 9-substituted berberine derivatives have gained much interest as selective DNA ligands that specifically target quadruplex DNA (G4-DNA) [23][24][25][26][27][28][29], i.e., non-canonical, biologically relevant DNA structures that are formed by the association of four guanine-rich DNA strands [30]. Due to their straightforward synthetic accessibility by substitution at the 9-hydroxy functionality of berberrubine (1b), 9-O-substituted berberine derivatives are particularly attractive. But so far, only functionalization at this position by alkylation [23][24][25][26] and nucleophilic substitution [17][18][19][20][21] of berberrubine (1b) have been performed, whereas the direct O-arylation of berberrubine has not been accomplished, yet, most likely because the substrate is not fully compatible with the usual conditions of the corresponding Ullmann or Buchwald-Hartwig coupling reactions. Nevertheless, it has been recently reported that 9-O-arylsubstituted berberine derivatives can be isolated by an Ullmanntype arylation of a tetrahydroberberrubine and a subsequent oxidation of the primary product to the respective berberine derivatives 2a-i (Figure 1) [31].</p><p>So far, a successful direct arylation of berberrubine (1b) has not been reported. At the same time, the Chan-Evans-Lam cross coupling has been established as a useful and relatively mild method for metal-mediated arylation of hydroxyarenes [32,33]. Therefore, we proposed that this method may be suitable for the arylation of berberrubine. And herein, we report the application of this Cu-mediated coupling reaction for the synthesis of 10-Oarylated berberine derivatives as unexpected reaction product, along with first experiments that demonstrate the G4-DNAbinding properties of this class of berberine derivatives.</p><!><p>In first orienting experiments, the reaction of berberrubine (1b) and phenylboronic acid (3a) was performed under typical conditions [34] of a Chan-Evans-Lam coupling with Cu(OAc) 2 as catalyst and triethylamine as base in CH 2 Cl 2 to give an isolated product in only 5% yield (Scheme 1). Further attempts to optimize the reaction conditions showed that the reaction mostly led to decomposition of the starting material in polar protic solvents or in the presence of bases or additives other than triethylamine (cf. Supporting Information File 1). However, the product was isolated in a moderate yield of 26% when the reaction was performed at room temperature in DMF as solvent. Notably, the yield was slightly lower (19%) when the reaction was performed at 40 °C, indicating that at higher temperatures side reactions are even more favored, such as the degradation of berberrubine under alkaline and/or oxidative conditions [35][36][37]. Most surprisingly, the NMR-spectroscopic data of the product were not consistent with the expected 9-Oarylated product. Moreover, in control experiments with 4-chlorophenylboronic acid (3b) the same unexpected product was formed. The product from the latter reaction was exemplarily studied by NOESY NMR experiments. Based on these results, the formation of the 10-O-arylated regioisomer 4b was unambiguously confirmed by a clear NOE crosspeak between the protons 2'-H and 6'-H of the O-aryl group and the 11-H proton of the berberine (Figure S12, Supporting Information File 1), whereas an NOE between the O-aryl group and the 8-H proton, as expected for the 9-O-arylated product, was not observed. Furthermore, mass spectrometric data and elemental analysis data were consistent with the structure assignment of product 4b.</p><p>To assess whether the formation of the regioisomeric product 4a is a general feature under these conditions, the reaction was conducted also with other representative boronic acids (Scheme 1). Although no products could be isolated after the reaction with 4-dimethylamino-, 3,4,5-trimethoxy-and 4-methylphenylboronic acid, a series of 10-O-arylated products 4b-f was available in low yields (16-26%) by the Chan-Evans-Lam coupling of berberrubine (1b, Scheme 1). The formation of the 10-O-arylated products may be explained by an initial Lewis acid-catalyzed demethylation of berberrubine (1b) [38] to derivative 6 (Scheme 2) that reacts subsequently in the Cu 2+ -catalyzed coupling reaction with the boronic acid (Scheme 2). The regioselectivity of the latter reaction step is most likely determined by a stronger nucleophilicity of the oxyanion in the 10-position of 6 that is caused by the particular electron distribution in the intermediate 6, specifically because of the decreased electron density at the 9-oxyanion by linear conjugation with the quaternary nitrogen atom [39]. Nevertheless, the reactivity of intermediate 6 is still relatively low, as indicated by the low yields of the Cu 2+ -catalyzed coupling reaction, thus resembling the parent berberrubine (1b), which has been shown to give no products in an Ullmann reaction [31]. Furthermore, the sensitivity of the isoquinolinium unit towards alkaline conditions and redox-active transition metal ions, especially under aerobic conditions, may also cause the low yields [35][36][37].</p><p>Unfortunately, the products 4a-f are hardly soluble in aqueous solution, presumably due to their zwitterionic structure, which hampers their use in biological studies. To ensure a sufficient solubility, these derivatives were methylated in the 9-O-posi-tion by the reaction with iodomethane under mild alkaline conditions to give the corresponding 9-methoxy-substituted derivatives 5a-e in moderate to good yields of 40-77% (Scheme 1). During the methylation of derivative 5f a transesterification occurred and a 2:1 mixture of the ethyl and methyl ester was formed that could not be further separated. The novel compounds 4a-f and 5a-e were identified and fully characterized by NMR-spectroscopic analysis ( 1 H, 13 C, COSY, HSQC, HMBC), mass spectrometric data and elemental analyses.</p><!><p>The absorption properties of the derivatives 5a-e resemble the ones of the berberine chromophore with long wavelength maxima between 403 nm (5e in aqueous buffer) and 426 nm (5d in CHCl 3 ) (Table 1, Table S2, Figure S1, Supporting Information File 1) [40]. The emission intensity of all compounds is very low (Φ fl << 0.01) and mostly not detectable in the series of tested solvents. To examine whether the low fluorescence quantum yields are caused by conformational changes in the excited state, the fluorescence was recorded in media with different viscosity, namely in glycerol at different temperatures (Figure 2, Figure S2, Supporting Information File 1). It was observed that the derivatives 5a-e have significantly higher, but still relatively low emission quantum yields in glycerol at room temperature (Φ fl = 0.009-0.031, 20 °C, η = 1412 cP) [41], whereas the emission intensity decreased with increasing temperature (Φ fl = 0.001-0.002, 80 °C, η = 32 cP) [41]. Such a behavior usually indicates a radiationless deactivation of the excited state by conformational changes, e.g., torsional relaxation [42,43]. Since the fluorescence quantum yield of the parent berberine (1a) does not correlate well with the viscosity of the medium [40], it was concluded that the weak light-up effect in glycerol is mainly caused by the suppressed rotation about the Ar-O bond. Nevertheless, as the emission quantum yield of the derivatives 5a-e still remained low, even at high viscosity of the medium, there obviously exist additional relaxation pathways in the excited state, most likely a photo-induced electron transfer (PET) from the 10-aryl substituent to the berberine chromophore. The latter has been shown to operate also in resembling cationic, biaryl-type dyes [42,43]. Along the same lines, the low intrinsic emission quantum yield of the parent berberine (1a) has been suggested to result from an internal charge transfer (ICT) process from the electron-rich benzodioxole unit to the isoquinolinium [44], which also contributes to the low fluorescence intensity of derivatives 5a-e.</p><!><p>Thermal DNA denaturation experiments</p><p>For a first screening of the interactions of the derivatives 5a-e with different G4-DNA forms, the effect of thermally induced unfolding of dye-labeled, quadruplex-forming oligonucleotides in the presence of the ligands was studied. In general, the binding of the ligand to the G4-DNA leads to a stabilization and thus to an increasing melting temperature of the DNA (Table 2). This effect of the ligands on the quadruplex melting tempera- TGAG 3 TG 3 TAG 3 TG 3 TA-tamra] and FkitT [fluo-AG 3 AG 3 CGCTG 3 AG 2 AG 3 -tamra] was determined by fluorimetric monitoring of the temperature-dependent Förster resonance energy transfer (FRET) between the dyes [46]. The particular oligonucleotide sequences were chosen because they are known to be involved in biologically relevant processes, namely in the transcription regulation of myc (FmycT) [47,48], kit (FkitT) [49] and insulin (a2) [50], or in telomerase inhibition (F21T) [30,51,52]. At a ligand-DNA ratio (LDR) of 5, the derivatives 5b-e induced a small, but significant shift of the melting temperature of the oligonucleotides F21T (∆T m = 1.8-3.9 °C), and FmycT (∆T m = 1.3-2.1 °C), while the melting temperatures of the sequences Fa2T and FkitT were only affected marginally by the presence of all ligands. Notably, derivative 5a exhibited only a stabilizing effect on F21T (∆T m = 3.2 °C), whereas the melting temperature of all other oligonucleotides was not affected significantly. The induced ∆T m values of F21T are comparable to the one of the parent berberine (1a) under similar conditions (∆T m = 3.1 °C) [27]. As compared to other known G4-DNA ligands [23,[27][28][29], these shifts of the melting temperature in the presence of 5b-e are rather small and indicate a relatively weak stabilizing effect on the quadruplex structures. Although the ∆T m values do not nec-essarily correlate directly with the binding affinity [23], as they only refer to the stabilization at elevated temperatures, the data show that these ligands do not bind extremely strong to G4-DNA. Nevertheless, as these screening experiments revealed the most pronounced effect of the ligands on the ∆T m values of F21T, the binding interactions with the corresponding unlabeled telomeric oligonucleotide sequence d[A(G 3 TTA) 3 G 3 ] (22AG) were studied in more detail.</p><!><p>The interactions of derivatives 5a-e with calf thymus (ct) DNA, as a representative duplex DNA, and 22AG were monitored by photometric and fluorimetric titrations. In almost all cases, a decrease of the absorption bands at 403-412 nm and 343 nm was observed upon addition of the DNA, along with a red shift of the absorption bands (Figure 3, Figure S3, Supporting Information File 1). Only the absorption of compound 5a increased upon association with ct DNA. In general, a significant red shift of the absorption bands was observed during all titrations, which was more pronounced in the presence of 22AG (∆λ = 18-24 nm) than in the presence of ct DNA (∆λ = 8-12 nm). However, no isosbestic points were observed during these titrations, which indicated different binding modes at the particular ligand-DNA ratios (LDRs). The resulting binding isotherms obtained from the photometric titrations were employed to determine the binding constants K b (Table 3, Figure S5, Supporting Information File 1) [53]. As a general trend, all derivatives showed a slightly higher affinity towards G4-DNA (K b = 5.2 × 10 5 M −1 to 8.7 × 10 5 M −1 ) than to ct DNA (K b = 2.5 × 10 4 M −1 to 5.1 × 10 4 M −1 ). As compared with the parent berberine (1a), the binding affinity towards G4-DNA (1a: K b = 4.5 × 10 5 M −1 ) [16] is a bit higher, whereas the binding constants towards ct DNA are somewhat lower (1a: K b = 9.7 × 10 4 M −1 ) [54]. Hence the affinity of the ligands 5a-e to duplex and quadruplex DNA lies in a similar range as the one of 1a, so that the same DNA-targeted bioactivity of these substrates is assumed [14].</p><p>Upon addition of DNA the intensity of the emission bands of derivatives 5a-d, that were hardly detectable in the absence of DNA, increased slightly with small shifts of the emission maxima (ct DNA: 520-529 nm; G4-DNA: 514-516 nm, Figure 4, Figure S4, Supporting Information File 1). Nevertheless, the fluorescence quantum yield remained rather low (Φ fl < 0.01) for all DNA-bound derivatives with a slightly more pronounced increase in the presence of G4-DNA (Φ fl = 0.010-0.017). In the case of derivative 5e, the emission intensity remained essentially not detectable upon addition of DNA, presumably caused by the nitro group that is known to be an efficient emission quencher. It was demonstrated with experiments in glycerol solution that the emission of 5a-d increases also upon suppression of conformational freedom of the molecule in media with high viscosity and limited free volume (Figure 2). Therefore, it is proposed that the association with the DNA causes a similar effect. Hence, the slightly increased emission intensity of the ligands 5a-d upon addition of DNA supposedly originates from the restricted conformational freedom of the ligand within the DNA binding pocket, which suppresses the torsional relaxation as non-radiative deactivation pathway [42,43]. Thus, the relatively stronger emission enhancement upon binding of the ligands 5a-d to G4-DNA is likely caused by a tighter accommodation of the aryl substituent within the G4-DNA binding pocket.</p><!><p>Solutions of ligands 5a-e in the presence of ct DNA were examined with flow linear dichroism (LD) and circular dichroism (CD) spectroscopy (Figure 5, Figure S6, Supporting Information File 1). The binding of ligands 5a-e resulted in the development of negative LD signals at 350-355 nm and 420-429 nm clearly indicating an intercalative binding mode of the ligand, since these bands result from a coplanar alignment of the aromatic system of the ligand to the DNA bases [55,56]. Furthermore, all derivatives developed positive induced CD (ICD) signals in the absorption range of the ligands upon addi-tion of ct DNA. However, except for ligand 5d the intensity of the ICD signals remained rather low, which has already been observed for several berberine derivatives [25,26,57,58]. Altogether, especially considering the steric demand of the aryloxy substituent, the LD-and CD-spectroscopic data suggest a coplanar arrangement between DNA base pairs and the isoquinolinium unit pointing into the groove. This model is further supported by the relatively low fluorescence quantum yields of all ligands upon complexation to duplex DNA because in this structure the aryl substituent still has some conformational flexibility in the binding pocket leading to reduced fluorescence quantum yields.</p><p>The characteristic CD spectra of the G4-DNA 22AG [57,58] only changed marginally in the presence of the derivatives 5a-e; namely, only a small increase (5b) or a small decrease (5a, 5c-e) of the characteristic CD band at 290 nm was observed with no obvious general trend. Most notably, a decrease of the shoulder at 255 nm was observed, in the case of 5e together with a development of a new, slightly blue-shifted negative band, with increasing content of derivatives 5a, 5b and 5d. This development of the CD spectrum is characteristic of a shift of the equilibrium between the [3 + 1] conformer, related to the shoulder at 255 nm [59], and the basket-type conformation of 22AG, assigned to the positive signal at 290 nm and the weak negative band at 260 nm [60,61] in favor of the latter. Hence, these observations showed that the ligands bind preferentially to the basket-type quadruplex structure and thereby shift the equilibrium to this form. Furthermore, during all titrations of the derivatives 5a-e to 22AG no clear ICD band was detected, which is usually interpreted as an indication of terminal π stacking of the ligand to the quadruplex structure [62][63][64], however, this interpretation of an absent signal has to be applied very carefully.</p><!><p>In summary, we demonstrated that 10-O-arylated berberine derivatives are accessible in low to moderate yield by the Chan-Evans-Lam coupling reaction of berberrubine (1b) and subsequent methylation. The straightforward synthetic route enables the synthesis of a new class of berberine derivatives from easily accessible starting materials. The derivatives bind with slightly higher affinity to G4-DNA as compared to the parent berberine (1a) and induce a moderate stabilization of telomeric quadruplex 22AG. CD-and LD-spectroscopic studies revealed an intercalative binding mode with ct DNA and most likely terminal stacking as binding mode with G4-DNA. Lastly, all derivatives experienced a weak light-up effect upon complexation to DNA, which was slightly more pronounced upon binding to G4-DNA as compared to ct DNA. Since the ligands show essentially the same DNA binding properties as the parent berberine (1a), they have the potential to exhibit a similar DNA-targeted bioactivity. The latter may even be improved due to the higher lipophilicity and thus more balanced bioavailability of the ligands [65]. Accordingly, the activity of these derivatives in the treatment of tuberculosis [7], diarrhea [8], diabetes [9], cardiovascular diseases [10] or high cholesterol levels [11] is worth to be tested, because the parent berberine is already employed as drug for these diseases. In summary, a new complementary class of arylated berberine derivatives was discovered, which may constitute a promising starting point for the development of lead structures in drug discovery.</p>

Beilstein

Thrombospondin-1 and Angiotensin II Inhibit Soluble Guanylyl Cyclase through an Increase in Intracellular Calcium Concentration

Nitric Oxide (NO) regulates cardiovascular hemostasis by binding to soluble guanylyl cyclase (sGC), leading to cGMP production, reduced cytosolic calcium concentration ([Ca2+]i) and vasorelaxation. Thrombospondin-1 (TSP-1), a secreted matricellular protein, was recently discovered to inhibit NO signaling and sGC activity. Inhibition of sGC requires binding to cell-surface receptor CD47. Here, we show that a TSP-1 C-terminal fragment (E3CaG1) readily inhibits sGC in Jurkat T cells, and that inhibition requires an increase in [Ca2+]i. Using flow cytometry, we show that E3CaG1 binds directly to CD47 on the surface of Jurkat T cells. Using digital imaging microscopy on live cells, we further show that E3CaG1 binding results in a substantial increase in [Ca2+]i, up to 300 nM. Addition of angiotensin II, a potent vasoconstrictor known to increase [Ca2+]i, also strongly inhibits sGC activity. sGC isolated from calcium-treated cells or from cell-free lysates supplemented with Ca2+ remains inhibited, while addition of kinase inhibitor staurosporine prevents inhibition, indicating inhibition is likely due to phosphorylation. Inhibition is through an increase in Km for GTP, which rises to 834 \xc2\xb5M for the NO-stimulated protein, a 13-fold increase over the uninhibited protein. Compounds YC-1 and BAY 41-2272, allosteric stimulators of sGC that are of interest for treating hypertension, overcome E3CaG1-mediated inhibition of NO-ligated sGC. Taken together, these data suggest that sGC not only lowers [Ca2+]i in response to NO, inducing vasodilation, but is also inhibited by high [Ca2+]i, providing a fine balance between signals for vasodilation and vasoconstriction.

thrombospondin-1_and_angiotensin_ii_inhibit_soluble_guanylyl_cyclase_through_an_increase_in_intracel

<!>Materials<!>Cell culture<!>Flow cytometry<!>Expression and purification of E3CaG1<!>Cloning and transient transfection of human soluble guanylyl cyclase<!>sGC immunoprecipitation and activity assays<!>sGC activity and cGMP accumulation in intact cells and cell lysates<!>Cell-free inhibition of sGC<!>sGC activity and cGMP accumulation in lysed Jurkat T cells<!>Calcium imaging<!>Statistical analysis<!>CD47 is necessary but insufficient for E3CaG1 binding to Jurkat T cells and inhibition of sGC<!>E3CaG1 induces an increase in [Ca2+]i<!>E3CaG1-dependent increases in [Ca2+]i requires CD47<!>Calcium inhibits NO-inducible sGC activity in Jurkat T cells<!>Angiotensin II and phytohemoagglutinin inhibit NO-driven cGMP accumulation<!>Phosphodiesterases have minimal effect<!>Compounds YC-1 and Bay 41-2272 overcome Ca2+ inhibition of sGC in live cells<!>Immunoprecipitated sGC remains inhibited, but activity is recovered upon binding compounds YC-1 and Bay 41-2272<!>Inhibited sGC exhibits an increase in Km<!>Inhibition of sGC requires a calcium-dependent kinase<!>DISCSUSSION

<p>Nitric oxide (NO)1 regulates numerous vital functions in animal physiology including blood pressure, memory formation, platelet aggregation, angiogenesis and tissue development (1). Dysregulation of NO signaling contributes to cardiovascular disease, difficulties in wound healing, diabetes, asthma, and aging. NO is produced through the conversion of L-arginine to L-citrulline by nitric oxide synthase (NOS) (2, 3) and may function in the same cell where it is produced and/or in nearby cells (autocrine/paracrine signaling). Three isoforms of NOS are found in mammals, endothelial NOS (eNOS), neuronal NOS (nNOS) and inducible NOS (iNOS). Both eNOS and nNOS are regulated by Ca2+ through its binding to calmodulin. The primary NO receptor is soluble guanylyl/guanylate cyclase (sGC), a heterodimeric protein of ~150 kDa that binds NO through a ferrous heme. NO binding stimulates cyclase activity, the production of cGMP from substrate GTP and the subsequent amplification of NO-dependent signaling cascades (4–7). In smooth muscle cells, this leads to a reduction in cytosolic calcium concentration ([Ca2+]i) and smooth muscle relaxation, a mechanism closely tied to the regulation of blood pressure. While regulation of NOS is relatively well studied (8), the mechanisms underlying sGC regulation are poorly understood (4).</p><p>Recently, thrombospondin-1 (TSP-1), a trimeric extracellular matrix protein of 450 kDa, was discovered to be an inhibitor of NO signaling (9). The NO-stimulated increase in endothelial cell proliferation, migration and adhesion, which are of importance for angiogenesis, wound healing and tumor progression, are potently blocked by TSP-1. TSP-1 also blocks smooth muscle relaxation, leading to vasoconstriction. The mechanisms behind TSP-1 attenuation of NO signaling are not yet known but involve inhibition at multiple steps, including those involving vascular endothelial growth factor receptor-2 (VEGFR2), eNOS, sGC and protein kinase G (PKG) (10, 11). Among these, inhibition of sGC is particularly prominent and the focus of the present investigation.</p><p>TSP-1 is a multi-domain protein consisting of a globular N-terminal domain, procollagen homology domain, three thrombospondin structural or properdin-like (type 1) repeats, three EGF-like (type 2) repeats, seven Ca2+-binding (type 3) repeats and a globular C-terminal cell-binding domain (Fig. 1) (12–14). The trimeric form of the protein is stabilized through disulfide bonds located just after the N-terminal domain. TSP-1 interacts with multiple cell surface receptors through each of its domains and elicits a multitude of physiological responses. Through its C-terminal domain, TSP-1 binds to CD47 (also called integrin-associated protein; IAP), which is required for sGC inhibition (9).</p><p>CD47 is an ~50 kDa integral-membrane protein expressed in most cell types. It is suspected to traverse the membrane five times, and has an IgV-like extracellular domain and a small alternatively-spliced intracellular domain (15). The two well-characterized ligands of CD47 are signal inhibitory receptor protein α (SIRPα) and TSP-1. The CD47/SIRPα interaction functions to regulate innate immunity and experiments using knockout mice reveal that CD47 could act as a "self" marker since lack of CD47 leads to cells being phagocytosed by macrophages (16). CD47 can be co-immunoprecipitated with G-protein Gi (17) and is implicated in triggering Gi-dependent apoptosis in both breast cancer cells (18) and T lymphocytes (19). This has led to the suggestion that CD47 might be a non-canonical G-protein coupled receptor (GPCR) and that CD47/integrin complexes mimic GPCRs (15, 20). When TSP-1 binds to CD47 at the cell surface there is a decrease in cGMP production due to the reduced ability of NO to stimulate sGC. Full length TSP-1, a peptide derived from the C-terminus of TSP-1 (4N1), and a C-terminal fragment of TSP-1 (E3CaG1) have all been shown to inhibit NO signaling through a reduction in sGC activity (9, 10). Previous studies indicate that TSP-1 inhibition of NO signaling is directly through sGC and not, for example, through inhibition of phosphodiesterases (21, 22). Additionally, 4N1 or 4N1K (modified 4N1) and TSP-1 binding to CD47 have each been shown to increase [Ca2+]i levels in mast cells (23) and fibroblasts (24).</p><p>sGC is a heterodimeric enzyme with one alpha subunit of ~77 kDa and one heme-containing beta subunit of ~70 kDa. Each subunit consists of an N-terminal H-NOX domain, central PAS and coiled-coil domains and a C-terminal catalytic domain (25). Sub-cellular localization (26, 27), dimerization status (28), phosphorylation (29–34), protein-protein interaction (35–38), S-nitrosation (39, 40) and [Ca2+]i levels (41–43) have all been implicated in sGC regulation. Calcium, nitric oxide and cGMP are intimately associated in controlling numerous cellular functions, especially vascular tone. High [Ca2+]i levels lead to attenuation of NO-induced cGMP accumulation in transformed HEK 293 cells (41), in primary astrocytes (44) and in primary pituitary gland cells (45), and micromolar calcium concentrations can directly inhibit isolated sGC (41, 42, 46).</p><p>Based on the foregoing, we hypothesized that TSP-1 inhibition of sGC was mediated through Ca2+ signaling. Here, we show that E3CaG1 binding to Jurkat T cells leads to an increase in [Ca2+]i, and that this pulse is required for inhibition of sGC. We also show that a potent vasoconstrictor, angiotensin II (Ang II), which induces an increase in [Ca2+]i through GPCR AT1 (47, 48), also inhibits sGC through a Ca2+-dependent mechanism.</p><!><p>FITC-conjugated monoclonal anti-human CD47 antibody (B6H12) and an isotype control antibody were obtained from BD Biosciences (San Jose, CA). Anti-integrin antibodies to αV (P2W7 and 272-17E6) were obtained from Abcam. Ionomycin, thapsigargin, PHA and BAPTA-AM were obtained from Invitrogen (Carlsbad, CA). Fura-2AM was obtained from CalBiochem/EMD Biosciences (San Diego, CA). 2-(N,N-Diethylamino)-diazenolate-2-oxide (DEA/NO) was a kind gift from Dr. Katrina Miranda (University of Arizona). Phosphate-buffered saline (PBS) was prepared as 10 mM KH2PO4, 10 mM Na2HPO4, 137 mM NaCl, 2.7 mM KCl, pH 7.4. Tris-buffered saline (TBS) was prepared as 10 mM Tris.HCl, 150 mM NaCl, pH 7.4. Krebs buffer was prepared as 25 mM HEPES, 120 mM NaCl, 4.75 mM KCl, 1.44 mM MgSO4, 11 mM glucose, pH 7.4. All other reagents were obtained from Sigma unless otherwise noted.</p><!><p>Sf9 cells were maintained in Grace's Insect Media (Invitrogen) supplemented with 10% fetal bovine serum (Atlanta Biologicals), gentamicin (10 mg/ml) and fungizone (0.25 µg/ml). Jurkat T cells (TIB-152™) were purchased from ATCC. Jurkat T cells lacking CD47 (JinB8, (49)) or integrin β1 (Jurkat A1, (50)) were the kind gift of Dr. David Roberts (NIH). All Jurkat cell lines were maintained in RPMI 1640 (Invitrogen) supplemented with 5% fetal bovine serum (FBS), penicillin (5 mg/ml) and streptomycin (1 mg/ml). Jurkat T cells were maintained below 2 × 106 cells/ml, unless otherwise noted, and were weaned from 5% FBS to serum-free conditions starting 48 h prior to all experiments. 3T3 L1 fibroblasts were the kind gift Dr. Tsu-Shuen Tsao (University of Arizona) and were maintained in DMEM (Invitrogen) supplemented with 10% FBS, penicillin (5 mg/ml) and streptomycin (1 mg/ml). MCF-7 cells were obtained from ATCC (HTB-22™) and maintained in DMEM supplemented with 10% FBS, penicillin (5 mg/ml) and streptomycin (1 mg/ml), and used for experiments within 10 passages after thawing.</p><!><p>1 × 106 Jurkat T cells resuspended in stain/wash buffer (PBS supplemented with 0.1% BSA, 0.01% NaN3), were used per assay condition. Cells were incubated with stain/wash buffer or stain/wash buffer supplemented with E3CaG1 (22 nM) for 1 h at 4 °C and were fixed with 4% paraformaldehyde prior to incubation with FITC-conjugated monoclonal anti-human CD47 antibody or isotype control. Cells were washed with stain/wash buffer to remove unbound antibody followed by addition of 4% paraformaldehyde. One-color flow cytometric analysis was performed at 488 nm using a FACScan flow cytometer (BD Biosciences). The emission fluorescence of FITC-conjugated CD47 antibody was detected using a 530/30 bandpass filter and recorded at a rate of 200–400 events per second for 10,000 events gated on FSC (forward scatter) vs. SSC (side scatter). Data were analyzed using CellQuest PRO software (BD Biosciences). Appropriate electronic compensation was adjusted by acquiring cell populations stained with each dye/fluorophore individually, as well as an unstained control.</p><p>To examine increased [Ca2+]i by flow cytometry, we loaded cells with 5 µM Fluo-3AM in Krebs buffer for 30 min at room temperature with gentle mixing every 10 min. The green fluorescence emission of calcium binding dye Fluo-3 was then analyzed following 488-nm laser excitation on a BD LSRII flow cytometer (Becton Dickinson, Inc.). Buffer or E3CaG1 (2.2–220 nM) was added to cell suspension (2 × 106 cells/500 µl), and data were collected after ten min. Where indicated, cells were incubated with anti-CD47 antibody (B6H12) for 20 minutes prior to the addition of E3CaG1. Data were analyzed using FlowJo (v.7.6.4).</p><!><p>Baculoviral vector pAcGP67.coco (COCO), encoding E3CaG1 was kindly provided by Dr. Deane Mosher (University of Wisconsin). Expression and purification were carried out as described (51). Briefly, Sf9 cells were grown at 27 °C and were maintained in Grace's Insect Media supplemented with 10% fetal bovine serum and gentamicin and fungizone. When the cells reached a density of 1 × 106 cells/ml, they were transferred to SF900II media (Invitrogen) and were infected with high titer virus at a multiplicity of infection of 5. Media was collected 65 h post infection. After the His-tagged E3CaG1 was purified by immobilized metal ion affinity chromatography, it was stored at −80 °C in TBS supplemented with 2 mM CaCl2. Protein concentrations were determined by the BCA assay (Thermo Scientific, Rockford, IL) using bovine serum albumin as the standard.</p><!><p>Primers 5'-ctcagtctcgagatctattcctgatgc-3' and 5'-cagtcaggatccgatgttctgcacgaagc-3' were used to amplify human sGC α1 cDNA (ATCC clone MGC-33150) for cloning into pCMV-3Tag-9 (Clontech, Mountain View, CA) between BamHI and HindIII sites, yielding a C-terminal myc-tagged protein (vector WM397). Human sGC β1 was cloned into pCMV-3Tag-3A (Clontech) between SacI and XhoI sites, yielding a C-terminal FLAG-tagged protein (vector WM434). Primers 5'-gcactcgaggtcatcatcctgctttg-3' and 5'-cactgtgagctcatgtacggatttgtg-3' were used to amplify the cDNA from plasmid pSTBlue1-Huβ1 bearing the human sGC β1 cDNA – a gift from Dr. Alan Nighorn (University of Arizona). The Stratagene QuikChange® Lightning Site-Directed Mutagenesis Kit (Agilent, La Jolla, CA) was used to correct all errors in both plasmids to match CCDS34085.1 (GUCY1A3) and CCDS47154.1 (GUCY1B3) (52). Transfection reagent TurboFect™ (Fermentas, Glen Burnie, MD) was used at a ratio of 20 µg plasmid DNA (1:1 ratio of sGCα:sGCβ) to 25 µl reagent per 10-cm dish of cells at 50% confluency. Cells were harvested by trypsinization 12 h after transfection and the cell pellet was quickly frozen in liquid nitrogen.</p><!><p>Transiently transfected MCF-7 cells were trypsinized and resuspended in Krebs buffer. To manipulate [Ca2+]i, cells were incubated with ionomycin (1 µg/ml), thapsigargin (400 nM) and 0.1 mM CaCl2 or vehicle control (DMSO) for 15 min. Cell pellets were lysed into homogenization buffer (50 mM Tris-HCl pH 7.5, 100 mM NaCl, 1 mM EDTA, 1 mM TCEP, 1 mM PMSF, protease inhibitor cocktail (10 µl/ml cell lysate)) using a homogenizer. Lysate was spun at 13000 × g and supernatant was combined with anti-FLAG agarose beads (Sigma, St. Louis, MO) for 1 h, 4 °C, on an Adams™ Nutator Mixer (BD Biosciences). After incubation, beads were washed with TBS and evenly divided into 0.6 ml eppendorf tubes for the sGC activity assays. Inclusion of equal quantities of immunoprecipitated sGC in each assay condition was confirmed by Western blot analysis (supplemental Fig. S1). Western blots were analyzed on an Odyssey Imaging System (LI-COR) and Image J software was used to analyze loading quantities. Reactions were carried out in a final volume of 100 µl containing reaction buffer (3 mM GTP, 8 mM MgCl2, 50 mM Hepes, pH 7.7, prepared at 10X concentration just prior to use) and, where indicated, 10 µM YC-1 or BAY 41-2272 or vehicle control, and 10 µM DEA/NO. YC-1 and Bay 41-2272 were dissolved in DMSO and then diluted to a final stock concentration of 1.1 mM in ethanol. DEA/NO was prepared as a 1 mM stock solution in 10 mM NaOH. Upon adding DEA/NO or vehicle control, the assay was allowed to proceed for 5 min at 37 °C. The reactions were stopped by pelleting the beads and transferring the supernatant to Cell Lysis Buffer (Molecular Devices, Sunnyvale, CA, or Cisbio, Bedford, MA). cGMP concentrations were determined by competitive ELISA using the CatchPoint™ cGMP assay (Molecular Devices), following the manufacturer's instructions, or the homogenous time resolved fluorescence (HTRF) assay (Cisbio), following the manufacturer's instructions and using a BioTek H1F plate reader.</p><p>For kinetic measurements, transiently transfected MCF-7 cells were either treated with DMSO (vehicle control) or ionomycin, thapsigargin and 2 mM CaCl2 for 5 min. Cell pellets were lysed as described above and incubated with anti-FLAG agarose beads for 1.5 hours at 4 °C. Following this, the beads were washed three times and resuspended in an appropriate volume of Tris-buffered saline (pH 7.5). Aliquots of this slurry were then used for activity measurements. Reactions were carried out at 37 °C in a final volume of 150 µl and initiated by the addition of reaction buffer (5–2000 µM GTP, 8 mM MgCl2, 50 mM HEPES, pH 7.5, prepared at 10X concentration just prior to use). Where NO-induced sGC activities were measured, DEA/NO (50 µM) was added immediately after the addition of reaction buffer. Reactions were quenched by the addition of cell lysis buffer from the cGMP kit, generally after 10 min (−NO) or 3 min (+NO). Catalytic rates were linear over these time periods for all GTP concentrations used. Inclusion of equal quantities of immunoprecipitated sGC was confirmed by Western blot analysis (supplemental Fig. S1). For each experiment, cGMP accumulation was measured in duplicate using the cGMP-ELISA kit from Molecular Devices or the HTRF kit from Cisbio; higher concentrations of GTP did not interfere with the measurements. Kinetic parameters were obtained by non-linear fitting of the Michaelis-Menten equation, using SigmaPlot (SPSS, Inc., Chicago). Km and Vmax are presented as the average and standard deviation of three independent experiments.</p><!><p>Jurkat T cells (1 × 106 per assay condition) were resuspended in Krebs buffer. Where indicated, cells were preincubated with treatment agents (E3CaG1, BAPTA-AM, etc.) or vehicle controls for the indicated time at room temperature, followed by addition of 10 µM DEA/NO. Reactions were stopped after 2 min by placing the cell suspensions on ice, pelleted and quickly frozen. For cGMP measurements, the cell pellets were thawed and resuspended with 100 µL Cell Lysis Buffer. The basal and NO-induced sGC activities of intact cells were expressed in terms of picomoles cGMP produced per minute per milligram of total protein content (pmol cGMP min−1 mg−1), using the CatchPoint cGMP assay kit. Protein concentrations were determined by the BCA assay (Thermo Scientific) using bovine serum albumin as the standard.</p><p>To examine E3CaG1 inhibition, cells were incubated with 22 nM E3CaG1 in Krebs buffer for 15 min at room temperature, followed by the addition of 10 µM DEA/NO. To manipulate [Ca2+]i, cells were incubated with ionomycin (1 µg/ml) and thapsigargin (400 nM), 20 mM EGTA or vehicle control, and 0–10 mM CaCl2 for 15 min, followed immediately by addition of 10 µM DEA/NO. For experiments examining intracellular calcium chelation, cells were incubated with BAPTA-AM (10 µM, added from a 2 mM stock solution in DMSO) or vehicle control for 15 min prior to the addition of E3CaG1 (16 nM) or buffer for 15 min, and then DEA/NO (10 µM for 2 min). BAPTA-AM is a membrane permeable Ca2+ chelator that is converted to BAPTA in the cytosol, where it becomes trapped.</p><p>To examine the effect of PHA or Ang II on sGC activity, Jurkat T cells were grown in serum free media 12 h prior to the experiment at a density of less than 1 × 106 cells/ml. Where indicated, cells were incubated with 5 µM BAPTA-AM or vehicle control (DMSO) for 15 min, followed by the addition of the indicated concentrations of PHA or 1 µM Ang II for an additional 2 min, and then DEA/NO (10 µM) for 2 min. To examine the effect of compounds YC-1 and Bay 41-2272 on E3CaG1 inhibition of sGC, cells were incubated with 22 nM E3CaG1 for 15 min prior to the addition of 10 µM YC-1, 10 µM BAY41-2272 or vehicle control, followed immediately by addition of DEA/NO.</p><p>To examine the effect of phosphodiesterases on cGMP accumulation in intact cells, MCF-7 cells transiently transfected with sGC were used. 14 h post-transfection, cells were trypsinized and incubated with vehicle/DMSO, IBMX (0.5 mM) or 8-methoxymethyl IBMX (0.4 mM) for 30 min, followed by addition of ionomycin (1 µg/ml), thapsigargin (400 nM) and calcium chloride (0.1 mM) to appropriate samples, followed immediately by the addition of DEA/NO (10 µM). After 2 min, cells were spun down and cell pellets frozen.</p><!><p>MCF-7 cells were transiently transfected with sGC; 14 h post-transfection, cells were trypsinized and pellets were lysed in homogenization buffer. Immunoprecipitation of sGC was performed as described above. Jurkat cell lysate was then incubated with the beads for 15 min at 37 °C with or without 250 nM Ca2+ and/or staurosporine (1 µM). Following this, the beads were washed five times with TBS and resuspended in an appropriate volume for activity assay. Where indicated, 10 µM DEA/NO and 10 µM YC-1 were included in the reactions.</p><!><p>25 × 106 Jurkat T cells were used for each assay condition and were incubated with buffer or E3CaG1. Cell pellets were lysed into 600 µl homogenization buffer (50 mM Tris-HCl pH 7.5, 100 mM NaCl, 1 mM EDTA, 1 mM TCEP, 1 mM PMSF, protease inhibitor cocktail (10 µl/ml cell lysate)) using a homogenizer. Lysate was spun at 13000 × g and supernatant was incubated with or without IBMX (0.5 mM) and 8-methoxymethyl IBMX (0.4 mM) for 10 min. This was followed by the addition of Mg-GTP reaction buffer and DEA/NO (10 µM). Reactions were stopped after 2 min by the addition of 250 µl cell lysis buffer (Molecular Devices, Sunnyvale, CA). cGMP concentrations were determined by competitive ELISA using the CatchPoint™ cGMP assay (Molecular Devices), following the manufacturer's instructions.</p><!><p>In order to assay [Ca2+]i in Jurkat T cells, which normally grow in suspension, 3T3 L1 fibroblasts were used to coat glass coverslips with extracellular matrix. Fibroblasts were hypotonically lysed and cellular debris was mechanically removed with a cell scraper. Jurkat T cells were then allowed to adhere to the matrix-coated coverslips. The cells were left undisturbed for a minimum of one hour before use, and remained attached to the coverslips under these conditions for up to four hours. Attached cells were loaded with Fura-2AM for 30 min at room temperature in the dark. Fura-2 fluorescence was observed on an Olympus (Center Valley, PA) IX70 microscope equipped with a 75 W xenon lamp while alternating between excitation wavelengths of 340 and 380 nm. Images of emitted fluorescence above 505 nm were captured by an ICCD camera (Photon Technology International, Birmingham, NJ) under ImageMaster software control (PTI). Effective [Ca2+]i was calculated from equations published in (53). Initial [Ca2+]i was assessed over 20 – 60 s to establish a consistent baseline, and changes in [Ca2+]i were monitored over time for each experimental condition. Depending on the experiment, measurements were taken every 0.6 sec (for 3 – 5 min experiments) up to 5 s (for experiments > 5 min). Cell morphology within the time period of measurement was assessed by differential interference contrast microscopy, and found not to vary.</p><!><p>Data are presented as mean ± S.D of independent experiments. Differences between groups were compared for significance using Student's t-test (calculated with Microsoft Excel software).</p><!><p>To examine the mechanism behind TSP-1 inhibition of sGC, we used E3CaG1, a C-terminal TSP-1 construct that retains robust activity and is more stable than full-length TSP-1. E3CaG1 consists of the last EGFβ-like type II repeat, all of the calcium binding type III repeats and the C-terminal cell-binding domain required for CD47-dependent activity (Fig. 1A). We chose Jurkat T cells for these experiments since these cells are one of the few immortalized cell lines with intact sGC signaling, and since they also respond to TSP-1 (54). Initial experiments with full-length TSP-1 suggested inconsistent inhibition of sGC activity (data not shown). To uncover the reason for this, we used E3CaG1 to study CD47 binding over time. We measured binding of E3CaG1 to Jurkat T cells through its ability to compete with a FITC-conjugated monoclonal anti-human CD47 antibody, using fluorescence activated cell sorting (FACS). We first examined cells that had been kept at low density (0.5 × 106 cells/ml) or that had been cultured for less than 2 weeks. These cells exhibited very little auto-fluorescence and only a small increase in fluorescence upon treatment with a FITC-conjugated isotype control antibody, indicating little non-specific binding occurs. When cells were incubated with the FITC-conjugated CD47 antibody, there was an ~100-fold increase in fluorescence, indicating the presence of CD47 on the surface of Jurkat T cells. Plots of scattering vs. fluorescence for these data ("dot plots", supplemental Fig. S2) indicated a homogenous population of positively stained cells, consistent with a uniform distribution of CD47 throughout the Jurkat T cell population. When E3CaG1 (22 nM) was added to cells prior to the addition of CD47 antibody, the mean fluorescence decreased significantly, indicating that E3CaG1 competes with the monoclonal antibody and binds to CD47 on the Jurkat T cell surface (Fig. 1B). Older cells (> 6 weeks), or cells that had been grown at higher density (3 × 106 cells/ml), could still bind antibody, but E3CaG1 was no longer able to compete with antibody binding (Fig. 1D).</p><p>Preparations of E3CaG1 had little effect on the basal activity of sGC in younger cells, but profoundly inhibited NO-stimulated sGC activity (Fig. 1C), much as previously reported in other cell types (9). We observed strong inhibition for E3CaG1 concentrations as low as 0.22 nM (42%, supplemental Fig. S3) and found inhibition to be maximal for concentrations above ~20 nM E3CaG1 (~67%), similarly to previous reports for E3CaG1 and full-length TSP1 (9). Subsequent experiments were performed with 22 nM E3CaG1.</p><p>When we examined older Jurkat T cells, no inhibition was seen (Fig. 1E), consistent with the lack of binding as shown in Fig. 1D. We conclude from these experiments that CD47 remains on the cell surface; however, CD47 or a complex that includes CD47 has changed and can no longer interact with the TSP-1 fragment. All subsequent experiments were therefore performed on cells that were within 3 weeks of growth and kept below 1 × 106 cells/ml.</p><!><p>TSP-1 and peptide 4N1K are known to increase [Ca2+]i in fibroblasts and mast cells through a mechanism thought to require direct binding to CD47 (23, 24). We hypothesized that E3CaG1 inhibition of sGC in Jurkat T cells also involved changes in [Ca2+]i and used digital imaging microscopy to examine this possibility (Fig. 2, supplemental movie M1). Jurkat T cells were transferred to matrix-coated coverslips (see Experimental Procedures) for these experiments, and allowed to attach for at least 1 hr, well beyond the time where attachment-associated Ca2+ spikes have previously been described, which persist for ~8 min post attachment (55). At rest, Jurkat T cells displayed an [Ca2+]i of 10 – 25 nM. Addition of E3CaG1 (22 nM final concentration) induced an increase in [Ca2+]i to 150 – 300 nM (Fig. 2A,B). Similar increases were not observed after washing with Hank's basal salt solution (HBSS, pH 7.4) alone (Fig. 2B). Calcium concentrations could be experimentally controlled within Jurkat cells using the Ca2+ chelator BAPTA (Fig. 2C), or the Ca2+ ionophore ionomycin and sarco/endoplasmic reticulum Ca2+ ATPase (SERCA) pump inhibitor thapsigargin (Fig. 2D). None of the treatment conditions altered cell morphology within the time period of measurement as assessed by differential interference contrast microscopy.</p><!><p>We examined the requirement for CD47 using flow cytometry and the fluorescent Ca2+ indicator Fluo-3. Binding of 2.2 nM or 22 nM E3CaG1 to Jurkat T cells in suspension led to an ~100-fold increase in average fluorescence over addition of buffer alone (Fig. 3A). Thus, Jurkat T cells in suspension behaved similarly to those attached to coverslips. Addition of anti-CD47 antibody B6H12 completely blocked E3CaG1-dependent calcium mobilization. Likewise, cell line JinB8, which is a modified Jurkat T cell lacking CD47 (49), is not sensitive to E3CaG1 (Fig. 3B). Similarly, antibody B6H12 abolishes E3CaG1-dependent inhibition of sGC (Fig. 3C). We conclude that E3CaG1 signaling requires CD47, as expected from previous studies (9). In contrast, antibodies to integrin αV, which is known to associate with CD47 (15), have no effect on E3CaG1-dependent increases in [Ca2+]i (supplemental Fig. S4) and subsequent inhibition of sGC (Fig. 3C). E3CaG1 also remains active toward cell line Jurkat A1, which are integrin β1 null (data not shown) (50). Additionally, pertussis toxin, which inhibits Gi protein, had no effect on E3CaG1-dependent increases in [Ca2+]i (supplemental Fig. S4), or on E3CaG1-dependent inhibition of cGMP production (data not shown).</p><!><p>Based on previous reports showing that calcium can inhibit sGC activity in HEK 293 cells and also with purified protein (41, 42, 46), we examined whether this was also the case for Jurkat T cells. Jurkat cells were resuspended in Krebs buffer containing 1 µg/ml ionomycin and 400 nM thapsigargin and varying concentrations of extracellular calcium ([Ca2+]e). Under these conditions, [Ca2+]i is effectively set by [Ca2+]e. NO-inducible cGMP accumulation was inversely proportional to [Ca2+]e and complete inhibition occurred at [Ca2+]e = 4 mM (Fig. 4A). Chelating of extracellular calcium with EGTA abolished inhibition. Approximately 99% of cells were viable under each experimental condition, as indicated by trypan blue dye exclusion.</p><p>Chelating intracellular Ca2+ with compound BAPTA also overcame inhibition of sGC by E3CaG1, indicating that E3CaG1 inhibits sGC through a mechanism requiring increased [Ca2+]i. In the absence of BAPTA, E3CaG1 reduced NO-stimulated cGMP production by 50% (Fig. 4B). However, after pre-loading the cells with BAPTA, E3CaG1 had no effect on cGMP production.</p><!><p>Angiotensin II (Ang II) is a hormone that induces vasoconstriction through binding to GPCR AT1 and inducing a sustained increase in [Ca2+]i in targeted cells (47, 48). Phytohemoagglutinin (PHA) is a natural agonist of the T-cell receptor that transiently increases [Ca2+]i (56). Since Jurkat T cells have Ang II and T-cell receptors (56, 57), we asked whether Ang II and PHA would inhibit cGMP production by sGC. Addition of PHA inhibited NO-stimulated sGC activity in a dose-dependent manner to 60% at the highest concentration examined (50 µg/ml, Fig. 5A). Addition of 1 µM Ang II to cells increased [Ca2+]i to a similar level as did E3CaG1 (supplementary Fig. S4) and inhibited NO-stimulated sGC activity by 40% (Fig. 5B). As with E3CaG1, chelating intracellular Ca2+ with BAPTA reversed this inhibition.</p><!><p>Previous studies have indicated that TSP1-dependent inhibition of cGMP was through inhibition of sGC and not through stimulation of phosphodiesterases (PDEs) (21, 22). To confirm that this was also true under the conditions of our experiments, we examined cGMP accumulation when PDE was inhibited. We first sought to directly inhibit PDE proteins in live Jurkat T cells using 3-isobutyl-1-methylxanthine (IBMX), a general PDE inhibitor, or 8-methoxymethyl IBMX, a specific inhibitor of calcium/calmodulin-dependent PDE1. Unfortunately, these compounds activate T cells, possibly through a cAMP-dependent mechanism (58), which interferes with the measurement of E3CaG1 activity. We therefore measured NO-dependent cGMP accumulation in Jurkat T cell lysate obtained from cells that were previously treated with E3CaG1 or buffer control; measurements were made in the presence of IBMX or 8-methoxymethyl IBMX (Fig. 6A). Under these conditions, calcium is diluted and PDEs inhibited. Inhibition of NO-stimulated sGC activity was pronounced in the cell-free lysate, consistent with only minor PDE affect on cGMP accumulation.</p><p>We further examined the role of PDEs using transiently expressed human sGC in MCF-7 cells, which do not normally express sGC. Raising [Ca2+]i in these cells led to pronounced inhibition of NO-stimulated sGC activity (Fig. 6B). Addition of IBMX or 8-methoxy IBMX led to small increases in basal, NO-stimulated and calcium-inhibited cGMP levels; however, the 60% reduction in NO-stimulated cGMP accumulation due to increased [Ca2+]i was unchanged in the presence of these compounds, indicating PDEs have at most a minor role in the observed loss of cGMP under the conditions of our experiments.</p><!><p>Compounds YC-1 and Bay 41-2272 are small molecule activators of sGC that act synergistically with NO and CO (59); the related compound BAY 63-2521 (Riociguat) is in clinical trial for pulmonary hypertension (60). When included at 10 µM, E3CaG1 inhibition of NO-stimulated sGC was completely overcome by either YC-1 or Bay 41-2272 (Figs. 7A and 7B), suggesting that Ca2+ inhibition of sGC is through an allosteric mechanism that can be overcome by allosteric stimulators. YC-1 also inhibits phosphodiesterases (61, 62), which may slightly contribute to the cGMP accumulation shown in Fig. 7A. Although Bay 41-2272 can also inhibit phosphodiesterases (63), inhibition is less pronounced (64, 65) and unlikely to be a factor in Fig. 7B.</p><!><p>That sGC in diluted cell extracts remains inhibited (Fig. 6A) suggests inhibition may be through covalent modification. To further examine this possibility, we transiently expressed human sGC in MCF-7 cells and isolated the protein through immunoprecipitation using a FLAG purification tag. The immunoprecipitated protein displays strong NO-stimulated activity (Fig. 7C). However, when cells were first treated with Ca2+/ionomycin, the isolated protein was substantially inhibited. As with cellular sGC, addition of YC-1 or Bay 41-2272 reversed the inhibition. Thus, higher [Ca2+]i leads to a modified sGC with reduced activity, and this activity can be overcome by allosteric stimulators.</p><p>Two reports indicate that Ca2+ can directly inhibit sGC isolated from bovine lung, with Ki values varying between 0.15 and 98.5 µM depending on conditions and laboratory (42, 46). We examined direct inhibition of immunoprecipitated human sGC and found that addition of 100 µM Ca2+ led to 50 ± 2% inhibition of the NO-stimulated protein (supplemental Fig. S5) while 1 mM Ca2+ brought the sGC activity completely back to basal levels; nanomolar levels of Ca2+, as found in vivo, displayed little inhibition in our hands (data not shown). Neither YC-1 nor Bay 41-2272 was able to overcome direct inhibition by 100 µM Ca2+ (supplemental Fig. S5), in contrast to the inhibition of sGC by E3CaG1 in whole cells. The fact that a high concentration of Ca2+ is needed to achieve substantial direct inhibition in vitro (micromolar versus nanomolar in the cell), and that YC-1 or Bay 41-2272 do not overcome this inhibition, indicate that direct binding of Ca2+ to sGC does not contribute to the observed intracellular sGC inhibition.</p><!><p>To further characterize the inhibited sGC, we measured steady-state kinetic parameters for the protein after immunoprecipitation (Fig. 8A, Table 1). The uninhibited protein displayed typical values for Km and Vmax (42, 46, 66, 67) and the expected decrease in Km and increase in Vmax upon stimulation by NO. In contrast, Km values for sGC isolated from calcium-treated cells were dramatically increased and unaltered by NO stimulation (Table 1). The inhibited value (Km ~ 870 µM) is about twice the value for cellular GTP concentrations, which are estimated to be ~470 µM in mammalian tissues (68). At this GTP concentration, the inhibited protein would operate in the cell at well below Vmax, while the uninhibited protein, with Km = 67 µM when bound to NO, would be nearly saturated with GTP and operating at near maximal velocity.</p><!><p>We developed a cell-free assay for evaluating calcium-dependent inhibition of sGC. Jurkat T cell lysate led to inhibited sGC upon addition of 250 nM Ca2+, but had no effect in its absence (Fig. 8B). Boiling of lysate prior to addition (data not shown), or addition of pan-kinase inhibitor staurosporine (Fig. 8B), prevented inhibition, indicating a calcium-dependent protein kinase was required for inhibition. Obvious candidates for this role are the multifunctional Ca2+/calmodulin-dependent protein kinases I, II and IV. However, addition of compounds KN-62 or KN-93, which inhibit these proteins, had no effect on E3CaG1 inhibition of sGC (data not shown).</p><!><p>The balance between vasoconstriction and vasodilation in mammals, as well as wound healing, angiogenesis and other related activities, relies on the give-and-take of numerous signaling pathways. Here, we reveal a new level of cross regulation: TSP-1 and Ang II, two factors of central importance for cell proliferation and vasoconstriction, can inhibit sGC activity by increasing [Ca2+]i. The present experiments were performed in Jurkat T cells, a convenient cell line for these studies since the cells perform well in tissue culture, retain endogenous sGC and also respond to TSP-1 and Ang II. This combination is rare in immortalized cells, which in general do not express sGC. Binding of the TSP-1-derived fragment E3CaG1 to CD47 in Jurkat T cells causes free [Ca2+]i to increase from resting levels of 5–10 nM to peak levels of 300 nM, leading to strong inhibition of sGC (Figs. 1–3). Blocking this increase with chelator BAPTA reverses the sGC inhibition (Fig. 4B). Inducing an increase in [Ca2+]i with Ang II or PHA (Fig. 5), or with the calcium ionophore ionomycin and SERCA inhibitor thapsigargin (Fig. 4A), also leads to sGC inhibition. These data make clear that Ca2+ regulates sGC activity in Jurkat T cells.</p><p>The link between Ang II and sGC is particularly interesting to discover. Ang II is part of the renin-angiotensin-aldosterone system for controlling blood pressure through the sensing of blood volume and the linking of kidney function to blood flow (69). The Ang II receptors are G-protein coupled receptors, the most common of which is angiotensin receptor type 1 (AT1). Binding leads to an increase in [Ca2+]i through production of inositol triphosphate (IP3) and subsequent binding to the IP3-sensitive calcium channel of the sarcoplasmic/endoplasmic reticulum. In vascular smooth muscle, Ca2+ stimulates myosin light chain kinase, which phosphorylates myosin, leading to vasoconstriction. Angiotensin converting enzyme (ACE) inhibitors, and AT1 inhibitors, are commonly used to block this pathway and vasoconstriction in the treatment of hypertension (70). NO-stimulated sGC produces cGMP, which lowers [Ca2+]i through multiple mechanisms, but in particular via phosphorylation of regulatory protein phospholamban by cGMP-dependent protein kinase G (PKG), which leads to stimulation of SERCA and the pumping of Ca2+ from the cytosol into cellular stores (71). Interestingly, TSP-1 can also inhibit PKG, further attenuating NO signaling (72). Additionally, fluctuations in [Ca2+]i may affect the availability of nitric oxide: NADPH oxidase 5 (NOX5), which is found in both vascular smooth muscle and endothelial cells, is stimulated by Ca2+ to produce superoxide, a free radical molecule that reacts at the diffusion limit with NO to yield peroxynitrite (73, 74). Thus, our data suggest a feedback mechanism that serves to balance vasodilation through NO and vasoconstriction through Ang II by directly raising and lowering [Ca2+]i levels (Fig. 9). That this may be the case in vivo is supported by a recent study on Ang II-induced hypertension in rats, which demonstrated that Ang II treatment led to blunted sGC activity (75).</p><p>A major finding in the present study is that increased [Ca2+]i leads to inhibition of sGC through covalent modification, most likely by phosphorylation. sGC inhibited in Jurkat T or MCF-7 cells (Figs. 6, 7), or in Jurkat lysate supplemented with 250 nM Ca2+ (Fig. 8), remains inhibited after the excess calcium is diluted or washed away, displaying a 13-fold increase in KmGTP in the presence of NO (Fig. 8, Table 1). Inhibition of kinases with staurosporine relieves the calcium-dependent inhibition of sGC (Fig. 8), suggesting inhibition is through direct phosphorylation of sGC. Interestingly, TSP-1-dependent inhibition of PKG also appears to be through covalent modification since inhibition is retained in cell-free extracts (72), indicating a common mechanism may be at work. Although Ca2+ can also stimulate Ca2+/calmodulin-dependent phosphodiesterases, particularly in neuronal cell extracts (43), phosphodiesterase activity is at most a minor contributor to the inhibition observed in the present experiments.</p><p>sGC allosteric activators YC-1 and Bay 41-2272, which are synergistic with CO and NO for stimulating sGC activity, completely restore NO-stimulated sGC activity. The compounds overcome E3CaG1 inhibition of sGC in Jurkat T cells (Figs. 7A and 7B), and overcome calcium-induced inhibition of sGC in MCF-7 cells (Fig. 7C) or in cell lysate (Fig. 8B). These results suggest the sGC modification leads to a protein that is stabilized in a low activity conformation, but still fully capable of catalysis. Allosteric stimulation by NO alone is insufficient to drive the modified protein into a fully active state, but in combination with YC-1 or Bay 41-2272, full activity is achieved. Similarly, recent data from Miller and co-workers indicate that stimulation of sGC by YC-1 and Bay 41-2272 in TSP-1 treated platelets and vascular smooth muscle cells is reduced, as was stimulation by NO alone (76); however, stimulation with both YC-1 and NO was not examined in that study. Taken together, these data indicate that both allosteric activator and NO are required to completely overcome TSP-1-dependent inhibition. The data also suggest that the YC-1 class of compounds may offer broad relief to hypertensive individuals, even where Ang II and TSP-1 levels are high, as occurs, for example, in older individuals or those suffering from type-II diabetes (70, 77).</p><p>The mechanism by which TSP-1/E3CaG1 causes increases in [Ca2+]i remains unknown. CD47 is clearly required for signal transduction (Fig. 3). However, E3CaG1 binding is lost as Jurkat T cells age despite the continued presence of CD47 on the cell surface (Fig. 1), suggesting that CD47 alone may not be sufficient for binding and signaling. CD47 was originally identified as integrin associated protein and is likely to function through a signaling complex. Interestingly, CD47 is required for integrin-mediated Ca2+ influx in endothelial cells (78), which may be related to the Ca2+ signal described herein. Several integrin complexes are known to induce Ca2+ influx (79–81), although others reduce [Ca2+]i (81). CD47 may also associate with Gi protein and thereby function as a non-canonical GPCR (15). In such a mechanism, binding of TSP-1/E3CaG1 would induce Ca2+ mobilization through IP3, much as happens with Ang II binding to AT1. However, pertussis toxin had no effect on Ca2+ mobilization (Fig. S4) or inhibition of sGC in Jurkat T cells, suggesting Gi is not involved, and initial experiments designed to interfere with specific integrins did not alter E3CaG1-dependent Ca2+ mobilization (Figs. 3C and S4). Furthermore, TSP-1 transiently decreases IP3 in A2058 melanoma cells (82).</p><p>Finally, it should be noted that autocrine NO signaling, which occurs in endothelial and neuronal cells, is complicated with respect to Ca2+. Increased [Ca2+]i stimulates eNOS and nNOS, leading to NO production, yet also inhibits sGC. Recent experiments by Isenberg and co-workers using endothelial cells demonstrated that TSP-1 binding through CD47 led to a decrease in the ability of ionomycin to increase [Ca2+]i and a subsequent decrease in NO production by eNOS (83).</p><p>In summary, we have shown that sGC is inhibited by a cellular increase in calcium, which can be induced by extracellular TSP-1 fragment E3CaG1 binding to transmembrane protein CD47 and associated proteins, or by Ang II binding to AT1. This inhibition of NO-stimulated sGC involves a post-translational modification and can be overcome through the binding of allosteric compounds YC-1 and Bay 41-2272.</p>

PubMed Author Manuscript

Activating mechanosensitive channels embedded in droplet interface bilayers using membrane asymmetry

Droplet microcompartments linked by lipid bilayers show great promise in the construction of synthetic minimal tissues. Central to controlling the flow of information in these systems are membrane proteins, which can gate in response to specific stimuli in order to control the molecular flux between membrane separated compartments. This has been demonstrated with droplet interface bilayers (DIBs) using several different membrane proteins combined with electrical, mechanical, and/or chemical activators. Here we report the activation of the bacterial mechanosensitive channel of large conductance (MscL) in a dioleoylphosphatidylcholine:dioleoylphosphatidylglycerol DIB by controlling membrane asymmetry. We show using electrical measurements that the incorporation of lysophosphatidylcholine (LPC) into one of the bilayer leaflets triggers MscL gating in a concentration-dependent manner, with partial and full activation observed at 10 and 15 mol% LPC respectively. Our findings could inspire the design of new minimal tissues where flux pathways are dynamically defined by lipid composition.

activating_mechanosensitive_channels_embedded_in_droplet_interface_bilayers_using_membrane_asymmetry

Introduction<!>Activity of MscL G22C F93W in lipid vesicles<!>LPC can activate MscL channels reconstituted into DIBs via membrane asymmetry<!>Discussion<!>Conclusions<!>Conflicts of interest

<p>The eld of bottom-up synthetic biology aims to reconstitute the form, function and behaviour of biological organisms from self-assembled chemical systems. [1][2][3][4][5] To this end, different pathways have been explored to create compartmentalised biomimetic microstructures capable of supporting functions such as chemical synthesis, [6][7][8] environment sensing, 9,10 information transduction 11 and motility. 12 One route involves the use of lipid monolayer-stabilised water-in-oil (w/o) droplets, where contact between two droplets leads to the spontaneous selfassembly of a lipid bilayer at the interface (Fig. 1A). These structures, which are intended to mimic natural lipid membranes found in biology, are known as droplet interface bilayers 13,14 (DIBs). DIBs offer several advantages over conventional planar bilayer systems (such as black lipid membranes (BLMs) or aperture suspended bilayers 15 ), including increased stability, 14 compartmentalisation of droplet content and the ability to support droplet volumes spanning three orders of magnitude from ml to pl. 16,17 Unlike other approaches, the DIB platform also offers the option to assemble bilayer networks by connecting additional lipid-monolayer coated droplets in series, enabling the construction of minimal articial tissues. [18][19][20] By supplying different lipids to each droplet compartment, 21 DIBs also offer a route for controlling membrane asymmetry, a feature that is ubiquitous in native biological membranes 22 and is essential in facilitating core biological functions such as apoptosis and phagocytosis. 23 This range of features combined with the ability to take quantitative measurements using either uorescence microscopy or by electrophysiology (supported by hydrogel-coated silver/silver chloride electrodes inserted into each droplet) have led to the use of DIBs in a number of different studies concerning the properties of lipid membranes, [24][25][26] the permeability of drug candidates/agrochemical compounds 27,28 and the incorporation of membrane proteins. 29,30 By using the DIB as an environment for the reconstitution of membrane proteins, the interplay between the protein and lipid bilayer can be interrogated ex vivo independently of the complex interaction networks found in cell biology. To-date multiple classes of membrane proteins have been reconstituted, including bacterial outer membrane proteins, 21 ion channels 31 and pore-forming peptide oligomers. 32,33 In these systems, the inserted protein is typically activated by stimuli such as transmembrane potential, 34,35 tension 36,37 and/or pH, 38,39 offering the user full control over the molecular ux across the membranea quality that can also be nely tuned through the use of mutants. Membrane asymmetry has also been shown to affect protein function, with an early study by Hwang et al. demonstrating the effect of charge asymmetry on the spontaneous gating probability of OmpG. 32 The bacterial mechanosensitive channel of large conductance (MscL) is a membrane protein of interest due to its relatively large ($3 nm, $3.5 nS) unselective pore size in the openstate. G22C F93W mutants enable the chemical activation 40 and spectroscopic detection of MscL, whereas the V23T 41 and G22S 42 gain-of-function (GOF) mutants offer lower tension thresholds for activation ($6 mN m À1 ). Although both GOF mutants have been reconstituted and activated in DIBs [41][42][43][44][45] with greater ease than their wild-type counterpart, they still require individual droplets to be prepared or manipulated by the user, meaning that ux pathways cannot be dened in real-time. This becomes a problem when constructing droplet networks, especially given that network architecture and composition can be used to dene the ow of molecular information throughout the tissue. 46 To this end, recent work has focused on using external stimuli such as light 19 or temperature 47 to dynamically control network activation, however this has only been achieved for the water-soluble alpha-toxin alpha haemolysin and not for waterinsoluble membrane proteins, highlighting the need to develop methods that offer new, orthogonal ways to dene information ow across a bilayer network.</p><p>Here we show for the rst time in a droplet system that the activation of MscL can be achieved using bilayer asymmetry, i.e. in the absence of any channel activators or applied external pressure. We achieve this by incorporating 1-oleoyl-2-hydroxysn-glycero-3-phosphocholine (LysoPC/LPC) into one of the bilayer leaets (Fig. 1B and C) to generate an asymmetric change in the lateral pressure prole that has been shown previously to activate the MscL channel, [48][49][50][51][52][53] and identify protein activity using single-channel electrical measurements (Fig. S1 †). Our method to control the full gating of the G22C F93W loss-offunction (LOF) channel in a DIB system using membrane patterning could be applied to other mutants and serve as a new strategy to control molecular ux in droplet networks, helping to design and build minimal tissues capable of increasingly complex information processing.</p><!><p>MscL G22C F93W was expressed in BL21(DE3) E. coli and puried via cobalt-immobilized affinity chromatography as in previous work 10,50 (Fig. S2 †). The activity of the expressed MscL channel was tested through reconstitution into 1 : 1 (mol : mol) 1,2-dioleoyl-sn-glycero-3-phosphocholine (DOPC) : 1,2-dioleoylsn-glycero-3-phospho-(1 0 -rac-glycerol) (DOPG) lipid vesicles containing a self-quenching (50 mM) concentration of entrapped calcein. 54 MscL-vesicles were produced via thin-lm hydration, extrusion and detergent mediated reconstitution, before separating unencapsulated calcein from the produced vesicles via size-exclusion chromatography to form vesicles $100 nm in diameter (Fig. S3 †).</p><p>MscL channel activity was then assayed via three different activation methods in parallel: (i) chemical modication of vesicle structure through the addition of LPC (Fig. 2A) (ii) chemical modication of vesicle structure through the enzymatic activity of secretory phospholipase A2 (sPLA2) 55 and (iii) chemical modication of the MscL channel through addition of the channel activator [2-(trimethylammonium)ethyl]methanethiosulfonate (MTSET). An LPC concentration gradient was added to vesicles AE MscL, and the calcein ux response was monitored over 5 hours (Fig. 2B). Vesicles containing the MscL channel displayed calcein ux that increased as a function of increasing LPC concentration, indicating that the lysolipid could successfully activate the channel. When the same LPC concentrations were added to vesicles lacking the MscL channel, negligible calcein ux occurred over the 300 minute experimental lifetime (Fig. 2C and S4 †), indicating that the channel is essential for triggered release in response to LPC, and that the expressed MscL protein is active.</p><p>Based on these ux measurements, the threshold concentration of LPC necessary for channel activation lies between 1- 5 mol% LPC. This value agrees well with previous investigations of the effect of LPC on MscL reconstituted into vesicles, 49,51,53 where less than 10 mol% LPC was necessary for channel activation. 53 To further conrm channel activity, enzymatic and chemical activation of the reconstituted channel was undertaken using sPLA2 and MTSET respectively (Fig. S5 †). Again, the presence of the MscL channel enabled calcein ux responses to enzymatic and chemical (channel labelling) stimuli, whilst negligible release was observed for vesicles lacking the channel, indicating high activity of the channel.</p><!><p>Aer conrming the activity of expressed MscL, the effect of LPC on MscL-functionalized DIBs was studied. MscL was reconstituted into vesicles with lower charge better suited for DIB formation (95 : 5 DOPC : DOPG vesicle composition), and used as the droplet-stabilising lipid source (lipid-in) for w/o droplets (1 ml). Droplets were manipulated manually on the tips of agar-coated silver/silver chloride electrodes to assemble DIBs and the membrane was probed using droplet electrophysiology (Fig. 1A and S1 †) exactly as described in our previous work, applying a À100 mV potential across the bilayer. 56 To generate asymmetric DIBs, the rst droplet was formed containing PC : PG : LPC vesicles with increasing LPC concentrations (5, 10 and 15 mol%), whilst a second droplet was formed containing PC:PG vesicles AE MscL (Fig. 3A, B and S9 † for 15, 10 and 5 mol% LPC traces respectively). Asymmetric DIBs incorporating MscL showed LPC-dependent gating behaviour, with 10-fold greater gating events occurring in MscL DIBs containing 15 mol% vs. 10 mol% respectively. We observed negligible gating events in DIBs containing 5 mol% LPC (#events ¼ 42/6/1 for 15/10/5% LPC respectively). These events could be clustered into seven gating states, where each state is dened as a function of the percentage of full channel conductance (S%). Six of these states align with previously proposed conductance states for the channel 57,58 (Table S1 and Fig. S6 †). A histogram analysis approach was used to characterise the step changes of all traces (see Note S1 for further information †).</p><p>The majority of events in our 15% LPC results occurred at S4.5 $14 pA (0.14 nS, n ¼ 23) and S6.6 $21 pA (0.21 nS, n ¼ 4). We additionally observed events occurring at S13 $42 pA (0.42 nS, n ¼ 3), which we assign to the sub-conducting state of MscL observed in our previous work. 56 At times, we saw combinations of such states leading to the observation of events $55 pA (Fig. 3A(iv)), which could represent the gating of two separate MscL proteins or a change in the conformational state of a single channel. Statistical testing between these three states observed in 15% LPC DIBs conrms that the events can be clustered into three populations in this manner (p < 0.001), as well as conrming that the S4.5 sub-conductance state was observed in both 10 and 15% LPC DIBs respectively (p < 0.001). As only the lowest two sub-conductance states are observed at 10% LPC, we conclude that 10% LPC is able to only partially gate the channel.</p><p>As mentioned above, the observed gating events $42 pA correlate well with our previous work 56 which is established as S12, a subconductance state $12% open conductance. 58 Similarly, the low sub-conductance state of S6.6 has been previously observed in MscL gating studies. 57 We note that subconductance states were maintained for extended lengths in our recorded 15% LPC traces (Fig. S7B and C †), indicating that the combination of MscL and asymmetric LPC may be useful for sustained ux across a DIB. Interestingly, the most frequent subconductance state observed here (S4.5) has not been observed in previous MscL electrophysiology experiments. Its appearance here is likely due to the combination of using a lossof-function channel mutant under high potential (À100 mV), which has been shown to increase sampling of low conductance states. 57 The asymmetric incorporation of 15% LPC in MscL DIBs also led to the occupation of higher gating states (Fig. 4). The states shown in Fig. 4A-C are attributed to the higher subconductance states of the channel, showing good agreement with previously established conductances for these states 58 (Table S1 and Fig. S6 †). The open state of the channel (S100 $3.2 nS, n ¼ 3) was observed to occur through these larger subconductance states (Fig. 4D, S13 and S14 †), indicative that asymmetry can drive full opening of the MscL channel. We note that LPC ip-op has previously been shown to be negligible on similar timescales to those within our system. 59,60 Here, LPC ip-op appears insignicant as MscL gating occurs throughout the $30 min experimental timescale (Fig. 3A).</p><p>If DIBs were produced containing MscL but without LPC, negligible gating events occurred, conrming that the protein is in the closed state in the absence of LPC (Fig. 3C). This agrees with our previous work, where gating of reconstituted MscL in symmetric DIBs was not observed. 56 Similarly, if DIBs were produced without MscL but LPC asymmetry was maintained (Fig. 3D), no gating events were detected, indicating that gating could not be attributed to transient pore formation in the DIB caused by the presence of non-bilayer forming LPC. 61 To probe the role of LPC asymmetry further, control experiments with symmetric DIBs containing 15 mol% LPC were performed. This composition attenuated bilayer formation resulting in poor DIB success rates (n ¼ 1/6). When formed, the recordings indicated bilayer instability such that protein activity could not be delineated from the trace (Fig. S12 †). All unltered traces for our complete data set can be observed in Fig. S7-S12 † in the ESI. †</p><p>The increased likelihood for occupying sub-conductance states was not observed in previous work where patch-clamp electrophysiology was used to analyse the effect of LPC on MscL. 51 We attribute this to differences in experimental setup between excised bilayer patches in patch-clamp and the DIB membrane utilised in droplet electrophysiology, the high applied potential of À100 mV leading to increased occupancy of subconductance states 57 and the LOF mutant used in the study (WT vs. G22C F93W here). Whilst further optimisation is necessary to generate more digital activation behaviour, our electrophysiology experiments indicate that LPC can be successfully employed to gate MscL reconstituted into asymmetric DIB membranes.</p><!><p>We have presented the rst evidence that bilayer asymmetry alone can be used to gate MscL channels reconstituted into DIBs. Two aspects of our electrophysiology results indicate that LPC-gating of MscL could represent a powerful tool in controlling ux through DIB networks. Firstly, we see extended gating of the MscL channel in LPC bilayers on the minute timescale (Fig. 3A, S6B and C, S13 and S14 †) which is ideal for sustained ux across the membrane. This is activated without externally applied tension, and hence could be used without the presence of any external mechanical actuation. The lower subconductance states observed here correlate with our previous electrophysiology work on the same mutant (triggered via MTSET-labelling of the channel) 56 and this has been shown to be sufficient for the ux of calcein (MW ¼ 622.6 Da) across MscL DIBs 62 and gating of solutes $6.6 kDa through MscL reconstituted into lipid vesicles. 63 We can therefore infer that LPCgated MscL DIBs should enable molecular ux with a MWCO between 0.6 and 6.6 kDa (and potentially higher than this based on our higher conductance gating events which generate the $3 nm diameter of the open MscL pore 64 ).</p><p>Secondly, the ux we have observed is actuated purely by asymmetry in the lipid bilayer and could be patterned during network assembly, or tuned dynamically using optical tweezers as demonstrated recently using vesicles. 65,66 Furthermore, LPC ip-op could be used to build pre-dened ux patterns into the network, as ip-op to a symmetric bilayer should close the MscL channel. 48 The slow kinetics of this process 67 could be potentially altered if combined with LPC-chelators such as BSA that have been shown to deactivate MscL channels upon removal of LPC from the membrane. 53 The activation mechanism of MscL in response to asymmetric membrane incorporation of LPC has been previously indicated to occur via asymmetric changes in the lateral pressure prole of the membrane, and not via-stretch-induced tension in the membrane. 48,51 In our DIB setup, the global curvature of the membrane is insignicant from the perspective of a single channel, whilst all our experiments are conducted without applied pressure or mechanical stimulation. We therefore attribute the gating observed here to an asymmetric curvature stress induced by LPC in the DIB perturbing the lateral pressure prole at the water-lipid interface. Such a change in membrane lateral pressure reduces the gating energy for the channel, enabling spontaneous gating without applied pressure. 68 MD simulations have indicated that the asymmetric presence of LPC can generate areas of high local curvature in the membrane that contribute to channel activation. 69 This would be energetically less likely to occur in DIBs compared to vesicles, as DIBs possess $30-60-fold higher surface tension ($1-2 mN m À1 (ref. 26 ) vs. $30 mN m À1 (ref. 70 )). Such an effect may decrease the probability of channel activation in DIBs compared to vesicles assuming local curvatures are truly necessary. Comparative electrophysiology experiments on the same MscL channel in both asymmetric DIBs and patch-clamp setups may help further elucidate the mechanism of LPC-activation of MscL.</p><p>To gain further insight to our system the monolayer surface tension of aqueous droplets stabilised with PC:PG:LPC vesicles containing 0-15 mol% LPC was quantied using droplet shape analysis (Fig. S15 †). A linear decrease in surface tension was observed from 0-15 mol%, with the surface tension difference between a droplet containing 0 and 15% LPC found to decrease by 0.74 mN m À1 , from 1.83 AE 0.23 mN m À1 for 0 mol% LPC to 1.09 AE 0.71 mN m À1 for 15 mol% LPC. This tension decrease is expected considering that LPC acts as a surfactant, and measured tensions are similar to those of previously studied asymmetric DIBs (1-2 mN m À1 (ref. 26 )). Indeed, a similar decrease in tension in response to LPC has been shown previously for phosphatidic acid-stabilised droplets. 71 These results appear to match well with the predicted effect of asymmetric LPC generating a pressure differential across the leaets of a DOPC membrane. 69 This further indicates that the primary driver of channel activation observed here is leaet asymmetry and not simply a high bilayer surface tension, as DIB surface tensions are an order of magnitude lower than the activation tension of the channel (>12-14 mN m À1 (ref. 51 )).</p><p>We note that the threshold activation concentration of LPC required for MscL gating differs between the vesicle and DIB model systems: gating occurs from 5 mol% LPC added to MscL vesicles, whilst 10 mol% LPC is necessary when the channel is reconstituted into DIBs. We hypothesise that the increased activation threshold may be due to LPC partitioning between the DIB and droplet monolayers or into the bulk hexadecane. 72 These mechanisms would reduce the effective LPC concentration in the bilayer to minimise the destabilising effect of the lyso-lipid 73 and hence minimise the free energy of the system.</p><p>Although we see full gating of the MscL channel in our work, the most frequently observed events appeared to be gating via sub-conductance states. This could be due to LPC partitioning, higher applied voltage or may reect the high activation energy of the mutant used here. 51 The sidedness of MscL insertion may also play a role: MscL is reconstituted without preference into vesicles (and hence into the DIB), and the presence of both orientations in the same membrane may affect which channels are gated by LPC asymmetry as well as channel gating probability. Interestingly, the amphipath 2,2,2-triuoroethanol was recently shown to activate MscL when added to either leaet of a patch-clamp setup, 74 indicating that LPC may also be able to activate both channel orientations in the DIB. Further investigation could be conducted by using protocols which reconstitute MscL in a single orientation into the DIB membrane as well as testing the LPC activation mechanism with wild-type or GOF channel mutants with lower gating energies. 51 Indeed, MscL V23T has shown a higher probability for the open state when reconstituted in DIBs in response to mechanical actuation, 44 and channel activation should be feasible by using LPC to create asymmetric bilayers.</p><p>Recently, MscL V23T was shown to possess voltagedependent gating in asymmetric DPhPC:DOPhPC DIBs under mechanical stimulation when negatively hyperpolarized. 45 This was achievable for the V23T mutant due to a dominant dielectric effect compared to the WT channel on account of increased pore solvation. The G22C F93W channel possesses increased pore hydrophobicity (and hence decreased pore solvation 75 ) compared to the WT, reducing the dielectric component of the channel in the presence of an electric eld compared to both V23T and WT. Charge asymmetry is therefore signicantly unlikely to gate either the WT or G22C MscL channels without applied tension, but such asymmetries may affect the onset of MscL gating to an LPC asymmetry. This may be useful as a tool to further differentiate channel function in droplet networks.</p><!><p>In conclusion, we have shown that LPC-induced asymmetry in DIB membranes can be utilised as a tool for the activation of incorporated mechanosensitive channels in a sustained manner. This work furthers recent application of the MscL protein as a model mechanosensitive channel 'part' for use in bottom-up synthetic biology, 10,76,77 extending its utility in droplet-interface bilayer networks. The combination of responsive protein channels and bilayer asymmetry shown here points a way towards using lipid composition to dene network functiona still underexplored parameter in the design of droplet systems and minimal tissues.</p><!><p>There are no conicts to declare.</p>

Royal Society of Chemistry (RSC)

Preferences of Specialist and Generalist Mammalian Herbivores for Mixtures Versus Individual Plant Secondary Metabolites

Herbivores that forage on chemically defended plants consume complex mixtures of plant secondary metabolites (PSMs). However, the mechanisms by which herbivores tolerate mixtures of PSMs are relatively poorly understood. As such, it remains difficult to predict how PSMs, singly or as complex mixtures, influence diet selection by herbivores. Although relative rates of detoxification of PSMs have been used to explain tolerance of PSMs by dietary specialist herbivores, few studies have used the rate of detoxification of individual PSMs to understand dietary preferences of individual herbivores for individual versus mixtures of PSMs. We coupled in vivo experiments using captive feeding trials with in vitro experiments using enzymatic detoxification assays to evaluate the dietary preferences and detoxification capacities of pygmy rabbits (Brachylagus idahoensis), dietary specialists on sagebrush (Artemisia spp.), and mountain cottontails (Sylvilagus nuttallii), dietary generalists. We compared preference for five single PSMs in sagebrush compared to a mixture containing those same five PSMs. We hypothesized that relative preference for individual PSMs would coincide with faster detoxification capacity for those PSMs by specialists and generalists. Pygmy rabbits generally showed little preference among individual PSMs compared to mixed PSMs, whereas mountain cottontails exhibited stronger preferences. Pygmy rabbits had faster detoxification capacities for all PSMs and consumed higher concentrations of individual PSMs versus a mixture than cottontails. However, detoxification capacity for an individual PSM did not generally coincide with preferences or avoidance of individual PSMs by either species. Cottontails avoided, but pygmy rabbits preferred, camphor, the PSM with the slowest detoxification rate by both species. Both species avoided \xce\xb2-pinene despite it having one of the fastest detoxification rate. Taken together our in vivo and in vitro results add to existing evidence that detoxification capacity is higher in dietary specialist than generalist herbivores. However, results also suggest that alternative mechanisms such as absorption and the pharmacological action of individual mixtures of PSMs may play a role in determining preference of PSMs within herbivore species.

preferences_of_specialist_and_generalist_mammalian_herbivores_for_mixtures_versus_individual_plant_s

INTRODUCTION<!>Animal Capture and Care.<!>Identification of Monoterpenes for In Vivo Feeding Studies and In Vitro Enzymatic Assays.<!>In Vivo Feeding Studies \xe2\x80\x93 Artificial Diets.<!>In Vivo Feeding Studies \xe2\x80\x93 Feeding Trials.<!>In Vitro Enzymatic Detoxification Assays.<!>Statistical Analysis.<!>RESULTS<!>DISCUSSION

<p>Plant secondary metabolites (PSMs) influence the foraging behavior of herbivores and may affect patterns of habitat selection at multiple scales (Duncan and Gordon 1999; Frye et al. 2013; Lawler et al. 2000; Moore and Foley 2005; Ulappa et al. 2014). High concentrations of PSMs often have deleterious effects on foraging herbivores (Degabriel et al. 2009; Estell 2010; Guglielmo et al. 1996; Sorensen et al. 2005a), and selective foraging is one mechanism to limit exposure to those PSMs (Frye et al. 2013; Moore and Foley 2005; Ulappa et al. 2014; Wiggins et al. 2006). Plants often contain complex mixtures of PSMs, the identities and concentrations of which can vary among taxa, populations, and individual plants within populations (Frye et al. 2013; Julkunen-Tiitto 1986; Hemming and Lindroth 1995; Lawler et al. 1998; Nyman and Julkunen-Tiitto 2005; O'Reilly-Wapstra et al. 2013; Richards et al. 2015; Thoss et al. 2007; Ulappa et al. 2014). This diversity of PSMs has wide-ranging physiological effects on vertebrate herbivores including reduced digestion, interference with cellular processes, and compromised energy budgets and reproductive success (Degabriel et al. 2009; Estell 2010; Guglielmo et al. 1996; Kohl et al. 2015; Sorensen et al. 2005a). Animals also cope with absorbed PSMs via different detoxification strategies (Sorensen et al. 2006; Sorensen and Dearing 2006), with specialist herbivores generally relying on faster and less expensive detoxification systems than their generalist counterparts (Boyle et al. 1999; Shipley et al. 2012; Sorensen and Dearing 2003a; Sorensen et al. 2004). The complexities of chemicals mixtures in plants and variable capacity of herbivores to detoxify PSMs make it difficult to identify which specific compounds, combinations, and concentrations drive observed patterns in diet selection by herbivores.</p><p>Three general approaches – field observations, in vivo captive studies, and in vitro enzymatic assays – have been used to understand how PSMs influence the foraging behavior of herbivores. Field-based, observational studies maintain the complexity inherent in natural systems while sacrificing a degree of causality in the relationships observed. These studies often identify correlations between intake and the concentration of individual PSMs and broad classes of PSMs (e.g., total monoterpenes or polyphenolics) that are thought to be representative of more complex mixtures of PSMs (Duncan et al. 1994; Moore and Foley 2005; Moore et al. 2010; Frye et al. 2013; Ulappa et al. 2014). The patterns that emerge from these studies may help predict habitat selection and foraging behavior, but are correlative, and must be considered in light of other habitat parameters (e.g., nutritional quality, predation risk, microclimate) that may complicate or obscure the interpretation of observed patterns.</p><p>In vivo laboratory studies address the mechanisms by which PSMs directly affect diet selection by manipulating concentrations of specific compounds and measuring food intake by captive animals (Dziba and Provenza 2008; Farentinos et al. 1981; Kirmani et al. 2010; Kimball et al. 2012; Shipley et al. 2012). Although better suited to establish causal relationships between PSMs and diet selection than field-based studies, captive studies often sacrifice natural chemical complexity by focusing on a single compound rather than complex mixtures of PSMs found in whole plants (McLean et al. 2007; Kirmani et al. 2010; Shipley et al. 2012; Wiggins et al. 2003). Captive studies that use artificial diets that contain whole plants or extracts from plants can preserve the chemical complexity of natural forage (Kohl et al. 2015; McIlwee et al. 2001; Sorensen et al. 2005a), but do not help identify which specific PSMs or combination of PSMs explain dietary preferences of herbivores. Additionally, many herbivores respond differently to diets containing individual versus mixtures of PSMs (Bernays et al. 1994; Dyer et al. 2003; Marsh et al. 2006; Richards et al. 2010, 2012; Wiggins et al. 2003). Generalist herbivores restricted to a single PSM may overload a specific detoxification pathway and consequently consume less food than when offered a diet containing an equivalent concentration of a mixture of PSMs (Burritt and Provenza 2000; Dearing and Cork 1999; Marsh et al. 2006; Wiggins et al. 2003). While the diversity and evenness of PSMs absorbed by specialist and generalist herbivores consuming natural plant diets has not, to our knowledge, been evaluated, greater PSM diversity can be inferred from studies demonstrating that the diversity of plants consumed is higher in generalists than specialists when both have equal access to plant communities (Crowell et al. 2018; Dial 1988). Specialist herbivores may show relatively higher tolerances for the PSMs they regularly encounter consuming primarily one plant species (Shipley et al. 2012; Sorensen et al. 2004, 2005a), but may have reduced tolerance for novel PSMs (Sorensen et al. 2005b). Captive feeding trials focused on individual PSMs do not capture the additive, synergistic, or potential inhibitory effects of consuming mixtures of PSMs. Likewise, trials employing artificial diets containing whole plants or plant extracts do not capture which combination or individual compound explain diet selection by herbivores nor do they reveal the mechanisms for variable tolerance among PSMs within a species or among species of herbivores.</p><p>In vitro pharmacological assays that quantify rates of enzymatic detoxification can provide insight into the mechanisms for variable tolerance of PSMs by herbivores. The majority of these studies focus on comparing enzymatic activity of microsomes from herbivores that vary in dietary selection using standard substrates (Dermauw and Van Leeuwen 2014; Green et al. 2004; Kumar et al. 2014; Labbé et al. 2011; Li et al. 2004; Skopec et al. 2007). Unfortunately, the majority of these assays only assess the rates of detoxification of standard substrates developed for use in model organisms or humans and do not assess how specific enzymes of herbivores detoxify the PSMs they encounter in natural forage. Even in human pharmacology, in vitro assays often do not predict in vivo outcomes (Karlsson et al. 2013; Tan et al. 2017). To our knowledge, no study has assessed whether the rate of detoxification of individual PSMs by metabolizing enzymes from wild vertebrate species can explain in vivo dietary preferences observed in the same species.</p><p>Incorporating biologically relevant mixtures of PSMs into captive feeding trials and coupling those trials with mechanistic understanding of the rates of detoxification of PSMs within the same mixture by specialists and generalists may lead to better predictions of diet selection in the field. To do this, we investigated the relationship between: 1) the relative preference of specialist (pygmy rabbit, Brachylagus idahoensis) and generalist (mountain cottontail, Sylvilagus nuttallii) mammalian herbivores for individual and mixtures of monoterpenes, a class of PSMs, in sagebrush (Artemisia tridentata spp.) and 2) the relative rates of detoxification of a mixture of the same individual monoterpenes by enzymes isolated from the specialist and generalist herbivores. Monoterpenes are a class of volatile PSMs that comprise approximately 2.5% of the dry weight (DW) of sagebrush leaves on plants browsed naturally by pygmy rabbits and cottontails (Crowell 2015). High concentrations of both total monoterpenes and specific individual monoterpenes have been correlated with reduced intake among a variety of free-ranging (Frye et al. 2013; Ulappa et al. 2014) and captive (Dziba and Provenza 2008; Kirmani et al. 2010; Lamb et al. 2004; Shipley et al. 2012) mammalian herbivores. Pygmy rabbits have a higher tolerance of sagebrush and specific monoterpenes than mountain cottontails (Camp et al. 2015; Camp et al. 2017; Shipley et al. 2012;). However, plant selection in the field and daily intake in laboratory studies by pygmy rabbits is compromised, at least in part, by increasing concentrations of monoterpenes (Camp et al. 2015; Camp et al. 2017; Ulappa et al. 2014; Utz et al. 2016;). The prevalence and variability of monoterpenes in sagebrush (Kelsey et al. 1982), their putative, differential, and dose-dependent effects on feeding behavior by a variety of specialist and generalist herbivores (Boyle et al. 1999; Lawler et al. 1998; Shipley et al. 2012; Wiggins et al. 2003), and commercial availability of pure forms of monoterpenes make them an ideal class of PSMs to assess the link between selection of individual versus mixtures of PSMs by herbivores and enzymatic detoxification rates of PSMs.</p><p>We first conducted in vivo assays to compare the relative preference of pygmy rabbits and cottontail rabbits between individual monoterpenes and mixtures. We also conducted in vitro enzymatic assays to compare the relative rate of detoxification of individual monoterpenes within a mixture using microsomal enzymes isolated from a pygmy rabbit (n = 1) and cottontail rabbits (n = 2). The relative proportions of monoterpenes in the mixture used in both in vivo and in vitro assays was representative of the composition and relative ratio of monoterpenes quantified in Wyoming big sagebrush (A. t. wyomingensis) from field sites where both pygmy rabbits and mountain cottontail rabbits forage. We hypothesized that specialists would be less selective between individual PSMs and a mixture of PSMs than their generalist counterparts due, in part, to faster rates of detoxification for all monoterpenes in sagebrush than generalists. In contrast, because PSMs consumed individually could overwhelm any single detoxification pathway (Estell 2010), we predicted that generalists would show stronger preferences for the mixture of monoterpenes which contained lower concentrations of any one monoterpene than specialists. Finally, we hypothesized that individual PSMs that were preferred compared to mixtures by a species would have the fastest rates of detoxification in that species.</p><p>By providing captive herbivores with a mixture of PSMs, we assessed how potential synergistic, antagonistic or neutral interactions among multiple PSMs influence diet selection by herbivores. By controlling the identities, concentrations, and ratios of PSMs within this mixture, we minimized the potentially confounding natural variation in concentrations and ratios of nutrients and PSMs found within whole plants. We propose that comparing preferences of herbivores between concentrations of mixtures of PSMs and equivalent concentrations of the individual PSM isolated from the mixtures occurring in whole plants would help identify which individual PSMs are most likely to influence foraging under natural conditions. Specifically, preference for a mixture over an equivalent concentration of an individual PSM might suggest selection and intake of whole plants is limited by concentrations of the avoided individual PSM. In contrast, preference for an individual PSM compared to a mixture may suggest relatively fast detoxification, low potential for toxic consequences, or high potential for beneficial consequences of that individual PSM. Although a simplified mixture is incapable of representing the full complexity of PSMs produced by wild plants, the individual compounds selected or avoided using this method could be targeted to establish and test hypotheses related to both the pattern and mechanism by which PSMs influence diet selection by wild herbivores.</p><!><p>We captured adult pygmy rabbits from sagebrush-dominated sites in Blaine, Camas, and Lemhi Counties in Idaho (Idaho Department of Fish and Game collection permits 100310 and 01813) and Beaverhead County, Montana (Montana Department of Fish, Wildlife, and Parks scientific collection permit 2014–062). We captured mountain cottontail rabbits in Pullman, Washington (Washington Department of Fish and Wildlife Scientific Collection Permit #14–206). When not undergoing trials, all animals were housed indoors in individual 1.2 × 1.8 m mesh cages at the Small Mammal Research Facility at Washington State University (Boise State University Institutional Animal Care and Use Committee Protocol # 006-AC12–009, Washington State University Institutional Animal Care and Use Committee Protocol # 04513–001). Animals not in trials were provided with ad libitum pelleted commercial rabbit chow (Purina Professional Rabbit Chow, Purina Mills LLC, St. Louise, MO) and fresh water and approximately 15 g/day of fresh mixed greens and greenhouse-grown basin big sagebrush (A. t. tridentata). The rabbit chow was the same used throughout experimental trials and was similar in fiber (36% by dry weight (DW)) and nitrogen (3.4% by DW) to sagebrush leaves (30% fiber and 2.5–4.5% nitrogen by DW, Camp et al. 2015). Rabbits were maintained at an average temperature of 7.66 °C (average minimum 1.58 °C, average maximum 13.42 °C) throughout trial period from 28 March through 16 April 2014.</p><!><p>To create a mixture of PSMs for in vivo feeding trials and in vitro enzymatic assays that mimicked the natural concentration of monoterpenes in sagebrush, we first analyzed the monoterpene profile of 420 individual Wyoming big sagebrush plants (Table 1). Plants were selected within a ~ 1000 ha area with evidence of browsing by both pygmy rabbits and mountain cottontails in southern Blaine County, Idaho (43°14' N, 114°19' W; elevation: 1470 m). Browsed plants were selected because previous work indicated that although the composition of monoterpenes does not differ between individual sagebrush within a species and within foraging patches browsed by vertebrate herbivores, including pygmy rabbits, the concentrations of individual monoterpenes can differ (Frye et al. 2013; Ulappa et al. 2014). As such, monoterpene profiles of browsed plants more accurately represent profiles that pygmy rabbits and cottontails would naturally consume. The monoterpene profile was analyzed from frozen leaf and stem material from each plant that was coarsely ground (< 2 mm particle size) in liquid nitrogen with a mortar and pestle. Relative concentrations of each monoterpene from each sample (100 mg wet weight) were determined using headspace gas chromatography. All samples were analyzed using an Agilent 6890N gas chromatograph (GC, Santa Clara, CA) coupled with a Hewlett-Packard HP7694 headspace autosampler (Palo Alto, CA). The headspace program was as follows: 100 °C oven temperature, 110 °C loop temperature, and 120 °C transfer line temperature. The vial equilibrium and pressurization times were each 0.20 minutes, the loop fill time was 0.50 minutes, the loop equilibrium time was 0.20 minutes, and the injection time was 0.50 minutes. One mL of headspace gas from each sample was injected into an Agilent J&W DB-5 capillary column (30 m × 250 μm × 0.25 μm, Santa Clara, CA) with helium as the carrier gas at a constant flow of 1.0 mL.min−1 and splitless injector temperature of 250 °C. The temperature program for the GC was as follows: 40 °C for 2.0 minutes, then increased by 3 °C.min−1 to 60 °C, then by 5 °C.min−1 to 120 °C and finally by 20 °C.min−1 to 300 °C where final temperature was held for seven minutes. Inlet pressure was 80 KPa and we used a flame ionization detector set at 300 °C. Retention times of individual monoterpenes and individual areas under the curve (AUC) were quantified using Hewlett-Packard ChemStation software version B.01.00 (Palo Alto, CA). Peaks were identified using co-chromatography with known standards. Samples were then dried at 60° C for 24 hours to correct for water content of sample and to calculate AUC per 100 μg of DW of sagebrush. Relative concentrations (AUC/100 μg DW) of individual monoterpenes were then averaged across all plants and divided by the total concentration of monoterpenes to obtain ratios among constituent compounds. To create a monoterpene mixture that represented whole sagebrush, we determined the proportions of the top five most prevalent individual monoterpenes in sagebrush based on relative AUC (α-pinene, β-pinene, camphene, camphor, 1,8-cineole at 99% purity or greater, Sigma Aldrich, St. Louis, MO, Table 1). These five compounds were added to food (in vivo assays) or microsomes (in vitro enzymatic assays) in the same average proportions in which they occurred naturally in sagebrush (Table 1).</p><!><p>For artificial diets, individual monoterpenes or the monoterpene mixture was added to pelleted rabbit chow at 1% of DW. Camphor and camphene are solids at room temperature and cannot be added homogenously to rabbit chow, whereas α-pinene, β-pinene, and 1,8-cineole are liquid and can be directly added to rabbit chow. Pure camphor (260 mg/mL) and camphene (248 mg/mL) were therefore dissolved together in methylene chloride (≥ 99.8% pure, Sigma Aldrich, St. Louis, MO). The methylene chloride mixture was thoroughly mixed with rabbit chow in a glass jar at a concentration of 25 μg/g DW of chow. The treated rabbit chow was then spread in a single layer in a fume hood for six hours to allow the highly volatile solvent to evaporate. The time needed for evaporation of the solvent relative to individual monoterpenes was determined by analyzing the concentration of methylene chloride and the camphor and camphene dissolved in methylene chloride added to rabbit chow over time until concentrations of methylene chloride were less than 1.0 μg/g DW of chow. The evaporation of camphor and camphene during the six hour period was minimal relative to the solvent, resulting in the desired final concentrations of monoterpenes (Table 1). In a preliminary study, we determined that pygmy rabbits and mountain cottontails did not discriminate between control rabbit chow and chow that was mixed with methylene chloride only (no camphor and camphene) and allowed to evaporate for six hours (Nobler 2016). After the solvent was evaporated, the remaining liquid monoterpenes were thoroughly mixed with the rabbit chow already treated with camphor and camphene in a glass jar. To prevent the volatilization of monoterpenes, all treated chow was stored at −20° C until offered to rabbits. Samples of treated rabbit chow were saved in sealed scintillation vials at −20° C before being analyzed for concentrations of monoterpenes via gas chromatography.</p><!><p>Before beginning feeding trials with monoterpene diets, all animals were acclimated to receiving commercial rabbit chow offered in equal portions at two feeding stations equal distances from a nest box over a period of three days. After acclimation, rabbits were offered a choice between rabbit chow treated with either 1% by DW of each individual monoterpene or 1% by DW monoterpene mixture (Table 1). This concentration represents the lower end of the range of monoterpene concentration by weight in sagebrush (Kelsey et al. 1982), and corresponds with concentrations at which individual monoterpenes reduce intake by mountain cottontails (Shipley et al. 2012). Individual monoterpene treatments that were paired with the mixture were administered sequentially, but in a randomly-determined order. Animals were also given rest periods of three to five days between treatments to prevent habituation. The mixture was first offered on a randomly determined side of the nest box, followed by alternating sides relative to the individual monoterpene treatment for three days to avoid directional bias (Utz 2012). We recorded the amount of food offered and remaining (orts) after 24 hours from each choice (individual monoterpene versus mixture) in each feeding trial (encompassing both diurnal and nocturnal intake), and corrected for DW by drying the orts and a sample of the treated rabbit chow offered at 100° C for ≥ 24 hrs. Five feeding trials were conducted (three days/trial) in which the monoterpene mixture was paired with each of the five individual monoterpenes.</p><!><p>Microsomes from a pygmy rabbit (n = 1) and mountain cottontails (n = 2) were prepared from livers obtained from freshly euthanized animals that had been in captivity for at least one year. Pygmy rabbits are a species of conservation concern with one population listed as endangered under the Endangered Species Act. Therefore, agencies were reluctant to issue permits that involved terminal outcomes for this species. Thus, the euthanasia of additional animals to increase sample sizes for rabbits was not possible. Tissues from euthanized animals were collected on dry ice and immediately transferred and stored at −70° C. All steps involved with sample preparation were carried out on ice. Partially thawed livers were cut into small pieces (< 3 mm2) and approximately 1.0 g of chopped tissue was combined with 3–4 mL of cold homogenizing buffer (150 mM KCl, 10 mM EDTA, 0.10 M Tris, pH 7.4). Tissue was homogenized with 5–8 short bursts using the probe of an Omni Tissue Master. The liver homogenates were then centrifuged at 12,500 × g for 15 minutes at 4 °C. The resulting supernatants were collected, then centrifuged at 105,000 × g for 70 minutes at 4 °C. Supernatants were discarded, and pellets were re-suspended in the original volume of homogenizing buffer. These samples were centrifuged again at 105,000 × g for 40 minutes at 4 °C. Supernatants were discarded, and the final pellet re-suspended in cold microsome buffer (10 mM EDTA, 20% glycerol, 0.050 M Tris, pH 7.5). The total protein concentration of the microsome suspensions was determined using a Biorad DC Protein assay kit according to manufacturer's directions and suspensions were adjusted to a final concentration of 20 mg/mL total protein prior to conducting enzymatic assays used to measure rates of detoxification of individual monoterpenes. Microsome suspensions were stored at −70 °C until use.</p><p>Rates of detoxification of individual monoterpenes within a mixture using microsomal enzymes isolated from a pygmy rabbit and mountain cottontails were monitored in vitro by measuring the percent difference in monoterpene concentration between paired enzyme reactions at time zero and at 15 minutes using headspace GC analysis. Concentrations of monoterpenes (α-pinene, β-pinene, camphene, camphor, 1,8-cineole) that represented proportions in whole sagebrush (Table 1) were dissolved as a mixture in DMSO at 50X final reaction concentrations. Assay tubes contained 864 μL of phosphate buffered saline solution (0.137 M NaCl, 0.01 M K2HPO4, 0.0027 M KCl, pH 7.4); 100 μL of 10 mM NADPH, and 26 μL of microsome (20 mg/mL in PBS). To start the reaction, 10 μL of the monoterpene mixture was added to microsomes in pairs. One paired reaction was incubated at 37 °C for zero minutes and the other paired reaction was incubated at 37 °C for 15 minutes. To terminate the reaction at zero or 15 minutes, the mixture was transferred to a 20 mL headspace vial containing 0.5 g NaCl, sealed, and heated for 1.0 minute at 200 °C. Rate of detoxification was determined as the percent difference in concentration of each monoterpene in the mixture between the enzyme reactions terminated at zero minutes and the reactions terminated at 15 minutes. Assays for each paired reaction for each microsomal enzyme sample were run in triplicate and thus represent pseudoreplication due to limited sample size of animals used to obtain microscomes. Negative control reactions included reactions that contained all components of enzyme reactions, but did not contain either NADPH nor microsomes or contained heat-denatured microsomes. Control reactions were used to confirm that loss of monoterpenes from assay tubes was only associated with microsomal enzyme activity.</p><!><p>To determine preferences for or against individual monoterpenes compared to a mixture, we divided the amount of each treatment consumed (i.e., individual monoterpene versus mixture) by the total amount of food consumed from both choices each day. The calculated proportion of total intake constituting a single monoterpene was averaged across the three day choice trial for each treatment for each animal. Preferences for the single monoterpene (compared to the mixture) are reported as the three-day mean proportion (± standard error) of the total food consumed constituting the individual monoterpene. Preferences were reported separately for each treatment comparison (n = 5), and for each rabbit species (i.e., pygmy rabbits and mountain cottontails). To evaluate the rabbits' preference for each treatment, we compared the proportion consumed of each treatment to 0.50 using a one sample t-test. Animals consuming an equal proportion (0.50) from the feeding station with the individual monoterpene and the feeding station with the monoterpene mixture were considered to have no preference between the treatments. To evaluate if the type of individual monoterpene influenced the proportion of the mixture, we used a mixed-effects linear model with the proportion of the individual monoterpene consumed as the response variable and rabbit species and treatment (i.e., type of individual monoterpene offered), and the interaction of species and treatment as fixed effects, and individual rabbit as a random effect. To investigate a potential relationship between preference and total daily intake, we added intake (total daily g DW consumed from both choices/g body mass) and the interaction between species and intake as fixed effects to the mixed-effects linear model with the proportion of the individual monoterpene consumed as the response variable and rabbit species and treatment (i.e., type of individual monoterpene offered), and the interaction between species and treatment as fixed effects, and individual rabbit as a random effect. To evaluate differences between species, we followed significant results with pairwise comparisons using a Tukey's HSD test adjusted p-value.</p><p>To compare rates of detoxification for monoterpenes, we used a generalized linear model with individual monoterpene and rabbit species, and the interaction of monoterpene and species as fixed effects. We used a Tukey's HSD test to compare rates of detoxification among monoterpenes within each species. All statistical analyses were conducted using R version 3.2.0 (R Foundation for Statistical Computing 2015) and JMP Pro 11.0 (SAS Institute Inc. 2013).</p><!><p>Both rabbit species responded to choices between individual monoterpenes and mixtures, but their preferences varied among individual monoterpenes. The proportion of individual monoterpenes consumed did not differ between species (F1,28 = 0.26, P > 0.05), but did differ with treatment (i.e., individual monoterpene offered, F4,28 = 18.04, P < 0.0001), and species × treatment interaction (F4,28 = 11.68, P < 0.0001). When offered choices between one of five individual monoterpenes compared to mixed monoterpenes, pygmy rabbits showed no preference when α-pinene (64% ± 0.11, t4 = −1.80, P > 0.05), β-pinene (43% ± 0.04, t4 = 2.06, P > 0.05), or camphene (52% ± 0.11, t4 = −0.27, P > 0.05) were paired with the mixture. However, pygmy rabbits preferred camphor (t4 = −4.37, P = 0.01) and 1,8-cineole (t4 = −4.93, P = 0.008) compared to the mixture (Fig. 1). The percentage of camphor (66% ± 0.08) or 1,8-cineole (70% ± 0.08) in the diet of pygmy rabbits was twice that of the monoterpene mixture (Fig. 1). The effect of total intake (daily g DW consumed from both choices/g body mass) on preferences was not significant (F1,26= 0.55, P = 0.46), nor was the interaction between species and total intake (F1,26 = 0.43, P = 0.51).</p><p>Similar to pygmy rabbits, mountain cottontails showed no significant preference between α-pinene (48% ± 0.13) and the monoterpene mixture (t3 = 0.20, P > 0.05). However, generalists showed significant preferences for both camphene (t3 = −9.77, P = 0.002) and 1,8-cineole (t3 = −23.81, P = 0.002), consuming more than five times as much camphene (85% ± 0.04) as the monoterpene mixture, and 24 times as much 1,8-cineole (96% ± 0.03) as the monoterpene mixture. Mountain cottontails consumed three times as much monoterpene mixture as β-pinene (25% ± 0.08, t11 = 0.643, P < 0.001) and twice as much monoterpene mixture as camphor (31% ± 0.08, t11 = 4.991, P < 0.001, Fig. 1).</p><p>Pygmy rabbits and cottontails did not differ in their preference for α-pinene (t28 = 1.81, P > 0.05, pygmy rabbit, 64% ± 0.11; cottontail, 48% ± 0.13), β-pinene (t28 = 2.08, P > 0.05, pygmy rabbit, 43% ± 0.04; cottontail, 25% ± 0.08), or 1,8-cineole (t28 = −3.00 P > 0.05, pygmy rabbit, 70% ± 0.08; cottontail, 96% ± 0.03, Fig 1) compared to the monoterpene mixture. However, the preferences between species differed significantly for camphene (t28 = −3.63, P = 0.03, Fig. 1), which was preferred by cottontails (85% ± 0.04) compared to the monoterpene mixture, but consumed in similar proportions (52% ± 0.11) to the monoterpene mixture by pygmy rabbits. Pygmy rabbits preferred camphor (66% ± 0.08) relative to the monoterpene mixture, and cottontails preferred the mixture relative to camphor (31% ± 0.08). Pygmy rabbits consumed twice the proportion of camphor as cottontails (t28 = 3.95, P = 0.01, Fig. 1).</p><p>Rates of detoxification were faster in the pygmy rabbit microsomes than in cottontails microsomes for all monoterpenes within the mixture (F1,35 = 371.6, P < 0.0001), and rates differed among individual monoterpenes (F4,35 = 27.0, P < 0.0001). The monoterpene by species interaction was removed because it was not significant (F4,35 = 0.66, P > 0.05). The percent difference for α-pinene (pygmy rabbit, 90.18% ± 3.23; cottontail, 31% ± 0.08), β-pinene (pygmy rabbit, 91.63% ± 1.74; cottontail, 39.9% ± 4.09), camphene (pygmy rabbit, 97.04% ± 0.54; cottontail, 41.76% ± 4.43), camphor (pygmy rabbit, 62.58% ± 0.42; cottontail, 8.22% ± 3.5), and 1,8-cineole (pygmy rabbit, 71.93% ± 0.43; cottontail, 15.74% ± 4.69) during a 15 minute reaction compared to a zero minute reaction was 2.0, 2.3, 2.3, 7.6, and 4.6 fold faster, respectively, for pygmy rabbit microsomes than for cottontail microsomes (Fig 2). In both pygmy rabbit and cottontail microsomes, camphor and 1,8-cineole did not differ from each other and had significantly slower rates of detoxification than α-pinene, β-pinene, and camphene which did not differ from each other (Fig. 2). After a 15 minute reaction with microsomes from a pygmy rabbit, there was only a 63% decline of camphor and 72% decline of 1,8-cineole compared to a decline of more than 90% for α-pinene, β-pinene, and camphene. Similarly, there was only an 8% decline of camphor and 16% decline of 1,8-cineole after reacting with cottontail microsomes compared to a decline of approximately 40% for α-pinene, β-pinene and camphene.</p><!><p>Dietary preferences of herbivores have long been hypothesized to be dictated by the physiological capacity of herbivores to process absorbed PSMs (Freeland and Janzen 1974; Freeland 1991; Foley et al. 1999). Specifically, faster rates of detoxification should increase tolerance and therefore relative intake of PSMs by herbivores. In support of expectations, the microsomes from specialist herbivores (pygmy rabbit) had faster rates of detoxification for all monoterpenes than the generalists (Fig 1) and are consistent with higher daily intake of single monoterpenes in captivity (cineole, Shipley et al. 2012) and higher proportion of sagebrush in the diet (Crowell et al. 2018) by specialist pygmy rabbits compared to generalist cottontails. In contrast, relative differences in detoxification rates among monoterpenes were not consistent with patterns of diet selection for individual monoterpenes within species in our study. For generalists, the hypothesis that monoterpenes with the fastest detoxification rates would be preferred over mixtures that contain monoterpenes with slower detoxification rates was only partially supported. Consistent with predictions, camphor had with the slowest detoxification rate in mountain cottontails and was associated with avoidance relative to the mixture. However, both mountain cottontails and pygmy rabbits preferred 1,8-cineole despite it having one of the slowest detoxification rates. In contrast to cottontails and in opposition to predictions, pygmy rabbits preferred camphor which had the slowest rate of detoxification. For both species, β-pinene had one of the fastest rates of detoxification, yet was associated with the lowest proportional intake of any individual monoterpene.</p><p>Preferences are likely a function of the dose-dependent pharmacological consequences of PSMs (Forbey et al. 2011; Kohl et al. 2015) that can be influenced by a variety of mechanisms. Limitations to enzymatic detoxification has received the most attention as an explanation limiting intake of any one plant by generalist herbivores like mountain cottontails (Dearing and Cork 1999; Dearing et al. 2000; Freeland and Janzen 1974; Shipley et al. 2009). Assuming different plants contain different types of PSMs that use different detoxification pathways, generalists are thought to avoid overwhelming a single detoxification pathway by mixing their diet. Diet mixing results in the intake of smaller amounts of any one plant and therefore smaller concentrations of any one PSM. In support, several generalist herbivores do consume more food when offered a diet containing mixed PSMs than when restricted to an individual PSM (Burritt and Provenza 2000; Dearing and Cork 1999; Wiggins et al. 2003). This pattern remains even when the diets are identical nutritionally (Bernays et al. 1994), supporting the hypothesis that saturated detoxification pathways can play a role in limiting intake (Freeland and Janzen 1974). The hypothesis that diet mixing by generalists minimizes saturation of detoxification pathways assumes that generalists have reduced capacity (lower diversity or expression) in the enzymes responsible for detoxifying individual PSMs compared to specialists and that individual PSMs use different detoxification pathways. Recent genomic studies provide evidence that insect (Calla et al. 2017) and vertebrate (Johnson et al. 2018; Kitanovic et al. 2018) specialists may have higher capacity to detoxify PSMs in host plants through relatively high diversification and duplication of the cytochrome P450 (CYP) enzymes. Although detoxification enzymes generally have broad substrate affinity, CYPs do have differential substrate selectivity for particular monoterpenes (Hernandez-Ortega et al. 2018) and affinity for one monoterpene can be shifted to another structurally similar monoterpene by mutations in the CYP enzyme (Bell et al. 2003). As such, genetic diversity of detoxification enzymes could result in differential capacity to detoxify individual monoterpenes.</p><p>Under the assumption that detoxification pathways are rate limited, dietary specialists have a greater diversity of detoxification pathways for PSMs in their host plant, and that individual monoterpenes have higher affinity for specific detoxification pathways, we expected mountain cottontails to prefer the monoterpene mixture that contained lower absolute concentrations of any individual monoterpene than diets containing a single monoterpene at higher concentrations (Table 1). However, cottontails preferred the monoterpene mixture only when paired with camphor and β-pinene, consumed equal proportions of the mixture and α-pinene, and preferred camphene and 1,8-cineole more than the mixture. Like cottontails, pgymy rabbits preferred 1,8-cineole more than the mixture, but also preferred camphor more than the mixture. However, pygmy rabbits did not demonstrate a preference for or against α-pinene, β-pinene, or camphene. A lack of preference for α-pinene by both specialists and generalists could indicate that the dose-dependent pharmacology of α-pinene is equivalent to that of a mixture of monoterpenes. Preference for individual monoterpenes relative to a mixture could indicate that 1% DW of the individual monoterpene was not at a high enough dose to have a negative pharmacological effect regardless of detoxification rate. Alternatively, preference for individual monoterpenes may indicate that the mixture at 1% DW had synergistic negative effects or contained individual compounds that are biologically active even at relatively low doses.</p><p>In vivo dietary preferences that are inconsistent with in vitro detoxification rates of liver microsomes may suggest differential rates of absorption among individual monoterpenes. Diet selection may be dependent on rates of detoxification by host and microbial enzymes in the intestine prior to absorption and mechanisms regulating the absorption of PSMs Cui, 2018, Kohl and Dearing 2017; Peters et al., 2016). Evidence exists that tolerance of PSMs by herbivores is linked to the functional attributes of microbial communities (Kohl et al. 2014) and mechanisms that limit absorption of ingested PSMs. For example, specialist woodrats absorbed five times less of the most abundant monoterpene in juniper (α-pinene) than generalist counterparts after receiving identical doses (Sorensen and Dearing 2003b) and specialist sage-grouse excrete PSMs from their diet of sagebrush unchanged in feces (Frye 2012, Thacker et al. 2012). In addition, inhibition of lymphatic absorption resulted in greater intake of PSMs in whole plants by generalist woodrats (Kohl and Dearing 2017). These studies provide examples of how in vivo experiments can be used to assess how intestinal absorption can explain tolerance of PSMs by herbivores. In addition, in vitro assays of efflux transporters and their substrates (see Sorensen et al. 2006) can be used to compare mechanisms that regulate absorption among taxa.</p><p>Evaluation of the matches and mismatches between in vitro rates of detoxification and in vivo diet selection from this study, coupled with physio-chemical properties of PSMs (e.g., tissue/blood partition coefficients, Daina et al., 2017) may help focus attention on particular PSMs most likely to influence foraging by vertebrate herbivores. For example, PSMs that are avoided at low concentrations by herbivores and have molecular structures that indicate high absorption may be particularly bioactive even at low concentrations in mixtures and could therefore serve as valuable predictors of intake by herbivores. For example, in vivo and in vitro results demonstrate that β-pinene comprised the lowest proportion of the total intake in both pygmy rabbits and mountain cottontails (Fig 1) despite it having one of the fastest rates of detoxification (Fig 2). In contrast, 1,8-cineole comprised the highest proportion of the total intake in both pygmy rabbits and mountain cottontails (Fig 1) despite it having one of the slowest rates of detoxification (Fig 2). Based on structural properties of β-pinene (lower molecular weight, lack of oxygen atom), this PSM is predicted to be more lipophilic and less water soluble and therefore has lower absorption than 1,8-cineole and is more likely to be an inhibitor of detoxification enzymes than1,8-cineole (from SwissADME, Daina et al., 2017). The predicted pharmacokinetic properties may explain the avoidance of individual β-pinene at 1% DW (10 mg/g DW) and why higher concentrations of cineole (at 1% DW, 10 mg/g DW) was preferred compared to low concentrations of β-pinene in the mixture (0.018% DW, 0.18 mg/g DW). The pharmacodynamic properties of PSMs may also explain preference patterns. For example, preference of pygmy rabbits and avoidance of cottontails for camphor may reflect differences in pharmacological mechanisms of action of this PSM. For example, camphor reduced digestive enzyme activity in a generalist more than in a specialist avian folivore (Greater sage-grouse, Centrocercus urophasianus, Kohl et al. 2015). We propose that pygmy rabbits may be more resistant to the pharmacological affects of camphor than cottontails. Although not tested in specialist pygmy rabbits, avian herbivores that specialize on sagebrush are more resistant to concentration-dependent inhibition of digestive enzymes by camphor than a generalist (Kohl et al. 2015). Similar resistance to this mechanism of action by pygmy rabbits may explain why pygmy rabbits can subsist almost entirely on sagebrush (Crowell et al. 2018) containing monoterpenes dominated by camphor (Table 1). The relatively high absolute concentration of camphor in the individual diet (10 mg/g DW) may also provide a more realistic olfactory cue for pygmy rabbits that naturally consume sagebrush containing camphor at similar concentrations (estimated at 14 mg/g DW of leaves, Table 1, Crowell 2015). Combined in vivo and in vitro assays could help isolate the olfactory cues that explain pre-ingestive diet selection (Finnerty et al. 2017, Schmitt et al. 2018) from the post-ingestive pharmacokinetic (absorption and detoxification, Kohl and Dearing 2017, Sorensen et al. 2006. Sorensen and Dearing 2006) and pharmacodynamic (mechanisms of action, Forbey et al. 2011, Kohl et al. 2015) consequences of subsequent dietary choices.</p><p>The role of PSMs in influencing patterns of foraging and habitat selection is slowly becoming better understood (Denno 2012; Frye et al. 2013; Lawler et al. 1998; Moore and Foley 2005; Moore et al. 2010; Rosenthal and Berenbaum 2012; Ulappa et al. 2014). However, the complexity of PSMs and the diverse effects PSMs have on the physiology and behavior of herbivores has made it difficult to identify the compounds and combinations of compounds most likely to drive complex patterns of foraging. When forced to choose at random from hundreds of potentially influential PSMs, chemical ecologists and physiologists have been hard pressed to narrow their focus and determine mechanistic relationships between compounds and the animals that consume them. Field-based studies can be used to identify and quantify the most common PSMs thought to influence habitat selection. Those data in turn, can inform the hybrid approach we present in this paper, in which simplified mixtures of PSMs can be used in in vivo and in vitro assays to identify the few compounds most likely to influence diet selection, either singly or in combinations. Moreover, combining in vivo captive studies, in vitro enzymatic assays, and predicted pharmacology based on the structure of PSMs could help establish and test a priori predictions of how specific mixtures of PSM influence both specialist and generalist herbivores in the field.</p>

PubMed Author Manuscript

Ice Recrystallization Inhibition by Amino Acids: The Curious Case of Alpha- and Beta-Alanine

Extremophiles produce macromolecules which inhibit ice recrystallization, but there is increasing interest in discovering and developing small molecules that can modulate ice growth. Realizing their potential requires an understanding of how these molecules function at the atomistic level. Here, we report the discovery that the amino acid l-α-alanine demonstrates ice recrystallization inhibition (IRI) activity, functioning at 100 mM (∼10 mg/mL). We combined experimental assays with molecular simulations to investigate this IRI agent, drawing comparison to β-alanine, an isomer of l-α-alanine which displays no IRI activity. We found that the difference in the IRI activity of these molecules does not originate from their ice binding affinity, but from their capacity to (not) become overgrown, dictated by the degree of structural (in)compatibility within the growing ice lattice. These findings shed new light on the microscopic mechanisms of small molecule cryoprotectants, particularly in terms of their molecular structure and overgrowth by ice.

ice_recrystallization_inhibition_by_amino_acids:_the_curious_case_of_alpha-_and_beta-alanine

<!>Author Contributions<!>

<p>The cryopreservation of biological materials is a key factor in regenerative medicine and cell-based therapies,1,2 but strategies are required to limit the cellular damage incurred at subzero temperatures.3 Organic solvents such DMSO and glycerol are widely used as cryoprotectants,4 and while they are effective, they are not suitable for all cell types and not all cells are typically recovered post-thaw.5,6 The cryopreservation of large tissues also remains extremely challenging. A major contributor to post-thaw damage is ice recrystallization (IR), an Ostwald ripening process7 whereby large ice crystals grow in favor of smaller ones, exerting mechanical and osmotic stress on the sample which often leads to post-thaw cell death. Ice recrystallization inhibition (IRI) represents an appealing strategy to improve cryopreservation outcomes, and ice recrystallization inhibitors (IRIs) have consequently received significant interest in recent years. However, the mechanism(s) by which these molecules inhibit IR and the underlying molecular determinants are poorly understood. In turn, this hampers the discovery of new materials that possess a sought-after balance of potent inhibition (at low concentrations), biocompatibility, and amenability to low cost, large scale production.</p><p>The IRI-active materials discovered to date are diverse, ranging from polymers,8 to the well-known ice-binding proteins,9 to small molecules.10 These materials can all give rise to the same macroscopic effect defined as IRI, but multiple molecular-level mechanisms appear to underpin the observable phenomenon. For certain materials, including many of the antifreeze proteins (AFPs) and glycoproteins (AFGPs), IRI activity is linked to a molecule's ability to bind to specific faces of an ice crystal, causing a local positive curvature of the ice front which inhibits further ice growth through the Gibbs–Thomson (Kelvin) effect.11,12 By arresting ice growth, surface adsorption also results in a non-colligative depression of the freezing point relative to the melting point, known as thermal hysteresis (TH). TH is often accompanied by dynamic ice shaping (DIS), whereby ice crystals display distinct morphologies reflecting the specific lattice face(s) onto which the AF(G)Ps adsorbed.13 The question remains: how do these materials first recognize and bind ice in a vast excess of liquid water? The answer is still unclear. A diversity of chemical motifs appear capable of binding to ice, but the relative contributions of hydrogen bonding, hydrophobic interactions,14,15 and ordered clathrate waters16 are still under debate.</p><p>Equally unresolved are the mechanistic details of IRI. Recent studies have confirmed that, although IRI and TH activities appear connected, the two properties are not directly correlated in AFPs.17,18 Meanwhile, for other IRI-active materials such as the small molecule carbohydrates reported by Ben and co-workers,19,20 there is no evidence of ice binding at all (i.e., no TH or DIS). These findings both point to an alternate mode of inhibition that is independent of ice binding and likely the primary mechanism of action for small molecules. One possible mechanism, proposed by Ben and colleagues,19 suggests that small molecules inhibit IR by disrupting the order of water in the interfacial region between bulk water and the preordered water layer surrounding an ice crystal. However, this hypothesis lacks extensive experimental support and is challenging to study computationally. Given that ice binding can often shape bound crystals into needlelike spicules that are injurious to biological materials, understanding this mode of inhibition represents a crucial step toward the identification of novel IR inhibitors which have clinical application. In this context, small molecules are also attractive targets because they can be efficiently manufactured, and obtaining structure–property relationships is simpler than for, e.g., polymers, which have intrinsic heterogeneity (e.g., dispersity).</p><p>Here, we report that the simple amino acid l-α-alanine (herein referred to as α-alanine) exhibits IRI activity at millimolar concentrations. We also show that its structural isomer, β-alanine, is IRI-inactive at equivalent concentrations, despite the structural similarity of these two forms. To unravel the origin of α-alanine's IRI activity we used atomistic molecular dynamics (MD) simulations. These simulations allow for ice growth kinetics to be studied in the presence of (α/β-)alanine, shedding new light on the mechanisms of small molecule IRIs and the molecular determinants of IRI activity.</p><p>We began by assessing the ability of α- and β-alanine to inhibit IR using the "splat" cooling assay. This assay involves rapidly cooling a droplet of solution to form a polycrystalline ice monolayer (Figure 1c), which is then annealed at −8 °C. The growth of crystals within this layer is determined by comparing the mean grain size (MGS) after 30 min to a positive control for ice growth. A smaller relative (%) MGS value therefore indicates stronger inhibition. This assay requires salt (or other additives) to ensure a eutectic phase is formed, in order to avoid false positives that arise from using pure water alone.21,22 When α-alanine was tested in phosphate buffered saline (PBS), which is typically used for this assay, no IRI activity was observed (Figure S1). However, in a series of controls using α-alanine and betaine (another IRI-inactive small molecule of similar molecular weight to α-alanine), we found that including 10 mM NaCl led to significant IRI with α-alanine (and not betaine) and is a sufficient quantity of saline to avoid false positive hits (see the Supporting Information and Figure S1).23 We also found that α-alanine remains active in 10 mM phosphate buffer but not under higher concentrations of saline (100 mM NaCl) (Figure S1); hence, we suggest that the IRI activity of this system is sensitive to high concentrations of salt rather than specific components of the PBS solution. We also note that, while this assay is usually performed in PBS, the use of saline solution for IRI measurements is not uncommon in this field.24,25</p><p>(a) Ice recrystallization inhibition activity of α-alanine and β-alanine. Error bars are ±1 SD from a minimum of three repeats. The percentage mean grain size (MGS) is reported relative to a saline control (10 mM NaCl). (b) Structures of α-alanine and β-alanine. (c) Example cryomicrographs of ice wafers from the "splat" cooling assay, grown in the presence of (50 mM) α-alanine or β-alanine.</p><p>Using 10 mM NaCl, we found that α-alanine is able to suppress ice growth almost entirely at millimolar concentrations, producing crystals which are dramatically smaller than in the controls (Figure 1a). This level of activity can be considered moderate in comparison to the most potent IRI-active materials, such as poly vinyl alcohol (PVA),26 but on the same magnitude as other small molecules tested under similar conditions.24 For example, PVA20 (where 20 is the degree of polymerization) achieves similar levels of inhibition to 100 mM α-alanine (∼10 mg/mL) at ∼1 mg/mL.27 However, the IRI activity of α-alanine must be considered in the context of the molecule's size. Given that its structure comprises merely 13 atoms (corresponding to a molecular mass below 90 atomic mass units), to our knowledge α-alanine represents the smallest ice recrystallization inhibitor discovered to date; it is significantly smaller than other materials (including small molecules) reported elsewhere.10,20,28</p><p>In the knowledge that α-alanine is inactive at high salt concentrations (100 mM and above), we note that the cryoprotective applications of α-alanine specifically may be limited to cases where biological materials are stored in low-salt buffers. However, we highlight that there are a number bacterial and plant growth media which satisfy these minimal salt requirements (e.g., 2xYT,29 M9,30 TB,31 and MK32 media), and such materials require effective cryopreservation for both industrial and research applications.</p><p>To confirm whether this property is unique to α-alanine, we also tested the IRI efficacy of its structural isomer, β-alanine, noting that the stereoisomers l- and d-α-alanine are equally IRI active. Strikingly, despite the structural similarity to the α-form, the β-isomer does not exhibit any IRI activity at equivalent concentrations (Figure 1a). This difference was observed across a range of concentrations, most notably at 100 mM where the % MGS of α-alanine and β-alanine were 5% and 65%, respectively. It is important to highlight that almost any material can inhibit ice growth at sufficiently high concentrations (e.g., as shown for β-alanine in Figure 1a). However, given that α-alanine demonstrates a clear inhibitory effect at low concentrations (100 mM and below), in multiple buffer systems, and in stark contrast to β-alanine, nonspecific inhibitory effects can be ruled out and the basis for the IRI activity of α-alanine can be addressed.</p><p>To investigate whether α-alanine has any effect on ice crystal morphology, and therefore binds ice, we used a modified version of the "sucrose-sandwich" assay. In this assay, concentrated sucrose solution (50% w/v) is used to produce segregated ice crystals whose individual shapes can be clearly observed. Crystals grown in the presence of α-alanine did not exhibit morphologies that are characteristic of ice-binding (Figure S2). This is reminiscent of other small molecule IRIs elsewhere discovered,19 in contrast to larger IRI-active materials (e.g., PVA, AFPs).33−35 In the standard "sucrose-sandwich" assay, we also observed no ice growth inhibition (Figure S3). We hypothesize that this is due to the high fraction of liquid (sucrose solution) in this assay, which means that diffusion effects are dominant and the effective concentration of amino acid at the ice/water interface is very low, resulting in no observable activity.</p><p>In the absence of ice shaping effects, we looked to identify an alternative mechanism to explain the IRI activities of α- and β-alanine that does not depend on ice binding. This amounts to a challenging task, given the transient nature of the ice/water interface during recrystallization and the size and similarities of these two structures. To overcome this, we used atomistic MD simulations to study the growth of ice in the presence of α- or β-alanine. These simulations comprise an ice/water interface featuring a central slab of ice and two adjacent slabs of water in contact with a vacuum, as depicted in Figure 2a. The ice slab is orientated so that a selected lattice face (Figure S4a) is exposed to the water in the xy-plane, seeding ice growth in the ±z-direction. We used the all-atomistic CHARMM36 force field36 along with the TIP4P/Ice model37 to simulate the amino acid and water molecules. Further details of the computational setup can be found in the Supporting Information. Using this setup, we collected 60 statistically independent trajectories for α-alanine as well as β-alanine: 20 for each of the prismatic and the basal faces. We also probed concentration effects by including either one or two molecules in each of the two water slabs in every trajectory. The simulations were run for 100 ns (prismatic faces) or 120 ns (basal face), by which time only a small quantity of liquid water typically remains in the simulation cell. We quantified ice growth by calculating the number of molecules in the seeded ice cluster over time, which was then converted to a growth rate. We note that while these simulations do not fully capture the complex recrystallization process (Ostwald ripening) in its entirety, the transfer of water molecules from the supercooled liquid fraction at the grain boundaries to the surface of a growing crystal is a fundamental step in ice recrystallization. Our simulations capture the kinetics of this process (i.e., the growth rate), which is thus directly proportional to the rate of ice recrystallization represented by the % MGS metric. This computational methodology has been validated extensively, and results obtained via this setup have shown excellent correlation with the experimentally observed IRI activity of both polymers38 and small peptides.39</p><p>(a) Computational setup illustrating the growth of a primary prismatic plane in the ±z direction. (b) Rates of ice growth in simulations containing α- (left, circles) or β-alanine (right, triangles). The rate of ice growth is calculated over the period from when (α/β-)alanine first binds ice until the end of the simulation. Box plots in part b show the median and quartiles of the distribution.</p><p>In the simulations concerning the primary and secondary prismatic planes, we found that the presence of α-alanine resulted in slower rates of growth compared to β-alanine (Figure 2b), consistent with our experimental % MGS data. The basal fronts displayed similar rates of growth in the presence of either α- or β-alanine, although we note that in the context of IRI, growth (inhibition) at the prismatic fronts is considered of greater relevance due to the rapid growth rates observed from these faces relative to the basal face.17,40 To quantify these results categorically, we define a simulation as showing ice growth inhibition when the observed growth rate is below 0.03 m/s. This corresponds to a minimum of 50% reduction in the growth rate compared to our control simulations which contain no (α/β-)alanine molecules, although the following trends hold regardless of the chosen cutoff value within the range of 0.01 and 0.06 m/s. Applying this definition, we observed 5 and 6 instances of inhibition by α-alanine, and 1 and 2 instances by β-alanine, for the primary and secondary prismatic planes, respectively (Table 1).</p><p>Numbers in parentheses show the outcomes for the simulations with two alanine molecules.</p><p>A trajectory was defined as overgrown (OG) if at least one alanine molecule is deposited within the ice at least two layers (∼8 Å) deep along the z-axis with respect to the water by the end of the simulation.</p><p>A trajectory was defined as showing ice growth inhibition (IRI) if the growth rate is below 0.03 m/s.</p><p>The numbers for trajectories shown here do not include those which are overgrown (OG).</p><p>With our simulation and experimental data in agreement, we examined the trajectories looking for differences between α- and β-alanine that could explain their IRI activities. Interestingly, a significant number of (α/β-)alanine molecules were found to become overgrown by the advancing ice front and then incorporated into the lattice, as depicted in Figure S4b. We defined a trajectory as overgrown if at least one molecule of (α/β-)alanine is deposited under two or more layers (∼8 Å) of ice by the end of the simulation. These outcomes, summarized in Table 1, revealed that β-alanine is more frequently overgrown than α-alanine by each of the three crystal fronts studied. Again, we highlight these differences for the primary and secondary prismatic planes, where β-alanine is overgrown in 10 and 15 instances, respectively, compared to just 2 and 7 cases for α-alanine. Hence, this process readily occurs under these conditions, in contrast to larger molecules such as AFPs and polymers, for which overgrowth is considered a rare event at similar levels of supercooling.41,42</p><p>Having established this difference between α- and β-alanine, we suggest that inhibition and overgrown outcomes might be linked, because once the molecule becomes overgrown, the ice front can advance unimpeded and any inhibitory capacity is lost. In contrast, when the (α/β-)alanine molecule is not overgrown, it is still able to disrupt the growth of ice at or near the interface. Indeed, such differences can clearly be observed by comparing the growth rates of nonovergrown and overgrown trajectories (Figure S5). We also suggest that the overgrowth of (α/β-)alanine molecules has a compounding effect, as it sequesters the amino acids in the ice fraction. Consequently, the effective concentration of α-alanine at the grain boundaries could increase over time relative to β-alanine, further slowing ice growth via greater surface coverage. It is necessary to point out that there are cases wherein individual simulations, as with their experimental counterparts, do not reflect this overall trend. Fundamentally, IRI activity is not an "on/off" property, especially when examined at the scale of these simulations, and a large number of independent trajectories were therefore required to validate these findings. We also note that in the simulations where each ice front is exposed to two (α/β-)alanine molecules, the same trends in terms of ice growth inhibition and overgrowth are observed (Table 1, Figure S6). In these simulations, ice also grows at a consistently slower rate compared to those containing a single amino acid molecule (Figure S6), reflecting the concentration effects we observe experimentally (Figure 1a). Nonetheless, we focus our analysis herein on simulations with one (α/β-)alanine molecule, as we can clearly define these cases as overgrown (or not) with respect to a single molecule.</p><p>To understand why β-alanine is more likely to become overgrown compared to α-alanine, we sought to identify relevant characteristics that differ between these two molecules. For AFPs, the size of the molecule or area of its ice binding site is known to correlate strongly with antifreeze activity.43,44 Our previous work38 also revealed that, for the flexible polymer PVA, the effective volume and contact area with the ice surface can also determine the strength of IRI. Therefore, we first computed the volume and solvent-accessible surface area (SASA) occupied by these two molecules throughout the course of the simulations. However, we found that these properties of α-alanine and β-alanine are almost indistinguishable, with average volumes and surface areas differing by approximately 1 Å3 and 3 Å2 in volume and SASA, respectively, corresponding to a relative difference of around 1% for each quantity (Figure S7). Moreover, while α-alanine typically occupies a larger volume than β-alanine, this trend is reversed with respect to the surface areas. Hence, molecular volumes or surface areas do not appear to be correlated with IRI activity for these small molecules. We also investigated the (binding) orientation of the molecule with respect to the ice front but found no correlation between this property and IRI activity (data not shown).</p><p>Next, we analyzed the hydrogen bonding interactions between these molecules and water/ice. We found that the hydrogen bonding capacity of α- and β-alanine differ significantly, considering both molecules share the same hydrogen bond donor and acceptor groups. Whilst β-alanine frequently forms three or four hydrogen bonds with water via its carboxylate group, the ability of α-alanine to form a full complement of bonds via this same group is impaired (Figure S8, left panel). In fact, we observed that α-alanine typically forms just one or two hydrogen bonds out of a possible four. The number of hydrogen bonds formed via the amine group, meanwhile, does not appear to differ between these two compounds (Figure S8, right panel). This difference arises from the relative positions of the carboxylate and amine groups in these two molecules. For α-alanine, these groups are bound to the same (α-)carbon atom, whereas for β-alanine they are separated by an additional methylene group (Figure 1b). Consequently, the carboxylate and amine groups in α-alanine are fixed within close proximity, and the rotation of the carboxylate group (defined by the O–C–C–C/O–C–C–N dihedral angle for α- and β-alanine, respectively) is restricted due to the electrostatic interaction between the protonated nitrogen atom of the amine group and the nearest oxygen of the carboxylate moiety. We confirmed this using well-tempered metadynamics, employing the aforementioned dihedral angles as the collective variables. The resulting free energy landscape revealed two energy minima in this phase space for α-alanine, corresponding to two conformations wherein the distance between the nitrogen atom and each oxygen in turn is minimized (Figure S9a). The high energy barrier that exists between these conformers prohibits free rotation about the C– C bond, consistent with the narrow dihedral distribution observed in our unbiased simulations (Figure S9b). Hydrogen bond formation is sterically hindered in these conformations, limiting the hydrogen bond interactions between α-alanine and water. In contrast, the free energy landscape of β-alanine features conformational energy barriers on the same order as thermal fluctuations at room temperature, allowing this phase space to be fully explored during the unbiased run and greater hydrogen bond formation compared to α-alanine.</p><p>Given the difference in the hydrogen bonding capacities of α- and β-alanine, we investigated differences in the solvation shells of these two molecules. We determined the hydration index, based on the definition provided by Tam et al.,19 to be 0.152 ± 0.023 and 0.162 ± 0.021 molecules/Å3 for α- and β-alanine, respectively. This small difference suggests a marginally greater entropic gain associated with the desolvation of β-alanine compared to α-alanine. To consolidate these results, we also computed the solvation free energy, ΔGsolv, for both species via MD simulations using the Bennett acceptance ratio method.45 The solvation energies of α- and β-alanine were found to be −36.9 ± 0.1 and −43.5 ± 0.2 kcal/mol, respectively, in line with solvation energies previously reported.46 The greater solvation energy of β-alanine compared to α-alanine is consistent with both hydrogen bonding and hydration index data. These results at first appear to counter intuition: β-alanine can form a greater number of hydrogen bonds with ice than α-alanine and stands to benefit from a larger entropic gain upon binding, yet it is less effective at inhibiting ice growth. We instead suggest that a stronger interaction with the ice front, confirmed by the aforementioned computational analyses, could be detrimental to the IRI activity of these small molecules as it increases the likelihood of becoming overgrown. Further, we also provide substantial evidence that the molecule's compatibility within the ice lattice is crucial to the overgrowth outcome. To demonstrate this, we computed the distances between the atoms in (α/β-)alanine that are able to participate in hydrogen bonding (nitrogen and oxygen). We found that the nitrogen–oxygen distances provide a close match with the lattice distances of the prismatic and basal planes for β-alanine but not for α-alanine (Figure 3a). The lattice distances represent the oxygen–oxygen distances between water molecules in ice (Figure 3b), and therefore a closer match to these distances means that the molecule can be incorporated into a growing crystal at the lattice sites and overgrown without significant disruption to the crystal order.</p><p>(a) (α/β-)alanine N–O distance distributions for all simulations. The solid blue lines represent the average ice lattice distance sampled from these trajectories. The shaded cyan area represents ±1 standard deviation. (b) Schematic showing the characteristic ice lattice distances for the primary prismatic, basal, and secondary prismatic faces (left to right). These faces are exposed to water in the xy-plane during the simulations.</p><p>Further, we also observed that β-alanine offers a more compatible "fit" within the crystal lattice than α-alanine with respect to its tetrahedral arrangement with neighboring water molecules. When a given water oxygen (OW) was replaced by an oxygen atom (OAla) or nitrogen atom (NAla) from β-alanine (Figure 4b), the corresponding angles between these atoms and water were in close agreement with the O···O···O angles observed in a tetrahedrally coordinated lattice structure consisting of just water molecules (Figure 4a and Figure S10). In contrast, the angles formed between the oxygen atoms of α-alanine showed a broader distribution and a greater deviation from the O···O···O angle in pure ice. Similarly, the geometry of (α/β-)alanine molecules that become overgrown (right panels, Figure 4a) also displayed greater tetrahedral character with coordinated water molecules than those that were not overgrown (left panels, Figure 4a). Hence, these data also provide rationale as to why certain molecules of either (α/β-)alanine become overgrown whereas others do not. Given that these small molecules have relatively few degrees of freedom with respect to their geometry, this highlights the subtlety of the features which can determine whether a molecule becomes overgrown and consequently the level of ice growth inhibition.</p><p>(a) O···O···O and O···N···O angle distributions for simulations of (α/β-)alanine with the primary prismatic plane of ice exposed. The solid green line represents the average O···O···O angle between tetrahedrally coordinated water molecules in the ice crystal, sampled from these trajectories. The shaded green area represents ±1 standard deviation of the sampled angles. These distributions are representative of the those observed for the basal and secondary prismatic simulations, which can be found in the Supporting Information (Figure S10). (b) Snapshot of α-alanine and water molecules showing representative O···O···O (top) and O···N···O (bottom) angles (θ). These angles are calculated for the three nearest water molecules (e.g., OW1–OW3) to (OAla) (top) and (NAla) (bottom), respectively, computed at every frame. For O···O···O angles, both O atoms of (α/β-)alanine are considered.</p><p>In summary, we have shown via α-alanine that very small molecules, with fewer than 15 atoms, can be effective IRI agents and represent a scaffold to understand structure-function relationships. We have also brought to attention amino acids as a new class of IRI-active materials, investigating the IRI activity of α-alanine alongside its isomer β-alanine, using quantitative experimental measurements and atomistic molecular simulations. Surprisingly, we found that the difference in IRI activity of these structures is dictated by their propensity to become engulfed and irreversibly overgrown by ice, underpinned by their compatibility to fit within the ice lattice. We note that the trends observed here with respect to ice binding significantly differ from those reported in the literature for, e.g., antifreeze proteins,15 highlighting the different structural determinants at play in small molecule IRIs. These findings provide new insights and avenues for the discovery and development of small molecule cryoprotectants, building upon the ubiquitous amino acid scaffold. In light of the limited IRI activity of α-alanine under high salt concentrations, the identification of saline-stable inhibitors represents a focal point for future work.</p><p>Additional experimental and computational details and methods including IRI and DIS assays and controls; MD and metadynamics simulations; geometry optimization and RESP charge fitting; determination of ice clusters and overgrowth; hydrogen bonding; calculation of molecular volumes, SASAs, hydration indices and solvation energies; supplementary figures including IRI controls; cryomicrographs from IRI and DIS assays; ice growth rates and data, volume, SASA and hydrogen bond distributions; free energy profiles; and O···O···O and O···N···O angle distributions (PDF)</p><p>Transparent Peer Review report available (PDF)</p><p>jz1c04080_si_001.pdf</p><p>jz1c04080_si_002.pdf</p><!><p>M.T.W. and I.G. performed the experiments and analysis. M.T.W and F.B. performed and analyzed the simulations. All authors interpreted the results. G.C.S. and M.I.G. conceived the research. M.T.W., F.B., G.C.S., and M.I.G. wrote the manuscript.</p><!><p>The authors declare no competing financial interest.</p>

PubMed Open Access

Enantioselective Catalytic [4+1]-Cyclization of\northo-Hydroxy-para-Quinone Methides with\nAllenoates

The first highly asymmetric catalytic synthesis of densely functionalized dihydrobenzofurans is reported, which starts from ortho-hydroxy-containing para-quinone methides. The reaction relies on an unprecedented formal [4+1]-annulation of these quinone methides with allenoates in the presence of a commercially available chiral phosphine catalyst. The chiral dihydrobenzofurans were obtained as single diastereomers in yields up to 90% and with enantiomeric ratios up to 95:5.

enantioselective_catalytic_[4+1]-cyclization_of\northo-hydroxy-para-quinone_methides_with\nallenoate

Introduction<!>Initial optimization of the racemic reaction<!>Development of an asymmetric catalytic protocol<!>Application scope<!>Mechanistic considerations<!>Conclusions<!>Experimental Section<!>General asymmetric [4+1]-cyclization procedure<!>Dihydrobenzofuran 5a

<p>The 2,3-dihydrobenzofuran scaffold is a prominent structural motif found in numerous biologically active (natural) compounds[1] and the development of novel synthesis strategies to access these targets has been a heavily investigated topic over the last years.[2–8] One especially appealing approach to access chiral 2,3-dihydrobenzofurans is the formal [4+1]-cyclization[9] between a suitable C1 building block and a carefully chosen (maybe in situ generated) acceptor–donor containing C4 building block. The most versatile class of C4 building blocks used to obtain the dihydrobenzofuran skeleton 1 via a formal [4+1]-cyclization are ortho-quinone methides (o-QMs) 2.[10] These, usually in situ generated, reactive compounds have recently been very successfully used for racemic as well as highly stereoselective [4+1]-annulations with either sulfonium or ammonium ylides,[4] α-halocarbonyl compounds,[5] or with diazocompounds as C1 synthons (Scheme 1A).[6] Alternatively, the hydroxy-containing para-quinone methides 3 have very recently emerged as powerful building blocks for formal (4+n)-annulations as well.[7, 11–13] Interestingly however, their applicability for asymmetric [4+1]-cyclizations to access dihydrobenzofurans 1 has so far been rather limited, with highly asymmetric protocols still being rare (Scheme 1B).[7] Two years ago we reported the first highly enantioselective synthesis of compounds 1 by reacting preformed chiral ammonium ylides with in situ formed o-QMs 2.[4c] More recently we found that the highly functionalized allenoates 4 can undergo a very unique (and up to then unprecedented) formal [4+1]-cyclization with acceptors 2 in the presence of a stoichiometric amount of PPh3 (Scheme 1C).[8] The (unexpected) outcome of this reaction was in sharp contrast to other previously described reactions between o-QMs 2 and (differently substituted) allenoates, which all resulted in formal [4+2]-annulations.[14] Unfortunately however, we were only able to carry out this reaction in racemic manner, as even the use of a stoichiometric amount of different commonly used chiral phosphine catalysts gave low yields and poor enantioselectivities only.[8] In especially we found that the in situ formed o-QMs 2 decomposed rather rapidly under the previously developed reaction conditions, thus making a catalytic approach difficult. Given these limitations in catalyst turnover and asymmetric induction, we thought that maybe an alternative and slightly more stable acceptor would be beneficial to address these challenges. Thus, we decided to investigate if this methodology may also be extended to the formal [4+1]-cyclization of the o-hydroxy-containing p-QMs 3 with allenoates 4.[12] We reasoned that this preformed and, compared to 2, more stable acceptor molecule may be better suited to establish a truly catalytic as well as asymmetric protocol, which would then allow for the first enantioselective [4+1]-annulation of p-QMs 3 to access the highly functionalized chiral dihydrobenzofurans 5 (Scheme 1D).</p><!><p>We started our investigations by carrying out the racemic reaction between p-QM 3a and diethyl allenoate 4a in the presence of PPh3 (Table 1 gives an overview of the most significant screening results). Our first reactions were carried out in analogy to the conditions developed for the annulation of o-QMs 2[8] (please note that we previously used a twofold excess of the quinone methide 2 to compensate for its competing decomposition under the reaction conditions). Gratifyingly the targeted dihydrobenzofuran 5a could be obtained as a single diastereomer in this initial attempt already (entry 1). The relative configuration of the product 5a was confirmed by NOESY experiments (as shown in Scheme 4) and we also observed the same correlations for other products 5 later (Scheme 2). As the reaction was found to be rather slow, with significant amounts of unreacted 3a being recovered (indicating its increased stability compared to o-QMs 2), we next increased the amount of base (entry 2), which however had a detrimental effect (complete decomposition of starting materials).</p><p>As decomposition of the acceptor 3a was not very fast in the first attempts with 2 equivalents of base, we next used an excess of allenoate 4a, which led to a measurable increase in yield (entry 3). The screening of different solvents revealed that toluene allows for a slightly higher yield (entry 4), but also accompanied with a more pronounced formation of various not identified side- or decomposition products. Other solvents did not give satisfactory results (see entry 5 for one example) and so further optimizations with CH2Cl2 were carried out. Very interestingly, lowering the amount of base (entry 6) significantly improved the yield and suppressed side product formation. By testing other bases, K2CO3 turned out to be the most promising (entry 7). It should be noted that other simple trialkylphosphines were tested as well,[15] but in analogy to our previous observation[8] these did not allow for this [4+1]-annulation.</p><p>With these first high yielding conditions set, we next lowered the amount of PPh3. Gratifyingly, and in sharp contrast to the reaction with o-QMs 2,[8] the use of 20 mol% PPh3 allowed for the same yield as when using a stoichiometric amount (compare entries 7 and 8). Further lowering of the catalyst amount unfortunately slowed down the reaction measurably (entry 9). Considering the beneficial effect of using less base when using Cs2CO3 (entry 6), we finally also lowered the amount of K2CO3 (entries 10, 11), and much to our surprise the reaction proceeded well even without any base (entry 11; the reaction was reproduced several times on different scales and also on 1 mmol scale).</p><!><p>Having established high yielding and robust catalytic procedures for the racemic synthesis of 5a we next focused on the use of chiral phosphine catalysts. As already mentioned before, we were not able to identify a suited asymmetric catalyst for our previous [4+1]-annulation of o-QMs 2. However, given the fact that p-QM 3a performed very well in the racemic reaction and also allowed for a catalytic approach, we were confident that the well-described bulky chiral phosphines A[16] or B[17] may allow for a truly catalytic enantioselective protocol (Table 2). We first used the binaphthyl-based phosphines A1–3, but unfortunately neither of them allowed for any product formation (entries 1–3). Gratifyingly however, by switching to the commercially available chiral spiro phosphine B ((R)-SITCP)[17] we observed a very clean and reasonably enantioselective product formation when using 20 mol% of this catalyst under base-free conditions in CH2Cl2 (entry 4). Lowering the reaction temperature to 0°C unfortunately did not allow for product formation anymore (entry 5). When carrying out the reaction in the presence of two equivalents of K2CO3, the outcome was only slightly affected in this solvent (entry 6).</p><p>Interestingly, when changing to toluene (other solvents like THF were found to be not suited), we were able to improve the enantioselectivity significantly (entries 7–9). At room temperature reactions in the presence of K2CO3 as well as under base-free conditions performed very similarly, with a slightly higher e.r. in the absence of base (compare entries 7 and 8). However, when we further investigated the application scope, we realized that the base-mediated conditions were more robust when using differently substituted starting materials 3, while not all of those allowed for good conversions under base-free conditions. Other bases were found to be less satisfactory (with for example, Cs2CO3 giving lower yields and K3PO4 giving no product at all). We thus tested if any further improvement in the presence of 2 equivalents of K2CO3 may be possible (entries 8–11). However, lowering of the reaction temperature was possible to some extent only (entries 9, 10), but reducing the catalyst loading to 10 mol% was unfortunately not possible anymore (entry 11). Accordingly, the best-suited and most robust catalytic enantioselective approach to access 5a as a single diastereomer was to carry out the reaction in toluene at 10°C in the presence of 2 equivalents of K2CO3 by using 20 mol% of the commercially available phosphine catalyst B (entry 9, the reaction was reproduced by different persons on 0.05–0.1 mmol 3a scale giving identical results).</p><!><p>Having established a high yielding and robust catalytic procedure for the synthesis of dihydrobenzofuran 5a, we next tested the use of differently substituted quinone methides 3 and allenoates 4 (Scheme 2).</p><p>First, we could show that replacement of one of the allenoate ethyl ester groups for a benzyl ester was tolerated very well (see product 5b). Then it turned out that a dimethyl-based p-QM 3 can be used as well to obtain the enantioen-riched product 5c (albeit with a slightly lower selectivity than for the parent tBu-based 5a). Interestingly, substituents in the 5 and 6-position of the benzofuran backbone were very well tolerated (see compounds 5d–f, 5i–k, 5m). In contrast, substituents in positions 4 and 8 turned out to be more limiting and product 5g was only accessible with a rather low enantiomeric ratio of 79:21. Surprisingly, compound 5h was not formed at all under the asymmetric conditions (even with longer reaction times). We were however able to obtain racemic 5h in high yield when using PPh3 as an achiral catalyst. Very interestingly, while we found initially that benzyl ester containing allenoates were tolerated similarly well as ethyl ester-based ones (see targets 5a and 5b), we found that tert-butyl esters resulted in somewhat lower enantiomeric ratios compared to ethyl and benzyl esters (compare 5k and 5l as well as 5m, 5n, and 5o). All asymmetric reactions were initially carried out on 0.05 mmol scale of the limiting agent 3 and we also reproduced selected reactions on up to 0.2 mmol scale without affecting the outcome, thus indicating that the asymmetric procedure is of similar robustness as the racemic one (Table 1, entry 11).</p><p>All substrate combinations gave the (+)-enantiomer as the major product, but unfortunately, it has not been possible to obtain suitable crystals of any of the products 5 to determine the absolute configuration by single-crystal X-ray analysis.</p><p>It has been described by others that the tert-butyl groups of the phenol derivatives obtained by addition of nucleophiles to QMs can be cleaved off under (Lewis) acidic conditions.[7a, 11c] We thus carried out a few (unoptimized) test reactions to see if a similar debutylation is also possible on the highly functionalized diester-containing dihydrobenzofuranes 5 (Scheme 3). Carrying out the reaction at elevated temperature only led to decomposition. In contrast, at room temperature, the slow formation of the debutylated diester 6a was observed by MS. Interestingly however, the major product was found to be the debutylated monoester 7a that was formed in around 30% after one day and around 50–60% after 3 days (accompanied with some decomposition products) and which could also be isolated after column chromatography (NMR clearly confirmed that the ester group on the stereogenic center was hydrolyzed). It should be noted that no further attempts to optimize this reaction were undertaken, but this result clearly shows that the highly functionalized compounds 5a can be used for further transformations and that the two ester groups have different reactivities.</p><!><p>Mechanistically this is a rather complex reaction and it should be admitted that so far, we only have some hints that may allow us to postulate the mechanistic scenario depicted in Scheme 4. This proposal is also based on our recent observations made for the racemic [4+1]-annulation of o-QMs 2 where we found that intramolecular rapid proton transfers are crucial to explain the outcome of this [4+1]-cyclization.[8]</p><p>Addition of the phosphine to the allenoate is supposed to give the required zwitterion I after proton transfer on the primary addition product. Following the reaction between PPh3 and 4a by 31P NMR shows the appearance of two new signals around 27 ppm (the parent PPh3 peak is at –5 ppm) substantiating the formation of alkylated phosphine species (these addition products decomposed very quickly in the absence of any electrophile). Upon addition of the quinone methide 3a immediately a strong red color evolves, which can be rationalized by the 1,6-addition of I to 3a to give the phenolate II. Similar color changes can also be observed when adding different nucleophiles to other p-QMs, substantiating the assumed initial 1,6-addition. With respect to the nature of the electrophile 3a one could however also postulate that a prototropic shift from the phenol to the para-QM moiety gives an ortho-QM in situ, which then reacts with I to give III directly.[18] However, we found compound 3a being rather stable under basic conditions and we never observed any other species or got any experimental hint that supports this pathway, but it should not be ruled out completely with the current state of knowledge. The phenolate II then needs to undergo two proton transfer reactions towards the betaine IV, which can then finally react to the product 5a via an Sn2'-type cyclization. We have recently shown for the cyclizations of o-QMs 2 that these proton transfers are rather likely processes and we reason that the presence of a base is beneficial for these reactions, which would be an explanation why the herein presented [4+1]-cyclization is more robust under basic conditions. This beneficial effect of base became especially pronounced in those cases where no electron-donating ring substituent para to the OH group is present (these reactions usually proceeded a bit slower as well). This observation supports a scenario where the final ring closure may be the rate-determining step, which also rationalizes why slightly larger amounts of catalyst were necessary to obtain satisfying catalyst turnover.</p><p>With respect to the observed stereoselectivity it is likely that the catalyst controls the absolute configuration in the 1,6-addition step. An alternative may be a less selective 1,6-addition followed by base-mediated isomerization of the benzylic position on one of the chiral catalyst-bound intermediates II or III. However, as the observed enantioselectivity was more or less the same under basic and base-free conditions (compare with Table 2), this option seems less likely. The diastereoselectivity is then controlled in the final proton transfer–cyclization sequence. Given the fact that SN2′ reactions usually proceed with a cis-orientation of nucleophile and leaving group[19] the proton transfer towards IV is supposed to be highly selective, and may be steered by electrostatic attraction between the phenolate anion and the phosphonium cation in the nonpolar reaction solvent. However, it should clearly be pointed out that this is just a mechanistic hypothesis and although we were able to observe the presence of some alkylated phosphonium species by 31P NMR during the reaction, none of these intermediates could be isolated or more carefully analyzed.</p><!><p>The first highly asymmetric catalytic formal [4+1]-annulation of o-hydroxy-p-quinone methides 3 with allenoates 4 has been developed. The outcome of this reaction is in sharp contrast to other recently reported reactions between quinone methides 3 and allenoates.[12] Key to success was the use of the commercially available chiral phosphine B as a catalyst under carefully optimized reaction conditions. This methodology allowed for the so far unprecedented synthesis of the chiral dihydrobenzofurans 5 as single diastereomers in yields up to 90% and with enantiomeric ratios up to 95:5.</p><!><p>General details can be found in the online Supporting Information. This document also contains detailed synthesis procedures and analytical data of novel compounds and reaction products as well as copies of NMR spectra and HPLC traces.</p><!><p>A mixture of the para-quinone methide 3 (0.05–0.2 mmol), K2CO3 (2 equiv), and chiral phosphine B (20 mol%) was cooled to 10°C and a solution of the allenoate 4 (2 equiv) in dry toluene (20 mL per mmol 4) was added. The resulting mixture was stirred at 10°C under an Ar atmosphere for approximately 20 h. The mixture was diluted by adding CH2Cl2 (5 mL), filtrated over a pad of Na2SO4 and the residue was rinsed with CH2Cl2 (5 × 5 mL). The combined organic layers were evaporated to dryness (under reduce pressure) and the products were purified by silica gel column chromatography (gradient of heptanes and EtOAc) giving the corresponding dihydrobenzofurans 5 in the reported yields and enantiopurities (Syntheses of racemic samples were carried out in analogy using PPh3 instead).</p><!><p>Obtained as a yellow residue in 89% and e.r.=94:6. [α]D23=64.6 (c = 0.15, CHCl3, e.r. = 94:6); 1H NMR (300 MHz, δ, CDCl3, 298 K): δ = 0.82 (t, J = 7.2 Hz, 3H), 1.36 (t, J = 7.1 Hz, 3H), 1.36 (s, 18H), 1.89 (d, J = 7.2 Hz, 3H), 3.52–3.77 (m, 2H), 4.26–4.36 (m, 2H), 5.11 (s, 1H), 5.24 (s, 1H), 6.40 (q, J = 7.1 Hz, 1H), 6.83 (s, 2H), 6.94 (t, J = 7.3 Hz, 1H), 7.01–7.09 (m, 2H), 7.19–7.26 ppm (m, 1H). 13C NMR (75 MHz, δ, CDCl3, 298 K): δ = 13.4, 14.2, 15.5, 30.2, 34.2, 55.8, 60.8, 61.1, 94.0, 110.2, 121.8, 125.7, 126.1, 128.9, 129.4, 130.0, 132.7, 133.2, 135.1, 153.0, 158.1, 167.2, 168.4 ppm; HRMS (ESI): m/z calcd for C31H40O6:509.2898 [M+H]+; found: 509.2897. The enantioselectivity was determined by HPLC (YMC Chiral Art Cellulose-SB, eluent: hexane/iPrOH = 95:5, 0.5 mLmin−1, 10°C, retention times: tmajor = 9.4 min, tminor = 11.0 min).</p>

PubMed Author Manuscript

Patterning of transition metal dichalcogenides catalyzed by surface plasmons with atomic precision

A surface plasmon polariton (SPP)-driven etching strategy is reported to etch various transition metal dichalcogenides (TMDs) into desired layers and lateral sizes by controlling the light power. The etching of TMD nanoflakes is highly dependent on the SPP propagating directions, whereby diverse patterns of TMD homostructures can be achieved. A SPP-induced synergetic effect of weakening the interlayer interaction and oxidizing TMDs by H 2 O 2 is proposed to elucidate the underlying etching mechanism.

patterning_of_transition_metal_dichalcogenides_catalyzed_by_surface_plasmons_with_atomic_precision

INTRODUCTION<!>RESULTS AND DISCUSSION<!>Conclusions<!>EXPERIMENTAL PROCEDURES<!>Materials availability<!>Data and code availability<!>Experimental setup<!>SPPE of TMD nanoflakes<!>Characterization<!>Calculations<!>DFT calculations<!>SUPPLEMENTAL INFORMATION

<p>Surface plasmons possess light-trapping and electromagnetic-field-concentrating properties and open up a wide range of applications in solar energy conversion, [1][2][3][4][5][6][7] metasurfaces, 8,9 and ultrasensitive sensing. [10][11][12] Localized surface plasmon resonance of metal nanostructures can occur in the metal surface in which confined free electrons oscillate upon the incident radiation, generating an intense and highly localized optical fields. 2,13 An important potential application of localized surface plasmon resonance is driving chemical reactions of molecules adsorbed on metal nanosctructures. 14 Electron-hole pairs could be generated in the metal nanostructures. Therefore, great efforts have been made to promote charge separation by fabricating metal-semiconductor Schottky junctions. 3,15,16 A variety of plasmonic metal/transition metal dichalcogenide (TMDs) heterostructures, such as Au nanorods/MoS 2 , Ag nanowires/MoS 2 , Au nanotriangles/WS 2 , and WSe 2 /Au trench, were developed with enhanced light-matter interactions to boost the efficiency of photocatalysis and photodetectors. [17][18][19][20][21][22][23][24] All these two-dimensional (2D) materials exhibited stable layered structures. The transformation of 2D TMDs during the plasmon-induced catalysis is very rare, 25,26 whereas the erosion of TMDs has not yet been reported.</p><p>The exotic layer-dependent properties of TMDs have aroused great interests in the fabrication of 2D TMDs with desired layers and lateral size. Currently, the common</p><p>The bigger picture Plasmonic metal/transition metal dichalcogenide (TMD) heterostructures have attracted a considerable interest owing to plasmon-induced interfacial charge transfer, which has emerged as an important topic in photocatalysis and photovoltaics. In this work, we found an exotic interaction of TMDs with surface plasmon polaritons, which can be exploited as a general method to fabricate TMDs with controllable layers and lateral sizes. Taking MoS 2 as an example, the strong plasmonic coupling at the Au/ MoS 2 interface can accumulate holes in the valence band of MoS 2 and weaken its interlayer interaction. Meanwhile, the plasmonic hot electrons would transfer on MoS 2 surface to produce the oxidizing H 2 O 2 and etch top layer MoS 2 in aqueous media. By controlling the light power, monolayer, bilayer, and trilayer MoS 2 can be achieved with a pristine lateral size. Our findings offer deep insights into the plasmonic coupling at metal-TMDs interfaces, opening up a new avenue for controlled fabrication of TMDs. routes to produce 2D TMDs often suffer from low efficiency, random layer distribution, and unsatisfied lateral sizes. Mechanical exfoliation can get desired layers of TMDs, but the yield is low. 27 Solution-based exfoliation often obtains random layers of TMDs and is afflicted by poor crystal crystalline owing to the introduction of defects or phase transitions. [28][29][30] Furthermore, the atomic layer etching has been used to control the MoS 2 layers, 31 but along with the experimental complexity and harsh conditions. Although chemical vapor deposition has been a prevalent method to prepare 2D TMDs and can obtain different layers, it is still plagued by the defect and polycrystalline with grain boundaries. 32,33 Therefore, it remains a great challenge to achieve TMD flakes with precise control of the layer number and lateral sizes.</p><p>Here, we report the etching of TMDs by planar surface plasmon polaritons (SPPs). Unlike the localized surface plasmon resonance confined in metal nanostructures, planar SPPs could be excited with a weak illumination and propagates along the planar metal interface. 34 We found the SPPs catalyzed the etching of TMDs, which was highly dependent on the SPP propagating directions. By using this capability, we demonstrated a general approach for the precise fabrication of highly uniform and crystalline TMDs with controllable layer number and lateral sizes. Taking MoS 2 as an example, desired layers of MoS 2 (monolayers, bilayers, and trilayers) can be controllably achieved by simply tuning the power density of exciting light.</p><!><p>To excite SPPs, we used a home-built inverted optical microscope with a high numerical aperture objective. 35,36 Through phase matching, red light with wavelength of 660 nm was irradiated on a gold surface at a certain incident angle, i.e., the resonant angle (Figure S1). The excited SPPs resulting from the coupling of surface electromagnetic polaritons to oscillating free electrons propagated along the Au/water interface, generating an evanescent field with a penetration depth of approximately 200 nm. We employed this bifunctional setup for triggering and monitoring layercontrolled etching of 2D TMD materials (Figure 1). In a typical procedure, we mechanically exfoliated multilayered MoS 2 nanoflakes from bulk crystals and transferred them onto a gold-coated glass coverslip (Figure S2). Surprisingly, we observed an etching phenomenon in multilayered MoS 2 nanoflakes upon illumination at the resonant angle under the microscope in deionized (DI) water. Optical images of a MoS 2 nanoflake before and after etching for 40 min clearly show the morphological changes when we used a low red-light power density (1.7 mW$mm À2 ) for exciting SPPs (Figures 2A and 2B). A video of the etching process, presented in Video S1, reveals that the top layers of a MoS 2 nanoflake were gradually etched away with increasing illumination time. Finally, a thinner MoS 2 nanoflake with the same lateral size as before the experiment remained on the Au substrate. Because the beam diameter (ca. 0.9 mm) is far beyond the SPP propagation length of ca. 6.5 mm, it can be expected that all the exfoliated TMD flakes within the beam diameter can be etched.</p><p>To quantify the layer number of the MoS 2 nanoflake remaining on the Au substrate, we measured the thickness of the MoS 2 nanoflake using atomic force microscopy (AFM). The height of the residual MoS 2 nanoflake was approximately 2.6 nm (Figure 2C), which corresponds to trilayer MoS 2 . Furthermore, we also employed Raman spectroscopy to characterize the MoS 2 nanoflake before and after etching. The Raman spectrum is sensitive to the layer number, by reading out a frequency difference of approximately 20, 22, and 23 cm À1 for monolayer, bilayer, and trilayer MoS 2 nanoflakes, respectively. 37 Raman spectra of pristine and residual MoS 2 nanoflake exhibit two typical characteristic vibrational modes, E 1 2g (in-plane) and A 1g (Figure 2D), respectively. The MoS 2 nanoflake after etching shows obvious smaller frequency difference of 23.1 cm À1 between these two peaks than that of pristine MoS2 nanoflake (24.9 cm À1 ), suggesting that the residual MoS 2 nanoflakes can be identified to three layers of MoS 2 , consistent with the abovementioned AFM analysis.</p><p>We further increased the illumination power density and found that the layer number of residual MoS 2 could be precisely regulated to the bilayer and monolayer. When the power density of the red light was increased to 3.4 mW$mm À2 , the residual MoS 2 was more transparent and thinner after etching (Figures 2E and 2F). After etching, the AFM height of ~1.83 nm corresponded to bilayer MoS 2 (Figure 2G). This layer identification was corroborated by the Raman spectra with a smaller frequency difference of 21.4 cm À1 (Figure 2H). As we continued to increase the power density to 6.8 mW$mm À2 , monolayer MoS 2 could be preserved on the Au substrate (Figures 2I and 2J). Both the AFM height of 0.87 nm and the frequency difference of 20.2 cm À1 validated the monolayer structure of MoS 2 (Figures 2K and 2L). More examples of trilayer, bilayer, and monolayer MoS 2 after SPPs-driven etching (SPPE) displayed similar frequency differences (see more details in Figure S3), implying that this approach could be a robust method for fabricating MoS 2 with controlled layers. Notably, this method can precisely control the layer number and retain the lateral size of pristine MoS 2 flakes, which is not feasible with other mechanical-force-driven exfoliation or bottom-up growth methods, providing a reliable preparation approach to meet the layer number and size demands of TMD nanoflakes for device fabrication.</p><p>We assumed that the underlying driving force of etching in our work originated from the role of SPPs. To verify this hypothesis, we conducted control experiments to demonstrate that the SPPs was responsible for the etching of MoS 2 nanoflakes. First, we changed the p-polarized light to s-polarized light and recorded unaffected MoS 2 nanoflakes (Figures S4A-S4C). For metal films, the p-polarized light commonly Article excites the propagating SPPs much more efficiently than s-polarized light, with a larger amplitude under the surface normal electric field. 38 Second, we illuminated the incident light vertically without SPPs excitation and found that the MoS 2 nanoflakes remained unchanged (Figures S4D-S4F). Third, to exclude the light irradiation angle effect, we used bare glass and an indium tin oxide-coated glass slide under the same total internal reflection configuration and observed unchanged MoS 2 nanoflakes (Figures S4G-S4L). Fourth, the theoretically determined relationship between the incident angle and the wavelength of p-polarized light shows that the SPPs could not be launched in the wavelength range from 520 to 570 nm (Figure S5). We used a 570 nm p-polarized light as the light source and found the MoS 2 nanoflakes could not be etched. These results provided solid evidence that the excitation of SPPs was pivotal to the etching of MoS 2 nanoflakes. Moreover, the local temperature variation can be calculated from the differential plasmonic image of MoS 2 nanoflakes based on the quantitative dependence of local reflectivity on temperature. 39 We found that the temperature increase was negligible, ruling out the possibility of the photothermal effect on the etching of MoS 2 nanoflakes (Figure S6).</p><p>To clarify the etching process, we tried to probe the potential active oxidizing species that might be associated with plasmonic excitation and etching of the MoS 2 . We first used ascorbic acid and isopropanol solutions to trap oxidizing $O 2À and $OH, which are representative reactive oxygen species (ROS). The etching of MoS 2 nanoflakes still occurred, ruling out the effect of $O 2À and $OH in our system (Figures S7A-S7D). The hydrogen peroxide can be another candidate oxidizing species, which presents strong oxidizing ability. 40,41 We used 0.01 mg/mL catalase solution as H 2 O 2 scavenger and observed that the etching of MoS 2 nanoflakes was significantly suppressed (Figures S7E and S7F). This finding indicated that the presence of H 2 O 2 played a critical role in the etching process of MoS 2 nanoflakes.</p><p>It is known that SPPs at Au/MoS 2 interface can generate hot (or highly energetic) carriers via two possible pathways: the non-radiative decay into hot carriers in Au and the resonant energy transfer to excite electrons and holes in MoS 2 directly. [42][43][44][45] In the former case, the hot electron with sufficient energy can transfer from Au into MoS 2 by overcoming the barrier at the interfaces (Figure 3A). The plasmonic hot electron will thermalize during the transport and react with oxygen and H 2 O to produce H 2 O 2 through O 2 + 2e À + 2H + /H 2 O 2 . In our case, H 2 O 2 can induce the oxidation dissolution and etching of MoS 2 multilayers into desired layers in the aqueous environment via</p><p>The formation of H 2 O 2 during SPP interaction was confirmed using a fluorescence probe (Figure S8). In addition, we did not observe a layer-by-layer etching of MoS 2 nanoflake when it was immersed in a 0. The participation of dissolved O 2 in the dissolution of MoS 2 was confirmed by irradiating the Au/MoS 2 sample in the air, where the etching process was completely suppressed (Figure S10). In addition, the etching was significantly suppressed when N 2 -saturated deionized water was used as the medium. These results validated In contrast, the resonant energy transfer from the plasmon decay can excite electrons and holes in MoS 2 directly. When the spectral functions of surface plasmon of Au film and exciton excitation in thin MoS 2 overlap, the strong plasmon-exciton coupling regime can be expected. [46][47][48][49] The band structure calculated with a two-layer MoS 2 represents the typical indirect band gap and notable interlayer features between the S atoms in the upper and lower layers (Figures 3C and 3D). Two kinds of exciton excitation can take place in MoS 2 flakes, which are the direct gap excitation at the K-K point (1.7 eV) and the indirect band-gap excitation at the G-K points (1.2 eV) mediated by intervalley scattering. 50,51 Considering the SPP has an energy of 1.85 eV, both the two kinds of the exciton can be excited in MoS 2 through the resonant energy transfer process. Thus, the holes are formed in the valence band at both G and K points, where those at K point will further decay to the valence band maximum (VBM) at G point. Moreover, the S-S distance between the upper and lower layers (d ss ) is found to increase dramatically as a function of increased hole doping in VBM (Figure 3E). Consequently, the accumulation of holes in VBM of MoS 2 can weaken the interlayer interactions in MoS 2 , which further facilitates the oxidation dissolution of top layers of MoS 2 . Therefore, we suggest that the holes in VBM of MoS 2 , produced via exciton-plasmon coupling, are responsible for the interlayer repulsion, and the electrons, produced in Au via plasmon decay, are responsible for the dissolution. These two processes together contribute to the overall etching process layer by layer.</p><p>The residual layers of MoS 2 suggest a strong interaction between the Au film and adjacent MoS 2 layer, which could stem from the strong affinity of Au for sulfur atoms with high bond strength. 52 Density functional theory (DFT) calculations show a binding energy value up to 1.43 J/m 2 between Au and MoS 2 , confirming the strong interaction (Figure S12). We also applied this method to prepare other 2D TMD materials, including WS 2 , MoSe 2 , and ternary MoS 2x Se 2(1Àx) (Figure S13), which showed similar etching features.</p><p>More interestingly, we found that the etching of MoS 2 top layers was highly dependent on the SPP propagation direction, which could be easily tuned by changing the light irradiation position in the optical path. A video of the SPPs propagation direction-dependent etching behavior is presented in Video S2. Figures 4B and 4D show several snapshots of the etching process. When the SPPs propagated from left to right, etching occurred first on the left side of the MoS 2 nanoflake and gradually extended toward the center along the SPPs propagation direction (Figures 4A and 4B). When we adjusted the SPP propagation direction to be from right to left, the right edge of the nanoflake began to be etched toward the central region (Figures 4C and 4D). Finally, the whole top layers of the MoS 2 nanoflake were completely etched.</p><p>To reveal the anisotropic etching behavior, we captured plasmonic images of MoS 2 nanoflakes. The scattering pattern of MoS 2 shows a high dependence on the SPP propagation direction (Figure 4E). The edge of MoS 2 encountering the SPPs exhibited a higher intensity contrast than the edge away from the SPPs. To quantify the local electrical field distribution, we employed COMSOL Multiphysics software to simulate the scattering pattern of MoS 2 nanoflakes (Figure S14). The simulated images clearly show the local electric field distribution on a MoS 2 nanoflake, with the electric field intensity gradually decreasing along the SPPs propagation direction (Figure The strongest electric field emerged at the edge of MoS 2 encountering the SPPs, consistent with the plasmonic images. Furthermore, we also simulated the local electric field distribution on MoS 2 with offset angle variation (Figure S15), which conformed to the above results. Thus, the experimental plasmonic images and theoretical simulation suggest that the electric field spatial distribution along the SPP propagation direction determines the etching direction of MoS 2 nanoflakes.</p><p>Based on the etching capability of the plasmonic coupling at Au/MoS 2 interface, here, we demonstrated a proof-of-concept application in MoS 2 patterning by manipulating the propagation direction of the SPP and the illumination duration (Figure 5A). Figure 5B presents the patterning process of a MoS 2 homostructure using two different SPP propagation directions. We obtained a multilayer MoS 2 domain on the top center after switching the direction of the light irradiation. Because the etching initiation was highly dependent on the propagation direction of the SPP, we designed MoS 2 homostructures with different features through precise selection of the SPP propagation direction and etching time. The desired shape, size, and thickness of multilayer domains remaining on the top center of MoS 2 flakes can be explicitly patterned (Figure 5C), providing new possibilities for realizing novel properties in MoS 2 -based electronic devices. Because the etching area is dependent on the beam diameter, we could enlarge the illumination size to achieve large-scale etching of TMD based on other setups, such as prism based configuration for SPP excitation. Our SPPE approach is versatile, being applicable not only to fabrication of TMDs with controlled layers and size but also to artificial patterning of desired architectures, which is a challenge for other techniques. Therefore, these results demonstrated that our method has great potential for sample processing on the nanoscale owing to the oxidizing capability and controlled manipulation of Article propagating polaritons, which can be expected to be utilized in future micro-/nanoprocessing technology.</p><!><p>In summary, we have successfully developed a general approach for the precise control of the tuneable thickness and size of 2D TMD materials. Thick MoS 2 multilayers can be precisely etched into monolayers, bilayers, and trilayers by simply varying the output power of the exciting light while preserving the pristine lateral size. In combination of the theoretical calculations, we revealed that the production of ROS from the plasmonic hot electrons and weakened interlayer interaction from the exited holes in MoS 2 were responsible for the etching event, and diverse MoS 2 homostructures can be patterned by manipulating the propagation direction of the SPPs. Our findings provide not only a deeper understanding of SPPs-driven photochemistry but also a new perspective on plasmon-matter interactions, offering a paradigm to precisely fabricate 2D TMD materials to obtain novel physical properties and devices.</p><!><p>Resource availability Lead contact Further information and requests for resources should be directed to and will be fulfilled by the Lead Contact, Xian-Wei Liu (xianweiliu@ustc.edu.cn).</p><!><p>This study did not generate new unique reagents.</p><!><p>All data supporting this study are available in the manuscript and Supplemental information.</p><!><p>The etching of TMD nanoflakes was conducted on a home-built optical microscope without lithography. The setup was built based on an inverted total internal reflection microscope (Nikon Ti-E) with a 603 oil-immersion objective. 35,36 A laser diode (Coherent OBIS, elength = 660 nm, output power adjustable from 0.5 to 70 mW) was adopted as the light source, and a polarizer was inserted in the optical path to generate p-polarized light to excite the SPPs. Note that the maximum effective output power of the red light for generating SPPs was ~35 mW because only 50% of the incident light reached the objective after passing through a 50/50 beam splitter. The actual power density of the red light (for etching events) was detected by a power meter (PM100D, Thorlabs). The propagation direction of the SPPs can be easily tuned by changing the light irradiation position in the optical path. Gold chips were prepared by evaporating 2 nm chromium as an adhesion layer followed by a 47 nm gold layer on BK-7 glass coverslips. The optical images were recorded using two CCD cameras (MER-500-14U3C, Daheng Imavision, China, and Pike F-032B from Allied Vision Technologies, Newburyport, MA).</p><!><p>Bulk MoS 2 , WS 2 , MoSe 2 , and ternary MoS 2x Se 2(1Àx) single crystals were purchased from Nanjing MKNANO. Taking MoS 2 as an example, bulk MoS 2 single crystals were used to prepare mechanically exfoliated MoS 2 nanoflakes on gold chips using the scotch tape method. A polydimethylsiloxane chamber with a volume of 700 mL was adhered onto the MoS 2 nanoflake-deposited gold chip. Afterward, 300 mL DI water was added into the chamber. When the MoS 2 nanoflakes were illuminated at the resonant angle under the optical microscope, an unprecedented etching phenomenon in the multilayered MoS 2 nanoflakes occurred. The etching degree of the MoS 2 nanoflakes strongly depended on the red-light power density for exciting SPPs, and we first adopted a power density of 1.7 mW$mm À2 to fabricate trilayer MoS 2 . To obtain bilayer and monolayer MoS 2 , the red-light power density was tuned to 3.4 and 6.8 mW$mm À2 , respectively. Successive optical images were recorded at 1-2 min intervals. To block electron transfer between the Au film and MoS 2 nanoflakes, the gold-coated coverslip was modified by evaporating 3 nm Si 3 N 4 as an insulating layer. Unless indicated otherwise, the red-light power density for exciting SPPs was set at 1.7 mW$mm À2 in general when we conducted control experiments.</p><!><p>Raman spectra were obtained using a Raman spectroscope (Horiba, LabRAM HR Evolution) with a laser excitation wavelength of 532 nm. AFM was performed using a Bruker Dimension ICON system in tapping mode. High-resolution transmission electron microscopy, high-angle annular dark-field scanning transmission electron microscopy, and corresponding energy dispersive spectroscopic mapping analyses were performed on a FEI Talos F200X electron microscope.</p><!><p>Finite element simulations Three-dimensional finite element simulations were performed with commercial COMSOL Multiphysics software. A gold film is sandwiched between water and glass layers, with a piece of MoS 2 above the gold film serving as a scatter. Surface plasmons at the interface between the gold film and water are excited by a p-polarized planar polariton (wavelength, 660 nm) incident from the glass layer. To better simulate the experimental conditions and avoid reflection owing to truncation of the simulation space, the dimensions of the gold film are set as 4 mm34 mm347 nm with the scattering boundary condition, whereas the size of MoS 54 and glass, respectively. The three-layer structure (water-gold-glass) supports two surface plasmonic modes with propagation constants of K/K 0 = 1.66 and 1.42, according to the 2D dispersion relationship. 55 The latter mode has a longer propagation length and a more confined electric field on the water side, which is the focus of the simulation.</p><!><p>Our ab initio calculation method employed the projector-augmented wave method 56 as implemented in the Vienna ab initio simulation package. 57,58 The Perdew-Burke-Ernzerhof 59 parametrization of the generalized gradient approximation was used for the exchange-correlation potential. For the optimization of the lattice constants, a planewave cutoff of 500 eV and a Monkhorst-Pack k point grid of 21 3 21 3 7 were used. The energy and force were converged with the threshold of 10 À6 eV and 10 À3 eV/A, respectively. The resulting lattice constants are a = 3.189 A, c = 12.405 A, which is in good agreement with the experimental ones. 60 To mimic the doping effect, a slab of 6 3 6 supercell containing two layers of MoS 2 with AB stacking was used. The Brillouin zone was sampled at the Gamma-point only. The hole doping was simulated by removing electrons from the slab whereas a homogeneous negative background charge is added to keep the charge neutrality of the slab.</p><!><p>Supplemental information can be found online at https://doi.org/10.1016/j.chempr. 2021.03.011.</p>

Chem Cell

Thymine DNA glycosylase recognizes the geometry alteration of minor grooves induced by 5-formylcytosine and 5-carboxylcytosine

The dynamic DNA methylation-demethylation process plays critical roles in gene expression control and cell development. The oxidation derivatives of 5-methylcytosine (5mC) generated by Tet dioxygenases in the demethylation pathway, namely 5-hydroxymethylcytosine (5hmC), 5-formylcytosine (5fC), and 5carboxylcytosine (5caC), could impact biological functions by altering DNA properties or recognition by potential reader proteins. Hence, in addition to the fifth base 5mC, 5hmC, 5fC, and 5caC have been considered as the sixth, seventh, and eighth bases of the genome. How these modifications would alter DNA and be specifically recognized remain unclear, however. Here we report that formyl-and carboxylmodifications on cytosine induce the geometry alteration of the DNA minor groove by solving two highresolution structures of a dsDNA decamer containing fully symmetric 5fC and 5caC. The alterations are recognized distinctively by thymine DNA glycosylase TDG via its finger residue R275, followed by subsequent preferential base excision and DNA repair. These observations suggest a mechanism by which reader proteins distinguish highly similar cytosine modifications for potential differential demethylation in order to achieve downstream biological functions.

thymine_dna_glycosylase_recognizes_the_geometry_alteration_of_minor_grooves_induced_by_5-formylcytos

Introduction<!>Results<!>TDG selectively recognizes 5fC and 5caC through the nger residue Arg275<!>Discussion<!>Synthesis and purication of modied DNA oligonucleotides<!>DNA crystallization, diffraction data collection, and structure determination<!>Modeling, structural parameter calculation, and MD simulation<!>TDG expression and purication<!>DNA glycosylase activity assay<!>Single turnover kinetics assay<!>Accession codes

<p>In mammals DNA methylation and demethylation at the C5 position of cytosine is a dynamic process which is critical for cell fate reprogramming and development. 1,2 Aberrant methylation occurring in the human genome leads to numerous diseases and cancers, such as myelodysplastic syndromes (MDS) and acute myeloid leukemia (AML). 3 Thus, this dynamic process has been considered as an important factor for studies of tumorigenesis mechanisms and the discovery of therapeutic targets. In the methylation pathway, cytosine (C) is converted to 5-methylcytosine (5mC) by DNA methyltransferases (DNMTs), which interferes with the recognition of transcriptional factors and silences gene expression. 4 While in the demethylation pathway, 5mC is oxidized to 5-hydroxymethylcytosine (5hmC), 5-formylcytosine (5fC) and 5-carboxylcytosine (5caC) in a stepwise manner by ten-eleven-translocation proteins (Tets). 3,5 Subsequently, the last two oxidization pyrimidine products 5fC and 5caC, but not 5hmC, are recognized and excised by thymine DNA glycosylase (TDG) to form an apyrimidinic site (AP), followed by base-excision repair (BER) to revert to unmodied cytosine. 6 Researchers established the DNA methylation pathway several decades ago. The indispensable biological roles of 5mC in fundamental processes such as genomic imprinting, X chromosome inactivation, suppression of transposable elements, and tumorigenesis have been broadly discussed. 1,4 In contrast, the biological function of the active DNA demethylation pathway has remained unclear.</p><p>Among the four cytosine modications (5mC, 5hmC, 5fC, and 5caC) involved in the demethylation pathway, the latter three were newly discovered in the human genome in 2009 (5hmC) and 2011 (5fC and 5caC). 7,8 In contrast to the distinguished gene suppression effect of 5mC in the genome, 5hmC, 5fC, and 5caC appeared to be intermediate products without any clear independent biological function. Along with the rapid development of sequencing and nucleic acid modication detection technologies against cytosine modications, these products are found to have tissue-specic distributions and may play functional roles in regulating the different stages of embryonic development or other potential biological processes. [9][10][11][12][13][14] For instance, 5hmC was found to be highly enriched in the genome of the brain and central nervous system, at a level nearly equal to that of 5mC. 15 5hmC recruits specic reader proteins such as MeCP2 to activate neuronal function-related gene expression by inducing the loss of H3K27me3. [16][17][18][19] Compared with numerous studies on the distribution and functions of 5hmC, there are few reports on the biological function of 5fC and 5caC. Unlike the stable geometry of 5hmC in dsDNA, 5fC was reported to alter the tertiary structure of dsDNA and facilitate its transformation from B-form to Z-form, or even to F-form. 18,20,21 Recent research further showed that 5fC is enriched in early embryos at approximately double the level in the parental genome aer fertilization, while 5caC is increased in breast cancers and gliomas and affects the transcriptional rate and specicity of RNA polymerase II. 14,22,23 These discoveries suggest additional roles of 5fC and 5caC in altering duplex DNA properties, in addition to functioning as intermediates of the oxidative demethylation pathway.</p><p>To date, thymine DNA glycosylase (TDG) has been the only functional "reader" protein to recognize 5fC and 5caC modications and achieve downstream biological functions. 6,24 TDG belongs to the uracil DNA glycosylase (UDG) superfamily, and is thought to be involved in the dsDNA repair of G$T and G$U mismatch in the mammalian genome for decades. 25 When repairing, TDG uses a "nger residue" Arg275 to recognize and ip out the damaged/aberrant base into the active site from mismatch-containing dsDNA, and performs subsequent cleavage of the glycosylic bond of these base lesions. [26][27][28] The mutagenesis of R275 (R275L) signicantly reduces the glycosylase activity of TDG, 29 which may be related to lung adenocarcinoma tumorigenesis (TCGA library data). In 2011, He et al. showed that TDG catalyzed the base excision of 5fC and 5caC, thus acknowledging TDG as a key factor to complete the DNA demethylation pathway. 6,27 Furthermore, TDG shows distinctive substrate preference for 5fC compared with 5caC with an unknown mechanism. 30,31 Here we have solved the crystal structures of a dsDNA decamer containing fully symmetric 5mC, 5hmC, 5fC or 5caC. Structural analysis suggests that 5fC and 5caC induce distinct geometry alteration of the dsDNA minor groove, which is subsequently recognized by the nger residue Arg275 of TDG. The mutagenesis of Arg275 to alanine (R275A) abolishes the excision of 5fC with a minor effect on the 5caC excision. Our ndings thus reveal the mechanism of substrate preference of TDG with regard to C, 5mC, 5hmC, 5fC, and 5caC in the demethylation pathway, providing insights into the principle on how reader proteins distinguish these highly similar cytosine modications in order to achieve downstream biological functions.</p><!><p>Overall structures of 5mC, 5hmC, 5fC and 5caC containing duplex DNA In order to investigate the structural characteristics of dsDNA modied by 5mC, 5hmC, 5fC and 5caC, a 10-bp self- complementary DNA decamer (5 0 -CCAGXGCTGG-3 0 , X refers to 5mC, 5hmC, 5fC or 5caC) was synthesized by using solid-phase synthesis. Since +4 to À4 base pairs aside from the central substrate modication are required for TDG recognition, 30,31 the modied base was set at the central position of the decamer. The oligo was further annealed, generating a 10bp decamer dsDNA containing a fully symmetric 5mC$G, 5hmC$G, 5fC$G or 5caC$G modication at the central CpG site. The crystal screening was performed using the hanging drop method at 4 C, and square-like crystals appeared under acidic conditions in 4 to 6 days. The 5mC-dsDNA, 5hmC-dsDNA, 5fC-dsDNA and 5caC-dsDNA structures were determined under resolution 1.40 Å, 2.85 Å, 1.56 Å and 1.06 Å with the space group C2, C2, P6 1 and P2 1 , respectively. The structures of 5mC-dsDNA and 5hmC-dsDNA were solved by using the molecular replacement method (MR) with a published dsDNA structure (pdb code: 1EN9; sequence: 5 0 -CCAGCGCTGG-3 0 ) as the search model, 32 while the structures of 5fC-dsDNA and 5caC-dsDNA were solved by using a direct method due to their extremely high resolutions. 33 All the structures were subsequently rened by using the maximumlikelihood renement method carried out with the Phenix soware package. 34 Table 1 and ESI Table S1 † summarize the data collection and structure renement statistics. Structural analysis suggests that the 5mC-dsDNA, 5fC-dsDNA and 5caC-dsDNA structures exhibit a single noncanonical right-hand double helix pattern. Interestingly, 5hmC-dsDNA displays two non-canonical right-hand double helixes in an asymmetric unit, where one strand is complemented with half of the other two stands at the same time, exhibiting a nonstandard "X" base paring pattern. Within the double stranded helix of the structures, 5mC-dsDNA, 5hmC-dsDNA and 5caC-dsDNA helixes exhibit B-form dsDNA patterns, which are similar to the published B-form 5C-dsDNA structure (pdb code: 1EN9), 32 while the 5fC-dsDNA helix exhibits an A-form pattern. All the modication groups point towards the major groove as expected. The C5 atoms from the modication groups in each structures are 7.4 Å (5mC), 7.7 Å/7.0 Å (5hmC), 10.4 Å (5fC) or 7.2 Å (5caC) away from each other (distance between C5m atoms) (Fig. 1 and ESI Fig. S1 and S2 †).</p><p>Further local rotational and translational base-step parameter analysis including the slide displacement and roll angle indicates structural variations among these crystal structures (Fig. 2a and c). Remarkably, we observed distinct structural alteration in 5fC and 5caC containing dsDNA. Regarding 5fC containing dsDNA, $2 Å shi displacement alteration at the 5fC$G site occurs compared with the adjacent base pairs, leading to a 3-5 Å opening in the major groove width and 2-3 Å narrowing in the minor groove (Fig. 2b, d and e). It is noteworthy that such alterations have been observed in previously reported 5fC-containing DNA structures under different crystallization conditions suggesting that 5fC modication specically contributes to DNA structure conformational alteration through inducing groove geometry alteration in the helix. 21 Within the 5fC$G base pair, the formyl-group does not interfere with the regular G$C base paring sterically, and the sp 2 hybridization plane of the formyl group lies in the same plane as the pyrimidine ring (Fig. 1b). The O5 atom of the formyl group forms an intramolecular hydrogen bond with the exocyclic N4 amino group of the pyrimidine, and locks the rotation of the O5-C5m-C5-C6 dihedral angle of the formyl group (Fig. 3c). Moreover, the formyl groups are stabilized at the right place via hydrogen bond networks in the major groove. In the center of the networks, a carbonate ion source from the DNA purication reagent triethylammonium bicarbonate (TEAB) mediates the crosstalk of the two fully symmetric formyl groups as well as the interactions among the formyl groups, water molecules, and the phosphodiester backbones as a hub. Similar to 5fC containing dsDNA, a $2 Å shi displacement alteration occurs at the 5caC$G site compared with the adjacent base pairs in 5caC containing dsDNA, suggesting that the negative charges at the C5m position of the cytosine pyrimidine ring could induce similar geometry alteration of dsDNA (Fig. 2b). Unexpectedly, further structural alteration in the major and minor grooves is not observed, thus indicating that 5caC has distinctive contribution to dsDNA structural alteration compared with 5fC (Fig. 2d and e). Within the 5caC$G base paring, the carboxyl group of 5caC does not interfere with the regular G$C base paring sterically, and the sp 2 hybridization plane of the carboxyl group lies in the same plane as the pyrimidine ring as well (Fig. 1d). Due to the signicantly shorter distance between two fully symmetric carboxyl groups in dsDNA, a bridging water molecule (W1) directly connects the two carboxyl groups, mediating the crosstalk between two 5caC instead of the carbonate ion in the 5fC-DNA structure. What is more, the carboxyl group is further locked by the intramolecular hydrogen bond with the exocyclic N4 amino group of the pyrimidine, and hydrogen bonds with the phosphodiester backbone are mediated by one (W2) or two (W3 and W4) water molecules (Fig. 1d).</p><p>Effects of C, 5mC, 5hmC, 5fC, and 5caC in inducing the geometry alteration of dsDNA 5hmC, 5fC, and 5caC generated by TET dioxygenases from 5mC share highly similar chemical structures. However, TDG accurately distinguishes these derivatives by excluding C, 5mC, and 5hmC and catalyzing the removal of 5fC signicantly faster than that of 5caC. 30,31 Density functional theory (DFT) analysis against the O5/H-C5m-C5-C6 dihedral angle suggests that the methyl group in the 5mC-dsDNA structures adopts the most stable conformation in the minimum energy state at À120 , 0 or 120 , while formyl and carboxyl groups adopt the most stable conformation in the minimum energy state at À180 or 180 and À180 , 0 or 180 , respectively (Fig. 3). Dihedral angle rotation of formyl and carboxyl groups along the C5m-C5 bond from 0 to 90 leads to an increase of potential energy and instability of the structures. The maximum energy state is reached at 90 , and subsequently decreases from 90 to 180 . At 180 , the carboxyl group returns to the most stable conformation in the minimum energy state. The energy of the formyl group at 0 is signicantly higher than that at 180 due to the loss of the intramolecular hydrogen bond with the exocyclic N4 amino group of the pyrimidine, suggesting a critical role of the intramolecular hydrogen bonding in stabilizing the formyl-and carboxylgroup conformation in the structure. What is more, the energy state landscape of 5hmC is totally different from that of 5mC, 5fC and 5caC. The minimum energy state of the 5hmC dihedral angle is around 136 or À133 rather than 0 or 180 in 5fC and 5caC structures and À120 , 0 or 120 in the 5mC structure. This is most likely caused by the sp 3 hybridization of the hydroxymethyl group, and such a distinctive orientation and energy state of 5hmC potentially leads to the unrecognition of TDG on 5hmC for catalysis.</p><p>In order to investigate the effects of the cytosine modications on dsDNA conformation, we employed a scrupulous method to dene the force eld and perform further molecular dynamic simulation according to our previous study 35,36 (see the method section), which will eliminate the interference from the DNA sequence and crystallization condition variations. As shown in Fig. 4, aer 400 ns simulation, the overall structures of 5C-, 5mC-, 5hmC-, 5fC-and 5caC-dsDNA tend to be canonical Bform dsDNA, conrming that the A-form conformation observed from the 5fC containing structure was caused by crystal packing rather than the variation of the cytosine modication.</p><p>The slide displacement analysis showed that all the modications have similar effects except that the slide displacements of 5mC containing dsDNA slightly decrease by 0.4 Å, while the roll angle of 5caC increases remarkably by 5 , leading to more bended structural conformations of 5caC-dsDNA, especially at the modication site compared with that of the canonical Bform dsDNA (Fig. 4a-c).</p><p>Further geometry analysis against major and minor grooves shows that the major grooves in 5mC-and 5hmC-dsDNA are signicantly opened by $1 Å and $0.7 Å compared to that of 5fC-and 5caC-dsDNA respectively, leading to a relatively narrower major groove in 5fC-and 5caC-dsDNA structures, and mildly looser major grooves in 5mC-and 5hmC-dsDNA structures (Fig. 4d). Moreover, the minor groove of 5caC-dsDNA is signicantly enlarged by $1 Å compared to other modication containing structures, which has not been observed previously, suggesting distinctive contribution of 5caC to minor groove geometry alteration (Fig. 4e). By analyzing the electrostatic potential energy of the minor groove, 5fC and 5caC are found to be in a less stable state in base pairing due to the repulsive force produced by the same negative charges of the carbonyl oxygen atom in 5fC or the carboxyl group in 5caC and the phosphodiester backbone (Fig. 4f). However, the repulsive force in 5caC-dsDNA squeezes the 5caC base aside from the phosphate backbone, resulting in an opposite and more stable electric potential energy at the midpoint of C5-P (phosphate in the backbone) with 5fC (Fig. 4g). Together with these observations, the 5fC and 5caC modications are found to variously inuence the geometry alteration of the dsDNA major and minor grooves. The various conformational alterations are predominantly induced by the distinctive repulsive force between the formyl or carboxyl group and the phosphodiester backbone, which would benet the preferential recognition and catalysis of 5fC and 5caC by various reader proteins, such as TDG.</p><!><p>To date, TDG has been the only enzyme discovered for the enzymatic removal of 5fC and 5caC from dsDNA with marked catalytic specicity. Structural analysis of the published TDG-dsDNA structure (PDB code: 3UO7) indicates that TDG recognizes DNA predominantly via two loops (Fig. S2 †). 30,31 Loop1 between residues 199-202 (GSKD) inserts into the major groove of the dsDNA. While loop2 between residues 274-277 (ARCA) inserts into the minor groove of the dsDNA, where the nger residue Arg275 penetrates into the DNA helix and pushes the modied base into the active pocket of TDG for catalysis. Considering the eminent inuences of 5fC and 5caC on dsDNA groove geometry, polar residues in these two loops were mutated (S200A, K201A, and D202A in loop1, and R275A in loop2) in order to evaluate their potential roles in recognizing the conformational alteration of dsDNA induced by 5fC and 5caC. Surprisingly, only the mutation of Arg275 (R275A) on loop2 exhibits signicant inuence on 5fC/5caC selection through glycosylase activity assay. As shown in Fig. 5a, the wildtype enzyme catalyzed 5fC and 5caC excision completely in a 30 min reaction, while the R275A mutation preferentially abolished the excision of 5fC compared with the partial excision of 5caC in dsDNA. A single turnover kinetics assay further conrms this result. As shown in Table 2, mutagenesis of residues on loop1 (S200A, K201A, and D202A) reduces the maximal catalytic activity (K max ) in catalyzing 5fC and 5caC excision by 1.1 to 4.7 fold as expected, 28,29 suggesting that loop1 unlikely contributes to selective recognition and catalysis of substrates. However, the R275A mutation in loop2 dramatically decreases the catalysis activity against 5fC by 47 fold versus to 4.8 fold in catalyzing 5caC, resulting in a nearly ten times slower catalysis rate of TDG for 5fC removal compared with that of 5caC (Fig. 5b). These observations indicate that TDG makes substrate selection against 5fC and 5caC by recognizing the geometry alteration of the dsDNA minor groove by its nger residue Arg275 on loop2. Notably, a sharp activity decrease of TDG R275A in catalyzing uracil was also observed, suggesting that Arg275 can specically recognize uracil besides 5fC (Fig. 5a, ESI Table S2 †). As the uracil does not have a positive charge modication on the C5 position of the pyrimidine ring, such recognition could be due to the wobble guanine-uracil base pairing, which does not follow Watson-Crick base pairing rules. 37 Our results here show that TDG R275 is a key residue that recognizes the distinctive geometry alterations in the minor groove of dsDNA induced by modied bases such as 5fC and 5caC, and ips the selected substrate base into the active pocket for catalysis to achieve independent downstream biological functions. Our results provide a potential mechanism by which reader proteins distinguish these highly similar 5-substituents of cytosine. This selective recognition mechanism further provides novel insights into the biological functions of these epigenetic modications in the active DNA demethylation pathway, as well as in development and tumorigenesis.</p><!><p>Our studies have revealed that methyl-, hydroxyl-, formyl-, and carboxyl-group modications on cytosine have distinctive inuences on the geometry of dsDNA which potentially allow for the preferential recognition and catalysis among these highly similar derivatives by various reader proteins, such as TDG. Since full methylation on a chromosome has been well known as a regular epigenetic event, all the oxidative modications (5hmC, 5fC and 5caC) generated from the oxidation of fully methylated cytosine have been detected by numerous whole genome sequencing studies (e.g. TAB seq for 5hmC and fCAB-seq for 5fC). 11,38 This indicated that fully symmetric 5hmC, 5fC or 5caC sites in the genome do have a signicant independent biological meaning. We synthesized and crystallized 10-bp self-complementary DNA decamers containing a fully symmetric 5mC$G, 5hmC$G, 5fC$G or 5caC$G modied CpG site in the middle position, and analyzed the conformation alteration by DFT computation and molecular dynamics simulations. The sp 2 hybridized formyl and carboxyl groups adopt plane conformations, and the ketone from the formyl and carboxyl groups forms an intramolecular hydrogen bond with the exocyclic N4 amino group of the pyrimidine, locking the sp 2 hybridization plane to lie in the same plane as the cytosine pyrimidine ring. In contrast, the hydroxylmethyl group adopts a distinctive non-planar sp 3 hybridization, and forms a $136 angle with the pyrimidine plane in base paring. 18 Furthermore, the negative charge carried by 5caC generates a repulsive force with the phosphodiester backbone of dsDNA, which inuences the geometry of dsDNA by predominantly widening the minor groove and loosening the dsDNA structures compared with canonical B-form dsDNA. Meanwhile, the repulsive force in 5caC-dsDNA squeezes the 5caC base aside from the phosphate backbone, causing more drastic geometry alteration in the minor groove of dsDNA and opposite electric potential energy at the midpoint of C5-P compared with 5fC.</p><p>Then, using two techniques, we investigated the key residue which could sense the geometry alteration in the minor groove, and observed that the mutagenesis of Arg275 on the loop2 of TDG to alanine (R275A) sharply decreases the activity of TDG against 5fC by 47 fold, thus suggesting a specic role of Arg275 in 5fC recognition and catalysis. However, the sharp drop of activity does not occur in 5caC catalysis to a similar degree. The activity of R275A against 5caC only decreases by 4.8 fold, which is similar to that of mutations in loop1 (S200A, K201A, and D202A), thus suggesting a non-specic recognition of Arg275 on 5caC. This difference could be explained by the deection of the 5caC base as well as the more stable status of 5caC compared to 5fC, which might weaken the role of Arg275 in 5caC recognition and ipping.</p><p>It should be noted that a previous study reported that the residue Asn157 inside the active pocket of TDG is crucial for selective 5caC excision. The mutagenesis of Asn157 to aspartic acid (N157D) signicantly reduces such preference at acidic pH, 31 while our results here suggest that the substrate preferential selection of TDG occurs prior to the ipping of the base from the dsDNA helix. Our results further suggest that TDG N157D and R275A double mutation would abolish the base excision of 5fC by either blocking the base ipping or blocking the catalysis of the base excision, which would cause the accumulation of 5fC in the genome. Hence, this TDG double mutant could be used as a chemical biology tool to further study the biological function of 5fC in the differentiation of embryo stem cells in vivo.</p><p>The mammalian DNA demethylation is a critical process for cell development, and how enzymes involved in this process recognize the highly similar chemical structures of the cytosine derivatives is one of the key scientic questions in understanding the biological function of this process. A recent study reported that ten-eleven-translocation proteins (Tets) and TDG physically interact with one another to avoid DNA double strand breaks (DSB) in the demethylation. 39 Our current results support their observation as we show that the oxidative cytosine derivatives such as 5fC and 5caC induce signicant geometry alteration in the DNA minor groove, which may decrease the stability of the genome, leading to DSB. The way to avoid such a possibility is that the two enzymes work together, and no free 5fC and 5caC site is exposed.</p><p>All these results provide novel insights into the mechanisms of cytosine modications in achieving independent downstream biological functions, which will in turn facilitate the understanding of biological functions of these epigenetic modications in active DNA demethylation pathways, as well as in development and tumorigenesis.</p><!><p>The oligonucleotides with sequences (5 0 -CCA GXG CTG G-3 0 and 5 0 -ATA GAA GAA TTC XGT TCC AG-3 0 , X refers to 5mC, 5hmC, 5fC or 5caC) used in crystallization and biochemistry studies were synthesized by using solid-phase synthesis. The synthesized oligonucleotides were puried by using high-performance liquid chromatography (reverse-phase C18 column) and lyophilized as previously reported. 40 The oligonucleotides containing normal bases (5 0 -CTG GAA CGG AAT TCT TCT AT-3 0 ) were purchased from Sangon Company. The oligonucleotides were subsequently dissolved in buffer containing 20 mM HEPES, pH 7.0, and 100 mM NaCl, and heated to 100 C for slow annealing.</p><!><p>Crystallization of 5mC, 5hmC, 5fC or 5caC containing DNA decamers was performed by using a sitting drop vapor diffusion method. 1 mL of 2 mM 5mC-, 5hmC-, 5fC-or 5caC-dsDNA sample was mixed with an equal volume of reservoir solution and P2 1 by using HKL3000, respectively. The phases of 5mC-dsDNA and 5hmC-dsDNA were determined by using the molecular replacement method (MR) with a published dsDNA structure (pdb code: 1EN9; sequence: 5 0 -CCAGCGCTGG-3 0 ) as the search model. 32 The phases of 5fC-dsDNA and 5caC-dsDNA were determined by using a direct method with the CCP4 suite, 33 followed by subsequent maximum-likelihood renement with the Phenix soware package. 34 The electron density-based model building was performed using the computer graphics program Coot, 41 and the nal structures were visualized by using PyMol soware. 42 Table 1 and ESI Table S1 † summarize the data collection and structure renement statistics.</p><!><p>Beside the determined crystal structures of 5mC-dsDNA, 5hmC-dsDNA, 5fC-dsDNA and 5caC-dsDNA, the 5C-dsDNA structure was obtained from the published crystal structure (PDB code: 1EN9), while canonical A-form and B-form dsDNA structures with the same sequence were modeled by using Maestro soware.</p><p>The DNA rigid-body parameters were calculated and analyzed by using the ensemble scripts provided by the 3DNA suit. 43 The data were processed using g-analyze from the Gromacs analysis tool. The vertical bars represent the error estimate of the average values calculated using the blocking method. All relaxed potential energy surface scans were carried out with the Gaussian 09 soware package at the M062x/6-31+g(d,p) level. 44,45 The dihedral torsion angle of O5/ H-C5m-C5-C6 in the modied deoxycytidines was scanned every 30 degrees with geometry optimization at each conformation. For MD simulation, the RESP charges of 5mC, 5hmC, 5fC and 5caC were calculated with Gaussian 09 soware at the HF/6-31(d,p) level and the force eld parameters were produced using an antechamber based on previous reports. 35,36 The force eld for DNA was ff-nucleic-OL15. 46 The solvent effects were involved using the polarizable continuum model with water as the solvent. 47 Tleap was used to model all simulation systems, namely 5C-dsDNA, 5mC-dsDNA, 5hmC-dsDNA, 5fC-dsDNA, 5caC-dsDNA, A-form dsDNA, and B-form dsDNA. In all simulation systems, the DNA structure was submerged in explicit TIP3P water in cubic boxes with an extra 10 Å extension along each axis of the DNA. The net charge of the system was neutralized by adding a suitable number of counterions. Amber topology and coordinates les were converted into Gromacs format by using Acpype. All MD simulations herein were performed with the Gromacs package (version 5.1). The particle mesh Ewald (PME) method was applied to handle the long-range electrostatics. 48 Nonbonded van der Waals forces and short-range electrostatic interactions between atoms were truncated at 10 Å. Periodic boundary conditions (PBCs) were used during the MD simulations. All the MD simulations were performed at 300 K and 1 atm. The LINCS was used to constrain the length of the hydrogen bonds, allowing the movements integrated numerically with a time step of 2 fs algorithm. 49 The starting structure of each model was energy-minimized by using the steepest-descent algorithm to remove unfavorable steric clashes. 50 Coordinates of each model were saved every 10 ps throughout the 400 ns production runs. The DNA structures were extracted from MD trajectories every 200 ps for further calculations. The Delphi program was used for electrostatic potential calculations. 51 The partial charge and atom radii were taken from the topology le used during MD simulations. The salt concentration was 0.145 M and a 1.4 Å probe sphere was applied for the calculation of the solute molecule surface. The interior of the solute molecule was assigned an internal dielectric constant of 2 whereas exterior regions were assigned a dielectric constant of 80. The size of the cubical grid was set to 165. The minor groove electrostatic potential and the electrostatic potential of the geometric midpoint between the C5 atom and P atom were calculated based on a previous report. 52,53</p><!><p>The catalytic domain of human TDG was expressed and puri-ed as previously reported. 30 Briey, the cDNA for the catalytic domain of human TDG (residues 111-308) was subcloned into a pMCGS19 plasmid and expressed in BL21 (DE3) cells containing the vector pRK1037. The cultures were grown at 37 C until OD600 reached 0.6, and then induced at 25 C with 1 mM isopropyl b-D-1-thiogalactopyranoside (IPTG) overnight. Subsequently, the cells were harvested and re-suspended with a lysis buffer containing 20 mM Tris, pH 7.4, 500 mM NaCl, 20 mM imidazole, 1 mM dithiothreitol (DTT) and 0.25 mM phenylmethylsulphonyl uoride (PMSF), and lysed using a French press. The lysate was centrifuged at 13 000g for 40 min, and the supernatant was used for further purication via loading onto an affinity column (Ni-NTA), ion-exchange column (HiTrap SP column), and gel-ltration column (16/60 Superdex 75). The puried protein was concentrated and quantied by using the Bradford reagent (Bio-Rad), ash-frozen, and stored at À80 C.</p><!><p>A 20-mer 5fC-or 5caC-containing strand was labeled with g-32 P-ATP by incubation with T4 DNA polynucleotide kinase (T4 PNK, NEB) at 37 C for 1 h (A 17-mer labeled strand was used as a control). Subsequently, the labeled oligonucleotide was annealed with a complementary strand and the duplex was puried for the glycosylase activity assay. The reactions were performed with 100 nM TDG and 10 nM DNA substrates ( 32 P-5 0 -ATA GAA GAA TTC C*GT TCC AG-3 0 and 5 0 -CTG GAA CGG AAT TCT TCT AT-3 0 ) at 22 C in the reaction buffer containing 25 mM HEPES, pH 7.4, 0.5 mM EDTA, 0.5 mg mL À1 BSA, and 0.5 mM DTT. Reactions were quenched by adding 1 M NaOH and 100 mM EDTA and incubating at 100 C for 5 min to break the DNA strand containing an abasic site. The samples were cooled down and loaded into a denaturing gel for electrophoresis.</p><!><p>Single turnover kinetics assay under saturating enzyme conditions was carried out to determine the rate constant (K max ) of TDG. The reaction was performed with 5 mM TDG and 0.5 mM DNA substrates at 22 C in the reaction buffer (25 mM HEPES, pH 7.4, 0.5 mM EDTA, 0.5 mg mL À1 BSA, and 0.5 mM DTT) and was quenched at specic time points with 50% (v : v) 0.1 M NaOH and 0.01 M EDTA. The samples were further boiled for 15 min at 85 C (5fC and U were quenched with 0.3 M piperidine and 0.03 M EDTA and incubated at 85 C for 15 min) and then cooled down for HPLC analysis. The products and reactants from different time points were separated and quantied by anion-exchange HPLC using reported denaturing conditions with a DNAPac PA200 column (Dionex). The oligonucleotides are detected by absorbance (260 nm), and the fraction product (F) is determined from the integrated peak areas for the product strands (A P1 and A P2 ) and target strand (A S ) using the equation: F ¼ (A P1 + A P2 )/(A P1 + A P2 + A S ), and the single turnover rate constant K max is calculated by using the equation: F ¼ A[1 À exp(ÀK max t)], in which A is the fraction of the substrate converted to the product at completion and t is the reaction time.</p><!><p>PDB: the atomic coordinates and structure factors for the reported crystal structures are deposited under accession codes 6JV5 (5mC-dsDNA), 6JV3 (5hmC-dsDNA), 5ZAS (5fC-dsDNA) and 5ZAT (5caC-dsDNA).</p>

Royal Society of Chemistry (RSC)

Action and Timing of BacC and BacD in the Late Stages of Biosynthesis of the Dipeptide Antibiotic Bacilysin

Biosynthesis of the dipeptide antibiotic bacilysin, encoded by the seven B. subtilis genes bacA-G, involves diversion of flux from prephenate to the noncognate amino acid anticapsin. The anticapsin warhead is then ligated to the C-terminus of l-alanine to produce mature bacilysin. We have previously noted the formation of two diastereomers of tetrahydrotyrosine (4S- and 4R-H4Tyr) by tandem action of the four purified enzymes BacABGF. BacC (oxidase) and BacD (ligase) have been hypothesized to be remaining late stage enzymes in bacilysin biosynthesis. Using a combination of BacCD in vitro studies, B. subtilis deletion mutants, and isotopic feeding studies, we were able to determine that the H4Tyr diastereomers are actually shunt products that are not on-pathway to bacilysin biosynthesis. Dihydroanticapsin and dihydrobacilysin accumulate in extracts of a \xce\x94bacC strain and are processed to anticapsin and then bacilysin on addition of BacC and BacD, respectively. These results suggest the epoxide group in bacilysin is installed in an earlier step of bacilysin biosynthesis, while BacC oxidation of the C7-hydroxyl followed by BacD ligation of anticapsin to l-Ala are the penultimate and ultimate steps of bacilysin biosynthesis.

action_and_timing_of_bacc_and_bacd_in_the_late_stages_of_biosynthesis_of_the_dipeptide_antibiotic_ba

INTRODUCTION<!>Materials and Instrumentation<!>Cloning, Expression, and Purification of the Enzymes BacABCDFG<!>Construction of \xce\x94bacB and \xce\x94bacCB. subtilis PY79 genomic deletion mutants<!>Production and LC/MS analysis of bacilysin and associated intermediates from B. subtilis PY79 strains<!>Production and purification of dihydroanticapsin hydrate for NMR analysis<!>BacC and BacD reactions with 4S-and 4R-H4Tyr<!>BacC and BacD reactions with B. subtilis PY79 \xce\x94bacC lyophilized extracts<!>Feeding studies with isotopically labeled [13C]-H4Tyr and [2H]-ex-H2HPP diastereomers<!>Evidence for 5,6-epoxy-7-hydroxycyclohexyl-Ala (dihydroanticapsin) in B. subtilis \xce\x94bacC extracts<!>Purified BacC oxidizes 2S, 4S, 7R- but not 2S, 4R, 7R-H4Tyr<!>Action of BacD as an L-Ala-L-amino acid dipeptide ligase<!>Tandem Action of BacC and BacD on 4S-H4Tyr to yield the oxidized dipeptide L-Ala-4S-cyclohexenonyl-Ala<!>BacC converts dihydroanticapsin to anticapsin in B. subtilis \xce\x94bacC fermentation extracts<!>Feeding studies with isotopically labeled [13C]-H4Tyr and [2H]-H2HPP<!>Proposed Biosynthetic Pathway to Anticapsin and Bacilysin

<p>Bacilysin, the dipeptide of l-Ala ligated to the N-terminus of anticapsin (epoxycyclohexanonyl-Ala) (Figure 1A), is a Trojan Horse antibiotic excreted by some Bacillus species.1, 2 Once bacilysin is transported into a neighboring cell, anticapsin can be freed by peptidase action. Free anticapsin inside a susceptible bacterial or fungal cell is a time-dependent, irreversible inactivator of the glutaminase domain of glucosamine synthetase3, 4 (Figure S1). The blockade of GlcNAc formation leads to interdiction of cell wall biosynthesis and subsequent cell demise.</p><p>The seven-gene region bacA-G in producing Bacillus strains has been identified by genetic studies1, 5 as the biosynthetic gene cluster responsible for bacilysin production (Figure 1, inset). In biochemical studies, we have previously established that BacABGF are four enzymes that act in tandem to divert some of the flux of prephenate away from production of l-Phe and l-Tyr into a four-step pathway leading to both 2S, 4S, 7R- and 2S, 4R, 7R-tetrahydrotyrosine (H4Tyr) diastereomers6–9 (Figure 1B). BacE is a proposed dipeptide permease involved in host resistance by pumping bacilysin from the intracellular to the extracellular environment1. In this work we evaluate the remaining two proteins BacC and BacD for their roles in the late stages of the bacilysin assembly pathway.</p><p>Bioinformatic analysis predicts BacC should be in the NAD+-dependent oxidoreductase family, as is BacG. We have recently shown that BacG acts to reduce the 3E- and 3Z-ex-dihydrohydroxyphenylpyruvate (H2HPP) geometric isomers to the 2S, 4R, 7R- and 2S, 4S, 7R-diastereomers of H4HPP, respectively, via conjugate hydride addition8 (Figures S2A and S2B). Another obvious redox step in the formation of anticapsin is oxidation of the C7-hydroxyl group found in prephenate to the ketone moiety found in anticapsin. We show here that purified BacC with NAD+ cofactor will dehydrogenate the hydroxyl group in the 4S- but not the 4R-H4Tyr diastereomer, consistent with the known 4S stereochemistry in anticapsin. Addition of purified BacD (promiscuous dipeptide ligase)10, 11, ATP, and l-Ala to such BacC incubations yields a l-Ala-l-4S-cyclohexenonyl-Ala dipeptide.</p><p>To shed light on the timing of epoxidation at the C5-C6 double bond vs. oxidation of the C7 hydroxyl to ketone vs. dipeptide ligation, we have undertaken two approaches. One was construction of a ΔbacCB. subtilis strain to characterize accumulating intermediates. The second approach involved feeding B. subtilis with isotopically labeled versions of 4S- and 4R-H4Tyr along with isotopically labeled earlier pathway intermediates, 3Z- and 3E-ex-H2HPP. Of these four compounds, only feeding with 3E-ex-H2HPP yields isotopically labeled bacilysin/anticapsin, suggesting both the H4Tyr diastereomers and 3Z-H2HPP may be in vitro shunt products in the absence of the (still unidentified) expoxidase. Additionally, we conclude that C5-C6 double-bond epoxidation occurs before C7 hydroxyl oxidation to the ketone, which is immediately followed by dipeptide ligation to form mature bacilysin.</p><!><p>Prephenic acid barium salt, β-nicotinamide adenine dinucleotide hydrate (NAD+), β-nicotinamide adenine dinucleotide reduced disodium salt hydrate (NADH), l-phenylalanine, l-alanine, adenosine 5′-triphosphate disodium salt, and Dowex resin were purchased from Sigma-Aldrich. Deuterium oxide (99.9%) (D2O) and 3-[13C]-l-alanine (99%) was purchased from Cambridge Isotope Laboratories. DNA oligonucleotide primers were obtained without purification from Integrated DNA Technologies. Restriction endonucleases and T4 DNA ligase were obtained from New England BioLabs. Bacillus subtilis sp. 168 genomic DNA and B. subtilis sp. PY79 cells were obtained from ATCC. 1H one-dimensional and two-dimensional NMR spectra were recorded at 25 °C on a Varian VNMRS 600 MHz spectrometer equipped with a triple-resonance probe. 13C one-dimensional NMR spectra were recorded on a Varian MR 400 MHz spectrometer (100.497 MHz for 13C) equipped with a ONE NMR probe. NMR data were processed with ACD/Laboratories software. High-resolution LC/MS data were collected on an Agilent Technologies 6520 Accurate-Mass Q-TOF LC/MS system and analyzed (including integration) using the accompanying Mass Hunter Qualitative Analysis software. HPLC was performed on a Beckman Coulter System Gold instrument. UV-vis measurements were collected using a Cary 50 BIO UV-vis spectrophotometer and analyzed using the accompanying software. DNA sequencing was performed by Genewiz. Purification of plasmid DNA and PCR amplified dsDNA was performed using kits from Qiagen.</p><!><p>The cloning of plasmids for production of BacABGF has been previously described7, 8. The gene for BacC (ywfD) was amplified from B. subtilis sp. 168 genomic DNA via PCR using primers encoded with BamHI and XhoI restriction sites (5′-AATCCGGATCCATGATCATGAACCTCACC-3′ and 5′-AATTCTCGAGCTATTGTGCGGTGTATCCTCC-3′, respectively). The gene for BacD (ywfE) was amplified from B. subtilis sp. 168 genomic DNA via PCR using primers encoded with NdeI and XhoI restriction sites (5′-CGGCAGCCATATGGAGAGAAAAACAGTATTGGTCA-3′ and 5′-GGTGCTCGAGTCATACTGGCAGCACATACTTTGCC-3′, respectively). Each amplified gene was ligated into vector pET-28a (Novagen) such that the target protein would be expressed as an N-terminally tagged His6-fusion. The ligated plasmid was transformed into chemically competent TOP10 E. coli cells (Invitrogen) and proper gene insertion was confirmed by DNA sequencing of the purified plasmid DNA. The sequence-confirmed plasmid was then transformed into chemically competent BL21(DE3) E. coli cells for protein expression. Protein expression and purification of BacC and BacD was performed exactly as previously reported for BacA7. Protein purity of both BacC and BacD was judged to be > 95% from SDS-PAGE analysis on an Any kD TGX gel (BioRad) with visualization by coomassie blue staining (Figure S3). The concentration of each protein was determined by UV-vis absorbance using the following extinction coefficients (ε280nm) calculated from the protein primary sequence using the ExPASy Bioinformatics Research Portal: 18 450 M−1 cm−1 for BacC and 40 340 M−1 cm−1 for BacD. Extinction coefficients for BacABGF have been previously reported7, 8. Protein stocks were flash frozen in liquid N2 and stored at −80 °C until use.</p><!><p>Markerless ΔbacB and ΔbacC deletions were made in the B. subtilis PY79 genome with use of the pMiniMAD plasmid12 (Figure S4). Plasmid pMiniMAD-ΔbacB was constructed by PCR amplifying the 800 bp genomic regions upstream and downstream of the bacB gene from B. subtilis 168 gDNA using the primers SAM-1 with SAM-2 and SAM-3 with SAM-4, respectively (Table S1)13. The two PCR amplified dsDNA fragments were gel purified and ligated into the pMiniMAD plasmid cleaved with BamHI (3-piece ligation) using the one-step isothermal DNA assembly method of Gibson and coworkers14. The isothermal assembly mastermix used to accomplish the ligation was made in-house according to the published recipe14. Plasmid pMiniMAD-ΔbacC was constructed by PCR amplifying the 500 bp genomic regions upstream and downstream of the bacC gene from B. subtilis 168 gDNA using the primers JBP-1 with JBP-2 and JBP-3 with JBP-4, respectively (Table S1). The remainder of the pMiniMAD-ΔbacC construction was identical to the pMiniMAD-ΔbacB assembly described above. The ligated plasmids were transformed into E. coli TOP10 chemically competent cells (Invitrogen), positive transformants selected on LB-agar containing 50 μg/mL ampicillin, and plasmid DNA was purified. Sequencing of both DNA strands confirmed proper construction of the mature pMiniMAD plasmids. The sequence confirmed plasmids were individually transformed into chemically competent recA+ BL21(DE3) cells (Invitrogen) and plasmid DNA purified from these cells was used for B. subtilis transformation.</p><p>Deletion of the target genes from the B. subtilis genome was accomplished by double-crossover homologous recombination, performed in two single-crossover steps (Figure S4). The pMiniMAD plasmid containing the appropriate deletion homology (either ΔbacB or ΔbacC) was transformed into B. subtilis PY79 cells (single-crossover integration step) using a one-step competence protocol provided by the Rudner lab at HMS. To transform, a freshly streaked colony of B. subtilis was inoculated into 1 mL of MC medium (100 mM potassium phosphate pH 7.0, 3 mM sodium citrate, 2% glucose, 22 mg/mL ferric ammonium citrate, 0.1% casein hydrolysate, 0.2% potassium glutamate, and 3 mM magnesium sulfate) and grown for 4 h at 37 °C in a roller drum traveling at 60 rpm. 2 μL of pMiniMAD plasmid (at various dilutions) were mixed with 200 μL of the 4 h culture and re-incubated in the roller drum for 2 h followed by plating on LB-agar + MLS and incubation at 37 °C for 20 h. Because pMiniMAD contains an erythromycin resistance cassette, positive transformants were identified by selecting for resistance to MLS antibiotic (1 μg/mL erythromycin and 25 μg/mL lincomycin). The second-crossover was accomplished by growing a single-crossover B. subtilis transformant in LB broth (containing no antibiotics) at 21 °C (pMiniMAD contains a temperature sensitive origin of replication) for 24 h, diluting the culture into fresh LB broth, and repeating the growth/dilution process twice more. Replication initiation stimulates homologous recombination and the integrated DNA has two regions of homology. Accordingly, recombination between one of these two regions and the homologous chromosomal region will "loop-out" the plasmid in a manner that regenerates the wild-type gDNA configuration, while recombination of the other integrated homologous region with the corresponding chromosomal region will result in the second-crossover and give expulsion ("loop-out") of the pMiniMAD backbone and target gene, creating the desired deletion (Figure S4). Second-crossover ("loop out") candidate colonies (plated on LB-agar) could be initially identified by loss of resistance to MLS antibiotic. Antibiotic sensitive colonies were then screened by cPCR to confirm the deletion junctions.</p><!><p>For general bacilysin production, B. subtilis PY79 WT cells were grown in PA minimal media15. One liter of PA media in H2O contained the following components: 1.1 g potassium phosphate monobasic; 0.55 g of magnesium sulfate heptahydrate; 0.55 g of potassium chloride; 4.4 g of l-glutamic acid monosodium salt monohydrate; 13.7 g sucrose; 0.1 g trisodium citrate dihydrate; 0.1 g iron(III) chloride hexahydrate; 1 mL of oligodynamic solution16; pH to 7.0 using sodium hydroxide. The media was sterilized via filtration using a 0.22 μm PES membrane.</p><p>Small-scale bacilysin production for LC/MS analysis was accomplished by inoculating a 5 mL culture of PA media with a single colony of B. subtilis PY79 WT cells from a freshly streaked LB-agar plate. The culture was incubated at 30 °C in a roller drum traveling at 60 rpm for 48 h. The culture was then clarified via centrifugation at 3000g for 15 minutes at 4 °C. The pellet was disposed and the bacilysin containing supernatant was kept for further processing by one of two ways as will be indicated in the figure legend of each reported experiment. Either the supernatant was diluted 2-fold with H2O, frozen, and lyophilized to dryness; or the supernatant was subjected to a Dowex purification. For Dowex purification, the supernatant was first diluted 2-fold in ice-cold ethanol, centrifuged as described above, and the pellet discarded. The supernatant was gravity fed through a 2 × 1 cm Dowex 50WX8-200 hand-poured column equilibrated in 50/50 ethanol/H2O. Unbound material was washed away with 5 mL of H2O and bound compounds were eluted with 2 mL of 4% ammonium hydroxide (aqueous) into a vessel submerged in liquid N2. The elution was immediately flash frozen and lyophilized to dryness. This small-scale growth/Dowex purification was also performed with B. subtilis PY79 ΔbacB and ΔbacC strains. Dowex purification of the ΔbacC culture yielded dihydrobacilysin and dihydroanticapsin products that had their epoxide hydrated. To avoid this fate the ΔbacC strain was re-grown and the supernatant was lyophilized without subjection to Dowex purification. Although this resulted in lower signal-to-noise ratio during LC/MS analysis (data not shown), the masses of dihydrobacilysin and dihydroanticapsin with the intact epoxide were obtained (Figure 2). All lyophilized samples were dissolved in ~ 1 mL of H2O for loading onto the LC/MS.</p><p>LC/MS detection of bacilysin, anticapsin, dihydrobacilysin, and dihydroanticapsin was performed in positive detection mode with 0.1% formic acid spiked into Buffer A (H2O) and Buffer B (acetonitrile). The mass spectrometer was set to the following parameters: 2500 V capillary voltage; 350 °C drying gas at a flow-rate of 10 L/min; 30 psi nebulizer pressure; and 125 V fragmentor voltage. The LC was set to a flow-rate of 0.4 mL/min through a 50 × 2.1 mm 5 μ Hypercarb column. The sample loading volume was 3 μL and the compounds were eluted using the following LC program: 0% Buffer B for 1 min; 0 – 95% linear gradient of Buffer B over 10 min; 95 – 0% Buffer B linear gradient over 2 min, 0% Buffer B reequilibration for 10 min before loading the next sample. Samples in which the mass signals saturated the detector were diluted in H2O and re-run.</p><!><p>To produce enough dihydroanticapsin to allow the collection of NMR spectra, a 500 mL culture of PA media (additionally buffered with 100 mM MOPS pH 6.8) was placed in a 2800 mL baffled flask and inoculated with a 5 mL starter culture grown in the same media. The starter culture had been inoculated with a freshly streaked colony of B. subtilis PY79 ΔbacC cells and grown for 16 h at 30 °C in a roller drum traveling at 60 rpm. After inoculation, the 500 mL culture was grown at 30 °C for 48 h with shaking at 200 rpm. The supernatant of the culture (containing dihydroanticapsin) was harvested via centrifugation at 3000g for 20 min at 4 °C. The pellet was discarded and the supernatant was diluted 2-fold with ice-cold ethanol. Nascent precipitate was removed by repeating the centrifugation described above and discarding the pellet.</p><p>The first step of purification consisted of loading the supernatant onto a 6 × 2.5 cm Dowex 50WX8-200 hand-poured column via gravity flow at room temperature. The Dowex resin had been previously equilibrated in 50/50 ethanol/H2O. Unbound material was washed away with 40 mL of H2O. Bound compounds were eluted with 40 mL of 4% aqueous ammonium hydroxide into a vessel submerged in liquid N2. After freezing, the elution was lyophilized to dryness.</p><p>For the next purification step, the lyophilized Dowex elution was dissolved in 3 mL of Buffer A (H2O + 0.1% formic acid). 1 mL of this mixture was loaded onto a 100 × 21.2 mm Hypercarb 5 μ column equilibrated in Buffer A using a flow-rate of 4 mL/min (3 runs total). Bound components were eluted with a 0 – 95% linear gradient of Buffer B (acetonitrile + 0.1% formic acid) over 45 min. Fractions (4 mL) containing dihydroanticapsin hydrate were identified by LC/MS using the same method as described above for bacilysin analysis. Once identified, the fractions containing dihydroanticapsin hydrate were frozen and lyophilized to dryness.</p><p>The final purification step of dihydroanticapsin hydrate was identical to the purification described in the previous paragraph, except with Buffer A as 10 mM potassium phosphate pH 8 and Buffer B as neat acetonitrile. The fraction (3 mL) containing dihydroanticapsin hydrate was frozen, lyophilized to dryness, and dissolved in 300 μL of D2O. The sample was placed in a 5 mm D2O matched Shigemi tube and NMR spectral data (1H, 13C, 1H-13C HSQC, 1H-13C HMBC, 1H-1H COSY, 1H-1H NOESY) were collected at 25 °C. Water suppression was accomplished via pre-saturation and proton signals were referenced to the residual H2O peak (4.79 ppm)17. Carbon peaks were referenced by spiking the NMR sample with 0.5 μL of acetonitrile and referencing to its methyl carbon signal (1.47 ppm)17. Pure bacilysin was expressed and processed exactly as described for dihydroanticapsin except the initial supernatant was obtained from B. subtilis PY79 WT cells (Figure S15).</p><!><p>H4Tyr diastereomer substrates were prepared in preparative amounts from potassium prephenate as previously described8. The concentrations of the substrates were determined from NMR proton spectra by referencing the integration of the C7 proton signal of H4Tyr to the proton signal of an internal standard of 1 mM sodium formate (Figure S5).</p><p>The ability of BacC to process 4S- and 4R-H4Tyr was initially screened via a UV-vis assay monitoring the production of NADH at 340 nm. The reactions contained 4 mM NAD+ and 2 mM H4Tyr diastereomer in 50 mM potassium phosphate buffer pH 8.0. Each reaction was initiated via addition of 25 μM BacC, placed in a 1 cm quartz cuvette, and monitored for 30 seconds in the spectrophotometer at 20 °C.</p><p>The reactivity of BacC and BacD with the H4Tyr diastereomers was further analyzed by LC/MS in negative detection mode using the same flow-rate, column, and LC method as the LC/MS analysis of bacilysin. However, in this experiment Buffer A and Buffer B contained 0.1% ammonium hydroxide instead of formic acid and the spectrometer parameters were set to the following: 3500 V capillary voltage; 300 °C drying gas at a flow-rate of 11 L/min; 30 psi nebulizer pressure; and 125 V fragmentor voltage. Reactions for this analysis contained 1 mM H4Tyr, 3 mM l-Ala, 3 mM ATP, 3 mM NAD+, and 5 mM MgCl2 in 50 mM potassium phosphate buffer pH 8.0. The reactions were initiated via the addition of 20 μM of BacC alone, BacD alone, or BacCD added simultaneously. After incubation at room temperature for 6 h, each 25 μL reaction was quenched by adding acetonitrile to 75% v/v and vortexing. H2O was added to the quenched reactions to reduce the concentration of acetonitrile to 25% v/v to facilitate freezing. The reactions were frozen, lyophilized to dryness, re-dissolved in 120 μL of H2O, centrifuged to pellet any insoluble material, and 3 μL of the supernatant was loaded onto the LC/MS for analysis.</p><p>Preparative scale reactions of BacD and BacCD were run with 4S-H4Tyr to obtain product dipeptides for NMR analysis. The BacD reaction contained 6 mM 4S-H4Tyr, 15 mM ATP, 15 mM l-Ala, 20 mM MgCl2, 1 mM DTT, and 20 μM BacD in 500 μL of potassium phosphate buffer pH 8.0. The BacCD reaction was identical to the BacD reaction except with the addition of 15 mM NAD+, 30 μM BacC, and the concentration of BacD was decreased to 10 μM. Both reactions were incubated at room temperature for 12 h before being quenched via addition of acetonitrile to 30% v/v, frozen in liquid N2, and lyophilized to dryness. The dried reactions were re-dissolved in 1 mL of Buffer A for purification (10 mM potassium phosphate buffer pH 8). The reaction mixtures were centrifuged to pellet insoluble material and the supernatants were loaded (individually) onto a 100 × 10 mm 5μ Hypercarb column equilibrated in Buffer A at a flow-rate of 1.5 mL/min. Bound compounds were eluted with a linear gradient of acetonitrile (no additives) and fractions were analyzed for the appropriate product via LC/MS in negative detection mode as described above for the analysis of the small-scale BacCD reactions. 3 mL of fractions containing the desired compounds were frozen, lyophilized to dryness, and dissolved separately in 300 μL of D2O. The dissolved compounds were analyzed by NMR spectroscopy as described above for dihydroanticapsin hydrate.</p><!><p>The ΔbacC extract used as substrate was identical to that described above (without Dowex processing) that contained masses of dihydrobacilysin and dihydroanticapsin with the intact epoxide (Figure 2). Each 50 μL reaction contained 33 μL of ΔbacC extract spiked with 2 mM NAD+, 2 mM l-Ala, 2 mM ATP, and 50 mM potassium phosphate pH 8.0. The reactions were initiated with 20 μM of BacC, BacD, or BacCD simultaneously. The reactions were incubated, quenched, and prepared for LC/MS analysis exactly as described above for the BacCD reactions with H4Tyr. However, the actual LC/MS analysis was identical to the positive detection mode method used to analyze the original ΔbacC extract (described above).</p><!><p>Purified 4R- and 4S-[13C]-H4Tyr diastereomers were prepared from 3, 5, 5'-[13C]-prephenate (non-uniformly labeled) as previously described for the generation of unlabeled H4Tyr diastereomers. The 3, 5, 5'-[13C]-prephenate was enzymatically prepared from 2, 6, 9-[13C]-chorismate as previously described7. The concentration of both [13C]-H4Tyr diastereomers was determined by NMR spectroscopy by referencing to an internal standard of sodium formate as described above for unlabeled H4Tyr. The extents of [13C]-labeling of the H4Tyr diastereomers were determined by LC/MS using the negative detection mode method described above (Figure S6).</p><p>Purified 3E- and 3Z-[2H]-ex-H2HPP diastereomers were prepared from reactions of prephenate in 95% D2O with either BacAB or AerDE, respectively, as previously described8. The concentration of each diastereomer was determined via UV-vis spectroscopy using previously reported extinction coefficients8. The extents of [2H]-labeling of the ex-H2HPP diastereomers were determined by LC/MS using the negative detection mode method described above (Figure S6).</p><p>4R- and 4S-[13C]-H4Tyr diastereomers were added (separately) to 5 mL cultures of PA media (additionally buffered with 100 mM MOPS pH 6.8) at a final concentration of 1 mM. The cultures were inoculated with a freshly streaked colony of B. subtilis PY79 WT and ΔbacB cells (separately) and incubated at 30 °C for 48 h in a roller drum traveling at 60 rpm. The culture was clarified via centrifugation (as described above) and the supernatant was diluted 2-fold into H2O before being froze and lyophilized to dryness. The dried supernatant was resuspended in 1 mL of H2O and analyzed via LC/MS using the positive detection mode method described above. To ensure that B. subtilis could take up labeled amino acids under our experimental conditions, this feeding experiment was repeated with B. subtilis PY79 WT cells being fed with 3-[13C]-l-alanine (99% labeling uniformity) instead of [13C]-H4Tyr. Because l-Ala can be transaminated in vivo to yield pyruvate, we saw multiple 13C labels incorporated into bacilysin indicating l-Ala was accepted into B. subtilis cells (Figure S7).</p><p>3E- and 3Z-[2H]-ex-H2HPP diastereomers were added (separately) to 5 mL cultures of PA media at a final concentration of 1 mM. (MOPS was not added to this media as we found the high concentration of MOPS present in the [13C]-H4Tyr feedings to saturate the detector.) The cultures were inoculated with a freshly streaked colony of B. subtilis PY79 WT, ΔbacB, and ΔbacC cells (separately). The cultures were grown at 37 °C for 24 h in a roller drum traveling at 60 rpm. The cultures were then clarified, the supernatant processed using Dowex 50WX8-200 resin, and LC/MS analyzed in positive detection mode exactly as described above for the analysis of bacilysin from small-scale B. subtilis PY79 cultures.</p><p>Because previous data had shown that the deuteriums in the [2H]-ex-H2HPP diastereomers can exchange with protons in H2O upon BacB action7, 8, the temperature of the incubations in the above paragraph was increased relative to that in the H4Tyr feeding incubations (described above) so that a shorter incubation time could be employed to limit "washing out" of the deuterium signal. To ensure that the new incubation conditions would provide the same qualitative result, albeit cleaner, as the conditions used for the [13C]-H4Tyr feedings, 3E-[2H]-ex-H2HPP (1 mM) was fed to a 5 mL B. subtilis PY79 WT culture and incubated/processed exactly as for the [13C]-H4Tyr feedings (Figure S8).</p><!><p>In parallel with in vitro studies on purified BacC described in a subsequent section, we undertook construction of a clean ΔbacC deletion mutant of the bacilysin-producing B. subtilis PY79 strain to evaluate accumulating intermediates. The ΔbacC deletion mutant was constructed by double-crossover homologous recombination using the pMiniMAD plasmid12. To obtain the deletion, two regions of homology were cloned into pMiniMAD: 1) the 500 bp sequence immediately upstream of bacC joined directly to 2) the 500 bp sequence located immediately downstream of bacC. However, because the stop codon of bacB (TGA) overlaps the start codon of bacC (ATG), the stop codon of bacB was mutated from TGA to TAA so the start codon of bacC (now ATA) would be disrupted to eliminate nonsense transcripts. These two regions of homology were ligated together and into the pMiniMAD plasmid using the method of Gibson et al.14 such that cloning scars were not introduced. The plasmid was then transformed into B. subtilis PY79 and positive transformants were identified via the erythromycin resistance cassette present in the pMiniMAD backbone (this integration event is single-recombination: integration of the entire plasmid into the chromosome (Figure S4)). Individual colonies were inoculated in LB liquid cultures without antibiotic and incubated in attempt to accomplish the second-recombination. The second-recombination removes the plasmid backbone, resistance cassette, and gene of interest from the chromosome. After several rounds of dilution and re-growth, culture was plated on LB agar without antibiotics. Colonies that successfully completed the second-recombination were initially identified via replica plating on LB-agar containing erythromycin, and then confirmed by colony-PCR followed by sequencing of the colony-PCR dsDNA product.</p><p>B. subtilis PY79 ΔbacC cells (and WT cells as control) were grown in minimal media and the cell mass and spent media (5 mL) were separated by centrifugation. The cell mass was discarded while the supernatant was frozen and lyophilized. The lyophilized media extract was then examined for anticapsin, bacilysin, and any dihydro-intermediates that might accumulate from loss of the predicted C7-hydroxyl oxidase activity of BacC (Figure 2A). High-resolution LC/MS analyses (Figures 2B and 2C) showed the WT extract contained a significant amount of bacilysin (calc. mass: 271.1288; observed: 271.1296) and a minor amount of anticapsin (calc. mass: 200.0917; observed: 200.0921), but the ΔbacC extract did not possess detectable amounts of either. However, the ΔbacC extract did contain accumulated amounts of masses corresponding to dihydroanticapsin (calc. mass: 202.1074, observed: 202.1072) and dihydrobacilysin (calc. mass: 273.1445; observed: 273.1446), which contain the intact epoxide but possess a C7-hydroxyl moiety in place of the C7-ketone found in mature anticapsin and bacilysin (Figure 2A). This finding is consistent with BacC assignment as the dehydrogenase used to oxidize the C7-hydroxyl, and strongly suggests that BacC oxidation occurs sometime after the epoxidation of the cyclohexenol double-bond.</p><p>To support the existence of the dihydroanticapsin/dihydrobacilysin intermediates suggested above by the LC/MS data, a fermentation of the ΔbacC deletion strain was scaled up in an effort to obtain sufficient materials for NMR characterization. 500 mL of B. subtilis ΔbacC culture was harvested and after discarding the cell mass, the spent media was subjected to an initial purification on a hand-poured Dowex 50WX8-200 cation-exchange column. Bound compounds were eluted with a 4% solution of aqueous ammonium hydroxide and then subjected to further purification on a preparative Hypercarb HPLC column. Even though dihydrobacilysin was already an initial minor component compared to dihydroanticapsin (Figure 2B), after the acidic purification on the Dowex and Hypercarb resins essentially all of the dihydrobacilysin dipeptide was hydrolyzed to the amino acid (data not shown). Additionally, the epoxide of dihydroanticapsin became irreversibly hydrated to 5, 6, 7-trihydroxycyclohexyl-Ala that we termed dihydroanticapsin hydrate (calc. mass: 220.1179; observed: 220.1183) (Figure 3, Table S2). As shown in Figure 3, the structure of dihydroanticapsin hydrate was definitively determined by 1H-NMR spectroscopy, fully consistent with the LC/MS data for the ΔbacC culture extracts presented above. Only a trace amount of this trihydroxy intermediate is detectable in the B. subtilis WT strain with the functional bacC gene (when processed with Dowex) (Figure S9A). Although a 1H – 1H NOESY spectrum was collected in attempt to determine the stereochemistry of the three hydroxyl groups present in dihydroanticapsin hydrate, spectral overlap between H5, H6, H7 and H3b, H4 made this task impossible (Figure 3).</p><!><p>Overproduction of B. subtilis BacC with an N-terminal His6-tag was carried out in E. coli BL21 (DE3) cells, yielding 16 mg of purified, soluble BacC protein per liter of culture (Figure S3). Incubations of BacC and NAD+ cofactor with the purified hydrate of dihydroanticapsin yielded unaltered substrate. Although we were not able to purify the presumed in vivo BacC substrate (dihydroanticapsin), due to reactivity of the epoxide, we did possess mg quantities of 2S, 4S, 7R- and 2S, 4R, 7RH4Tyr diastereomers, remaining from our previous study8, to test as potential surrogate substrates. (For simplicity, 2S, 4S, 7R-H4Tyr will be abbreviated as 4S-H4Tyr and 2S, 4R, 7R-H4Tyr will be abbreviated as 4R-H4Tyr for the remainder of this manuscript.) When 2 mM of these two substrates were individually exposed to 25 μM of purified BacC and 4 mM NAD+ cofactor, the incubation of the 4S-H4Tyr diastereomer showed 100-fold higher velocity than the incubation with the 4RH4Tyr diastereomer, as noted by the increase in light absorption at 340 nm (Figure 4A). (It is possible that the detected velocity of the 4R-H4Tyr incubation is actually due to 4S-H4Tyr contamination from how the H4Tyr diastereomers are prepared8.) Additionally, LC/MS analysis (positive detection mode) of the 4S-H4Tyr reaction showed the appearance of a new mass corresponding to H4Tyr with an oxidized C7 hydroxyl, cyclohexenonyl-Ala (calc. mass: 184.0968; observed 184.0969) (Figure S10B). LC/MS analysis of the 4R-H4Tyr reaction showed no detectable amount of new product (Figure S10A)</p><p>Given that anticapsin has 4S stereochemistry, it was gratifying to observe that BacC oxidized the 4S-H4Tyr diastereomer but possessed negligible activity with the 4R-H4Tyr diastereomer (Figure 4A). We anticipated the product enone might be susceptible to both intermolecular addition of exogenous nucleophiles as well as intramolecular capture by its own amino group18. Indeed, prolonged incubations of BacC with 4S-H4Tyr to obtain NMR quantities of the BacC product yielded multiple product peaks as observed by LC/MS (as also seen in Figure S10B). The identity of those metabolites is the subject of a further study that will be reported separately and relates to the formation of the bicyclic 2-carboxy-6-hydroxyoctahydroindole scaffold found in cyanobacterial aeruginosin peptide toxins19, 20. As detailed below, we turned to a coupled assay with purified BacD ligase enzyme to generate a stable dipeptide product to enable further characterization.</p><!><p>Bioinformatics predicts that BacD is in the ADP-forming dipeptide ligase superfamily. In support of this prediction, workers have reported initial kinetic studies of BacD as an L-Ala-L-X promiscuous dipeptide ligase with potential practical utility10, 11. They had not, however, probed specificity for the possible in vivo substrates of BacD (i.e. X= anticapsin, H4Tyr, its C7-oxidation product cyclohexenonyl-Ala, or dihydroanticapsin). To undertake substrate studies and to couple with purified BacC incubations, we overproduced BacD in E. coli BL21 (DE3) cells yielding 6 mg of highly purified BacD per liter of culture (Figure S3).</p><p>As expected from the previous published work10, BacD supplied with ATP readily ligated L-Tyr onto the C-terminus of L-Ala (data not shown). We then found that BacD would ligate L-Ala onto the N-terminus of both the 4S and 4R diastereomers of H4Tyr, as confirmed by LC/MS (Figures 4B, S10A, and S10B). Given the positive BacC result with only the 4S-H4Tyr diastereomer, we focused our efforts only on this diastereomer and subsequently confirmed the identity of the LAla-4S-H4Tyr dipeptide by 1H-NMR (Figure S11A and Table S3). We viewed the LAla-4S-H4Tyr dipeptide as a potential substrate for subsequent C7-hydroxyl oxidation by BacC. However, BacC did not utilize the L-Ala-4S-H4Tyr as a substrate (Figure S10B). This data strongly argues that the BacC oxidase acts before the BacD ligase in the bacilysin biosynthetic pathway.</p><!><p>Given the oxidoreductase activity of BacC on 4S-H4Tyr and the instability of the BacC enone product, we examined whether the BacD ligase could capture that presumed 4S-cyclohexenonyl-Ala BacC product and sweep it through to the dipeptide. A simultaneous incubation of BacC and BacD with 4S-H4Tyr, L-Ala, and ATP did in fact yield the L-Ala-4S-cyclohexenonyl-Ala dipeptide product, (Figure 4C), corroborated both by LC/MS (calc.: 253.1194; observed: 253.1198) (Figure S10B) and by 1H-NMR (Figure S11B and Table S4). This result indicates the BacD ligase can intercept the nascent 4S-cyclohexenonyl-Ala and ligate its N-terminus to L-Ala before its free amino group undergoes intramolecular addition. Although we have not performed comprehensive kinetics with BacD, the fact that the simultaneous BacCD incubations yield predominantly LAla-L-cyclohexenonyl-Ala dipeptide and not L-Ala-L-H4Tyr dipeptide suggests that cyclohexenonyl-Ala is a better substrate than H4Tyr for the BacD ligase (Figure S10B).</p><!><p>In the aforementioned experiments, we viewed the 4SH4Tyr diastereomer as a surrogate substrate for BacC and the 4S-cyclohexenonyl-Ala BacC oxidation product as a surrogate substrate for BacD ligase. Because we do not currently have a route to purify dihydroanticapsin (with the epoxide intact) from ΔbacC strain extracts, we decided to add exogenous BacC and BacD into crude ΔbacC extracts and analyze any dihydroanticapsin transformations by LC/MS.</p><p>To supercharge any potential BacC and/or BacD activity, extracts were first doped with 2 mM each of L-Ala, ATP, and NAD+. The loaded extracts were then exposed to 20 μM of BacC alone, BacD alone, or BacCD simultaneously. LC/MS analysis of these reactions (Figure 5) revealed four important conclusions: 1) BacC oxidizes the C7-hydroxyl of dihydroanticapsin to generate anticapsin; 2) BacC has no oxidation activity on the dihydrobacilysin dipeptide (as was inferred previously with the L-Ala-4S-H4Tyr dipeptide); 3) BacD can readily ligate dihydroanticapsin with L-Ala to yield dihydrobacilysin; 4) simultaneous BacCD exposure sweeps through dihydroanticapsin to almost exclusively generate bacilysin. These data corroborate our previously drawn conclusion from the surrogate substrates where the tetrahydro amino acid (4S-H4Tyr) was a better substrate for the BacC oxidase than the BacD ligase, and verifies that BacD dipeptide ligation is the last step of bacilysin biosynthesis.</p><!><p>The data we have presented thus far revealed that epoxidation occurs before C7-hydroxyl oxidation, which occurs before L-Ala ligation. However, it was still not clear at what point epoxidation occurs and if the BacABGF-produced H4Tyr was an on-pathway intermediate or merely a shunt product. To begin answering these questions, we first prepared mg quantities of 4S- and 4R-[13C]-H4Tyr diastereomers (Figures 6A and S6), from 2, 6, 9-[13C]-chorismate utilized in one of our previous studies7, 1 millimolar amounts of each of these [13C]-H4Tyr diastereomers was fed to a small scale growth of B. subtilis PY79 WT cells and LC/MS used to detect if any 13C-label was incorporated into bacilysin (Figure S12A). LC/MS analysis showed no increase in the 13C content of detected bacilysin despite the [13C]-H4Tyr feeding (Figure 6C). This result confirmed our previous suspicion that H4Tyr is a shunt product and is not on-pathway for bacilysin biosynthesis.</p><p>We then decided to go back two enzymatic steps and produce mg quantities of 3E- and 3Z-[2H]-ex-H2HPP by tandem action of BacAB and AerDE8, respectively, in D2O (= 2H2O) (Figures S6A and S6C). (We chose 13C-labeling of the H4Tyr diastereomers in the above paragraph because running the 4-enzyme tandem of either BacABGF or AerDE + BacGF in D2O was severely inefficient due to cumulative solvent isotope effects.) When the [13C]-H4Tyr feeding described above was carried out instead with the [2H2]-ex-H2HPP diastereomers, LC/MS analysis revealed that deuterons from 3E-ex-H2HPP (and less so 3Z-ex-H2HPP) appeared in bacilysin upon B. subtilis WT fermentation (Figures 6D and S12B and Table S5). (We did not expect the results of this feeding to be perfectly clean as we have previously shown that BacB can interconvert the 3E- and 3Z-ex-H2HPP isomers7.) To confirm this result we fed the [2H]-ex-H2HPP diastereomers to the B. subtilis ΔbacC deletion strain and were able to detect deuterons primarily from 3E-ex-H2HPP in both dihydroanticapsin and dihydrobacilysin (Figures 7 and S13). This result was initially baffling because our previous experiments clearly demonstrated that the 3Z (and not the 3E) diastereomer undergoes BacG-mediated hydride reduction to give the 4S stereochemistry that is seen in anticapsin/bacilysin (Figure 1)8, 18, 21. A probable explanation for this result is that 3E-ex-H2HPP is the substrate for epoxidation by a currently unidentified enzyme activity in the pathway.</p><p>In addition to the ΔbacC B. subtilis PY79 strain, we also possessed a ΔbacB deletion strain constructed by equivalent methodology that has been previously reported13. When bacB was deleted there was no detectable production of anticapsin, bacilysin, or dihydro intermediates (found in the ΔbacC strain) in the fermentation extracts (Figures S9A and S9B). This result is consistent with BacB playing an essential, but earlier role than BacC in the anticapsin/bacilysin pathway. Indeed, when the B. subtilis PY79 ΔbacB strain was fed with the [2H]-ex-H2HPP diastereomers, neither the 3E nor the 3Z isomer rescued bacilysin/anticapsin/dihydroanticapsin production (data not shown). As a member of the bicupin enzyme family, which are known to possess a wide range of activities22, 23, it is possible that BacB is the missing epoxidase24 although our assays with purified BacB have not revealed such an activity.</p><!><p>Given the results from the studies described here and building on prior efforts6–8, we can fill in missing steps in the anticapsin/bacilysin pathway and show that BacC and BacD catalyze the last two steps. In particular BacC is a NAD+-dependent alcohol dehydrogenase, working to oxidize the C7-hydroxyl as the last step in anticapsin assembly (Figure 8). BacD is the dipeptide ligase that adds l-Ala to the amino group of anticapsin as a self-protection strategy of B. subtilis producers against inactivation of their own glucosamine synthase by anticapsin. BacD is promiscuous for the C-terminal substrate, also accepting 2S-Tyr, 2S, 4R, 7R- and 2S, 4S, 7R-H4Tyr diastereomers, and also dihydroanticapsin for coupling with L-Ala.</p><p>Our previous efforts7, 25 have defined BacA as the founding member of a novel class of non-aromatizing prephenate decarboxylases that initiates the pathway by diverting some of the prephenate pool to the endocyclic dienyl product 7R-en-H2HPP. BacB acts next to accelerate the isomerization of one of the double bonds into conjugation with the 2-keto moiety to yield the thermodynamically favored 7R-exocyclic-H2HPP. We have demonstrated that BacB equilibrates the Δ3-geometric isomers to a 7:3 E:Z ratio7.</p><p>Two findings of this study are consistent with epoxidation of the double-bond occurring next, by an as yet uncharacterized enzyme activity. One of those findings is that 3E-H2HPP (but not 3Z) is on pathway (deduced from feeding studies) in the parental B. subtilis strain and in a ΔbacC but not a ΔbacB strain. The second finding is that dihydroanticapsin accumulates in the ΔbacC strain and can be processed by pure BacC and then BacD to anticapsin and bacilysin, respectively.</p><p>We thus feature epoxidation occurring at the level of the 3E-ex-H2HPP isomers. Epoxidation after reduction to the 4S-H4HPP level is ruled out by failure of the 3Z-H2HPP isomer (known to give 4S-H4HPP by BacG action8) to yield anticapsin/bacilysin in the feeding studies noted above. We further posit that in vitro action of BacG and BacF on the 3E- and 3Z-H2HPP isomers8 reflects the permissiveness of these reactants as surrogate substrates. The resultant H4Tyr diastereomers are shunt products (and in labeled form do not go on to anticapsin) but the stereochemical outcome from BacG action8 is a useful constraint: 3E-ex-H2HPP gives the 4R-H4Tyr product (after coupled transamination by BacF) while 3Z-ex-H2HPP gives the 4S-H4Tyr product. Anticapsin (and by inference dihydroanticapsin) has 4S-stereochemistry. Thus, a putative epoxy-H2HPP substrate for BacG could be the 3Z- isomer with the same facial selectivity for hydride addition to C4. This would suggest BacB could equilibrate the epoxy-3EH2HPP to the 3Z isomer before BacG acts (Path 2 in Figure 8). Alternatively, the prior introduction of the epoxy group on the 3E isomer could direct hydride addition by BacG preferentially/exclusively to the ring face opposite to the epoxide (Path 1) and yield epoxy-4S-H4HPP, a transamination away from the observed dihydroanticapsin (Figures S2C and S2D). Further insights will require detection of the earliest epoxygenated scaffold and the catalyst responsible.</p><p>With regard to antibiotic activity, we compared bacilysin with dihydrobacilysin as noted in Figure S14B against a lawn of S. aureus RN4220 as the test organism. Bacilysin releases the proximal inhibitor anticapsin and has the anticipated growth inhibition. Dihydrobacilysin is inactive in that initial assay, most probably because dihydroanticapsin, while possessing the epoxide, lacks the C7 ketone functionality, implicated in the glucosamine synthetase inactivation mechanism by Baldwin and colleagues4. In terms of the importance of the epoxide to the epoxyketone warhead, we found that the l-Ala-4S-cyclohexenonyl-Ala dipeptide described above from action of BacC and BacD on L-Ala and 2S, 4S, 7RH4Tyr also does not show antibiotic activity under those assay conditions (Figure S14A). Future efforts will be required to deconvolute whether the 4S-cyclohexenonyl-Ala is released by peptidase action in the target bacterial cell, lasts long enough to reach the active site of glucosamine synthase, and whether the enone is or is not an effective warhead compared to the epoxyketone in anticapsin.</p>

PubMed Author Manuscript

New bactericide derived from Isatin for treating oilfield reinjection water

Isatin, an extract from Strobilanthes cusia (Nees) Kuntze, was the base for synthesizing derivatives that were screened for antibacterial activity against oilfield water-borne bacteria. The bacterial groups are sulfate reducing, iron and total. The derivatives were characterized by spectrums and they showed good to moderate activity against sulfate reducing bacteria.

new_bactericide_derived_from_isatin_for_treating_oilfield_reinjection_water

Background<!><!>Synthesis of isatin derivatives<!>X-ray data collection and structure refinement<!><!>Microbiological monitoring<!>Chemistry<!><!>Chemistry<!><!>Bioactivity<!><!>Bioactivity<!>Competing interests<!>Authors' contributions

<p>The roots and the leaves of the plant, Strobilanthes cusia (Nees) Kuntze of the Acanthaceae family that is widely distributed in northern and central China, have been used in traditional Chinese medicine to treat a variety of ailments caused by microorganisms and virus. It is suggested that the demonstrated use can be extended to processing oilfield water to remove or reduce bacteria before the water is re-injected into formations via wells.</p><p>The alkaloid isatin or indole-2-3-dione (Figure 1) is a compound found in Strobilanthes cusia (Nees) Kuntze and many other plants such as genus Isatis, Calanthe discolor LINDL, Couroupita guianensis Aubl. and in mammalian tissue [1]. It has versatile bioactivity [2] and it is used to synthesize a large variety of heterocyclic compounds in preparing drugs [3-7]. Isatin Schiff bases are reported to have antibacterial activity against Bacillus subtilis[8], Gram(+) and Gram(−) bacterial strains [9] and Magnaporthe grisea[10] among others. The compound has been produced industrially and can thus be used for large-scale applications such as treating oilfield water before re-injection.</p><!><p>Development of new bactericide for oilfield reinjection water treatment from traditional Chinese medicine.</p><!><p>Isatin (1 mmol) was dissolved in methanol (20 ml) and a methanol solution of 1.2 mmol amino compound (10 ml) was added dropwise, until the disappearance of isatin, as evidenced by thin-layer chromatography. The solvent was removed in vacuo and the residue was separated by column chromatography (silica gel, petroleum ether/ ethyl acetate = 1:1 ~ 1:3 v/v), to give the product. Single crystals of the compound 4 suitable for X-ray analysis was obtained on slow evaporation of a methanol solution (30 ml) of the product (30 mg) over a period of 7 d.</p><!><p>Intensity data for colorless crystals of compound 4 was collected at 150 K on a Bruker SMART 1000 CCD fitted with Mo Ka radiation. The data sets were corrected for absorption based on multiple scans [11] and reduced using standard methods [12]. The structures was solved by direct-methods [13] and refined by a full-matrix leastsquares procedure on F2 with anisotropic displacement parameters for non-hydrogen atoms, carbon-and nitrogen bound hydrogen atoms in their calculated positions and a weighting scheme of the form w = 1/σ2(Fo2 ) + (αP)2 + bP where P = (Fo2 + 2Fc2)/3) [14]. Crystal data and refinement details were given in Table 1.</p><!><p>Experimental data of compound A and B</p><!><p>Viable counts of SRB, TGB and FB were determined with the "most probable number" method, People's Republic of China Standard of Petroleum and Natural Gas Industry, the national method of the bactericidal agent's performance, SY/T 5890–1993). The produced water containing the three kinds of bacteria was gathered from Zichang Oilfield Factory, Yanchang Oilfield.</p><!><p>The isatin derivatives were synthesized as shown in Scheme 1. All the isatin derivatives were characterized by 1 H-NMR (400 MHz) and MS (EI) spectra and the results were summarized in Table 2. The entire spectra consist with the anticipated structures.</p><!><p>Synthesis of isatin derivatives by condensation reaction.</p><p>The1 H-NMR (400 MHz) and MS (EI) spectra of the isatin derivatives</p><!><p>Besides, single crystal of compound 4 was analysized by X-ray, which confirms the assignment of the structure from spectroscopic data. The values of the geometric parameters of compound 4 are within normal ranges and experimental errors. The X-ray structural analysis confirmed the assignment of its structure from spectroscopic data. The molecular structure is depicted in Figure 2, and a packing diagram of compound 4 is depicted in Figure 3. Geometric parameters of compound 4 are in the usual ranges. The indol-2-one ring system is substantially planar. In the crystal structure, intermolecular N—H—N and O—H—O hydrogen bonds (Table 3) are effective in the stabilization of the structure and are responsible for the formation of a one-dimensional network. The angle of C1—C2—N2 is 115.734°, and the angle of C2—N2—O2 is 112.199°.</p><!><p>An ORTEP-3 drawing of compound 4, with the atom-numbering scheme and 30% probability displacement ellipsoids.</p><p>Packing of compound 4, dashed lines indicating hydrogen bonds.</p><p>Hydrogen-bond geometry in the crystal of compound 4 (Å, °)</p><p>*Symmetry codes: (i) − x + 1, y + 1/2, −z + 2; (ii) − x + 1, y − 1/2, −z + 2.</p><!><p>Produced water is a consequence of an oilfield exploitation that uses waterflood or steam injection or has an aquifer linked to the reservoir. The most usual disposal ways for high volumes of produced water is re-injected after treatment, which will meet some requirements imposed by environmental regulations [15]. Microbiologically influenced corrosion (MIC) caused by growth of sulfate reducing bacteria (SRB), iron bacteria (IB) and total general bacteria (TGB) in oil pipelines, is considered a major problem for water treatment in the oil industry [16]. MIC can result in different types of attack: pitting, crevices, dealloying and erosion in pipelines [17]. Corrosion products produced by microorganisms are production of hydrogen sulfide, molecular hydrogen, hydrogen ions and destabilization of metal oxide films. In addition, microbial degradation of crude oil can lead to increased acidity in the oil phase, and oil containing acids is a problem concerning corrosion of pipelines. The reported results showed that the interaction of IB, SRB and TGB accelerated the corrosion rate, and the corrosion in the mixture of IB, SRB and TGB was more serious than in a single microbial system. If this is the case, different treatment system to inhibit corrosion should be considered, among which bactericide agent has received the greatest acceptance. Currently, oxidizer, aldehyde, quatemary ammonium salt and heterocycle compounds has been used as bactericide agents, and Cl2, ClO2, formaldehyde, pentane-1, 5-dial, trichloroisocyanuric acid (TCCA) and ect [18], but the toxicity tests have been conducted on a limited selection.</p><p>In this work, isatin and amino compounds condensed to form the new C = N bond, and it is the isostere of C = O in the structure of isatin, which may ensure the bioactivity of these derivatives similar to isatin. The antifungal activity of these compounds against oilfield microorganism was tested under the concentration of 0.20 g/L and 0.02 g/L, and the results were summarized in Table 4.</p><!><p>The antifungal activity of isatin derivatives against MIC</p><!><p>From the table, it can be found that isatin is antifungal active against SRB, but inactive against IB and TGB under both concentrations. For the 5-substitued isatin, compound 2 and 3, the antifungal active against SRB is similar to isatin, slightly more potent against IB, but both, as well as isatin, are inactive against TGB. From the results of the 3-imine indole-2-one (compound 4–10), it was found that the SRB inhibitions are more effective potent under both concentration. While only compound 10 is active against IB with the microbial concentration of 0.5 /mL under the concentration of 0.20 g/L and 0.9 /mL under the concentration of 0.02 g/L. Only compound 7 is active against TGB with the microbial concentration of 2.5 /mL under both concentration.</p><!><p>The authors declare that they have no competing interests.</p><!><p>GC has conceived the study, formulated the research idea and prepared the manuscript draft version, HS, MZ and FH carried out the chemical synthesis, JZ carried out the Microbiological monitoring, and XH and JZ participated in its design and coordination. All authors have read and approved the final manuscript.</p>

PubMed Open Access

Analysis of the potential effect of ponatinib on the QTc interval in patients with refractory hematological malignancies

PurposeCardiac dysfunction, particularly QT interval prolongation, has been observed with tyrosine kinase inhibitors approved to treat chronic myeloid leukemia. This study examines the effects of ponatinib on cardiac repolarization in patients with refractory hematological malignancies enrolled in a phase 1 trial.MethodsElectrocardiograms (ECGs) were collected at 3 dose levels (30, 45, and 60 mg) at 6 time points. Electrocardiographic parameters, including QTc interval, were measured, and 11 morphological analyses were conducted. Central tendency analyses of ECG parameters were performed using time-point and time-averaged approaches. All patients with at least 2 baseline ECGs and 1 on-treatment ECG were included in the analyses. Patients with paired ECGs and plasma samples were included in the pharmacokinetic/pharmacodynamic analysis to examine the relationship between ponatinib plasma concentration and change from baseline in QT intervals.ResultsThirty-nine patients at the 30-, 45-, and 60-mg dose levels were included in the central tendency and morphological analyses. There was no significant effect on cardiac repolarization, as evidenced by non-clinically significant mean QTcF changes from baseline of −10.9, −3.6, and −5.0 ms for the 30-, 45-, and 60-mg dose levels, respectively. The morphological analysis revealed 2 patients with atrial fibrillation and 2 with T wave inversion. Seventy-five patients were included in the pharmacokinetic/pharmacodynamic analysis across all dose levels. The slope of the relationship for QTcF versus plasma ponatinib concentration was not positive (−0.0171), indicating no exposure–effect relationship.ConclusionsPonatinib is associated with a low risk of QTc prolongation in patients with refractory hematological malignancies.

analysis_of_the_potential_effect_of_ponatinib_on_the_qtc_interval_in_patients_with_refractory_hemato

Introduction<!>Study design<!>Analysis population<!>ECG evaluation<!>Statistical analysis of ECG parameters: central tendency analysis and outlier analysis<!>Pharmacokinetic/pharmacodynamic evaluations<!><!>Time-averaged central tendency analysis<!>Morphological analysis<!><!>Discussion

<p>Ponatinib (AP24534) is a novel, synthetic, orally administered, multi-targeted tyrosine kinase inhibitor (TKI) and a potent pan–BCR-ABL inhibitor [1–3]. The product of a computational and structure-based approach to the design of a small-molecule TKI, ponatinib binds with high affinity to the active site of BCR-ABL and renders binding less susceptible to any single amino acid substitution [1]. Ponatinib contains an unique carbon–carbon triple bond linkage that avoids the steric hindrance to other drugs caused by the bulky isoleucine residue at position 315 in the T315I mutant.</p><p>Based on the results in patients with chronic myeloid leukemia (CML) and Philadelphia chromosome–positive acute lymphoblastic leukemia (Ph+ ALL) in phase 1 testing and phase 2 clinical trials [4, 5], ponatinib (45 mg once daily) has been approved in the United States for the treatment of patients with CML and Ph+ ALL that is resistant or intolerant to prior TKI therapy [6].</p><p>Cardiac dysfunction has been noted with other TKIs approved for the treatment of patients with CML. For example, the imatinib prescribing information includes a warning regarding congestive heart failure and left ventricular dysfunction [7]. The nilotinib prescribing information includes "QT prolongation" as a boxed warning [8], and the dasatinib prescribing information carries "QT prolongation" as a precaution [9]. During phase 1 testing of ponatinib, treatment-related QTc prolongation was observed in 4 % of patients [4].</p><p>The QT interval is a measure of the duration of the electrical depolarization and repolarization of the ventricles of the heart and serves as a surrogate marker for the risk of torsades de pointes, which can lead to sudden death. The International Conference on Harmonisation E14 guidelines [10] outline requirements for studies of the effects of drugs on the QT interval. Specifically, the ideal QT study would include a placebo and control drug along with evaluation of a supratherapeutic dose. However, it is not often possible to implement such a study design with cancer patients, particularly the use of a placebo and positive controls.</p><p>The cardiac safety of ponatinib was initially investigated in an in vitro assay conducted in human embryonic kidney cells stably expressing the hERG potassium channel. In this study, 5 ponatinib doses were compared with the positive control cisapride (unpublished data, ARIAD Pharmaceuticals, Cambridge, MA). This preclinical study revealed that ponatinib inhibits hERG current, which is implicated in the prolongation of cardiac repolarization, at concentrations above 1 μM, which is substantially in excess of the steady-state ponatinib maximal concentrations (Cmax) observed in patients treated at the clinical dose of 45 mg orally once daily (geometric mean, 77.4 ng/mL or 0.145 μM) [4]. The cardiac safety of ponatinib was also investigated in vivo in 4 conscious telemetered dogs (unpublished data, ARIAD Pharmaceuticals, Cambridge, MA). In this study, dogs received vehicle and 3 doses of ponatinib (2, 5, and 10 mg/kg) administered 1 week apart; electrocardiographic (ECG), heart rate, and arterial pressure measurements were taken to assess the effects of ponatinib on cardiovascular parameters. This in vivo study showed that oral administration of single doses of ponatinib up to 10 mg/kg was not associated with biologically relevant effects on cardiac or circulatory function.</p><p>This report is a safety analysis of the phase 1 trial focused on the potential effects of ponatinib on cardiac repolarization in patients with refractory hematological malignancies.</p><!><p>The design of this phase 1 trial has been previously described [4]. There were 7 dose levels, with doses ranging from 2 to 60 mg. The primary objective was to determine the maximal tolerated dose, and secondary objectives included safety/tolerability, anti-leukemia activity, and pharmacokinetics (PK)/pharmacodynamics (PD). An additional secondary end point was introduced through a protocol amendment allowing analysis of ECG parameters including the QTc interval, which was primarily assessed using the Fridericia-corrected method (QTcF). The protocol amendment, under which all patients included in the central tendency and outlier analyses were evaluated, required baseline QTc to be less than 450 ms and prohibited concomitant use of medications known to prolong the QTc interval.</p><p>All patients provided signed informed consent. The protocol, amendments, and consent forms were approved by the institutional review board at each center. The study was conducted in accordance with the Guidelines for Good Clinical Practice and the Declaration of Helsinki.</p><!><p>All patients in the 3 dose groups (30, 45, and 60 mg) with at least 2 available baseline ECGs and 1 on-treatment ECG were included in the central tendency and outlier analyses of ECG parameters (39 of 57 patients at these dose levels met these criteria). All patients across all dose levels with paired ECG and plasma concentrations for ponatinib were included in the PK/PD analysis (75 of 81 patients met these criteria).</p><!><p>Resting 12-lead ECGs were collected across the 3 dose levels (30, 45, and 60 mg) and at 6 time points: baseline (day 1, in triplicate), predose; day 15, predose (single); day 29 (cycle 2/day 1), predose (in triplicate); day 29, 2 h post-dose (in triplicate); day 29, 4 h post-dose (in triplicate); and day 29, 6 h post-dose (in triplicate). Patients were supine and at rest during ECG recording, which was performed from all 12 leads simultaneously for 10 s. Electrocardiograms were recorded using GE MAC1200 ECG recorders (version 6.1) at each study site and transmitted to a central laboratory for analysis conducted by a cardiologist.</p><p>Six cardiac interval durations were measured: heart rate, PR interval, QRS interval, QT interval, QTcF, and Bazett-corrected QT (QTcB). In addition, 11 morphological analyses were conducted to identify the onset of new morphological abnormalities: atrial fibrillation and atrial flutter, second- and third-degree heart block, complete left and right bundle branch block, ST-segment change (elevation and depression separately), wave abnormalities (negative T waves only), myocardial infarction pattern, and abnormal U waves.</p><!><p>The central tendency analysis of all ECG interval parameters, defined as a change from baseline to post-treatment time points (except for cycle 1/day 15, predose), was performed using 2 approaches: time point and time averaged. For the time-point analysis, 3 ECGs were to be collected at each time point (baseline and 4 post-treatment visits); however, post-dose time points with only 2 ECGs were included in this analysis. The data from the 2 or 3 ECGs were averaged to provide a single set of ECG intervals for each time point. Data were summarized using descriptive statistics. Changes from baseline to 4 post-treatment time points were described with data-based (not model-based) 2-sided 90 % CI statistics. For QTc measurements, the QTcF method was the primary measurement; QTcB was considered secondary, provided for historical purposes only. For the time-averaged analysis, baseline time points were averaged and the value obtained was subtracted from the mean of all combined 4 post-treatment ECG time points.</p><p>Outlier or categorical analysis was also performed to identify patients who experienced a significant effect on any ECG interval parameter (heart rate, PR interval, QRS interval, QT interval, QTcF, and QTcB) that would not be revealed by the central tendency analysis and should be considered exploratory in nature. This analysis used a time-averaged approach that compared the baseline ECG interval value with all post-treatment ECG time points, and then, the value that represented the greatest positive change from baseline was chosen to determine whether each patient fell into the outlier criterion. For heart rate, both the largest negative and positive value compared with baseline was chosen.</p><!><p>Plasma samples were collected concomitantly with ECG assessments. A linear mixed-effects modeling approach was used to quantify the relationship between the plasma concentration of ponatinib and the change from baseline in QT intervals. This model was used to estimate the population slope and the standard error of the slope of the relationship between the change from baseline in QTc intervals and plasma concentrations of ponatinib. As this model is meant solely to determine the relationship of QTc change with the degree of change in exposure, the time points are not relevant; therefore, all plasma concentration and time point QTc pairs were used irrespective of the time point and the dose group from which such pairs were taken. A linear relationship was declared if the P value of the slope was less than 0.05.</p><!><p>Demographic characteristics of patients included in the cardiac analysis</p><p>AML acute myeloid leukemia, CML chronic myeloid leukemia, ECOG Eastern Cooperative Oncology Group, Ph + ALL Philadelphia chromosome–positive acute lymphoblastic leukemia, TKIs tyrosine kinase inhibitors</p><p>Electrocardiographic interval parameters (time-averaged central tendency analysis), outlier analysis, and morphological abnormalities, by dose level</p><p>bpm beats per minute, ms milliseconds, QTcF Fridericia-corrected QT, QTcB Bazett-corrected QT, RBBB right bundle branch block, LBBB left bundle branch block</p><p>aThe Bazett correction method is often less reliable than the Fridericia correction method; QTcB is provided for historical purposes only</p><!><p>The findings from the outlier analysis of absolute QTcF duration identified few patients experiencing QTcF prolongation (Table 2). One patient at the 45-mg dose level (5 %) had a QTcF >500 ms; 1 patient at the 60-mg dose level (8 %) had a change in QTcF >60 ms from baseline; and 3 patients at the 45-mg dose level (14 %) had a 30- to 60-ms change in QTcF from baseline (Table 2). The patient at the 45-mg dose level with a QTcF >500 ms had a Cmax of 57.6 ng/mL, which was below the geometric mean Cmax at the 45-mg dose level (77.4 ng/mL) [4]. This patient was receiving concomitant Darvocet (acetaminophen and propoxyphene), a medication known to prolong the QTc interval.</p><!><p>Atrial fibrillation and T wave inversion were observed in 2 chronic-phase CML patients each (Table 2). Three of these 4 patients had a history of cardiovascular disease (e.g., stroke, hypertension, intermittent sinus bradycardia, and palpitations), suggesting that these morphological abnormalities may reflect the patient population being studied rather than representing an effect of the study medication. Given their age (median, 49 years) and ECOG performance status, the patients in this study are representative of the patient population that will be treated with ponatinib in the clinic. The fourth patient was taking concomitant moxifloxacin, a medication known to be associated with cardiac arrhythmias (in violation of the study protocol) [11].</p><!><p>Mean (±90 % CI) change from baseline in electrocardiographic interval parameters for 4 on-treatment time points (time-point central tendency analysis). CI confidence interval, bpm beats per minute, ms milliseconds, QTcF Fridericia-corrected QT, QTcB Bazett-corrected QT</p><p>Change from baseline versus ponatinib plasma concentration</p><p>QTcF Fridericia-corrected QT, QTcB Bazett-corrected QT</p><p>QTcF change from baseline by ponatinib plasma concentration across 7 dose levels (N = 69). QTcF Fridericia-corrected QT, ms milliseconds</p><p>Estimates from linear mixed model QTcF and QTcB</p><p>QTcF Fridericia-corrected QT, QTcB Bazett-corrected QT, ms milliseconds</p><!><p>This analysis of QTc intervals in patients with refractory hematological malignancies who received daily doses of 30, 45, or 60 mg of ponatinib in a phase 1 clinical trial revealed no significant effect of ponatinib on cardiac repolarization. The recommended dose of ponatinib is 45 mg. Initial characterization of cardiac safety, including QT prolongation, was previously described across all 81 patients included in this phase 1 trial [4]. Although dose-limiting toxicities identified in phase 1 did not include cardiovascular findings, among the adverse events reported in the trial (n = 81), 3 patients (4 %) experienced treatment-related QT prolongation: 1 patient each at the 2-, 4-, and 45-mg dose levels. Of these 3 patients, 2 (3 %) experienced grade 3 treatment-related QT prolongation (at the 4- and 45-mg dose levels). All 3 patients had low steady-state Cmax (4.5–57.6 ng/mL), suggesting that QT prolongation was not due to increased ponatinib exposure. Two of the 3 patients were enrolled before protocol amendment, and all 3 patients were found to have prolongation of QTc at baseline or to have received concomitant medications known to be associated with QTc prolongation. There were no clinical consequences of the ECG findings in these patients.</p><p>The results of this cardiac analysis suggest that ponatinib is associated with a low risk of QTc prolongation. Other targeted agents approved for the treatment of CML have been found to be associated with cardiac toxicities [7–9]. Imatinib has been associated with left ventricular dysfunction and heart failure, particularly in patients with comorbidities and risk factors [7, 12]. In the phase 3 International Randomized Study of Interferon and STI571 (IRIS) in 1,106 patients with newly diagnosed Ph+ CML, severe cardiac failure and left ventricular dysfunction were observed in 0.7 % of patients taking imatinib compared with 0.9 % of patients taking interferon alfa plus cytarabine [7, 13, 14]. The dasatinib prescribing information carries QT prolongation as a precaution. In a phase 1 trial (NCT01392703) in 75 healthy subjects, a clear QT prolongation effect was not detected [15]. However, this adverse event emerged in a phase 3 trial conducted in patients newly diagnosed with CML: QTc intervals between 450 and 500 ms were observed in 2 % of the patients taking dasatinib, compared with 4 % of patients taking imatinib [16]. The nilotinib prescribing information includes a boxed warning regarding QT prolongation [8]. Results of ECG analyses conducted on about 400 patients with CML who participated in a phase 1/2 trial (NCT00109707) showed a significant association between nilotinib concentration and a change from baseline in QTcF, indicating a prolongation of the QTc interval associated with nilotinib [17–19]. A modest linear correlation between nilotinib concentration and a change from baseline in QTcF along with a higher incidence of developing ischemic heart disease in the nilotinib arms was also found in the phase 3 trial Evaluating Nilotinib Efficacy and Safety in Clinical Trials—Newly Diagnosed Patients (ENESTnd) conducted in patients with newly diagnosed CML [20, 21]. Finally, the effects of bosutinib on cardiac repolarization were studied in a randomized, crossover, placebo- and moxifloxacin-controlled study. In the healthy adult subjects enrolled in this study, therapeutic and supratherapeutic bosutinib exposures were not associated with QTc prolongation [22].</p><p>This study had 2 primary limitations. First, the study was not designed as a true thorough QT study as outlined by the International Conference on Harmonisation of Technical Requirements for Registration of Pharmaceuticals for Human Use E14 guidelines. However, it is worthwhile noting that formal thorough QT studies are difficult, if not impossible, to conduct ethically in this patient population owing to the requirement for placebo and positive controls. Second, the number of subjects enrolled in the study was relatively small, particularly for dose-level analyses. Unlike the healthy subject QTc analysis conducted with nilotinib, this study evaluated the cardiac effects of ponatinib in a patient population with refractory hematological malignancies, which may increase confidence that these results are consistent with what will be seen in clinical practice.</p><p>The results of this QTc analysis in CML patients treated with ponatinib at clinically relevant doses suggest that ponatinib is associated with a low risk of QTc prolongation.</p>

PubMed Open Access

Solar Hydrogen Generation from Lignocellulose

AbstractPhotocatalytic reforming of lignocellulosic biomass is an emerging approach to produce renewable H2. This process combines photo‐oxidation of aqueous biomass with photocatalytic hydrogen evolution at ambient temperature and pressure. Biomass conversion is less energy demanding than water splitting and generates high‐purity H2 without O2 production. Direct photoreforming of raw, unprocessed biomass has the potential to provide affordable and clean energy from locally sourced materials and waste.

solar_hydrogen_generation_from_lignocellulose

<!>Introduction<!><!>Introduction<!>Photocatalytic Reforming of Biomass<!><!>Photocatalytic Reforming of Biomass<!>PR of Lignocellulose Components<!>Sugars<!>Oligosaccharides and Polysaccharides<!>Cellulose<!>Lignin<!>Raw Biomass PR<!><!>Raw Biomass PR<!>The PR Mechanism<!><!>The PR Mechanism<!>Biomass PR Beyond H2 Generation<!>Conclusion and Outlook<!>Conflict of interest<!>Biographical Information

<p>M. F. Kuehnel, E. Reisner, Angew. Chem. Int. Ed. 2018, 57, 3290.</p><!><p>Biomass is Earth's most abundant renewable resource and has been a source of energy to mankind since the Stone Age. Today, our economy depends on fossil fuels, which are derived from ancient biomass. With the gradual consumption of these non‐renewable resources and problems associated with CO2 emission, finding a sustainable source of energy is imperative.1 H2 is a promising energy carrier for a post‐fossil era, but current H2 production relies on fossil fuel reforming and is thus not sustainable.2 Generating H2 fuel directly from waste biomass without the timescales of fossilization has the potential to afford renewable energy at large scale and low cost, without competition with food production.</p><p>Lignocellulose is the most abundant form of biomass. It has a multi‐component structure, evolved to provide mechanical and chemical stability (Figure 1).3 Its primary component, cellulose, forms strong, poorly soluble fibrils comprising linear glucose β‐1,4‐homopolymer chains linked by hydrogen bonds. Cellulose fibrils are cross‐linked by hemicellulose, a branched co‐polymer of different pentose and hexose sugars. The major non‐carbohydrate component, lignin, is a polyether derived from different phenol monomers in varying compositions. It cross‐links the fibril structure and protects it from UV damage.4 Lignocellulose utilization is therefore kinetically challenging, as it requires disruption of this robust structure.</p><!><p>The structural components of lignocellulose.3</p><!><p>A number of strategies have been developed to produce fuels directly from biomass.5 Alcohol production from combined cellulose saccharification and fermentation is a field of intense research,6 but cellulose hydrolysis is slow and separation of the resulting alcohol is uneconomical at low concentrations. Thermochemical processes such as biomass gasification and reforming require high temperatures and pressures, and the generated H2 contains impurities that must be removed before use.7</p><!><p>Photocatalytic reforming (PR) of biomass uses the photo‐excited state of a semiconductor to drive reforming at ambient conditions (Figure 2 A). When the semiconductor absorbs light of energies greater than its band gap, an electron is excited from the valence band (VB) to the conduction band (CB). CB electrons are highly reducing and can promote the fuel‐forming hydrogen evolution reaction [HER, Eq. (1)], while the oxidizing holes left in the VB can drive the biomass oxidation reaction [BOR, shown for glucose in Eq. (2)].</p><!><p>A) Photocatalytic biomass reforming process. B) The solar spectrum as it reaches the earth's surface (AM 1.5G).</p><!><p>H2 generation from water splitting [Eqs. (3) and (4)] has a large thermodynamic barrier (ΔE 0=−1.23 V) due to the energy‐demanding oxygen evolution reaction [OER, Eq. (3)]. It also generates explosive mixtures of H2 and O2. In contrast, the overall biomass reforming reaction [Eq. (5)] is almost energy neutral (ΔE 0=+0.001 V),8 meaning energy is only needed to overcome activation barriers. In theory, biomass PR is therefore possible using low‐energy photons (visible and IR light), which are highly abundant in the solar spectrum (Figure 2 B).</p><p></p><p>Throughout this Minireview, catalyst performance is compared on the basis of the PR rate [mmolH2  gcat −1 h−1] and external quantum efficiency (EQE). H2 production is given as yield [mmolH2  gbio −1].</p><!><p>Photocatalytic conversion of biomass to CO2 and H2 was first reported in 1980 using TiO2 modified with Pt and RuO2 as hydrogen evolution and biomass oxidation co‐catalysts, respectively.9 The field has progressed significantly since then, but the majority of studies are still performed with TiO2‐based photocatalysts.10 While these materials are robust and inexpensive, their large band gaps (3.2 eV) limit solar light utilization to the UV region (Figure 2 B). PR studies initially focused on generating H2 from biomass‐derived feedstocks. The higher solubility and reactivity of these feedstocks facilitate reaction kinetics,10 but they are valuable chemicals themselves, and thus biomass PR must focus on using inedible waste material without any additional processing.</p><!><p>Sugars have been widely studied as model substrates for biomass photoreforming, since the majority of lignocellulose is based on saccharide monomers (cellulose and hemicellulose).</p><p>Glucose PR is most established using Pt/TiO2.11 These UV light‐absorbing photocatalysts achieved performances up to 1.15 mmolH2  gcat −1 h−1,12 and 8.5 % EQE.11a Other co‐catalysts (Rh,13 Ru,13b, 14 Pd,15 Au)13b, 15b, 16 showed enhanced activity, with AuPd/TiO2 reaching 8.8 mmolH2  gcat −1 h−1 and 17.5 % EQE.17 Non‐precious co‐catalysts (Ni,15b, 18 Fe,19 Cu)13a gave up to 2.0 mmolH2  gcat −1 h−1 and 59 mmolH2  gbio −1 yield. Performing PR at elevated temperature (30–60 °C) improved activity15a and allowed quantitative H2 yield.13b, 20 Moreover, heteroatom doping (B/N,21 S,22 F)23 or sensitization with upconverting Er:YAlO3 particles was employed to improve the light absorption of TiO2.24 Pt/TiO2 also demonstrated PR activity towards other sugars (fructose,12c, 17, 25 galactose,26 mannose,26a sorbose,26a arabinose,25 xylose12d, 27).</p><p>Visible‐light driven glucose reforming was reported using Pt/CdZnS with rates up to 0.485 mmolH2  gcat −1 h−1,28 whereas a related ZnS/ZnIn2S4 solid solution offered a lower performance.29 Non‐precious co‐catalysts were shown to be superior over Pt, with a MoS2/CdS composite30 achieving up to 55 mmolH2 gcat −1 h−1 and 9.3 mmolH2 gbio −1 and 81 mmolH2 gcat −1 h−1 reported for Co/CdS/CdOx quantum dots.31 Narrow‐band gap metal oxides, such as Zn:Cu2O (3.82 mmolH2 gcat −1 h−1)32 and Fe2O3/Si (4.42 mmolH2 gcat −1 h−1)33 have shown promising activities for visible‐light driven glucose PR. Other suitable materials include LaFeO3,34 BixY1‐xVO4,35 CaTa2O6,36 La:NaTaO3,37 and SrTiO3.38</p><!><p>Disaccharides (cellobiose,25, 26 maltose,26b, 34b sucrose,9, 11a, 12a,12b, 13b, 21, 26a, 39 lactose)26b generally gave lower PR rates than monosaccharides, with a maximum activity of 3.69 mmolH2  gcat −1 h−1 reported for sucrose PR over Pt/B,N:TiO2 and a maximum yield of 20 mmolH2 gbio −1 over Pd/TiO2.13b PR of soluble polysaccharides proceeded at even lower rates,9, 12c, 26b presumably due to their higher molecular weights and stable hydrogen‐bonded structures. Soluble starch gave 3.14 mmolH2 gcat −1 h−1 and 26 mmolH2  gbio −1 yield over Pd/TiO2 13b and 1.8 % EQE over Pt/TiO2.11a Visible‐light driven PR of polysaccharides has only been investigated for hemicellulose with Co/CdS/CdOx, with a rate of 2.04 mmolH2  gcat −1 h−1.31</p><!><p>Only a handful of examples have demonstrated cellulose PR. While the thermodynamics of cellulose reforming are similar to that of oligosaccharides,40 the kinetics are more challenging due to the compact tertiary structure of cellulose.</p><p>Direct cellulose PR was first demonstrated using Pt/TiO2/RuO2 at low activities (0.012 mmolH2  gcat −1 h−1);9 comparable performance was achieved with Pt/TiO2.11a Improved cellulose solubility at alkaline conditions led to enhanced activity (0.041 mmolH2  gcat −1 h−1) and 1.3 % EQE.9, 11b Optimization of catalyst loading, cellulose concentration, and pH further increased the performance of Pt/TiO2 to 0.223 mmolH2  gcat −1 h−1.41 Remarkably, cellulose photoreforming proceeded with comparable activity under natural sunlight, demonstrating real‐world applicability. Immobilizing cellulose on the photocatalyst surface enhanced the rate of photocatalysis and produced 67 mmolH2  gbio −1 under UV light; 14 mmolH2  gbio −1 yield were produced under natural sunlight.42 Further enhancement was reported upon raising the reaction temperature (0.61 mmolH2  gcat −1 h−1 at 40 °C).26b An inexpensive Ni/TiO2 photocatalyst achieved a performance of 0.12 mmolH2  gcat −1 h−1 at 60 °C.15b Visible‐light driven cellulose PR was reported at Co/CdS/CdOx in alkaline solution with rates up to 4.9 mmolH2  gcat −1 h−1 and 7.4 mmolH2  gbio −1.31</p><!><p>Although lignin is considered a promising renewable feedstock,43 it has received little attention as a PR substrate. Lignin PR is hampered by its redox stability and brown color, limiting light absorption by the photocatalyst. Pt/TiO2 generated 0.026 mmolH2  gcat −1 h−1 from lignin under UV light (0.6 % EQE).44 Visible‐light driven lignin PR was reported using CdS/CdOx (0.26 mmolH2  gcat −1 h−1)31 and C,N,S‐doped ZnO/ZnS.45</p><!><p>Direct PR of unprocessed biomass is highly desirable to lower H2 production cost, but is hampered by low substrate solubility. Light is scattered from insoluble biomass and absorbed by colored components. The recalcitrance of raw biomass causes a large overpotential for the BOR reaction, requiring strongly oxidizing VB holes.</p><p>PR of various plants (Table 1) was first shown over Pt/TiO2 at rates comparable to pure cellulose (0.004–0.018 mmolH2  gcat −1 h−1).11a, 11b Enhanced performance was achieved in alkaline solution, or upon addition of the OER catalyst RuO2 (0.058 mmolH2  gcat −1 h−1). Elevated temperatures (60 °C) allowed PR of Fescue grass over Pt/TiO2 at 0.061 mmolH2  gcat −1 h−1, albeit only after removal of chlorophyll.15b Natural sunlight‐driven PR of plant matter proceeds in neutral water at rates up to 0.095 mmolH2  gcat −1 h−1 over Pt/TiO2.41 H2 yields were found to vary widely across the different types of biomass (Table 1), with aquatic plants generally demonstrating higher rates and yields than terrestrial plants under similar conditions, presumably due to their lower lignin content. 3.3 mmolH2  gbio −1 were produced from laver with 3.3 % EQE.11a A visible‐light absorbing Co/CdS/CdOx photocatalyst showed high PR activity under simulated sunlight.31 Bagasse, wood, grass and sawdust gave H2 production rates and yields of up to 5.3 mmolH2  gcat −1 h−1 and 0.49 mmolH2  gbio −1. Strongly alkaline conditions enhanced biomass solubility and photocatalyst stability.</p><!><p>Selected examples of photocatalytic reforming of unprocessed lignocellulose.</p><!><p>Biomass solubility is crucial for high PR performance. Adding detergents was shown to enhance the PR rate of castor oil at aqueous Pt/TiO2.46 PR of cotton subjected to hydrothermal liquefaction (250 °C, 40 bar)47 was 50 times faster than with untreated cotton under similar conditions,11b but the overall H2 yield was lower. Dilute acid hydrolysis of pinewood (160 °C, 10 bar) gave a hydrolysate suitable for high‐yield PR over Pt/TiO2 (0.813 mmolH2  gbio −1).48 Alternatively, raw biomass can be digested at mild conditions using natural enzymes. PR of various cellulase/xylanase‐treated grasses27, 49 over Pt/TiO2 achieved rates up to 1.9 mmolH2  gcat −1 h−1 and a yield of 34.6 mmolH2  gbio −1. Protease A‐digested chlorella produced 30 mmolH2  gbio −1 at rates up to 0.234 mmolH2  gcat −1 h−1[50] in neutral water (cf. 0.73 mmolH2  gbio −1 and 0.024 mmolH2  gcat −1 h−1 for untreated chlorella under these conditions).11a Although the yields and rates of pre‐treated biomass compare favorably to PR without pre‐treatment, pre‐processing adds considerable cost and time to the overall process.</p><!><p>Photoreforming consists of two separate half‐reactions (see Section 2). HER is substrate‐independent, and typically proceeds at metal co‐catalysts such as Pt. This co‐catalyst acts both as a Schottky barrier that suppresses charge recombination and as a HER catalyst. PR in D2O has shown that the generated H2 originates from the aqueous solvent rather than the biomass.11, 31</p><p>BOR is a more complex multi‐step process that directly involves the substrate. PR rates with various substrates differ depending on the substrates' adsorption to the photocatalyst surface.11c, 12a, 13b, 28b, 42, 51 This is consistent with the Langmuir‐type kinetics observed for glucose PR on TiO2.13b, 15a Infrared (IR) spectroscopy,51a electron energy loss spectroscopy (EELS)51a and X‐ray absorption near edge structure (XANES)52 measurements confirm that glucose chemisorbs on TiO2. Improving this binding by changing the ionic strength,28b using α‐glucose instead of β‐glucose,53 or immobilizing the substrate42 enhances the PR rate. Chemisorption promotes electronic interactions such as substrate‐photocatalyst charge transfer,51a shifting the flat band potential11c, 12a and hole trapping at the substrate.54 BOR is therefore believed to involve direct hole transfer to the chemisorbed substrate (Figure 3 A),51b, 52, 54 generating surface‐bound radicals on the sub‐ns timescale, as evidenced for glucose by transient absorption spectroscopy (TAS)52 and electron paramagnetic resonance (EPR)55 spectroscopy. Fragmentation of these radicals leads to C−C bond cleavage starting from C1,55 resulting in a step‐wise degradation of glucose to arabinose, erythrose etc. with concomitant formic acid formation (Figure 3 B).13c Metal co‐catalysts can be involved in BOR, presumably by interaction with chemisorbed intermediates.51c</p><!><p>Mechanism of biomass PR on metal‐oxide surfaces. A) Mechanistic pathway depending on the substrate reproduced from Ref. 51b with permission from Elsevier. B) Mechanistic proposal for glucose reforming on TiO2 reproduced from Ref. 55 with permission from the ACS.</p><!><p>Alternatively, involvement of OH. radicals has been suggested15b, 30, 34a, 41 on the basis of spin‐trapping EPR experiments in the absence of biomass.14, 23, 29 However, biomass PR is known to proceed on photocatalysts incapable of generating OH. radicals.13c, 31</p><!><p>The low market value of H2 renders alternative PR products desirable and, consequently, the selective photocatalytic transformation of renewable feedstocks into valuable organic products is a field of intense research.56 The radical nature of glucose PR over M/TiO2 gives rise to a number of trace by‐products such as CO,12e, 14 CH4,14, 19, 22 formic acid16 and others.19 PR of cellulose or raw biomass over Pt/TiO2 generated traces of C2H6, ethanol and acetone.11b Polymorph‐dependent selectivity control was observed in glucose PR over Rh/TiO2. Rutile showed preferred decarboxylation of glucose to give arabinose and erythrose, while further oxidation to CO2 was suppressed.13c LaFeO3 produced only H2 and gluconate,34b because further oxidation was slow on the less oxidizing VB compared to TiO2. Impregnating Pt/TiO2 with cellulose promoted glucose, cellobiose and formic acid formation during PR.42 The produced glucose could be further photoreformed at Pt/TiO2 to hydroxymethyl furfural.41 Accumulation of formate was seen during cellulose PR at CdS/CdOx,31 as formic acid PR was slower than cellulose PR. Formic acid could be further photoreformed at CdS to H2 or CO.57</p><p>Alternatively, reducing equivalents generated upon biomass photo‐oxidation can be used for organic transformations instead of H2 generation. Photocatalytic conversion of glucose to arabinose and erythrose over Pd/TiO2 could be coupled with the reduction of nitroarenes and aldehydes to anilines and alcohols, respectively, thus producing high‐value products from both half‐reactions.58 This approach was recently adapted using lignin as both reductant and oxidant.59 Photo‐oxidation of lignin alcohol moieties to ketones with simultaneous reductive C−O bond cleavage in the lignin backbone resulted in an overall transfer hydrogenolysis of lignin to substituted phenols.</p><!><p>Biomass PR is a promising approach to sustainably generate fuels and feedstock chemicals. The simplicity of this room‐temperature process to produce clean H2 fuel is of considerable advantage over thermochemical methods, but efficiencies are yet to match conventional processes. This field has historically focused on materials and catalysts designed for solar water splitting, limiting photocatalytic activity to UV light. Future work should focus on designing narrow band‐gap materials specifically for biomass PR to enhance the performance under natural sunlight. Tailor‐made biomass oxidation catalysts will be needed to lower the required driving force and to improve the selectivity towards high‐value products. Ultimately, integrating PR with other solar fuel production systems by utilizing low‐energy photons unsuitable for water splitting may be the key to translate PR into a scalable and economically viable process.</p><!><p>A patent covering biomass photoreforming has been filed by Cambridge Enterprise (PCT/EP2017/080371) that names M.F.K. and E.R. as inventors.</p><!><p>Moritz F. Kuehnel received his PhD from the Freie Universität Berlin (with Dieter Lentz). He was awarded the Schering Prize for his doctoral thesis on carbon‐fluorine bond activation. After a postdoctoral stay at the HU Berlin (with Thomas Braun), he joined the group of Erwin Reisner (Cambridge) as a DFG fellow, before his promotion to Senior Postdoc. Recently, he started his independent career as a Chemistry Lecturer at Swansea University. His research encompasses the application of semiconductor nanocrystals for solar fuel production from biomass and CO2.</p><p></p><p>Erwin Reisner obtained his PhD at the University of Vienna (with Bernhard K. Keppler), followed by postdoctoral research at the Massachusetts Institute of Technology (with Stephen J. Lippard) and the University of Oxford (with Fraser A. Armstrong). He is currently the Professor of Energy and Sustainability in the Department of Chemistry at the University of Cambridge, head of the Christian Doppler Laboratory for Sustainable SynGas Chemistry, and director of the UK Solar Fuels Network. His group develops solar‐driven chemistry by combining chemical biology, synthetic chemistry and materials chemistry.</p><p></p>

PubMed Open Access

Necroptosis-Inducing Rhenium(V) Oxo Complexes

Rhenium(V) oxo complexes of general formula [ReO(OMe)(N^N)Cl2], where N^N = 4,7-diphenyl-1,10-phenanthroline, 1, or 3,4,7,8-tetramethyl-1,10-phenanthroline, 2, effectively kill cancer cells by triggering necroptsosis, a non-apoptotic form of cell death. Both complexes evoke necrosome (RIP1-RIP3)-dependent intracellular ROS production and propidium iodide uptake. The complexes also induce mitochondrial membrane potential depletion, a possible downstream effect of ROS production. Apparently, 1 and 2 are the first rhenium complexes to evoke cellular events consistent with programmed necrosis in cancer cells. Furthermore, 1 and 2 display low acute toxicity in C57BL/6 mice and reasonable stability in fresh human blood.

necroptosis-inducing_rhenium(v)_oxo_complexes

Introduction<!>Synthesis and Characterization<!>In Vitro Potency<!>Cellular Mechanism of Action and Mode of Cell Death<!>Characterization of Necroptotic Features<!>PARP-1- and p53-Independent Necroptosis<!>Cell Cycle Analysis<!>In Vivo Toxicity and Stability in Whole Human Blood<!>Conclusion<!>Material<!>Physical Measurements<!>Synthesis of [ReO(OMe)(4,7-diphenyl-1,10-phenanthroline)Cl2] (1)<!>Synthesis of [ReO(OMe)(3,4,7,8-tetramethyl-1,10-phenanthroline)Cl2] (2)<!>Electrochemistry<!>Cytotoxicity MTT assay<!>Reactivity of 1 and 2 with necrostatin-1<!>Intracellular ROS Assay<!>JC-1 Assay<!>Propidium Iodide Uptake<!>Fluorescence Microscopy<!>Immunoblotting Analysis<!>Cell Cycle

<p>Cancer is a leading cause of death and suffering throughout the world and continues to impose a huge socio-economic burden on society. According to the latest statistics released by the World Health Organization, an estimated 8.2 million cancer-related deaths occurred in 2012, a 0.6 million increase from the previous estimation in 2008.1 With the global cancer burden rising, the development of new cancer treatments is crucial. Since the discovery of their antineoplastic properties in 1969, platinum drugs such as cisplatin, oxaliplatin, and carboplatin have become a mainstay chemotherapy for cancer.2,3 Their use, however, is limited by side effects and inherent or acquired resistance.4–6 These limitations have driven the search for new treatments, including investigations of other transition metal compounds. Many ruthenium, osmium, titanium, copper, iron, and other metal compounds have been tested for their anticancer activity, and some of the most promising candidates have been studied clinically.7</p><p>Several rhenium-based compounds have been employed as in vitro and in vivo imaging agents, but in-depth studies exploring their anti-proliferative properties are relatively rare.8,9 The most active rhenium compounds reported to date contain Re(I)-carbonyl centers bound to mono-, di-, tri-dentate ligands.10–14 This class of compounds are proposed to induce their cytotoxic effects through covalent interactions with DNA bases and/or protein side chains. A number of photolabile rhenium(I) derivatives that trigger cell death upon irradiation have also been devised.15–17 These complexes offer temporal and spatial control over activation and therefore could be useful in overcoming systemic toxicity. Recently, octahedral Re(IV) complexes bearing polypyridyl ligands were discovered to exhibit potent in vitro antiproliferative activity against breast, ovarian, and prostate cancer cells.18 The complexes were hypothesized to interact with nuclear DNA upon hydrolysis, inducing apoptotic cell death. The anticancer potential of dinuclear rhenium compounds has also been investigated.19–21 In addition to displaying promising anticancer activity, paddle-wheel dirhenate(III) complexes have attractive pharmacological features such as low neuro-, hepato- and nephrotoxicity.21–24 Dirhenate(III) units with propionate ligands and quadruple Re-Re bonds have varying degrees of efficacy in sarcoma-, leukemia- and melanoma-bearing mice models.25 Subsequent studies found that the anticancer activity of the rhenium clusters depends on the identity and orientation of the ligands around the two rhenium(III) centers rather than the presence of a quadruple bond. Remarkably, dinuclear rhenium(III) analogues bearing alkylcarboxylates and zwitterionic aminocarboxylate ligands inhibited tumor growth by up to 60% in Wistar rats inoculated with tumor carcinoma Guerink (T8) cells. Furthermore when combined with cisplatin, the rhenium(III) clusters completely inhibited tumor progression.21–24,26 Dirhenate(III) complexes are also regarded as promising modulators of cisplatin. According to in vitro and in vivo studies, carboxylate-bridges dirhenate(III) complexes stabilized red blood cells (RBC) against haemolysis, thereby diminishing dose-limiting toxicity associated with cisplatin.21–24,27</p><p>Many clinically used anticancer agents act by targeting and damaging nuclear DNA, eventually leading to apoptosis.28–31 Cytotoxic compounds may also kill cells through non-apoptotic cell death pathways such as necrosis.32–34 Although necrosis was previously believed to be a random, unregulated process, it is now understood that programmed necrosis, also known as necroptosis, does occur in certain cell types.35 Necroptosis is activated by the formation of a complex between receptor interacting protein (RIP) kinases, RIP1 and RIP3.36 The RIP1-RIP3 complex, also known as a necrosome, is thought to modulate oxidative stress and generate mitochondrial reactive oxygen species (ROS) capable of inducing bioenergetics-related cell death.37 The relationship between ROS production and necroptotic cell death is poorly understood, although some reports link RIP1 and ROS-induced cell death.38 Owing to the persistent use of apoptosis-inducing anticancer drugs, many cancers have evolved resistance to apoptosis.39–41 Therefore chemotherapies capable of inducing non-apoptotic cell death such as necroptosis warrant further investigation. Here we present the synthesis, characterization, and cell-based studies of two necroptosis-inducing rhenium(V) oxo complexes.</p><!><p>The rhenium(V) oxo complexes 1 and 2 were prepared by the reaction of [ReOCl3(PPh3)2] with 1.5 equivalents of the corresponding bidentate ligand in methanol (Scheme 1). The complexes were isolated in reasonable yields as pale green solids and fully characterized by NMR and IR spectroscopy, and ESI-MS spectrometry. The purity of the complexes was confirmed by elemental analysis. Variable-temperature 1H NMR spectroscopic studies in DMSO revealed the complexes to be stable and remain intact at elevated temperatures (up to 75 °C, see Figure S1).</p><p>The cyclic voltammograms of 1 and 2 display only a single irreversible reduction within the potential window spanning −1.00 to 1.00 V versus Fc/Fc+ (Figure S2A & B). This event is ascribed to the two-electron reduction of the metal center in each complex to form, initially at least, a Re(III) product. The more negative cathodic peak potential of 2, −1.48 V, compared to that of 1, −1.32 V, is consistent with the more electron donating nature of the 3,4,7,8-tetramethyl-1,10-phenanthroline ligand, which better stabilizes the Re(V) oxidation state. To further rationalize the different redox potentials of the two compounds, DFT calculations were performed at the uPBEPBE/LANL2DZ level of theory in Gaussian 03.42 Solvent effects of DMSO were included implicitly through the self-consistent reaction field approach, as implemented in the default PCM model. The calculations suggested that both compounds preferentially adopt a low-spin configuration, which is significantly lowerer in energy than the corresponding high-spin one (Figure S2C & D). Partial atomic charges were calculated by means of Mulliken population analysis.43 As shown in Table S1, the calculated HOMO energy of 2 is higher than that of 1, which is consistent with observed redox potentials.44 The Mulliken charges on the Re(V) atom also predict that 2 has a lower redox potential than 1. These results can be attributed to the stronger electron-withdrawing effect of the phenyl group (χp = 2.49),45 which can presumably stabilize reduced 1. Overall, the electrochemical studies and calculations shows that the reduction potential of rhenium(V)-oxo complexes such as 1 and 2 can be tuned by subtle ligand modifications.</p><p>The lipophilicity of the rhenium(V)-oxo complexes (1 and 2) was determined by measuring the extent to which they partition between octanol and water, Po/w or P. The experimentally determined log P values are 1.20 for 1, and 0.95 for 2. The hydrophobic character of the rhenium(V)-oxo complexes suggests that they will be taken up well by cells.</p><!><p>In vitro effect of rhenium(V) oxo complexes 1 and 2 toward a panel of human cell lines was determined by the colorimetric MTT [[3-(4,5-dimethylthiazol-2-yl)-2,5-diphenyltetrazolium bromide] assay. Cisplatin was also included as a control. The IC50 values or concentration required to induce 50% cell death, were derived from dose-response curves and are summarized in Table 1. Compounds 1 and 2 display nanomolar potency toward cancer cells, with good selectivity over normal fibroblast cells (up to 10-fold). The cytotoxicity of the rhenium complexes was significantly higher than that of cisplatin for the cell lines tested. Notably, 1 displayed 34-fold higher toxicity for lung carcinoma A549 cells compared to cisplatin. Furthermore, the rhenium complexes are not cross-resistant with cisplatin, as demonstrated by their ability to kill cisplatin-resistant ovarian carcinoma cells (A2780CP70) with up to 15-times better efficacy than cisplatin-sensitive cells (A2780).</p><p>Given the ability of 1 and 2 to selectively kill ovarian cisplatin-resistant over sensitive cells, we evaluated their efficacy against other cisplatin-resistant cell lines derived from tissues such as HT-29 (colorectal adenocarcinoma), MDA-MB-231 (breast adenocarcinoma), PC-3 (prostate adenocarcinoma), MCF-7 (breast adenocarcinoma), and DU 145 (prostate carcinoma). The rhenium complexes displayed nanomolar and sub-micromolar toxicities toward the cisplatin-resistant cells (Table 2). Remarkably, 1 and 2 killed colorectal adenocarcinoma HT-29 cells over 300-fold better than cisplatin. Although cisplatin is one of the most successful broad-spectrum anticancer drugs in clinical use, several tumors exhibit resistance. A plethora of molecular mechanisms account for cisplatin resistance, including reduced intracellular accumulation, increased sequestration by scavengers, efficient DNA repair, and deregulation of proteins involved in the DNA damage and apoptotic cell death pathways.47 Therefore compounds such as 1 and 2, which can overcome cisplatin-resistance, hold significant therapeutic potential. To further investigate this potential, cytotoxicity studies were conducted with quiescent A549 cells (Table 2). The potency of 1 and 2 toward quiescent A549 cells was over 40-fold less than toward proliferating A549 cells. This result highlights the ability of 1 and 2 to selectively kill fast growing cancer cells.</p><!><p>To gain insight into how the rhenium complexes induce cell death, 1 and 2 were analyzed by a recently developed functional strategy employing a RNAi signature assay to predict the mechanism of cytotoxic drug action.48–50 This RNAi-based methodology relies on a fluorescence competition assay with lymphoma cells that are partially infected with eight distinct short hairpin RNAs (shRNAs). shRNA-bearing cells will either enrich or deplete relative to the uninfected population based on the survival advantage or disadvantage conferred by a given shRNA. The responses of these cells compose signatures, which have been obtained from all classes of clinically used cytotoxic agents. These signatures comprise a reference set which is then informatically classified by a probabilistic K-nearest neighbors algorithm to determine whether a new compound belongs to a class in the reference set or requires a new category not yet represented. Neither 1 nor 2 classified as belonging to any category of drug mechanism present in the reference set and thus represent novel mechanisms of drug action (Figure 1).</p><p>In order to determine the cell killing mechanism of 1 and 2, we carried out cytotoxicity studies in the presence of apoptosis and necrosis inhibitors. Upon addition of z-VAD-FMK, a potent inhibitor of caspase-mediated apoptosis,51 the ability of 1 and 2 to kill A549 cells remained largely unaltered, indicative of a non-apoptotic cell death program (Figure 2A). By contrast, the IC50 values for known apoptosis-inducing agents such as etoposide and cisplatin increased significantly (t test, p < 0.05) in the presence of the inhibitor (Figure 2A). Immunoblotting studies showed that proteins implicated in the apoptotic cell death pathway, namely, cleaved-caspase 7 and 9, were not detected in 1- and 2-treated A549 cells (100 – 400 nM for 72 h; Figure S3), providing further evidence for a non-apoptotic program. We next investigated the possibility that 1- and 2-induce necroptosis. Necroptosis is a well regulated mode of cell death that is different from unregulated necrosis and apoptosis.35 Unlike unregulated necrosis, which can be induced by H2O2 or heat, necroptosis relies on the interaction of protein kinases, RIP1 (receptor-interacting protein 1) and RIP3 (receptor-interacting protein 3), to initiate cell disintegration. This process can be blocked by necrostatin-1, a potent RIP1 kinase inhibitor.52,53 To determine whether 1 and 2 induced necroptosis and/or uncontrolled necrosis, cytotoxicity studies were conducted in the presence of necrostatin-1 (60 μM) and IM-54 (10 μM), an inhibitor of H2O2-induced necrosis.54 Co-incubation with necrostatin-1 markedly decreased the toxicity of 1 and 2 (t test, p < 0.05) against A549, PC-3, and HT-29 cells (Figures 2B and S4). A similar effect was also observed for shikonin, a naturally occurring compound known to induce necroptosis in certain cell types (Figure S4A).38 In contrast, co-treatment with IM-54 did not significantly affect the toxicity of 1 and 2 (Figure 2B). Taken together, the cytotoxicity data suggest that 1 and 2 induce RIP1-RIP3 (necrosome) mediated necroptosis, rather than uncontrolled necrosis or apoptosis. Immunoblotting studies revealed that the overall expressional level of RIP1 and RIP3 in A549 cells remained unchanged with increasing 1 and 2 dosages (Figure S3). Therefore 1- and 2-induced cell death relies on RIP1-RIP3 complex formation and not on the expressional levels of the individual protein kinases. RIP1 can also form a cytosolic complex with Fas-associated death domain (FADD) and caspase 8, known as a ripoptosome, to initiate apoptosis (through caspase 8 cleavage).55–57 Immunoblotting studies showed that FADD and cleaved caspase 8 expression remained unaltered with increasing 1 and 2 concentration (Figure S3), indicating that ripoptosome formation was not responsible for 1- and 2-induced cell death. This finding is consistent with the fact that 1- and 2-treated A549 cells do not undergo apoptosis.</p><!><p>Having established that necrosome formation is a determinant of 1 and 2 activity, we performed additional studies to understand the cascade of events leading from necrosome formation to cell death. Necrosomes generate abnormally high levels of mitochondrial reactive oxygen species (ROS),58–60 leading to ATP depletion and eventual degradation of the mitochondrial membrane potential.61,62 With this fact in mind, we quantified intracellular ROS production by flow cytometry using 6-carboxy-2′,7′-dichlorodihydrofluorescein diacetate (DCFH-DA), a well-established ROS indicator. A549 cells incubated with 1 and 2 (20 μM for 12 h) displayed markedly higher levels of ROS than untreated control cells (Figure 3A & B). A549 cells dosed with H2O2 (1 mM for 1 h, ROS-inducer) and shikonin (20 μM for 12 h, necroptosis-inducer) also exhibited significantly higher levels of measurable ROS than untreated cells (Figure 3C & 3D). Remarkably, 1- and 2-induced ROS production was attenuated in the presence of necrostatin-1 (60 μM) (Figure 3A & B), suggesting that the RIP1-RIP3 kinase complex plays a role in modulating intracellular ROS production.</p><p>The effect of 1 and 2 on the mitochondrial membrane potential was assessed by flow cytometry, using the JC-1 assay (5,5′,6,6′-tetrachloro-1,1′3,3′-tetraethyl benzimidazolyl carbocyanine iodide).63,64 JC-1 is a cationic lipophilic dye that localizes in the mitochondria of healthy cells as red-emitting aggregates. If the mitochondrial membrane potential is disrupted, JC-1 forms green-emitting monomers. A549 cells incubated with 1 and 2 (20 μM for 12 h) displayed increased green fluorescence compared to untreated cells, indicative of mitochondrial membrane disruption (Figure S5A & B). A similar result was observed for A549 cells dosed with carbonyl cyanide m-chlorophenyl hydrazone (CCCP) (5 μM for 12 h), a known mitochondrial membrane depolarizer (Figure S5C), and shikonin (20 μM for 12 h), a necroptosis inducing agent (Figure S5D). Notably, 1- and 2-induced mitochondrial membrane depletion was amplified with necrostatin-1, suggesting that 1 and 2 may inherently target mitochondria and induce mitochondrial dysfunction, independent of RIP1-RIP3 formation. This result could explain the high residual toxicity (>1 μM) observed for A549 cells co-incubated with the rhenium complexes (1 and 2) and necrostatin-1 (Figure 2B).</p><p>Intracellular ROS production and mitochondrial membrane depletion contribute to necroptosis.61,62 Cells undergoing necroptosis display necrosis-like morphological features such as loss of cell membrane integrity, increase in organelle and cell volume (oncosis), and intact nuclear membrane.65–67 To further test whether 1- and 2-treatment is able to trigger necroptosis, Hoechst 33258/propidium iodide (PI) double staining was carried out to determine nuclear membrane morphology and integrity. Hoechst 33258 is a DNA minor groove binder that is routinely used to visualize the nucleus without the need for cell permeabilization.68 When used without cell permeabilization agents, PI stains the nuclei of necrotic cells.69 Early stage apoptotic and normal cells maintain cell membrane integrity and thus are not stained by PI. A549 cells were treated with 1 and 2 (20 μM for 12 h), incubated with Hoechst 33258 and PI, and imaged using a fluorescent microscope. Untreated A549 cells exhibited bright blue nuclei, owing to Hoechst 33258 uptake (Figure 4A). Cells incubated with 1 and 2 displayed pink nuclei, owing to Hoechst 33258 and PI uptake, which is consistent with necroptosis (Figure 4B & C). Furthermore 1- and 2-treated cells showed clear signs of plasma membrane disintegration with undamaged nuclei. A549 cells co-incubated with 1 or 2 and necrostatin-1 (60 μM for 12 h) were unstained by PI suggesting that necrostatin-1 is able to block 1- and 2-induced necroptosis (Figure 4E & F). Overall, the microscopy data suggest that necrosome formation contributes to the necrosis-like morphological features observed upon 1- and 2-treatment. To further validate this result, A549 cells were treated under the same conditions as above, stained with PI, and analyzed by flow cytometry. Complementary to the microscopy results, 1- and 2-treated cells expressed higher PI uptake compared to untreated control cells, indicative of necrotic cell death (Figure S6A & B). The flow cytometry data also showed that necrostatin-1 could block 1- and 2-mediated PI uptake. Additional studies showed that pre-treatment of A549 cells with N-acetylcysteine (3mM for 1 h), a ROS inhibitor, significantly decreased 1- and 2-induced PI uptake (Figure S6C & D). This result suggests that intracellular ROS generation is an integral part of the necroptotic mechanism of action of 1 and 2.</p><!><p>Apart from necrosome formation, necroptosis can also result from the overactivation of poly(ADP–ribose) polymerase (PARP-1).66,67 PARP-1 is a nuclear enzyme that is involved in DNA repair and transcriptional regulation.70 DNA damage can trigger PARP-1 activity, resulting in ATP and NAD depletion and bioenergetic-mediated cell death.71 To determine whether PARP-1 activity is a factor in 1- and 2-mediated cell death, cytotoxicity assays were conducted with wild-type mouse embryonic fibroblast cells (MEFs PARP-1+/+) and the corresponding PARP-1-null cells (MEFs PARP-1−/−). The IC50 values for 1 and 2 were similar for MEFs PARP-1+/+ and MEFs PARP-1−/− cells, indicating that 1 and 2-induced necroptosis is independent of PARP-1 function (Figure 5A). This result is consistent with immunoblotting studies, which revealed that 1- and 2-treatment did not upregulate canonical markers of DNA damage such as the phosphorylated forms of H2AX γH2AX) and CHK2 (Figure S7). Recently, p53 has also been reported to play a role in necroptosis. p53 induces cathepsin Q, a lysosomal protease that cooperates with ROS to execute necrosis.72 To investigate whether p53 might play a role in 1- and 2-mediated necroptosis, cytotoxicity studies were conducted with HCT116 p53+/+ and HCT116 p53−/− cells. The potency of 1 and 2 was statistically similar for HCT116 p53+/+ and HCT116 p53−/− cells, indicating that 1 and 2 induce necroptosis in a manner that is independent of p53 (Figure 5B). This concludion is consistent with the RNAi signatures, which reveal that p53 is not important in the cellular response evoked by the complexes, especially for 1. Apart from the implications of this result on the mechanism of action of 1 and 2, it is clinically very appealing because p53 is mutated, defective, or inactivated in several chemoresistant cancers.</p><!><p>To gain a more complete understanding of the cellular response evoked, DNA-flow cytometric studies were conducted to determine the effect of 1 and 2 on the cell cycle. A549 cells were treated with 1 or 2 (2 μM) and the cell cycle was determined over the course of 72 h (Figure S8). After 24 h treatment, both complexes stalled the cell cycle at the G1-phase. Cells treated with 1 remained stalled at the G1-phase after 48 h. Upon further incubation (72 h), large populations of cell debris were detected (32%), indicative of cell death. Cells incubated with 2 for 48 h and 72h also displayed large populations of debris (26 and 38% respectively). G1-phase cell cycle arrest followed by immediate cell death is characteristic of programmed necrosis.73,74</p><!><p>Given the impressive in vitro data acquired for 1 and 2, an in vivo study was conducted with C57BL/6 mice to determine acute toxicity and possible side effects. Single doses of 1 and 2 (3, 7, 11, 15, 20, 36 mg/ kg) in saline, were administered by intra-peritoneal injection. The mice were then monitored for signs of pain, distress, and weight loss for 6 days post-treatment. The compounds exhibited no toxicity in mice, as gauged by a lack of weight loss after treatment. The change in weight of mice after a single dose of 1 and 2 at the maximum solubility of the complexes (36 mg/ kg) is depicted in Figure S9. A single 30 mg/ kg dose of cisplatin induces acute nephrotoxicity in C57BL/6.75 The in vivo data highlight the relatively low toxicity of 1 and 2 compared to cisplatin in C57BL/6 mice. The pharmacological toxicity profile of 1 and 2 is very appealing in terms of further pre-clinical studies.</p><p>The stability of biologically active compounds in human blood is vitally important for their potential application in clinical settings. We therefore investigated the stability of 1, in whole human blood using a recently developed protocol.76 This method exploits the ability of octanol to extract hydrophobic metal complexes such as 1. The rhenium complex, 1 (500 μM) was incubated with fresh human blood at 37 °C and aliquots were extracted into octanol at various time points. The amount of 1 in the octanol extracts (corresponding to unreacted 1) was measured by graphite furnace atomic absorption spectroscopy (GFAAS). The data present in Figure S10, revealed that the half-life of 1 in human blood is 29.1 min, comparable to that reported for cisplatin (t1/2 = 21.6 min).77</p><!><p>Two rhenium(V) oxo complexes were prepared and their in vitro properties were investigated. The complexes selectively kill cancer cells over normal cells and display markedly higher cell toxicity than cisplatin. Remarkably, 1 and 2 display two orders of magnitude higher toxicity than cisplatin against colorectal adenocarcinoma cells. Cells treated with 1 and 2 displayed features consistent with programmed necrosis (necroptosis), including RIP1-RIP3 dependent intracellular ROS production, cell membrane disruption, PI uptake, mitochondrial damage, and G1 cell cycle arrest. Given the inherent and/or acquired resistance of tumors toward apoptosis-inducing chemotherapies, compounds such as 1 and 2, capable of killing cancer cells through necroptosis, are highly sought after when selecting preclinical drug candidates for chemoresistant malignancies.</p><!><p>All synthetic procedures were performed under normal atmospheric conditions without exclusion of oxygen or moisture. The bidentate aromatic ligands 4,7-diphenyl-1,10-phenanthroline and 3,4,7,8-tetramethyl-1,10-phenanthroline were purchased from Sigma Aldrich and used as received. ReOCl3(PPh3)2 was prepared as previously reported.78 The synthesis of 1 has been reported previously,79 but the procedure reported here is novel. Analytical grade acetone and dichloromethane were used as solvents.</p><!><p>NMR measurements were recorded on a Bruker 400 MHz spectrometer in the MIT Department of Chemistry Instrumentation Facility (DCIF) at 20 °C. 1H and 13C{1H} NMR chemical shifts were referenced internally to residual solvent peaks or relative to tetramethylsilane (SiMe4, δ = 0.00 ppm). Fourier transform infrared (FTIR) spectra were recorded with a ThermoNicolet Avatar 360 spectrophotometer upon preparation of the samples as KBr disks. The spectra were analyzed using the OMNIC software. Graphite furnace atomic absorption spectrometry was carried out using a Perkin-Elmer AAnalyst600 GFAAS.</p><!><p>[ReOCl3(PPh3)2] (63.8 mg, 0.08 mmol) was suspended in methanol (15 mL) and heated to 50 °C. To this mixture was added a methanolic solution (5 mL) of 7-diphenyl-1,10-phenanthroline (34.0 mg, 0.10 mmol). The resultant mixture was heated under reflux for 24 h to give a deep purple solution with a pale green precipitate. The precipitate was filtered, washed with hot methanol, cold methanol, and diethyl ether. The rhenium(V) oxo complex was isolated as a pale green solid. Yield: 19.8 mg (37%). Mp 247 °C (dec). 1H NMR (400 MHz, DMSO-d6): δ 10.08 (d, 2H), 8.48 (d, 2H), 8.31 (s, 2H), 7.83 (m, 4H), 7.73 (m, 6H), 2.56 (s, 3H). IR (KBr, cm−1): 941.51 (Re=O), 508.42 (Re-OMe). ESI-MS (MeOH/ DMSO): m/z. 605.0 ([M-OMe]+, calcd. 605.0). Anal. Calcd. for 1, C25H19Cl2N2O2Re: C, 47.17; H, 3.01; N, 4.40. Found: C, 46.79; H, 3.05; N, 4.35.</p><!><p>[ReOCl3(PPh3)2] (50.0 mg, 0.06 mmol) was suspended in methanol (15 mL) and heated to 50 °C. To this mixture, a methanolic solution (5 mL) of 3,4,7,8-tetramethyl-1,10-phenanthroline (21.85 mg, 0.09 mmol) was added. The resultant mixture was heated under reflux for 24 h to give a deep purple solution with a pale green precipitate. The precipitate was filtered, washed with hot methanol, cold methanol, and diethyl ether. The rhenium(V) oxo complex was isolated as a pale green solid. Yield: 14.8 mg (41%). Mp > 276 °C (gradual darkening and decomposition). 1H NMR (400 MHz, DMSO-d6): δ 9.69 (s, 2H), 8.60 (s, 2H), 2.99 (s, 6H), 2.74 (s, 6H), 2.36 (s, 3H). IR (KBr, cm−1): 954.85 (Re=O), 492.85 (Re-OMe). ESI-MS (MeOH/ DMSO): m/z. 509.0 ([M-OMe]+, calcd. 509.0). Anal. Calcd. for 2, C17H19Cl2N2O2Re: C, 37.78; H, 3.54; N, 5.18. Found: C, 37.76; H, 3.63; N, 4.99.</p><!><p>Electrochemical experiments were performed at room temperature using a VersaSTAT3 potentiostat from Princeton Applied Research operated with the V3 Studio software. Cyclic voltammetry was performed using a three-electrode system comprising a glassy carbon working electrode, a platinum wire auxiliary electrode, and a silver wire pseudo-reference electrode. Potentials were referenced internally to the ferrocene/ferrocenium redox couple. Solutions of 1 and 2, at concentrations of approximately 2 mM, were prepared in DMSO containing 0.2 M tetra-n-butylammonium hexafluorophosphate. All solutions were sparged with nitrogen for 5 min prior to measurement. Voltammograms were obtained on quiescent solutions at a scan rate of 200 mV/s.</p><!><p>The colorimetric MTT assay was used to determine the toxicity of 1, 2, and cisplatin. Cells (2 × 103 cells/well) were seeded in a 96-well plate. After incubating the cells overnight, various concentrations of platinum as determined by GF-AAS (0.3–100 μM) were added and incubated for 72 h (total volume 200 μL). Cisplatin was prepared as a 5 mM solution in PBS and diluted using media. 1 and 2 were prepared as 10 mM solutions in DMSO and diluted using media. The final concentration of DMSO in each well was 0.5% and this amount was also present in the untreated control. After 72 h, the medium was removed, 200 μL of a 0.4 mg/mL solution of MTT in DMEM, RPMI or, McCoy's 5A was added, and the plate was incubated for an additional 1–2 h. The DMEM/MTT, RPMI/MTT, or McCoy's 5A/MTT mixture was aspirated and 200 μL of DMSO was added to dissolve the resulting purple formazan crystals. The absorbance of the solution wells was read at 550 nm. Absorbance values were normalized to DMSO-containing control wells and plotted as the concentration of test compound versus % cell viability. IC50 values were interpolated from the resulting dose dependent curves. The reported IC50 values are the average from at least three independent experiments, each of which consisted of six replicates per concentration level.</p><p>For specific cell death inhibitor assays, inhibitors of necroptosis (necrostatin-1, 60 μM), H2O2-induced necrosis (IM-54, 10 μM), apoptosis (v-VAD-FMK, 5 μM) were added to cells and incubated for 1 h prior to treatment with the test compounds.</p><!><p>Mixing the rhenium(V)-oxo complexes (1 and 2) (20 μM) with necrostatin-1 (60 μM) in DMSO and cell culture media (DMEM, RPMI, and McCoy's 5A) did not result in a precipitate. Incubation of the rhenium(V)-oxo complexes (1 and 2) with necrostatin-1 (1:3 ratio) for up to 6 hr in DMSO-d6 did not lead to a chemical reaction, as determined by 1H NMR analysis (Figure S11 and S12). Despite the presence of sulfur and nitrogen atoms in necrostatin-1, the 1H NMR spectra unequivocally prove that the reactivity/bioactivity of 1 and 2 is not compromised by necrostatin-1.</p><!><p>Untreated and treated A549 cells (1.5 × 106 cells/ well) grown in 6-well plates were incubated with 6-carboxy-2′,7′-dichlorodihydrofluorescein diacetate (20 μM) for 30 min. The cells were then washed with PBS (1 mL × 3), harvested, and analyzed by using the FACSCalibur-HTS flow cytometer (BD Biosciences) (20,000 events per sample were acquired). The FL2 channel was used to assess intracellular ROS levels. Cell populations were analyzed using the FlowJo software (Tree Star).</p><!><p>The JC-1 Mitochondrial Membrane Potential Assay Kit (Cayman) was used. The manufacturer's protocol was followed to carry out this experiment. Briefly, to untreated and treated A549 cells (1.5 × 106 cells/ well) grown in 6-well plates was added the JC-1 staining solution (100 μL/ mL of cell media). The cells were incubated for 30 min, harvested, and analyzed by using the FACSCalibur-HTS flow cytometer (BD Biosciences) (20,000 events per sample were acquired). The FL2 channel was used to assess mitochondrial depolarization. Cell populations were analyzed using the FlowJo software (Tree Star).</p><!><p>Untreated and treated A549 cells (1.5 × 106 cells/ well) grown in 6-well plates, were washed with PBS (1 mL × 3), harvested, incubated with propidium iodide (5 μM), and analyzed by using the FACSCalibur-HTS flow cytometer (BD Biosciences) (20,000 events per sample were acquired). The FL2 channel was used to assess intracellular PI uptake. Cell populations were analyzed using the FlowJo software (Tree Star).</p><!><p>A549 cells (1.5 × 106 cells/ well) were incubated with and without 1 and 2 (20 μM) for 12 h. The media were then removed and the cells were washed with additional media (2 mL × 2). After incubation of the cells with more media containing Hoechst 33258 (7.5 μM) and propidium iodide (5 μM), the nuclear regions were imaged using a fluorescent microscope. Fluorescence imaging experiments were performed using a Zeiss Axiovert 200M inverted epifluorscence microscope with a Hamamatsu EM-CCD digital camera C9100 and a MS200 XY Piezo Z stage (Applied Scientific Instruments, Inc.). An X-Cite 120 metal halide lamp (EXFO) was used as the light source. Zeiss standard filter sets 49 was employed for imaging the nuclear region. The microscope was operated with Volocity software (version 6.01, Improvision). The exposure time for acquisition of fluorescence images was kept constant for each series of images at each channel.</p><!><p>A549 cells (1.5 × 106 cells/ well) grown in 6-well plates were incubated with 1 and 2 (concentrations, μM) for 72 h at 37 °C. Cells were washed with PBS, scraped into SDS-PAGE loading buffer (64 mM Tris-HCl (pH 6.8)/ 9.6% glycerol/ 2% SDS/ 5% β-mercaptoethanol/ 0.01% Bromophenol Blue), and incubated at 95 °C for 10 min. Whole cell lysates were resolved by 4–20 % sodium dodecylsulphate polyacylamide gel electrophoresis (SDS-PAGE; 200 V for 25 min) followed by electrotransfer to polyvinylidene difluoride membrane, PVDF (350 mA for 1 h). Membranes were blocked in 5% (w/v) non-fat milk in PBST (PBS/0.1% Tween 20) and incubated with the appropriate primary antibodies (Cell Signalling Technology and Santa Cruz). After incubation with horseradish peroxidase-conjugated secondary antibodies (Cell Signalling Technology), immuno complexes were detected with the ECL detection reagent (BioRad) and analyzed using an Alpha Innotech ChemiImager™ 5500 fitted with a chemiluminescence filter.</p><!><p>In order to monitor the cell cycle, flow cytometry studies were carried out. A549 cells (1.5 × 106 cells/ well) grown in 6-well plates were incubated with and without the test compounds for 24, 48, and 72 h at 37 °C. Cells were harvested from adherent cultures by trypsinization and combined with all detached cells from the incubation medium to assess total cell viability. Following centrifugation at 1000 rpm for 5 min, cells were washed with PBS and then fixed with 70% ethanol in PBS. Fixed cells were collected by centrifugation at 2500 rpm for 3 min, washed with PBS, and centrifuged as before. Cellular pellets were re-suspended in 50 μg/mL propidium iodide (Sigma) in PBS for nucleic acids staining and treated with 100 μg/mL RNaseA (Sigma). DNA content was measured on a FACSCalibur-HTS flow cytometer (BD Biosciences) using laser excitation at 488 nm and 20,000 events per sample were acquired. Cell cycle profiles were analyzed using the ModFit software.</p>

PubMed Author Manuscript

Ferromagnetic Nanoscale Electron Correlation Promoted by Organic Spin-Dependent Delocalization

We describe the electronic structure and the origin of ferromagnetic exchange coupling in two new metal complexes, NN-SQ-CoIII(py)2Cat-NN (1) and NN-Ph-SQ-CoIII(py)2Cat-Ph-NN (2) (NN = nitronylnitroxide radical, Ph = 1,4-phenylene, SQ = S = 1/2 semiquinone radical, Cat = S = 0 catecholate, and py = pyridine). Near-IR electronic absorption spectroscopy for 1 and 2 reveals a low energy optical band that has been assigned as a \xce\xa8u \xe2\x86\x92 \xce\xa8g transition involving bonding and antibonding linear combinations of delocalized dioxolene (SQ/Cat) valence frontier molecular orbitals. The ferromagnetic exchange interaction in 1 is so strong that only the high-spin quartet state (ST = 3/2) is thermally populated at temperatures up to 300 K. The temperature-dependent magnetic susceptibility data for 2 reveals that an excited state spin doublet (ST = 1/2) is populated at higher temperatures, indicating that the phenylene spacer modulates the magnitude of the magnetic exchange. The valence delocalization within the dioxolene dyad of 2 results in ferromagnetic alignment of two localized NN radicals separated by over 22 \xc3\x85. The ferromagnetic exchange in 1 and 2 results from a spin-dependent delocalization (double exchange type) process and the origin of this strong electron correlation has been understood in terms of a valence bond configuration interaction (VBCI) model. We show that ferromagnetic coupling promoted by organic mixed-valency provides keen insight into the ability of single molecules to communicate spin information over nanoscale distances. Furthermore, the strong interaction between the itinerant dioxolene electron and localized NN electron spins impacts our ability to understand the exchange interaction between delocalized electrons and pinned magnetic impurities in technologically important dilute magnetic semiconductor materials. The long correlation length (22 \xc3\x85) of the itinerant electron that mediates this coupling indicates that high-spin \xcf\x80-delocalized organic molecules could find applications as nanoscale spin-polarized electron injectors and molecular wires.

ferromagnetic_nanoscale_electron_correlation_promoted_by_organic_spin-dependent_delocalization

INTRODUCTION<!>General Considerations<!>Preparation of Quinone-Nitronylnitroxide (Q-NN)<!>Preparation of CoIII(py)2(Cat-NN)(SQ-NN) (1)<!>Preparation of CoIII(py)2(Cat-Ph-NN)(SQ-Ph-NN) (2)<!>Magnetism<!>NIR Electronic Absorption Spectroscopy<!>X-Ray Crystallography<!>Electronic Structure Calculations<!>X-ray Crystallography<!>Electronic Absorption Spectroscopy<!>Magnetic Susceptibility Measurements<!>Delocalization, the Nature of the Low-Energy Charge Transfer Band, and Interstate Vibronic Coupling<!>Magnetic Exchange Coupling<!>Solvent Dependence of IVCT/\xce\xa8u \xe2\x86\x92 \xce\xa8g NIR band<!>Spin-dependent Delocalization<!>DISCUSSION<!>CONCLUSIONS<!>

<p>The exchange interaction between localized magnetic moments via delocalized conduction band electrons can lead to long-range magnetic order and ferromagnetic spin alignment in solid-state materials.1–4 This delocalized electron-, or carrier-, mediated ferromagnetism also figures prominently in spintronic materials including dilute magnetic semiconductors (DMS). There have been considerable theoretical and experimental studies on DMS materials with the goal of understanding the nature of the ferromagnetic exchange interactions between doped magnetic impurity ions mediated by itinerant electron(s).5–7 The mechanism of nanoscale electron correlation leading to ferromagnetic exchange between a delocalized electron and localized magnetic moments is complex and not completely understood. Notably, long-range ferromagnetic exchange between localized spins mediated by delocalized electrons has not been observed in molecular systems. Here we show that long-range electron correlation of localized radical spins, leading to ferromagnetic spin alignment, can be promoted by organic mixed-valency and spin-dependent electron delocalization. A high-spin ST = 3/2 ground state is found for NN-SQ-CoIII(py)2Cat-NN (1) (py = pyridine), which possesses mixed-valent catecholate (Cat) and radical semiquinone (SQ) ligands elaborated with localized nitronylnitroxide (NN) radicals. The ferromagnetic exchange is so strong in this complex that there exists no evidence of low-spin ST = 1/2 states being populated at room temperature. In addition, electron delocalization of the itinerant electron within the mixed-valent SQ/Cat dyad in NN-Ph-SQ-CoIII(py)2Cat-Ph-NN (2) (Ph = 1,4-phenylene) results in a remarkable ferromagnetic alignment of localized NN radical spins separated by distances greater than 22 Å. These results clearly indicate that very long-range electron correlation can result in the transfer of spin information over nanoscale distances in magnetically dilute metallo-organic hybrid and related systems. Importantly, the itinerant electron in 1 and 2 couples with the localized NN moments via an excited state donor-acceptor charge transfer configuration. That this occurs in a molecular system is of extreme interest since similar charge transfer mechanisms have been used to explain high Tc ferromagnetism in DMSs.6,7</p><!><p>Elemental Analyses (C, H, and N) were performed by Atlantic Microlab, Inc, Norcross GA. Infrared spectra were collected on a Perkin Elmer Spectrum RX-1 FT-IR Spectrometer for samples cast as a film from CH2Cl2 on a NaCl plate. X-band EPR spectra were recorded on an IBM-Brüker E200SRC continuous-wave spectrometer in fluid solution. Samples were dissolved in CH2Cl2 solution and spectra collected in quartz sample tubes. Fluid solution EPR spectra were simulated using WinSim to obtain accurate hyperfine coupling constants (aN).8 Toluene and CH2Cl2 were collected from an alumina column solvent purification system.9 Q-Ph-NN and Cat-NN were prepared as described previously.10,11 Lead dioxide and pyridine were purchased from Sigma Aldrich and used as received without further purification. Dicobalt octacarbonyl was purchased from Strem Chemicals and used as received without further purification.</p><!><p>A 100 mL round bottom flask containing Cat-NN (102 mg, 0.32 mmol) and 75 mL of CH2Cl2 is stirred at room temperature under nitrogen. An excess of PbO2 is added (ca. 500 mg, 2.09 mmol) as a black solid, changing from a dark blue to a dark green solution. The mixture is stirred under nitrogen for 3 hours to ensure completion. Excess PbO2 was removed via filtration and the solution evaporated to afford a green residue that was found to be unstable to atmosphere and degraded over time. The product was therefore carried through to the next step without purification. IR (film from CH2Cl2) ν in cm−1: 2961, 2940, 2873, 1775, 1736, 1659, 1615, 1553, 1454, 1415, 1372, 1304, 1272, 1241, 1222, 1166, 1142, 1093, 977, 900, 876, 796, 734, 693, 652. EPR: fluid solution ca. 0.2 mM in CH2Cl2, aN = 7.31 G.</p><!><p>A 100 mL Schlenk flask containing Q-NN (101 mg, 0.32 mmol) and 40 mL of a 1:1 CH2Cl2/toluene solution is stirred at room temperature under nitrogen atmosphere. A solution of dicobalt octacarbonyl (27 mg, 0.08 mmol) in 15 mL of 1:1 CH2Cl2/toluene is added via cannulation followed by pyridine (0.03 mL, 0.32 mmol) added neat via syringe. The dark blue-green solution is heated to ca. 35°C, shielded from light, and left to stir for 6 hours. Concentration under reduced pressure precipitates a dark blue solid (60 mg, 44%). IR (film from CH2Cl2) ν in cm−1: 6091, 3573, 2951, 2865, 1769, 1660, 1598, 1575, 1447, 1429, 1396, 1373, 1281, 1212, 1143, 1000, 871, 831, 765, 730, 693. Analysis for C44H56CoN6O8·C6H12·2 CH2Cl2: Calc. C, 56.27; H, 6.54; N, 7.57. Found C, 56.03; H, 6.47; N, 7.67.</p><!><p>A 100 mL Schlenk flask containing Q-Ph-NN (96 mg, 0.24 mmol) and 35 mL of a 1:1 CH2Cl2/toluene solution is stirred at room temperature under nitrogen atmosphere. A solution of dicobalt octacarbonyl (21 mg, 0.06 mmol) in 10 mL of 1:1 CH2Cl2/toluene is added via cannulation followed by pyridine (0.02 mL, 0.24 mmol) added neat via syringe. The dark blue-green solution is heated to ca. 35°C, shielded from light, and left to stir for 5 hours. Concentration under reduced pressure precipitates a dark blue solid (137 mg, 56%). IR (film from CH2Cl2) ν in cm−1: 5941, 3563, 2941, 2588, 1597, 1476, 1450, 1417, 1384, 1362, 1274, 1211, 1132, 1073, 1006, 973, 926, 872, 830, 763, 734, 696, 637. Analysis for C56H64CoN6O8·C6H12·CH2Cl2: Calc. C, 64.28; H, 6.68; N, 7.14. Found C, 64.34; H, 6.47; N, 6.89.</p><!><p>Magnetic susceptibilities were measured on a Quantum Design MPMS-XL7 SQUID Magnetometer using an applied field of 0.7 T for Curie plots. Microcrystalline samples were prepared by two methods: 1) samples were loaded into gel cap/straw sample holders and mounted to the sample rod with Kapton tape; 2) samples were loaded into a Delrin sample holder and screwed into the sample rod. Data from the gelcap samples were corrected for the sample container and molecular diamagnetism using Pascal's constants as a first approximation. The data was further corrected for inaccuracies in the diamagnetic correction by incorporation of a straight line into the overall fit expression, the slope of which represents the residual diamagnetic correction. The data for the Delrin samples were corrected for molecular diamagnetism through Pascal's constants and for the inherent diamagnetism of the sample holder by a background scan subtraction.</p><!><p>Spectra were collected on a Perkin Elmer Spectrum RX-1 FT-IR Spectrometer for samples cast as evaporated thin-films from CH2Cl2 on a NaCl plate. Solutions of 1 and 2 were prepared using four different solvents (1,2-Cl2C2H4, CH2Cl2, CCl4 and CS2, see Supporting Information) and loaded into a KBr Cavity Cell for sample collection. Pure solvent spectra were collected for background correction.</p><!><p>X-Ray crystallography for complex 1 was conducted at the X-Ray Structural Facility at North Carolina State University. Data collection and refinement for complex 2 was conducted at the X-Ray facility at the University of Michigan. A complete description of the data collected is reported in the Supporting Information.</p><!><p>Spin unrestricted gas phase geometry optimizations for compounds 1 and 2 were performed at the density functional level of theory using the Gaussian 03W software package.12 All calculations employed the B3LYP hybrid functional. A 6–311G++ basis set, including diffuse and polarizability functions, was used for all atoms. Input files were prepared using the molecule builder function in the Gaussview software package. Frontier molecular orbitals (MO) were generated for the optimized ground states and the contributions of each MO were further analyzed using the program AOMix. Time-dependent DFT calculations were performed on the optimized ground state geometries, and the first 40 excited states were calculated. Electron density difference maps (EDDMs) were constructed using the GaussSum suite of programs.</p><!><p>Schematic structures of both 1 and 2 are depicted in Figure 1 along with ORTEP drawings of their X-ray structures (see Supporting Information for crystallographic details). Compound 1 exists as two crystallographically independent molecules in the unit cell. Important bond lengths for 1 and 2 are compared with those of bona fide SQ and Cat ligands in Table 1. The data in Table 1 show that both the dioxolene (dioxolene = SQ/Cat) C-O and C-C bond lengths for 1 and 2 are roughly the average of those observed in TpCum,MeCoII(3,5-DBSQ)13 and MnIV(py)2(3,5-DBCat)2,14 suggesting Class III15 delocalization in the SQ/Cat dyad. However, crystallographically-imposed inversion centers coincident with the molecular C2 axes are found in the structures of both 1 and 2, and give the impression that the SQ/Cat dyads are completely delocalized Class III species.15 Therefore, X-ray crystallography cannot distinguish between averaged bond lengths due to positional disorder and averaged bond lengths due to electron (SQ/Cat) delocalization. A close inspection of the bond lengths within the dioxolene rings (Table 1) and comparison to typical values for SQ and Cat merely confirm that the complex exists in the [CoIII(diox)2]0 oxidation state.</p><!><p>We obtained IR-NIR electronic absorption spectra of 2 in four solvents and as an evaporated thin film cast from CH2Cl2 (see Supporting Information for additional spectra of 1 and 2). The CS2 solution and thin film spectra for 2 are presented in Figure 2. These spectra display a low-energy band maximum in the 3,300–3,600 cm−1 range with absorption intensity that extends out to ~6,000 cm−1. Although instrument limitations preclude the collection of higher quality spectra, the absorption envelope appears to include some vibronic structure. Similar, albeit less structured, low-energy electronic absorption bands have been observed in other SQ-CoIII(diimine)Cat complexes,16 including the related NN-Ph-SQ-CoIII(bpy)Cat-Ph-NN,17 that display effective C2 symmetry within the Co(diox)2 core and an ~45° dihedral angle between the SQ and Cat planes. In all of these cases, the IR-NIR absorption band has been assigned as a classic Cat→SQ intervalence charge transfer (IVCT) band, and the complexes have been described as being in the Robin and Day Class II limit since the degree of electron delocalization in solution was not able to be directly determined.18–20</p><!><p>The nature of the ground spin state, radical-radical exchange couplings, and the ground state electronic structure have been probed by variable temperature magnetic susceptibility studies. The variable temperature magnetic susceptibility data for 1 and 2 were collected from 2 – 300 K in an applied magnetic field of 0.7 T, and the data are plotted as the product of the paramagnetic susceptibility and the temperature (χpara•T) in Figure 3. It is interesting to note that the χpara•T product for 1 is temperature independent above 50 K and displays Curie-Weiss behavior, consistent with an ST = 3/2 paramagnetic and indicating that only a single spin state is populated at room temperature (Figure 3A). In contrast to the behavior of 1, the χpara•T data for 2 (Figure 3B) decreases for temperatures above ~50 K, and this is indicative of lower multiplicity spin states being populated at these elevated temperatures. The maximum value of χpara•T for both 1 and 2 is very close to 1.88 emu·K·mol−1, the theoretical χpara•T value for an ST = 3/2 ground state. An ST = 3/2 ground state can only result from dominant pairwise ferromagnetic exchange coupling between SQ and the two NN spin centers in these molecules. The χpara•T product is observed to decrease at temperatures below ~50 K for both 1 and 2. This is a characteristic feature of these complexes and most likely results from intermolecular interactions that are treated using the Weiss correction. Zero-field splitting (ZFS) of an S > 1/2 ground state could also result in a reduction of the χpara•T product at low temperatures. However, the spin-orbit coupling constants of the second row atoms that constitute the NN and SQ radicals are very small. This results in a reduced out-of-state spin-orbit coupling between the ground and excited states, and a negligibly small ZFS that will be dominated by dipolar terms.11</p><!><p>Complexes 1 and 2 both possess planar Co(diox)2 cores with an effective C2h symmetry. This is in marked contrast to NN-Ph-SQ-CoIII(bpy)Cat-Ph-NN, which possesses effective C2 symmetry and a distinctly nonplanar Co(diox)2 core with a ~45° dihedral angle between the dioxolene planes.17 Due to the planarity of the Co(diox)2 core in 1 and 2, we hypothesized that the degree of electron delocalization should be notably larger than that observed in NN-Ph-SQ-CoIII(bpy)Cat-Ph-NN and other C2 type CoIII(diimine)(Cat)(SQ) systems. Electronic structure calculations performed on the high spin (ST = 3/2) ground state indicate planar Co(diox)2 cores and a complete delocalization of the positive spin density across the entire molecule in both 1 and 2 (Figure 4). This differs from the degree of delocalization determined for C2-symmetric NN-Ph-SQ-CoIII(bpy)Cat-Ph-NN, which was estimated to be < 15% from the ratio of the magnetic exchange coupling parameters, JNN(B)-SQ(A)/JNN(A)-SQ(A) (vide infra, see Supporting Information), where A and B indicate NN and dioxolene sites on the left and right hand side of the molecule, respectively. Taken together, these results indicate that the planarity of the Co(diox)2 core in both 1 and 2 may facilitate a considerably greater degree of electron delocalization compared to their C2 analogues.</p><p>For Class III valence delocalized species, a spin- and dipole-allowed optical transition should be observed with an energy that is directly related to the magnitude of the electronic coupling matrix element, HAB. In the intermediate Class II delocalization limit, this is described as a classical IVCT transition involving a one electron promotion from an orbital, φB, localized on Cat to an acceptor orbital, φA, localized on SQ.15,21 However, in the limit of complete Class III electron delocalization, this transition is a bonding to antibonding Ψu → Ψg transition, where Ψu,g = 2−1/2 (φA ± φB) (Figure 5).15,22,23</p><p>Time dependent DFT (TDDFT) calculations performed on the delocalized ground states of 1 and 2 indicate an intense Ψu → Ψg transition at 3,700 cm−1, in very good agreement with the experimental 3,500 cm−1 absorption band maximum that we observe for both 1 and 2 (Figure 2; see Supporting Information for the spectrum of 1). In the effective C2h symmetry of the delocalized Co(diox)2 core present in 1 and 2, the Ψu2Ψg1 configuration yields a state of 4Au symmetry (considering also the symmetries of the two half-filled NN-based orbitals), while the Ψu1Ψg2 first excited configuration yields an excited state of 4Bg symmetry. Two energetically proximate states of the same multiplicity are subject to distortion through a vibration which transforms as the direct product of the two states. Thus, the 4Au ground state is susceptible to a distorting force, F, which results from vibronic coupling with the 4Bg excited state (Eq. 1). Eq. 1F=〈Au4|(∂V∂Q)Q*|Bg4〉 </p><p>This will warp the ground state potential energy surface and result in a double well potential with a more localized (i.e. Class II) electronic structure description. The symmetry of the vibration that leads to this localization is determined from the direct product ΓAu ⊗ ΓBg = Γ(∂V/∂Q)Q*. As shown in Figure 6, the distorting mode which couples the 4Au and 4Bg states transforms as bu.</p><p>The 4Bg excited state is expected to dominate vibronic coupling contributions to the 4Au ground state due to the very small energy gap (~0.5 eV) between these two states, and this results in an in-plane distorting vibration of bu symmetry that can lead to electronic localization within the Co(diox)2 cores of 1 and 2 (Figure 6). Thus, the stabilization leading to electron delocalization is in direct opposition to the stabilization gained by vibronic coupling leading to localization. Although the calculated spin density distributions and nature of the low energy electronic transitions strongly suggest an electronic structure description of 1 and 2 that approaches the Class III limit, the apparent disorder in the crystal structures of 1 and 2 precludes us from determining whether a static bu distortion is responsible for Class II behavior, or whether these molecules possess true inversion symmetry and are significantly more delocalized, at the Class II/III or Class III limits. However, the quality of the structures is sufficient to determine that the Co(diox)2 cores of 1 and 2 are indeed planar. The observed planarity in 1 and 2 is important, since higher energy MLCT and LMCT states can also vibronically couple with the 4Au ground state and lead to non-planar distortions of the Co(diox)2 cores.</p><!><p>In order to further explore the degree of electron delocalization in 1 and 2, and to understand the electronic origin of their high-spin (ST = 3/2) ground states, we have probed the nature of the ground state electronic structure via variable temperature magnetic susceptibility studies. The χpara•T product for 1 is temperature independent above 50 K, and the data have been fit to a Curie-Weiss model (Eq. 2) consistent with an S = 3/2 paramagnet (Figure 7). Here T is the temperature, S is the total spin value (ST = S = 3/2), N is Avogadro's number, g is the isotropic g-value, β is the Bohr magneton, and kB is Boltzmann's constant. Eq. 2χpara•T=Ng2β2T3kB(T−θ)S(S+1);S=32 </p><p>The best fit of Eq. 2 to the data yields a negative Weiss parameter (θ = −2K), which indicates that weak intermolecular antiferromagnetic interactions are present and responsible for the downward turn in the data below ca. 50 K. In marked contrast to the S = 3/2 paramagnetic nature of 1, the χpara•T data for 2 decreases at temperatures above ~50 K, indicating the population of lower multiplicity spin states. We begin the analysis of 2 by treating the χpara•T data in the localized Class I/II limit, which derives from vibronic trapping due to a localizing bu distorting mode or a lack of resonance delocalization. The appropriate Heisenberg spin Hamiltonian for this situation is given in Eq. 3 for an unsymmetrically-coupled system, which explicitly treats the NN(A)-SQ(A) exchange interaction (JNN(A)-SQ(A)) on the left hand side of 2 and the NN(B)-SQ(A) exchange interaction (JNN(B)-SQ(A)) on the right hand side of 2. Eq. 3ĤHDνV=−2JNN(A)−SQ(A)(ŜNN(A)•ŜSQ(A))−2JNN(B)−SQ(A)(ŜNN(B)•ŜSQ(A)) </p><p>Equation 3 describes the two magnetic exchange pathways present in 1 and 2 in the Class I/II limit and these pathways are depicted in Figure 8. Operating with this Hamiltonian on the |SNN(A), SSQ(A), SNN(B)> spin bases yields a spin quartet and two spin doublet states with the following energies, where JA = JNN(A)-SQ(A) and JB = JNN(B)-SQ(A). Eq. 4EQ=−JA+JB2EDA,DB=JA+JB2±JA2+JB2−JAJB </p><p>Inspection of the structure for 2 indicates that the shortest through-bond exchange coupling pathway between SQ(A) and NN(B) is over 11 bonds, and should be appreciably weaker than the NN(A)-SQ(A) exchange interaction, which is only over 5 bonds. From the distance dependence of J (Eq. 5),24,25 the anticipated ratio of 11-bond- to 5-bond J-values can be calculated using distances from the X-ray crystal structures. A β-value of 0.5 is typical for a 1,4-phenylene bridge,26 where the distances between NN radicals obtained from the X-ray structures of 2 and TpCumMeZn(SQ-Ph-NN) yield r ≈ 16 Å and ro ≈ 7 Å, respectively. This leads to a ratio of J11-bond/J5-bonds = 10−2. Eq. 5J=Joe−β(r−ro) </p><p>Thus, the exchange parameter over 11 bonds is expected to be 100-times weaker than the corresponding exchange over 5 bonds. To obtain experimental values of JNN(A)-SQ(A) (5-bonds) and JNN(B)-SQ(A) (11 bonds) the exchange energies that derive from Eq. 4 can be inserted into the field-independent van Vleck expression (Eq. 6), where the best fit of the equation to the χpara•T data for 2 is shown in Figure 9, with values summarized in Table 3. Eq. 6χpara•T=Ng2β2T3kB(T−θ)[∑SS(S+1)(2S+1)e−EskBT∑SS(S+1)e−EskBT] </p><p>Analysis of the magnetic data allows for the degree of spin-dependent delocalization in 1 and 2 to be probed through the ratio JNN(B)-SQ(A)/JNN(A)-SQ(A). In the limit of complete SQ/Cat localization (i.e., Class I behavior in the Robin and Day scheme15), the localized SQ(A) spin would preclude any measurable exchange interaction between SQ(A) and NN(B) due to the extremely long 11 bond pathway, and the JNN(B)-SQ(A)/JNN(A)-SQ(A) ratio would approach 10−2 (vide supra). However, full Class III behavior would yield an energetically stabilized S = 3/2 ground state due to spin-dependent delocalization and a JNN(B)-SQ(A)/JNN(A)-SQ(A) ratio of unity. Values of JNN(B)-SQ(A)/JNN(A)-SQ(A) between 0 and 1 would define the Class II regime. We recall that the ferromagnetic exchange coupling in 1 is so strong that only the ST = 3/2 ground state is thermally-populated at all temperatures measured (Figure 3a). Thus the JNN(B)-SQ(A)/JNN(A)-SQ(A) ratio cannot be determined for compound 1. However, for compound 2 the best fit of the susceptibility expression derived through the application of Eqs. 4 and 6 yields JNN(B)-SQ(A) = JNN(A)-SQ(A) = +234 cm−1. The energy level spectrum for 2 is depicted in Figure 10. The fact that JNN(B)-SQ(A) = JNN(A)-SQ(A) provides strong evidence for delocalization and an apparent Class II/III to Class III description for the mixed-valent Cat/SQ ligands in 2 and, by inference, compound 1 as well.</p><!><p>With the magnetic data supporting strong delocalization in this system, we turn to a more classical approach to establishing mixed-valency and delocalization in the spectroscopic analysis of the NIR band to support this analysis. It is well established that systems in the Class II mixed-valent regime are very sensitive to solvent, and changing the dielectric constant of the solvent will shift the band maximum, often in excess of 1000 cm−1.27–29 There are also systems that are presented as Class III delocalized yet still demonstrate solvent dependence (ca. 100 – 400 cm−1).21 For those systems described as borderline Class II/III IVCT band shifts of 1000 – 2500 cm−1 have been reported.28,29 Electronic absorption spectra for 2 were collected in several different solvents (see also Supporting Information), to assess solvent dependence on band maxima and band shapes. The results are presented in Table 2, where the band maximum and band width at half-height are taken as the maximum possible band width for the primary asymmetric feature ≈3500 cm−1.</p><p>The NIR band near 3500 cm−1 is clearly solvent dependent, but the shift is only ca. 210 cm−1, far less than that expected for a Class II mixed-valent system, and contrary to the behavior predicted by Marcus theory, where low dielectric solvents should shift the band to higher energy.21 Also, the maximum possible ratio Δν1/22νmax is ca. 25% of the Hush prediction of 2312 cm−1.21 The Hush prediction is most accurate in the Class II regime, and near the Class I/Class II border. Once delocalization approaches the Class II/III and Class III limits, the utility of the Hush relationships vanishes. Therefore, values of Δν1/22νmax less than 2312 cm−1 suggests delocalization either at the Class II/III limit, or deeper into the Class III regime. This presents a limitation of the spectroscopic analysis of delocalized and mixed-valent systems: the distinction between Class I and Class II is very clear -- evidenced by the presence or absence of an IVCT band -- whereas there is not always a clear distinction between Class II and Class III. Thus, the latter regimes exist as a continuum. Spectral features of band shape, solvent effects, and band asymmetry are presented as evidentiary for both Class II and Class III systems.21 In the present case, the use of exchange coupling of pendant, localized NN radicals to report on the level of delocalization within the mixed-valent SQ/Cat dyad is perhaps superior to band shape analyses, as the coupling between the localized NN spins on opposite ends of the molecule can only be achieved in the limit of strong delocalization within the mixed-valent SQ/Cat dyad. Spectroscopic evidence combined with the modeling of variable-temperature magnetic susceptibility provides strong support for delocalization in 2 at the Class II/III border or perhaps deeper into the Class III regime.</p><!><p>The HDvV exchange analysis indicates a high level of delocalization in 1 and 2, a result supported by the spectroscopic studies of the NIR band and computational results. In order to further elaborate on their mixed-valent nature and to correlate the ground state of 2 with the excited state electronic structure, we have applied a more suitable spin-dependent delocalization (SDD) model22,31–33 to this problem. A spin-dependent delocalization model is appropriate for analyzing the magnetic and spectroscopic properties of these systems since an itinerant electron is present in the mixed-valent dioxolene fragments of 1 and 2. The spin-dependent delocalization Hamiltonian matrices for the ST = 3/2 and 1/2 states are given in Eq. 7. This Hamiltonian yields a modified energy spectrum with twice the number of ST states generated by the HDvV Hamiltonian.22,31,32,34–36 Eq. 7ĤST=32=|ST,SA,SB〉|32,1,12〉|32,12,1〉  |32,1,12〉−JHAB  |32,12,1〉HAB−JĤST=12=|ST,SA,SB〉|12,1,12〉|12,12,1〉|12,0,12〉|12,12,0〉  |12,1,12〉−JHAB2032HAB  |12,12,1〉HAB2−J32HAB0  |12,0,12〉032HABJ−HAB2  |12,12,0〉32HAB0−HAB2Jwhere J=JSQA−NNA </p><p>In the SDD analysis of the magnetic data, HAB represents a transfer integral whose magnitude reflects the degree of electronic communication and delocalization within the SQ/Cat dyad, and the exchange parameter is JNN(A)-SQ(A), formally the same as in the HDvV analysis (Figure 8). In order to minimize the fit parameters in this SDD model, we fixed the value of HAB to the spectroscopic value of 1750 cm−1 determined from electronic absorption spectroscopy, and set JNN(B)-SQ(A) = 0 due to the long superexchange pathway between SQ(A) and NN(B).37 Thus, a high spin ground state can only result from a non-zero HAB. As a result, the only parameters used in fitting the magnetic data of 2 are JNN(A)-SQ(A), the electronic g-value, and an intermolecular interaction (θ). The best fit of the SDD model to the data is presented in Figure 11, and the best fit parameters are JNN(A)-SQ(A) = 580 cm−1, g = 2, and θ = −2.06 K (Table 3). This affords an energy gap between the ground state quartet (QA) the first doublet (DA) of 229 cm−1.</p><p>The resultant energy level spectrum (Figure 12) that derives from Eq. 7 requires some comment, as it differs substantially from that commonly encountered for transition metal systems that have been observed to display spin-dependent delocalization.32,38 This difference is due to the comparatively weak "single-site" exchange resulting from the 1,4-phenylene bridge39 and parameterized by the magnitude of JNN(A)-SQ(A). Coupled with the magnitude of HAB, this leads to a strong configurational mixing between the |ST=1/2, SA=1, SB=1/2> and |ST=1/2, SA=0, SB=1/2> states, and this wavefunction mixing is clearly evident in Figure 12. In marked contrast to transition metal systems, the "single-site" exchange term in 2 is significantly reduced by 2–3 orders of magnitude. Thus, for transition metal systems there is negligible mixing between energy levels with the same ST that derive from different electronic configurations, and are therefore completely ignored in the analysis of transition metal systems that display spin-dependent delocalization phenomena. The incorporation of the |ST=1/2, SA=0, SB=1/2> "non-Hund"32 states is extremely important in the development of a proper description of both the ground and excited state properties of these organic mixed-valence systems that possess high-spin ground states resulting from spin-dependent electron delocalization. The resulting energy levels obtained from the antisymmetric HDvV and SDD analyses are summarized in Figure 13.</p><!><p>We have attempted to assess the degree of electronic delocalization in both 1 and 2 by X-ray crystallography, optical spectroscopy, analyses of the magnetic susceptibility data, bonding calculations, and spin density distributions. The observation of (1) a low-energy absorption feature in the IR-NIR region of the spectrum demonstrating a slight solvent dependence, (2) sharp deviations in band width from the Hush prediction, (3) a JNN(B)-SQ(A)/JNN(A)-SQ(A) ratio of unity determined from the HDvV analysis, and (4) calculated ground state structures and spin density distributions, all point toward a high degree of delocalization and a Class II/III to Class III electronic description for the Cat/SQ dyads in 1 and 2.29 Provided 1 and 2 reside in the Class II/III limit, the dominant contribution to electronic localization and a barrier to thermal electron transfer most likely results from an in-plane distortion of bu symmetry due to vibronic coupling between the low-lying 4Bg excited state and the 4Au ground state. Unfortunately, the crystallographic data do not allow for a direct determination of this structural distortion due to the presence of a crystallographically imposed inversion center at the Co ion. In light of the planar Co(diox)2 cores, it is of interest to note that mixed-valent ortho-dimethyltetrathiafulvalene cation radical dimers that possess either dimethylsilicon or dimethylgermanium bridges display planar geometries and isotropic proton hyperfine coupling constants consistent with Class III behavior and complete delocalization of the radical electron over the entire molecule.40 The calculated HOMO and SOMO for the ortho-dimethyltetrathiafulvalene cation radical dimers are remarkably similar to 1 and 2 and possess au and bg symmetry, respectively.40 Despite a very low HOMO-LUMO gap (~500 cm−1.) calculated for the neutral species, the data do not provide any evidence for localization due to the presence of an in-plane bu distorting vibration resulting from vibronic coupling.40</p><p>It is of interest to understand how electron delocalization could be mediated by a low-spin diamagnetic Co(III) center that bridges the two dioxolene units in 1 and 2. In the high delocalization limit (i.e. a low thermal barrier for electron transfer) effective C2h symmetry is appropriate in the analysis of the data. Here, the local symmetry of the doubly-occupied Ψu orbital lacks the appropriate symmetry to mix with any of the Co d-orbitals, as the latter possess gerade symmetry. However, the Ψg dioxolene orbital does possess the correct symmetry to mix with the Co dxz orbital to provide a metal-assisted pathway promoting electron delocalization of the Ψg electron. Thus, an effective electron delocalization is primarily facilitated by p-d-p π-overlap (HAB ~ 1750 cm−1) between p-orbitals on different dioxolene ligands bridged by the Co dxz orbital.</p><p>We can now describe how localized NN spins separated by 22 Å can be coupled via valence delocalization within the SQ/Cat dyad, and relate this mechanism to that proposed for high Tc ferromagnetism in DMSs. Schematic energy level diagrams that describe this mechanism within a valence-bond configuration interaction framework are presented in Figure 14. It is important to note that the JAB and JBC exchange interactions which couple NN(A) and NN(B) via the delocalized itinerant electron do not derive from direct NN-(SQ/Cat) wavefunction overlap. This is because the localized NN SOMOs are orthogonal to singly-occupied Ψg. Thus, the coupling of these localized NN moments occurs via two excited-state charge transfer configurations (A and C, Figure 14). Excited configurations A and C result from (SQ/Cat)→NN* donor-acceptor charge transfer and contribute directly to the stabilization of the ST = 3/2 ground state via configurational mixing. Within this VBCI model, excited configuration B represents an excitation within the SQ/Cat dyad and does not couple the localized NN moments. The contributions of excited state configurations A and C to the ST = 3/2 ground state follow the same paradigm we previously used to explain the electronic origin of strong ferromagnetic exchange within individual SQ-NN donor-acceptor biradical ligands.41,42 Mixing of these excited configurations into the ground state wavefunction is the mechanism by which covalency is introduced into the NN*-(SQ/Cat) bonding scheme to form delocalized Ψu and Ψg wavefunctions. This admixture of NN* character into the delocalized (SQ/Cat) wavefunction is clearly evident upon inspection of the delocalized Ψu and Ψg wavefunctions in Figure 5. When coupled to a strong ferromagnetic single site NN*-NN exchange (2K ≈ 4000 cm−1),42 this results in ferromagnetic coupling of localized moments over nanoscale distances. Thus, an itinerant electron can couple to a donor-acceptor type charge transfer state to yield very long coherence lengths for the itinerant electron, and this itinerant electron can interact via a strong single site exchange integral (K) to promote long-range ferromagnetic exchange coupling in molecular systems, and ferromagnetism in materials of higher dimensionality including DMSs.</p><!><p>Compounds 1 and 2 are prototypes of strong spin-dependent delocalization and ferromagnetic exchange using purely organic spin-bearing units. It is of fundamental importance that the high-spin nature of 1 and 2 are only made possible via SQ/Cat valence delocalization in the Class II/III to Class III limit with the excess electron being coupled to excited state donor-acceptor charge transfer configurations that involve virtual orbitals on the NN fragments. In this respect, the strong interaction between a delocalized electron and the "localized" NN moments resembles the exchange interaction between delocalized electrons and pinned magnetic impurities in magnetically-doped semiconductors leading to the observed ferromagnetism in these materials. We also suggest that modeling magnetic susceptibility of mixed-valent systems having localized radicals covalently attached to each partner of a mixed-valent dyad may provide an excellent handle for evaluating delocalization. Furthermore, we have shown that valence delocalization mediated by the organic π system can lead to very long-range electron correlation (> 22 Å) that is difficult to obtain in classic molecular mixed-valence transition metal systems.</p><!><p>Supporting Information Available: Crystallographic details, CIF files, saturation magnetization plots, electronic absorption spectra and supporting magnetic susceptibility plots. This material is available free of charge via the Internet at http://pubs.acs.org</p>

PubMed Author Manuscript

Antibiotic Resistance in Burkholderia Species

The genus Burkholderia comprises metabolically diverse and adaptable Gram-negative bacteria, which thrive in often adversarial environments. A few members of the genus are prominent opportunistic pathogens. These include B. mallei and B. pseudomallei of the B. pseudomallei complex, which cause glanders and melioidosis, respectively. B. cenocepacia, B. multivorans, and B. vietnamiensis belong to the B. cepacia complex and affect mostly cystic fibrosis patients. Infections caused by these bacteria are difficult to treat because of significant antibiotic resistance. The first line of defense against antimicrobials in Burkholderia species is the outer membrane penetration barrier. Most Burkholderia contain a modified lipopolysaccharide that causes intrinsic polymyxin resistance. Contributing to reduced drug penetration are restrictive porin proteins. Efflux pumps of the resistance nodulation cell division family are major players in Burkholderia multidrug resistance. Third and fourth generation \xce\xb2-lactam antibiotics are seminal for treatment of Burkholderia infections, but therapeutic efficacy is compromised by expression of several \xce\xb2-lactamases and ceftazidime target mutations. Altered DNA gyrase and dihydrofolate reductase targets cause fluoroquinolone and trimethoprim resistance, respectively. Although antibiotic resistance hampers therapy of Burkholderia infections, the characterization of resistance mechanisms lags behind other non-enteric Gram-negative pathogens, especially ESKAPE bacteria such as Acinetobacter baumannii, Klebsiella pneumoniae and Pseudomonas aeruginosa.

antibiotic_resistance_in_burkholderia_species

1. Introduction<!>2. Bacterial antibiotic resistance factors<!>3. Burkholderia pseudomallei complex organisms<!>3.1 \xce\xb2-lactam resistance in Bpc bacteria<!>3.2. Efflux pump mediated multidrug resistance in Bpc members<!>3.3. Bpc species outer membrane permeability barrier<!>3.4. Alterations in drug targets<!>4. Burkholderia cepacia complex organisms<!>4.1 \xce\xb2-lactam resistance in Bcc bacteria<!>4.2. Efflux pump mediated multidrug resistance in Bcc members<!>4.3. Bcc species outer membrane permeability barrier<!>4.4. Alterations in Bcc bacteria drug targets<!>5. Conclusions

<p>Antimicrobial resistance is rapidly becoming an unavoidable public health crisis with the potential to radically alter the standard of medical care across the globe (Brown and Wright, 2016; Lushniak, 2014; Michael et al., 2014). Without an increased understanding of the mechanisms that drive resistance to therapy for bacterial infection, especially those considered drugs of last resort, the treatment obstacles posed by these factors could ultimately prove insurmountable. If current resistance trends continue and the pipeline for new drugs with activity against Gram-negative remains stagnant, infections caused by these organisms possess the most potential for complete emergence into a post-antibiotic state. Species within this group have been established as organisms of concern in both environmental and nosocomial settings, with members of the Pseudomonas, Acinetobacter, Klebsiella and Burkholderia genera already known to be intransigent to standard first-line therapy as a result of both acquired and intrinsic resistance factors (Boucher et al., 2009; Michael et al., 2014). Many of these factors are found in all four clades and some members, including P. aeruginosa, A. baumannii and K. pneumonia, are ESKAPE bacteria (Boucher et al., 2009).</p><p>The Burkholderia genus is large clade within the β-proteobacteriaceae class, containing over 70 species (Sawana et al., 2014). Of these, only B. mallei is considered an obligate parasite of eukaryotic hosts, while the rest are found as environmental saprophytes (Galyov et al., 2010). Despite this, many species other than B. mallei are opportunistic pathogens and capable of causing disease. B. pseudomallei is the etiologic agent of a serious and often fatal syndrome known as melioidosis (Cheng and Currie, 2005; Limmathurotsakul and Peacock, 2011; Peacock, 2006; Wiersinga et al., 2012; Wiersinga et al., 2006). The B. cepacia complex (BCC) consists of a group of at least 17 species (Mahenthiralingam et al., 2005; Mahenthiralingam and Vandamme, 2005; Vandamme and Dawyndt, 2011; Vanlaere et al., 2009). Several members of this complex including B. cenocepacia, B. multivorans, and B. vietnamiensis, are opportunistic pathogens. These organisms are particularly problematic in patients suffering from cystic fibrosis, and are responsible for high mortality rates within this patient cohort (Drevinek and Mahenthiralingam, 2010; Jassem et al., 2011). An underlying theme within the genus is the ability to evade the action of multiple classes of antimicrobials, a capacity that is partially responsible for the seriousness of infections caused by its member species (Dance, 2014; Golini et al., 2004; Jassem et al., 2011; Peeters et al., 2009; Rajendran et al., 2010; Schweizer, 2012a; Wuthiekanun and Peacock, 2006).</p><!><p>Before we review antibiotic resistance in Burkholderia species, it is prudent to present a brief overview of the factors that govern bacterial resistance. This will provide a perspective of the commonalities and differences of the types of resistance determinants that Burkholderia species employ when compared to other bacteria, especially other Gram-negatives.</p><p>Three types of mechanisms contribute to antimicrobial resistance in bacteria. Intrinsic antibiotic resistance is caused by physicochemical properties of a bacterium that are not subject to genetic change in response to antibiotic exposure. Possibly the best recognized example of intrinsic resistance is exclusion of drug molecules from Gram-negative bacteria by constraints of the cell envelope, mostly the outer membrane and its lipopolysaccharide and porin constituents. Acquired resistance is caused by acquisition of a previously absent resistance trait, for instance mutation of a chromosomally encoded target, transfer of foreign resistance genes via mobile plasmids, phage-mediated transduction, and transformation using free DNA (Fig. 1). Mutations of a regulatory element for a normally not expressed resistance trait such as an enzyme, an efflux pump, etc., also contribute to acquired resistance.</p><p>Bacteria can exhibit tolerance to antimicrobials that is either dependent or independent of genetic change. This includes bacterial lifestyle adaptations, for instance planktonic versus biofilm growth, intracellular anaerobiosis or persister cell formation (Balaban et al., 2013; Mah, 2012; Monroe, 2007).</p><p>An overview of mechanisms that bacteria use to resist antibiotics is presented in Fig. 2. (Blair et al., 2015; Walsh and Wencewicsz, 2016). They include the following: 1) Exclusion from or reduced penetration into the cell by constraints mediated by the cell envelope; 2) Active efflux from the cell; 3) Enzymatic inactivation, either by substrate cleavage or chemical modification, for instance acetylation, adenylation, glucosylation and phosphorylation; 4) Target alteration by point mutations or, rarely, deletion. Targets can also be enzymatically modified, for instance by methylation of ribosomal RNA; 5) Metabolic bypass by substitution of a susceptible target with a resistant target; 6) Target overproduction by either increased transcription or gene multiplication; and 7) Drug sequestration by specific binding proteins (Schweizer, 2012a; Walsh and Wencewicsz, 2016).</p><p>Bacteria frequently employ disparate mechanisms that act synergistically to achieve elevated resistance. The often high-level acquired or intrinsic resistance of non-enteric bacteria such as P. aeruginosa and Burkholderia species is in no small part attributable to synergy between reduced penetration into and efflux from the cell (Schweizer, 2012b).</p><p>In this review we will provide an overview of major resistance mechanisms described in select pathogenic members of the genera, specifically the Burkholderia pseudomallei complex (Bpc) and the Burkholderia cepacia complex (Bcc).</p><!><p>Members of the Bpc consist of B. pseudomallei, B. mallei, B. humptydoensis and B. thailandensis and are characterized by their similarities to B. pseudomallei at a genetic and phenotypic level (Brett et al., 1998; Ginther et al., 2015; Holden et al., 2004; Nierman et al., 2004; Yu et al., 2006). These organisms are thought to share a common progenitor, most likely very similar to B. pseudomallei itself. B. mallei is considered a clone of B. pseudomallei having diverged from the latter in an animal host approximately 3.5 million years ago and developed into an obligate pathogen by in-host evolution (Losada et al., 2010; Song et al., 2010). Both B. pseudomallei and B. mallei are capable of causing serious disease (Galyov et al., 2010). B. thailandensis, although generally considered non-pathogenic, is known to cause sporadic human disease (Glass et al., 2006).</p><p>Mortality rates of B. pseudomallei infections still can exceed 50% without initiation of rapid and appropriate antibiotic therapy, especially in patients with underlying risk factors such as diabetes (Cheng and Currie, 2005; Limmathurotsakul and Peacock, 2011; Peacock, 2006; Wiersinga et al., 2012; Wiersinga et al., 2006). The majority of the resistance factors encoded by B. pseudomallei are also found in B. mallei, and B. thailandensis. However, B. mallei is generally more susceptible to antibiotics than B. pseudomallei, presumably because ongoing in-host evolution leads to genome reduction, including loss of antibiotic resistance determinants (Nierman et al., 2004).</p><p>Glanders, caused by B. mallei, is a rare but frequently fatal infection mostly affecting solipeds but occasionally also humans (Whitlock et al., 2007). The organism is known to have been weaponized during both modern and ancient conflicts, most likely because of its potentially devastating effect on cavalry horses and pack animals (Cheng et al., 2005; Dance, 2005; Van Zandt et al., 2013). Few modern human cases have been documented, but symptoms in both animals and humans range from suppurating abscess of the mucosa and solid organs, to sepsis, and pneumonia, depending on the route of infection (Van Zandt et al., 2013).</p><p>Melioidosis, caused by B. pseudomallei, is characterized by a variety of symptoms ranging from self-limiting abscess, to sepsis, necrotizing pneumonia, osteomyelitis, and dissemination to the solid organs and brain (Bartley et al., 1999; Caldera et al., 2013; Currie et al., 2010; Jane et al., 2012; Maguire et al., 1998; McLeod et al., 2015; Morse et al., 2013; St John et al., 2014; Wiersinga et al., 2012). Current recommended therapy for melioidosis consists of two weeks of intravenous ceftazidime or a carbapenem, followed by up to six months of oral trimethoprim+sulfamethoxazole (co-trimoxazole)(Chetchotisakd et al., 2014; Dance, 2014; Lipsitz et al., 2012; Pitman et al., 2015). In either phase of therapy, the administration of amoxicillin+clavulanate (co-amoxiclav) is used in instances where other drugs are contraindicated (Dance, 2014; Lipsitz et al., 2012; McLeod et al., 2015). Based largely on the similarity of the in vitro antibiotic susceptibility patterns of B. mallei and B. pseudomallei treatment of glanders in humans follows a similar scheme (Kenny et al., 1999; Lipsitz et al., 2012; Peacock et al., 2008).</p><!><p>Treatment failure during the administration of ceftazidime occurs in approximately 11–17 percent of clinical cases, although it was found that only a small minority of these cases were due to primary resistance when isolates are assessed using in vitro methods (Chierakul et al., 2005; Wuthiekanun and Peacock, 2006). A survey of over 4,000 B. pseudomallei patient isolates obtained over a period spanning 20 years showed that primary resistance to β-lactam antibiotics is rare, but does exist. For instance, 24 of 4,021 (or 0.6%) of isolates were resistant to ceftazidime (n=8), amoxicillin+clavulanic acid (n=8) or both drugs (n=13). None of the isolates was resistant to carbapenems.</p><p>A Class A β-lactamase encoded by penA located on chromosome 2 is responsible for primary resistance to β-lactam antibiotics in the majority of clinical B. pseudomallei isolates (Godfrey et al., 1991)(Fig. 3). Two types of mutations have been implicated in PenA-mediated ceftazidime resistance in clinical B. pseudomallei isolates. The majority of these cause changes to or near conserved β-lactamase domains, the so-called Ambler motifs (Ambler et al., 1991). Notable amino acid substitutions implicated in β-lactam resistance in clinical isolates include Cys69Tyr, Ser72Phe and Pro167Ser, which cause ceftazidime (Cys69Tyr and Pro167Ser) and clavulanic acid (Ser72Phe) resistance, respectively (Rholl et al., 2011; Sam et al., 2009; Sarovich et al., 2012a; Sarovich et al., 2012b; Tribuddharat et al., 2003). Several strains have been identified that contain amino acid substitutions leading to simultaneous ceftazidime and clavulanate resistance, thereby rendering β-lactamase inhibitors of the clavulanate family ineffective in their presence. Some ceftazidime resistant clinical B. pseudomallei isolates also contain a point mutation in the penA upstream region, which probably causes an increased expression of PenA via increased penA transcription (Sarovich et al., 2012b). This notion is supported by results of a recent study with in vitro selected ceftazidime resistant B. thailandensis mutants (Yi et al., 2012a). Treatment of ceftazidime resistant B. pseudomallei is still possible as these isolates remain susceptible to carbapenems. A recent study characterizing B. pseudomallei PenA found that it is a membrane bound lipoprotein, secreted by the twin-arginine transport (TAT) system (Randall et al., 2015; Rholl et al., 2011). The unique membrane association of the enzyme could be a productive route of inquiry as investigative drugs capable of targeting lipoprotein synthesis, as well as TAT-mediated protein export may exhibit efficacy against ceftazidime resistant penA strains. Indeed, tat mutants are susceptible to β-lactam antibiotics and TAT inhibitors could potentially be used to sensitize B. thailandensis to β-lactams (Rholl et al., 2011; Vasil et al., 2012).</p><p>Both B. mallei and B. thailandensis express a conserved PenA Class A β-lactamase, and therefore can be anticipated to exhibit similar molecular properties and phenotypes (Winsor et al., 2008). Most observations made with B. pseudomallei β-lactamase have been corroborated by in vitro studies with B. thailandensis (Yi et al., 2012a; Yi et al., 2012b). Studies on PenA β-lactamase expression and its contribution to β-lactam resistance are hampered by the diversity of potential functionally important PenA structural features and factors governing penA transcription in sequenced B. pseudomallei strains. While we are beginning to understand the roles that structural features of PenA play in its function and substrate profile, virtually nothing is known about factors governing the enzyme's expression. Reliable information about the role of PenA in clinically significant β-lactam resistance continues to be obtained by analysis of isogenetic pre- and post-ceftazidime therapy B. pseudomallei patient isolates (Tribuddharat et al., 2003; Sam et al., 2009; Sarovich et al., 2012a; Sarovich et al., 2012b).</p><p>β-lactam resistance in this organism can also occur as the result of large scale rearrangements at the chromosomal level. In a 2011 study examining ceftazidime treatment failures in a Thai hospital, several B. pseudomallei isolates were found to have deleted large segments of chromosome 2. All isolates encompassed deletion of a common 71 kb segment (Chantratita et al., 2011). This segment contained three genes encoding putative penicillin-binding proteins (PBPs), which are known targets for β-lactam antibiotics. Of the two genes encoding PBP3s and one gene encoding a putative PBP6, one PBP3 homolog was shown to be responsible for the severe growth defect manifested as a filamentous appearance by microscopy, as well as high-level ceftazidime resistance (Chantratita et al., 2011). Because of the propagation of distinct populations arising in vivo, the highly-resistant subtype had initially been overlooked during diagnostic testing and was only discernable on specialized Ashdown's selective medium, which contained glycerol that supported growth of the otherwise unstable mutants (Ashdown, 1979). This may represent a common persistence mechanism in cases where clinical ceftazidime resistance occurs, but cannot be corroborated by standard susceptibility analysis ex vivo.</p><!><p>Efflux is a major resistance mechanism in Bpc organisms, more so in B. pseudomallei and B. thailandensis than in B. mallei. A recent review focused on efflux-mediated drug resistance in Burkholderia, which will be covered herein in an abbreviated manner (Podnecky et al., 2015).</p><p>Bacterial genomes encode at least six efflux pump families (Fernandez and Hancock, 2012ß; Hassan et al., 2015; Nikaido and Pages, 2012; Piddock, 2006). These include: 1) The major facilitator (MFS) superfamily; 2) The resistance nodulation cell division (RND) family; 3) The small multidrug resistance (SMR) family; 4) The multi-drug and toxic compound extrusion (MATE) family; 5) The ATP-binding cassette (ABC) family; and 6) The proteobacterial antimicrobial compound efflux (PACE) family. Most bacteria encode several members of each of these efflux pump families, and Burkholderia species are no exception. In Gram-negative bacteria, efflux pumps of the RND family are of major significance because of their unique ability to span the entire cell envelope, which can lead to high-level resistance by synergy between the outer membrane permeability barrier and efflux into the extracellular milieu (Fernandez and Hancock, 2012; Schweizer, 2012b).</p><p>All B. pseudomallei strains encode at least 10 RND systems, seven of which are encoded by chromosome 1 and three by chromosome 2 (Holden et al., 2004; Kumar et al., 2008; Podnecky et al., 2015). Three of these RND pumps have been characterized in B. pseudomallei. AmrAB-OprA is expressed in most B. pseudomallei strains and responsible for intrinsic resistance to aminoglycosides and macrolides, and also confers some resistance to tetracyclines (Moore et al., 1999; Trunck et al., 2009). AmrAB-OprA overexpression confers high-level resistance to cethromycin (Mima et al., 2011). Rare aminoglycoside-susceptible B. pseudomallei environmental or clinical isolates either do not express AmrAB-OprB due to regulatory mutations, or lack the amrAB-oprA operon as a result of a chromosomal deletion, or express a non-functional efflux pump due to mutations that affect the AmrB RND transporter (Podin et al., 2014; Trunck et al., 2009). Several strains of B. mallei lack AmrAB-OprA and are therefore aminoglycoside and macrolide susceptible (Nierman et al., 2004).</p><p>In regulatory mutants, BpeAB-OprB confers low-level resistance to chloramphenicol, fluoroquinolones, macrolides, and tetracyclines (Chan et al., 2004; Mima and Schweizer, 2010). In some isolates this pump has been implicated in additional functions in host adaptation and quorum sensing (Chan and Chua, 2005). The clinical significance of BpeAB-OprB remains unclear. BpeEF-OprC is expressed in regulatory mutants, and extrudes chloramphenicol, fluoroquinolones, sulfamethoxazole, tetracyclines, and trimethoprim (Hayden et al., 2012; Kumar et al., 2006; Podnecky et al., 2015; Podnecky et al., 2013). It was shown that the multidrug resistant phenotype in a clinical relapse isolate is likely due to constitutive BpeEF-OprC expression as a result of a regulatory mutation caused by a 800 kb chromosomal inversion (Hayden et al., 2012; Podnecky et al., 2015).</p><p>The pump repertoire, expression pattern and ensuing drug resistance profile of B. thailandensis parallels that of B. pseudomallei: AmrAB-OprA confers resistance to aminoglycosides, macrolides, and tetracyclines; BpeAB-OprB extrudes tetracyclines; and and BpeEF-OprC effluxes chloramphenicol, fluoroquinolones, sulfamethoxazole, tetracyclines, and trimethoprim (Biot et al., 2013; Biot et al., 2011).</p><p>Together, AmrAB-OprA, BperEF-OprB, BpeEF-OprC effectively render six entire classes of compounds at least partially inactive depending on expression level, and evidence obtained mainly with B. thailandensis suggests that they may act in concert to tightly control the intrusion of these compounds into the bacterial cell (Biot et al., 2013).</p><!><p>Although it has been known for some time that the relative outer membrane permeation of non-enteric bacteria like Burkholderia is significantly lower that that of E. coli, the underlying mechanisms remain poorly understood (Hancock, 1998). However, alterations to membrane permeability and structure have been implicated as resistance determinants in these species. For instance, atypical lipopolysaccharide (LPS) structure plays a crucial role in intrinsic resistance of Burkholderia species to cationic peptides, notably polymyxin B (Loutet and Valvano, 2011). A common LPS modification leading to resistance to polymyxin B in Gram-negative bacteria is the modification of lipid A with a positively charged 4-amino-4-deoxy-arabinose (Ara4N) moiety that masks the negative charges of the two phosphate moieties attached to the lipid A (Loutet and Valvano, 2011; Olaitan et al., 2014). The major lipid A species of B. pseudomallei and B. thailandensis was reported to contain a biphosphorylated disaccharide backbone modified with Ara4N at both phosphate groups (Novem et al., 2009). The presence of Ara4N effectively reduces the net negative charge of the cell envelope, reducing the permeation of cationic antimicrobials like polymyxins. The LPS core structure also plays a role in B. pseudomallei polymyxin B resistance. One gene (waaF) encodes a protein required for LPS core oligosaccharide biosynthesis that when mutated increases polymyxin B susceptibility. However, resistance to polymyxins is not restricted to a single pathway but is highly complex. Other mutants exhibiting a polymyxin susceptible phenotype had mutations in a predicted UDP-glucose dehydrogenase and an enzyme called IspH (formerly LytB) necessary for the synthesis of isoprenoids (Rohdich et al., 2002).</p><p>Outer membrane porin proteins also play a role in antimicrobial resistance, especially in concert with resistance enzymes or efflux pump expression (Fernandez and Hancock, 2012; Pages et al., 2008). Little is known about porins and their possible roles in B. pseudomallei drug resistance. In vitro studies with B. pseudomallei Omp38 in reconstituted liposomes indicated that this protein functions as a diffusion porin for neutral sugars and charged antibiotics (Siritapetawee et al., 2004). Subsequent black lipid membrane reconstitution studies, liposome swelling assays and expression in a porin-deficient E. coli strain confirmed translocation of antibiotics through Omp38 and suggested a possible contribution of this porin in B. pseudomallei ceftazidime and carbapenem resistance (Aunkham et al., 2014; Suginta et al., 2011). However, assessment of the true contribution of this porin to B. pseudomallei antibiotic resistance awaits genetic and functional analyses in the source organism.</p><!><p>Point mutations that alter drug targets are one of the most common means that bacteria utilize to resist antibiotics (Blair et al., 2015; Walsh and Wencewicsz, 2016). Aside from the aforementioned PenA mutations that increase affinity to β-lactams that were previously poor substrates and the PBP3 deletion causing ceftazidime resistance in B. pseudomallei, reports on target mutations in B. pseudomallei leading to resistance to other antibiotics are scarce. To date there is one report on target mutation-based fluoroquinolone resistance in B. pseudomallei. Fluoroquinolones selectively target bacterial topoisomerase type II enzymes, DNA gyrase and topoisomerase IV (Walsh and Wencewicsz, 2016). Each enzyme is an A2B2-type heterotetramer, consisting of two GyrA and GyrB monomers for DNA gyrase and two ParC and ParD monomers for topoisomerase IV. Fluoroquinolones inhibit the respective "A" subunit of the topoisomerase heterotetramer, e.g. GyrA and ParC. Resistance to fluoroquinolones is frequently due to mutations affecting the quinolone resistance-determining regions (QRDR) of GyrA or ParC. In E. coli the GyrA QRDR encompasses a relatively small region (amino acids 67 to 106) of the protein (Yoshida et al., 1990). Quinolone resistance in E. coli and other bacteria is caused most often by mutations affecting amino acids 81 through 87. In E. coli roughly 50% of mutations affect Ser83 (Yoshida et al., 1990). In B. pseudomallei, all in vitro selected fluoroquinolone resistant mutants contained a Thr83Ile mutation (Viktorov et al., 2008). This is consistent with the finding that most Burkholderia species contain a GyrA Thr83 instead of the Ser83 found in other bacteria (Winsor et al., 2008). Notably, P. aeruginosa also contains Thr83 in GyrA and a frequent amino acid change seen in fluoroquinolone resistant clinical isolates is a Thr83Ile mutation (Kureishi et al., 1994).</p><!><p>The Burkholderia cepacia complex (Bcc) organisms are opportunistic nosocomial pathogens capable of causing severe disease in immunocompromised individuals, especially those with cystic fibrosis (CF) (Mahenthiralingam et al., 2005). In these patients, infection with Bcc organisms causes "cepacia syndrome", typically characterized by necrotizing pneumonia, sepsis and an overall negative prognosis (Mahenthiralingam and Vandamme, 2005). Although several species from this group have been isolated from the lungs of CF patients, B. cenocepacia and B. multivorans appear to cause the most serious forms of disease in these individuals and account for 85% of all Bcc infections (Drevinek and Mahenthiralingam, 2010; Mahenthiralingam and Vandamme, 2005). Treatment of Bcc infections relies on ceftazidime and other extended-spectrum cephalosporins, as intrinsic resistance prevents the action of many other classes of antimicrobials. One study of several thousand clinical isolates isolated from CF patients found that resistance to many standard therapies was overwhelming; greater than 50 percent of isolates were resistant to chloramphenicol, co-trimoxazole, ciprofloxacin, tetracycline, rifampin, avibactam, and co-amoxiclav (Zhou et al., 2007). However, these data may overestimate the occurrence of resistance in Bcc organisms as the study was carried out on patient isolates solicited because they were in fact multidrug resistant. Despite this caveat, resistance patterns, both intrinsic and acquired, must not be discounted in these organisms.</p><!><p>Resistance to β-lactam antibiotics such as ceftazidime is caused by class A β-lactamases encoded by Bcc organisms, first described in the PenA-PenR system of B. cepacia 249 (Trepanier et al., 1997), which are now named PenB and PenR (AmpR) (Hwang and Kim, 2015). A recent study in B. cenocepacia identified ceftazidime-driven mutations to the peptidoglycan recycling enzyme AmpD as a putative cause for up-regulation of the PenB and AmpC β-lactamases (Hwang and Kim, 2015). PenB is a Class A penicillinase with broad spectrum carbapenemase character that highly conserved across the Bcc and shares significant protein homology with PenA of Bpc organisms (Poirel et al., 2009; Hwang and Kim, 2015). Both penB and ampC promoters were associated with binding sites for LysR-type repressor PenR (AmpR). PenR is a bifunctional protein. It is a repressor when it binds the D-alanine-D-alanine pentapeptide stem terminus of the peptidoglycan precursor UDP-MurNAc-pentapeptide. This protein was previously found to bind 1,6-anhydro-MurNac peptides, which are produced in the presence of β-lactam antibiotics or after disruption of AmpD, an enzyme that normally degrades them. This binding causes PenR to become an activator and leads to transcription of its penB and ampC targets (Hwang and Kim, 2015; Vadlamani et al., 2015). As a result, susceptibilities to ceftazidime, cefotaxime, and meropenem were greatly reduced.</p><p>B. cenocepacia also contains a Class A PenA β-lactamase, which is 55% identical and 67% similar to B. pseudomallei strain 1026b PenA (Winsor et al., 2008). As in the Bpc bacteria, the penA gene is located on chromosome 2 and its genetic surroundings are very similar to that of B. pseudomallei penA (Fig. 3). However, unlike B. pseudomallei or B. thailandensis PenA (the B. thailandensis PenA is now also referred to as PenL in some publications (Hwang and Kim, 2015)) the B. cenocepacia enzyme has not yet been shown to be involved in β-lactam resistance. B. thailandensis mutants with in vitro selected ceftazidime resistance solely contain PenA (PenL) structural and/or regulatory mutations, whereas in B. cenocepacia the same selection leads to regulatory mutants that either overexpress PenB or AmpC (Hwang and Kim, 2015; Yi et al., 2012a). PenB and AmpC exhibit similar substrate spectra and both are under control of the LysR-type PenR regulatory protein (Fig. 3). PenB and AmpC are absent from Bpc bacteria. Of note is that the sequenced B. cenocepacia strain J2315 PenB β-lactamase contains a Ser72Tyr substitution, which may explain this strain's intrinsic clavulanate resistance (Hwang and Kim, 2015).</p><p>B. multivorans contains a PenA enzyme (Bmul_3689 in B. multivorans ATCC 17616), that is closely related to PenB in BCC bacteria (Poirel et al., 2009; Hwang and Kim, 2015). It is also similar to KPC-2, which is the most clinically significant serine carbapenemase (Papp-Wallace et al., 2013). B. multivorans PenA is closely related to B. pseudomallei PenA (also called PenI by Dr. Bonomo's group (Papp-Wallace et al., 2013)). However, the B. multivorans enzyme is an inhibitor-resistant carbapenemase, whereas the B. pseudomallei enzyme is an extended spectrum β-lactamase. The role of PenA in clinically significant B. multivorans β-lactam resistance compromising therapy is not well established.</p><!><p>The roles of efflux pumps in antibiotic resistance of members of the Bcc have recently been reviewed (Podnecky et al., 2015). In B. cenocepacia, at least six efflux pumps of the RND family have been implicated in drug resistance – RND-1, RND-3, RND-4, RND-8, RND-9, and RND-10 (Bazzini et al., 2011; Buroni et al., 2014; Buroni et al., 2009; Coenye et al., 2011; Nair et al., 2004). Of these RND-3, RND-4, and RND-10 (also known as CeoAB-OpcM) correspond to B. pseudomallei AmrAB-OprA, BpeAB-OprB and BpeEF-OprC, respectively, although the resistance patterns bestowed by the respective pumps are somewhat different, with the exception of RND-10 and BpeEF-OprC whose resistance profiles are very similar (Podnecky et al., 2015). An inventory of B. cenocepacia resistance mechanisms showed that efflux pump activity is prevalent in this bacterium and that mutations in the RND-3 pump regulator are the major cause of efflux pump activity and RND-3 mediated antibiotic resistance (Tseng et al., 2014). In B. vietnamiensis, aminoglycoside resistance emerges during chronic infection or after in vitro exposure to aminoglycosides and is the result of AmrAB-OprM efflux pump expression, which is most likely a homolog of B. pseudomallei and B. thailandensis AmrAB-OprA (Jassem et al., 2014; Jassem et al., 2011).</p><p>The B. vietnamiensis NorM pump, a member of the multidrug and toxic compound extrusion family of efflux systems, was shown to contribute to polymyxin B resistance, but curiously only in the presence of exogenously added tetracycline (Fehlner-Gardiner and Valvano, 2002).</p><!><p>Similar to Bpc organisms, Bcc bacteria are typically resistant to polymyxins (Loutet and Valvano, 2011; Olaitan et al., 2014). As described above, this is thought to be partially attributable to a unique LPS structure that inhibits polymyxin binding to the outer membrane (Loutet and Valvano, 2011). In B. cenocepacia, an amino arabinose biosynthesis operon is responsible for the synthesis of 4-amino-4-deoxy-L-arabinose (Ara4N) used for modifications of lipid A that alter the total charge of the LPS molecule, thereby decreasing susceptibility to cationic antimicrobial peptides and polymyxins (Isshiki et al., 1998; Loutet and Valvano, 2011; Olaitan et al., 2014; Ortega et al., 2007).</p><p>The alternative sigma factor RpoE, which controls the expression of a regulon of genes required for bacteria to respond to extracytoplasmic stress, plays a significant role in polymyxin B resistance in B. cenocepacia at 37°C but not at 30°C (Loutet and Valvano, 2011).</p><p>As in Bpc bacteria, polymyxin resistance in Bcc species is complex and the pathways found in the former are also found in Bcc bacteria, as reviewed in Loutet et al. (Loutet and Valvano, 2011).</p><p>In B. multivorans, genes needed for the synthesis of hopanoid compounds are also implicated in polymyxin resistance (Malott et al., 2012). Interestingly, the use of fosmidomycin may potentiate colistin activity against B. multivorans by interrupting this process. Through disruption of the isoprenoid synthesis pathway, fosmidomycin prevents hopanoid synthesis and alters membrane composition, ultimately resulting in a 64 fold decrease in the colistin MIC (Malott et al., 2014).</p><p>The role of porins in decreased antibiotic susceptibility of Bcc species has long been established. Bcc complex bacteria such as B. cenocepacia contain general porins that exhibit a permeability that is similar of P. aeruginosa, and approximately 10 times less than E. coli (Parr et al., 1987). β-lactam resistant CF B. cenocepacia isolates and a resistant mutant were shown to have decreased porin content (Aronoff, 1988).</p><!><p>Drug target modification in these species has been mostly associated with resistance to fluoroquinolones. Depending on the selection scheme and drug concentration used, in vitro selection of ciprofloxacin resistant B. cenocepacia mutants yielded mostly Thr83Ile or Asp87Asn mutations in the GyrA QRDR (Pope et al., 2008). These resulted in a 12–64 fold increase in the ciprofloxacin minimal inhibitory concentration (MIC). High-level ciprofloxacin resistance (MIC>256 μg/ml) required an additional Ser80Leu mutation in the ParC QRDR (Pope et al., 2008). Similar gyrA mutations were identified in a majority of levofloxacin resistant isolates studied in a survey of resistance mechanisms in Bcc (Tseng et al., 2014). The resistant isolates contained Gly81Asp, Thr83Ile, and Asp87His mutations. None of the isolates contained mutations in the parC QRDR.</p><p>Dihydrofolate reductase is the target of trimethoprim. Purification of the enzyme from trimethoprim susceptible and trimethoprim resistant B. cepacia strains indicated that the protein from the resistant strain was indeed refractory to trimethoprim inhibition (Burns et al., 1989). While this finding is consistent with target alteration, the molecular and genetic basis for this reduced inhibition was not established.</p><!><p>Although the resilience of Burkholderia species to antimicrobials has been recognized for quite some time, a literature review quickly reveals that our overall understanding of resistance in these bacteria is still rather rudimentary. Fig. 4 summarizes the current state of knowledge of the various resistance determinants that have been observed in antibiotic resistant Burkholderia species. First, the outer membrane permeability barrier is an important contributor to Burkholderia drug resistance. The two major players involved are lipopolysaccharide (LPS) and outer membrane porins. LPS typically plays a general role in drug resistance, but in most Burkholderia species it plays a major role in resistance to cationic peptides. Lipid A modification by aminoarabinose and changes to the LPS core are major determinants for the widespread intrinsic polymyxin resistance in Burkholderia. Restrictive porin proteins are contributing factors to drug resistance, especially in combination with other determinants such as efflux. RND pump-mediated efflux makes major contributions to intrinsic and acquired multidrug resistance. Depending on species, either periplasmic or membrane-bound β-lactamases play important roles in intrinsic β-lactam resistance and acquired resistance to clinically significant β-lactam antibiotics. Resistance due to antibiotic target mutations has mostly been associated with fluoroquinolones, but has also been implicated with resistance to other antibiotics, e.g. trimethoprim. Of note is that most, if not all, of the Burkholderia resistance determinants identified to date are encoded by the genome of the respective organisms. The genomes of Burkholderia species consist of at least two chromosomes, e.g. Bpc bacteria, but other species contain additional genetic elements, e.g. B. cenocepacia, whose sequenced prototype strain J2315 contains a third chromosome and a plasmid (Holden et al., 2009; Holden et al., 2004; Nierman et al., 2004). Plasmids have not yet been demonstrated in Bpc bacteria. The constellation of both intrinsic and acquired resistance mechanisms in this genus combines to create a unique and often difficult challenge for researchers and clinicians. Further study is necessary to understand the interplay of these factors and their effect on antimicrobial therapy.</p>

PubMed Author Manuscript

Siloxy Esters as Traceless Activator of Carboxylic Acids: Boron-Catalyzed Chemoselective Asymmetric Aldol Reaction

The catalytic asymmetric aldol reaction of carboxylic acids is among the most useful reactions for the synthesis of biologically active compounds and pharmaceuticals. Despite the existence of many prominent reports, no general method is available to incorporate the aldol motif into complex carboxylic acids and their derivatives at late stages. Chemoselective catalytic asymmetric aldol reaction of multifunctional carboxylic acids is difficult to achieve, due to the high basicity required for enolization and the poisonous chelation of β-hydroxy acid products to Lewis acid catalysts. Herein, we identified that preconversion of carboxylic acids to siloxy esters facilitated the boron-catalyzed direct aldol reaction, leading to the development of carboxylic acid-selective, catalytic asymmetric aldol reaction applicable to multifunctional substrates. The asymmetric boron catalyst stereodivergently controlled the products' stereochemistry depending on the catalyst's chirality, not on the stereochemical bias of substrates. Computational studies rationalized the mechanism of the catalytic cycle and the stereoselectivity, and proposed Si/B enediolates as the active species for the asymmetric aldol reaction. The silyl ester formation facilitated both enolization and catalyst turnover through acidifying the α-proton of substrates and attenuating poisonous Lewis bases to the boron catalyst.

siloxy_esters_as_traceless_activator_of_carboxylic_acids:_boron-catalyzed_chemoselective_asymmetric_

INTRODUCTION<!>RESULTS AND DISCUSSION<!>Table 2. Substrate Scope a<!>Scheme 2. Comparison between Siloxy Ester and tBu Ester<!>Acceleration of Mannich and Allylation Reactions by Siloxy Ester Formation<!>Conclusions

<p>The carboxyl group is a common functional group existing in a wide range of organic molecules, especially in biologically active natural products and pharmaceuticals: e.g. nonsteroidal anti-inflammatory drugs (NSAIDs) 1 and antibiotics. 2 Therefore, catalytic, chemoselective, and asymmetric C-C bond-formation of carboxylic acids containing multiple functional groups is useful in complex molecule synthesis and late-stage structural diversifications of drug leads for optimizing their pharmaceutical properties. 3 Due to the high acidity of carboxylic acids compared to other common functional groups, chemoselective recognition and activation of carboxylic acids is possible through reversible catalyst-substrate covalent bond-or salt-formation. Yamamoto reported a pioneering example of boronic acid-catalyzed electrophilic activation of carboxylic acids in the presence of amines for amidation (Scheme 1a). 4,5 Electron-withdrawing arylboronic acids chemoselectively formed acyloxyboron intermediates with substrate carboxylic acids to enhance their electrophilicity. Yu reported site-selective, palladium-catalyzed C-H bond functionalizations of carboxylic acids at distal positions (βor farther positions) to the carboxyl group and their applications to natural product synthesis (Scheme 1b). 6 The carboxylate group worked as a directing group of the palladium catalyst by coordination, realizing highly practical site-selective C-H functionalizations. Our group reported chiral boronate-catalyzed nucleophilic activation of carboxylic acids through enolization, enabling chemoselective and asymmetric Mannich reaction and α-allylation of carboxylic acids (Scheme 1c). 7,8 Chiral diboron enediolate 1 9 was proposed as the active species for those reactions. Due to the high oxophilicity of the boron catalyst, however, this enolization method was not effective in Scheme 1. Catalytic Chemoselective Activation of Carboxylic Acids promoting a catalytic asymmetric aldol reaction of carboxylic acids, 10 another fundamental and synthetically useful C-C bond-forming reaction. [11][12][13][14] Here we found that in-situ pre-conversion of carboxylic acids to siloxy esters dramatically facilitated the boron-catalyzed, carboxylic acid-selective asymmetric aldol reaction. The siloxy group worked as a traceless activator, allowing for the direct use of carboxyl group-containing multifunctional drugs and natural products themselves as substrates of catalytic asymmetric aldol reactions. 15 We propose Si/B enediolates 2 as the active species for this reaction. Since the boron catalyst did not promote aldol reactions of simple esters (e.g. tBu ester), the reactivity is unique to siloxy esters.</p><!><p>We began optimization of the reaction between benzaldehyde (3a) and propionic acid (4a) using a chiral boron catalyst generated from either BH3•SMe2 or (AcO)4B2O with ligand L1 7,10 (Table 1). In the absence of any additives, the yield of product 5aa was up to the loading amount of the boron catalyst (entries 1 and 2). We attributed the lack of catalyst turnover to the high stability of the catalytically inactive boron aldolate intermediate 6. To facilitate catalyst turnover, we added silylating reagents. 13a Although using BH3•SMe2 as a boron source did not produce 5aa in the presence of Me3SiCl (entry 3), (AcO)4B2O afforded the product in 42% yield (entry 4). Treatment of 4a with Me3SiCl prior to the aldol reaction further improved the yield to 48% (entry 5). NMR studies revealed that 4a was converted to its silyl ester by this treatment. 16 Screening silylating reagents led us to identify (EtO)3SiCl as the optimum additive regarding product yield, affording 5aa in 92% yield with 13/1 dr, but with only 3% ee (entry 6). The selectivity slightly improved to 15/1 dr and 6% ee in THF, despite yield lowered to 71% (entry 7). No reaction proceeded in the absence of the boron catalyst (entry 8). To improve enantioselectivity, we next studied valine-derived ligands with various N-aryl sulfonyl groups. Enantioselectivity increased according to the number of fluorine substituents on the aryl group of ligands L1-L4, 7b,10a,17 but diastereoselectivity decreased accordingly (entries 7 and 9-11). To optimize the electronic properties of the sulfonyl group, we substituted the 4-fluorine atom of L4 with an electron-donating MeO (L5: entry 12) or Me2N (L6: entry 13) group: using L6 afforded balanced dr and ee (entry 13). Finally, reducing the amount of DBU to 3.5 equiv and concentration to 0.1 M afforded 5aa in 69% yield with 16/1 dr and 86% ee (entry 14). We then examined the substrate scope under the optimized conditions (Table 2). The scope of carboxylic acids was first investigated using benzaldehyde (3a). 3-Phenylpropionic acid (4b) furnished 5ab in high yield and stereoselectivities (89%, >20/1 dr, 90% ee). The amount of boron catalyst could be reduced to 8 mol % without significant erosion in reactivity and selectivity (76%, 19/1 dr, 89% ee). The reaction was applicable on a gram scale (64%, >20/1 dr, 90% ee). Carboxylic acids containing potentially reactive functional groups such as unsaturated C-C bonds, halogens, and a hydroxy group were investigated (4c-4k). In all cases, the reaction proceeded smoothly with high diastereo-and enantioselectivity (5ac-5ak). The reaction was chemoselective at the α-position of the carboxyl group in the presence of amide, ester, ketone, and nitrile functional groups, which are intrinsically more prone to enolization than the carboxyl group (5ag-5aj).</p><p>Next, the scope of aldehydes 3 was examined using carboxylic acid 4b. A series of aromatic aldehydes bearing an electron-donating, an electron-withdrawing, and a naphthyl group were competent substrates (5bb-5fb). Introduction of a substituent at the ortho or meta position of the aromatic ring did not affect the result (5gb and 5hb vs. 5cb). Reactions with aliphatic aldehydes also proceeded in high yield and selectivity by increasing DBU to 5 equiv and the concentration to 0.3 M (5ib-5nb). Aliphatic aldehydes are especially difficult substrates for catalytic asymmetric direct aldol reactions because they can enolize easily under basic conditions. 13 Table 1. Optimization of Reaction Conditions a a Standard conditions (for entries 7-14): A mixture of 4a (2 equiv), silylating reagent (2 equiv), and DBU (4 equiv) in THF (0.5 mL) was stirred at room temperature (rt) for 30 min (solution A). A boron source (B: 20 mol %) and a ligand (20 mol %) in THF (0.5 mL) were stirred at rt for 30 min in another vessel (solution B). Solution B and 3a (1 equiv) were added successively to solution A, and the mixture was stirred at rt for 12 h. b In toluene. c The silylating reagent was added as the final component. d DBU (3.5 equiv) and 0.1 M.</p><!><p>a General reaction conditions: 4 (0.60 mmol), 3 (0.30 mmol), (EtO)3SiCl (0.60 mmol), DBU (1.05 mmol), (AcO)4B2O (0.03 mmol), L6 (0.06 mmol), THF (3.0 mL), room temperature, 12 h. Isolated yield, enantiomeric excess (ee), and diastereomeric ratio (dr) shown in the Table were determined after conversion to the corresponding methyl esters, except for 5ak. 16 b 4 mol % (AcO)4B2O and 8 mol % L6 were used. c 6 mmol scale reaction. d NMR yield is shown in parentheses. e Concentration was 0.2 M. f 4 equiv of DBU was used. g Concentraion was 0.3 M, and 5 equiv of DBU was used. h 4 equiv of (EtO)3SiCl was used. i 6 equiv of DBU was used. j ent-L6 was used. k 10 equiv of aldehyde 3 was used. l 30 mol % (AcO)4B2O and 60 mol % L6 were used.</p><p>Finally, we applied our method to more complex carboxylic acids containing multiple functional groups. By using commercially available NSAIDs, acemetacin (4l) and oxaprozin (4m), the corresponding aldol products 5al and 5am were obtained in high yield and selectivity. An immunosuppessant drug, mycophenolic acid (4n), containing a phenolic hydroxy group afforded 5an in 69% yield with 17/1 dr and 87% ee. The reaction between 3a with dehydrocholic acid (4o) bearing three keto groups proceeded chemoselectively at the α-position of the carboxyl group. The reaction was stereodivergent, producing isomeric products depending on the chirality of the boron catalyst: using ligand L6 or ent-L6, product 5ao or iso-5ao was obtained in high stereoselectivity, respectively. Cholic acid bearing three hydroxy groups afforded products 5ap and iso-5ap in excellent selectivity, which was again controlled by the catalyst. The stereodivergent reaction also proceeded from mupirocin, an antibiotic drug containing hydroxy, epoxy, α,β-unsaturated ester, and carboxyl groups, in high chemo-and stereoselectivity (5aq and iso-5aq).</p><!><p>To gain mechanistic insights, we conducted a series of experiments. First, we compared the reactivity between siloxy esters and alkyl esters. Whereas the reaction between aldehyde 3a and carboxylic acid 4a proceeded to afford product 5aa in 69% yield through the corresponding siloxy ester, the reaction between 3a and tBu ester 7 did not proceed at all, irrespective of the presence or absence of (EtO)3SiCl (Scheme 2). The acidity of carbonyl α-proton may dictate the contrasting reactivity: calculated pKa values for CH3COOSi(OEt)3 and CH3COOtBu were 20 and 26, respectively. 16,18 The significantly higher acidity of siloxy esters compared to alkyl esters might be partly due to divalent coordination of carboxylate to a hypervalent silicone atom, which was previously proposed by Yamamoto and colleagues on the basis of 29 Si NMR in mechanistic studies of the silyl ester-mediated peptide coupling reaction. 19</p><!><p>Acidification of the α-proton of carboxylic acids may not be the sole cause for rate acceleration by the silylating reagent, however. During the initial optimization, we noticed that increasing the amount of carboxylate decreased the reactivity. Inhibitory effects of an excess carboxylate were likely due to coordination to the boron catalyst, which attenuated its Lewis acidity by forming a borate. The following experiments support our hypothesis (Scheme 3); whereas 5aa was obtained in 41% yield for 30 min under the standard conditions, yield decreased to 26% in the presence of 1 equiv propionate. Therefore, silyl ester formation would decrease the detrimental coordination of free carboxylate to the boron catalyst. If the acceleration effects of silylating reagents are caused by acidification of the α-proton and/or attenuation of the carboxylate coordination to the boron catalyst through silyl ester formation, the effects would be general for other boron-catalyzed α-functionalizations of carboxylic acids. Therefore, we applied the present conditions to previously developed reactions, 7 and found that this is indeed the case (Scheme 4). Mannich reaction 7a between 9 and 4a or 4m proceeded in the presence of 2 mol % boron catalyst and (EtO)3SiCl, producing 10a or 10m in 62% or 59% yield, respectively (Scheme 4a). Allylation reaction 7b between 11 and 12 afforded 13 in 89% yield using 5 mol % boron catalyst and 5 mol % palladium catalyst in the presence of (EtO)3SiCl (Scheme 4b). The products were obtained in up to only 7% yield under the previous conditions without the silylating reagent. We propose the active nucleophile for this catalytic asymmetric aldol reaction to be chiral Si/B enediolate 2 based on the following non-linear effects experiments. 20 We first observed a linear relationship between enantiomeric excesses of the catalyst (20, 40, 60, and >99% ee) and 5ab (19, 36, 54, 71, and 90% ee, respectively) for the aldol reaction between 3a and 4b the presence of (EtO)3SiCl. Because the catalytic aldol reaction did not in the absence of (EtO)3SiCl (Table 1, entry 2), however, it was not possible to compare this result to control conditions in the absence of (EtO)3SiCl. Therefore, we investigated boron-catalyzed asymmetric Mannich reaction using BINOL-derived ligand L7, 7a an optimized chiral ligand for the Mannich reaction, which proceeded either in the absence or the presence of (EtO)3SiCl (Scheme 5). Positive non-linear effects were observed in the absence of (EtO)3SiCl (Condition A), supporting our idea that the reaction proceeds through diboron enediolate 1. In the presence of (EtO)3SiCl (Condition B), however, the relationship was linear. Thus, the reaction may not involve intermediates or active species containing more than one boron atom. The most probable active species is Si/B enediolate 2. The fact that enantio-and diastereoselectivity depended on the silyl group (Table 1, entries 5 and 6) is also consistent with the hypothesis that the silyl group is involved in the stereo-determining step. We then performed density functional theory (DFT) calculations to rationalize the overall reaction pathway. 16 Computed free energy profile is shown in Figure 1. First, we searched possible intermediates for deprotonation of triethoxysilyl ester derived from 4a (Figure S1). 16 Among them, deprotonation of I1, where the siloxy ester coordinated to the Lewis acidic chiral boron catalyst, showed the lowest free energy barrier (TS1, 4.1 kcal/mol, Figure 2A). Other intermediates furnished relatively large free energy barriers. Therefore, we chose I1 as the starting point of the reaction pathway. After deprotonation, intermediate I2, which was 8.2 kcal/mol more stable than I1, was formed. Then, reconstitution of I2 generated Si/B hetero enediolate intermediate I3 or I4 with only 3.3 or 5.9 kcal/mol higher energy than I2, respectively. Because the energy difference between I3 and I4 was only 2.6 kcal/mol, we searched transition states for the asymmetric aldol reaction starting from I3 or I4. The asymmetric aldol reaction from I3 proceeded in an innersphere fashion through a six-membered chair transition state TS2, which was only 4.4 kcal/mol higher energy than I3 (Figure 2B). We also calculated an outer-sphere transition state from I4 and aldehyde 3a coordinating to another chiral boron complex. The transition state was, however, 11.7 kcal/mol higher than TS2, indicating that the outer-sphere mechanism did not contribute to the reaction. Thus, we concluded that the asymmetric aldol reaction proceeded through I3 and TS2, leading to the (2R,3R)-product. Among calculated transition states, TS2 existed in 80% probability. 16 Other transition states affording (2S,3S)-, (2R,3S)-, and (2S,3R)-products existed in 12.5%, 7.5%, and 0% probabilities, respectively. Thus, the calculated syn/anti ratio was 93/7. This value is consistent with the experimental result of 16/1 dr for 5aa (Table 2). Furthermore, the computed ee value was 73%, which is qualitatively in good agreement with the experimental data (86% ee). After asymmetric aldol reaction, boron aldolate I5 with 29.1 kcal/mol below the entry point (I1) was generated. Due to the existence of an oxophilic silyl group in the molecule, ligand exchange proceeded between boron and silicon atoms in I5 to generate a catalytically active boron carboxylate I7, which was only 1.7 kcal/mol higher energy than I5.</p><!><p>We developed a chiral boron-catalyzed, carboxylic acid-selective, and asymmetric aldol reaction, which was for the first time applicable to multifunctional substrates at late stages. The reaction proceeded chemoselectively at the α-position of a carboxyl group, even when substrates contained functional groups of intrinsically more acidic protons, such as ketones, esters, nitriles, and amides. Catalyst-controlled stereodivergent reactions were also possible. The transformation of carboxylic acids to siloxy esters in the reaction mixture, which can be easily cleaved in the workup operation, was critical for this catalysis.</p><p>Mechanistic studies suggested three main roles of the siloxy ester formation: 1) acidification of the α-proton for acceleration of the enolization step, 2) attenuation of detrimental coordination of carboxylate to the boron catalyst, and 3) facilitation of the catalyst turnover through silylation of the boron aldolate intermediate. DFT calculations confirmed that deprotonation of the siloxy ester has a low energy barrier, and the subsequent asymmetric aldol reaction between the Si/B hetero enediolate intermediate and aldehydes proceeded in an inner sphere fashion. The computed reaction selectivity was consistent with the experimental results. We are currently applying this method to late-stage diversification of complex natural products to identify new compounds with better pharmaceutical properties.</p>

ChemRxiv

Seeing the Forest in Lieu of the Trees: Continuum Simulations of Cell Membranes at Large Length Scales

Biological membranes exhibit long-range spatial structure in both chemical composition and geometric shape, which gives rise to remarkable physical phenomena and important biological functions. Continuum models that describe these effects play an important role in our understanding of membrane biophysics at large length scales. We review the mathematical framework used to describe both composition and shape degrees of freedom, and present best practices to implement such models in a computer simulation. We discuss in detail two applications of continuum models of cell membranes: the formation of microemulsion and modulated phases, and the effect of membrane-mediated interactions on the assembly of membrane proteins.

seeing_the_forest_in_lieu_of_the_trees:_continuum_simulations_of_cell_membranes_at_large_length_scal

1. Introduction<!>2. General formulation of the model<!>2.1. Derivation of the model in real space<!>2.2. Real space implementation<!>2.3. Formulation in Fourier space<!>2.4. Fourier space implementation<!>3. Application: Composition fluctuations<!>3.1. Deriving the general model<!>3.2. Results<!>3.3. Phase diagram<!>3.3.1. Numeric calculation<!>4. Application: Hybrid models<!>4.1. General Elastic Model<!>4.2. Hybrid Membrane\xe2\x80\x93Particle model<!>5. Discussion and Outlook

<p>The plasma membrane serves a vital biological role as the cellular boundary. Not only is it the primary barrier that prevents the uncontrolled exchange of material between the cell and its surroundings, it is also instrumental in the spatial organization of cellular components such as the cytoskeleton and proteins embedded in or associated with the membrane. The length scale over which this ordering occurs can be much larger than the size of the molecular components of the membrane, and ranges from tens to hundreds of nanometers.</p><p>The principal chemical structure of the plasma membrane is a bilayer of many different types of lipid molecules. While early models considered these lipids as a homogeneous and passive matrix whose main function was to provide an environment in which membrane proteins could exist [1], it has become apparent that the membrane exhibits significant heterogeneities in both lipid and protein composition. In particular, the concept of lipid rafts, domains enriched in sterol- and sphingolipids less than 100 nanometers in size [2], has received significant attention (see, for example, Ref. [3] and references therein).</p><p>The complexity of the plasma membrane poses a challenge to both the design and the interpretation of experiments that aim to probe its lateral organization. The study of model membrane systems, in which the composition of just a select few lipid types can be controlled, has therefore been instrumental in elucidating mechanisms of spatial ordering. For example, the ability of ternary mixtures to separate into two coexisting liquid phases of differing composition and local order demonstrates that segregation can be achieved even in the absence of proteins [4, 5]. While these composition heterogeneities are too large to be considered rafts, they provide valuable insight into their physical and chemical properties [6–8]. They also indicate that one might be able to employ established tools of statistical mechanics to describe phase separation and equilibria, such as the classical mean field theory of Landau and Ginzburg [9], to describe composition heterogeneities in biological membranes.</p><p>Composition is not the only property of biological membranes that exhibits spatial heterogeneity. The shape of the membrane can also exhibit long-range correlations, and can affect other parts of a cell over large distances. The pioneering work of Canham and Helfrich has shown that both the ground state and the fluctuation behavior of membranes can be understood in terms of basic geometric properties, such as integrals over the local curvature [10, 11]. This enabled the prediction of a large variety of possible cell and vesicle shapes [12].</p><p>There has been a renewed interest in shape deformations, in part due to the discovery of a class of proteins that significantly deform the cellular membrane in order to perform specific biological functions [13]. The membrane responds to the local adhesion of proteins by adjusting its shape, which in turn can induce a long-range interaction between them. Similarly, geometric constraints imposed on the membrane by the actin cytoskeleton can give rise to a remodeling of the actin network [14].</p><p>The present work focuses on the computational modeling of membrane composition heterogeneities and membrane shape deformations on large length scales. Even though they are physically distinct phenomena, they are well characterized by models that are very similar in their mathematical structure. We will therefore introduce these models in a generic form in the next chapter, and present detailed guidance on two possible implementations of these models in a computer program. We then present two specific applications of these models. In Section 3 we show how many of the experimentally observed structures of composition heterogeneities, including separated and modulated phases, can be studied in a unified model. In Section 4 we discuss how even a simple model of a purely geometric protein-membrane coupling can result in novel protein–protein interactions. We finally conclude with an outlook on future developments.</p><!><p>As we will see, both the local composition and the local shape of a membrane can under certain well-defined circumstances be described by a single scalar order parameter, which is typically denoted ϕ(r) or h(r), respectively. Even though composition and shape are two very different physical quantities, the models describing them are mathematically very similar, and in this section we will use the field ϕ(r).</p><!><p>Consider a finite patch of a cell membrane that spans a square of side length L. Let ϕ(r) represent a scalar quantity, such as composition or shape, that is defined at every point r of the two-dimensional patch. We construct an energy functional E for that field as (1)E=∫Lddrα2ϕ(r)2+σ2∣∇ϕ(r)∣2+κ2(∇2ϕ(r))2+b4ϕ(r)4. Here d = 2 is the dimensionality of the system, and α, σ, κ and b are constants that are chosen based on the specific physical system. For example, if the field ϕ(r) represents deviations of the membrane shape from a flat reference configuration, then σ and κ correspond to the surface tension and bending rigidity, respectively (see Section 4).</p><p>We next choose dynamic rules for the field ϕ(r) that allow us to study its time evolution and to calculate statistical ensemble averages. We choose Langevin dynamics of the form (2)∂ϕ(r,t)∂t=−∫Lddr′Λ(r−r′)δEδϕ(r′)+ξ(r,t). The kernel Λ(r) is a generalized mobility that captures the effect of dissipating energy from the field ϕ(r) to an implicit environment, such as the solvent. It is chosen based on the specific physical problem. The force is given by the functional derivative (3)δEδϕ(r)=αϕ(r)−σ∇2ϕ(r)+κ∇4ϕ(r)+bϕ(r)3. The last term in (2) is a stochastic force that satisfies the fluctuation-dissipation theorem, (4)〈ξ(r,t)ξ(r′,t′)〉=2kBTΛ(r−r′)δ(t−t′), where kBT is the thermal energy, and δ denotes the Dirac delta function. This choice enables the system to explore the equilibrium ensemble, i.e, each realization of the field ϕ(r) is sampled according to its Boltzmann weight, exp (–E/kBT).</p><!><p>While the dynamics of the field ϕ(r), as given by (2), depend on the specific choice of the generalized mobility Λ(r – r′), equilibrium averages do not. If one is only interested in the latter, one chooses any kernel function that is computationally convenient. In this section we limit ourselves to a simple point function, (5)Λ(r−r′)=Λ0δ(r−r′).</p><p>To solve the equations (2), (3) and (4) numerically, we discretize the space into a two-dimensional grid of mesh size Δx. For each cell (m, n) we define a coarse-grained variable ϕ‒m,n as the average value of the field ϕ(r) in that cell: (6)ϕ‒m,n(t)=1(Δx)2∫cellm,ndrϕ(r,t)</p><p>If the length Δx is sufficiently small, then (ϕm,n3)¯≈(ϕ‒m,n)3, and we obtain (7)dϕ‒m,ndt=−Λ0[αϕ‒m,n−σ∇2¯ϕ‒m,n+κ∇2¯(∇2¯ϕ‒m,n)+bϕ‒m,n3]+ξ‒m,n(t), where ∇2¯ is the discretized Laplace operator, which can be represented by the 3 × 3 matrix (8)1(Δx)2(1212−122121). Application of this operator is executed as a convolution of this matrix with the field values ϕ‒m,n. Note that one must pay special attention to the boundary conditions. Some numerical libraries provide functions to compute "wrap-around" convolutions that have the effect of assuming periodic boundary conditions. The latter can also be implemented by adding an additional row and column to the array storing the values of ϕ‒m,n that replicate the first row and column, respectively.</p><p>The last term in (7) is the coarse-grained stochastic force. Its statistics are given by (9)〈ξ‒m,n(t)ξ‒m′,n′(t′)〉=2kBTΛ0(Δx)2δm,m′δn,n′δ(t−t′), where δi,j is the Kronecker delta, which is equal to one if i = j, and zero otherwise.</p><p>The stochastic differential equation (7) can be numerically solved using the Euler-Maruyama scheme [15] (10)ϕ‒m,n(t+Δt)=ϕ‒m,n(t)−ΔtΛ0[αϕ‒m,n−σ∇2¯ϕ‒m,n+κ∇2¯(∇2¯ϕ‒m,n)+bϕ‒m,n3]+R, where R is a normally distributed random number with mean zero and variance 2kBTΛ0Δt/(Δx)2. The timestep Δt must be smaller than the fastest time scale of the problem to eliminate numerical instabilities. Using dimensional analysis of the parameters of our model, this implies that Λ0Δt should be small compared to 1/|α|, (Δx)2/|σ|, (Δx)4/κ and (assuming the field ϕ is dimensionless) 1/b.</p><!><p>It is sometimes beneficial to solve the Langevin equation (2) in Fourier space rather than in real space. To that end, we define the Fourier transform of a function f(r), (11)∫Lddrf(r)e−ikr, which we denote interchangeably as f̃k or {f}k. If the function f(r) is real, then the Fourier coefficients satisfy the Hermitian symmetry (12)f~−k=f~k∗, where the star symbol denotes complex conjugation. This implies that f̃0 must be real. The function f(r) can be recovered from these coefficients via the inverse transform (13)f(r)=1Ld∑kf~keikr, where the sum is over all wavevectors k consistent with the problem domain, i.e., each Cartesian component of k is a multiple of 2π/L.</p><p>With these definitions at hand, we can rewrite equations (1), (2) and (3) in terms of the Fourier coefficients ϕ~k: (14)E=12Ld∑k(α+σk2+κk4)∣ϕ~k∣2+b4L3d∑k∑k′∑k″ϕ~kϕ~k′ϕ~k″ϕ~−k−k′−k″(15)∂ϕ~k∂t=−Λ~k{δEδϕ}k+ξ~k(t)(16){δEδϕ}k=αϕ~k+σk2ϕ~k+κk4ϕ~k+bL2d∑k′∑k″ϕ~k′ϕ~k″ϕ~k−k′−k″.</p><p>The last term in (15) is the Fourier transform of the stochastic force. Its statistical properties, given in real space by (4), are (17)〈Re(ξ~k(t))Re(ξ~k′(t′))〉=Λ~κkBTLdδ(t−t′)(δk,k′+δ−k,k′)(18)〈Im(ξ~k(t))Im(ξ~k′(t′))〉=Λ~κkBTLdδ(t−t′)(δk,k′−δ−k,k′)(19)〈Re(ξ~k(t))Im(ξ~k′(t′))〉=0 Note that the k = 0 mode of the stochastic force is unique in that its real part has variance 2Λ~kkBTLdδ(t−t′) and its imaginary part is always zero due to (12), while the real and imaginary parts of all other modes have variance Λ~kkBTLdδ(t−t′). From these equations it follows that (20)〈ξ~k(t)ξ~k′(t′)〉=2Λ~kkBTLdδ(t−t′)δ−k,k′.</p><p>Equations (15) and (16) illustrate the main advantages of the Fourier representation. First, the convolution with the kernel Λ(r – r′) in (2) is replaced by a simple multiplication with its Fourier transform Λ~k. Second, if the model is completely linear (i.e., b = 0) then all Fourier modes ϕ~k are independent dynamical variables (up to the symmetry requirement (12)). In this case, the characteristic timescale for each mode is (21)τk=1Λ~k(α+σk2+κk4). If b ≠ 0 the nonlinear term in (16) introduces coupling of modes, which significantly changes both the physics of the problem and the numerical implementation of these equations.</p><!><p>Since our primary interest is the simulation of fields that represent composition or height fluctuations of biological membranes, we now limit ourselves to two spatial dimensions, i.e., d = 2. In this case, the allowed wavevectors are of the form kp,q = (2πp/L, 2πq/L), and we use the shorthand notation ϕ~p,q for ϕ~kp,q. While the inverse transform (13) in principle involves a sum over an infinite number of such modes, in practice one has to choose a finite number of wavevectors. This is usually done by imposing a high-frequency (or small-wavelength) cutoff, which limits the number of modes to –P ≤ p ≤ P and –Q ≤ q ≤ Q for positive integers P and Q. The logical arrangements of these (2P + 1)(2Q + 1) Fourier modes is illustrated in Fig. 1a.</p><p>Because of the symmetry relation (12) only approximately half of these modes are independent. We choose the modes with {(p, q) : (p ∈ [0, P] and q = 0) or (p ∈ [–P, P] and q ∈ [1,Q])} as the independent modes. This set, which we refer to as k ≥ 0, is also shown in Fig. 1a. It contains 2PQ + P + Q complex-valued modes and the real-valued k = 0 mode, totaling 4PQ + 2P + 2Q + 1 degrees of freedom.</p><p>As in the real space case, we again use the Euler–Maruyama method to integrate (15) for each of the independent Fourier modes over a finite timestep Δt: (22)ϕ~k(t+Δt)=ϕ~k(t)−ΔtΛ~k{δEδϕ}k+R, where R is a complex random number whose independent real and imaginary parts are normally distributed with mean zero and variance Λ~kkBTLdΔt (unless k = 0, in which case R is real and has variance 2Λ~0kBTLdΔt).</p><p>In a computer program one has to store the Fourier amplitudes ϕ~k of these independent modes, for which there are many possible implementations. In many applications it is necessary to also calculate the real-space field ϕ(r), for example to calculate the non-linear term in (16) as shown below, or to couple the dynamics of the field ϕ(r) to other degrees of freedom. This can be done using the inverse Discrete Fourier Transform (DFT), which can be efficiently calculated using the Fast Fourier Transform algorithm. In these cases it is beneficial to store the Fourier amplitudes ϕ~k in a format that can be directly passed to the DFT subroutine. However, as we will see shortly, one must consider some technical details of the DFT to avoid impacting the physics of the model by this design decision driven by computational convenience.</p><p>For a rectangular array of real numbers fm,n, where 0 ≤ m < M and 0 ≤ n < N, the two-dimensional DFT and its inverse are (23)f^p,q=∑m=0M−1∑n=0N−1fm,ne−i2π(mp∕M+nq∕N)(24)fm,n=1MN∑p=0M−1∑q=0N−1f^p,qei2π(mp∕M+nq∕N)(25)=1MN∑p=−M∕2+1M∕2∑q=−N∕2+1N∕2f^p,qei2π(mp∕M+nq∕N), where the last equality holds if M and N are even. In addition to the Hermitian symmetry f^−p,−q=f^p,q∗, the coefficients f̂p,q also satisfy (26)f^p+M,q=f^p,q+N=f^p,q.</p><p>If one wants to use the inverse DFT (25) to calculate the real-space field ϕ(r) on a sampling grid of size (M, N) = (2P, 2Q), then these additional symmetries reduce the number of independent Fourier modes ϕ~k to 2PQ – 2 complex-valued and 4 real-valued modes, leaving 4PQ degrees of freedom. This behavior is illustrated in Fig. 1b.</p><p>There are several complications to this approach. First, limiting Fourier modes with non-zero wavevector to purely real amplitudes breaks the translational symmetry of the underlying problem. Second, this constraint necessitates special handling of the stochastic force term ξ~k(t) in (15) for those wavevectors, which then also must be real. One possibility is to strengthen those forces by a factor of 2 in order to maintain (20). This approach is taken, for example, in Refs. [16–18]. One could also choose to satisfy (17) instead, which is a direct consequence of the fluctuation-dissipation theorem (4). Third, applying the inverse DFT directly to the Fourier coefficients ϕ~p,q does not directly yield a sampling of the function ϕ(r) on a regular sampling grid rm,n = (mL/M, nL/N). Instead one finds (27)ϕ(rm,n)=1L2[MNϕm,n+∑q=−N∕2+1N∕2(−1)mϕ~M∕2,qei2πnq∕N+∑p=−M∕2+1M∕2(−1)nϕ~p,N∕2ei2πmp∕N+(−1)m+nϕ~M∕2,N∕2], where the ϕm,n are the inverse DFT (25) of ϕ~p,q.</p><p>All these complications arise from the behavior of the DFT at the boundary modes p = M/2 and q = N/2 when (M, N) = (2P, 2Q). While the latter two can be corrected for by redefining the Fourier coefficients at those modes, we here propose an alternative approach that can be trivially implemented: we embed the matrix of (2P + 1) × (2Q + 1) Fourier modes in an array of size (M, N) = (2P + 2, 2Q + 2) by adding additional modes that are constrained to have zero amplitude, as shown in Fig. 1c. These "phantom" modes are located at the new boundaries p = M/2 = P + 1 and q = N/2 = Q + 1, and absorb all the artificial symmetries imposed by using the DFT. For example, one can now directly obtain a sampling of ϕ(r) on a regular grid: (28)ϕ(rm,n)=MNL2ϕm,n. These regular sampling points can then be interpolated to any arbitrary point r, as is done, for example, in Refs. [17, 19–22]. In some applications, it can be advantageous to avoid the DFT and instead use (13) directly, as illustrated in Section 4.</p><p>The necessity of choosing a finite basis set in a computer implementation leads to a complication when evaluating the non-linear term in (16), which couples Fourier modes with wavevectors k, k′, k″ and k – k′ – k″. This implies that Fourier modes outside of the represented region |p| ≤ P, |q| ≤ Q are excited, and contribute to the time evolution of those modes that are explicitly propagated. This requires an additional choice in the model for how energy is transferred across the boundary of represented and non-represented modes. One solution is to project out at each timestep the modes that lie beyond the chosen wavevector cutoff.</p><p>A computationally efficient way to evaluate the non-linear term is based on the DFT. Rather than evaluating the double sum over all wavevectors k′ and k″ for each Fourier mode ϕ~k, which would require (MN)3 operations, one uses the inverse DFT to obtain a sampling of the function ϕ(r) on a regular mesh, raises those values to the third power to obtain a sampling of ϕ(r)3 on the same grid, and then uses the DFT to transform back to Fourier space. Because the computational cost of the DFT is on the order of MN log(MN), this approach results in significantly faster performance.</p><p>When evaluating such non-linear terms, one must be cautious of potential artifacts that can arise from the aliasing property of the DFT, which we illustrate for the one-dimensional case: if a signal is sampled on a grid with spacing Δ = L/M, then modes with wavevectors greater than π/Δ will be folded onto Fourier modes with wavevectors less than π/Δ by the DFT. This phenomenon can result in an artificial increase in the amplitudes of Fourier modes close to the resolution limit [23]. In the current application, the signal to be transformed is the function ϕ(r)3, which is known at M evenly spaced points over the range 0 ≤ x ≤ L (similar for the y direction). Because the chosen basis for ϕ(r) contains Fourier modes with |p| ≤ P, the cubic nonlinearity will generate modes at wavevectors |p| ≤ 3P. To fully resolve all these modes in the DFT would require a minimum of M = 6P + 2 sampling points.</p><p>Increasing the number of sampling points can be easily accomplished by padding the array of Fourier coefficients ϕ~k with additional phantom modes, as illustrated in Fig. 1d. By inserting 2δp – 1 columns and 2δq – 1 rows of modes constrained to have zero amplitude in the large wavevector part of the spectrum, one obtains a sampling grid of dimension (M, N) = (2P + 2δp, 2Q + 2δq) using the inverse DFT. Based on the considerations above, one needs a padding of δp = 2P + 1 and δq = 2Q + 1 to determine all Fourier modes of ϕ(r)3.</p><p>This, however, turns out to be wasteful: if we only need to recover the modes |p| ≤ P without aliasing artifacts, it is sufficient to choose the padding δp = P + 1 and δq = Q + 1, which leads to a smaller array of coefficients used in the DFT. In this case aliasing will still occur, but only onto modes with high wavevectors, while the coefficients of interest (|p| ≤ P) are not affected (see Fig. 2).[24] The high wavevector modes are set to zero after each iteration of (22), which corresponds to applying a low-pass filter to the field ϕ(r) at every timestep.</p><!><p>The idea that the cell membrane is inhomogeneous [25] has inspired fruitful theoretical and experimental research [3]. It is motivated by the theme of compartmentalization seen in many levels of biology. Small membrane domains, rafts, are assumed to specialize in specific tasks, such as the enhancement of protein activity by increasing their local concentration. Whether the driving force for the formation of the domains is protein-protein interactions, protein-lipid interactions, or lipid-lipid interactions is not clear. Experiments performed on artificial vesicles composed of a ternary mixture of saturated lipids, unsaturated lipids, and cholesterol have found large scale phase separation into liquid ordered (rich in saturated lipid and cholesterol) and liquid disordered (rich in unsaturated lipid) domains [4]. Such membrane composition is a good model system for the outer leaflet of a mammalian plasma membrane. However, large scale phase separation has not been observed in the plasma membrane of mammalian cells. This may be due to the interaction of the outer leaflet with the inner leaflet, or to the interaction with the cytoskeleton.</p><p>Membrane rafts are small domains (10-200nm) with a short life time. It was proposed that the raft may be seen as a microemulsion [26, 27]. A microemulsion appears when the line tension between two phases is reduced to zero. In a mixture of water and oil, this reduction is produced by the addition of amphiphiles. The mechanism that leads to the reduction of the line tension between liquid ordered and liquid disordered regions is unknown. Brewster et. al. suggest that a hybrid lipid with one saturated and one unsaturated tail can serve as such an agent [28] in a mixture of lipids with two saturated tails and two unsaturated tails. However, the cell membrane contains only a small amount of lipids with two saturated tails. Later work suggests that the hybrid lipids serve both as the bulk and the line agents [29, 30]. In a recent paper [26], Schick shows that the interaction between membrane curvature and membrane composition can produce a modulated phase; a phase that exhibits periodic order in the composition over long length scales. Examples of modulated phases include stripe and hexagonal phases. As the temperature increases, this phase melts into a microemulsion fluid in which the typical size of the microemulsion is similar to the wavelength of the modulation.</p><p>Recent observation by Toulmay et. al. found phase separation in yeast vacuoles in addition to the formation of stripe and hexagonal phases [31]. On the plasma membrane of yeast, small domains were observed [32]. In the following sections we show that these phenomena can be explained by the tendency of a system to phase separate, in conjunction with a reduction of the energy of the boundary between domains.</p><!><p>The model presented in Eq. 1 is general and can be derived from different mechanisms that reduce the line energy between two domains. Take for example a mechanism that couples the membrane composition to the spontaneous curvature [26]; the free energy of the system is given by Etot = Em + Ep + Emp, where Ep is the energy of the composition ϕ in the membrane, Em is the elastic energy of the membrane (see eq. (40) in the following section), and Emp is the interaction of the membrane curvature with the lipids [26, 27]: (29)Ep[ϕ]=∫dr[a2(ϕ)2+γ2(∇ϕ)2](30)Em[h]=∫dr[κ2(∇2h)2+σ2(∇h)2](31)Emp[h,ϕ]=−∫drΓ(∇2h)ϕ. Here, h(r) is the height of the bilayer relative to some reference plane, and κ and σ are the bilayer bending modulus and surface tension, respectively. The parameter a represents the balance between the entropically favored homogeneous state and the attractive interaction energy between like lipids. It is proportional to T – Tc, where Tc is the phase transition temperature in the mean field approximation. The interface energy between domains is controlled by γ. Finally, Γ is the strength of the coupling between the membrane curvature and the membrane composition.</p><p>Taking the Fourier transform and minimizing with respect to h~(k), we obtain the following free energy: (32)Etot[ϕ]=A2(2π)2∫dk{a+γ2[1−(Γ2∕γσ)(1+κk2∕σ)]k2}ϕ~(k)ϕ~(−k).</p><p>The ground state of the system in dominated by wavevectors k with low energetic cost. Hence it is sufficient to describe correctly the free energy around its minimum. For Γ2 < γσ the minimum is at k = 0, while for Γ2 > γσ the minimum is at (Γ−γσ)∕κγσ. Expanding the free energy around the minimum to fourth order in k, we find (33)Etot[ϕ]≈A2(2π)2∫dk{a+γ2[1−(Γ2∕γσ)]k2+Γ2κ2σ2k4}ϕ~(k)ϕ~(−k) for Γ2 < γσ, and (34)Etot[ϕ]≈A2(2π)2∫dk{a+(γσ)3∕22κΓ(Γ2γσ−1)3+γ(γσΓ2−1)k2+γ2κ2Γγσk4}ϕ~(k)ϕ~(−k). for Γ2 > γσ. For Γ2 = γσ the two approximations coincide. As we show below, the system is a microemulsion above the critical Γ.</p><p>These two equations have the same form as the general model introduced in Section 2; see for example (14). By taking the inverse Fourier transform, we recover equation (1) with γ[1−(Γ2∕γσ)]∕2 or γ(γσ∕Γ2−1) as the coefficient of (∇ϕ)2, and Γ2κ∕2σ2 or γ2κ∕(2Γγσ) as the coefficient of (∇2ϕ)2, depending on the value of Γ.</p><!><p>To simplify the analysis, the number of the parameters in the equation is reduced using rescaling. By rescaling time and space, we can eliminate two parameters. In a non-linear equation, as the one above, we can also rescale the field ϕ to eliminate a third parameter. Because we are interested in the effects of a, the line tension σ, and the noise, we rescale Λb and Λκ. By substituting the rescaled parameters r~=r∕r0,t~=t∕t0, recalling that δ(t)=δ(t~)∕t0 and δ(r)=δ(r~)∕r02, we obtain (35)∂ϕ∂t=−[a~ϕ+ϕ3−σ~∇2ϕ+∇2(∇2ϕ)]+ξ(36)〈ξ(r,t)ξ(r′,t′)〉=2kBTκbδ(t−t′)δ(r−r′) with σ~=σ∕(κb)1∕2, a~=a∕b. The derivatives are taken with respect to the dimensionaless variables. The natural time and length scales of the system are 1/Λb and κ∕b4, respectively.</p><!><p>To study the phase diagram of the model, we start with a mean field approximation. This approximation assumes that the system is in a state that minimizes the energy of the system: (37)δEtotδϕ~k=0(38)δEtotδk=0. To understand the nature of the stable state, we calculate the structure factor S(k)=〈ϕ~(k)ϕ~(−k)〉−〈ϕ~(k)〉〈ϕ~(−k)〉. This function is the Fourier transform of the pair correlation function g(r), which is defined in Section 4.2. The bracket 〈·〉 indicates the thermodynamic average or average over time when doing the numerical calculation. Using the Langevin equation guarantees that the two averages coincide if the simulation time is sufficiently long.</p><p>The phase diagram of the model presented in Eq. 1 was calculated using the mean field approximation [33–36] and is presented in [37]. For positive values of a, the system is a regular fluid with 〈ϕ〉 = 0 and the structure factor decays monotonically. As a is reduced, the peak in the structure factor increases (~ 1/a) and diverges at the critical point a = 0. At a < 0 the system phase separates into two coexisting phases, liquid ordered and liquid disordered, with 〈ϕ〉=±ϕo≠0, ϕo=−a∕2b. For σ < 0 and a > σ2/4κ the system is in a microemulsion state with 〈ϕ〉 = 0 and a structure factor that peaks at kc=−σ∕2κ. This fluid state is of particular interest as its fluctuations have a typical length scale 2π/kc, creating domains with finite lifetimes and characteristic sizes. As a is reduced, the time scale of the fluctuation increases until a = σ2/4κ, where the modulations become stable and a modulated phase appears. The peak of the structure factor S(kc) diverges at the transition.</p><p>For σ < 0, as a is reduced a first order transition to a two coexisting phases occurs at a=−σ2∕2κ(1+3∕2). As this transition is first order, near the transition line the modulated phase coexists with the two liquid phases (triple line). Close to the transition, in addition to the stable state, a metastable state that satisfies (37) exists. Hence, the final state depends on the initial condition of the system. As the metastable state loses its stability along the spinodal line, the size of the fluctuation increases. At the point where the triple line intersects with the transition to the fluid phase (the tricritical point), the spinodal meets the line of first order transition. Hence, close to the tricritical point we can approximate the transition from the modulated phase to the two-phase coexistence using the enhancement of the fluctuations.</p><!><p>We now turn to find the phase digram beyond the mean field approximation and understand the effect of fluctuations on the phase diagram. As in the mean field approximation, we expect three kinds of phases: a fluid phase, two coexisting liquids phases, and a modulated phase.</p><p>To identify the transitions between the phases we use three indicators: 1) The peak of the structure factor Smax(a) is expected to be maximal at the second order transition. 2) The distribution P(ϕ) of the field. One peak in P(ϕ) around zero indicates the fluid phase, while two symmetric peaks indicate a modulated phase, and asymmetric peaks (which can result from a bias in the system's initial condition) indicate coexisting liquids. 3) The average 〈ϕ(k ≠ 0)〉, which is zero in the fluid phase or in coexisting liquids because undulations are short lived and the phase of ϕ(k ≠ 0) changes rapidly. In the modulated phase, ϕ(k ≠ 0) is changing due to drift in the direction of the stripe, but the change is slow, hence 〈ϕ~(k)〉〈ϕ~(−k)〉≈〈ϕ~(k)ϕ~(−k)〉 for k close to kc.</p><p>Integrating the system (35) over time, we accumulate statistics of ϕ(r), from which we find S(k, a) (Fig. 4). We denote kmax the wavevector for which S(k, a) has a maximum: Smax(a) = S(kmax, a). Plotting Smax(a) for fixed σ, we find the point where the fluctuation are the largest, which we identify with the phase transition.</p><p>Note that in a first order transition, increasing fluctuations are generally a signature of the spinodal line; the line where the metastable state loses its stability. However, close to the tricritical point the spinodal line is close the the transition line. Far from the tricritical point one can compare the mean energy of the two stable states, where the point of equal energy denotes the first order transition.</p><p>Fig. 4 shows Smax for two cases with σ < 0. The first case, where σ is close to zero, shows only one peak in Smax. The second case of σ < < 0, shows two peaks as expected from the mean field approximation. Calculating kmax, we find that the former corresponds to a transition from the microemulsion phase to two-phase coexistence. This transition does not exist in the mean field approximation. It suggests that systems which show phase separation can support the formation of a microemulsion.</p><p>P (ϕ) and 〈ϕ~(k)〉〈ϕ~(−k)〉 confirm that the transition is from a microemulsion phase directly to two-phase coexistence if σ is negative but not too, small, while for σ ≲ −1 there are two transitions: one from a microemulsion to a modulated phase, and one from there to two-phase coexistence.</p><p>To find the first order transition far from the tricritical point, we calculate the mean energy, 〈Etot〉, as a function of a, starting from two different stable states of the system. Plotting the two curves 〈Etot〉 vs. a, the point of intersection indicates the triple line. Close to the tricritical point, the slope of the two curves are very similar, and the intersection point is difficult to determine. Hence, we use the peak of the structure factor to identify the transition, as described above.</p><p>Figure 5 shows a time frame from the simulation near the critical line. In the first case, the system temperature a is raised and the system stays in a uniform state, while in the second the temperature is reduced and the system remains in a modulated phase. This shows, as expected, that near the first order transition the final state depends on the way the system was prepared. We show S(k), 〈ϕ~(k)〉〈ϕ~(−k)〉, and the the distribution P(ϕ) for each phase.</p><p>To summarize this section, we find that our simple model can explain different structures observed on membranes: the transition of a uniform phase to two coexisting phases, and the formation of modulated phases like stripes and hexagonal phase. Including the effects of fluctuations, we find that the critical temperature is reduced, and that there is a direct transition from a microemulsion fluid to two-phase coexistence. Lastly, we show that as a modulated phase melts to a uniform phase, this phase is a microemulsion with a characteristic length scale that resembles that of the modulated phase.</p><!><p>At length scales much larger than its thickness, a biological membrane behaves like a two-dimensional fluid sheet, and its properties are dominated by two material constants: surface tension and bending rigidity. By treating the membrane as a two-dimensional surface embedded in three-dimensional space, one can derive computationally efficient models of membrane shape dynamics, which we will interface with a general model of membrane-associated proteins.</p><!><p>The standard model for understanding the shape and fluctuations of a biological membrane was developed by Canham and Helfrich [10, 11, 38, 39]. It asserts that the energy of membrane conformation can be written as (39)Em=∫dS[σ+κ2(1R1+1R2−2C0)2+kg1R1R2], where σ and κ are the surface tension and bending rigidity, respectively, R1 and R2 are the principal radii of curvature, C0 is the spontaneous curvature, kg is the saddle-splay modulus, and the integral is taken over the membrane surface.</p><p>According to the Gauss-Bonnet theorem, the Gaussian curvature term in (39) is a topological invariant. Because the membrane topology cannot change within our model, this term merely adds a constant to the total energy and can therefore be discarded. For simplicity we will focus on membranes without spontaneous curvature, C0 = 0, such as a homogeneous symmetric bilayer.</p><p>We limit ourselves to membrane conformations that are deviations from a completely flat membrane without overhangs. In this case, the shape of the membrane can be parametrized by a single height function h(r), where r = (x, y). This is known as the Monge gauge representation. If the deviations from the reference shape are small, then we can expand the energy (39) to quadratic order in h and its derivatives, and finally obtain (40)Em=∫L2dr[σ2(∇h(r))2+κ2(∇2h(r))2]. This expression is a special case of the general energy functional (1), where ϕ(r) has been replaced with h(r). Following the procedure outlined in Section 2.3, we express the energy in terms of the Fourier components h~k of the membrane height field: (41)Em=1L2∑k(σ2k2+κ2k4)∣h~k∣2. This transformation implies the use of periodic boundary conditions. The equation of motion for the Fourier coefficients is the Langevin equation (42)∂h~k∂t=−Λ~k{δEmδh(ri)}k+ξ~k(43)=−Λ~k(σk2+κk4)h~k+ξ~k. In these expressions Λ~k is the Fourier transform of the Oseen tensor, which accounts for the viscosity of the surroundings. It plays the role of a generalized mobility in the Langevin equation that captures hydrodynamic effects. For a membrane surrounded by a solvent of viscosity η it takes the form: [16, 39] (44)Λ~k=14ηk.</p><p>The last term in (43) is a Gaussian stochastic noise term that satisfies the fluctuation-dissipation theorem (20): (45)〈ξ~k(t)〉=0(46)〈ξ~k(t)ξ~k′(t′)〉=2kBTL2Λ~kδk,−k′δ(t−t′)</p><p>The dynamical scheme encoded in these equations is known as Fourier Space Brownian Dynamics [16, 39–41]. In the present form it simulates the dynamics of a free membrane embedded in implicit solvent. After integrating (43), information about the thermal fluctuations of the membrane is obtained. From the equipartition theorem, the average fluctuations of the membrane height are given by (47)〈∣h~k∣2〉=kBTL2σk2+κk4. Obtaining this fluctuation spectrum is an important test for the correctness and convergence of a computer simulation of the dynamics of a free membrane.</p><!><p>While the structure and dynamics of a free membrane is well understood within the Canham-Helfrich framework, there has been recent interest in coupling this model with a particle description of other cellular components, such as membrane proteins. Trying to combine a continuum representation with a discrete representation poses severe challenges both in the mathematical formulation and in the computational implementation of such models.</p><p>We are guided in the development of our model by the interaction between the plasma membrane and the actin cytoskeleton. The presence of actin filaments locally quenches membrane fluctuations, mainly due to steric interactions. For simplicity we will neglect all chemical detail of the filaments, and only maintain the position at which the membrane height is pinned to a specific value by the protein. This system is illustrated in Fig. 6.</p><p>In our model, the energy of the system can be decomposed into a contribution from the free membrane, membrane-protein interactions, and protein-protein interactions: (48)Etotal({h~k},{ri})=Em({h~k})+Emp({h~k},{ri})+Epp({ri}). The first term is the energy of the free membrane, (41). The second term represents the coupling of the membrane with the proteins, which we write as (49)Emp({h~k},{ri})=12ε∑i=1N(h(ri)−l)2. Here, the two-dimensional vector ri is the location at which protein i limits the membrane height to small fluctuations around the fixed length l. The parameter ε describes the strength of the constraint imposed by the protein.</p><p>The equation of motion for the membrane height is derived using the Langevin formalism described in Section 2.3. In the presence of membrane-protein coupling, equation (15) becomes (50)∂h~k∂t=−Λ~k[(κk4+σk2)h~k+ε∑ie−ikri(h(ri)−l)]+ξ~k.</p><p>For the protein-protein interaction we assume a pairwise additive potential that depends only on the separation between two proteins, (51)Epp=∑i<jV(∣ri−rj∣), where V(r) is the pair potential. We choose a generic and purely repulsive potential that has the functional form of the Weeks-Chandler-Andersen (WCA) potential [42], (52)V(r)={0ifr≥21∕6σpp4εpp[(σppr)12−(σppr)6]+εppifr<21∕6σpp.} The parameters εpp and σpp set the energy and length scale for the protein-protein interaction, respectively.</p><p>We still need to specify the dynamics for the filaments. We assume that they undergo Brownian motion in the (x, y) plane, governed by the Langevin equation (53)∂ri∂t=−γp∇iEtotal+ζi(t). Here ∇iEtotal is the gradient of the energy with respect to the position of particle i and γp is the mobility coefficient. The latter is related to the bare diffusion constant through D0 = kBTγp. Evaluating the gradient yields (54)∂ri∂t=−γp[ε∑i(h(ri)−l)∇h(ri)+∇iEpp]+ζi(t). We note that the gradient of the height field is given by (55)∇h(ri)=1L2∑k(ik)h~keikr→i. The final term in (53) is a stochastic force acting on protein i. Its statistics are governed by the fluctuation-dissipation theorem (56)〈ζi,α(t)ζj,β(t′)〉=2γpkBTδijδαβδ(t−t′), where α and β denote the Cartesian components of the vector ζi.</p><p>Numerically integrating equations (50) and (54) in time allows us to simulate the time evolution of both membrane geometry and protein positions. For this we need to choose a time step, Δt, that is small enough to resolve all pertinent dynamic processes [17]. There are multiple bounds to the time-step that must be considered. The first depends on the relaxation time (21) of the fastest membrane mode in the system, which requires that (57)Δt≪4ησkmax+κkmax3, where kmax is the largest wavevector in the system. To avoid proteins moving over distances comparable to their diameter in a single timestep, we also require Δt≪σpp2∕4D0. Similarly, the distance travelled in a single timestep should be small compared to the shortest wavelength of the membrane Fourier modes, Δt≪π2∕Dkmax2.</p><p>At this point we emphasize similarities and differences with existing hybrid models. The effect of locally suppressing membrane height fluctuations at static pinning sites has been studied in Ref. [16] to investigate the role of the spectrin network on the membrane of red blood cells. The diffusion of single inclusions within the membrane has been investigated in Refs. [19–21]. In these studies, the protein is coupled to the membrane curvature, rather than the membrane height. The effect of multiple proteins that locally pin the membrane height to fixed values has been studied in a lattice model [43, 44]. In our model, we focus on the effect of the membrane on the structure of proteins diffusing in the (x, y) plane.</p><p>Our approach also differs in details of the numerical implementation. Most authors use the DFT to convert the membrane Fourier modes to real space. This results in a sampling of the height field h(r) on a regular grid. If the proteins diffuse in the continuum, one must interpolate those sampling points, which can result in artifacts that must be corrected [21, 22]. Since the number of proteins is relatively small, we instead use equations (13) and (55) directly to compute the membrane height and its derivative at the positions of the proteins.</p><p>To quantify the effect of the membrane-induced interaction among the proteins, we calculate the radial distribution function (58)g(r)=L2N2∑i<j〈δ(r−(rj−ri))〉, which is a measure of correlations among proteins.</p><p>This function is shown in Figure 7 for a membrane with N = 100 proteins. If the bending rigidity is infinite, then the membrane is completely flat, and the proteins behave like a two-dimensional WCA fluid. Lowering κ, and thereby increasing membrane fluctuations, leads to a broad peak at a distance r ≳ σpp. This illustrates the attractive nature of membrane-induced attractions that originate from suppressing membrane fluctuations.</p><!><p>The present work illustrates how modeling of either the composition or the shape as a continuum field can shed light on the complicated behavior of biological membranes at large length scales. A natural question to ask is whether additional phenomena set in when one follows the coupled dynamics of both fields explicitly. Experiments suggest that there are indeed situations where these distinct physical quantities are tightly coupled, see for example Refs. [45–48]. While both fields were considered in Section 3.1, integrating over the shape fluctuations implies that these geometric degrees of freedom are constantly in equilibrium, which might not always be the case in an experimental systems. Using the methods outlined in Section 2, it is straightforward to study the coupled dynamics of both fields [49, 50].</p><p>The plasma membrane is asymmetric, i.e., the chemical compositions of the inner and the outer leaflet differ significantly. Therefore, the leaflets can have different tendencies to phase separate or to exhibit other forms of spatial organization. This is also true for the coupling of each leaflet's composition to the shape of the membrane. These effects can be captured in models with not one, but two composition fields (one for each leaflet). Coupling of these fields gives rise to complex phenomena that are currently being explored [51, 52].</p><p>To study the effect of membrane-mediated interactions between proteins, we have presented a hybrid model that couples a continuum description of the membrane with a particle representation of the proteins. We have assumed a form of the protein-membrane interaction that describes a harmonic constraint on the membrane height above a reference plane, as is appropriate for proteins that suppress height fluctuations. Other proteins, such as sca olding proteins [13], can couple to other geometric quantities of the membrane, for example its curvature. Hybrid models that take such coupling into account have been developed in Refs. [19, 21, 22]. While we have focused on proteins whose interaction can be described by a generic isotropic pair potential, it is desirable to incorporate into the model more complicated protein–protein interactions. The first step in this direction would be to endow proteins with rotational, in addition to translational, degrees of freedom, for example in a rigid body framework of protein motion.</p><p>While continuum models of the membrane are the central topic of this report, we would like to end with a brief discussion of particle models of biological membranes. The recent development of coarse-grained force fields has significantly increased the length scale accessible to molecular modeling techniques. Membranes consisting of thousands of lipid molecules can be simulated over microsecond timescales with models that maintain a significant degree of chemical fidelity, such as the Martini force field [53, 54]. This allows the simulation of phase separation processes over tens of nanometers. Even bigger system sizes can be used with more aggressively coarse-grained and solvent-free models [55–57]. For example, a microsecond simulation of a 120 nm diameter vesicle has recently been reported [56]. While each lipid molecule is represented in these models by so few degrees of freedom that further coarse-graining at the molecular level seems unlikely, it is possible to design particle models in which each fundamental unit represents multiple lipid molecules or small patches of membrane [58–62]. These models accurately reproduce the mesoscopic elastic properties of membranes, and can therefore complement continuum models to study the coupling of membrane shape and protein organization.</p>

PubMed Author Manuscript

Ganoderic Acid A Attenuates LPS-Induced Neuroinflammation in BV2 Microglia by Activating Farnesoid X Receptor

Neuroinflammation plays an important role in the onset and progression of neurodegenerative diseases. Microglia-mediated neuroinflammation have been proved to be the main reason for causing the neurodegenerative diseases. Ganoderic acid A (GAA), isolated from Ganoderma lucidum, showed anti-inflammatory effect in metabolism diseases. However, little research has been focused on the effect of GAA in neuroinflammation and the related mechanism. In the present study, lipopolysaccharide(LPS)-stimulated BV2 microglial cells were used to evaluate the anti-inflammatory capacity of GAA. Our data showed that GAA significantly suppressed LPS-induced BV2 microglial cells proliferation and activation in vitro. More strikingly, GAA promoted the conversion of BV2 microglial cells from M1 status induced by LPS to M2 status. Furthermore, GAA inhibited the pro-inflammatory cytokines release and promoted neurotrophic factor BDNF expression in LPS-induced BV2 microglial cells. Finally, we found that the expression of farnesoid-X-receptor (FXR) was prominently downregulated in LPS-stimulated BV2 microglial cells, antagonism of FXR with z-gugglesterone and FXR siRNA can reverse the effect of GAA in LPS-induced BV2 microglial cells. Taking together, our findings demonstrate that GAA can significantly inhibit LPS-induced neuroinflammation in BV2 microglial cells via activating FXR receptor.

ganoderic_acid_a_attenuates_lps-induced_neuroinflammation_in_bv2_microglia_by_activating_farnesoid_x

Introduction<!><!>Cell Culture<!>Cell Counting Kit-8 Assay<!>Drug Treatments<!>Transient Transfection with siRNA<!>ELISA for IL-1β, IL-6, TNF-α and BDNF<!>Western Blot<!>Immunofluorescent Staining<!>Image Analysis<!>Statistical Analysis<!><!>Discussion<!>

<p>Neuroinflammation, inflammation of the central nervous system (CNS), is an immune response often initiated against a variety of harmful stimuli, including pathogens, trauma and neural damage, etc. Accumulative evidence strongly suggested that neuroinflammation is a common feature of neurodegenerative diseases, such as Parkinson's diseases (PD), multiple sclerosis (MS) and Alzheimer's diseases (AD), and is associated with the progressive loss of neuronal structure and function [1–3]. The inflammation reaction is an automatic defense response of the body to external stimuli. In some cases, it is usually beneficial because it can promote the clearance of pathogenic factors and the healing of damaged tissue; but in other cases, it is detrimental because it can aggravate the damage of injured tissue or cells and worsen the condition [4]. The strategies to modulate the inflammatory processes are increasingly considered as the candidate options to therapy inflammation related disease.</p><p>Microglia are the resident macrophages of the CNS and plays an important role in immune surveillance, homeostasis and neuroinflammation [5]. Under normal conditions, microglia not only provide immune surveillance but also respond to harmful stimuli; under pathologic conditions, microglia can be activated in order to respond to the detrimental signals. Similar to macrophages, microglia was heterogeneous [6]. Generally speaking, activated microglia can be categorized as the classic pro-inflammatory M1 type or the anti-inflammatory M2 type. M1 type microglia were characterized by an overproduction of inflammatory cytokines and inflammatory mediators, including tumor necrosis factor(TNF)-α, interleukin(IL)-6, IL-1β, inducible nitric oxide synthase(iNOS) and prostaglandin G2(PG2), etc., [7]. On the contrary, M2 type microglia were characterized by the secretion of anti-inflammatory cytokines including IL-4, IL-10 and transforming growth factor(TGF)-β [8]. M1 type microglia play a detrimental effect while M2 type microglia exert a neuroprotective and regenerative effect. Therefore, it is of great importance to regulate the differentiation of microglia and reduce the inflammatory damage.</p><p>Taking these factors into consideration, researchers focus their interest on natural products with potential anti-inflammatory and neuroprotective effects. Previous studies have discovered many natural products, which can converse the polarization of microglia from M1 to M2 in vitro and in vivo [9–11]. Ganoderic acid A (GAA), isolated from Ganoderma lucidum, is proved to exert anti-tumor, anti-oxidant, anti-inflammatory and hepatoprotective effects [12–14]. The protective role of Ganoderma lucidum extracts on neurons has also been well studied [15]. However, the specific effect of GAA on neuroinflammation remains unknown, even though GAA is a major pharmaceutically active compound of Ganoderma lucidum. Based on these findings, we hypothesize that GAA has an inhibitory effect on neuroinflammation and can interfere with microglial polarization.</p><p>The farnesoid-X-receptor (FXR, NR1H4), also known as a bile acid receptor, was a ligand-activated transcriptional factor and belongs to the nuclear hormone receptor superfamily. FXR has been extensively studied in human metabolic disease [16, 17]. Recently, the researcher found that FXR plays a neuroprotective role in multiple sclerosis [18]. Growing evidence indicated that GAA can activate FXR [19]. However, whether GAA can inhibit inflammation via activation FXR, it remains unclear. Therefore, this study aimed to investigate the effects of GAA on LPS-induced inflammation of microglial cells and to explore the involved mechanisms.</p><!><p>Chemical structure of Ganoderic acid A</p><!><p>Murine BV2 microglial cell line was provided by Dr. Qi Yan, Yunnan University of Traditional Chinese Medicine. The cells were cultured in DMEM high glucose complete medium (Cat: 10-013-CVRC), supplemented with 10% fetal bovine serum (FBS) (Cat: 04-0011-1ACS) and 1% penicillin streptomycin solution at 37 °C in a humidified incubator under 5% CO2 in T25 flasks. When reached over 80% confluence, cells were seeded onto 96-well, 24-well or 6-well plate for further experiments.</p><!><p>BV2 microglial cells were plated in 96-well plates at a density of 5 × 103 cells per well, all of the study was conducted 24 h after the cells were seeded. Cells were then treated with LPS (0.1, 0.25, 0.5, 0.75, 1 and 2 μg/ml), GAA (1, 25, 50, 75, 100 and 200 μg/ml) and GS (10, 20, 40, 60, 80 and 100 μM) for 24 h. After treatment, the cells were rinsed with PBS for twice and the medium was changed to 100 µl DMEM high glucose medium and 10 μl of CCK-8 was added into the culture plate. Followed by incubation at 37 °C for 2 h, the optical density value at the wavelength of 450 nm was detected by using a microplate reader (Epoch, BioTek Instruments, Winooski, USA). After correction by subtracting the optical density value of wells that did not contain cells, experimental data were shown as relative cell viability normalized to the control group [13].</p><!><p>BV2 microglial cells were stimulated with 0.5 μg/ml LPS as an inflammation state in vitro. For the study of GAA on LPS induced neuroinflammation, GAA was administrated in simultaneously with LPS to the BV2 cells. After 24 h treatment, cytokines, FXR and microglial biomarker were detected by western blot and immunofluorescence methods.</p><p>For the study of GS (a selective FXR receptor antagonist) on GAA and LPS co-treated BV2 microglial cells, GS was administrated to BV2 microglial cells 2 h before LPS and GAA treatment. After 24 h treatment, cytokines were detected by western blot method.</p><!><p>When BV2 microglial cells were confluent to 60–70%, they were transfected with FXR siRNA (1.5 μg) or negative control siRNA (1.5 μg) using the DNAfectin™ Plus Transfection Reagent (Cat: G2500, Applied Biological Materials Inc). The siRNA sequence targeting FXR 5′-GGCGUAGCAUUACCAAGAATT-3′ was designed and supplied by GenePharma. After 36 h, the DNAfectin™ Plus Transfection Reagent were removed and the cells were treated with GAA and LPS. 24 h later, the inhibition of siRNA on FXR expression and the expression of TNF-α and BDNF in BV2 microglial cells were detected by western blot.</p><!><p>After 24 h treatment, the levels of IL-1β, IL-6, TNF-α and BDNF in cell culture supernatant were measured according to manufacturer's instructions using ELISA kits. Results were expressed as pg/ml of supernatant.</p><!><p>After 24 h treatment, the cell culture medium was discarded and the cells were washed three times with ice-cold PBS. 200 μl of RIPA cell lysis buffer mixed with protease and phosphatase inhibitors were added to each well, then cells were incubated on ice for 30 min, and the lysate was collected by spinning at the speed of 12,000 rpm for 10 min at 4 °C, the supernatants were used for following study. Protein concentrations were determined using a BCA protein assay kit (Pierce Biotechnology, Rockford, USA). Equal amounts of proteins were subjected to 10–12.5% SDS-PAGE gel electrophoresis and transferred to 0.22 µm polyvinylidene difluoride (PVDF) membranes (Cat: ISEQ00010, Merck Millipore Ltd). Antibodies against Iba1, iNOS, Arg-1, IL-1β, IL-6, TNF-α, BDNF and FXR were used as primary antibodies. Secondary antibodies, including Donkey anti-Goat IgG (H + L) HRP, Goat anti-Rabbit IgG (H + L) HRP and Goat anti-Mouse IgG (H + L) HRP. The anti-GAPDH and anti-Tubulin antibodies were applied for loading calibration. Immunoreactive bands were visualized using the ECL detection system (Millipore, Billerica, USA). The images were acquired by the chemiluminescent imaging system (Amersham Imager 600, GE) and quantified using Image Pro Plus version 6.0 software (Media Cybernetics, Rockville, USA).</p><!><p>After LPS and GAA treatments, cultured BV2 microglial cells were washed thrice with cold 1 × PBS and fixed in 4% paraformaldehyde in PBS for 20 min at room temperature. The cells were then incubated with blocking buffer (1% BSA and 0.2% Triton X-100 in PBS) for 1 h at room temperature. Next, cells were incubated with primary anti-Iba1, anti-iNOS, anti-Arg1 and anti-FXR antibodies at 4 °C overnight. Cells were then washed with PBST for three times, appropriate secondary antibodies labeled with Alexa Fluor 488 or Alexa Fluor 594 was prepared in PBST containing 5% BSA. After washing, cells were incubated with second antibody solution for 1 h at room temperature and rinsed with PBST thrice. After washing, the cells were mounted onto slides with anti-fade mounting media containing DAPI solution.</p><!><p>All slides were photographed and digitized using a video camera mounted on a Leica microscope (Leica DM2500, Germany). All images were taken under exactly the same conditions, including laser output strength, exposure time, gain, offset, etc. BV2 microglial cells were randomly photographed, with 5 or more images obtained for each coverslip to ensure that conditions for each coverslip in each treatment group were the same. Pictures were further processed using Adobe Photoshop CS5 (Adobe Systems Software, Ireland).</p><!><p>All data were analyzed with one-way ANOVA followed by Turkey post hoc test. All data were analyzed using Graph Pad Prism Ver. 5.0 (Graph Pad Software, Inc., San Diego, CA) and expressed as the mean ± SEM. P values less than 0.05 were considered statistically significant. Figures were generated by GraphPad Prism version 5 software.</p><!><p>GAA suppressed the LPS-induced BV2 microglial cells proliferation and activation in vitro. a BV2 cells were cultured with different concentration of GAA for 24 h. b BV2 cells were stimulated with different concentration of LPS for 24 h. c BV2 cells were cultured with different concentration of GAA in the presence of 0.5 μg/ml LPS for 24 h. Cell proliferation was detected by CCK-8 assay. d Immunofluorescence images showing the BV2 microglial cells after LPS stimulation which was labeled with anti-Iba1 antibody, With GAA, the expression of Iba1 is decreased. Scale bar equals to 100 μm. e The protein levels of Iba1 were detected by Western blot. After normalization to the control, data were analyzed using one-way ANOVA followed by post hoc Turkey tests and were presented as Mean ± SEM for three independent experiments. (a–c *P < 0.05 LPS 0.5 µg/ml vs. CON, **P < 0.01 LPS 0.75 µg/ml vs. CON, #P < 0.05 LPS + GAA 50 µg/ml vs. LPS; Fig. 1e, **P < 0.01 LPS vs. CON; ##P < 0.01 LPS + GAA vs. LPS)</p><p>GAA suppressed the up-regulation of iNOS and the down-regulation of Arg-1 in LPS-stimulated BV2 microglial cells. a Immunofluorescence images showing the BV2 microglial cells after LPS stimulation which were labeled with anti-iNOS or anti-Arg-1 antibody. With GAA, the expression of iNOS was decreased and the expression of Arg-1 was increased. Scale bar equals to 100 μm. b The protein levels of iNOS were detected by Western blot. c The protein levels of Arg-1 were detected by Western blot. After normalization to the control, data from three independent experiments was analyzed using one-way ANOVA followed by post hoc Turkey tests and were presented as Mean ± SEM. (*P < 0.05, ***P < 0.001 LPS vs. CON; ##P < 0.01, ###P < 0.001 LPS + GAA vs. LPS)</p><p>The effects of GAA on IL-1β, IL-6, TNF-α and BDNF expression levels in LPS-stimulated BV2 microglial cells. The protein levels of IL-1β (a), IL-6 (b), TNF-α (c) and BDNF (d) were detected by Western blot. After normalization to the control, data from three independent experiments were analyzed using one-way ANOVA followed by post hoc Turkey tests and were presented as Mean ± SEM. The protein levels of IL-6 (e) and TNF-α (f) were detected by ELISA assay. Data were analyzed using one-way ANOVA followed by post hoc Turkey tests and were presented as Mean ± SEM. N = 5–6 each group. (*P < 0.05, **P < 0.01, ***P < 0.01 LPS vs. CON; #P < 0.05 LPS + GAA vs. LPS)</p><p>GAA reversed the down-regulation of FXR in LPS-stimulated BV2 microglial cells. a Immunofluorescence images showing the BV2 microglial cells after LPS stimulation which were labeled with anti-FXR antibody. With GAA, the expression of FXR is significantly up-regulated. Scale bar equals to 100 μm. b The protein levels of FXR were detected by Western blot. After normalization to the control, data from three independent experiments was analyzed using one-way ANOVA followed by post hoc Turkey tests and were presented as Mean ± SEM. (**P < 0.05 LPS vs. CON; #P < 0.05 LPS vs. LPS + GAA)</p><p>GS dose-dependently blocked the expression of FXR in LPS-stimulated BV2 cells. a BV2 cells were stimulated with different concentration of GS for 24 h. b The protein levels of FXR were detected by Western blot. After normalization to the control, data from three independent experiments was analyzed using one-way ANOVA followed by post hoc Turkey tests and were presented as Mean ± SEM. (*P < 0.05 GS 30 μM vs. control; ***P < 0.001 GS 100 μM vs. control)</p><p>GS inhibited the anti-inflammatory effects of GAA in LPS-stimulated BV2 microglial cells. The protein levels of TNF-α (a) and BDNF (b) were detected by Western blot. GS was administrated to BV2 microglial cells for 2 h before LPS and GAA treatment for 24 h. After normalization to the CON, data from three independent experiments was analyzed using one-way ANOVA followed by post hoc Turkey tests and were presented as Mean ± SEM. (*P < 0.05, ***P < 0.001 LPS vs. CON; ##P < 0.01, ###P < 0.001 LPS + GAA vs. LPS; $$P < 0.01, $$$P < 0.001 LPS + GAA + GS vs. LPS + GAA)</p><p>The effects of FXR knock-down on GAA-mediated expression of FXR, TNF-α and BDNF levels in LPS-stimulated BV2 microglial cells. The protein levels of FXR (a), TNF-α (b) and BDNF (c) were detected by Western blot. After normalization to the CON, data from three independent experiments was analyzed using one-way ANOVA followed by post hoc Turkey tests and were presented as Mean ± SEM. (***P < 0.001 FXR-NC vs. FXR-si-RNA; **P < 0.01, ***P < 0.001 LPS vs. CON; ##P < 0.01, ###P < 0.001 LPS + GAA vs. LPS; $$$P < 0.001 LPS + GAA + FXR-si-RNA vs. LPS + GAA)</p><!><p>We sought to study the effects of GAA on LPS-induced neuroinflammation in BV2 microglial cells and its underlying mechanisms. We found that (1) GAA significantly inhibits LPS-induced BV2 microglial cells proliferation and activation in vitro; (2) GAA promoted the conversion of LPS-induced microglial cells from M1 status to M2 status; 3) GAA prominently attenuated pro-inflammatory cytokines IL-1β, IL-6 and TNF-α and enhanced neurotrophic factor BDNF expression in LPS-induced BV2 microglial cells; (4) GAA reversed LPS-induced FXR down-regulation in BV2 microglial cells; (5) the effects of GAA were blocked after FXR antagonist GS or FXR siRNA treatment in LPS-treated BV2 microglial cells.</p><p>Microglia-mediated neuroinflammation is a hall mark of neurodegenerative diseases, including AD, PD, amyotrophic lateral sclerosis (ALS) and MS [6]. Microglia are the resident neuroimmune cells of the central nervous system and play an important role in maintaining homeostasis in normal conditions [22]. In response to injury or stimuli, microglia become readily activated and consequently modulates their phenotypes to adapt the activated state. Accumulating evidence strongly showed that LPS can activate BV2 microglial cells to produce various cytokines, nitric oxide, PGE2, COX2 and iNOS, hence LPS-stimulated BV2 microglial cells were often used as an in vitro neuroinflammation model [23–25]. Previous studies have shown that LPS can induce BV2 microglial cells and brain resident microglia proliferation and activationin vitro and vivo [26–28]. Subsequently, activated microglia releases inflammatory mediators such as TNF-α, IL-1β, IL-6. These inflammatory factors in turn act on microglia and brain and lead to neurodegenerative diseases [29].Cumulative studies have showed that pharmacologic regulation of microglia activation is effective in the treatment of neurodegenerative diseases [30, 31]. Consistent with previous results, our results showed that GAA inhibited LPS-induced BV2 microglial cells proliferationand activation, indicating that it plays an important role in neuroimmune regulation.</p><p>Activated microglia, as in macrophages, the phenotypes were heterogeneous, can be divided into either M1 or M2 type [6], which are considered neurotoxic or neuroprotective, respectively [32]. Activation of M1 type microglia releases diverse pro-inflammatory cytokines and oxidative stress-induced free radicals that promotes neuroinflammation and inhibits brain repair. Conversely, activation of M2 type microglia improves brain repair and inhibits neuroinflammation by releasing anti-inflammatory cytokines, neurotrophic cytokines, and enhancing phagocytosis. However, previous studies have shown that most of compounds can suppress neuroinflammation simply by inhibiting M1 microglia activation [10, 33, 34], few compounds can suppress neuroinflammation by promoting the conversion of M1 type to M2 type microglia [35, 36]. Our results showed that GAA treatment significantly inhibited the up-regulation of iNOS and the down-regulation of Arg-1 expression, which indicate that GAA worked as the molecular switch to convert microglia from M1 to M2 type and alleviated inflammation.</p><p>Stimulating of BV2 microglial cells by LPS lead to the production of pro-inflammatory cytokines, such as IL-1β, IL-6 and TNF-α, which have been confirmed that could cause neural cell damage, initiate and amplify the inflammatory response, and lead to the development of neurodegenerative diseases [29]; Therefore, the suppression of their production is pivotal for prevention of neurodegenerative diseases [2, 11]. On the contrary, microglia can also secrete anti-inflammatory cytokines and some neurotrophic factors to ameliorate neurodegenerative disease [37, 38], such as BDNF. The expression of IL-4, TGF-β and BDNF were detected in this study, unfortunately we did not discover the any changes of IL-4 and TGF-β in LPS-induced BV2 microglial cells. Previous study had reported that GAA could not decrease the level of TNF-α, IL-6 and IL-1β in the cell culture supernatants of LPS-stimulated primary mouse microglia, but GAA could decrease the expression of TNF-α, IL-6 and IL-1β in the cell lysates of LPS-stimulated primary mouse microglia [39]. Our results showed that GAA significantly attenuated pro-inflammatory cytokines IL-1β, IL-6 and TNF-α and enhanced neurotrophic factor BDNF expression in cell lysates of LPS-stimulated BV2 microglial cells, but not in the cell culture supernatant of LPS-stimulated BV2 microglial cells, which was in consistent with previous findings.</p><p>FXR has been extensively studied in liver disease, such as innate hepatic inflammation, cholestatic liver disease and non-alcoholic fatty liver disease (NASH) [17]. Intriguingly, FXR agonist has been tested in clinic trial for treatment of liver disease, demonstrating that FXR has become an attractive target in human metabolic disease. In fact, FXR was not only expressed in liver, gut and kidney [40], but also expressed by immune cells, OPCs and mature oligodendrocytes, like microglia and astrocyte [41]. Previous studies have shown that FXR expression was significantly decreased after LPS stimulation in monocytes [42] and IFNγ stimulation in macrophages [43], which indicated a link between chronic autoimmune inflammation and FXR expression. However, whether or not FXR expression is changed after LPS stimulation in BV2 microglial cells remains unclear. In the present study, our results further confirmed that FXR plays an important role in regulating chronic inflammation.</p><p>FXR activation has been proved to confer protection in LPS-induced neuroinflammation in BV2 microglial cells [44, 45]. However, in order to further validate the effect of FXR in LPS-induced neuroinflammation in BV2 microglial cells, GS and FXR-siRNA, were chosen to block FXR in this study. The present study found that GS or FXR-siRNA treatment can significantly reverse the effect of GAA in inhibiting TNF-α and promoting BDNF expression in LPS-induced BV2 microglial cells. These results indicatedthat GAA inhibit LPS-induced neuroinflammation through activation of FXR.</p><p>In conclusion, this study demonstrates that GAA suppressed LPS-induced BV2 microglial cells proliferation and activation, promoted the conversion of M1 type microglia to M2 type, inhibited the LPS-induced pro-inflammatory cytokine release, and enhanced the neurotrophic factor BDNF expression. The precise mechanism of GAA in inhibiting LPS-induced neuroinflammation was mainly via activating FXR. Our results have strongly supported that GAA exert an anti-inflammation role in the context of neuroinflammation. Therefore, GAA may be a valuable anti-inflammatory and neuroprotective candidate for the treatment of brain diseases associated with inflammation.</p><!><p>Publisher's Note</p><p>Springer Nature remains neutral with regard to jurisdictional claims in published maps and institutional affiliations.</p><p>Yue Jia and Dandan Zhang contributed equally to this work.</p>

PubMed Open Access

Chemoproteomics-Enabled Discovery of Covalent RNF114-Based Degraders that Mimic Natural Product Function

Summary The translation of functionally active natural products into fully synthetic small molecule mimetics has remained an important process in medicinal chemistry. We recently discovered that the terpene natural product nimbolide can be utilized as a covalent recruiter of the E3 ubiquitin ligase RNF114 for use in targeted protein degradation (TPD) \xe2\x80\x93 a powerful therapeutic modality within modern day drug discovery. Using activity-based protein profiling-enabled covalent ligand screening approaches, we herein report the discovery of fully synthetic RNF114-based recruiter molecules that can also be exploited for PROTAC applications, and demonstrate their utility in degrading therapeutically relevant targets such as BRD4 and BCR-ABL in cells. The identification of simple and easily manipulated drug-like scaffolds that can mimic the function of a complex natural product is beneficial in further expanding the toolbox of E3 ligase recruiters, an area of great importance in drug discovery and chemical biology.

chemoproteomics-enabled_discovery_of_covalent_rnf114-based_degraders_that_mimic_natural_product_func

Introduction<!>Covalent ligand screening against RNF114<!>EN219-Based RNF114 Recruiter in TPD Applications<!>Discussion<!>Significance<!>Lead Contract:<!>Materials Availability<!>Data and Code Availability<!>Cell Lines<!>Chemicals<!>Cell-based degrader assays<!>Gel-Based ABPP<!>EN219-alkyne probe labeling in situ and pulldown studies<!>LC\xe2\x80\x93MS/MS analysis of pure RNF114 EN219 modification<!>IsoTOP-ABPP<!>Mass spectrometry analysis<!>TMT-based quantitative proteomic profiling<!>RNF114 ubiquitination assay<!>Western blotting<!>(E)-tert-butyl (4-(4-(3-(4-bromophenyl)-3-oxoprop-1-en-1-yl)phenoxy)butyl)carbamate (2):<!>tert-butyl (4-(4-(3-(4-bromophenyl)-1-(2-chloroacetyl)-4,5-dihydro-1H-pyrazol-5-yl)phenoxy)butyl)carbamate (3):<!>ML 2\xe2\x80\x9314:<!>(E)-tert-butyl 2-(4-(3-(4-bromophenyl)-3-oxoprop-1-en-1-yl)phenoxy)acetate (4):<!>tert-butyl 2-(4-(3-(4-bromophenyl)-1-(2-chloroacetyl)-4,5-dihydro-1H-pyrazol-5-yl)phenoxy)acetate (5):<!>ML 2\xe2\x80\x9322:<!>ML 2\xe2\x80\x9323:<!>EN219-alkyne:<!>(E)-tert-butyl (7-(4-(3-(4-bromophenyl)-3-oxoprop-1-en-1-yl)phenoxy)heptyl)carbamate (8):<!>tert-butyl (7-(4-(3-(4-bromophenyl)-1-(2-chloroacetyl)-4,5-dihydro-1H-pyrazol-5-yl)phenoxy)heptyl)carbamate (9):<!>ML 2\xe2\x80\x9331:<!>(E)-tert-butyl (2-(2-(2-(2-(4-(3-(4-bromophenyl)-3-oxoprop-1-en-1-yl)phenoxy)ethoxy)ethoxy)ethoxy)ethyl) carbamate (10):<!>tert-butyl (2-(2-(2-(2-(4-(3-(4-bromophenyl)-1-(2-chloroacetyl)-4,5-dihydro-1H-pyrazol-5 yl)phenoxy)ethoxy)ethoxy)ethoxy)ethyl)carbamate (11):<!>ML 2\xe2\x80\x9332:<!>tert-butyl (R)-3,5-bis(4-bromophenyl)-4,5-dihydro-1H-pyrazole-1-carboxylate ((R)-12):<!>(R)-1-(3,5-bis(4-bromophenyl)-4,5-dihydro-1H-pyrazol-1-yl)-2-chloroethan-1-one ((R)-EN219):<!>tert-butyl (S)-3,5-bis(4-bromophenyl)-4,5-dihydro-1H-pyrazole-1-carboxylate ((S)-12):<!>(S)-1-(3,5-bis(4-bromophenyl)-4,5-dihydro-1H-pyrazol-1-yl)-2-chloroethan-1-one ((S)-EN219):<!>Deschloro-EN219:<!>QUANTIFICATION AND STATISTICAL ANALYSIS<!>

<p>Natural products have remained a cornerstone of drug discovery research for decades resulting in numerous FDA-approved medicines and tools for biomedical research across a wide range of therapeutic areas (Newman and Cragg, 2016). Historically, a large percentage of natural product-inspired medicines have utilized the natural product as a starting point, wherein the tools of synthetic chemistry are used to fine tune compound properties (i.e a semisynthetic approach) to achieve the desired endpoint. Alternatively, the function of natural products can also serve as motivation for medicinal chemistry, where the design of fully synthetic small molecules are less constrained by availability, synthetic manipulation limitations, and physicochemical and metabolic liabilities. Modern-day examples of this approach include the translation of the alkaloid cytisine into the smoking cessation drug varenicline, the development of the blockbuster cardiovascular drug atorvastatin from the polyketide lovastatin, and the discovery of the proteasome inhibitor carfilzomib inspired by the natural polypeptide epoxomicin (Fig. 1). Such approaches, however, are greatly facilitated by an understanding of the binding mode of the small molecule to its protein target as well as the identification of key pharmacophores within the parent natural product (Fig. 1).</p><p>We recently discovered that the anti-cancer natural product nimbolide, a limonoid-type triterpene isolated from Azadirachta indica (neem), covalently reacts with an N-terminal cysteine (C8) within an intrinsically disordered region of the E3 ubiquitin ligase RNF114 in human breast cancer cells (Spradlin et al., 2019). Covalent targeting of RNF114 by nimbolide led to impaired ubiquitination of its endogenous substrate, the tumor suppressor p21, through a nimbolide-dependent competition of the RNF114-substrate binding interaction, thus providing a potential mechanism for the anti-cancer effects of this natural product. The realization that nimbolide targeted a substrate recognition domain within RNF114 suggested that nimbolide could potentially be used as a recruiter of RNF114 for targeted protein degradation (TPD) applications. Consistent with this premise, we showed that a proteolysis-targeting chimera (PROTAC) formed by linking nimbolide to a Bromodomain and extraterminal domain (BET) inhibitor JQ1 led to proteasome- and RNF114-dependent degradation of BRD4 in cells. While TPD has arisen as a powerful drug discovery paradigm for tackling the undruggable proteome by targeting intracellular proteins for proteasomal degradation rather than classic inhibition, the lack of a broad range of E3 ligase recruiters represents a known limitation in this arena (Bondeson and Crews, 2017; Chamberlain and Hamann, 2019; Lai and Crews, 2017). Indeed, while over 600 different E3 ligases have been annotated, only a small handful of these potential targets have succumbed to the PROTAC strategy. Discovering additional and more synthetically tractable E3 ligase recruiters is thus an important topic in expanding the scope of TPD and may help to address resistance mechanisms (Bond et al., 2020; Ottis et al., 2019), promote differing selectivity or kinetic profiles of degradation (Bondeson et al., 2018; Huang et al., 2018; Tong et al., 2020), and lead to cell-type or location-specific degradation.</p><p>While nimbolide has demonstrated success in TPD applications, its high molecular weight, modest chemical stability, and limited points for synthetic modification have prompted us to continually search for RNF114-based ligands for PROTAC development. Herein we realize the successful translation of the binding site of nimbolide into a simple, easily manipulated covalent small molecule. Because the N-terminal region of RNF114 that includes C8 is intrinsically disordered, structure-guided ligand discovery and optimization was not possible thus requiring unbiased approaches to ligand discovery. Activity-based protein profiling (ABPP)-enabled covalent ligand screening has been previously used to discover covalent recruiters against E3 ligases RNF4 and DCAF16 (Ward et al., 2019; Zhang et al., 2019) and has also facilitated ligand discovery against cysteines targeted by covalently-acting natural products (Grossman et al., 2017). Using this technique, we were able to discover fragments that could be used to replace nimbolide as the covalent E3 ligase recruitment module in fully functional degraders against several oncology targets.</p><!><p>To discover a fully synthetic covalent ligand that could access the same cysteine (C8) targeted by nimbolide on RNF114, we screened 318 cysteine-reactive chloroacetamide and acrylamide ligands in a gel-based competitive activity-based protein profiling (ABPP) assay, in which we competed ligands against binding of a rhodamine-functionalized cysteine-reactive iodoacetamide probe (IA-rhodamine) to pure RNF114 protein (Fig. 2a, 2b, Fig. S1, Table S1) (Bachovchin et al., 2010; Grossman et al., 2017; Ward et al., 2019). Through this screen, chloroacetamide EN219 emerged as the top hit, showing the greatest inhibition of IA-rhodamine binding to RNF114 (Fig. 2c). Dose-response studies showed that EN219 interacted with RNF114 with a 50 % inhibitory concentration (IC50) of 470 nM (Fig. 2d, 2e). EN219 also inhibited RNF114-mediated autoubiquitination and p21 ubiquitination in vitro, similarly to our previously observed findings with nimbolide (Fig. 2f). Given that EN219 possesses a stereogenic center, we also synthesized enantiomeric (R)- and (S)-EN219 derivatives (with >90% ee) (Mahé et al., 2010) and showed that these enantioenriched molecules showed similar potency for binding to RNF114 as the racemate we previously employed (Fig. S2a). We also synthesized a non-reactive acetamide analog of EN219, deschloro-EN219 (JNS 2–229), and showed that this compound no longer interacted with RNF114 (Fig. S2a). Mapping the sites of EN219 covalent modification on pure RNF114 protein by LC-MS/MS also confirmed C8 as the only site of modification (Fig. S2b). While EN219 represents an early lead compound within a drug discovery setting and recognize that through medicinal chemistry efforts, it could further be tuned, we nonetheless pursued further characterization of EN219 as a fully synthetic ligand for RNF114.</p><p>To characterize the proteome-wide interaction profile and the potential for engagement of RNF114 C8 by EN219 in cells, we next mapped the proteome-wide cysteine-reactivity of EN219 in situ in 231MFP breast cancer cells through several different approaches. First, we crudely assessed general cysteine-reactivity of EN219 compared to a previously identified promiscuous scout ligand YP 1–44 (Ward et al., 2019), and showed that EN219 did not broadly inhibit IA-rhodamine cysteine reactivity in 231MFP breast cancer cell lysate even at 250 μM, compared to YP 1–44 (Fig. S2c). We next assessed proteome-wide selectivity of EN219 by competitive isotopic tandem orthogonal proteolysis-ABPP (isoTOP-ABPP) using previously well-validated methods (Backus et al., 2016; Grossman et al., 2017; Spradlin et al., 2019; Wang et al., 2014; Weerapana et al., 2010). Cells were treated in situ with vehicle or EN219 and cells were subsequently lysed and labeled with an alkyne-functionalized iodoacetamide probe (IA-alkyne), followed by attachment of isotopically light or heavy, Tobacco Etch Virus (TEV)-cleavable biotin-azide enrichment handles through copper-catalyzed azide-alkyne cycloaddition (CuAAC). Probe-labeled proteins were enriched by avidin, digested with trypsin, and probe-modified peptides were eluted by TEV protease and analyzed by liquid chromatography-mass spectrometry (LC-MS/MS). Only 4 proteins showed >2-fold light versus heavy or control versus EN219-treated probe-modified peptide ratios with adjusted p-value <0.05 out of 686 probe-modified cysteines quantified. These proteins were RNF114 C8, TUBB1 C201, HSPD1 C442, and HIST1H3A C97, among which RNF114 was the only E3 ligase (Fig. 2g, Table S2).</p><p>To further investigate the selectivity of EN219 and to confirm EN219 engagement of RNF114 in cells, we also synthesized an alkyne-functionalized EN219 probe (EN219-alkyne) (Fig. 2h). Initially, we labeled 231MFP cells in situ with this EN219-alkyne probe and visualized labeled proteins by gel-based ABPP and investigated labeled protein bands that were competed by EN219 pre-treatment (Fig. S2d). This experiment showed about 8 potential EN219-alkyne-specific proteins that were competed by EN219 itself in cells, with several bands that were specific to EN219-alkyne (Fig. S2d). In some cases, we observed increased EN219-alkyne protein labeling of specific proteins in cells pre-treated with excess EN219, indicating increased labeling of EN219-alkyne specific off-targets when EN219-specific sites were occupied (Fig. S2d). To further elucidate these targets, we performed Tandem mass-tagging (TMT)-based quantitative proteomic profiling to identify proteins that were enriched by EN219-alkyne in situ labeling of 231MFP cells and were competed by EN219 in situ treatment (Fig. S2e, Table S3). While we did not identify RNF114 in this proteomics experiment, likely due to its relatively low abundance or other many EN219-alkyne specific targets that occluded detection of EN219-specific targets, we did identify 7 additional potential off-targets of EN219 that showed >3-fold competition with adjusted p-value <0.05 against EN219-alkyne labeling in cells (Fig. S2b; Table S3). We found 10 total targets that showed >3-fold enrichment with EN219-alkyne compared to DMSO-treated cells with adjusted p-value < 0.05 (Table S3). This number of potential EN219 off-targets would be consistent with the ~8 protein bands observed by gel-based ABPP in Fig. S2d. None of these 7–10 potential EN219 off-targets were E3 ligases. Collectively, our results suggested that EN219 was a moderately selective covalent ligand against C8 of RNF114. While we did not observe RNF114 by TMT-based proteomic experiments, RNF114 was clearly enriched from 231MFP cells treated with EN219-alkyne in situ compared to vehicle-treated controls after CuAAC-mediated appendage of biotin-azide and subsequent avidin-pulldown and RNF114 blotting (Fig. 2i).</p><!><p>We next tested whether EN219 could be exploited for TPD applications. We had previously demonstrated that nimbolide could be linked to the BET inhibitor ligand JQ1 to selectively degrade BRD4. Thus, we benchmarked our EN219 RNF114 recruiter by linking EN219 to JQ1 through three different linkers—ML 2–14, ML 2–31, and ML 2–32 with C4 alkyl, C7 alkyl, and polyethylene glycol (PEG4) linkers, respectively (Fig. 3a; Fig. S3a). Similar to effects seen with nimbolide-based degraders, ML 2–14 with the shortest linker showed the most robust degradation of BRD4 in 231MFP breast cancer cells compared to ML 2–31 and ML 2–32 with longer linkers, with 50 % degradation concentration (DC50) values of 36 and 14 nM for the long and short isoforms of BRD4, respectively (Fig. 3b, 3c; Fig. S3b, S3c) (Spradlin et al., 2019). EN219 treatment alone did not affect BRD4 levels, compared to ML 2–14 (Fig. S4a). This ML 2–14 mediated degradation was fully averted by pre-treating cells with the proteasome inhibitor bortezomib as well as the E1 activating enzyme inhibitor TAK-243 (Fig. 3d–3f, Fig. S4b).</p><p>To further validate that ML 2–14 degradation of BRD4 was driven through RNF114, we showed that nimbolide pre-treatment completely attenuated BRD4 degradation by ML 2–14 in 231MFP breast cancer cells (Fig. 3g–3h). BRD4 degradation in HAP1 cells was also significantly attenuated in RNF114 knockout cells compared to wild-type counterparts in two independent experiments (Fig. 3i–3j, Fig. S4c–S4d). We note that these purchased RNF114 knockout cells showed residual RNF114 protein expression, likely indicating a mixed population of cells (Fig. 3i–3j). Furthermore, we showed that this loss of BRD4 protein levels was not due to transcriptional downregulation of BRD4 expression since ML 2–14 treatment in 231MFP cells did not alter BRD4 mRNA levels (Fig. S4e). TMT-based quantitative proteomic profiling of ML 2–14-mediated protein expression changes showed selective degradation of BRD3 and BRD4, but not BRD2. We also observed stabilization of two known or putative RNF114 substrates, including the tumor suppressor CDKN1A (p21) and CTGF (Fig. 3k, Table S4) (Han et al., 2013; Spradlin et al., 2019).</p><p>To further demonstrate the utility of our fully synthetic RNF114 recruiter EN219 in degrading other more challenging protein targets, we synthesized a degrader linking EN219 to the BCR-ABL inhibitor dasatinib, ML 2–23 and ML 2–22, bearing a longer PEG3 linker and a shorter C3 alkyl linker, respectively (Fig. 4a, Fig. S4f). For this particular target, ML 2–23 with the longer linker showed more robust degradation of BCR-ABL in K562 leukemia cells compared to ML 2–22, consistent with previously observed structure-activity relationships of nimbolide-based BCR-ABL degraders (Fig. 4b–4c, Fig. S4g) (Tong et al., 2020). Consistent with ML 2–23 engaging BCR-ABL in cells, we observed inhibition of CRKL phosphorylation, a downstream substrate of BCR-ABL signaling (Fig. 4b–4c). Interestingly, EN219 showed preferential degradation of BCR-ABL compared to c-ABL, compared to several previous BCR-ABL degraders utilizing cereblon or VHL recruiters that showed opposite selectivity (Fig. 4b–4c) (Burslem et al., 2019; Lai et al., 2016). This preferential degradation was also observed with the equivalent nimbolide-based degrader (Tong et al., 2020). We further showed that this loss of BCR-ABL and c-ABL was not due to transcriptional downregulation of these genes, since BCR-ABL mRNA levels remained unchanged and we observed a likely compensatory increase in c-ABL mRNA levels (Fig. S4h). Furthermore, we do not believe that the loss of BCR-ABL shown here is because of general cytotoxicity since ML 2–23 treatment at the timepoints used to assess BCR-ABL degradation only exerted relatively modest cell viability impairments in K562 cells; EN219 compromises K562 viability only at higher concentrations (Fig. S4i). While rescue experiments with proteasome inhibitors proved challenging due to the cytotoxicity of proteasome inhibitors at the long timepoints required for robust BCR-ABL degradation, we observed significant rescue of early-stage ML 2–23-mediated BCR-ABL and c-ABL degradation with pre-treatment of K562 cells with the proteasome inhibitor MG132 at a shorter time point (Fig. 4d, 4e).</p><!><p>We previously discovered that the natural product nimbolide targets a predicted intrinsically disordered cysteine in RNF114 at a substrate recognition site (Spradlin et al., 2019). Here we report EN219 as a moderately selective covalent ligand that exploits this same binding modality. We show that EN219 can be linked to the BET inhibitor JQ1 to degrade BRD4 in a nimbolide-sensitive and proteasome- and RNF114-dependent manner. We further show that EN219 can be linked to the kinase inhibitor dasatinib to selectively degrade BCR-ABL over c-ABL in a proteasome-dependent manner in leukemia cells. These findings also further highlight that moderately selective cysteine-targeting ligands can still lead to robust protein degraders (Ward et al., 2019; Zhang et al., 2019). While EN219 represents a promising initial chemical scaffold for RNF114 recruitment, future medicinal chemistry efforts can further optimize the potency and metabolic stability of RNF114 recruiters; additionally, the facile synthesis of pyrazoline derivatives makes this a much more tractable synthetic problem when compared to the necessary chemistry to realize numerous nimbolide derivatives.</p><p>Overall, our results highlight the utility of chemoproteomic platforms for discovering chemical scaffolds that can be used as E3 ligase recruiters for TPD applications. Moreover, these studies further speak to the power of unbiased chemoproteomic approaches for identification of synthetic ligands mimicking the function of covalently acting natural products, particularly those in which structural binding information is unavailable (Nomura and Maimone, 2019).</p><!><p>We previously discovered that the natural product nimbolide could be used as a unique and covalent RNF114 E3 ubiquitin ligase recruiter for targeted protein degradation (TPD) applications. However, the adoption of this natural product in degraders has been hindered by the synthetic difficulty of using a complex natural product as a starting point for degrader synthesis. Here, using chemoproteomics-enabled covalent ligand screening approaches, we have discovered a more synthetically tractable covalent ligand EN219 that targets RNF114 and mimics nimbolide mode of action. We demonstrate that EN219, like nimbolide, can be incorporated into bifunctional degrader compounds to degrade neosubstrate proteins in an RNF114-dependent manner. Our study highlights the utility of covalent ligand screening in target-based screening endeavors to expand the toolbox of E3 ligase recruiters for TPD applications—an area of great important in drug discovery and chemical biology.</p><!><p>Further information and requests for resources and reagents should be directed to and will be fulfilled by the Lead Contact, Daniel K. Nomura, dnomura@berkeley.edu</p><!><p>Plasmids, compounds generated in this study will be made available upon reasonable request.</p><!><p>Data generated in this study will be made available upon reasonable request. No code was developed for this study.</p><!><p>The 231MFP cells were obtained from Prof. Benjamin Cravatt and were generated from explanted tumor xenografts of female MDA-MB-231 cells as previously described (Jessani et al., 2004). The 231MFP cells were cultured in L15 medium containing 10% FBS and maintained at 37 °C with 0% CO2. K562 chronic myeloid leukemia cell lines of female origin were purchased from ATCC. The K562 cells were cultured in Iscove's Modified Dulbecco's Medium containing 10% FBS and maintained at 37 °C with 5% CO2. HAP1 RNF114 wild-type and knockout cell lines of female origin were purchased from Horizon Discovery. The RNF114 knockout cell line was generated by CRISPR/Cas9 to contain a frameshift mutation in a coding exon of RNF114. HAP1 cells were grown in Iscove's Modifed Dulbecco's Medium in the presence of 10% FBS and penicillin/streptomycin.</p><!><p>Covalent ligands screened against RNF114 were purchased from Enamine LLC, including EN219. Structures of compounds screened can be found in Table S1. See Supplementary Information for synthetic methods and characterization for EN219 and degraders. Nimbolide was purchased from Cayman Chemicals (Item No. 19230).</p><!><p>For assaying degrader activity, cells were seeded (500,000 cells for 231MFP and HAP1 cells, 1,000,000 for K562 cells) into a 6 cm tissue culture dish (Corning) in 2.0–2.5 mL of media and allowed to adhere overnight for 231MFP and HAP1 cells. The following morning, media was replaced with complete media containing the desired concentration of compound diluted from a 1,000 × stock in DMSO. For rescue studies, the cells were pre-treated with proteasome inhibitors, nimbolide, or E1 activating enzyme inhibitors 30 min prior to the addition of DMSO or degrader compounds. At the specified time point, cells were washed once with PBS on ice, before addition of 120 μL of lysis buffer (20 mM Tris-HCl at pH 7.5, 150 mM NaCl, 1 mM Na2EDTA, 1 mM EGTA, 1% Triton, 2.5 mM sodium pyrophosphate, 1 mM beta-glycerophosphate, 1 mM Na3VO4, 1 μg/ml leupeptin) with Complete Protease Inhibitor Cocktail (Sigma) was added. The cells were incubated in lysis buffer for 5 min before scraping and transferring to microcentrifuge tubes. The lysates were then frozen at −80 °C or immediately processed for Western blotting. To prepare for Western blotting, the lysates were cleared with a 20,000g spin for 10 min and the resulting supernatant protein concentrations were quantified via BCA assay. The lysates were normalized by dilution with PBS to match the lowest concentration lysate, and the appropriate amount of 4 × Laemmli's reducing buffer was added.</p><!><p>Gel-Based ABPP methods were performed as previously described (Ward et al., 2019). Pure recombinant human RNF114 was purchased from Boston Biochem (K-220). RNF114 (0.25 μg) was diluted into 50 μL of PBS and 1 μL of either DMSO (vehicle) or covalently acting small molecule to achieve the desired concentration. After 30 min at room temperature, the samples were treated with 250 nM of tetramethylrhodamine-5-iodoacetamide dihydroiodide (IA-Rhodamine) (Setareh Biotech, 6222, prepared in anhydrous DMSO) for 1 h at room temperature. Incubations were quenched by diluting the incubation with 20 μL of 4 × reducing Laemmli SDS sample loading buffer (Alfa Aesar) and heated at 90 °C for 5 min. The samples were separated on precast 4–20% Criterion TGX gels (Bio-Rad Laboratories, Inc.). Fluorescent imaging was performed on a ChemiDoc MP (Bio-Rad Laboratories, Inc.). Inhibition of target labeling was assessed by densitometry using ImageJ.</p><!><p>Experiments were performed following an adaption of a previously described protocol (Thomas et al., 2017). The 231MFP cells were treated with either DMSO vehicle or 50 μM EN219-alkyne probe for 90 min. Cells were collected in PBS and lysed by sonication. For preparation of Western blotting samples, the lysate (1 mg of protein in 500 μl) was aliquoted per sample and then the following were added: 10 μl of 5 mM biotin picolylazide (900912 Sigma-Aldrich) and 50 μl of click reaction mix (three parts TBTA 5 mM TBTA in butanol:DMSO (4:1, v/v), one part 50 mM Cu(II)SO4 solution and one part 50 mM TCEP). Samples were incubated for 1 h at room temperature with gentle agitation. After CuAAC, proteomes were precipitated by centrifugation at 6,500 g and washed twice in ice-cold methanol (500 μl). The samples were spun in a prechilled (4 °C) centrifuge at 6,500 g for 4 min allowing for aspiration of excess methanol and subsequent reconstitution of protein pellet in 250 μl PBS containing 1.2% SDS by probe sonication. Then the proteome was denatured at 90 °C for 5 min, the insoluble components were precipitated by centrifugation at 6,500g and soluble proteome was diluted in 1.2 ml PBS (the final concentration of SDS in the sample was 0.2%) to a total volume of 1450 μl, with 50 μl reserved as input. Then 85 μl of prewashed 50% streptavidin agarose bead slurry was added to each sample and samples were incubated overnight at room temperature with gentle agitation. Supernatant was aspirated from each sample after spinning beads at 6,500 g for 2 min at room temperature. Beads were transferred to spin columns and washed three times with PBS. To elute, beads were boiled 5 min in 50 μl LDS sample buffer. Eluents were collected after centrifugation and analyzed by immunoblotting. The resulting samples were also analyzed as described below for TMT-based quantitative proteomic profiling.</p><!><p>Purified RNF114 (10 μg) in 50μl PBS was incubated 30 min at room temperature either with DMSO vehicle or EN219 (50 μM). The DMSO control was then treated with light iodoacetamide while the compound treated sample was incubated with heavy iodoacetamide for 1h each at room temperature (200 μM final concentration, Sigma-Aldrich, 721328). The samples were precipitated by addition of 12.5 μl of 100% (w/v) trichloroacetic acid and the treated and control groups were combined pairwise, before cooling to −80 °C for 1h. The combined sample was then spun for at max speed for 10 min at 4 °C, supernatant was carefully removed and the sample was washed with ice-cold 0.01 M HCl/90% acetone solution. The pellet was resuspended in 4M urea containing 0.1% Protease Max (Promega Corp. V2071) and diluted in 40 mM ammonium bicarbonate buffer. The samples were reduced with 10 mM TCEP at 60 °C for 30min. The sample was then diluted 50% with PBS before sequencing grade trypsin (1 μg per sample, Promega Corp, V5111) was added for an overnight incubation at 37 °C. The next day, the sample was centrifuged at 13,200 rpm for 30 min. The supernatant was transferred to a new tube and acidified to a final concentration of 5% formic acid and stored at −80 °C until mass spectrometry analysis.</p><!><p>IsoTOP-ABPP studies were done as previously reported (Spradlin et al., 2019). Cells were lysed by probe sonication in PBS and protein concentrations were measured by BCA assay35. For in situ experiments, cells were treated for 90 min with either DMSO vehicle or covalently acting small molecule (from 1,000× DMSO stock) before cell collection and lysis. Proteomes were subsequently labeled with N-5-Hexyn-1-yl-2-iodoacetamide (IA-alkyne) labeling (100 μM) for 1 h at room temperature. CuAAC was used by sequential addition of tris(2-carboxyethyl) phosphine (1 mM, Sigma), tris[(1-benzyl-1H-1,2,3-triazol-4-yl)methyl]amine (34 μM, Sigma), copper(II) sμlfate (1 mM, Sigma) and biotin-linker-azide—the linker functionalized with a tobacco etch virus (TEV) protease recognition sequence as well as an isotopically light or heavy valine for treatment of control or treated proteome, respectively. After CuAAC, proteomes were precipitated by centrifugation at 6,500g, washed in ice-cold methanol, combined in a 1:1 control:treated ratio, washed again, then denatured and resolubilized by heating in 1.2% SDS–PBS to 80 °C for 5 min. Insoluble components were precipitated by centrifugation at 6,500g and soluble proteome was diluted in 5 ml 0.2% SDS–PBS. Labeled proteins were bound to avidin-agarose beads (170 μl resuspended beads per sample, Thermo Pierce) while rotating overnight at 4 °C. Bead-linked proteins were enriched by washing three times each in PBS and water, then resuspended in 6 M urea and PBS (Sigma), and reduced in TCEP (1 mM, Sigma), alkylated with iodoacetamide (18 mM, Sigma), before being washed and resuspended in 2 M urea and trypsinized overnight with 0.5 μg μl−1 sequencing grade trypsin (Promega). Tryptic peptides were eluted off. Beads were washed three times each in PBS and water, washed in TEV buffer solution (water, TEV buffer, 100 μM dithiothreitol) and resuspended in buffer with Ac-TEV protease and incubated overnight. Peptides were diluted in water and acidified with formic acid (1.2 M, Spectrum) and prepared for analysis.</p><!><p>Total peptides from TEV protease digestion for isoTOP-ABPP or tryptic peptides for mapping EN219 site of modification on pure RNF114 were pressure loaded onto 250 mm tubing packed with Aqua C18 reverse phase resin (Phenomenex no. 04A-4299), which was previously equilibrated on an Agilent 600 series high-performance liquid chromatograph using the gradient from 100% buffer A to 100% buffer B over 10 min, followed by a 5 min wash with 100% buffer B and a 5 min wash with 100% buffer A. The samples were then attached using a MicroTee PEEK 360 μm fitting (Thermo Fisher Scientific no. p-888) to a 13 cm laser pμlled column packed with 10 cm Aqua C18 reverse-phase resin and 3 cm of strong-cation exchange resin for isoTOP-ABPP studies. Samples were analyzed using an Q Exactive Plus mass spectrometer (Thermo Fisher Scientific) using a five-step Mμltidimensional Protein Identification Technology (MudPIT) program, using 0, 25, 50, 80 and 100% salt bumps of 500 mM aqueous ammonium acetate and using a gradient of 5–55% buffer B in buffer A (buffer A: 95:5 water:acetonitrile, 0.1% formic acid; buffer B 80:20 acetonitrile:water, 0.1% formic acid). Data were collected in datadependent acquisition mode with dynamic exclusion enabled (60 s). One full mass spectrometry (MS1) scan (400–1,800 mass-to-charge ratio (m/z)) was followed by 15 MS2 scans of the nth most abundant ions. Heated capillary temperature was set to 200 °C and the nanospray voltage was set to 2.75 kV, as previously described.</p><p>Data were extracted in the form of MS1 and MS2 files using Raw Extractor v.1.9.9.2 (Scripps Research Institute) and searched against the Uniprot human database using ProLuCID search methodology in IP2 v.3 (Integrated Proteomics Applications, Inc.) (Xu et al., 2015a). Probe-modified cysteine residues were searched with a static modification for carboxyaminomethylation (+57.02146) and up to two differential modifications for methionine oxidation and either the light or heavy TEV tags (+464.28596 or +470.29977, respectively) or the mass of the EN219 adduct. Peptides were required to be fully tryptic peptides and to contain the TEV modification. ProLUCID data was filtered through DTASelect to achieve a peptide false-positive rate below 5%. Only those probe-modified peptides that were evident across two out of three biological replicates were interpreted for their isotopic light to heavy ratios. Light versus heavy isotopic probe-modified peptide ratios are calculated by taking the mean of the ratios of each replicate paired light vs. heavy precursor abundance for all peptide spectral matches (PSM) associated with a peptide. The paired abundances were also used to calculate a paired sample t-test p-value in an effort to estimate constancy within paired abundances and significance in change between treatment and control. P-values were corrected using the Benjamini/Hochberg method.</p><!><p>TMT proteomic profiling was performed as previously described (Spradlin et al., 2019).</p><!><p>Recombinant Myc-Flag-RNF114 proteins were purchased from Origene (Origene Technologies Inc., TP309752) or were purified as described previously(Spradlin et al., 2019). For in vitro auto-ubiquitination assay, 0.2 μg of RNF114 in 25 μl of TBS was pre-incubated with DMSO vehicle or the covalently acting compound for 30 min at room temperature. Subsequently, 0.1 μg of UBE1 (Boston Biochem. Inc., E-305), 0.1 μg UBE2D1 (Boston Bichem. Inc., e2–615), 5 μg of Flag-ubiquitin (Boston Bichem. Inc., u-120) in a total volume of 25 μl Tris buffer containing 2 mM ATP, 10 mM DTT and 10 mM MgCl2 were added to achieve a final volume of 50 μl. For substrate-protein ubiquitination assays, 0.1 μg of p21 (Origene) was added at this stage. The mixture was incubated at 37 °C with agitation for 1.5 h. Then, 20 μl of Laemmli's buffer was added to quench the reaction and proteins were analyzed by Western blot assay.</p><!><p>Antibodies to RNF114 (Millipore Sigma, HPA021184), c-ABL (Santa Crus, 24–11), p-CRKL (Tyr207, Cell Signaling Technology, 3181), GAPDH (Proteintech Group Inc., 60004–1-Ig), BRD4 (Abcam plc, Ab128874), and beta-actin (Proteintech Group Inc., 6609–1-Ig) were obtained from various commercial sources and dilutions were prepared per the recommended manufacturers' procedures. Proteins were resolved by SDS–PAGE and transferred to nitrocellulose membranes using the iBlot system (Invitrogen). Blots were blocked with 5% BSA in Tris-buffered saline containing Tween 20 (TBST) solution for 1 h at room temperature, washed in TBST and probed with primary antibody diluted in diluent, as recommended by the manufacturer, overnight at 4 °C. Following washes with TBST, the blots were incubated in the dark with secondary antibodies purchased from Ly-Cor and used at 1:10,000 dilution in 5% BSA in TBST at room temperature. Blots were visualized using an Odyssey Li-Cor scanner after additional washes. If additional primary antibody incubations were required, the membrane was stripped using ReBlot Plus Strong Antibody Stripping Solution (EMD Millipore, 2504), washed and blocked again before being re-incubated with primary antibody. Blots were quantified and normalized to loading controls using Image J.</p><!><p>To a solution of enone 1 (226 mg, 0.75 mmol) in anhydrous DMF (8 mL) was added 4-(Boc-amino)butyl bromide (376 mg, 1.5 mmol) and K2CO3 (414 mg, 3 mmol) and the reaction mixture was stirred at 60 °C for 5 h under an atmosphere of nitrogen. Upon cooling, the inorganic salts were filtered off, the solution was diluted with EtOAc, washed with water, and the volatiles removed in vacuo. The crude material was purified by silica gel chromatography (25% EtOAc/hexanes) to yield 310 mg (87%) of 2: 1H NMR (400 MHz, CDCl3): δ 7.96 – 7.88 (m, 2H), 7.83 (d, J = 15.6 Hz, 1H), 7.72 – 7.59 (m, 4H), 7.39 (d, J = 15.6 Hz, 1H), 7.01 – 6.91 (m, 2H), 4.66 (s, 1H), 4.07 (t, J = 6.2 Hz, 2H), 3.25 (q, J = 6.8 Hz, 2H), 1.94 – 1.82 (m, 2H), 1.79 – 1.69 (m, 2H), 1.49 (s, 9H).</p><!><p>To a solution of enone 2 (550 mg, 1.16 mmol) in EtOH was added hydrazine monohydrate (116 mg, 2.32 mmol) and the reaction mixture heated at 80 °C for 5 h under a nitrogen atmosphere. The reaction was cooled to room temperature, diluted with water, and extracted with DCM. The combined organic phase was dried over anhydrous magnesium sulfate and then concentrated in vacuo to ~ 5 mL. [NOTE: All of the workup procedures should be performed quickly (<1.5 h total time) and the rotovap bath kept cool as the crude product can easily undergo autooxidation]. To the concentrated DCM solution was immediately added chloroacetyl chloride (157 mg, 1.39 mmol) and triethylamine (152 mg, 1.5 mmol) at 0°C. The reaction mixture was stirred in an ice bath for 30 minutes, then warmed to room temperature and stirred overnight under N2. Upon completion of the reaction, the reaction mixture was diluted with EtOAc, washed with brine, and concentrated in vacuo. The crude was purified by silica gel column chromatography (40% EtOAc/hexanes) to yield 381 mg (58%) of 3: 1H NMR (400 MHz, CDCl3): δ 7.60 (q, J = 8.8 Hz, 4H), 7.19 – 7.10 (m, 2H), 6.87 – 6.77 (m, 2H), 5.55 (dd, J = 11.7, 4.7 Hz, 1H), 4.55 (s, 2H), 3.94 (t, J = 6.2 Hz, 2H), 3.74 (dd, J = 17.8, 11.7 Hz, 1H), 3.24 – 3.11 (m, 3H), 1.85 – 1.75 (m, 2H), 1.65 (p, J = 7.1 Hz, 2H), 1.44 (s, 9H).</p><p> </p><!><p>i. Chloroacetamide 3 (185 mg, 0.327 mmol) was dissolved in DCM (2.5 mL) and trifluoroacetic acid (2.5 mL) was added dropwise slowly over 20 minutes. After an additional 20 minutes of stirring, the solvent was removed in in vacuo. To remove residual TFA, the crude material was dissolved in 3mL of DCM and concentrated in vacuo and this process repeated two additional times. The deprotected amine was used directly in the next step without purification.</p><p>ii. The resulting TFA salt was dissolved in DCM (8 mL) and JQ1-acid (157 mg, 0.4 mmol), 1-[Bis(dimethylamino)methylene]-1H-1,2,3-triazolo[4,5-b]pyridinium 3-oxide hexafluorophosphate (HATU) (202 mg, 0.530 mmol), and N,N-Diisopropylethylamine (168 mg, 1.31 mmol) were added. The mixture was stirred overnight with monitoring by TLC (5% MeOH in DCM, 100% EtOAc). Upon completion, the reaction mixture was concentrated in vacuo and directly purified by silica gel flash column chromatography (1–5% MeOH/DCM). The eluted fractions were insufficiently pure and those containing product were combined, concentrated and purified again by flash silica chromatography (100–0% EtOAc/DCM followed by 0–5% MeOH/DCM) to afford 88.4 mg (32%) of ML 2–14: 1H NMR (400 MHz, CDCl3): δ 7.66 – 7.46 (m, 4H), 7.39 (d, J = 8.2 Hz, 2H), 7.29 (d, J = 8.3 Hz, 2H), 7.13 (d, J = 8.3 Hz, 2H), 6.86 (t, J = 5.9 Hz, 1H), 6.80 (d, J = 8.3 Hz, 2H), 5.54 (dt, J = 11.7, 4.1 Hz, 1H), 4.63 (t, J = 7.0 Hz, 1H), 4.54 (s, 2H), 3.90 (t, J = 6.1 Hz, 2H), 3.72 (dd, J = 17.8, 11.7 Hz, 1H), 3.54 (dd, J = 14.3, 7.5 Hz, 1H), 3.33 (ddt, J = 30.5, 13.3, 6.8 Hz, 3H), 3.17 (dd, J = 17.9, 4.7 Hz, 1H), 2.64 (s, 3H), 2.39 (s, 3H), 1.84 – 1.72 (m, 2H), 1.69 (d, J = 7.4 Hz, 2H), 1.65 (s, 3H); 13C NMR (151 MHz, CDCl3) δ 170.6, 164.1, 161.4, 158.8, 155.8, 154.5, 151.4, 150.0, 136.9, 136.7, 132.9, 132.3, 132.2, 131.1, 131.0, 130.6, 130.0, 129.9, 128.9, 128.4, 127.2, 125.3, 120.7, 115.1, 67.6, 60.4, 54.7, 42.3, 42.1, 39.7, 39.4, 31.7, 26.7, 26.4, 22.8, 14.5, 14.3, 13.2, 12.0, 11.6. HRMS (ESI): calcd. C40H39BrCl2N7O3S ([M+H]+): m/z 846.1390, found: 846.1399.</p><!><p>To a solution of 1 (153 mg, 0.50 mmol) in anhydrous DMF (8 mL) was added tert-Butyl bromoacetate (146 mg, 0.75 mmol) and K2CO3 (138 mg, 1 mmol) and the reaction mixture was heated to 60 °C for 5 hours under an atmosphere of nitrogen. Upon cooling, the inorganic salts were filtered off, the solution was diluted with EtOAc, washed with water, and concentrated in vacuo. The crude was purified by silica gel chromatography (25% EtOAc/hexanes) to yield 150 mg (72%) of 4: 1H NMR (400 MHz, CDCl3): δ 7.97 – 7.89 (m, 2H), 7.82 (d, J = 15.6 Hz, 1H), 7.73 – 7.60 (m, 4H), 7.41 (d, J = 15.6 Hz, 1H), 7.01 – 6.93 (m, 2H), 4.61 (s, 2H), 1.54 (s, 9H).</p><!><p>To a solution of 4 (417 mg, 1.00 mmol) in EtOH was added hydrazine monohydrate (250 mg, 5.00 mmol) and the reaction mixture was heated at 80 °C for 5 hours under an atmosphere of nitrogen. The reaction was cooled to room temperature, diluted with water, and extracted with DCM. The combined organic phase was dried over anhydrous magnesium sulfate and then concentrated in vacuo to ~ 5 mL [NOTE: All of the workup procedures should be performed quickly (<1.5 h total time) and the rotovap bath kept cool as the crude product can easily undergo autooxidation]. To the concentrated DCM solution was immediately added chloroacetyl chloride (169 mg, 1.50 mmol) and triethylamine (303 mg, 3.00 mmol) at 0 °C under an atmosphere of nitrogen. The reaction was maintained at this temperature for 30 minutes and then warmed to room temperature and stirred overnight under N2. The reaction mixture was diluted with EtOAc, washed with brine, and concentrated in vacuo. The crude material was purified by silica gel chromatography (35 – 40 % EtOAc/hexanes) to yield 140 mg (28%) of 5: 1H NMR (400 MHz, CDCl3): δ 7.69 – 7.58 (m, 4H), 7.24 – 7.16 (m, 2H), 6.93 – 6.84 (m, 2H), 5.60 (dd, J = 11.7, 4.6 Hz, 1H), 4.58 (d, J = 1.6 Hz, 2H), 4.52 (s, 2H), 3.78 (dd, J = 17.8, 11.8 Hz, 1H), 3.23 (dd, J = 17.8, 4.7 Hz, 1H), 1.52 (s, 9H).</p><!><p>i. Compound 5 (15.2 mg, 0.030 mmol) was dissolved in DCM (1.5 mL) and trifluoroacetic acid (0.5 mL) added slowly over the course of 20 minutes. After stirring for an additional 20 minutes, the solvent was removed in vacuo. To remove residual TFA, the crude material was dissolved in 3mL of DCM and concentrated in vacuo and this process repeated two additional times. The crude material was used directly in the next step.</p><p>ii. The aforementioned crude carboxylic acid was dissolved in 2 mL DCM and amine 6 (15.0 mg, 0.030 mmol), 1- [Bis(dimethylamino)methylene]-1H-1,2,3-triazolo[4,5-b]pyridinium 3-oxide hexafluorophosphate (HATU) (17.1 mg, 0.045 mmol), and N,N-Diisopropylethylamine (193.5 mg, 1.50 mmol) were added (Tong et al., 2020). The reaction was stirred overnight monitoring by TLC (20% Methanol in DCM). Upon completion, the reaction mixture was diluted with EtOAc, washed with brine, and concentrated in vacuo. The crude material was purified by silica gel flash column chromatography (1–5% MeOH/DCM). The eluted fractions were insufficiently pure and those containing product were combined, concentrated and purified again by flash silica chromatography (100–0% EtOAc/DCM followed by 10–35% MeOH/DCM) to afford 17.1 mg (60%) of ML 2–22: 1H NMR (600 MHz, DMSO-d6) δ 11.47 (s, 1H), 9.88 (s, 1H), 8.23 (s, 1H), 8.11 (d, J = 18.2 Hz, 1H), 7.78 – 7.71 (m, 2H), 7.71 – 7.64 (m, 2H), 7.40 (dd, J = 7.8, 1.7 Hz, 1H), 7.34 – 7.23 (m, 2H), 7.18 – 7.10 (m, 2H), 6.97 – 6.88 (m, 2H), 6.06 (s, 1H), 5.54 (dd, J = 11.7, 4.7 Hz, 1H), 4.78 – 4.64 (m, 2H), 4.46 (s, 2H), 3.86 (dd, J = 18.2, 11.8 Hz, 1H), 3.50 (s, 4H), 3.31 – 3.08 (m, 4H), 2.41 (s, 6H), 2.31 (s, 2H), 2.25 (s, 3H), 1.64 (s, 2H); 13C NMR (151 MHz, DMSO) δ 167.0, 164.7, 162.8, 162.1, 161.9, 159.5, 156.6, 156.5, 154.3, 140.4, 138.4, 133.8, 133.1, 132.0, 131.4, 129.5, 129.0, 128.6, 128.4, 127.7, 126.5, 125.3, 123.7, 114.4, 82.2, 66.7, 59.2, 55.1, 54.5, 51.9, 43.1, 42.0, 41.4, 36.5, 29.5, 28.0, 25.1, 23.0, 22.0, 17.9, 13.5, 10.5; HRMS (ESI): calcd. C42H44BrCl2N10O4S ([M+H]+): m/z 933.1823, found: 933.1814.</p><!><p>i. Compound 5 (16.2 mg, 0.032 mmol) was dissolved in DCM (1.5 mL) and trifluoroacetic acid (0.5 mL) added slowly over the course of 20 minutes. After stirring for an additional 20 minutes, the solvent was removed in vacuo, To remove residual TFA, the crude material was dissolved in 3mL of DCM and concentrated in vacuo and this process repeated two additional times. The crude material was used directly in the next step.</p><p>ii. The aforementioned crude carboxylic acid was dissolved in DCM (2 mL) and amine 7 (19.8 mg, 0.032 mmol), 1- [Bis(dimethylamino)methylene]-1H-1,2,3-triazolo[4,5-b]pyridinium 3-oxide hexafluorophosphate (HATU) (18.2 mg, 0.048 mmol), and N,N-Diisopropylethylamine (206.0 mg, 1.60 mmol) were added (Tong et al., 2020). The reaction mixture was stirred overnight with monitoring by TLC (20% Methanol in DCM). Upon completion, the reaction mixture was diluted with EtOAc, washed with brine, and concentrated in vacuo. The crude material was purified by silica gel flash column chromatography (1–5% MeOH/DCM). The eluted fractions were insufficiently pure and those containing product were combined, concentrated and purified again by flash silica chromatography (100–0% EtOAc/DCM followed by 10–35% MeOH/DCM) to afford 20.1 mg (59%) of ML 2–23: 1H NMR (400 MHz, Methanol-d4) δ 8.16 (s, 1H), 7.77 – 7.71 (m, 2H), 7.61 (dd, J = 8.8, 2.3 Hz, 2H), 7.40 – 7.34 (m, 1H), 7.31 – 7.15 (m, 4H), 7.00 – 6.90 (m, 2H), 6.00 (s, 1H), 5.57 (dd, J = 11.7, 4.7 Hz, 1H), 4.73 (d, J = 13.8 Hz, 1H), 4.60 (d, J = 13.9 Hz, 1H), 4.52 (s, 2H), 4.12 (q, J = 7.1 Hz, 1H), 3.88 (dd, J = 18.2, 11.7 Hz, 1H), 3.62 (tdd, J = 13.9, 10.2, 6.7 Hz, 16H), 3.48 (t, J = 5.5 Hz, 2H), 3.19 (dd, J = 18.1, 4.8 Hz, 1H), 2.66 (dd, J = 11.8, 5.7 Hz, 6H), 2.47 (s, 3H), 2.34 (s, 3H); 13C NMR (151 MHz, DMSO) δ 167.2, 164.6, 162.8, 161.9, 159.5, 156.6, 154.4, 140.4, 137.6, 133.3, 133.1, 132.0, 131.4, 131.3, 129.1, 128.6, 128.4, 127.3, 126.5, 114.4, 69.3, 69.2, 68.4, 66.7, 59.3, 53.6, 42.0, 41.4, 41.3, 37.8, 35.3, 30.3, 29.5, 27.4, 25.1, 23.0, 22.0, 17.9, 17.6, 15.6, 13.5, 11.9, 10.5; HRMS (ESI): calcd. C47H54BrCl2N10O4S ([M+H]+): m/z 1051.2453, found: 1051.2445</p><!><p>i. Compound 5 (17.6 mg, 0.035 mmol) was dissolved in DCM (1.5 mL) and trifluoroacetic acid (0.5 mL) added slowly over the course of 20 minutes. After stirring for an additional 20 minutes, the solvent was removed in vacuo. To remove residual TFA, the crude material was dissolved in 3mL of DCM and concentrated in vacuo and this process repeated two additional times. The crude material was used directly in the next step.</p><p>ii. The aforementioned crude carboxylic acid was dissolved in DCM (2 mL) and propargylamine (2.3 mg, 0.042 mmol), 1-[Bis(dimethylamino)methylene]-1H-1,2,3-triazolo[4,5-b]pyridinium 3-oxide hexafluorophosphate (HATU) (19.8 mg, 0.052 mmol), and N,N-Diisopropylethylamine (223.9 mg, 1.750 mmol) were added. The reaction mixture was stirred overnight with monitoring by TLC (60% EtOAc in hexane). Upon completion, the reaction mixture was diluted with DCM, washed with brine, and concentrated in vacuo. The crude material was purified by silica gel chromatography (25 – 70 % EtOAc/hexanes) to yield 14.6 mg (88%) of EN219-alkyne: 1H NMR (400 MHz, CDCl3): δ 7.68 – 7.60 (m, 4H), 7.27 – 7.21 (m, 2H), 6.95 – 6.89 (m, 2H), 6.81 (s, 1H), 5.60 (dd, J = 11.8, 4.8 Hz, 1H), 4.64 – 4.54 (m, 2H), 4.52 (s, 2H), 4.18 (dd, J = 5.6, 2.6 Hz, 2H), 3.81 (dd, J = 17.8, 11.8 Hz, 1H), 3.22 (dd, J = 17.9, 4.8 Hz, 1H), 2.31 (t, J = 2.6 Hz, 1H); 13C NMR (101 MHz, CDCl3): δ 167.7, 163.9, 156.7, 154.3, 134.5, 132.1, 129.6, 128.1, 127.4, 125.2, 115.1, 78.9, 71.8, 67.3, 60.3, 60.0, 42.0, 41.9, 29.6, 28.7, 21.0, 14.1; HRMS (ESI): calcd. C22H19BrCl1N3O3Na ([M+Na]+): m/z 510.0191, found: 510.0186.</p><p> </p><!><p>To a solution of 1 (153 mg, 0.50 mmol) in anhydrous DMF (8 mL) was added N-Boc-7-bromoheptan-1-amine (300 mg, 1.00 mmol) and K2CO3 (276 mg, 2.00 mmol). The reaction mixture was heated at 60 °C for 5 hours under an atmosphere of nitrogen. Upon cooling to room temperature, the inorganic salts were filtered off, the solution was diluted with EtOAc, washed with water, and the volatiles removed in vacuo. The crude material was purified by silica gel chromatography (25% EtOAc/hexanes) to yield 238 mg (92%) of 8: 1H NMR (400 MHz, CDCl3): δ 7.97 – 7.89 (m, 2H), 7.83 (d, J = 15.6 Hz, 1H), 7.73 – 7.59 (m, 4H), 7.39 (d, J = 15.6 Hz, 1H), 7.00 – 6.92 (m, 2H), 4.55 (s, 1H), 4.04 (t, J = 6.5 Hz, 2H), 3.16 (q, J = 6.8 Hz, 2H), 1.94 – 1.76 (m, 2H), 1.57 – 1.50 (m, 4H), 1.49 (s, 9H), 1.44 – 1.37 (m, 4H).</p><p> </p><!><p>To a solution of 8 (223 mg, 0.43 mmol) in EtOH was added hydrazine monohydrate (43 mg, 0.86 mmol) and the reaction mixture heated at 80 °C for 3 h under an atmosphere of nitrogen. The reaction mixture was cooled to room temperature, diluted with water and extracted with DCM. The combined organic phase was dried over anhydrous magnesium sulfate and then concentrated in vacuo to ~ 5 mL [NOTE: All of the workup procedures should be performed quickly (<1.5 h total time) and the rotovap bath kept cool as the crude product can easily undergo autooxidation]. To the concentrated solution was quickly added chloroacetyl chloride (58 mg, 0.52 mmol) and triethylamine (57 mg, 0.56 mmol) at 0 °C. The reaction mixture was stirred in an ice bath for 30 minutes, then warmed to room temperature and stirred overnight under N2. The reaction mixture was diluted with EtOAc, washed with brine, and concentrated in vacuo. The crude was purified by silica gel chromatography (25 – 40 % EtOAc/hexanes) to yield 104 mg (40%) of 9: 1H NMR (400 MHz, CDCl3): δ 7.69 – 7.53 (m, 4H), 7.23 – 7.12 (m, 2H), 6.89 – 6.82 (m, 2H), 5.57 (dd, J = 11.7, 4.7 Hz, 1H), 4.57 (s, 2H), 3.93 (t, J = 6.5 Hz, 2H), 3.76 (dd, J = 17.8, 11.7 Hz, 1H), 3.21 (dd, J = 17.9, 4.7 Hz, 1H), 3.13 (q, J = 6.8 Hz, 2H), 1.85 – 1.70 (m, 2H), 1.55 – 1.46 (m, 4H), 1.47 (s, 9H), 1.41 – 1.32 (m, 4H).</p><p> </p><!><p>i. Compound 9 (104 mg, 0.171 mmol) was dissolved in DCM (2.5 mL) and TFA (2.5 mL) was added slowly over 20 minutes followed by stirring for an additional 20 minutes at which point the reaction mixture was concentrated in vacuo. To remove residual TFA, the crude material was dissolved in 3mL of DCM and concentrated in vacuo and this process repeated two additional times. The crude material was used without further purification for amide coupling.</p><p>ii. The aforementioned TFA salt was dissolved in DCM (8 mL) and JQ1-acid (82 mg, 0.205 mmol), 1-[Bis(dimethylamino)methylene]-1H-1,2,3-triazolo[4,5-b]pyridinium 3-oxide hexafluorophosphate (HATU) (97.5 mg, 0.257 mmol), and N,N-Diisopropylethylamine (552 mg, 4.28 mmol) were added. The reaction mixture was stirred overnight with monitoring by TLC (5% MeOH in DCM, 100% EtOAc). Upon completion, the reaction mixture was directly concentrated in vacuo and the crude material purified by silica gel flash chromatography (1–5% MeOH/DCM). The eluted fractions were insufficiently pure and those containing product were combined, concentrated and purified again by flash silica chromatography (100–0% EtOAc/DCM followed by 0–5% MeOH/DCM) to afford 66 mg (43%) of ML 2–31: 1H NMR (400 MHz, CDCl3): δ 7.68 – 7.57 (m, 4H), 7.48 – 7.39 (m, 2H), 7.39 – 7.32 (m, 2H), 7.22 – 7.11 (m, 2H), 6.90 – 6.81 (m, 2H), 6.67 (t, J = 5.8 Hz, 1H), 5.58 (ddd, J = 11.8, 4.7, 2.1 Hz, 1H), 4.69 – 4.61 (m, 1H), 4.59 (d, J = 2.5 Hz, 2H), 3.93 (t, J = 6.4 Hz, 2H), 3.84 – 3.67 (m, 2H), 3.56 (dd, J = 14.3, 7.4 Hz, 1H), 3.42 – 3.30 (m, 2H), 3.28 – 3.16 (m, 3H), 2.70 (s, 3H), 2.47 – 2.35 (m, 3H), 1.82 – 1.71 (m, 2H), 1.71 – 1.67 (m, 3H), 1.56 (p, J = 7.0 Hz, 2H), 1.45 – 1.43 (m, 4H); 13C NMR (151 MHz, CDCl3) δ 170.4, 163.9, 158.9, 155.7, 154.6, 149.9, 136.8, 136.6, 132.6, 132.1, 132.1, 130.9, 130.9, 130.5, 129.8, 128.7, 128.2, 127.0, 125.2, 115.1, 115.1, 114.9, 67.9, 60.2, 54.9, 54.5, 43.0, 42.2, 42.1, 39.6, 39.4, 29.7, 29.4, 29.1, 29.0, 26.8, 25.9, 18.5, 17.2, 14.4, 13.1, 12.4, 11.8; HRMS (ESI): calcd. C43H45BrCl2N7O3S ([M+H]+): m/z 888.1865, found: 888.1858.</p><p> </p><!><p>To a solution of 1 (153 mg, 0.50 mmol) in anhydrous DMF (8 mL) was added tert-Butyl (2-(2-(2-(2-bromoethoxy)ethoxy)ethoxy)ethyl)carbamate (344 mg, 0.97 mmol) and K2CO3 (276 mg, 2.00 mmol), and the reaction mixture was stirred at 60 °C for 5 hours under an atmosphere of nitrogen. Upon cooling, the inorganic salts were filtered off, the solution was diluted with EtOAc, washed with water, and the volatiles removed in vacuo. The crude was purified by silica gel chromatography (50% EtOAc/hexanes) to yield 251 mg (88%) of 10: 1H NMR (400 MHz, CDCl3): δ 7.97 – 7.89 (m, 2H), 7.89 – 7.78 (m, 1H), 7.72 – 7.55 (m, 4H), 7.40 (d, J = 15.6 Hz, 1H), 7.04 – 6.95 (m, 2H), 5.07 (s, 1H), 4.23 (dd, J = 5.7, 4.0 Hz, 2H), 3.97 – 3.84 (m, 2H), 3.83 – 3.77 (m, 2H), 3.76 – 3.72 (m, 2H), 3.71 – 3.64 (m, 4H), 3.58 (t, J = 5.1 Hz, 2H), 3.35 (d, J = 6.3 Hz, 2H), 1.48 (s, 9H).</p><p> </p><!><p>To a solution of 10 (244 mg, 0.42 mmol) in EtOH was added hydrazine monohydrate (42 mg, 0.84 mmol) and the reaction solution was heated at 80 °C for 5 h under an atmosphere of nitrogen. Upon cooling to room temperature, the solution was diluted with water, extracted with DCM, and dried over anhydrous magnesium sulfate. The combined organic phase was concentrated in vacuo to ~ 5 mL [NOTE: All of the workup procedures should be performed quickly (<1.5 h total time) and the rotovap bath kept cool as the crude product can easily undergo autooxidation]. To the concentrated solution was quickly added chloroacetyl chloride (58 mg, 0.52 mmol) and triethylamine (57 mg, 0.56 mmol) at 0°C. The reaction mixture was stirred for 30 minutes in an ice bath and then warmed to room temperature and stirred overnight under N2. Upon completion of the reaction, the mixture was diluted with EtOAc, washed with brine, and concentrated in vacuo. The crude material was purified by silica gel chromatography (30 – 65 % EtOAc/hexanes) to yield 82 mg (30%) of 11: 1H NMR (400 MHz, CDCl3): δ 7.67 – 7.58 (m, 4H), 7.21 – 7.14 (m, 2H), 6.92 – 6.85 (m, 2H), 5.57 (dd, J = 11.7, 4.7 Hz, 1H), 5.11 (s, 1H), 4.57 (d, J = 1.1 Hz, 2H), 4.12 (dd, J = 5.6, 4.2 Hz, 2H), 3.88 – 3.83 (m, 2H), 3.75 – 3.73 (m, 2H), 3.72 – 3.69 (m, 2H), 3.67 – 3.62 (m, 4H), 3.55 (t, J = 5.1 Hz, 3H), 3.32 (q, J = 4.3, 3.4 Hz, 2H), 3.21 (dd, J = 17.8, 4.7 Hz, 1H), 1.46 (s, 9H).</p><p> </p><!><p>i. Compound 11 (81.7 mg, 0.120 mmol) was dissolved in DCM (2.5 mL) and trifluoroacetic acid (2.5 mL) was added slowly over 20 minutes followed by stirring for an additional 20 minutes at which point the reaction mixture was concentrated in vacuo. To remove residual TFA, the crude material was dissolved in 3mL of DCM and concentrated in vacuo and this process repeated two additional times. The crude material was used directly in the next step.</p><p>ii. The aforementioned crude material was dissolved in DCM (8 mL) and JQ1-acid (57.6 mg, 0.150 mmol), 1-[Bis(dimethylamino)methylene]-1H-1,2,3-triazolo[4,5-b]pyridinium 3-oxide hexafluorophosphate (HATU) (68.4 mg, 0.180 mmol), and N,N-Diisopropylethylamine (774 mg, 6.00 mmol) were added. The reaction was stirred overnight with monitoring by TLC (10% MeOH in DCM, 100% EtOAc). Upon completion, the reaction mixture was concentrated in vacuo and purified by silica gel flash chromatography (1–10% MeOH/DCM). The eluted fractions were insufficiently pure and those containing product were combined, concentrated and purified again by flash silica chromatography (100–0% EtOAc/DCM followed by 0–10% MeOH/DCM) to afford 46 mg (40%) of ML 2–32: 1H NMR (400 MHz, CDCl3): δ 7.66 – 7.57 (m, 4H), 7.45 – 7.41 (m, 2H), 7.34 (d, J = 8.5 Hz, 2H), 7.16 (dq, J = 7.9, 3.1 Hz, 2H), 6.98 (d, J = 5.1 Hz, 1H), 6.92 – 6.83 (m, 2H), 5.57 (ddd, J = 11.8, 7.3, 4.7 Hz, 1H), 4.68 (t, J = 6.9 Hz, 1H), 4.57 (s, 2H), 4.12 (dd, J = 5.7, 4.0 Hz, 2H), 3.90 – 3.84 (m, 2H), 3.78 – 3.66 (m, 10H), 3.65 – 3.58 (m, 2H), 3.54 – 3.50 (m, 2H), 3.44 – 3.37 (m, 1H), 3.20 (dd, J = 17.8, 4.7 Hz, 1H), 2.68 (d, J = 1.5 Hz, 3H), 2.42 (s, 3H), 1.69 (s, 3H); 13C NMR (151 MHz, CDCl3) δ 170.4, 163.8, 163.7, 158.4, 155.6, 154.3, 149.7, 145.2, 136.6, 136.6, 132.9, 132.1, 131.9, 131.7, 130.8, 130.6, 130.3, 130.2, 129.9, 129.8, 128.6, 128.1, 126.9, 125.0, 114.9, 70.7, 70.5, 70.5, 70.3, 69.7, 69.6, 67.4, 60.1, 54.3, 54.3, 42.1, 41.9, 39.3, 39.0, 29.6, 14.3, 13.0, 11.7; HRMS (ESI): calcd. C44H47BrCl2N7O6S ([M+H]+): m/z 950.1863, found: 950.1852.</p><p> </p><!><p>(E)-1,3-bis(4-bromophenyl)prop-2-en-1-one (109.8 mg, 0.3 mmol, 1.0 eq.), tert-butyl carbazate (43.6 mg, 0.33 mmol, 1.1 eq.), potassium phosphate (82.8 mg, 0.39 mmol, 1.3 eq.), and catalyst A (13.3 mg, 0.03 mmol, 10 mol %) were combined in THF (0.6 mL, 0.5 M) under nitrogen and stirred at 0 °C for 18h (Mahé et al., 2010). The mixture was diluted with EtOAc, filtered to remove salts, concentrated in vacuo and purified by silica gel chromatography (0–15% EtOH/Hex) to provide 94 mg (65%) of (R)-12: 1H NMR (400 MHz, Chloroform-d): δ 7.61 (d, J = 8.6 Hz, 2H), 7.55 – 7.43 (m, 4H), 7.14 – 7.08 (m, 2H), 5.31 (dd, J = 11.9, 5.1 Hz, 1H), 3.73 (dd, J = 17.5, 12.1 Hz, 1H), 3.08 (dd, J = 17.5, 5.5 Hz, 1H), 1.35 (s, 9H).</p><p> </p><!><p>(R)-12 (45 mg, 0.094 mmol, 1.0 eq) was dissolved in DCM (0.47 mL, 0.2 M), cooled to 0 °C, and 4.0 M HCl in dioxane (0.234 mL, 0.94 mmol, 10.0 eq.) was added under nitrogen. The mixture was stirred at room temperature until the starting material was consumed (1.5 h), then additional DCM was added (0.47 mL, reducing to 0.1 M) and the mixture cooled to 0 °C. Chloroacetyl chloride (22 μL, 0.28 mmol, 3.0 eq.) was added followed by TEA (0.17 mL, 1.22 mmol, 13.0 eq.), and the reaction allowed to warm to room temperature and stirred for 1h. The reaction mixture was concentrated in vacuo and the crude material purified by silica gel chromatography (15% EtOAc/Hex) to yield 36 mg (85%) of (R)-EN219 as a waxy solid. The enantiomeric excess of (R)-EN219 was determined by chiral HPLC (CHIRALCEL® OD-H column, i-PrOH/hexane (10:90), flow rate 1 mL/min) and found to be 91% ee. 1H NMR (400 MHz, Chloroform-d): δ 7.63 – 7.55 (m, 4H), 7.49 – 7.43 (m, 2H), 7.11 (d, J = 8.5 Hz, 2H), 5.55 (dd, J = 11.8, 4.9 Hz, 1H), 4.60 – 4.48 (m, 2H), 3.77 (dd, J = 17.9, 11.8 Hz, 1H), 3.16 (dd, J = 17.9, 4.9 Hz, 1H); 13C NMR (101 MHz, CDCl3): δ 164.2, 154.4, 139.8, 132.4, 132.3, 129.7, 128.4, 127.7, 125.6, 122.2, 60.3, 42.1, 42.1; HRMS (ESI): calcd. C17H14ON279Br235Cl ([M+H]+): m/z 454.9156, found: 454.9160.</p><p> </p><!><p>(E)-1,3-bis(4-bromophenyl)prop-2-en-1-one (116 mg, 0.32 mmol, 1.0 eq.), tert-butyl carbazate (46 mg, 0.35 mmol, 1.1 eq.), potassium phosphate (88 mg, 0.41 mmol, 1.3 eq.), and catalyst B (56 mg, 0.13 mmol, 0.4 eq) were combined in THF (2mL) under nitrogen and stirred at 0 °C for 18h (Mahé et al., 2010). The mixture was diluted with EtOAc, filtered to remove salts, concentrated and purified by silica gel chromatography (0–35% EtOAc/Hex) to provide 71 mg of (S)-12 intermediate as a yellow oil which was immediately carried forward into (S)-EN219.</p><p> </p><!><p>(S)-12 (71 mg, 0.148mmol, 1.0 eq) was dissolved in DCM (4mL), cooled to 0 °C, and 4.0 M HCl in dioxane (0.370 mL, 1.48mmol mmol, 10.0 eq.) was added under nitrogen. The mixture was stirred at room temperature until the starting material was consumed (2h), then the mixture was cooled again to 0 °C. Chloroacetyl chloride (36 μL, 0.44 mmol, 3.0 eq.) was added followed by TEA (0.268 mL, 1.92 mmol, 13.0 eq.), and the reaction allowed to warm to room temperature and stirred for 30 minutes. The reaction mixture was extracted in DCM and washed with water 3x. The reaction mixture was then concentrated in vacuo and the crude material purified by silica gel chromatography (15% EtOAc/Hex) to yield 35 mg (52%) of (S)-EN219 as a light yellow solid. The enantiomeric excess of (S)-EN219 was determined by chiral HPLC (CHIRALCEL® OD-H column, i-PrOH/hexane (10:90), flow rate 1 mL/min) and found to be 91% ee. 1H NMR (600 MHz, CDCl3) δ 7.63 – 7.56 (m, 4H), 7.48 – 7.44 (m, 2H), 7.14 – 7.09 (m, 2H), 5.55 (dd, J = 11.8, 4.9 Hz, 1H), 4.56 (d, J = 13.5 Hz, 1H), 4.51 (d, J = 13.4 Hz, 1H), 3.77 (dd, J = 17.8, 11.8 Hz, 1H), 3.16 (dd, J = 17.8, 4.9 Hz, 1H). 13C NMR (151 MHz, CDCl3) δ 164.2, 154.4, 139.8, 132.4, 132.3, 129.7, 128.4, 127.7, 125.6, 122.2, 60.3, 42.1, 42.1; HRMS (ESI): calcd. C17H13ON279Br235ClNa ([M+Na]+): m/z 476.8975, found: 476.8976</p><!><p>Hydrazine monohydrate (50% conc., 80.16 μL, 0.8 mmol, 2.0 eq.) was added to a suspension of (E)-1,3-bis(4-bromophenyl)prop-2-en-1-one (146.4 mg, 0.4 mmol, 1.0 eq.) in EtOH (1.33 mL, 0.3 M). The resulting reaction mixture was stirred at reflux temperature for 4 h before it was concentrated under reduced pressure. The crude pyrazoline was then dissolved in DCM (2mL, 0.2 M) and cooled to 0 °C. Triethylamine (167 μL, 1.2 mmol, 3.0 eq.) was added dropwise, followed by chloroacetyl chloride (42.6 μL, 0.6 mmol, 1.5 eq.). The resulting reaction mixture was stirred at ambient temperature for 1 hour before it was diluted with DCM. The organic phase was sequentially washed with satd. aq. NaHCO3 solution, brine, dried over Na2SO4, and concentrated in vacuo. Purification by column chromatography (15% EtOAc/Hex) to afforded 157 mg (96%) of deschloro-EN219: 1H NMR (400 MHz, CDCl3): δ 7.61 (q, J = 8.6 Hz, 4H), 7.49 (d, J = 8.4 Hz, 2H), 7.24 – 7.10 (m, 2H), 5.58 (dd, J = 12.3, 4.9 Hz, 1H), 3.77 (dd, J = 17.6, 12.0 Hz, 1H), 3.24 – 3.06 (m, 1H), 2.44 (s, 3H).; 13C NMR (101 MHz, CDCl3): δ 168.9, 152.7, 140.7, 132.10, 132.1, 130.2, 128.0, 127.4, 124.8, 121.7, 59.6, 42.0, 21.9; HRMS (ESI): calcd. C17H15ON279Br235 ([M+H]+): m/z 420.9546, found: 420.9550.</p><!><p>For quantification of Western blots, bands were quantified using Image J and normalized to protein loading controls. Statistical analysis was performed using a Student's unpaired two-tailed t-test.</p><p>For isoTOP-ABPP data analysis, data were extracted in the form of MS1 and MS2 files using Raw Extractor v.1.9.9.2 (Scripps Research Institute) and searched against the Uniprot human database using ProLuCID search methodology in IP2 v.3 (Integrated Proteomics Applications, Inc.)(Xu et al., 2015b). Cysteine residues were searched with a static modification for carboxyaminomethylation (+57.02146) and up to two differential modifications for methionine oxidation and either the light or heavy TEV tags (+464.28596 or +470.29977, respectively). Peptides were required to be fully tryptic peptides and to contain the TEV modification. ProLUCID data were filtered through DTASelect to achieve a peptide false-positive rate below 5%. Only those probe-modified peptides that were evident across two out of three biological replicates were interpreted for their isotopic light to heavy ratios. For those probe-modified peptides that showed ratios greater than two, we only interpreted those targets that were present across all three biological replicates, were statistically significant and showed good quality MS1 peak shapes across all biological replicates. Light versus heavy isotopic probe-modified peptide ratios are calculated by taking the mean of the ratios of each replicate paired light versus heavy precursor abundance for all peptide-spectral matches associated with a peptide. The paired abundances were also used to calculate a paired sample t-test P value in an effort to estimate constancy in paired abundances and significance in change between treatment and control. P values were corrected using the Benjamini–Hochberg method.</p><!><p>Table S1. Related to Figure 2. Structures of covalent ligands screened against RNF114</p><p>Table S2. Related to Figure 2. IsoTOP-ABPP analysis of EN219 in 231MFP breast cancer cells. IsoTOP-ABPP analysis of EN219 in 231MFP breast cancer cells. 231MFP cells were treated in situ with DMSO vehicle or EN219 (1 μM) for 90 min. Control and treated cell lysates were labeled with IA-alkyne (100 μM) for 1 h, after which isotopically light (control) or heavy (EN219-treated) biotin-azide bearing a TEV tag was appended by CuAAC. Proteomes were mixed in a 1:1 ratio, probe-labeled proteins were enriched with avidin and digested with trypsin, and probe-modified peptides were eluted by TEV protease and analyzed by LC-MS/MS.</p><p>Table S3. Related to Figure 2. TMT-based proteomic analysis of EN219-alkyne pulldown competed by EN219. 231MFP cells were treated with DMSO vehicle or EN219 (20 μM) 30 min prior to treating cells with DMSO or EN219-alkyne probe (2 μM) for 90 min. Resulting cell lysates were subjected to CuAAC with biotin-azide to append a biotin enrichment handle onto EN219-alkyne labeled proteins ex situ. EN219-alkyne labeled proteins were subsequently avidin-enriched, digested with trypsin, and resulting tryptic peptides from each treatment group were labeled with TMT reagents and combined and fractionated for LC-MS/MS analysis.</p><p>Table S4. Related to Figure 3. TMT-based proteomic analysis of ML 2–14-mediated protein level changes in 231MFP cells. TMT-based quantitative proteomic data showing ML 2–14-mediated protein level changes in 231MFP cells. 231MFP cells were treated with DMSO vehicle or ML 2–14 for 8 h.</p>

PubMed Author Manuscript

Cell cycle regulation by the intrinsically disordered proteins, p21 and p27

Synopsis Today, it is widely accepted that proteins that lack highly defined, globular structures, 3D, termed \xe2\x80\x9cintrinsically disordered proteins (IDPs)\xe2\x80\x9d, play key roles in myriad biological processes. Our understanding of how intrinsic disorder mediates biological function is, however, incomplete. Here, we review disorder-mediated cell cycle regulation by two intrinsically disordered proteins, p21 and p27. A structural adaptation mechanism involving a stretchable, dynamic linker helix allows p21 to promiscuously recognize the various Cdk/cyclin complexes that regulate cell division. Disorder within p27 mediates transmission of an N-terminal tyrosine phosphorylation signal to a C-terminal threonine phosphorylation, constituting a signaling conduit. These mechanisms are mediated by folding upon binding p21/p27\xe2\x80\x99s regulatory targets. However, residual disorder within the bound state contributes critically to these functional mechanisms. Our studies provide insights into how intrinsic protein disorder mediates regulatory processes and provide opportunities for designing drugs that target cancer-associated IDPs.

cell_cycle_regulation_by_the_intrinsically_disordered_proteins,_p21_and_p27

INTRODUCTION<!>Functional features of IDPs<!>Structural features of IDPs<!>p21 and p27 as complex regulators of cell cycle<!>Non-cell cycle functions of p21 and p27<!>p21 and p27 as dual regulators of oncogenesis<!>Primary structure of p21 and p27<!>Induced structure of p21 and p27 in the Cdk/cyclin-bound state<!>Nascent secondary structure of p21 and p27 in their free states<!>Sequential binding mechanism of p27<!>Binding promiscuity mechanism of p21<!>p21 and p27 as flexible signaling conduits<!>CONCLUDING REMARKS<!><!>The kinase-inhibitory domains (KIDs) of p21 and p27 regulate cell cycle progression .<!>p27-KID binds to Cdk2/Cyclin A through a two-step mechanism.<!>The stretchable linker helix of p21 mediates promiscuous binding to the various Cdk/cyclin complexes.<!>p27 serves as a signaling conduit that controls cell cycle progression.

<p>Over the past 15 years, there has been increasing awareness that globular 3D structure is not a prerequisite for protein function. While 3D protein structure is required for certain protein functions, another class of proteins lack stable secondary or ternary structure under physiological conditions and still perform important functions in cells. These proteins are termed intrinsically disordered proteins (IDPs). As the number of IDPs that play critical biological roles continues to grow, it has been proposed that the classic concept of structure-function relationships needs to be revised and extended to include IDPs 1,2. Despite broad recognition of the biological importance of IDPs, our understanding of disorder-mediated biological processes is incomplete. In this review, we provide a general introduction to IDPs and discuss how the intrinsic disorder of two prototypical IDPs, p21 and p27, mediate cell cycle regulation. These IDPs are of particular importance due to their critical roles in cell cycle regulation but also due to their potential as anticancer drug targets 3.</p><!><p>IDPs are involved in various cellular functions, notably in processes involving signaling and regulation 4. The highly dynamic features of IDPs enable diverse functions in cells. Although detailed information about how intrinsic disorder of an individual protein is associated with its specific function is limited, several examples provide insight into "disorder-function relationship".</p><p>Often, a single IDP performs multiple functions by associating with various targets to perform a specific function, which is mediated by its intrinsic disorder. For example, the cell cycle regulatory IDP, p21Cip1 (p21), binds several different cyclin-dependent kinase (Cdk)/cyclin complexes through structural adaptation to accommodate similar but topologically distinct binding sites 5.</p><p>Bioinformatics studies have shown that post-translational modifications (PTMs) predominantly occur within intrinsically disordered protein regions, which provides important regulatory mechanisms in cells 6,7. For example, the cell cycle regulatory IDP, p27Kip1 (p27), can switch the activity of Cdks from an inhibited state to an activated state through tyrosine phosphorylation 8.</p><p>Other biological processes mediated by intrinsic disorder are molecular movement and transport, as shown in, for example, motor proteins kinesin 9 and dynein 10, and FG-nups in the nuclear pore complex 11. Also, IDPs can serve as scaffolds for the assembly of multi-component, macromolecular complexes. As shown in several examples, axin 12, CBP 13 and BRCA1 14, multiple short interaction motifs in the disordered scaffold proteins 15 confer the ability to interact with and promote the co-assembly of many different partners.</p><!><p>IDPs have distinct primary structural features compared with those of globular proteins. Based on an analysis of the IDPs and intrinsically disordered regions deposited in the DisProt database 16, IDPs are primarily enriched in disorder-promoting residues (polar and/or charged amino acids: D, M, K, R, S, Q, P, and E) and are depleted in order-promoting residues (hydrophobic amino acids: C, W, Y, I, F, V, L, H, T, and N). Such distinct amino acid composition limits formation of highly defined protein structure by these proteins in isolation.</p><p>Although IDPs are highly disordered in solution, some retain partially populated secondary structure. For example, p21 and p27 in their free state do not exhibit completely random conformations but exhibit some degree of partially folded secondary structure, as evidenced by NMR spectroscopy, molecular dynamics (MD) simulations and circular dichroism (CD) spectroscopy 17–19.</p><p>Interestingly, in such many cases, nascent secondary structure in isolation is stabilized to form stable secondary structure upon binding to a target. This process is termed coupled folding and binding. In the case of p27, its partially populated secondary structure 18 is consistent with the secondary structure of its final bound state in p27/Cdk2/cyclin A crystal structure 20. This observation strongly suggests that the final bound state-like conformations, termed "intrinsically folded structural units" (IFSUs) when observed in isolation prior to binding 18, are populated and preferred for interaction with a binding partner. This is an example of conformational selection, one of two binding mechanism proposed to describe folding and binding by IDPs—conformational selection and induced folding—in which the target protein selects a conformation that is close to its bound state from the conformational ensemble of an IDP in its free state.</p><p>However, the partially populated secondary structure of an IDP does not always reflect the bound state. For the IDPs that adopt different conformations when bound to different partners, for example, the C-terminal region of p53 21, the secondary structure of the bound state and the intrinsic secondary structural propensity in the free state may differ. In this case, even if an IDP assumes a partially populated conformation consistent with binding to one of many targets, the generally flat energy landscape of IDPs enables other conformations to either be selected or induced to bind other targets. In the induced folding mechanism, an IDP binds to its target in a fully disordered state and is induced to fold upon binding to its target. For p27, only some regions exhibit IFSUs that suggests the conformational selection mechanism and other, highly disordered regions bind targets via the induced folding mechanism. Therefore, the p27 binding mechanism is likely a composite of the two extremes noted above and this complex behavior is likely exhibited in many other IDPs.</p><!><p>Progression through the mammalian cell cycle is driven by the sequential activation of Cdk/cyclin complexes (Fig. 1A). Cdk activity is negatively regulated by the Cip/Kip protein family. The Cip/Kip family members, including p21, p27, and p57, associate with the full repertoire of Cdk/cyclin complexes and inhibit their kinase activities at the G1/S and G2/M checkpoints (Fig. 1A). The binding promiscuity of the Cip/Kip proteins is a functional advantage afforded by their disordered features.</p><p>Although p21 and p27 were first identified as negative regulators of the cell division cycle 22,23, later studies have demonstrated that they can positively regulate Cdk/cyclin complexes 8,24–28. Contrary to the nuclear roles of p21 and p27 in inhibiting Cdk/cyclin complexes, their cytoplasmic roles are to mediate the assembly and nuclear import of Cdk4(6)/cyclin D complexes 24–2629. Furthermore, phosphorylation of p27 on Tyr 88 relieves the catalytic inhibition of Cdk2 8 and Cdk4 (6)27. As for p21, p21 phosphorylation on Tyr 77 also partially restores the kinase activity of Cdk2/cyclin A (M. Yoon, C. Park and R. Kriwacki, manuscript submitted), providing a mechanism to positively regulate Cdk2 activity. Moreover, it has been widely believed that the stoichiometry of p21 relative to Cdk/cyclin complexes determines whether a Cdk is active or not. At high levels of p21, Cdk2/cyclin A or Cdk4/cyclin D1 have been reported to be active even in the presence of p21 26,28.</p><p>Taken altogether, the mechanisms of regulation of Cdk activity by p21 and p27 are much more complicated than initially thought, depending on subcellular localization and tyrosine phosphorylation of p21 and p27, and the relative levels of p21 and Cdk/cyclin complexes.</p><!><p>In addition to their roles in cell cycle regulation, p21 and p27 are involved in other cellular functions, many of which are independent of Cdk/cyclin complexes. For example, p21 and p27 regulate transcription by directly binding to transcription factors, in addition to their ability to indirectly suppress transcription of cell cycle-related genes through inhibition of Cdk/cyclin complexes. p27 directly interacts with Neurogenin-2 (Ngn-2) via its N-terminal region and promotes transcription of its target genes, whereas p21 directly interacts with E2F 30, c-Myc31, and signal transducer and activator of transcription 3 (STAT3) 32 and inhibits their transcriptional activities.</p><p>Both p21 and p27 enhance cell migration by inhibiting the Rho/ROCK/LIMK/Cofilin signaling pathway 33 in the cytoplasm. These functions require phosphorylation on specific residues to enforce cytoplasmic localization (e.g., Thr157 34 or Thr198 35 on p27; Thr 145 on p21 36). Cytoplasmic p27 binds to RhoA and inhibits its activation by guanine-nucleotide exchange factors (GEFs), resulting in decreased actin stress fiber and focal-adhesion formation and subsequent increased cell migration 35. In contrast, cytoplasmic p21 binds to ROCK, a downstream target of RhoA, and inhibits its kinase activity, resulting in decreased actin stress fiber formation 37.</p><p>The unique C-terminal region of p21 directly associates with DNA polymerase δ processivity factor (PCNA) and inhibits its activity during DNA replication by blocking other DNA replication factors from binding PCNA 38. The C-terminal PCNA-binding region of p21 is overlapped with other interacting proteins, for example, the E7 oncoprotein of human papilloma virus 16 (HPV-16) 39 and c-Myc 31.</p><p>Interestingly, p21 has conflicting roles in apoptosis, having been demonstrated to both promote and inhibit programmed cell death. The pro-apoptotic activity of p21 has been attributed to both p53-dependent and p53-independent regulation of the apoptotic effector protein, Bax 40. On the contrary, cytoplasmic p21 protects cells against apoptosis both by directly binding to pro-apoptotic proteins, including procaspase 3, caspase 8, caspase 10, stress-activated protein kinases (SAPKs) and apoptosis signal-regulating kinase 1 (ASK1), and inhibiting their pro-apoptotic activities, and by suppressing the induction of pro-apoptotic genes by Myc and E2F1, thus inhibiting their transcription activity 41.</p><p>Other known p21 functions include regulation of DNA repair, senescence, cell differentiation, stem renewal and commitment 41–43. Furthermore, it is likely that other roles remain to be elucidated. It is remarkable that relatively small proteins such as p21 and p27 are entangled in such complex interaction networks. Their disordered features enable promiscuous protein-protein interactions within complex functional networks.</p><!><p>p21 was first identified as a mediator of p53 tumor suppressor function by serving as a downstream effector of p53-dependent inhibition of cell cycle progression through inhibition of Cdk/cyclin complexes at the G1/S and G2/M phase transitions and PCNA during S phase. However, such anti-proliferative activity of p21 was challenged by the later finding that it can also serve as an oncogene both by promoting cell growth and by inhibiting apoptosis, as discussed above.</p><p>Now it appears that the oncogenic function of p21 is associated with its cytoplasmic localization. As discussed above, when p21 is phosphorylated on Thr 145 by Akt, which lies downstream of the anti-apoptotic signaling protein PI3K, p21 is localized to the cytoplasm 36. Subsequently, cytoplasmic p21 inhibits the activity of pro-apoptotic proteins, mediates the assembly and activation of Cdk4(6)/cyclin D, and increases cell migration, invasion and metastasis by inhibiting the kinase activity of ROCK.</p><p>Interestingly, cytoplasmic mislocalization is also associated with the oncogenic activity of p27. Phosphorylation of p27 on Thr 157, Ser 10, or Thr 198 leads to cytoplasmic localization of p27, resulting in proliferation of cancer cells 34, promoting migration of hepatocellular carcinoma cells 44, or RhoA-dependent promotion of cell migration 35, respectively.</p><p>Furthermore, Grimmler, et al. 8, discovered that Tyr 88 of p27 was phosphorylated by the BCR-ABL fusion oncoprotein in chronic myelogenous leukemia (CML) cells and that this modification was associated with p27 ubiquitination and 26S proteasome-dependent degradation. As will be discussed in more detail below, phosphorylation of p21 on Tyr 77 may also lead to oncogenesis through a similar mechanism.</p><p>Taken altogether, although p21 and p27 exhibit tumor-suppressing activity by inhibiting Cdk/cyclin complexes in the nucleus, they also show oncogenic activity by altered subcellular localization and degradation.</p><!><p>The kinase-inhibitory function of p21 and p27 is mediated by a highly homologous N-terminal KID (Fig. 1B). The KID consists of 3 sub-domains, D1, linker helix (LH), and D2. Sub-domain D1, spanning 10 residues at the N-terminus of the KID, interacts with cyclin A through the RxL cyclin binding motif. The ~30 residues at the C-terminus of the KID, termed the D2 sub-domain, bind Cdk2 and inhibit its catalytic activity. The D1 and D2 domains are connected by sub-domain LH, which adopts a dynamic, helical conformation.</p><p>In contrast to conservation within the KIDs, the C-terminal domains of these proteins are more diverse, which is thought to mediate the unique aspects of their functions. The sequences of these proteins are rich in signaling and interaction motifs that encode diverse functionality within relatively short amino acid sequences. Different motifs are utilized differently in the three proteins, giving rise to their distinct biological roles.</p><!><p>Highly disordered, isolated p27 19,45, is induced to fold upon binding to its target Cdk2/cyclin A 20. In 1996, Russo, et al., reported the crystal structure of the p27-KID/Cdk2/cyclin A ternary complex (Fig. 2A), providing structural insight into the mechanism by which p27 inhibits Cdk2 activity. Sub-domain D1 binds in an extended conformation on the surface of cyclin A, whereas sub-domain D2 forms a β-hairpin and an intermolecular β-sheet on the surface of the N-terminal lobe of Cdk2. Sub-domain LH forms a 22 residue-long α-helix, spanning the ~40 Å gap between Cdk2 and cyclin A. These structural features reveal that p27 inhibits Cdk2/cyclin A in three different ways. First, the RxL motif within sub-domain D1 blocks the substrate binding site on cyclin A. Second, sub-domain D2 displaces the first β-strand of Cdk2, remodeling the catalytic cleft. Finally, the inhibitory 310-helix occupies the ATP binding pocket of Cdk2.</p><p>The NMR secondary 13Cα chemical shift values of p21-KID bound to Cdk2/cyclin A indicate that the secondary structure of bound p21-KID is generally consistent with that of p27-KID complexed with Cdk2/cyclin A in the crystal structure 5. This suggests that the sequence similarity between p21- and p27-KID leads to structural similarity in the Cdk/cyclin-bound state, and indicates that the two disordered proteins have similar intrinsic abilities to experience induced folding.</p><!><p>Despite being highly disordered, p21 and p27 do not completely lack secondary structure and exhibit partially folded secondary structure in their free states. CD spectra of p27 19 and p21 17 indicate a partially formed α-helix, within the LH sub-domain, as evidenced by the 13Cα secondary chemical shift values 45 (For p21, M. Yoon, C. Park and R. Kriwacki, unpublished results). In contrast to the nascent helical conformation of the LH sub-domain, only sub-domain D1 exhibits negative {1H}-15N heteronuclear NOE (HetNOE) values consistent with a high degree of flexibility 45. Interestingly, this nascent secondary structure of p27 45 is similar to that of p27-KID bound to Cdk2/cyclin A, suggesting that, at least for the motionally restricted regions of p27, the conformational selection mechanism provides a thermodynamic advantage by reducing the entropic penalty associated with folding upon binding. Furthermore, the 13Cα and 13C' secondary chemical shift analyses of free p21-KID (M. Yoon, C. Park, and R. Kriwacki, unpublished results) demonstrate that the nascent secondary structure of p21-KID is generally consistent with that of its Cdk2/cyclin A-bound structure 5, again suggesting a significant role for the conformational selection mechanism in this coupled folding and binding process. Taken together, p21 and p27 may derive thermodynamic advantage by assuming highly populated conformations, even before associating with their targets, that are similar to their conformations when bound to Cdk2/cyclin A.</p><!><p>Lacy, et al., provided insights into the mechanism associated with the binding of p27-KID to the Cdk2/cyclin A complex in terms of thermodynamics and kinetics, using isothermal titration calorimetry (ITC) 45 and surface plasmon resonance (SPR) 46. In the sequential binding mechanism identified in these studies (Fig. 2B), the highly dynamic D1 sub-domain rapidly binds to cyclin A though its RxL cyclin binding motif, and then LH sub-domain folds into an α-helix, finally followed by folding of three IFSUs, a β-hairpin, an intermolecular β-strand and a 310-helix within the D2 sub-domain. Rapid association of D1 sub-domain with cyclin A may be due to the large capture radius of disordered p27 through the "fly casting" mechanism 47. This fast initial step is followed by the slow association of less flexible, D2 sub-domain with the surface of Cdk2. Furthermore, this sequential binding mechanism, which combines induced fit and conformational selection mechanism, is also observed for binding to the Cdk4/cyclin D1 complex (L. Ou, B. Waddell, R. Kriwacki, unpublished results).</p><!><p>Recently, Wang, et al., reported the structural mechanism by which p21 promiscuously binds to various Cdk/cyclin complexes 5 (Fig. 3). Unlike the D1 and D2 sub-domains, the LH sub-domain of p21 exhibits sequence variation among Cip/Kip family members and highly dynamic features, as evidenced by analysis of HetNOE values. Inspection of B-factors for this region of p27 in the complex crystal structure revealed increased disorder, in contrast to a high degree of order observed for sub-domains D1 and D2 20. Furthermore, in this crystal structure, the LH sub-domain of p27 exhibited a conformation corresponding to a stretched α-helix, elongated over its 22 residue length by about 4 Å. Finally, Wang, et al. 5, deduced that helix stretching was responsible for the dynamic features of p21 when bound to Cdk2/cyclin A, and that this "structural adaptation" by both p21 and p27 was associated with their ability to promiscuously bind to the entire Cdk/cyclin repertoire.</p><p>Experiments with p21 variants with different length LH sub-domains further confirmed this hypothesis. Also, the crystal structures of Cdk4/cyclin D1 48 and Cdk4/cyclin D3 49 revealed that the LH sub-domain would have to contract and pivot in order for the D1 and D2 sub-domains to bind conserved surfaces on the D-type cyclins and Cdk4, respectively (Figure 3). The lack of tertiary contacts between the different sub-domains of the p21 and p27 KIDs creates a flat energy landscape that enables structural adaptation while folding upon binding to Cdk/cyclin complexes.</p><!><p>It has been well known that the Ser/Thr phosphorylation of p21 and p27 regulate their stability, activity, and subcellular localization. For example, phosphorylation of p27 Thr 187 50 or p21 Ser 130 51,52 by the Cdk2/cyclin E (A) complexes creates phosphodegrons, which are required for recognition by the SCFSkp2 E3 ubiquitin ligase complex for ubiquitination and subsequent proteasomal degradation of p21 and p27. Interestingly, Cdk2 phosphorylates p21 or p27 when bound to these so-called inhibitors. In other words, p21 or p27 serves as both an inhibitor and a substrate of Cdk2. This paradox was resolved by a previous study that demonstrated a molecular mechanism by which the activity and stability of p27 is regulated 8. In the two step phosphorylation mechanism (Fig. 4), p27 can be phosphorylated on Tyr 88 within the inhibitory 310-helix by Abl and Lyn when bound to Cdk2/cyclin A, leading to the restoration of Cdk2 activity by ejecting the 310-inhibitory helix from the active site of Cdk2. Subsequently, this reactivated Cdk2 phosphorylates p27 on Thr187 in the highly flexible C-terminal tail of p27 53 within the same ternary complex via a pseudo-unimolecular mechanism, resulting in the ubiquitination by SCFSkp2 E3 ligase and subsequent degradation.</p><p>Furthermore, we reasoned that, due to the functional and sequence similarities between p21 and p27, the same two-step phosphorylation mechanism may be applied also for p21. Indeed, there are critical Tyr and Ser/Thr residues that can participate in this signaling cascade in p21, Tyr 77 and Ser 130, equivalent to Tyr 88 and Thr 187 in p27, respectively. Also, the intrinsic disorder within the CTD of p21 54 may allow p21 to conduct the Tyr 77 signal within the KID to Ser 130 signal within the CTD, as observed for p27. Furthermore, as discussed above, phosphorylation of Ser 130 has been reported to be a signal for p21 ubiquitination and 26S proteosomal degradation 51,52. As expected, our biochemical and NMR studies show that p21 can be phosphorylated on Tyr77 and that the Tyr77-phosphorylated p21 couples the signal for Ser 130 phosphorylation by partially restoring the Cdk2 activity (M. Yoon, C. Park and R. Kriwacki, manuscript submitted).</p><p>This molecular mechanism illustrates how the intrinsic disorder of p21 and p27 allows them to function as flexible signaling conduits, enabling transmission of an intramolecular N-terminal Tyr phosphorylation signal to a C-terminal Ser/Thr phsophorylation site, followed by inter-molecular ubiquitination signaling.</p><!><p>We have reviewed how the intrinsic disorder of p21 and p27 mediates cell cycle regulation and have highlighted functional advantages associated with disorder. p21 promiscuously binds to various Cdk/cyclin complexes through a helix stretching, structural adaptation mechanism. Also, p27 serves as a flexible signaling conduit by transmitting tyrosine phosphorylation signals to a threonine degradation signal through the protein, with this mechanism facilitated by flexibility and disorder. Although these studies provide molecular insights into disorder-function relationship for p21 and p27 in cell cycle regulation, it remains to be elucidated how disorder mediates other cellular functions, for example, transcriptional regulation and control of apoptosis. Also, it would be interesting to address how p21 or p27 utilizes the same (as for cell cycle regulation) or different disordered regions to mediate their various functions and to compare these new functional mechanisms with those known to mediate cell cycle regulation.</p><p>More complex than initially found, p21 and p27 serve not only as negative regulators of cell cycle but also as assembly factors for and activators of several Cdk/cyclin complexes. This pro-proliferation effect for p21, due to the positive regulation of Cdk/cyclin complexes together with anti-apoptotic activity, makes p21 as an attractive therapeutic target in many cancers 55. We believe that the aforementioned studies involving disorder-mediated cell cycle regulation provide a structural basis for the design of anti-cancer drugs that target p21 or p27. Additionally, it may be possible to utilize the concepts associated with disorder-function relationships that have emerged from studies of p21 and p27 to develop molecules that target Cdk/cyclin complexes; such molecules hold promise against cancer as the Cdks are valid anti-cancer drug targets 56. The highly specific, potent peptide-based drug inhibiting Cdk activity may be designed using the intrinsically disordered features of p21 and p27. We believe that this "disorder-based drug design" will open a new door for designing and developing anti-cancer drugs, as more intrinsic disorder is found in cancer-associated proteins 4.</p><!><p>circular dichroism</p><p>cyclin-dependent kinase</p><p>C-terminal domain</p><p>guanine-nucleotide exchange factors</p><p>heteronuclear Overhauser effect</p><p>human papilloma virus 16</p><p>intrinsically disordered proteins</p><p>intrinsically folded structural units</p><p>isothermal titration calorimetry</p><p>kinase-inhibitory domain</p><p>linker helix</p><p>nuclear localization signal</p><p>nuclear magnetic resonance</p><p>non-receptor tyrosine kinase</p><p>DNA polymerase δ processivity factor</p><!><p>(A) p21 and p27 inhibit the activity of Cdk1(2)/cyclin E(A) and Cdk1/cyclin B(A) that are required for progression from G1 to S phase and for mitosis, respectively. During G1 phase, p21 and p27 mediate assembly and activation of Cdk4(6)/cyclin D in the cytoplasm, as well as inhibiting their activity in the nucleus. (B) Sequence alignment of the KIDs of p21 and p27. Residues identical in two sequences are shaded pink and similar residues are shaded green. The sub-domains are schematically represented as D1 (cyan), linker helix (LH) (light yellow), and D2 (blue). The secondary structure of p27-KID bound to Cdk2/cyclin A 20 is illustrated. (C) Domain organization in p21 and p27. The KID, PCNA-binding domain (PCNA), secondary cyclin–binding domain (Cy2), and QT domain (QT) are illustrated lime, pink, green, and purple, respectively.</p><!><p>(A) Crystal structure of p27-KID in complex with Cdk2 (orange) and cyclin A (magenta); PDB accession number 1JSU. Sub-domain D1 (cyan) binds and inhibits substrate binding to cyclin A, while sub-domain D2 (blue) binds Cdk2, inserting a 310 helix into the ATP binding pocket of Cdk2. Sub-domain LH (light yellow) connects the other two sub-domains. (B) Sequential mechanism of p27-KID association with Cdk2/Cyclin A; first, the highly disordered and dynamic RxL motif within sub-domain D1 rapidly associates with cyclin A; next, sub-domain LH fully folds, positioning sub-domain D2 near Cdk2; finally, sub-domain extensively folds upon binding and remodeling Cdk2 (adapted and modified from Lacy, et al. 45). Cdk2 and Cyclin A are colored as in panel A.</p><!><p>(A) Comparison of a standard α-helix, 22 residues in length, with the LH sub-domain of p21 that structurally adapts through helix stretching to accommodate binding to Cdk2/cyclin A. (B) The binding of p21 to different Cdk/cyclin complexes requires structural adaptation by the LH sub-domain through helix stretching/contraction and pivoting; the distance between conserved features of the p21 binding sites within the Cdk and cyclin subunits are illustrated for Cdk2/cyclin A (orange/magenta) and Cdk4/cyclin D3 (blue/pink). The locations of and distance between Val 30 and Leu 255, and Leu 34 and Leu 101, respectively, within these two complexes are illustrated. (adapted and modified from Wang, et al. 5).</p><!><p>Cell cycle is arrested by p27 inhibition of Cdk/cyclin complexes. The inhibitory effect is released through phosphorylation of p27 at Tyr 88 by NRTKs (Step 1). The now liberated kinase active site is able to phosphorylate p27 at Thr 187 (Step 2), which in turn recruits the SCFSkp2 ubiquitin ligase complex (Step 3). Poly-ubiquitinated p27 is finally degraded by the 26S proteasome (Step 4). Note that the ubiquitination pattern shown in the figure is solely for illustrative purposes. Figure adapted and modified from Galea, et al.57.</p>

PubMed Author Manuscript

Progress in Electrocatalytic Hydrogen Evolution Based on Monolayer Molybdenum Disulfide

Energy and environmental issues raise higher demands on the development of a sustainable energy system, and the electrocatalytic hydrogen evolution is one of the most important ways to realize this goal. Two-dimensional (2D) materials represented by molybdenum disulfide (MoS2) have been widely investigated as an efficient electrocatalyst for the hydrogen evolution. However, there are still some shortcomings to restrict the efficiency of MoS2 electrocatalyst, such as the limited numbers of active sites, lower intrinsic catalytic activity and poor interlayer conductivity. In this review, the application of monolayer MoS2 and its composites with 0D, 1D, and 2D nanomaterials in the electrocatalytic hydrogen evolution were discussed. On the basis of optimizing the composition and structure, the numbers of active sites, intrinsic catalytic activity, and interlayer conductivity could be significantly enhanced. In the future, the study would focus on the structure, active site, and interface characteristics, as well as the structure-activity relationship and synergetic effect. Then, the enhanced electrocatalytic activity of monolayer MoS2 can be achieved at the macro, nano and atomic levels, respectively. This review provides a new idea for the structural design of two-dimensional electrocatalytic materials. Meanwhile, it is of great significance to promote the study of the structure-activity relationship and mechanism in catalytic reactions.

progress_in_electrocatalytic_hydrogen_evolution_based_on_monolayer_molybdenum_disulfide

Introduction<!><!>Strategies for Electrocatalytic Hydrogen Evolution<!>Monolayer MoS2<!>Composition With 0D Materials<!><!>Composition With 1D Materials<!><!>Composition With 1D Materials<!>Composition With 2D Materials<!><!>Composition With 2D Materials<!>Outlook<!>Author Contributions<!>Conflict of Interest Statement<!>

<p>The continuous growth of the population and the development of the industrialization process have accelerated the consumption of fossil energy, and brought serious environmental problems. Therefore, the development of sustainable energy system is one of the most important challenges today (Wang and Mi, 2017; Chi and Yu, 2018). At present, a promising method is to produce renewable energy through the electrochemically catalytic reaction, which converts the common materials, such as water, carbon dioxide, and nitrogen, into the high-energy carriers (hydrogen, oxygen, hydrocarbons, ammonia, etc.). After years of research and practice, many important advances have been made in electrochemical energy conversion (Gu et al., 2018; Mao et al., 2018; Xiong J. et al., 2018). Among them, hydrogen energy is considered as the most powerful candidate to alternate fossil energy due to its clean, renewable, and environmentally friendly properties and high energy density (Lin et al., 2017; Zhang S. et al., 2017). Among various methods of hydrogen energy production, electrocatalytic water splitting has attracted tremendous attention because of its advantages of low cost, non-pollution and high efficiency (Wang et al., 2016). Moreover, the electrocatalytic cathode in this method is the key to determine the efficiency of water decomposition. So far, the Pt cathode possessing the near zero overpotential is considered to be the most effective catalytic cathode. However, it is difficult to be practically applied or industrialized due to its high cost and scarce resource (Eftekhari, 2017; Hou et al., 2017). Therefore, seeking for low-cost, abundant, high efficient, and environmentally friendly catalytic cathode materials has become a research hotspot. In the view of this point, many materials have been extensively explored, such as carbides, nitrides, sulfides, selenides, phosphides, and Mo-based non-noble metal electrocatalysts (Xie et al., 2014; Pu et al., 2016a,b,c, 2017, 2018; Voiry et al., 2016; Wei et al., 2016; Xie and Xie, 2016; Kou et al., 2017, 2018a,b; Jin et al., 2018). Among these materials, molybdenum disulfide (MoS2) has attracted much more attention due to its low cost, high catalytic activity, high stability, large in-plane carrier mobility and good mechanical properties (Tan et al., 2017; Li et al., 2018; Wang et al., 2018a). Studies have shown that monolayer MoS2 has higher electrocatalytic activity for hydrogen evolution. However, there are still some shortcomings, such as the limited numbers of active sites, lower intrinsic catalytic activity and poor interlayer conductivity. In order to further improve the electrocatalytic activity of monolayer MoS2, researchers usually composite them with other materials. In this paper, the composite of monolayer MoS2 with 0D, 1D, and 2D materials and its application in electrocatalytic hydrogen evolution are reviewed in order to provide guidance for related research. At present, there are two kinds of methods for preparing monolayer MoS2. The first method is top-down approach, including mechanical stripping (Li et al., 2012), ion intercalation (Nurdiwijayanto et al., 2018) and liquid phase stripping (Zhao et al., 2016), and the second one is bottom-up approach, including chemical vapor deposition(CVD) (Liu et al., 2018) and wet chemical stripping (Zeng et al., 2017). The development strategy of sustainable energy pattern and catalyst based on the electrocatalysis are shown in Figure 1.</p><!><p>Schematic diagram of sustainable energy pattern and catalyst development strategies based on the electrocatalysis.</p><!><p>As the electrocatalyst plays an important role in improving conversion efficiency in energy conversion process, the research of electrocatalyst is a crucial part in these conversion technologies. Up to now, the electrocatalysts suffer from the lack of types and low efficiency. What's more, the high expense leads them difficult to be practically used on a large scale. Many efforts have been made to solve these problems. For example, in order to improve the electrocatalytic activity, three strategies are usually proposed: one is to increase the number of active sites (from the view of the "quantity" aspect); the other is to increase the intrinsic activity of active sites (which belongs to the "quality" aspect); the third is to improve the conductivity of electrocatalysts by forming composites. These strategies are not mutually exclusive, but can be mutually complementary to improve the activity of catalysts simultaneously (Seh et al., 2017; Tang C. et al., 2018).</p><p>Two-dimensional materials such as MoS2 have been extensively studied in the electrocatalytic hydrogen evolution due to their promising potential application prospect. However, there is still a big gap compared with Pt catalyst. Therefore, great efforts have been made to improve the electrocatalytic activity of MoS2, including phase transformation (Tang and Jiang, 2016; Jiao et al., 2018; Wang J. et al., 2018), defect engineering (Xie et al., 2013a, 2017, 2019; Xie and Yi, 2015), nanocrystallization (Yun et al., 2017), doping (Xie et al., 2013b, 2016; Sun et al., 2014, 2018; Xiong Q. et al., 2018), modification (Benson et al., 2017; Wang Q. et al., 2018) and compounding (Jayabal et al., 2017; Zhai et al., 2018), etc.</p><p>The bulk phase MoS2 is inert for the electrocatalytic hydrogen evolution, and the free energy of hydrogen adsorption on the base surface of MoS2 is 1.92 eV. However, the theoretical results show that the ΔGH of Mo (1010) is 0.08 eV at 50% hydrogen coverage, which is close to the optimum value (≈0 eV) and exhibits the good electrocatalytic activity (Hinnemann et al., 2005). In addition, this propose is confirmed by the experimental results (Jaramillo et al., 2007). Theoretical and experimental studies have proved that the edge of MoS2 is active. Therefore, exposing more edge sites of MoS2 is an important method to enhance its electrocatalytic activity. Thus, the way to improve the electrocatalytic performance is classified to increase the "quantity" of active sites (Zhang J. et al., 2017). The electrocatalytic hydrogen evolution reaction is a two-electron transfer process, and the reaction rate depends largely on ΔGH. If the bonding between H2 and the surface is too weak, the adsorption (Volmer) step will limit the overall reaction rate; if the bonding is too strong, the desorption (Heyrovsky/Tafel) step will limit the reaction rate (Parsons, 1958; Wang et al., 2018b). Therefore, a highly active catalyst should have neither too strong nor too weak bonding intermediates. According to these points, by controlling the surface/interface properties of MoS2, the surface electronic properties, surface adsorption behavior and hydrogen evolution reaction path can be optimized, which can promote the kinetic process of the electrocatalytic hydrogen evolution and enhance the intrinsic catalytic ability (Otyepková et al., 2017; Chen et al., 2018). The research in this field is to improve the electrocatalytic activity by optimizing the "quality" of the active site. It has been proved that the transport of electrons between MoS2 layers needs to overcome certain barriers. The electron transport is dominated by the jump transport mode leading to the low transport efficiency, which limits the improvement of their electrocatalytic activities. So, the acceleration of the electron transport between layers is also an effective way to enhance the catalytic activity (Yu et al., 2013).</p><!><p>Monolayer MoS2 exhibits relatively high electrocatalytic activity due to the exposure of more active sites, which can enhance the "quantity" of active sites. Zhang et al. prepared monolayer MoS2 by low-voltage CVD method (Shi et al., 2014). By changing the growth temperature or the distance between source and substrate, the controllable boundary length of MoS2 was successfully realized. The electrocatalytic hydrogen evolution results showed that the exchange current density increased linearly with the increase of boundary length. By changing the morphology of monolayer MoS2, the boundary length could be further extended. The dendritic morphology enriched the boundary of monolayer MoS2 to a great extent, which contributed greatly to the enhancement of the electrocatalytic activity (Zhang et al., 2014; Xu et al., 2018). Fractal monolayer MoS2 can also promote the efficiency of the electrocatalytic hydrogen evolution reaction. The fractal monolayer MoS2 synthesized on the surface of fused quartz can expose a large number of active sites at its edge. Besides, the existence of large internal stresses in the fractal monolayer MoS2, causes more electrons to migrate to the edge active sites, further improving the electrocatalytic performance (Wan et al., 2018). This study also manifests that there is a linear relationship between the electrocatalytic hydrogen evolution activity and the number of marginal active sites of MoS2. The inert surface of MoS2 can be tuned into ordered porous structure by using template. The porous structure can increase the proportion of edge atoms (the number of active sites), resulting in the enhanced electrocatalytic performance (Su et al., 2018). MoS2 nanosheets with rich 1T phase content can be prepared by the chemical peeling. These nanosheets possess many defects, which benefit to the good catalytic activity in the electrocatalytic hydrogen evolution (Voiry et al., 2013; Chang et al., 2016). Doping monolayer MoS2 can activate the activity of the base surface and improve the catalytic activity. For example, the doping of transition metal element Co atoms can change the surface electronic structure of MoS2 and the adsorption energy of hydrogen atoms, improving the catalytic performance (Hai et al., 2017; Lau et al., 2018). The post treatment on monolayer MoS2 is also a strategy to enhance catalytic capacity. Processing with oxygen plasma can increase defects and interfaces in a large extent, which play a certain role in increasing active sites and enhancing intrinsic catalytic activity (Ye et al., 2016).</p><p>However, the electrocatalytic hydrogen evolution of monolayer MoS2 is still limited. The number of active sites, intrinsic catalytic activity and interlayer conductance limit the further improvement of the electrocatalytic hydrogen evolution performance of monolayer MoS2. The process for improving the electrocatalytic properties of monolayer MoS2 is complicated or needs special equipments which may limit its practical application. In order to overcome the defects of monolayer MoS2, it is necessary to composite monolayer MoS2 with other low-dimensional materials.</p><!><p>The electronic structure of the surface and the binding energy of the active intermediates can be modulated by compositing the single layer MoS2 and 0D, 1D, and 2D materials, leading to the improvement of the electrocatalytic performance by means of the active sites on the "quality" aspect. At the same time, the interlayer conductance of MoS2 can be enhanced by forming the composite, which further improves the electrocatalytic activity from another aspect. Zhang et al. composited Pd, Pt, and Ag nanoparticles with monolayer MoS2 by wet chemical method (Huang et al., 2013). The Pt nanoparticles with the size of 1-3 nm are successfully composited on the surface of monolayer MoS2. The Pt-modified monolayer MoS2 showed the excellent electrocatalytic performance with the neglected overpotential and the comparable Tafel slope of 40 mV/dec compared with pure MoS2 and Pt, which could be ascribed to the effective collection and transport of electrons in the presence of Pt.</p><p>Polyoxometallates (POMs) possess excellent performance in catalysis, which is attributed to the abundant oxygen on the surfaces and rich negative charges (Huang J. et al., 2017; Huang et al., 2018b). Polyoxometallates have the abilities to accept multiple electrons and reversible redox properties, which means that they have the outstanding electronic transport properties (Ammam, 2013). The MoS2 nanosheets were successfully exfoliated using the liquid phase exfoliation method assisted by formamide solvothermal treatment (Huang et al., 2018a). The monolayer MoS2 and POM were stacked into a multilayer heterostructure by the layer-by-layer (LBL) method, and the process for buildup of multilayer films is shown in Figure 2A. The electrocatalytic performance was improved due to the high electron transport performance of POMs and the electrochemical test results were plotted in Figures 2B–E.</p><!><p>(A) Process for building up the multilayer films (PMo12/MoS2)n, (B) Polarization curves of multilayer (PMo12/MoS2)n, (C) Current density as a function of layer number (inset of (c) Photograph of thin films of (PMo12/MoS2)n deposited on ITO with different number of layer), (D) Polarization curves of multilayer (PMo12/MoS2)4 and (MoS2)4, (E) The EIS spectra of multilayer (PMo12/MoS2)4 and (MoS2)4 (Huang et al., 2018a).</p><!><p>One-dimensional (1D) nanostructures offer the unique electronic transport channels. By compositing them with monolayer MoS2, the carrier transport capacity can be improved and the carrier recombination can be reduced. At the same time, the composite structure can bring the modification of the interface and the change of electronic structure, then the electrocatalytic performance can be further improved. Kim group prepared monolayer MoS2 by Li intercalation method (Ahn and Kim, 2017). 1D carbon nanotubes and MoS2 nanosheets were composited by LBL method to form a multilayer structure. The fabrication process was shown in Figure 3A. The hydrogen evolution performance reached the optimum value with the Tafel slope of 62.7 mV/dec for the number of layers of 14. The enhanced catalytic performance was attributed to the high conductivity of carbon nanotubes, which increased the conductivity of interlayer of MoS2, as shown in Figures 3B,C.</p><!><p>(A) Schematic of the LbL assembly of (MoS2/MWNT)n multilayer electrode, (B) Polarization curves of hybrid multilayer (MoS2/MWNT)n electrodes (Inset of B is the experimental setup of the three-electrode system), (C) Corresponding Tafel plots (Ahn and Kim, 2017).</p><!><p>Xia et al. combined Au nanorods with MoS2 to achieve surface plasmon resonance under auxiliary illumination, which increased the carrier concentration. Moreover, the improved carrier injection and carrier separation efficiency benefited from 1D structure can enhance the electrocatalytic efficiency (Shi et al., 2015).</p><!><p>The calculation results pointed out that the combination of graphene oxide and MoS2 could change the interface electronic structure, improve electron transport, and enhance electrocatalytic performance (Tang S. et al., 2018). The combination of graphene and monolayer MoS2 could increase the number of active sites, accelerate the desorption rate of H2 and enhance the efficiency of electron injection, and thus greatly boosted the electrocatalytic activity (Huang H. et al., 2017).</p><p>Sasaki team successfully exfoliated bulk MoS2 to obtain monolayer by Li intercalation method, and the monolayer MoS2 was verified to be 1T phase structure (Xiong P. et al., 2018). Then, it can be seen that the monolayer MoS2 was successfully restacked with graphene to form composite structure by the flocculation method (Figures 4A–E). The electrochemical measurements (Figures 4F–J) showed that this structure exhibited excellent electrocatalytic hydrogen evolution performance with the overpotential of 88 mV and Tafel slope of 48.7 mV/dec. The long-term stability was also manifested at 10 mA/cm2 for 10,00,00 s. The outstanding electrochemical properties could be originated from the enhanced electron transport and reduced Gibbs free energy of this unique structure.</p><!><p>(A) AFM images and height profiles for the exfoliated metallic MoS2 nanosheets, (B) AFM images and height profiles for the PDDA-graphene nanosheets, (C,D) SEM images of the MoS2/graphene superlattice with different magnifications, (E) SEM image and corresponding elemental mapping images of the MoS2/graphene, (F) Polarization curves, (G) Tafel plots, (H) The EIS spectra, (I) The polarization curves of the MoS2/graphene superlattice before and after the 105 s test, (J) Long-term stability measurement (Xiong P. et al., 2018).</p><!><p>According to the characteristics of 0D, 1D, 2D materials, it can play different roles in the composite structure, which can serve as an enhanced electron transport function, as well as to increase the active site, or to activate the in-plane properties. In the actual operation, the electrocatalytic performance can be improved with diverse composite structure in different aspects, such as intrinsic catalysis, increasing the number of active sites, and improving conductivity.</p><!><p>Monolayer MoS2 has attracted extensive attention for the electrocatalytic hydrogen evolution. In order to overcome the limitations of active sites, low intrinsic catalytic activity and poor interlayer conductivity, surface modification and composite structure are carried out to improve the electrocatalytic performance. However, there are still some challenges to be worthy of further investigation. Firstly, the properties of composite structure, active site and interface of composite materials are not clear, and need to be studied by more detailed characterization methods; secondly, the comprehensive utilization of monolayer MoS2 and its composite structures at the macro, nano and atomic levels will improve the efficiency of the electrocatalytic hydrogen evolution in principal, but involving the preparation, test and mechanism explanation of materials and devices. It belongs to the multi-disciplinary frontier field and can be studied through in-depth research. In-depth research on these issues can provide the clue for the improvement of the efficiency of the electrocatalytic hydrogen evolution and the deep insight of catalytic mechanism.</p><!><p>All authors listed have made a substantial, direct and intellectual contribution to the work, and approved it for publication.</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. This work was supported by the Natural Science Foundation of Shandong Province (Grant No. ZR2016FM30).</p>

PubMed Open Access

Chemoenzymatic synthesis of glycopeptides bearing rare N-glycan sequences with or without bisecting GlcNAc

N-Linked glycopeptides have highly diverse structures in nature. Herein, we describe the first synthesis of rare multi-antennary N-glycan bearing glycan chains on 6-OH of both a1,6and a1,3-linked mannose arms. To expedite divergent generation of N-glycan structures, four orthogonal protective groups were installed at the branching points on the core tetrasaccharide, which could be removed individually without affecting one another. In addition, the synthetic route is flexible, allowing a bisecting glucosamine moiety to be introduced at a late stage of the synthesis, further expanding the diversity of sequences that could be achieved. The bisecting glucosamine unit significantly reduced the glycosylation yields of adjacent mannoses, which was attributed to steric hindrance imposed by the glucosamine based on molecular modelling analysis. The N-glycans were then transformed to oxazoline donors and ligated with a glycopeptide acceptor from haptoglobin promoted by the wild type Arthrobacter endo-b-N-acetylglucosaminidase (Endo-A). Endo-A exhibited interesting substrate preferences depending on donor sizes, which was rationalized through molecular dynamics studies. This is the first time that a glycopeptide bearing a bisecting N-acetyl glucosamine (GlcNAc), the rare N-glycan branch, and two Lewis X trisaccharide antennae was synthesized, enabling access to this class of complex glycopeptide structures.

chemoenzymatic_synthesis_of_glycopeptides_bearing_rare_n-glycan_sequences_with_or_without_bisecting_

Introduction<!>General experimental procedures<!>Solid-phase peptide synthesis using the Fmoc-strategy<!>General procedure for Fmoc removal<!>General procedure for Lev removal<!>General procedure for Troc removal<!>General procedure for TBDPS removal<!>General procedure for glycosylation catalyzed by NIS/AgOTf<!>General procedure for global deprotection<!>General procedure for synthesis of glycosyl oxazolines<!>Computational modeling of compounds 17 and 20<!>Computational modeling of Endo-A complexes with sugar oxazolines 39 and 41<!>Results and discussion<!>Conclusion<!>Conflicts of interest

<p>Protein glycosylation, one of the most common posttranslational modications, [1][2][3] plays critical roles in directing biological functions, 4 stabilities 5 and conformations of the parent proteins. 6 Among glycoproteins, the N-linked structures form a major subset, in which carbohydrate residues are attached to the core protein backbone through amide bonds with asparagine residues. 7 N-glycans share a pentasaccharide core, the two mannose units (D and E) at the non-reducing end of which can bear multiple branches (Fig. 1). While the most common points of attachment are C2/C6 hydroxy groups of mannose D (a1,6-arm) and C2/C4 hydroxy groups of mannose E (a1,3-arm), a novel b1,6-N-acetylglucosaminyltransferase (b1,6GnT-IX also designated as b1,6GnT-Vb) has been identied that can add a glycan to 6-OH of mannose E. 8 b1,6 branching is known to be important to malignant phenotypes of prostate cancer, neuroblastoma and melanoma. 9,10 In addition to modications of D and E mannoses in the periphery, 4-OH of mannose C can be modied by a glycan such as N-acetyl glucosamine (GlcNAc), bisecting mannoses D and E. The bisecting GlcNAc is important for many biological processes, including tumor development, 11 immune response 12 and cellcell communication. 13 However, it is very challenging to obtain bisected N-glycans through either isolation from nature or synthesis.</p><p>During the past decade, tremendous advances have been made in synthesis of highly complex N-glycans with many innovative methods developed. [14][15][16][17][18][19][20][21][22] However, to date, N-glycans bearing a branch at 6-OH of mannose E (a1,3-arm) have not been synthesized. Given the structural complexity of N-glycans, synthesis of asymmetric antennae at each branching point remains a highly challenging task. To address this, we report our strategy enabling synthetic access to rare structures of Nglycans bearing antennary glycans on 6-OH of mannose E. To generate diverse glycan structures, a core tetrasaccharide was designed with potential branching points strategically protected with temporary protective groups that can be orthogonally removed. Furthermore, suitable conditions have been identied for the late stage installation of the bisecting GlcNAc moiety. The reducing ends of the synthetic N-glycans were then converted to oxazoline and transferred to a GlcNAc functionalized glycopeptide by endo-b-N-acetylglucosaminidase (Endo-A), [23][24][25][26][27][28] providing access to glycopeptides bearing these rare Nglycan sequences.</p><p>To improve the overall efficiency of glyco-assembly, our Nglycan synthesis design hinges on divergent modications of a key tetrasaccharide 1 bearing four protective groups, i.e., tertbutyl-diphenyl silyl (TBDPS), uorenylmethyloxycarbonyl (Fmoc), 2,2,2-trichloroethoxycarbonyl (Troc) and levulinoyl (Lev), which are strategically placed at the potential branching points on mannoses D and E and may be removed orthogonally. Orthogonal deprotections have been shown to be powerful strategies for divergent synthesis of a large number of glycans including N-glycans. 16,[29][30][31] The free 4-OH of the mannose-C unit in 1 was kept free for late stage installation of the bisecting GlcNAc.</p><!><p>All chemical reactions were carried out under nitrogen with anhydrous solvents in ame-dried glassware, unless otherwise noted. Glycosylation reactions were performed in the presence of molecular sieves, which were ame-dried right before the reaction under high vacuum. Glycosylation solvents were dried using a solvent purication system and used directly without further drying. Chemicals used were reagent grade as supplied except where noted. Compounds were visualized by UV light (254 nm) and by staining with a yellow solution containing Ce(NH 4 ) 2 (NO 3 ) 6 (0.5 g) and (NH 4 ) 6 Mo 7 O 24 $4H 2 O (24.0 g) in 6% H 2 SO 4 (500 mL). Flash column chromatography was performed on silica gel 60 (230-400 mesh). NMR spectra were referenced using residual CHCl 3 (d 1 H-NMR 7.26 ppm) and CDCl 3 (d 13 C-NMR 77.0 ppm). Peak and coupling constants assignments are based on 1 H-NMR, 1 H-1 H gCOSY and/or 1 H- 13 C gHMQC and 1 H- 13 C gHMBC experiments.</p><!><p>Most of amino acids and resins were purchased from Chemimpex. Reaction vessels (10 mL, disposable) and the Domino Block Synthesizer were purchased from Torvig. All peptides and glycopeptides were synthesized according to the Fmocchemistry based on solid phase peptide synthesis procedure. Resins with pre-loaded amino acid were loaded into a plastic syringe tted with a lter and swelled in DCM for at least 1 h. For the coupling reactions, the Fmoc-amino acid (5.0 equiv.) was activated by O-(1H-benzotriazol-1-yl)-N,N,N 0 ,N 0 -tetramethyluronium hexauorophosphate (HBTU) (4.9 equiv.), 1-hydroxybenzotriazole (HOBt) (4.9 equiv.), N,N-diisopropylethylamine (DIPEA) (10.0 equiv.) and anhydrous DMF (5 mL) for 30 min. Then this mixture containing activated Fmoc amino acid was transferred to the syringe containing the resin, which was then rotated on a tube rotator to mix the solvent with the resins. Aer completion of coupling, the resin was washed with DCM (3  5 mL) and DMF (3  5 mL) for 1 minute, each time followed by cleavage of the Fmoc group by treatment of the resin with a solution of piperidine (20%) in DMF for at least 2  20 min at r.t. Aer every coupling step, unreacted amino groups were capped by treatment with a mixture of Ac 2 O (0.5 mL), and DIPEA (1 mL) in DMF (3.5 mL) (capping reagent) for 2 times at 15 min each. For coupling of the glycosylated amino acid building block, (Fmoc-[GlcNAc(OAc)3b1-]Asn-OH) (2 equiv.) was dissolved in DMF and activated with 1-[bis(dimethylamino)methylene]-1H-1,2,3-triazolo[4,5-b]pyridinium 3-oxid hexauorophosphate (HATU) (2 equiv.), 1-hydroxy-7-azabenzotriazole (HOAt) (2 equiv.) and DIPEA (4 equiv.). Aer completion of the glycopeptide chain, the resin was washed and the glycopeptide was cleaved from the resin by treatment with triuoroacetic acid (TFA)/ triisopropylsilyl ether (TIPS)/H 2 O (95% : 2.5% : 2.5%) solution for 2.5 h. Aer ltration, the resins were washed with triuoroacetic acid (2  10 mL), and the volume of the combined ltrates was concentrated to 1 mL, then absolute Et 2 O (15 mL) at 0 C was added dropwise to the residues. The precipitates were separated from the mother liquor by centrifugation and washed with cold Et 2 O (10 mL). The crude products were dissolved in methanol and treated with NaOMe to cleave the ester protecting group on sugar, then neutralized with AcOH and the solvent was removed. Finally, the crude mixture was dissolved in H 2 O and subjected to a SUPELCOSILTM LC-18 HPLC column (25 cm  10 mm, 5 mm) or (25 cm  4.6 mm, 5 mm) for purication. The solvent systems used were A (0.1% TFA in H 2 O) and B (acetonitrile (MeCN)) with detection at 220 nm and 254 nm. Mass spectra were obtained by ESI mass spectra (Water Xevo G2-S Q-TOF LC-MS instrument).</p><!><p>The starting material was dissolved in triethylamine and ethyl acetate (1/1, 0.5 mmol coupling product in 5 mL) and the reaction stirred at room temperature for 6 hours, when TLC indicated the completion of reaction. The solvent was evaporated under reduced pressure, and co-evaporated three times with ethyl acetate. Then the residue was puried by ash column chromatography.</p><!><p>To a solution of a glycan (0.1 mmol) in DCM : CH 3 OH (1/1) was added hydrazine acetate (1 mmol), which was prepared in situ by mixing acetic acid and 51% hydrazine (4/1). The reaction was stirred at room temperature for 3 h, aer which it was diluted with ethyl acetate and washed with sat. NaHCO 3 and brine. The organic phase was dried (Na 2 SO 4 ), ltered, and the ltrate was concentrated under reduced pressure. The residue was puried by ash column chromatography.</p><!><p>To a solution of a glycan (0.1 mmol) in THF and acetic acid (4 : 1, 5 mL) was added zinc dust (200 mg). The reaction was stirred at room temperature for 20 min, aer which it was ltered by celite and the ltrate was concentrated under reduced pressure. The residue was puried by ash column chromatography.</p><!><p>To a solution of a glycan (0.1 mmol) in pyridine (3 mL) was added 70% HF$Py (1 mL) at 0 C. The reaction was stirred at room temperature for 3 h, aer which it was neutralized by sat. NaHCO 3 . The mixture was extracted with ethyl acetate and washed with 10% HCl, sat. NaHCO 3 and brine. The organic phase was dried (Na 2 SO 4 ), ltered, and the ltrate was concentrated under reduced pressure. The residue was puried by ash column chromatography.</p><!><p>A mixture of acceptor (20 mM in DCM), thioglycoside donor (3 equiv to each -OH), and 4 Å molecular sieves in DCM was stirred at room temperature for 30 min. Then it was cooled to À30 C followed by addition of N-iodosuccinimide (1.2 equiv. to donor) and AgOTf (0.1 equiv. to donor). The reaction was stirred at À30 C for 1.5 hours, aer which it was quenched by DIPEA and ltered by celite, and the ltrate was concentrated under reduced pressure. The residue was puried by ash column chromatography.</p><!><p>A mixture of protected oligosaccharide (0.01 mmol) and ethylene diamine : nBuOH (5 mL, 1/4) was stirred at 90 C overnight. The volatiles were evaporated, and the residue was puried by Sephadex LH-20. The crude product was dissolved in 0.5 mL pyridine followed by the addition of 0.05 mL acetic anhydride-1,1 0 -13 C 2 and the reaction was stirred overnight. The residue was puried by Sephadex LH-20. The products were then de-acetylated using sodium methoxide in MeOH (1.5 mL) overnight. The reaction mixture was neutralized by IR-120, ltered and concentrated in vacuum, and puried by Sephadex LH-20. To the products dissolved in MeOH : H 2 O : DCM (4/ 1/1, 2 mL), Pd(OH) 2 (100% by weight) was added, and the mixture was hydrogenated overnight. The reaction mixture was ltered through cotton and concentrated. The residues were puried by G-15 gel ltration chromatography using water as the eluent. The collected solution containing the product was lyophilized to obtain the product as a white powder.</p><!><p>The free oligosaccharide was dissolved in H 2 O and cooled in an ice bath. Triethylamine (45 equiv.) and 2-chloro-1,3dimethylimidazolinium chloride (DMC) (20 equiv.) were added and the reaction mixture was stirred in the ice bath. The reaction was monitored using High pH Anion Exchange Chromatography (HPAEC). Upon completion, purication was carried out on a Sephadex G10 gel ltration chromatography column eluting with 0.1% aqueous triethylamine. The pooled product fractions were then lyophilized to afford the product glycan oxazolines. The products were characterized using HPAEC and ESI mass spectrometry.</p><!><p>To obtain conformers covering a wide range of the conguration space, plain MD simulations were performed at 900 K. From the resulting MD frames, 1000 conformers were extracted and further optimized using the AM1 semi-empirical method. Then, the optimized 1000 conformers were sorted by their AM1 total electronic energies and the conformers within 6.0 kcal mol À1 of the lowest energy conformation of each compound were selected for geometrical comparison. The bonds, angles, and dihedral torsion parameters involving Si, that were absent in the amber general force eld (gaff), were generated by tting amber energies to the B3LYP/6-31G* energies of various conformers of TBDPS-OMe obtained from scanning all its dihedral angles. Amber 14 (Tools 15) 32 and Gaussian 33 were used for MD and quantum chemical simulations, respectively.</p><!><p>Initial coordinates of Endo-A were obtained from the Protein Data Bank 34 (PDB ID: 3FHA). 35 As the focus of the study was on pocket residue-ligand interaction, missing segments and residues outside the pocket region were capped using Molecular Operating Environment v.2016.08 (MOE). 36 Gate-keeper residues, W216 and W244 are positioned parallel to one another during transglycoslyation. 35 W244 was rotated from its original perpendicular orientation to parallel with W216. Protein was initially minimized in MOE under the AMBER ff10 force eld 37 and Extended Hückel Theory.</p><p>The volumes of N-glycan oxazolines 39 and 41 were calculated using the van der Waals volume QSAR descriptor of MOE, using a connection table approximation to calculate 2D molecular descriptors. The compounds were then noncovalently docked with the docking program within MOE. Binding poses were rened using an induced t renement method.</p><p>The geometries of the N-glycan oxazoline compounds were optimized using the Gaussian 16 program package. The optimizations were performed using the AM1 method. 38 The obtained Mulliken charges were used with the antechamber module of Amber 16 in the generation of parameters for the Nglycan compounds. The systems were prepared using the Leap module of AmberTools16 (ref. 32) under the AMBER ff14SB and GAFF force elds. Each enzyme complex was solvated in a 14 Å cube of TIP4P-Ew water beyond the solute and 100 mM sodium chloride. The systems were relaxed under NVT conditions over six minimization procedures with decreasing restraints on the protein of 500.0, 200.0, 20.0, 10.0, 5.0 kcal mol À1 ÅÀ2 to no restraints. The systems were then heated to 300 K over 30 ps. Atomistic molecular dynamics simulations were performed for 30 ns at 300 K and 1 atm using AMBER 16. The SHAKE algorithm constrained bonds involving hydrogen. 39 The trajectories were produced using Langevin dynamics and the pressure of the system was regulated with isotropic position scaling. Longrange electrostatic effects were modeled using the particle-mesh Ewald method with a 10 Å cutoff. The produced trajectories were analyzed using AMBER 16 and visualized with MOE and the UCSF Chimera package. Free energy of binding was calculated for every picosecond using the Poisson Boltzmann model form the MMPBSA.py module of AmberTools and AMBER 16. The relative free energy trends between models were compared so solute entropy was neglected.</p><!><p>The synthesis of key tetrasaccharide 1 commenced from disaccharide 2 27 (Scheme 1). Upon removal of the p-methoxybenzyl (PMB) group from 2, the glycosylation between the newly generated acceptor 3 and donor 4 went smoothly as promoted by N-iodosuccinimide (NIS) and TfOH to provide trisaccharide 5 in 88% yield. Trisaccharide 5 was treated with 80% AcOH to remove the benzylidene moiety followed by glycosylation with donor 7 to afford tetrasaccharide 1 bearing four different protecting groups at the branching points.</p><p>We next focused on the introduction of bisecting GlcNAc to the free 4-OH of tetrasaccharide 1. Glycosylation of the 4-OH group is known to be challenging as it is anked by two glycans. 18,40,41 Prior strategies to synthesize N-glycans with the bisecting glucosamine unit typically introduced the bisecting residue at 4-OH rst followed by the removal of the protective group on 6-OH and installation of the a1,6-arm. 27,[42][43][44][45][46][47] The direct Scheme 4 Glycans can be installed at the strategic branching points after orthogonal deprotections. glycosylation of an acceptor containing free 4-OH with both a1,3and a1,6-arms is attractive as it reduces protective group manipulations needed. Unverzagt and coworkers successfully developed this approach although a large excess (10 equiv.) of glycosyl donors was needed to overcome the low nucleophilicity of the 4-OH. 18,40,41 In our synthesis, we rst tested the direct glycosylation of tetrasaccharide 1 with 1.5 equiv. of glucosamine donor 8 (ref. 48) in the presence of NIS/TfOH. Surprisingly, while the desired pentasaccharide was detected from the reaction mixture by mass spectrometry, a pentasaccharide side product containing an iodide group was also formed, which likely resulted from electrophilic substitution of the product by the electron-decient iodonium ion generated from NIS. The side product could not be separated from the desired pentasaccharide. While NIS/TfOH is a promoter system widely used to activate thioglycosides, 49 the side reaction of iodination was not unprecedented as the Yu group reported aglycon iodination during NIS promoted thioglycosylation of a steroid derivative. 50 As an alternative to NIS/TfOH, we tested the reaction of 1 and 8 using p-TolSCl/AgOTf 51 as the promoter. In this case, however, a pentasaccharide side product containing a p-TolS moiety was found.</p><p>Scheme 5 Orthogonal deprotection and glycosylation for the synthesis of complex N-glycans 31-34.</p><p>We hypothesize that the most likely location in tetrasaccharide 1 that was electrophilically substituted would be the methyl or methylene group a to the ketone carbonyl of the Lev. As electron-withdrawing protective groups, i.e., Troc and Lev, are on the same mannose unit of 1, the acidity of a-H of ketone may be higher. This consideration led us to switch the protective groups of Fmoc and Lev on the mannosyl donors, i.e., using new donors 9 and 12. The coupling between donor 9 and disaccharide acceptor 3 went smoothly, which followed by subsequent benzylidene removal gave the trisaccharide 11 in 77% yield (Scheme 2). Selective glycosylation at the primary hydroxyl group of 11 by mannose donor 12 generated tetrasaccharide 13. The 4-OH of tetrasaccharide 13 was protected as the O-acetate (tetrasaccharide 14) to synthesize N-glycans without the bisecting moiety.</p><p>The bisecting glucosamine was successfully introduced via the reaction of 13 with 2 equiv. of donor 8 with NIS/TfOH promoter to form pentasaccharide 15 (Table 1, entry 1). Although the yield was a modest 37%, the undesired iodide substitution products were not detected. When 3 equiv. of donor 8 was used, the yield increased to 53% (Table 1, entry 2). Further increasing the amount of donor did not improve the yield signicantly, but rendered purication more complicated. The major side product was due to the reaction of succinimide generated from NIS with the activated glycosyl donor, which competed with the formation of the desired product. The promoter p-TolSCl/AgOTf was utilized next instead of NIS/TfOH. With 1.5 equiv. of donor 8, the desired pentasaccharide 15 was obtained in 44% yield (Table 1, entry 3). Increasing the amount of donor from 1.5 to 3.0 equiv. led to a signicant increase of the yield of glycosylation to 81% (Table 1, entries 3-5). This compared favorably with several literature approaches 18,40,41 for introduction of the bisecting glycan as good glycosylation yield was obtained without the need for a large excess (10 equiv.) of the glycosyl donor.</p><p>With the key pentasaccharide 15 in hand, the possibility of orthogonally removing each of the temporary protective groups was explored. As shown in Scheme 3, treatment of 15 with triethylamine selectively removed the Fmoc group in 95% yield without affecting any other protective groups. The Lev group in 15 could be deprotected by hydrazine acetate, while Troc and TBDPS moieties were cleaved by Zn/AcOH and HF$pyridine respectively in good yields.</p><p>With the orthogonality of protective group removal established, glycosylations of the newly liberated hydroxyl groups were tested for the construction of asymmetrically branched Nglycans. To understand the impact of the bisecting glycan on glycosylation, both pentasaccharide 17 and tetrasaccharide 20 obtained from Lev removal of 14 were utilized as acceptors.</p><p>Glycosylation of pentasaccharide 17 with donor 8 promoted by NIS/AgOTf gave hexasaccharide 21 in 53% yield (Scheme 4). Other promoter systems such as NIS/TMSOTf, NIS/TfOH and p-TolSCl/AgOTf were also tested, but did not lead to better yields. Compound 21 was treated with Et 3 N followed by glycosylation by disaccharide 23 and removal of TBDPS, affording octasaccharide 24 in 43% overall yield. Lewis X trisaccharide thioglycosyl donor 26 was prepared through a one pot procedure in gram scale. 52 Introduction of donor 26 onto the primary OH of octasaccharide 24 proceeded to give undecasaccharide 27 in 33% yield, where 21% of the unreacted acceptor was recovered in the reaction.</p><p>To synthesize glycans without the bisecting glycan, analogous glycosylations with donors 8, 23, 26 followed by subsequent deprotections were performed on the tetrasaccharide acceptor 20 (Scheme 4). The penta-, hepta-and decasaccharides 22, 25 and 28 were efficiently synthesized in 71%, 70% and 74% yields respectively, signicantly higher than the analogous yields for forming 21, 24 and 27 (53%, 43% and 33%). These results demonstrate that the bisecting Scheme 6 Synthesis of 13 glucosamine negatively impacted glycosylation yields at branching points.</p><p>To better understand the effect of the bisecting glycan, molecular modeling studies were performed. Both glycans 17 and 20 were subjected to molecular dynamics (MD) simulations at 900 K. From the resulting MD frames, 1000 conformers were extracted and optimized using the AM1 semi-empirical method. The conformers within 6.0 kcal mol À1 from the lowest energy conformation were selected for comparisons. It was noteworthy that the glycan 17 was more rigid than 20, as there were fewer distinct conformers of 17 than 20 in this energy range (Table S1 †). Conformational analysis showed that the free 2-OH of compound 17 was much more sterically hindered than that in compound 20. The average number of atoms within 5 Å of the OH group in the low energy conformers of 17 was quantied and found to be signicantly higher than that in compound 20 (42.9 vs. 35.9 Table S1 †). Furthermore, the OH group was pointing at the bisecting glucosamine unit in many conformers of 17 (Fig. 2). There were no additional hydrogen bond donors interacting with the free OH group in the optimized structures of 17 compared to that in 14. Therefore, the lower yields of glycosylation of 17 were mainly attributed to the increased steric hindrance around the OH group caused by the bisecting glucosamine unit. As N-glycans are attached to proteins, we next explored the possibility of introducing the rare N-glycan to peptides with haptoglobin peptide (amino acids 233-254) 53,54 as a representative acceptor. The transglycosylation strategy by endoglycosidases [23][24][25][26][27][28] pioneered by the Wang lab can enable convergent synthesis of glycopeptides by transferring glycosyl oxazoline donors to peptide acceptors bearing a GlcNAc on an asparagine residue. While the successful transfer of a bisected N-glycan by the endoglycosidase Endo-A has been reported, 27 the donor tested did not contain a1,3or a1,6-glycan branches. Furthermore, it is unclear how the bisecting GlcNAc may impact the transglycosylation reaction.</p><p>To synthesize N-glycans for haptoglobin glycopeptides, compounds 15 and 14 were rst treated with Zn/AcOH and HF/ Py giving diols 29 and 30. Lewis X trisaccharide thioglycosyl donor 26 glycosylated diols 29 and 30 using the NIS/AgOTf promoter system followed by subsequent Lev and Fmoc removal producing undeca-and deca-saccharides 31 and 32 in 40% and 51% yields respectively (Scheme 5a). To prepare triantennary N-glycan, the Lev group in 29 and 30 was selectively removed. The resulting triols were glycosylated with Lewis X trisaccharide thioglycosyl donor 26 leading to tetradeca-and trideca-saccharides 33 and 34 (Scheme 5b).</p><p>The four oligosaccharides 31-34 were deprotected by rst incubating with ethylenediamine in n-butanol at 90 C, 31 which removed all ester and carbonate groups as well as the Phth to generate free amines (Scheme 6). To facilitate future quantication by mass spectrometry, 55,56 the free amines were protected with 13 C labeled acetic anhydride to introduce 13 C labels into the GlcNAcs. Subsequent hydrogenolysis with Pearlman's catalyst under a hydrogen atmosphere 57 led to free N-glycans 35-38. With the free N-glycans in hand, they were transformed to the respective oligosaccharide oxazolines 39, 40, 41 and 42 using a mild chloroformamidinium dehydrating agent 43 (Scheme 7a), which were directly subjected to the enzymatic transglycosylation reactions with GlcNAc bearing haptoglobin glycopeptide 44 as the acceptor (Scheme 7b). Glycosyl oxazoline 40 was rst tested using endoglycosidases Endo-A (wild type) and Endo-M (wild type), and their respective Endo-A N171A, Endo-M N175A and Endo-M N175Q mutants as catalysts. Interestingly, we found that wild type Endo-A was able to transfer the glycan to give glycopeptide 46 in 73% overall yield, while all other enzymes tested failed to generate signicant quantities of desired products. Next, the bisecting GlcNAc containing oxazoline 39 was utilized as the donor. Similar to donor 40, Endo-A promoted the transglycosylation reaction efficiently producing glycopeptide 45 in 65% yield. These results suggest that the presence of the bisecting GlcNAc does not negatively affect the donor abilities of 39 and glycosyl oxazolines containing bisecting GlcNAc and complex glycans such as rare a1,3and a1,6-branches can be suitable donors for transglycosylation to glycopeptides.</p><p>To test the limit of the Endo-A enzymes, tri-antennary Nglycan oxazolines 41 and 42 bearing three Le X moieties were subjected to transglycosylation reactions with acceptor peptide 44. However, no desired glycopeptides were observed with Endo-A. Nor did the Endo-M or any of the Endo-A N171A, Endo-M N175A and Endo-M N175Q mutants tested give desired glycopeptides.</p><p>To better understand the divergent behavior of glycosyl oxazolines 39 vs. 41, computational studies were performed on their complexes with Endo-A. The initial coordinates of Endo-A were obtained from the Protein Data Bank 34 (PDB ID: 3FHA) 35 and N-glycan oxazolines 39 and 41 were docked into the binding pocket of Endo-A using the docking program within Molecular Operating Environment v.2016.08 (MOE). 36 Binding poses were scored using the GBVI/WSA DG scoring function. It is known that W93, N171, E173, Y205, F125, W216, F243, W244 and Y299 are the active site residues interacting with the donor substrate. 35 Poses with the oxazoline ring positioned within range of these residues were simulated using atomistic MD with the AMBER 16 soware package. 32 The lowest energies poses with the oxazolines remaining in the binding pockets were obtained and analyzed. MD simulations demonstrate that the binding energy of 39 and Endo-A is on average À81.44 AE 11.10 kcal mol À1 (Table 2). In contrast, the binding energy of 41 is signicantly lower at À56.24 AE 9.95 kcal mol À1 , suggesting the much weaker binding of 41 by Endo-A. In crystal structures of Endo-A as well as that of Endo-A complexed with a tetrasaccharide oxazoline substrate, 35 the indole rings of W216 and W244 are perpendicular to each other, possibly forming a gate sealing off the active site. When oxazoline 39 was docked into Endo-A, the indole of W244 was moved to be parallel to that of W216 to allow the substrate to enter the active site. In all poses generated from subsequent MD simulations, the indole of W244 returned to the original perpendicular orientation to W216 (Fig. 3a). However, in the complex of the tri-antennae bearing donor 41 with Endo-A, the additional antenna of 41 is located between these two tryptophan residues, possibly preventing the rotation of the indole ring. As a result, the two indoles of W244 and W216 remained close to parallel to each other (Fig. 3b). This may have precluded the formation of a closed active site for catalysis, reducing the yields for transglycosylation.</p><!><p>A exible synthetic route to rare N-glycan structures has been designed with the core tetrasaccharide 13 bearing four orthogonal protecting groups serving as a linchpin. Through this key intermediate, many N-glycan oligosaccharides could be accessed including asymmetric antennary structures through orthogonal deprotection followed by glycosylations. During the synthesis, it was found that if Troc and Lev groups were on the same mannosyl donor, subsequent glycosylation could not give the desired product in good purity due to the introduction of an iodine into the molecule. This problem was successfully solved by switching the protecting groups of Fmoc and Lev in the two mannosyl donors. Furthermore, the synthesis was designed to enable late stage introduction of bisecting glucosamine. Suitable conditions have been identied to install the bisecting glycan in a high yield without resorting to a large excess of the glycosyl donor. The bisecting glucosamine was found to reduce the glycosylation yields of each branching point. Molecular modeling results indicated that the bisecting glycan sterically hindered the access of the free hydroxyl groups of the mannose branches for glycosylation. Finally, the limit and scope of endoglycosidase promoted transglycosylation reactions were tested. Two homogenous haptoglobin glycopeptides bearing an undeca-and a dodeca-saccharide respectively could be produced via wild type Endo-A catalyzed transglycosylation reactions. Sugar oxazoline donors containing three Le X in the antennae did not undergo productive transfer by the enzymes, which was likely because the bulky glycan substrate reduced binding affinity and prevented the necessary reactive conformations of the enzymes. This is the rst time that glycopeptides bearing bisecting GlcNAc and rare branches on a1,3-mannose have been produced, expanding the diversity of N-glycans and glycopeptides that can be accessed.</p><!><p>The authors declare no conicts of interests.</p>

Royal Society of Chemistry (RSC)

Design of Enzymatically Cleavable Prodrugs of a Potent Platinum-Containing Anticancer Agent

Using a versatile synthetic approach, a new class of potential ester prodrugs of highly potent, but systemically too toxic, platinum\xe2\x80\x93acridine anticancer agents was generated. The new hybrids contain a hydroxyl group, which has been masked with a cleavable lipophilic acyl moiety. Both butanoic (butyric) and bulkier 2-propanepentanoic (valproic) esters were introduced. The goals of this design were to improve the drug-like properties (e.g., logD) and to reduce the systemic toxicity of the pharmacophore. Two distinct pathways by which the target compounds undergo effective ester hydrolysis, the proposed activating step, have been confirmed: platinum-assisted, self-immolative ester cleavage in a low-chloride environment (LC-ESMS, NMR spectroscopy) and enzymatic cleavage by human carboxylesterase-2 (hCES-2) (LC-ESMS). The valproic acid ester derivatives are the first example of a metal-containing agent cleavable by the pro-drug-converting enzyme. They show excellent chemical stability and reduced systemic toxicity. Preliminary results from screening in lung adenocarcinoma cell lines (A549, NCI-H1435) suggest that the mechanism of the valproic esters may involve intracellular deesterification.

design_of_enzymatically_cleavable_prodrugs_of_a_potent_platinum-containing_anticancer_agent

Introduction<!>Design and chemistry<!>Metal-assisted ester hydrolysis<!>Ester cleavage by human carboxylesterase-2 (hCES-2)<!>Cytotoxicity studies<!>Conclusion<!>Materials, general procedures, and instrumentation<!>Synthesis of [PtCl(en)C22H27N4O2](NO3)2 (2)<!>Synthesis of [PtCl(en)C26H35N4O2](NO3)2 (3)<!>Synthesis of [PtCl(NH3)2C26H35N4O2](NO3)2 (4)<!>Synthesis of [PtCl(pn)C26H35N4O2](NO3)2 (5)<!>Synthesis of [PtCl(en)C28H39N4O2](NO3)2 (6)<!>Synthesis of [PtCl(en)C24H31N4O2](NO3)2 (7)<!>Synthesis of [PtCl(pn)C28H39N4O2](NO3)2 (8)<!>Time-dependent NMR spectroscopy<!>Chemical hydrolysis assay<!>Enzymatic cleavage assay<!>Determination of partition coefficients (logD)<!>Cell culture maintenance<!>Cytotoxicity assay

<p>Traditional chemotherapies often suffer from high systemic toxicity and a narrow therapeutic window. To improve the pharmacokinetics (PK) and toxicity profiles of anticancer drugs, various avenues are being pursued, such as nano-sized delivery platforms, receptor-targeted conjugates, and prodrug designs.[1, 2] The rationale behind the latter approach is to generate a precursor molecule, which is converted post administration to the bioactive form of the drug either enzymatically or in response to a chemical stimulus. Bioactivation may occur during absorption, circulation, or at the tumor site.[3] Potential benefits of lipophilic prodrugs include efficient retention in, and absorption from, circulation, as well as improved penetration of membranes and accumulation in target tissues.[1]</p><p>Compound 1 (PT-ACRAMTU, Figure 1; ACRAMTU =1-[2-(acridin-9-ylamino)ethyl]-1,3-dimethylthiourea) represents the prototype of a class of DNA-targeted platinum–acridine hybrid agents, which have shown exquisite potency in several solid tumor models.[4] Non-small cell lung cancer (NSCLC) cells prove to be particularly sensitive to this pharmacophore, with the newer derivatives showing IC50 values in NSCLC cell lines in the low-nanomolar range and activity in tumor xenografts.[4, 5] Unlike cisplatin [cis-diamminedichloridoplatinum(II)] and its analogues, platinum–acridines derived from PT-ACRAMTU do not cross-link DNA bases but produce structurally unique hybrid adducts that are an intrinsically more severe form of DNA damage than the former bifunctional adducts.[6–8] The classical structure–activity relationship (SAR) approach based on modular library screening has previously been used to tune the chemical stability and reduce the off-target reactivity of the pharmacophore.[9] The desired improvements were achieved essentially by modifying the ligand and donor sets around the electrophilic metal.[4] These efforts have led to the development of a derivative (1‴, Figure 1) that shows three orders of magnitude higher potency than cisplatin (the IC50 values for 1‴ in NCI-H460 and A549 lung cancer cells were 1.3 and 3.9 nM, respectively; see the Supporting Information for dose–response curves).</p><p>Despite their promising cell-kill properties in chemoresistant, intractable cancer cells, platinum–acridines show unfavorable ADME[1] (absorption, distribution, metabolism, and excretion) properties, which hamper their preclinical development. Compound 1′, for instance, while inhibiting the growth of xenografted NCI-H460 tumors in mice, showed signs of severe toxicity in the test animals, resulting in a low maximum tolerated dose (MTD).[5] Mice, which were necropsied after treatment with 1′, showed high levels of platinum in normal tissues, but insufficient accumulation in tumors, as well as discoloration of the kidneys, a possible sign of hepatotoxicity or nephrotoxicity.[5] To improve the pharmacological properties of platinum–acridines and to potentially open the therapeutic window for systemic treatment with these agents, we have now designed lipophilic ester-based derivatives that can be tuned to specifically undergo enzymatic hydrolysis. This concept has resulted in the first case of a platinum-containing agent that is recognized as a substrate by human carboxyl-esterase-2 (hCES-2), a key enzyme involved in the activation of several anticancer prodrugs.[1]</p><!><p>Platinum–acridines show excellent cytotoxicity but unfavorable drug-like properties. The most cytotoxic derivatives exist in their fully protonated [pKa(9-aminoacridine) ≈ 9.5–10], dicationic form.[4 Because these hybrids are too hydrophilic, they show poor tissue distribution and are most likely removed too rapidly from circulation via renal clearance.[10] Thus, the goal of the structural modifications introduced here was to increase the lipophilicity of the agents while maintaining good water solubility. The alkyl residue of the amidine donor group (residue R, for Y =NH, see Figure 1) was chosen as the site of attachment for the carboxylic acid ester groups. The design involved introduction of a hydroxymethyl (MeOH) or an extended 3-hydroxy-propyl (Pr3-OH) group as R in place of simple Me and Et in 1′–1‴ and masking of the terminal OH function as lipophilic esters (see Scheme 1). A primary carboxylic acid, butanoic (butyric) acid, and a bulkier, branched derivative, 2-propylpen-tanoic (valproic) acid, were introduced as acyl components (see Scheme 1). Use of the latter residue was inspired by an analogous valproic amide-based prodrug of the anticancer drug gemcitabine, which is activated by hCES-2.[11]</p><p>Two distinct mechanisms of ester cleavage have to be considered: chemical and enzymatic hydrolysis. The former mechanism has previously been observed in chemically related carboxylic acid-modified platinum–acridines containing a reversed ester linkage (Pt-linker-C(O)OR′, instead of Pt-linker-OC(O)R′ used in this study).[12] It is dominated by platinum-assisted ester cleavage in a chloride-ion concentration dependent manner. Low intracellular chloride favors aquation of platinum, which serves as a Lewis-acidic metallohydrolase and accelerates ester cleavage. The second mode of activation involves carboxylesterase isoenzymes, in particular hCES-2, which is not only expressed at high levels in the gastrointestinal tract and the liver, but also in tumor tissue.[1] The choice of hCES-2 rather than hCES-1 as the target enzyme was based on the well-documented substrate selectivity of the two forms, according to which hCES-2 preferentially recognizes esters containing bulky alcohols (here, the hydroxyl-modified hybrid agent itself) in combination with relatively smaller acyl moieties.[13]</p><p>A total of seven ester-protected hybrids were synthesized. Structural diversity in this set of compounds was achieved by varying the chain length, n, and the nature of the acyl residue, R′ (Scheme 1). In addition, different non-leaving amines (L) were introduced to tune the reactivity of the platinum moiety (see Discussion section below). Compounds 2–8 were generated from the platinum precursors (12a–g) containing the appropriate ester-modified nitrile ligands (11 a–g, for synthetic details see the Supporting Information). The ligand substitution reactions leading to the precursor complexes containing short-chain nitrile esters (n =1, 12a–d) produced significantly lower yields than reactions performed with the long-chain derivatives (n =3, 12e–g) (20–30 vs 80–90 %). This previously observed effect can be attributed to the high CH acidity and reactivity of the nitrile ligands in the former set of derivatives.[12] (Attempts to generate analogous derivatives with n =2 failed due to unexpected β-elimination of the hydroxyl group; data not shown.) The final step affording hybrid agents 2–8 (yield >75 %, analytical purity >95 %) involved addition of the NHMe group in N-(acridin-9-yl)-N′-methylethane-1,2-diamine (13) across the metal-activated nitrile triple bond, [14] producing the desired amidine linkage, and subsequent protonation to generate the dinitrate salts.</p><p>An anomaly in the stereochemistry of the addition reactions leading to hybrids 2–5 was observed. The extended-chain ester derivatives (6, 7, and 8) and the prototype (1′) almost exclusively (>90 %) exist as E isomers in which the platinum moiety and amidine-NMe group adopt a trans configuration with respect to the N(imino)=C double bond, as typically observed in amidine ligands formed from secondary amines.[15] By contrast, hybrids 2–5 form a relatively high amount of the Z isomer (>25 %, based on 1 D and 2 D NMR analysis, see the Supporting Information). We attribute this outcome to the increased steric hindrance produced by the short-chain (n =1) acyl groups around the nitrile group. Intramolecular hydrogen bonding between the imino proton and the ester group (NH···O=C–O) may also contribute to this effect (Supporting Information).[16]</p><p>All newly synthesized hybrids maintain excellent solubility of >10 mg mL−1 in relevant aqueous media. To demonstrate the effect of the pendant ester groups on the hydrophilicity/lipophilicity balance of the compounds, we studied the partitioning of selected derivatives between octanol and phosphate-buffered saline (PBS) (expressed as log[octanol]/[PBS] =logD, the distribution coefficient for protonable pharmacophores[3]). The experiment was performed in PBS at pH 7.4 rather than water to suppress complex aquation and platinum-mediated ester hydrolysis, which is described in the following section. This setup also takes into consideration the pH dependence of logD to faithfully mimic conditions in plasma. For the unmodified, hydrophilic agent, 1‴, a logD of −0.98 (±0.19) was determined. By contrast, compound 8, the presumably most lipophilic derivative (n =3, valproic ester, L =pn), preferentially partitions into the octanol phase with a logD of 0.73 (±0.06), which reflects an increase in lipophilicity by 50-fold relative to compound 1′. An intermediate logD of −0.31 (±0.06) was determined for compound 7 (n =3, L =en, butyric ester). The logD value generated for this compound has to be interpreted with caution because of minor ester hydrolysis, which was unavoidable under the conditions of the experiment (<10 %, see the following section).</p><!><p>One of the proposed mechanisms of activation of compounds 2–8 as prodrugs involves platinum-promoted ester cleavage. To mimic the chloride ion concentration differential that exists between serum during circulation and after uptake into target cells, compounds were incubated at 37 °C in PBS (≈150 mM NaCl, pH 7.4) and in phosphate buffer (PB, pH 7.4), respectively. The reaction mixtures were analyzed at appropriate time points by in-line high-performance liquid chromatography-electrospray mass spectrometry (LC-ESMS). Reaction products were identified as 1+ or 2+ charged molecular ions in mass spectra recorded in positive-ion mode and quantified by integrating UV-visible HPLC traces at an acridine-specific wavelength (see the Supporting Information for a complete set of data).</p><p>The time course of the ester hydrolysis yielding hydroxyl-modified platinum–acridine and butyric/valproic acid is summarized in Figure 2 for both media. Generally, in sets of analogues characterized by common spacers linking the platinum and ester moieties, (CH2)n, the valproic esters show significantly slower conversion than the butyric esters, or no conversion at all (3–5 vs 2, and 6, 8 vs 7). In phosphate-buffered solution in the absence of chloride (Figure 2A), the most extensive hydrolysis is observed for butyric ester-based compounds 2 (n =1) and 7 (n =3), with the former producing approximately twofold higher levels of cleaved product after 36 h of continuous incubation. Hydrolytic activity is also observed for the valproic ester derivatives 3, 4, and 5 (all n =1), but at a much slower rate. Most strikingly, hybrids 6 and 8, which contain the same secondary acyl moiety but on an extended linker (n =3), are completely resistant to cleavage under these conditions. When incubations were performed in buffer supplemented with physiological chloride, a major reduction in ester hydrolysis of up to 75 % was observed (Figure 2B) compared to reactions in chloride-free media, consistent with the notion that (reversible) aquation of the platinum moiety plays a role in the cleavage mechanism.</p><p>The LC-ESMS profiles of compounds 2–5 share common features and support the proposed mechanism of platinum-mediated ester cleavage. We have chosen compound 2, which was also suitable for a kinetic study by 1H NMR spectroscopy, for a detailed discussion (for complete sets of LC-ESMS data for all other analogues, see the Supporting Information). The only hydrolysis product formed in incubations of compound 2 in PB was identified as a chelate in which the chloro leaving group has been replaced with the unprotected hydroxyl oxygen of the cleaved butyric ester (see peak labeled "a" in Figure 3A and the corresponding mass spectrum/structure in Figure 3D/ E, respectively). A minor amount of chemically unchanged 2 and a product resulting from substitution of chloride by phosphate containing intact ester are also observed after 36 h of incubation (peaks b and c). The same array of products was observed when compound 2 was incubated in the presence of sodium chloride in PBS, but ester cleavage and chelate formation were significantly suppressed based on the relative abundances of each species (Figure 3B). The absence of an opened-chelate form in this high-chloride environment attests to the inertness of the five-membered amidine-N/hydroxyl-O chelate. This was also confirmed in incubations with biological nitrogen and sulfur model nucleophiles, in which this product was completely unreactive (N-acetylcysteine, 2′-deoxyguanosine; data not shown). Unlike compounds 2–5 (n =1), compound 7 (n = 3) exclusively forms hydrolysis products containing a dangling hydroxyl group, confirming that a seven-membered, presumably less stable, chelate does not form (Supporting Information).</p><p>The dramatic effect of chloride ion on the kinetics of ester hydrolysis was confirmed for compound 2 in arrayed 1H NMR experiments (for an analysis of 1H NMR spectra see the Supporting Information). Cleavage of the butyric ester follows a (pseudo-) first-order rate law with rate constants of k =3.9 ± 10−5 s−1 in PB and k =1.2 ±10−5 s−1 in PBS, which corresponds to half-lives of 5 and 16 h, respectively (Figure 4). Thus, chloride slows cleavage of the pendant ester in complex 2 by approximately 70 %.</p><p>On the basis of the above product analysis, a mechanism of platinum-mediated, intramolecular ester cleavage is proposed (Scheme 2). Cleavage is triggered by (reversible) aquation of the platinum complex. The Lewis-acidic metal assists in the de-protonation of the aqua to a hydroxo ligand, which undergoes a nucleophilic attack on the acyl carbon to promote cleavage of the ester linkage. Cleavage results in the loss of the acyl protecting group as carboxylic acid and in a dangling or chelated alcohol/alkoxide moiety. High chloride concentrations shift the aquation reaction toward the chloro-substituted hybrid, which quenches ester cleavage. An intramolecular attack by platinum-associated hydroxide is also supported by the following observations: 1) Hydrolysis of ester is approximately twice as efficient for complex 4 bearing ammine (NH3) non-leaving groups than for the en-substituted analogue 3. This effect is consistent with the lower pKa value of the aqua ligands in Pt–ammine than in Pt–en complexes, [17] which produces higher concentrations of the more reactive hydroxo ligand. 2) For derivatives containing the same acyl moiety, ester cleavage is dramatically reduced for n =3 versus n =1. This effect can be rationalized with the fact that internal nucleophilic attack by the hydroxo ligand in the former compound requires formation of a 9-membered, thermodynamically less favorable, macrocyclic intermediate.</p><!><p>To determine if the compounds, especially those not undergoing chemical ester cleavage (6 and 8), are susceptible to deesterification by a pharmacologically relevant prodrug-converting enzyme, we also performed incubations with recombinant human carboxylesterse-2 (hCES-2). The particular engineered form of the protein used was sufficiently robust to provide enzymatic activity over long reaction times (>30 min), which allows for identification of cleavage products for non-classical, slowly metabolized substrates.[18] Reactions were performed in a buffer supplemented with chloride to suppress non-enzymatic hydrolysis and monitored for a relatively short period of time to minimize the effects of loss of enzyme activity over time and chemical hydrolysis. Unavoidable contributions from the latter pathway have been subtracted, where possible, from the data presented in Figure 2C. LC-ESMS analysis of the reaction mixtures shows minimal or no enzyme-mediated cleavage of ester for compounds 2–5, respectively, which contain short spacers (n =1). By contrast, derivatives 6–8 (n =3) show deesterification under these conditions with yields of ≈20 % at the 10 h time point. Importantly, the bulky valproic esters in compounds 6 and 8, which are completely resistant to chemical hydrolysis, are efficiently cleaved by the enzyme. The HPLC trace recorded for compound 2 after 10 h of incubation with enzyme (Figure 3C) shows the same reaction products formed in PB and PBS, as well as an additional peak (d, ≈5 %) not observed for chemical hydrolysis. Positive-ion mass spectra unequivocally identified this product as the platinum-chloro complex containing a dangling hydroxyl group (Figure 3D, E), consistent with a non-metal-mediated cleavage mechanism.</p><p>The fact that no opened chelate was detected for chemical hydrolysis of derivatives with n =1, but only five-membered N,O-chelate, can be taken as evidence that the former must be an enzyme-specific product. The synthesis of cisplatin derivatives containing chelated aminoalcoholato ligands requires basic conditions to help deprotonate the alcohol, while opening of such chelates only occurs under acidic conditions.[19] This is in agreement with the observation that the two forms (peaks a and d, Figure 3C) do not inter-convert under the conditions and on the time scale of the assay. Unlike the butyric ester derivative 7, which shows dual cleavage reactivity resulting in the dangling hydroxyl form of the hybrid agent, the valproic ester derivatives 6 and 8 only undergo enzyme-mediated cleavage to produce the corresponding deesterified form (shown for compound 6 in Figure 5). It can be concluded that only esters installed on an extended linker are viable substrates for hCES-2 and that only the chemically robust valproic ester confers true selectivity for enzymatic cleavage.</p><!><p>The anticancer properties of the newly generated compounds 2–8 were assessed using a colorimetric cell proliferation assay and compared to that of prototype 1′ (chloride salt). (Note: attempts to synthesize the deesterified, proposed active hydroxyl forms of 2–8 in pure form for screening in this assay have been unsuccessful due to the chemical instability of the platinum-nitrile precursor complexes.) Two NSCLC adenocarcinoma cell lines were screened for chemosensitivity: A549, which responds well to treatment with platinum–acridines, and NCI-H1435, a DNA repair-proficient and highly chemoresistant form of this cancer.[20] The latter cell line has also been shown to express significantly higher levels (three–fourfold) of hCES-2 enzyme than A549 and other lung adenocarcinomas.[21] Thus, cytotoxicity levels observed in the two cell lines may also provide hints about potential intracellular prodrug activation. Both cell lines were dosed with hybrid agents for 72 h at three selected concentrations, which were based on relative chemo-sensitivities established for compound 1′. A549 was dosed at 5.0, 50, and 500 nM, whereas 20-fold higher concentrations, 0.1, 1.0, and 10 μM, were chosen for NCI-H1435 to account for the relatively higher resistance observed in the latter cell line. The dose-response data resulting from the chemosensitivity screen are presented in Figure 6.</p><p>As a general trend, the ester-modified derivatives show reduced cytotoxicity levels compared to the unmodified hybrid, 1′, except for compound 7, which shows a response similar to that of the prototype in both cell lines. Because the butyrate-protected derivative is efficiently cleaved both chemically and enzymatically on a time scale relevant to the cell culture assay (see previous section), it is proposed that the hydroxyl form of this agent is the major contributor to the cell kill. It also suggests that the nPr3-OH residue in deesterified 7 in place of the Et residue in 1′ does not compromise the potency of the pharmacophore, possibly pointing to a similar mechanism at the DNA level. By contrast, the derivatives containing bulky valproic ester show greatly reduced activity. Compound 6, for instance, which would generate the same active hydroxyl form as compound 7, and compound 8 show no cell kill in A549 at the highest dose and significantly reduced activity in NCI-H1435. This observation is in agreement with the less efficient intracellular cleavage of the bulkier valproic esters 6 and 8, which, in their intact form, are less cytotoxic, possibly due to their reduced reactivity with cellular DNA. The same trend is observed for compounds 3–5, which proved to be relatively resistant to chemical and enzymatic hydrolysis. For compounds sharing the same linkers and ester moieties, the nature of the non-leaving group had only minor or no effects on the activity. Finally, compound 2, although cleaved very efficiently in chloride-free media (see previous section), showed activity inferior to the prototype. This outcome was expected because of the inertness of the 5-membered chelate generated in the process.</p><p>Compounds 6 and 8, which are cleaved by hCES-2, show a more pronounced enhancement in activity relative to compound 1′ in NCI-H1435 (high hCES-2) than in A549 (low hCES-2). The opposite is the case for compound 7. As a preliminary measure of chemo-selectivity, we defined the selectivity index, S: S = [CVi,NCI-H1435/CV1′,NCI-H1435]/[CVi,A549/CV1′,A549], where CVi and CV1′ are the % viabilities of the ester-based compounds and compound 1′ for their highest doses, respectively. CV1′ was introduced to normalize for differences in chemosensitivities between the two cell lines. Assuming that CV1′ is the highest cell kill that can be achieved with any of the cleaved esters, S values smaller than "1" would indicate a relative advantage of a given derivative in NCI-H1435, and vice versa. For compounds 6, 7, and 8, S values of 0.6, 1.4, and 0.5 were calculated, respectively. It can be concluded that the esters that are not cleaved chemically (6, 8) but show hCES-2-mediated cleavage perform relatively better in NCI-H1435. Conversely, compound 7, which efficiently converts to its active form via an enzyme-independent pathway, appears to have a relative advantage in A549. These findings may indicate that compounds 6 and 8 act as prodrugs at high levels of hCES-2, which confers sensitivity to NCI-H1435 cells. It should be noted, however, that a direct comparison of cytotoxic responses between two different cell lines has to be interpreted with caution. Because cellular uptake and efflux, among other factors, may also contribute to the observed differences, additional experiments in hCES-2 knockdown or hCES-2 transfected cells[11] of the same type are necessary.</p><!><p>Lipophilic, ester-modified derivatives of platinum–acridines have been synthesized and evaluated with the goal of improving the drug-like properties of these highly cytotoxic anticancer agents. On the basis of their physicochemical properties, chemical reactivities, their ability to serve as hCES-2 substrates, and their performance in cell lines, compounds 6 and 8 can be considered candidates for slow cleavage by hCES-2. Compound 7 holds promise of providing a dual mode of cleavage, in which chemical activation may be necessary for applications requiring extended circulation of precursor in plasma and self-immolative activation in cells lacking hCES-2. By contrast, the more lipophilic and chemically less reactive analogues 6 and 8, which should achieve even longer half-lives in circulation, would require activation by target enzyme expressed in the liver (similar to the anticancer prodrug irinotecan) or in tumor tissue. Whether the butyric and valproic esters are prone to hydrolysis by other ubiquitous serum esterases remains to be determined. To this end, incubations of 2–8 with fetal bovine serum (FBS, Thermo Scientific HyClone), which is commonly used as a model for human serum, [23] have shown no evidence for this unwanted reactivity (data not reported).</p><p>Another potential advantage of bulky prodrugs of platinum–acridines may be their poor DNA-binding properties. If ester cleavage occurs primarily intracellularly in tumor tissue, differential DNA recognition by the ester-protected (inactive) and hydroxo (active) forms of the hybrid agent might confer pro-drug selectivity to cancer cells while sparing normal cells. In addition to an improved ADME profile, lipophilic prodrugs also have the potential benefit of improving drug safety.[1] To test if this supposition holds for our prodrug design, dose escalation studies were performed with the chemically robust derivative 8 in Swiss Webster mice. Mice tolerated this analogue without showing signs of toxicity and weight loss when injected intra-peritoneally (i.p.) once a day for five consecutive days (qd × 5) at a dose of 1.6 mg kg−1 (Supporting Information). For comparison, compound 1′ was significantly more toxic and showed an MTD of 0.1 mg kg−1 when the same dosing schedule was applied.[5]</p><p>In conclusion, we have developed a versatile platform for tuning the pharmacological parameters of potent platinum–acridines and have demonstrated for the first time that a metal-based pharmacophore is compatible with the classical concept of carboxylesterase-mediated prodrug activation. This feature may provide a strategy of improving the ADME and safety of platinum–acridines and other metallopharmacophores.</p><!><p>All reagents were used as obtained from commercial sources without further purification unless indicated otherwise. Compound 1′ [5] (chloride salt) and N-(acridin-9-yl)-N′-methylethane-1,2-diamine[22] (13) were prepared according to published procedures. The platinum-nitrile precursors (12 a–g) were synthesized from the corresponding nitriles (11a–d) and silver-ion activated diam(m)inedichloridoplatinum(II) complexes (Supporting Information). 1H NMR spectra of the target compounds and intermediates were recorded on Bruker Advance DRX-500 and 300 MHz instruments. Proton-decoupled 13C NMR spectra were recorded on a Bruker DRX-500 instrument operating at 125.8 MHz. 2 D 1H-13C gradient-selected Heteronuclear Multiple Bond Coherence (gsHMBC) experiments and temperature-dependent spectra were acquired on a Bruker DRX-500 instrument equipped with a TBI probe and a variable-temperature unit. 2 D HMBC spectra were collected with 2048 pts in t2(sw =6510 Hz), 256 pts in t1 (sw = 27 670 Hz), 128 scans per t1 increment, and a recycle delay (d1) of 1.5 s. Chemical shifts (δ) are given in parts per million (ppm) relative to internal standard tetramethylsilane (TMS). 1H NMR data is reported in the conventional form including chemical shift (δ, ppm), multiplicity (s =singlet, d =doublet, t =triplet, q =quartet, m =multiplet, br =broad), coupling constants (Hz), and signal integrations. 13C NMR data are reported as chemical shift listings (δ, ppm). The NMR spectra were processed and analyzed using the MestReNova software package. HPLC-grade solvents were used for all HPLC and mass spectrometry experiments. LC-ESMS analysis was performed on an Agilent 1100LC/MSD ion trap mass spectrometer equipped with an atmospheric pressure electrospray ionization source. Eluent nebulization was achieved with a N2 pressure of 50 psi and solvent evaporation was assisted by a flow of N2 drying gas (350 °C). Positive-ion mass spectra were recorded with a capillary voltage of 2800 V and a mass-to-charge scan range of 150 to 2200 m/z. To establish the purity of target compounds, samples were diluted in methanol containing 0.1 % formic acid and separated using a 4.6 mm × 150 mm reverse-phase Agilent ZORBAX SB-C18 (5 mm) analytical column at 25 °C, by using the following solvent system: solvent A, optima water, and solvent B, methanol/0.1 % formic acid, with a flow rate of 0.5 mL min−1 and a gradient of 95 to 5 % A over 30 min. HPLC traces were recorded with a monitoring wavelength range of 363–463 nm. Peak integration was done using the LC/ MSD Trap Control 4.0 data analysis software. Analytical purity of greater than 95 % was confirmed this way for all target compounds prior to analytical and biological experiments (for analytical data and details of the NMR spectral assignments, including atom numbering schemes, see the Supporting Information).</p><!><p>Precursor complex 12a (240 mg, 0.5 mmol) was dissolved in 10 mL of anhydrous DMF and the solution was cooled to −20 °C. Acridine-amine 13 (138 mg, 0.55 mmol) was added to the solution, and the suspension was stirred at 4 °C for 24 h. The reaction mixture was added dropwise into 200 mL of anhydrous ethyl ether, and the resulting yellow slurry was vigorously stirred for 30 min. The precipitate was recovered by membrane filtration, dried in a vacuum overnight, and dissolved in 20 mL of methanol containing 1 equivalent of HNO3. After removal of the solvent by rotary evaporation, the crude product was recrystallized from hot ethanol, affording 2 as a 3:1 mixture of E and Z isomers. Yield: 238 mg (71 %). 1H NMR (300 MHz, [D4]MeOH) δ=8.61–8.51 (m, 2 H, acridine-H1/8), 8.13–7.75 (m, 4 H, acridine-H3–6), 7.75–7.45 (m, 2 H, acridine-H2/7), 6.87–6.57 (m, 1 H, Pt-NH=CCH2O-), 5.50 (s, 1.5 H, Pt-NH= CCH2O-, E-isomer), 5.49–5.39 (m, 4 H, -NH2CH2CH2NH2-), 4.76 (s, 0.5 H, Pt-NH=CCH2O-, Z-isomer), 4.59 (m, 0.5 H, -NHCH2CH2N(CH3)-, Z-isomer), 4.45 (t, J =6.3 Hz, 1.5 H, -NHCH2CH2N(CH3)-, E-isomer), 4.02 (t, J =6.3 Hz, 1.5 H, -NHCH2CH2N(CH3)-, E-isomer), 3.56–3.39 (m, 0.5 H, -NHCH2CH2N(CH3), Z-isomer), 3.14 (s, 2.25 H, -NHCH2CH2N-(CH3)-, E-isomer), 2.61–2.57(m, 4 H, -NH2CH2CH2NH2-), 2.38–2.06 (m, 2 H, -COCH2CH2CH3), 1.76–1.40 (m, 2 H, -COCH2CH2CH3), 1.02–0.75 ppm (m, 3 H, -COCH2CH2CH3); 13C NMR (75 MHz, [D4]MeOH) δ=172.50 (-COCH2CH2CH3, E-isomer), 171.76 (-COCH2CH2CH3, Z-isomer), 165.14 (Pt-NH=CCH2O-, Z-isomer), 163.38 (Pt-NH=CCH2O-, E-isomer), 158.48 (acridine-C9), 139.87 (acridine-11C/13C), 135.21(acridine-C3/6), 124.99 (acridine-C1/8), 124.02 (acridine-C2/7), 118.28 (acridine-C4/5), 112.63 (acridine-C10/12), 63.55 (Pt-NH=CCH2O-, E-isomer), 61.87 (Pt-NH=CCH2O-, Z-isomer), 35.03 (-COCH2CH2CH3), 17.69 (-COCH2CH2CH3), 12.43 ppm (-COCH2CH2CH3); MS (ESI, positive-ion mode): m/z: calcd for C24H35ClN6O2Pt [M–H] +: 669.22; found: 669.3.</p><!><p>Compound 3 was prepared according to the procedure described for 2 from precursor 12 b with a yield of 69 %. 1H NMR (300 MHz, [D4]MeOH) δ = 8.82–8.40 (m, 2 H, acridine-H1/8), 8.11–7.78 (m, 4H, acridine-H3–6), 7.77–7.50 (m, 2 H, acridine-H2/7), 5.52 (s, 1.5 H, Pt-NH=CCH2O-, E-isomer), 5.42–5.13 (m, 4 H, -NH2CH2CH2NH2-), 4.64–4.54 (m, 0.5 H, NHCH2CH2N(CH3)-, Z-isomer), 4.48 (t, J = 6.5 Hz, 1.5 H, NHCH2CH2N(CH3)-, E-isomer), 4.03 (t, J = 6.4 Hz, 1.5 H, -NHCH2CH2N-(CH3)-, E-isomer), 3.15 (s, 2.25 H, -NHCH2CH2N(CH3)-, E-isomer), 2.75–2.47 (m, 4 H, -NH2CH2CH2CH2NH2-), 2.44–2.11 (m, 1 H, -COCH(CH2CH2CH3)2), 1.54–1.13 (m, 8 H, -COCH(CH2CH2CH3)2), 0.97–0.71 ppm (m, 6 H, -COCH(CH2CH2CH3)2); 13C NMR (75 MHz, [D4]MeOH) δ=176.65 (-CO(CH2CH2CH3)2, E-isomer), 176.01 (-COCH(CH2CH2CH3)2, Z-isomer), 167.30 (Pt-NH=CCH2O-, Z-isomer), 164.85 (Pt-NH=CCH2O-, E-isomer), 159.90 (acridine-C9), 141.38 (acridine-C11/13), 136.77 (acridine-C3/6), 126.49 (acridine-C1/8), 125.61 (acridine-C2/7), 119.88 (acridine-C4/5), 114.12 (acridine-C10/12), 66.90 (Pt-NH=CCH2O-, E-isomer), 65.27 (Pt-NH=CCH2O-, Z-isomer), 49.87 (-NH2CH2CH2NH2-), 48.16 (-NHCH2CH2N(CH3)-), 46.35 (-COCH(CH2CH2CH3)2), 35.48 (-COCH(CH2CH2CH3)2), 21.58 (-COCH(CH2CH2CH3)2), 14.31 ppm (-COCH(CH2CH2CH3)2); MS (ESI, positive-ion mode): m/z: calcd for C28H43ClN6O2Pt [M–H] +: 724.28; found: 724.4.</p><!><p>Compound 4 was prepared according to the procedure described for 2 from precursor 12 c with a yield of 74 %. 1H NMR (300 MHz, [D4]MeOH) δ = 8.56 (d, J = 8.7, 2 H, acridine-H1/8), 8.03 (ddd, J = 9.2, 5.8, 1.8 Hz, 2 H, acridine-H3/6), 7.95–7.79 (m, 2 H, acridine-H4/5), 7.75–7.51 (m, 2 H, acridine-H2/7), 5.56 (s, 1.5 H, Pt-NH=CCH2O-, E-isomer), 4.67–4.54 (s, 0.5 H, -NHCH2CH2N(CH3)-, Z-isomer), 4.48 (t, J =6.5 Hz, 1.5 H, -NHCH2CH2N(CH3)-, E-isomer), 4.18 (bs, 3 H, NH3), 4.02 (t, J =6.5 Hz, 1.5 H, NHCH2CH2N(CH3)-, E-isomer), 3.93 (bs, 3 H, NH3), 3.14 (s, 2.25, H-NHCH2CH2N(CH3)-, E-isomer), 2.36–2.16 (m, 1 H, -COCH(CH2CH2CH3)2), 1.56–1.03 (m, 8 H, -COCH(CH2CH2CH3)2), 0.90–0.68 ppm (m, 6 H, -COCH(CH2CH2CH3)2); 13C NMR (75 MHz, [D4]MeOH) δ=176.67 (-CO(CH2CH2CH3)2, E-isomer), 176.07 (-COCH(CH2CH2CH3)2, Z-isomer), 164.74 (Pt-NH=CCH2O-, E-isomer), 159.89 (acridine-C9), 141.39 (acridine-C11/13), 136.76 (acridine-C3/6), 126.77 (acridine-C1/8), 125.60 (acridine-C2/7), 119.88 (acridine-C4/5), 114.20 (acridine-C10/12), 65.32 (Pt-NH=CCH2O-, E-isomer), 46.14 (-COCH(CH2CH2CH3)2), 35.49 (-COCH(CH2CH2CH3)2) 21.58 (COCH(CH2CH2CH3)2), 14.29 ppm (-COCH(CH2CH2CH3)2); MS (ESI, positive-ion mode): m/z: calcd for C26H41ClN6O2Pt [M–H] +: 699.26; found: 699.3.</p><!><p>Compound 5 was prepared according to the procedure described for 2 from precursor 12 d with the yield of 77 %. 1H NMR (300 MHz, [D4]MeOH) δ=8.61 (dd, J =8.8, 4.5 Hz, 2 H, acridine-H1/8), 8.14–7.95 (m, 2 H, acridine-H3/6), 7.87 (dd, J =9.0, 4.2 Hz, 2 H, acridine-H4/5), 7.75–7.53 (m, 2 H, acridine-H2/7), 6.83–6.73 (m, 1 H, Pt-NH= CCH2O-), 5.53 (s, 1.5 H, Pt-NH=CCH2O-, E-isomer), 5.29–4.88 (m, 4 H, -NH2CH2CH2CH2NH2-), 4.68–4.58 (m, 0.5 H, -NHCH2CH2N(CH3), Z-isomer), 4.55 (t, J =6.6 Hz, 1.5 H, -NHCH2CH2N(CH3)-, E-isomer), 4.03 (t, J =6.4 Hz, 1.5 H, -NHCH2CH2N(CH3)-, E-isomer), 3.68–3.44 (m, 0.5 H, m, 0.5 H, - NHCH2CH2N(CH3)-, Z-isomer), 3.15 (s, 2.25 H, -NHCH2CH2N(CH3)-, E-isomer), 2.95–2.52 (m, 4 H, -NH2CH2CH2CH2NH2-), 2.41–2.07 (m, 1 H, -COCH(CH2CH2CH3)2), 1.92–1.69 (m, 2 H, -NH2CH2CH2CH2NH2-), 1.59–0.98 (m, 8 H, -COCH(CH2CH2CH3)2), 0.98–0.64 ppm (m, 6 H, -COCH(CH2CH2CH3)2); 13C NMR (75 MHz, [D4]MeOH) δ=176.68 (-CO(CH2CH2CH3)2, E-isomer), 176.22 (-COCH(CH2CH2CH3)2, Z-isomer), 164.89 (Pt-NH= CCH2O-, E-isomer), 159.90 (acridine-C9), 141.41 (acridine-C11/13), 136.76 (acridine-C3/6), 126.56 (acridine-C1/8), 125.60 (acridine-C2/7), 119.88 (acridine-C4/5), 113.48 (acridine-C10/12), 46.29 (-COCH(CH2CH2CH3)2), 44.42 (-NH2CH2CH2CH2NH2-), 43.63 (-NH2CH2CH2CH2NH2-), 35.54 (-COCH(CH2CH2CH3)2), 29.39 (-NH2CH2CH2CH2NH2-), 21.60 (-COCH(CH2CH2CH3)2), 14.31 ppm (-COCH(CH2CH2CH3)2); MS (ESI, positive-ion mode): m/z : calcd for C29H45ClN6O2Pt [M–H] +: 739.29; found: 739.4.</p><!><p>Compound 6 was prepared according to the procedure described for 2 from precursor 12 e with the yield of 83 %. 1H NMR (300 MHz, [D4]MeOH) δ = 8.46 (dd, J = 8.5, 4.3 Hz, 2 H, acridine-H1/8), 7.91 (m, 2 H, acridine-H3/6), 7.77 (dd, J =9.0, 4.2 Hz, 2 H, acridine-H4/5), 7. 54 (m, 2 H, acridine-H2/7), 6.08 (s, 1 H, Pt-NH=CCH2CH2CH2O-), 5.36–5.11 (m, 4 H, -NH2CH2CH2NH2-), 4.33 (t, J = 6.6 Hz, 2 H, -NHCH2CH2N-(CH3)-), 4.03 (t, J = 6.4 Hz, 2 H, -NHCH2CH2N(CH3)-), 3.88 (m, 2 H, Pt-NH=CCH2CH2CH2O-), 3.03 (s, 3 H, -NHCH2CH2N(CH3)-), 2.97 (m, 2 H, Pt-NH=CCH2CH2CH2O-), 2.55–2.37 (m, 4 H, -NH2CH2CH2NH2-), 2.10–2.03 (m, 3 H, Pt-NH=CCH2CH2CH2O-,-COCH(CH2CH2CH3)2), 1.59–0.98 (m, 8 H, -COCH(CH2CH2CH3)2), 0.98–0.64 ppm (m, 6 H, -COCH(CH2CH2CH3)2); 13C NMR (75 MHz, [D4]MeOH) δ=178.08 (-CO(CH2CH2CH3)2), 170.02 (Pt-NH=CCH2O-), 159.98 (acridine-C9), 141.37 (acridine-C11/13), 136.71 (acridine-C3/6), 126.52 (acridine-C1/8), 125.53 (acridine-C2/7), 119.85 (acridine-C4/5), 114.10 (acridine-C10/12), 64.83 (Pt-NH=CCH2CH2CH2O-), 46.53 (-COCH(CH2CH2CH3)2), 35.81 (-COCH(CH2CH2CH3)2), 29.39 (-NH2CH2CH2CH2NH2-), 27.15 (Pt-NH=CCH2CH2CH2O-), 21.67 (-COCH(CH2CH2CH3)2), 14.34 ppm (-COCH(CH2CH2CH3)2); MS (ESI, positive-ion mode): m/z calcd for C30H47ClN6O2Pt [M–H] +: 752.31; found: 752.4.</p><!><p>Compound 7 was prepared according to the procedure described for 2 from precursor 12 f with the yield of 89 %. 1H NMR (300 MHz, [D4]MeOH) δ = 8.72–8.46 (m, 2 H, acridine-1/8), 8.01 (ddd, J = 8.1, 6.9, 1.0 Hz, 2 H, acridine-H3/6), 7.92–7.80 (m, 2 H, acridine-H4/5), 7.64 (ddd, J =8.3, 6.9, 1.2 Hz, 2 H, acridine-H2/7), 6.16 (s, 1 H, Pt-NH=CCH2CH2CH2O-), 5.44–5.22 (m, 4 H,-NH2CH2CH2NH2-), 4.42 (t, J = 6.5 Hz, 2 H, -NHCH2CH2N(CH3)-), 4.18 (t, J = 6.5 Hz, 2 H, -NHCH2CH2N-(CH3)-), 3.98 (t, J = 6.5 Hz, 2 H, Pt-NH=CCH2CH2CH2O-), 3.12–3.02 (m, 5 H, Pt-NH=CCH2CH2CH2O-,-NHCH2CH2N(CH3)-), 2.74–2.46 (m, 4 H, -NH2CH2CH2NH2-), 2.34–1.99 (m, 4 H, Pt-NH=CCH2CH2CH2O-, -COCH2CH2CH3), 1.58 (d, J = 7.3 Hz, 2 H, -COCH2CH2CH3), 0.90 ppm (t, J =7.4 Hz, 3 H, -COCH2CH2CH3); 13C NMR (75 MHz, [D4]MeOH) δ = 175.33 (-COCH2CH2CH3), 170.11 (Pt-NH=CCH2CH2CH2O-), 160.04 (acridine-C9), 141.38 (acridine-C11/13), 136.69 (acridine-C3/6), 126.43 (acridine-C1/8), 125.50 (acridine-C2/7), 119.87 (acridine-C4/5), 114.12 (acridine-C10/12), 64.7 (Pt-NH=CCH2CH2CH2O-), 36.84 (-COCH2CH2CH3), 32.07 (Pt-NH=CCH2CH2CH2O-), 27.10 (Pt-NH= CCH2CH2CH2O-), 19.37 (-COCH2CH2CH3), 13.94 ppm (-COCH2CH2CH3); MS (ESI, positive-ion mode): m/z: calcd for C26H39ClN6O2Pt [M–H] +: 697.25; found: 697.3.</p><!><p>Complex 8 was prepared according to the procedure described for 2 from precursor 12 g with the yield of 76 %. 1H NMR (300 MHz, [D4]MeOH) δ = 8.54 (d, J = 8.6 Hz, 2 H, acridine-H1/8), 8.10–7.76 (m, 4 H, acridine-H3–6), 7.59 (t, J =7.4 Hz, 2 H, acridine-H2/7), 4.40 (t, J = 6.1 Hz, 2 H, -NHCH2CH2N(CH3)-), 4.17 (t, J = 6.2 Hz, 2 H, -NHCH2CH2N-(CH3)-), 3.97 (t, J = 6.0 Hz, 2 H, Pt-NH=CCH2CH2CH2O-), 3.21–2.98 (m, 5 H, Pt-NH=CCH2CH2CH2O-,-NHCH2CH2N(CH3)-), 2.94–2.54 (m, 4 H, -NH2CH2CH2CH2NH2-), 2.40–1.95 (m, 3 H, Pt-NH=CCH2CH2CH2O-,-COCH(CH2CH2CH3)2), 1.79 (m, 2 H, -NH2CH2CH2CH2NH2-), 1.56–0.99 (m, 8 H, -COCH(CH2CH2CH3)2), 0.92–0.68 ppm (m, 6 H, -COCH(CH2CH2CH3)2); 13C NMR (75 MHz, [D4]MeOH) δ=178.02 (-CO(CH2CH2CH3)2), 169.87 (Pt-NH=CCH2CH2CH2O-), 159.85 (acridine-C9), 141.33 (acridine-C11/13), 136.67 (acridine-C3/6), 126.61 (acridine-C1/8), 125.50 (acridine-C2/7), 119.87 (acridine-C4/5), 114.13 (acridine-C10/12), 64.86 (Pt-NH=CCH2CH2CH2O-), 46.56 (-COCH(CH2CH2CH3)2), 44.31 (-NH2CH2CH2CH2NH2-), 43.63 (-NH2CH2CH2CH2NH2-), 35.83 (-COCH(CH2CH2CH3)2), 29.29 (-NH2CH2CH2CH2NH2-), 26.76 (Pt-NH=CCH2CH2CH2O-), 21.69 (-COCH(CH2CH2CH3)2), 14.37 ppm (-COCH(CH2CH2CH3)2); MS (ESI, positive-ion mode): m/z: calcd for C31H49ClN6O2Pt [M–H] +: 767.33; found: 767.4.</p><!><p>NMR spectra in arrayed experiments were collected at 37 °C on a Bruker 500 DRX spectrometer equipped with a triple-resonance broadband inverse probe and a variable temperature unit. Reactions were performed with 2 mM platinum complex dissolved in either 600 μL of 10 mM phosphate buffer (PB, D2O, pH* = 7.0) or in 600 mL of phosphate-buffered saline (1 × PBS, D2O, pH* = 7.0). The 1 D NMR kinetics experiments were carried out as a standard Bruker arrayed 2 D experiment using a variable-delay list. Each incremented spectrum was processed using the same procedure, and suitable signals of the ester moiety were integrated. Data were processed with MestReNova NMR software. The concentrations of platinum complex at each time point were deduced from relative peak intensities, averaged over multiple signals to account for differences in proton relaxation, and fitted to a first-order exponential decay function in Origin 8.0 (OriginLab, Northampton, MA).</p><!><p>The ester hydrolysis study of compounds 2–8 was carried out by incubating 1 mM of each test compound in 10 mM phosphate buffer (pH 7.4) or 1 × PBS containing ≈150 mM NaCl at 37 °C. At various time points samples were withdrawn from the reaction mixture and analyzed by in-line LC-ESMS. Chromatographic separations were performed with a 4.6 × 150 mm reverse-phase Agilent ZORBAX SB-C18 (5 μm) analytical column with the column temperature maintained at 25 °C. The following solvent system was used: solvent A, optima water, and solvent B, methanol/0.1 % formic acid, at a flow rate of 0.5 mL min−1 and a gradient of 95 % A to 5 % A over 15 min.</p><!><p>To study ester cleavage in compounds 2–8 by recombinant human carboxylesterase-2 (rhCES-2), 30 μM of each compound was incubated with 400 μg mL−1 hCES-2 (BD Biosciences, San Jose, CA, USA) at 37 °C in 1 × PBS. Aliquots were withdrawn at various time points, quenched in an equal volume of methanol, and centrifuged for 5 min at 10 000 g to denature and precipitate protein. The supernatant was collected and subjected to product separation and analysis using in-line LC-ESMS. Chromatography was performed on a 4.6 × 150 mm reverse-phase Agilent ZORBAX SB-C18 (5 μm) analytical column with the column temperature maintained at 25 °C. The following solvent system was used: solvent A, optima water, and solvent B, methanol/0.1 % formic acid, at a flow rate of 0.5 mL min−1 and a gradient of 95 % A to 5 % A over 15 min.</p><!><p>To obtain octanol-saturated water and water-saturated octanol, 100 mL of PBS was stirred with 100 mL of octanol for 24 h, followed by centrifugation for 5 min. The platinum complexes were dissolved in 1.0 mL of octanol-saturated PBS to a typical concentration of 0.1 mM and then mixed with 1.0 mL water-saturated octanol. Triplicates of each experiment were mixed in a multi-tube vortexer incubator for 16 h at room temperature and then centrifuged for 5 min. The layers were separated carefully, and the content of platinum–acridines was determined spectrophotometrically at 413 nm (with ε413 = 10 000 M −1 cm−1 in octanol-saturated PBS and ε413 = 8600 M −1 cm−1 in PBS-saturated octanol). The partition coefficients (D) of the samples were then determined as the quotient of the concentration of compound in octanol and the concentration in the aqueous layer. Reported logD values are the mean ± standard deviations of three determinations.</p><!><p>The human non-small cell lung cancer cell lines, NCI-H1435 and A549 (adenocarcinomas) were obtained from the American Type Culture Collection (Rockville, MD, USA). A549 cells were cultured in HAM's F12K media (Gibco) supplemented with 10 % fetal bovine serum (FBS), 10 % penstrep (P&S), 10 % L-glutamine, and 1.5 g L−1 NaHCO3. NCI-H1435 cells were cultured in serum-free 1:1 DMEM/ F12 media (Gibco) containing 2.436 g L−1 NaHCO3, 0.02 mg mL−1 insulin, 0.01 mg mL−1 transferrin, 25 nM sodium selenite, 50 nM hydrocortisone, 1 ng mL−1 epidermal growth factor, 0.01 mM ethanol-amine, 0.01 mM phosphorylethanolamine, 100 pM triiodothyronine, 0.5 % (w/v) bovine serum albumin (BSA), 10 mM HEPES, 0.5 mM sodium pyruvate, and an extra 2 mM L-glutamine (final concentration 4.5 mM). Cells were incubated at a constant temperature at 37°C in a humidified atmosphere containing 5 % CO2 and were subcultured every 2–3 days in order to maintain cells in logarithmic growth, except for slowly proliferating NCI-H1435, which was subcultured every seven days.</p><!><p>The cytotoxicity studies were carried out according to a standard protocol using the Celltiter 96 aqueous nonradioactive cell proliferation assay kit (Promega, Madison, WI). Stock solutions (5–10 mM) of 1′–8 were made in DMF and serially diluted with media prior to incubation with cancer cells. All drugs and controls were tested at the indicated concentrations in triplicate wells on duplicate plates. Incubations were carried out for 72 h and cell viabilities were determined by comparing drug-treated wells with control cells.</p>

PubMed Author Manuscript

Response to Comment on "Following molecular mobility during chemical reactions: no evidence for active propulsion" and "Molecular diffusivity of click reaction components: the diffusion enhancement question"

comment on the diffusion NMR measurements of molecules during a copper-catalyzed azide-alkyne cycloaddition (CuAAC) "click" reaction. Here we respond to these comments and maintain that no measurable diffusion enhancement was observed during the reaction. We expand on the physical arguments presented in our earlier JACS paper regarding the appropriate reference state for the diffusion coefficient and present new data showing that the use of other reference states as suggested by Huang et al. will still support our conclusion that the two reactants and one product of click reaction do not exhibit boosted mobility during the reaction.

response_to_comment_on_"following_molecular_mobility_during_chemical_reactions:_no_evidence_for_acti

<!>Binding-unbinding equilibria<!>Figure 2. (a)<!>Flocculation in some NMR experiments

<p>Reaction-induced boosted mobility is an exciting paradigm that has been comprehensively verified for micrometer-sized objects, while its relevance to the nano-and sub-nanometer scale, i.e. for enzymes and small molecules is less clear. There have been theoretical studies, e.g. by one of the authors of this letter, 1-2 which propose that "molecular swimmers" are possible and lay out the conditions under which they could be experimentally observed. In recent years, a number of experimental studies have reported diffusion enhancement in nanometer objects such as single enzymes. [3][4][5] However, these reports have been scrutinized on theoretical and experimental grounds and, accordingly, the measurability of diffusion enhancement in enzymes in the existing experimental setups has been critiqued, [6][7][8][9] most recently in ref. 10. On this background, a recent Science paper from Granick's group reporting pronounced diffusion enhancement for small molecular reactants has drawn the attention of a larger scientific community to the idea of molecular diffusion enhancement. 11 A particular case in their original and follow-up papers is the Cu(I)-catalyzed Azide-Alkyne Cycloaddition (CuAAC) click reaction, for which the components of the reaction are claimed to exhibit diffusion enhancement in NMR diffusion measurements. [11][12][13] These reports have been the subject of a series of critical exchanges, in which the existence of measurable diffusion enhancement for the CuAAC reaction components has been seriously debated. [12][13][14][15][16][17][18] Recently, through carefully-designed NMR diffusion measurements and analyses, including devising two novel post-acquisition NMR analysis methods, we reported that there was no measurable enhanced diffusion of the two reactants (alkyne and azide) and single product (triazole) of the CuAAC click reaction, and the observed alterations in their diffusion coefficients (Deff) pointed to the role of relatively large reaction intermediates diffusing more slowly than both the reactants and the product. 19 In their reaction to our article, Huang and Granick present a number of criticisms, 20 which we address below:</p><p>1. The choice of reference state for reactants In our article, we used the diffusion coefficient of each reactant in D2O in the absence of the second reactant, co-catalyst (sodium ascorbate) and catalyst (copper sulfate) as the reference diffusion coefficient (D0). 19 We did not use the limiting diffusion coefficient (D∞) of reactants towards the end of reaction as D0, because considering the known mechanism of click reaction and the formation of various reaction intermediates, we deemed it not to be a physically appropriate reference state, as also previously shown in 16 . To avoid any further complication arising from the known coordination of alkyne or azide with copper ions or their possible ascorbatecatalyzed redox reactions, we did not use any mixture as the reference state either. Our choice of reference state was criticized by Huang and Granick for being "artificial" and not "physically meaningful". The reference state used by us (reactant alone) actually corresponds to the reference state that Wang et al. 11 misleadingly claimed to have used in their original work, in which (on page 1) they defined their measured values ΔDapp/D0 as "the relative diffusion increase over the Brownian diffusion coefficient of the same molecules". Any reader would almost certainly have understood this definition as implying that D0 corresponds to the diffusion coefficient of the molecule under consideration in isolation. The true meaning of D0 as the measured diffusion of the corresponding signal at the end of the reaction was only made clear in their subsequent publications. [12][13] Regarding this, we find rather disingenuous to use the words "boosted" or "increased" when referring to cases in which the diffusivity is not being compared to a past state, but rather to a future state, particularly in cases in which the diffusivity decreases monotonically in time throughout the reaction. Huang and Granick now argue that "the relevant comparison should be the mixture, with and without chemical reaction, because physically this is the more meaningful way to isolate the effects of the chemical reaction". In fact, when we measure diffusion coefficients in different one-, two-, or three-component mixtures (Figure 1a) and use those of the three-component mixture as the reference state as proposed by them, our results are supported even more clearly (Figure 1b,c): the alkyne starts with a Deff close to its new D0 and undergoes a gradual monotonic decay afterwards (in particular for the Deff associated to its terminal proton, signal #1), and the Deff of azide remains close to its D0 value during the first 60-90 minutes of click reaction, but shows a rapid decay afterwards, in accordance with its later entry point to the reaction cycle, or as suggested in 18 , due to peak overlap with ascorbate or its oxidation products. Importantly, throughout the course of click reaction the Deff of both alkyne and azide remained lower than the new D0 values, indicating no measurable diffusion enhancement also with respect to the new reference state. As shown in Figure 1a, no other choice among possible reference states would compromise our conclusion that the components of click reaction do not show any reaction-induced boosted mobility. Indeed, all the changes in their diffusion can be attributed to the role of reaction intermediates. Our results do not reproduce the ~5% transient diffusivity increase of azide shown in Figure 2E of ref. 13. Moreover, they claim a 50% transient diffusivity increase for 2Cu-alkyne, which is however not measured directly but rather indirectly calculated, resulting in quite noisy, potentially unreliable values. In this regard, it is important to note that the variable proton magnetization recovery (due to T1 relaxation) over the course of reaction, as shown in ref. 14, precludes quantitative analysis of NMR signals in terms of reactant and intermediate concentrations. In any case, we believe that not only Fillbrook et al.'s work 18 and our work, 19 but also Huang et al.'s own later work, 13 already debunk a substantial fraction of the results reported in ref. 11 for the CuAAC reaction.</p><!><p>We propose that the measured diffusivity for each signal is consistent with a population-weighted average of the Brownian diffusion of the various components (reactants, reaction intermediates, and product) that carry it, and changes over time of this measured diffusivity reflect changes in the population distribution of these components as the reaction proceeds. We do not understand why Huang and Granick believe that, according to our proposal, the Deff of the reactants "should increase monotonically with time" and that of the product "show slowing down", or why they think we "do not explain" our proposed mechanism, because the mechanism is very simple and already explained in ref. 19. In short, we observe a monotonic decrease in the diffusivity of the signals corresponding to the reactants, and a monotonic increase in the diffusivity of the signal corresponding to the product. The decrease for the reactants can be explained as being due to an increasing fraction of reactant molecules being in larger, slower-diffusing complexed forms (Cu-alkyne, 2Cu-alkyne, azide coordinated with 2Cu-alkyne complex) as the reaction proceeds. Conversely, the increase for the product can be explained as resulting from an increasing fraction of product being free, rather than in the form of larger, slower-diffusing copper triazolide and copper metallacycle complexes, as the reaction proceeds.</p><!><p>Top: A minimal model for catalytic conversion of a reactant (R) to a product (P) by a catalyst (C) is displayed in the inset, which involves six kinetic rates 𝑘 𝑖 , and five populations: free reactant R, free product P, free catalyst C, reactant-catalyst complex CR, and product-catalyst complex CP. From these populations, the free-state fractions of reactant and product 𝑓 (R) and 𝑓 (P) can be calculated as shown. The effective diffusion coefficients of each, 𝐷 eff (R) and 𝐷 eff (P) , are under ideal measurement conditions given by a population average of the diffusion coefficients associated to the free-state (𝐷 R , 𝐷 P ) and complex-state (𝐷 CR , 𝐷 CP ), where generally we expect 𝐷 R > 𝐷 CR and 𝐷 P > 𝐷 CP . Bottom: Schematic showing how, as the reaction progresses, the free fractions of reactant and product progressively decrease and increase, respectively, causing the respective effective diffusion coefficients to progressively decrease and increase as well. (b) Numerical solution of the kinetic scheme in (a) displaying exactly this behavior. Parameters chosen:</p><p>While the CuAAC reaction is rather complex and not all the many intervening steps are well described, the basic mechanism we propose can already be understood within a minimal representative model of a catalyzed reaction (Figure 2). Numerical solution of the kinetic scheme shown in (a), displayed in (b), shows how indeed the fractions of free reactant and product may respectively decrease and increase monotonically as the reaction progresses. Under ideal conditions, in which the free and complex states have the same chemical shift or are in fast exchange, the relaxation rates of both species are the same and constant, and NMR recycle delays are long enough to ensure Boltzmann magnetization recovery (intensity proportional to concentration), these fractions determine the measured effective diffusion coefficient through a population-weighted average. We note, however, that even the simple kinetic scheme in Figure 2 already includes a large number of competing timescales, and thus other behaviors (including non-monotonic ones) of the freestate fractions are possible for different parameter choices. This highlights how even observation of a transient increase in diffusion coefficient is not necessarily a sign of an 'active' enhancement, and may be explained by complex reaction kinetics. Nevertheless, in all cases, a generic feature of such population averages is that the free fractions of reactant and product may never exceed one, implying that their effective diffusion coefficients are bounded from above by the diffusion coefficient of their free form.</p><!><p>In another comment, Huang and Granick point to the occurrence of flocculation in the mixture of alkyne and catalysts (without azide, see Figure 3a) and rightly state that this system is not suitable for quantitative diffusion measurements. The yellow precipitation is due to the formation of Cu(I) -acetylide of propargyl alcohol and its progression towards insoluble highly colored polymeric compounds. [21][22] We should however highlight the fact that in none of the samples in which we measured diffusion before or during the click reaction any flocculation was present (Figure 3b,c). Therefore, we do not see any relevance for this comment in connection with the main results and conclusion of our article. 19 The purpose of using that sample was only to show whether -coordination and/or -bond formation between alkyne and copper ions without further progression into the click reaction cycle would reproduce the alkyne's diffusional changes observed during the click reaction. Our results clearly, albeit qualitatively, showed that the mere alkyne-copper binding in the absence of azide was not able to reproduce alkyne's diffusional changes. 19 In their comment, Huang and Granick could be understood as portraying us as unreasonably resistant or skeptical to the idea of microscopic energy consumption being transduced into mechanical motion. On the contrary, we have worked for many years in trying to understand the mechanisms by which microscopic objects, from colloids to enzymes, do (or don't) convert chemical activity into motion. 1-2, 6-7, 23-24 However we should still exercise utmost caution and scrutinize the results with physical reasoning, to avoid artificially shoehorning the idea of active propulsion into molecular scale systems. For example, the authors of refs. 11-13 like to mention the idea that "boosted motion accompanied by reorientations from rotational Brownian diffusion would produce a random walk with an effective diffusion coefficient larger than that from just Brownian motion". [11][12][13] The same proposal was put forward by them in order to explain observations of enhanced enzyme diffu-sion. [25][26] However, such a mechanism has a very strong dependence on the size of the propelling object. In particular, a particle of size a propelling with speed v will have an effective diffusion enhancement going as ΔD ≈ v 2 /Dr, where Dr ∝1/a 3 is the rotational diffusion coefficient of the particle. Dividing this by the Brownian translational diffusion coefficient, Dt ∝1/a, one finds that the relative diffusion increase scales with particle size as ΔD/Dt ∝ v 2 a 4 . 6 Thus, to obtain the same relative values of diffusion enhancement (say, a few percent), a molecule with a = 0.5 nm needs to propel 100 times faster than an enzyme with a = 5 nm, or 4 million times faster than an active colloid with a = 1 μm. Previous work by us and others 6,8 has shown how the high speeds required make selfpropulsion an unrealistic mechanism for enhanced diffusion already at the scale of enzymes, and this problem only becomes worse at the smaller scale of molecules. This highlights how physical concepts cannot always be transferred across vastly different scales, as appealing as it may be. While we remain open to the idea of chemical activity being transduced into motion at the molecular scale, we believe that every claim must be evaluated and carefully scrutinized according to its own merits. Indeed, the two post-acquisition NMR methods introduced in our article are intended to enable detecting slight enhancements in molecular diffusion through NMR diffusion measurements less prone to artefacts. 19 The observations currently existing for the CuAAC reaction are better and more succinctly explained by Brownian diffusion of the various populations that take part in the chemical reaction.</p>

ChemRxiv

The Synthetic Potential of Thiophenium Ylide Cycloadducts

3+2) cycloaddition reactions are undeniably one of the most robust and versatile synthetic tool in heterocyclic chemistry. The classical 1,3-dipolar cycloaddition, which uses 1,3-dipoles, are however limited to three-atom sequences demonstrating stabilized formal charges in their Lewis structure. The scope of three-atom groupings possible in (3+2) cycloadditions can be greatly expanded by taking of advantage neutral three-atom components (TACs). These also provide an additional degree of chemical possibilities by resulting in zwitterionic (3+2) cycloadducts adaptable to multiple outcomes depending on structure and conditions. In this article, the intramolecular (3+2) cycloaddition reaction between alkynyl sulfides (neutral TAC) and alkynes to provide key thiophenium ylide intermediates is first reported. These highly reactive intermediates provide access to highly substituted fused thiophenes following predictable chemical sequences. Structural features on the obtained thiophenes was found to be highly configurable by judicious choice of both alkynyl sulfide substitution and reaction conditions.

the_synthetic_potential_of_thiophenium_ylide_cycloadducts

Introduction<!>Results & Observations<!>Table 1. Exploration and Optimization of Alkyl Alkynyl Sulfides in (3+2) Cycloadditions with a Tethered Alkyne.<!>Thiophenium Ylide Intermediate and Electrophilic<!>Procedure Simplification and Synthetic Applications.<!>Refinement of Reaction Model with<!>Conclusion

<p>(3+2) cycloadditions represent one of the most useful synthetic platforms for rapidly generating 5-membered heterocycles. The classical 1,3-dipolar cycloaddition pioneered by the group of Huisgen 1,2 has been exploited extensively not only from a synthetic standpoint, 3,4 but also as a tool driving advances in multiple other fields, such as biorthogonal chemistry 5,6 and material science. 7 There are, however, large synthetic limitations inherently associated with 1,3-dipolar cycloadditions. 1,3-dipoles are forcibly limited to three-atom sequences possessing heteroatoms capable of bearing stabilized charges in their Lewis structure. 2 This large limitation can be lifted with the use of "neutral" three-atom components (TACs) 8 capable of (3+2) cycloaddition. The use of neutral TACs is rare, [9][10][11][12][13][14][15][16][17][18][19] but these underexplored chemical entities present great potential for generating synthetic diversity towards heterocyclic structures. While classical 1,3-dipoles 2 produce stable, neutral cycloadducts 3, neutral TACs 4 produce unstable, zwitterionic cycloadducts 5. Therefore, the role of the neutral TAC is two-fold: operating both as a unique three-atom synthon towards 5-membered cycles, but also as a potential zwitterionic handle providing access to additional chemistry unaccessible with classical 1,3-dipoles.</p><p>Recently, our group reported that ynamides 7 -being N-C-C neutral TACs -react intramolecularly with alkynes 1 to produce a wide variety of polycyclic fused pyrroles via the postulated intermediacy of a (3+2) pyrrolium ylide cycloadduct 6. 20 It was observed that all ynamides accessed -derived from carbamates, urea, sulfonamide and sulfamides -cleanly displayed a 1,2-migration of the electron withdrawing group (EWG) in this process to produce 2-EWG substituted pyrroles (Figure 1, Path A). This reliable reactivity was used as a basis to experimentally and theoretically investigate the nature of the unprecedented (3+2) cycloaddition step. Following our discovery, the group of Anderson reported the unique case of yndiamides, which could undergo the (3+2) cycloaddition under either thermal or goldcatalyzed conditions, leading to analogous fused pyrroles. 21 Although pyrroles are of great synthetic interest, the presence of EWGs on the N-substituted alkyne substrates imposes an inherent structural limitation in this transformation. Ynamines, the simpler analogs lacking the EWG, are generally more difficult to access and manipulate due to their instability thus limiting their potential use for the development of more complex processes. In the case of neutral N-alkynyl TACs, the formation of the key pyrrolium ylide of interest can be considered to some extent as tied to the 1,2-migratory outcome.</p><p>In order to study the synthetic potential of the elusive heteroarylium ylides 6 in further details, a neutral TAC easy to access, and more stable under diverse structural modifications would be required. Although various heteroatomsubstituted alkynes have been synthesized in the past, alkynyl sulfides 8 were found to best fit these criteria. Alkynyl sulfides have high thermal stability and have been employed in a wide array of reactions under both basic and mild acidic conditions. 22 Alkynyl sulfides also benefit from being generally easy to synthesize, the traditional approach being the addition of metal acetylides onto electrophilic sulfide species (disulfides or thiosulfonates). A multitude of complementary methods have been more recently developed for the synthesis of alkynyl sulfides with much milder conditions, 23 such as cross coupling reactions, which has widened the diversity of structures onto which they can be installed, even to macrocyclic peptides. 24,25 To date, the (3+2) cycloaddition between alkynyl sulfides 8 and alkynes 1 has not been addressed. Other than the recent work on ynamides, 20,21 the literature on heteroatomsubstituted alkynes as TACs is limited to the pyrolysis of neat compounds 17 or the use of highly reactive benzynes as 2-atom cycloaddition partners, 18,19 which both impose great limits on functional group tolerance and structural diversity. With the prediction that alkynyl sulfides and alkynes may produce thiophenium ylides upon heating, ynetethered alkynyl sulfides presented a great platform for studying a variety of S-substituted thiophenium ylides.</p><p>In this article, we report the successful use of alkynyl sulfides as neutral TACs in (3+2) cycloadditions with tethered alkynes to produce a wide range of polysubstituted thiophenes. We report multiple reactivity outcomes of the intermediate thiophenium ylide depending on the starting alkynyl sulfide substitution, such as 1,2-migration or elimination (Figure 1, Paths A and C). The thiophenium ylide intermediates were also successfully trapped in situ in the presence of electrophiles (Figure 1, Path B), further diversifying synthetic options in this process. On the basis of theoretical and experimental studies, this divergence in reaction outcome, and the mechanistic implications, were studied in further details.</p><!><p>Alkynyl Sulfides as Viable TACs -Optimization and Reaction Scope. By analogy to the use of ynamides as TACs, our investigations of alkynyl sulfides in (3+2) transformations began with the synthesis and reactions of yne-alkynyl sulfides 9a & 9b possessing EWGs directly attached to the sulfur atom (Figure 2). Gratifyingly, we observed the two substrates cleanly rearranging to the desired 2-ester substituted thiophenes 10a & 10b in high yields. This result was a convincing proof that alkynyl sulfides were viable TACs for (3+2) transformations. Considering the 1,2-migratory outcome to be similar to that observed for ynamide derivatives and therefore easily applicable to other EWG substituted S-alkynyl derivatives, we then focused our attention to other alkynyl sulfides bearing substituents that would not undergo a 1,2-shift as easily, and may display other complementary reactivity. A series of alkyl-substituted alkynyl sulfides 9c-h were synthesized and their reactivity evaluated under similar conditions (Table 1). Interestingly, they were all thermally converted into the same trisubstituted thiophene 10c that was observed to be the major and sole isolable reaction product (38-91%). The structure of 10c was confirmed by a series of NMR experiments (see Supporting Information) and single crystal X-ray diffraction studies. Furthermore, by monitoring the reactions of substrates 9c-h by 1 H NMR spectroscopy in sealed NMR tubes, appreciable amounts of alkenes derived from elimination of the corresponding alkyl chains could be identified, thus suggesting that the hydrogen atom recovered at position C(5) of the thiophene originated from this side chain. Although substrates 9c & 9h cannot proceed through an elimination mechanism, these still provided large amounts of the same thiophene 10c, and small amounts of ethylene and cis-stilbene 26 byproducts, respectively. These observations are further explored in the mechanistic section of this report (see Figure 6). Alkynyl sulfides bearing electron neutral or enriched Saryl derivatives were also evaluated, but provided only</p><!><p>a All reactions performed in sealed NMR tubes at the 0.05 mmol scale in toluene-d8 (0.1M) with 2.0 µL mesitylene as internal NMR standard. b Yields assessed by 1 H NMR after the indicated time. c Isolated yield. Table 2. Reaction Scope for the (3+2) Cyclization in Yne-Alkynyl Sulfides: Alkyne Substitutions and Linkers a Unless otherwise specified, reactions were performed in sealed reaction tubes at the 0.1 mmol scale in toluene (0.1M) under air. b Brought to the indicated temperature using a microwave reactor. decomposition products upon heating under the same conditions (see Supporting Information). The tert-butyl alkynyl sulfide 9g, in particular, gave an excellent isolated yield of thiophene 10c (85%). In addition, this model transformation could be successfully scaled up to the gramscale with minimal loss of efficiency, affording 0.85g (80% yield) of the desired thiophene (Table 2, Part A). For these reasons, the tert-butyl group was employed as the S-substituent of choice for the study of other structural parameters influencing the formation of trisubstituted thiophenes.</p><p>The modification of the alkyne substituent was first studied (Table 2, Part A). Variations of the aryl group involving either substitution on the phenyl ring (10k) or replacement with other heteroaromatic cycles (10l & 10m) were all well tolerated, giving good to excellent yields (61-92%) of trisubstituted thiophenes in short reaction times. We then turned our attention to alkynyl, alkenyl, alkyl, as well as non-substituted alkynyl derivatives. Pleasingly, the use of a conjugated 1,3-diyne rapidly afforded the desired alkynyl thiophene 10n with an excellent yield (90%). Cyclohexenyl derivative 10o could be cleanly obtained as well, albeit with a longer reaction time (4.5 h). These two last examples emphasize the favourable nature of the studied (3+2) process when other thermal cycloaddition events (hexadehydroand tetrahydro-Diels-Alder reactions) 27 could be possible. Notably, a higher reaction temperature was required for Csp 3 methyl substituted alkyne 9p which could be converted into thiophene 10p in a moderate 44% yield. A similar reactivity was observed for terminal alkyne 9q (10q: 43%). The lower yields obtained for thiophenes 10p and 10q can be attributed in part to their low molecular weight and volatility, which made their isolation more challenging on small scale. Acyl substituted alkyne 9r was highly active in the (3+2) process, showing some formation of the desired thiophene 10r even at room temperature, and full consumption after only a few minutes of heating. Both silyl and sulfide-substituted alkynes also cleanly provided hetero-substituted thiophenes 10s and 10t. Overall, this (3+2) process was tolerant of all types of alkynyl groups that were accessed, and presents an alkyne scope noticeably larger than that found in the case of ynamides. 20 A variety of substrates possessing a three, four and fivemembered linker were synthesized, many promoting the yne-alkynyl sulfide (3+2) cycloaddition (Table 2, Part B). Exchanging the model ether linker by a sulfonamide or a malonate group led to moderate to good yields of the corresponding thiophenes 10u and 10w. Even the simplest allmethylene linker, upon heating to a higher temperature, led to a good yield (69%) of the corresponding thiophene 10v. Inserting an ester functionality in the linker led to a very clean formation of lactone-fused thiophene 10x (quant.). Further substitution/functionalization of the linker was found advantageous, such as can be seen for 10y and spirocyclic fused thiophene 10z, where a Thorpe-Ingold effect not only permitted the use of a lower temperature but also provided cyclized products in higher yields (91-97%)! When using higher reaction temperatures (200 °C), substrates with four-membered linkers were also found to efficiently undergo the (3+2) cycloaddition, expanding the access to [3.4.0] bicycles in moderate yields for 10aa and 10ab (62% and 37%, respectively). An improved yield was obtained when a geometrically-constrained linker was employed (see 10ac, 75%). Following this success, substrates with a five-membered linker were also synthesized, but their thermal reaction only led to degradation with no</p><!><p>Trapping. Hypothesizing the advent of a thiophenium ylide cycloadduct (6, Figure 1), we considered that this intermediate could be trapped in the presence of an appropriate electrophile, prior to the transfer of a hydrogen at position C(5), to provide fully substituted thiophenes. This method would permit access to more complex thiophenes (such as seen in Figure 2) without recourse of the 1,2-migration pathway, and starting instead from simpler, more accessible S-alkyl derivatives. Gratifyingly, interception of the intermediate ylide was indeed observed when using methyl chloroformate, and so the propensity for a series of alkylsubstituted alkynyl sulfides 9c-h towards electrophilic trapping was evaluated (Table 3). Interestingly, the nature of the alkyl substituent was found to be a key factor affecting the amount of thiophene resulting from the trapping event. Notably, while S-benzyl and S-tert-butyl substituted substrates gave the highest yields of simple thiophene 10c in the previous optimization (See Table 1), they provided ester-substituted thiophene 10a with the poorest yields. n-Propyl substrate 9d was chosen as the most suitable substituent for studying the cyclization / electrophilic trapping sequence, due to a combination of both its ease of synthesis and the highest ratio obtained in favour of the trapping product 10a. As can be reasonably suspected, increasing the amount of methyl chloroformate present also significantly increased the amount of thiophene 10a obtained (up to 85% isolated yield for 10 equiv. of chloroformate used). A control experiment was performed showing that thiophene 10c could not be converted into 10a in the presence of methyl chloroformate in toluene at 130 °C, thus demonstrating that the trapping occurs prior to the hydrogen atom transfer (see Supporting Information for more details).</p><p>Considering that the efficiency of the thiophene functionalization should be directly dependent on the electrophilic strength of the trapping agent, we then investigated the reaction outcomes of different electrophilic species under the same optimized conditions (Table 4). Similarly to methyl chloroformate, propargyl chloroformate and pivaloyl chloride were found to be effective electrophiles affording the acylated derivatives 10ag and 10ai in moderate yields (62% and 58%, respectively). Trapping with the comparatively weaker electrophilic diethylcarbamoyl chloride was possible as demonstrated by the formation of amide 10ah, which was isolated however in a much lower 29% yield. We found that trapping of both aldehydes and ketones was possible, providing secondary and tertiary alcohols 10aj and 10ak, respectively. Interestingly, when an Nsulfonylimide derivative was used as electrophile, a transfer of the propyl chain from the sulfur to the nitrogen was observed in 10am alongside the predicted product 10al in a 1:1 ratio and for an overall 46% yield. Analogously, when carbon dioxide (dry ice) was directly added to a reaction mixture containing methyl alkynyl sulfide 9c, the same model product 10a was obtained without need of any dealkylation (25% yield). With the aim to further diversify the 5-position of the thiophene, thiolation with electrophilic sulfide sources was carried out, gratifyingly affording 10an (79% yield) and trifluoromethylthio derivative 10ao (23% yield). Electrophilic trapping of silyl, stannyl, and phosphoryl derivatives were all successful, leading to hetero-substituted thiophenes 10ap-10ar in moderate yields. Exploiting the nucleophilic nature of the proposed thiophenium ylide intermediate allowed for a rapid and efficient synthesis of 5-functionalized thiophenes using an analogous yne-alkynyl sulfide substrate (9d). The electrophilic trapping approach explored in this scope provides a much more diverse and convergent route to fully substituted fused thiophenes than by accessing more complex substrates, such as those initially studied (9a & 9b, Figure 2).</p><!><p>While isolation of the yne-alkynyl sulfide substrates was necessary for their comparative study in the previous scopes, the particularly simple reaction conditions (only solvent and heat!) necessary for the (3+2) cyclization renders itself attractive for one-pot procedures involving consecutive synthetic steps (Figure 3, Part A). It was found that Sonogashira cross-coupling conditions are compatible with the reactivity of these yne-alkynyl sulfides. For example, coupling of 2-iodothiophene and terminal alkyne 9q in dry toluene rapidly afforded yne-alkynyl sulfide substrate 9m at room temperature, which then cyclized upon heating yielding the desired thiophene 10m in 52% overall yield. Alternatively, in situ condensation of propargyl amine 11 and aldehyde 12 followed by (3+2) cycloaddition upon mild heating provided imine derivative 10as in 69% yield. Finally, the alkynyl sulfide itself could be synthesized in-situ from a C-S coupling. 25 To avoid using foul-smelling tert-butyl thiol as the source of sulfide, tert-dodecyl mercaptan was used as a thiol bearing an alternative tertiary alkyl chain. This grapefruit-scented thiol, along with bromoalkyne 14, smoothly afforded alkynyl sulfide 15 which underwent cyclization towards thiophene 10u in 59% overall yield.</p><p>Interestingly, the temperature needed for the (3+2) process can be greatly reduced under specific structural requirements (Figure 3, Part B). In analogy to the recent study by the Anderson group on yndiamides, 21 the introduction of an amide-substituent directly attached to the alkynyl sulfide in the case of 9at permits the (3+2) transformation to be catalyzed with a gold(I) complex at room temperature. This feature is, however, limited to thioynamides, and its application to other substrates was not found successful.</p><p>Polythiophenes are an important class of conjugated structures capable of efficient charge transport, and have diverse applications in material science. 28 The synthetic accessibility of a broad scope of yne-tethered alkynyl sulfides and their efficient cyclization into thiophenes makes them great precursors for oligothiophene building blocks. For instance, two terminal yne-alkynyl sulfides units of 9q were easily coupled via Sonogashira reaction to afford oligothiophene precursor 9au (Figure 4, Part A). This precursor smoothly underwent a (3+2) cyclization generating the symmetrical tris-thiophene unit 10au in 82% yield. Since the elimination of the tert-butyl groups lead to the formation of 2,3,4-trisubstituted thiophenes, the C(5) position is free to be brominated afterward, affording building block 10av. The findings that the thiophenium ylide intermediate can be trapped by various electrophiles (see Table 4), particularly with tributyltin chloride, can be exploited for the synthesis of oligothiophene building blocks containing contrasting synthetic handles. Precursor 9aw can be readily synthesized from a sequence of Sonogashira coupling, propargylation and thiolation reactions. Using the previously optimized reaction conditions (see Table 3), cyclization of 9aw in the presence of tributyltin chloride allows for regioselective stannylation upon cyclization yielding the asymmetrical bis-thiophene 10aw in 58% yield. Exploiting the nucleophilic reactivity of the intermediate ylide allows to access a bis-thiophene containing both bromine and tin as two synthetically useful handles for subsequent functionalization or polymerization.</p><p>Further application of this method involves the core transformation of the obtained fused thiophenes by exploiting the diene character of their oxidized counterpart, thiophene S,S-dioxides (Figure 4, Part B). These are known to undergo Diels-Alder reactions generating substituted benzenes upon extrusion of SO2. 29,30 Oxidation of model product 10c was achieved using mCPBA and sodium bicarbonate yielding thiophene S,S-dioxide 16 in 41% yield. Thermally induced (4+2) cycloaddition of 16 in the presence of dimethyl acetylenedicarboxylate or 1,4-naphtoquinone as dienophiles afforded the corresponding pentasubstituted benzene derivatives 17 and 18 in 22% and 34% yields, respectively. With the intermediacy of a thiophene S-S-dioxide, this route provides an alternative stepwise metal-free approach to the (2+2+2) cyclization of alkynes. Although this fleeting species could never be directly observed, two case studies in DFT calculations (performed at the B3LYP/6-31+G(d) level of theory) confirmed this type of thiophenium ylide is a viable intermediate, with energies of 6.3 kcal/mol (III, R=t-Bu) and 11.4 kcal/mol (III, R=CO2Me)(Figure 5, Part B).</p><!><p>The (3+2) cycloaddition leading to the thiophenium ylide intermediate may occur via either a concerted (Path A) or stepwise diradical (Path B) mechanism. Using the B3LYP functional, no stable closed shell transition state for a concerted (3+2) cycloaddition could be located for either case studies. 31 However, a stepwise diradical pathway could be easily located in both cases, with rate-limiting steps of 24.9 kcal/mol (TS-I, R=t-Bu) and 25.6 kcal/mol (TS-I, R=CO2Me). Interestingly, the diradical species involved in the stepwise (3+2) pathway is in common with the diradical mechanism proposed for the tetradehydro Diels-Alder (TDDA) reaction (Path C). 32 However, the barrier diverging from the (3+2) mechanism towards a TDDA outcome is quite high at 33.3 kcal/mol for the tert-butyl case study, which explains why the (3+2) cyclization process is the one exclusively observed for these substrates. 33 From the common endocyclic thiophenium ylide, the 1,2-migration outcome (Path D) is theoretically possible for both S-alkyl and S-ester species. DFT studies show, however, that the barriers for this step, 17.2 kcal/mol (TS-III, R=CO2Me) and 27.3 kcal/mol (TS-III, R=t-Bu), largely varies. The barrier for the 1,2-migration step being so high for the S-alkyl species, this may explain divergence of outcomes from the thiophenium ylide depending on substrate substitution.</p><p>While electrophiles can clearly be trapped by the intermediate ylide when present in the reaction medium (Figure 6-A, Path E), it may be what follows protonation of the ylide (Figure 6-A, Path F) which dictates the efficiency of production of the observed thiophene.</p><p>Despite our effort, no concerted protonation-elimination transition state could be located that would directly link intermediate III (R=t-Bu) to the observed thiophene product IV-A and isobutene. Although an alternate pathway making use of a reactive thioketene could be theorized, as tert-butyl alkynyl sulfides have previously been proposed to fragment to these and isobutene, 34 this is highly unlikely when taking in account the high dependence of reaction rates and yields to the nature of the alkyne and its tether to the alkynyl sulfide. In order to better understand the outcome of S-alkyl thiophenium ylides, we consequently shifted our efforts to experimental studies. First, deuteration studies were performed with tert-butyl and n-propyl alkynyl sulfides 9g and 9d and in the presence of several deuterated additives (AcOH-d4, MeOH-d4 and MeCN-d3) (Figure 6, Part B). For both 9g and 9d, the percentage of deuteration observed at position C(5) of the thiophene product 10c is consistent with the acidity of the various additives. It is also in agreement with the amount of external deuterium source employed, substrate 9d leading up to 96% deuteration of the resulting thiophene when MeOH-d4 was used as a co-solvent. These observations are in line with the intermediacy of an endocyclic thiophenium ylide, which is both nucleophilic and basic at the 5-position of the thiophene. A control study on the thiophene product (Figure 6, Part C) also indicates that the deuteration occurs along the reaction pathway, and not in the product. Interestingly, the presence of AcOH-d4 in the case of the n-propyl alkynyl sulfide 9d greatly increased the yield of observed thiophene product from ~40% all the way to 85%! Since the (3+2) process occurs independently from the presence of AcOH-d4 additive, this result seems to indicate that the intermediate derived from the n-propyl alkynyl sulfide 9d is inefficient in producing the desired thiophene, but rendered efficient when protonated. Along with a weaker C-S bond, the tert-butyl-substituted thiophenium ylide differs from the n-propyl by the absence of α-hydrogens, and so the former only proceeds via an elimination mechanism (Figure 6-A, Path G). In the n-propyl case, proton transfer may occur to provide a more stable exocyclic thiophenium ylide (Path H), which may lead to secondary degradation pathways. To test this idea, a probe 9ax was designed where the hypothesized stabilized exocyclic thiophenium ylide 20 (Figure 6, Part D) has been previously shown to undergo ring expansion via a [1,2] alkenyl shift to produce 2Hthiopyrans of type 21. [35][36][37] Indeed, upon heating of the probe with either MeOH or no additive, 2H-thiopyran 21 was observed as the major product of the reaction. In line with the effect of AcOH-d4 observed in the study described in Part B, when the probe 9ax was heated in the presence of a stoichiometric amount of AcOH, only the thiophene 10c was observed, along with a stoichiometric amount of acetate substitution by-product 19. This interesting set of observations shows yet another controllable outcome of the (3+2) cycloaddition of yne-alkynyl sulfides dictated by both substrate substitution and reaction conditions.</p><p>It is important to note, however, that the observation of a significant amount of thiophene 10c upon heating 9ax when no additive was present indicates that elimination of a β-hydrogen is not the only pathway leading to thiophene 10c. The same can be concluded from substrates 9c and 9h (see Table 1), which upon heating, produce large amounts of thiophene 10c along with by-products derived from the "dimerization" of the corresponding alkyl groups used. These observations suggest an alternative pathway to 10c, where a sequence of alkylation-elimination between two ylide intermediates provide two units of thiophene 10c along with a symmetrical alkene by-product (Figure 6-A, Path I).</p><!><p>To summarize, the concept of using neutral TACs in (3+2) cycloadditions has been successfully applied to the formation of polyfunctionalized thiophenes from easily accessible alkyl and acyl alkynyl sulfides tethered with alkynes. Due to their more stable nature, alkynyl sulfides were used as representative heteroatom-substituted alkynes to study the outcome of diverse heteroarylium ylide (3+2) cycloadducts stemming from X-alkynyl species. It was found that acyl derivatives lead to 2-acylthiophenes via a 1,2-migration step, and alkyl derivatives lead to trisubstituted thiophenes mainly via a β-elimination mechanism.</p><p>The scope of alkynes that may participate as two-atom components in this (3+2) cycloaddition is quite diverse. Essentially, all variations on the alkyne leading to an isolable yne-alkynyl sulfide substrate were tolerated in the (3+2) process. The scope of linkers capable of promoting the transformation was also quite varied, the best results obtained with 3-atom linkers. The effect of larger linkers on reaction rate could be generally alleviated by increasing the temperature, allowing for the use of 4-atom linkers in this transformation. The simplicity in required conditions (only heating!) for this reliable cyclization method renders it easy to be used in conjunction with multiple other reactions to rapidly build molecular diversity. Finally, S-alkyl thiophenium ylide intermediates were successfully intercepted using a wide range of electrophiles, providing a series of uniquely functionalized tetrasubstituted fused thiophenes, and thus highlighting their synthetic potential for the generation of molecular diversity. This work demonstrates that endocyclic heteroarylium ylides are tangible species that can be taken advantage of to obtain specific heterocyclic structures of interest. The synthetic potential of these underexplored intermediates, which may extend to the realm of redox transformations, has yet to be exploited. Dominic Campeau -Department of Chemistry and Biomolecular Sciences, University of Ottawa, K1N 6N5 Ottawa, Canada; orcid.org/0000-0003-0260-8952</p>

ChemRxiv

Photoredox-catalyzed stereoselective alkylation of enamides with <i>N</i>-hydroxyphthalimide esters <i>via</i> decarboxylative cross-coupling reactions

Stereoselective b-C(sp 2 )-H alkylation of enamides with redox-active N-hydroxyphthalimide esters via a photoredox-catalyzed decarboxylative cross-coupling reaction is demonstrated. This methodology features operational simplicity, broad substrate scopes, and excellent stereoselectivities and functional group tolerance, affording a diverse array of geometrically defined and synthetically valuable enamides bearing primary, secondary or tertiary alkyl groups in satisfactory yields.

photoredox-catalyzed_stereoselective_alkylation_of_enamides_with_<i>n</i>-hydroxyphthalimide_esters_

Introduction<!>Results and discussion<!>Conclusions

<p>As a crucial subclass of enamines endowed with a delicate balance of reactivity and stability, enamides have attracted increasing attention among the chemical community as pivotal and versatile building blocks which are of recognized synthetic value in the construction of biologically and pharmaceutically active molecules, 1 especially small but complex nitrogencontaining compounds. 2 In the past few decades, we have witnessed a booming development in new synthetic strategies for the regio-and stereo-selective functionalization of enamides, especially at their b-C(sp 2 )-H bond, which are capable of producing enamides bearing a diverse array of functional groups through arylation, 3 alkenylation, 4 triuoromethylation, 5 diuoroacetylation, 6 alkynylation, 7 acylation, 8 sulfonylation 9 and other useful transformations. 10 Nevertheless, the coupling of alkyl moieties to enamides has been considered a more challenging task with scarce advances demonstrated. 11 One of the existing scenarios for the direct C-H alkylation of enamides was achieved by using electron-decient bromides as alkylating agents as established by Yu and co-workers through visible-light photoredox-catalysis (Scheme 1a, eqn (1)) 11a and by our group through a palladium-catalyzed strategy (Scheme 1a, eqn (2)) 11b . Recently, another elegant methodology for the branch-selective alkylation of enamides with terminal olens was demonstrated by Dong and co-workers (Scheme 1b). 11c However, the success to date has been somewhat restricted with respect to the limited scope of both enamides and alkylating reagents and the relatively strict reaction conditions. Thus, the development of a robust and generally applicable method for the preparation of enamides bearing a diverse range of alkyl groups with versatile functionalities has been considered a remaining challenge.</p><p>The redox-active alkyl N-hydroxyphthalimide esters (NHP) derived from alkanoic acids, as demonstrated for the rst time by Okada 12 and Overman, 13 have entered into an era of "Renaissance" in the past few years in a myriad of crosselectrophile coupling reactions as C(sp 3 ) radical equivalents Scheme 1 Alkylation of enamides.</p><p>through single-electron-transfer reduction and decarboxylation. Recent advances in this arena have witnessed a rapid development in a broad range of decarboxylative cross-coupling reactions to forge C(sp 3 )-C or C(sp 3 )-X (X ¼ Si, B, Se, etc.) bonds via transition-metal 14,15 and photoredox catalysis, 16,17 as elegantly established by the groups of Baran, 14a-g,o,15b Weix, 14h,i Fu, 16c-g,17a,b Oestreich, 15 Phipps, 16a Xiao, 16l and many others. 18 Very recently, Fu and co-workers demonstrated a brand new catalytic combination of sodium iodide and triphenylphosphine for the crosscoupling of redox-active esters with silyl enol ethers or heteroarenes without resorting to the use of dye or transition-metal based photocatalysts. 19 Enlightened by these seminal breakthroughs, we herein demonstrate a robust and practical protocol for stereoselective decarboxylative cross-coupling of NHP esters with enamides, forging a diverse array of geometrically dened alkylated enamides bearing various functional groups under mild conditions (Scheme 1c). Notably, this approach allows the incorporation of various primary, secondary and tertiary alkyl groups into enamides, which represents a signicant advance and a crucial complement to existing methods 11a,b which only enable the incorporation of electron-decient secondary alkyl groups.</p><!><p>At the outset of our investigation, N-benzyl-N-(1-phenylvinyl) acetamide ( 1a) and 1,3-dioxoisoindolin-2-yl cyclohexanecarboxylate (2a) were selected as model substrates for the screening of optimal reaction conditions (Table 1). Initial screening of common photocatalysts showed that fac-Ir(ppy) 3 was superior to Ru(bpy) 3 Cl 2 and Eosin Y (Table 1, entry 1 vs.</p><p>entries 2 and 3). Further investigation of solvents revealed that DMF was the optimal choice for the transformation (Table 1, entry 1 vs. entries 7-9) and the most appropriate concentration of the enamides was 0.3 M (Table 1, entries 11 and 12 vs. entries 1 and 10). The optimal loading of the photocatalyst proved to be 1.0 mol% with respect to reaction time and efficiency (Table 1, entries 4-6 vs. entry 1). The employment of 1.2 eq. of NHP esters instead of 1.5 eq. led to an increase of the product yield (Table 1, entry 12 vs. entry 11). Control experiments revealed that the photoredox catalyst and light were both of crucial importance for this transformation, and no desired product was formed in the absence of the photocatalyst or without irradiation (Table 1, entries 13 and 14).</p><p>With the optimal reaction conditions in hand, we next examined the substrate scope with regard to different enamides or enecarbamates 1a-1s with NHP esters 2a or 2s; the results are summarized in Table 2. It was found that substrates bearing either electron-withdrawing (1b-1h) or electron-donating groups (1i-1n) were viable in this transformation to furnish the desired products 3ba-3na in considerable yields. The substrates with ortho-or meta-substituents were well tolerated to give 3ma, 3ha and 3la in synthetically applicable yields, respectively. Substrates bearing halogen atoms (-Cl, -Br, or -I) also afford 3ca-3ea in excellent yields, enabling them to be amenable for further functionalization through cross-coupling reactions. A range of useful functional groups such as CF 3 and CO 2 Et were also applicable to this reaction to give 3fa and 3ga in 77% and 87% yields, respectively. Notably, a heterocyclic skeleton such as a 3-thienyl moiety was also well tolerated to give the target product 3oa in 72% yield. Replacing the N-protecting benzyl group with methyl and Boc did not attenuate the reaction efficiency, affording 3pa and 3qa in synthetically useful yields. Gratifyingly, a handful of enecarbamates N-Boc (1r) or N-Cbz (1s) substituents could smoothly react with redoxactive ester 2s, giving rise to alkylated enamides 3rs and 3ss in 65% and 73% yields, respectively. Notably, in all cases, this transformation proceeded smoothly in a stereoselective manner to afford geometrically dened E-type alkylated enamides (see the ESI † for details); the stereochemistry has been unambiguously conrmed through X-ray crystallography of 3ce as shown in Table 3. 20 Next, we investigated the generality of this reaction with respect to the scope of various NHP esters (Table 3). A broad range of NHP esters with different cyclic moieties were amenable to this transformation to give 3ab-3af and 3ce in moderate to good yields. It is worth noting that the protecting groups on piperidine such as tert-butyloxycarbonyl (Boc), p-toluenesulfonyl (Ts) or even heterocyclic 2-furancarbonyl were well tolerated. A plethora of NHP esters with primary alkyl groups were also readily applicable to this reaction to forge 3ag-3ak smoothly. Several useful functional groups such as phenol and ketone were also compatible with this transformation to give 3ai and 3ak in good yields, respectively. Especially noteworthy was the excellent compatibility of tertiary alkyl groups for this transformation, enabling the formation of enamides 3al and 3am bearing a quaternary carbon centre which were relatively difficult to be produced through other synthetic methods. In addition, various natural amino acid-derived NHP esters were viable substrates, affording synthetically valuable products 3an-3aq in moderate to good yields. Gratifyingly, an NHP ester bearing a naturally occurring dehydrocholic acid fragment containing three base-sensitive ketone groups was readily amenable to the transformation to afford 3ar in 82% yield.</p><p>To showcase the synthetic utility and practicality of this transformation. We have conducted a range of further transformations of the alkylated enamides. A gram-scale reaction of 1a with 2a proceeded smoothly, affording 3aa in good yield and stereoselectivity (Scheme 2a). Notably, upon treatment with triuoroacetic acid at 110 C, the E-congured enamides 3aa, 3fa and 3ae could be converted to their Z-isomers in moderate yields (which might be attributed to decomposition), allowing us to easily control the stereochemistry of the alkylated enamides (Scheme 2b). 21 The alkylation of enamide 4 with NHP ester 5 proceeded smoothly under standard reaction conditions to give the desired product 6 in 65% yield, which underwent a subsequent palladium-catalyzed intramolecular Heck coupling to furnish a synthetically and pharmaceutically crucial isoquinoline derivative 7 in 76% yield (Scheme 2c). Next, Pd/Ccatalyzed hydrogenation of enamide 3aa was successfully conducted under mild conditions to give benzylamine 8 in 69% yield (Scheme 2d). To our delight, the hydrolysis of alkylated enamides in the presence of concentrated HCl (aq.) afforded a broad range of a-alkylated ketones in excellent yields (Scheme 2e). Interestingly, when 3ak was applied under the hydrolysis condition, a cascade hydrolysis-intramolecular cyclization reaction occurred to give 9e in 65% yield (Scheme 2f). Gratifyingly, when alkylated enamide 3aa was treated with m-chloroperoxybenzoic acid (m-CPBA), a-acyloxyketone 10 was obtained in 75% yield aer a tandem epoxidation-intramolecular nucleophilic addition-elimination-hydrolysis process (Scheme 2g). It is worth noting that the N-Boc protecting group of 3qa could be removed efficiently by treatment with zinc bromide to give the desired product 11 under mild reaction conditions (Scheme 2h).</p><p>A number of preliminary mechanistic studies were conducted to shed more light on the reaction pathway. Initially, a radical-trapping experiment in the presence of a radical scavenger 2,2,6,6-tetramethyl-1-piperidinyloxy (TEMPO) was performed. A complete inhibition of the reaction was observed the alkyl radical could be intercepted by TEMPO to generate intermediate 12 as detected by GC-MS, which suggested that the reaction went through a plausible radical mechanism (Scheme 3a). Secondly, the coupling of a radical-clock-containing NHP ester 13 with enamide 1a afforded the ring-opening product 14, which strongly supported the participation of radical intermediates (Scheme 3b). In addition, we have determined a quantum yield of F ¼ 0.71 for the model reaction of 1a with 2a (see the ESI † for details), 22 implying that it is highly possible for the reaction to proceed through a photoredox catalytic pathway rather than a radical-chain mechanism.</p><p>Based on the above observations, we have proposed a plausible mechanism for the photoredox-catalyzed decarboxylative alkylation of enamides with NHP esters. Initially, the iridium photocatalyst</p><!><p>We have developed a novel, efficient and generally applicable approach for the chemo-and stereo-selective alkylation of enamides with NHP esters. A wide array of enamides and NHP esters bearing various functional groups were viable for this protocol to afford synthetically important and geometrically dened enamides bearing primary secondary or tertiary alkyl groups in moderate to good yields and excellent stereoselectivities. A plethora of further transformations were applied to showcase the synthetic value of this transformation. A radical reaction pathway was proposed through mechanistic investigation. The simple operation and the easy availability of the starting materials also allowed this method to pave a new way for the preparation of synthetically crucial alkylated enamides.</p>

Royal Society of Chemistry (RSC)

Reactivity of Metal-Free and Metal-Associated Amyloid-β with Glycosylated Polyphenols and Their Esterified Derivatives

Both amyloid-β (Aβ) and transition metal ions are shown to be involved in the pathogenesis of Alzheimer's disease (AD), though the importance of their interactions remains unclear. Multifunctional molecules, which can target metal-free and metal-bound Aβ and modulate their reactivity (e.g., Aβ aggregation), have been developed as chemical tools to investigate their function in AD pathology; however, these compounds generally lack specificity or have undesirable chemical and biological properties, reducing their functionality. We have evaluated whether multiple polyphenolic glycosides and their esterified derivatives can serve as specific, multifunctional probes to better understand AD. The ability of these compounds to interact with metal ions and metal-free/-associated Aβ, and further control both metal-free and metal-induced Aβ aggregation was investigated through gel electrophoresis with Western blotting, transmission electron microscopy, UV-Vis spectroscopy, fluorescence spectroscopy, and NMR spectroscopy. We also examined the cytotoxicity of the compounds and their ability to mitigate the toxicity induced by both metal-free and metal-bound Aβ. Of the polyphenols investigated, the natural product (Verbascoside) and its esterified derivative (VPP) regulate the aggregation and cytotoxicity of metal-free and/or metal-associated Aβ to different extents. Our studies indicate Verbascoside represents a promising structure for further multifunctional tool development against both metal-free Aβ and metal-Aβ.Alzheimer's disease (AD) is a growing concern for global public health, spurred on, in part, by a lack of effective treatments or cures 1 . While potential therapeutics have been developed for AD, their clinical success has been hindered by a limited molecular level understanding of the disease's etiology [2][3][4] . AD is commonly considered a protein misfolding disease characterized by the presence of protein aggregates, including senile plaques composed primarily of amyloid-β (two main Aβ forms, Aβ 40 and Aβ 42 ) 4-6 . Aβ undergoes a progressive aggregation process, advancing from a small, intrinsically disordered peptide to intermediate oligomers of various sizes and structures, finally forming extended fibers; individual aggregates are believed to have varying degrees of toxicity and relevance in AD 4,6,7 .The senile plaques in the AD-affected brain have also been shown to contain elevated concentrations of transition metals, specifically Cu, Zn, and Fe, which suggests that these metal ions interact with and alter the aggregation of Aβ 4,[8][9][10] . Understanding the interactions between metal ions and Aβ in vitro could help elucidate potentially toxic mechanisms of both factors in AD [10][11][12] . These metal ions could accelerate the aggregation of Aβ while simultaneously generating a variety of aggregates which may have biological functions distinct from those formed in the absence of metals 9,10 . Additionally, the binding of Aβ to redox active metals (i.e., Cu) can facilitate redox cycling and lead to the production of reactive oxygen species (ROS) resulting in an oxidative stress environment, a known characteristic of the AD-afflicted brain [12][13][14] . The effects of these interactions in vivo remain unclear, however;

reactivity_of_metal-free_and_metal-associated_amyloid-β_with_glycosylated_polyphenols_and_their_este

<!>Results<!>Metal Binding Properties of Polyphenols.<!>Interaction of Polyphenols with Multiple Aβ Forms.<!>Antioxidant Properties.<!>Regulating Cytotoxicity Related to Metal-Free and Metal-Associated Aβ. Previous studies of<!>Discussion<!>Cu(II)<!>Zn(II)<!>Methods<!>Amyloid-β (Aβ) Inhibition and Disaggregation Experiments.<!>Transmission Electron Microscopy.<!>Cell Viability Measurements.

<p>in-depth understanding is hindered by the multifactorial nature of the disease which makes it difficult to identify and quantify the influence of any of the potentially causative agents.</p><p>The application of chemical probes which can modulate the various factors associated with AD (e.g., Aβ, metal ions) may advance our understanding of the disease and uncover different toxic factors by isolating individual potential culprits. Unfortunately, there remains a lack of understanding about the relationship between small molecules and their subsequent biological functions with regards to the multiple aspects associated with AD. It is believed that hydrophobic interactions drive early stages of Aβ aggregation and that compounds with similarly hydrophobic regions may effectively disrupt these aggregation-promoting forces and act as modulators of amyloid formation; some compounds of this nature have been investigated previously 6 . It is unclear, though, which chemical moieties on these compounds may be the most potent at altering this process. To increase the specificity of anti-amyloidogenic compounds, multifunctional compounds have been designed to target and modulate additional aspects associated with AD (i.e., metal-associated Aβ (metal-Aβ), ROS) along with metal-free Aβ simultaneously [15][16][17] . Compounds known to interact with Aβ have been appended with known metal binding moieties to generate molecules capable of targeting both metal-free Aβ and metal-Aβ, and these multifunctional compounds have demonstrated an ability to modulate both factors to differing extents 15,16 . As was the case with Aβ interaction, however, rational design of these metal binding moieties is hindered by a limited understanding of how these chelating agents function in the complex AD environment. Additionally, while it is desirable for compounds to also mediate oxidative stress, both through the modulation of ROS and free radicals, this function is similarly difficult to rationally incorporate into a structural entity. Efforts have previously been made to design multifunctional compounds for investigating the role of metal-free Aβ, metal-Aβ, and ROS in AD 18 , but the advancements in the field are generally slowed by limited information on the structure-function relationships between small molecules and their ability to regulate these disease-related features.</p><p>In order to gain better insight to this complex disease and to broaden current understanding of the connection between chemical structure and its function in the presence of factors associated with AD, three naturally occurring polyphenolic glycosides (Phlorizin, Verbascoside, and Rutin; Fig. 1) were chosen for a selective reactivity study towards both metal-free Aβ and metal-Aβ. The investigation of naturally occurring compounds gives the substantial advantage of a minimal toxicity profile and a significant amount of background information from traditional medicine. Natural polyphenolic products have been previously shown to possess potential anti-amyloidogenic activity 19,20 . Both Verbascoside and Rutin have demonstrated the capacity to alter the aggregation and toxicity of metal-free Aβ aggregation towards nontoxic species [21][22][23][24] , and phloretin, the non-glycosidic version of Phlorizin, has exhibited an ability to prevent membrane-associated aggregation of Aβ 25 . While these preliminary studies examined aggregation in the absence of metal ions, the presence of known metal interaction moieties, phenol and catechol groups in Phlorizin and Verbascoside/Rutin, respectively, suggests that these compounds may be capable of simultaneously interacting with both Aβ and metal ions [26][27][28][29] . Furthermore, all three compounds are known antioxidants, indicating that they could also help mitigate oxidative stress associated with the AD-affected brain [30][31][32] . Finally, these compounds are naturally glycosylated, which has been proposed to improve bioavailability and distribution in the brain [33][34][35] , as well as redirect the folding of metal-free Aβ species 36,37 . Combined, these structural and chemical features suggest that Phlorizin, Verbascoside, and Rutin possess unique chemical features desirable for a probe to target multiple factors (i.e., metal-free and metal-bound Aβ) and mitigate their toxicity leading to AD.</p><p>To provide insight into structure-activity relationships, selectively esterified derivatives, F2, VPP, and R2 (Fig. 1), were prepared and investigated alongside their parent compounds [30][31][32] . Esterification could provide multiple benefits in the quest for an effective multifunctional probe while also acting as a general proof of concept that a prodrug approach is a viable method in the search for molecular tools in AD research. Ester formation may improve trafficking across lipid bilayers and facilitate targeted delivery of unprotected compounds within cells following cleavage by esterases; for these reasons, it is a commonly used modification in prodrug design 38 . The ester groups may also improve the ability of the compounds to passively diffuse across the blood brain barrier (BBB) due to greater lipophilicity 16,33 . Finally, the increased hydrophobicity could promote interactions between the compounds and the more aggregation prone regions of the Aβ sequence, which are similarly hydrophobic 4 . Overall, we believe that esterification may tune the ability of these three compounds to interact with both metal-free Aβ and metal-Aβ species while simultaneously improving the bioavailability, making them more suitable for future in vivo applications than their non-esterified counterparts. Using these six compounds, we aim to further expand the understanding of structure-function relationships between multifunctional probes towards both metal-free Aβ and metal-Aβ through a detailed, molecular level characterization.</p><!><p>Influence of Polyphenols on Metal-Free and Metal-Induced Aggregation In Vitro. Initially, to investigate the effects of the six polyphenols (Fig. 1) on the structure and formation of both metal-free Aβ and metal-Aβ aggregation pathways, disaggregation (Figs 2 and 3, and S1) and inhibition (Fig. S2) experiments were performed. Both the more prevalent Aβ 40 and the more aggregation-prone Aβ 42 isoforms were employed in both inhibition and disaggregation studies 4 . For disaggregation experiments, Aβ was allowed to aggregate in either the absence or presence of either CuCl 2 or ZnCl 2 for 24 h to form aggregated species. The compounds were then added to these aggregates and incubated for either 4 or 24 h to determine their ability to redirect the size and structure of pre-aggregated Aβ species. In inhibition experiments, Aβ (for metal-Aβ samples, Aβ samples were treated with CuCl 2 or ZnCl 2 ) and the compounds were incubated for either 4 or 24 h to identify how they are capable of modulating the early steps in aggregation of both metal-free Aβ and metal-Aβ. Gel electrophoresis with Western blot (gel/Western blot) using an anti-Aβ antibody (6E10) and TEM were utilized to visualize the size distribution and morphology, respectively, of the resultant Aβ species upon treatment with polyphenols in the absence and presence of metal ions for both disaggregation and inhibition experiments 18,39 .</p><p>Neither Phlorizin nor F2 demonstrated a significant ability to interact with metal-free Aβ or metal-Aβ species in either a disaggregatory or inhibitory manner. In disaggregation experiments, gel/Western blot revealed that neither of these compounds was able to transform preformed Aβ aggregates regardless of Aβ isoforms or metal presence (Figs 2a and 3a, lanes 2 and 3). Phlorizin did appear to slightly change the morphology of species to marginally more disordered than untreated fibrils; however, large and extended Aβ aggregates were observed with and without metal ions (Fig. S1). A similar lack of reactivity was observed in inhibition experiments (Fig. S2, lanes 2 and 3). These two compounds appear unable to influence Aβ 40 and Aβ 42 aggregation behaviors regardless of the presence or absence of metal ions.</p><p>Both Rutin and R2 showed mild reactivity with metal-free Aβ and/or metal-Aβ. Under disaggregation conditions, R2 produced low MW aggregates of Aβ 40 (< 15 kDa) after 4 h in the absence of metal ions; after 24 h in the absence of metal ions, both Rutin and R2 triggered relatively low MW aggregates (< 35 kDa) (Fig. 2a, lanes 6 and 7). The morphology of these species was slightly more compact that that of untreated Aβ 40 aggregates (Fig. S1). A relatively similar range of Aβ 40 aggregates was generated by both Rutin and R2 when CuCl 2 was present, after 24 h (Fig. 2a), and their morphology was even more compact than in the absence of metal ions (Fig. S1). Neither Rutin nor R2 appeared to have any impact on preformed Zn(II)-Aβ 40 aggregates. The two compounds displayed a minimal or no effect upon preformed Aβ 42 aggregates, regardless of the presence or absence of metal ions (Fig. 3a, lanes 6 and 7). Rutin and R2 presented a modest ability to redirect Cu(II)-induced aggregation of Aβ 40 or Aβ 42 after 24 h incubation in inhibition studies (Fig. S2, lanes 6 and 7). This suggests that both compounds have mild reactivity towards preformed Aβ aggregates, favoring those formed in the presence of CuCl 2 . Unlike the other four compounds, both Verbascoside and VPP were able to control Aβ 40 and Aβ 42 aggregation, though to different extents. When preformed, metal-free Aβ 40 aggregates were treated with VPP for 4 h, a wide MW range (10-260 kDa) of species was formed (Fig. 2a, lane 5). These species were still observed after 24 h incubation. Verbascoside, the non-esterified parent compound of VPP, did not indicate any reactivity against preformed ). In the case of Zn(II)-containing aggregates; the initial disaggregation by Verbascosde and VPP was slower still. The Aβ 40 sample treated with Verbascoside for 24 h exhibited smaller-sized species while VPP triggered a wider range of aggregates, similar to that observed in both the absence of metals and the presence of Cu(II) (Fig. 2a). For all aggregates, both Verbascoside and VPP were able to convert large extended fibrils into smaller, generally amorphous structures (Fig. 2b). This same reactivity trend was observed when Verbascoside and VPP were added to metal-free and metal-associated, preformed Aβ 42 aggregates (Fig. 3). These compounds were capable of slightly altering Aβ 42 aggregates in the absence of metal ions. Additionally, VPP generated various-sized aggregates of Zn(II)-Aβ 42 after 4 h only, a process which required 24 h with Aβ 40 (Fig. 3a). Similar to Aβ 40 , unstructured Aβ 42 aggregates induced by treatment with the two compounds were shown by TEM (Fig. 3b).</p><p>Verbascoside and VPP also displayed reactivity towards Aβ 40 and Aβ 42 in inhibition experiments (Fig. S2, lanes 4 and 5). Only VPP affected the aggregation of metal-free Aβ 40 /Aβ 42 , which was evident just after 4 h with both isoforms. Both Verbascoside and VPP demonstrated inhibitory activity towards metal-Aβ 40 /Aβ 42 . In the presence of Cu(II), the two compounds promoted a variety of aggregates to different extents after 4 h incubation; Verbascoside generated a smaller MW range of aggregates than VPP, suggesting that their activity may be size and/or conformation dependent. A similar trend in Aβ aggregate size was observed in the presence of Zn(II) when Verbascoside and VPP were introduced (Fig. S2). Verbascoside and VPP presented the similar time dependence in the presence of Zn(II) for the inhibition of Aβ 40 aggregation as they did with Aβ 40 disaggregation; their reactivity towards Zn(II)-Aβ 40 was noticeable only after 24 h (Fig. S2a). This suggests that the timescale of reactivity may be dependent upon solution conditions, specifically the metal(s) present in solution and the morphology of aggregates, at a given time. Taken together, the results from these disaggregation and inhibition experiments suggest that Verbascoside and VPP have a distinct capacity to modulate the aggregation of metal-Aβ (as well as metal-free Aβ mainly in the case of VPP) which is generally absent for the other four compounds analyzed in this study. It is interesting to observe that among the selected polyphenols the reactive Verbascoside and VPP both feature distinctive double catechol moieties, bearing the two characteristic ortho-hydroxy groups. Meanwhile, Rutin and R2 which only show mild reactivity have a single catechol moiety. Phloririzin and F2 have no catechol moieties while also presenting no noticeable reactivity with Aβ species. This points to a potential role for the catechol moiety in the activity of these compounds. Additionally, from this set of compounds, it appears that structural modification of the sugar moiety (i.e., selective esterification) can drastically alter the function of the parent compound, as was seen for both Verbascoside/VPP and Rutin/R2.</p><!><p>In order to elucidate the molecular level interactions responsible for the varied anti-amyloidogenic activity of the polyphenols towards metal-Aβ species (vide supra), it is necessary to understand the potential binding of each compound with the component parts of system: metal ions and Aβ monomers/aggregates. The potential interaction between the polyphenols and metal ions was investigated first. The OH groups found on the aromatic rings of all six compounds could potentially interact with transition metals; the catechol moieties of Verbascoside, VPP, Rutin, and R2 are especially likely to bind metal ions in a manner similar to previously reported polyphenols 20,28,[40][41][42] . It has already been suggested that both Verbascoside and Rutin could interact with Cu(II) in solution [43][44][45] . The Cu(II) binding properties of Phlorizin/F2, Verbascoside/VPP, and Rutin/R2 were examined by UV-Vis while the Zn(II) binding of Verbascoside and VPP was monitored by 1 H NMR due to a lack of significant optical changes in the spectra when the ligand was treated with Zn(II) (Fig. 4, S3 and S4).</p><p>The UV-Vis spectra of the six polyphenols showed different levels of spectral changes following the titration of CuCl 2 in buffered aqueous solution (pH 7.4; Fig. 4a and S3). Addition of CuCl 2 to both Verbascoside and VPP in solution induced a slight change in the absorption bands at ca. 330 nm and 405 nm, implying potential interaction between the catechol moieties of both ligands and Cu(II) in solution (Fig. 4a) 46 . The spectral change is more drastic in VPP than in Verbascoside, which suggests that esterification of the OH groups in the sugar ring could reduce their competitive interaction with Cu(II) and promote binding by the catechol moieties. Neither compound presented significantly noticeable spectral shifts indicative of complex formation, however, suggesting that interaction of either Verbascoside or VPP with Cu(II) in solution could be weak. Previous studies on complex formation between Verbascoside and Cu(II) are unclear about the extent of complex formation, which indicates that the complex formation could depend upon experimental conditions (i.e., buffer, salt concentration, pH) 43,47 . Addition of CuCl 2 to solutions of either Phlorizin or F2 induced no significant shift in spectral features, possibly due to the minimal interaction of the phenol moiety in the compounds for Cu(II) in solution (Fig. S3). The presence of Cu(II) did induce modest changes in the spectrum of Rutin, including a decrease in the primary peak at ca. 330 nm and an increase in a shoulder at ca. 425 nm (Fig. S3). This may be caused by the interaction between the catechol moiety and Cu(II). These shifts are similar, though of smaller magnitude, to those previously observed for Rutin-Cu(II) complexes 44,45 . Finally, R2 exhibited slight spectral changes over the course of the titration, suggesting minimal or no interaction with Cu(II) in solution (Fig. S3), suggesting that esterification of Rutin could change the interaction between the ligand and Cu(II). These results suggest that only Verbascoside, VPP, and Rutin have an observable ability to interact with Cu(II) under the conditions of this study.</p><p>The interaction of Verbascoside and VPP with Zn(II) was also investigated by 1 H NMR in DMSO-d 6 (Fig. 4b). When Zn(II) was titrated into Verbascoside up to 2 equiv, all phenolic protons exhibited minor sharpening while titration up to 10 equiv resulted in selective broadening of the resonances associated with the caffeic acid moiety. The hydroxytyrosine phenolic protons showed minimal change upon addition of more Zn(II). Selective broadening suggests that Zn(II) preferentially associates with the catechol moiety in caffeic acid which causes proton exchange with residual water present in the solvent. A similar pattern was observed when Zn(II) was added to VPP. During the titration of VPP, the protons of the catechol group in caffeic acid broadened, suggesting interaction with the metal and subsequent proton exchange with the residual aqueous solvent. No other proton resonances demonstrate any shifts or broadening over the course of the titration for either compound (Fig. S4). This would suggest that in both Verbascoside and VPP, Zn(II) is observed to relatively weakly interact with the phenolic protons of the caffeic acid moiety over all other portions of the compounds, though weakly. Overall, both Verbascoside and VPP are shown to interact with both Cu(II) and Zn(II) with a modest affinity under the conditions employed in this study, implying that direct interaction with the metal ions could be partially responsible for the ability of these compounds to redirect metal-Aβ aggregation (vide supra). This is likely not the direct result of metal chelation, however, given the relatively strong affinity of Aβ for both metal ions 10 .</p><!><p>Given the ability of both Verbascoside and VPP to redirect the morphology and aggregation of Aβ, the direct interaction of the two, along with both Phlorizin and R2, with monomeric Aβ 40 was investigated (Fig. 5 and S5). 2D band-Selective Optimized-Flip-Angle Short-Transient Heteronuclear Multiple Quantum Coherence (SOFAST-HMQC) NMR experiments were performed to identify potential residue-specific interactions between monomeric Aβ peptide and the compounds 48 . The chemical shift perturbation (CSP) for each resolvable residue was calculated by comparing the spectrum of ligand-free Aβ 40 with that of Aβ 40 in the presence of excess ligand (10 equiv) and these CSP values were compared to the average CSP values (Figs 5a, 5b and S4) to identify potentially favored interactions 20,28 . All four compounds induced some mild to moderate (0.02-0.035 ppm) chemical shifts in varying regions of the peptide. Verbascoside (Fig. 5c) triggered a relatively distinct shift in D23 and also weakly perturbed Q15 while VPP (Fig. 5d) caused some shifts in Y10, E11, Q15, or F20. The interactions of Verbascoside seem to favor both polar and charged residues that are centrally located in the sequence, suggesting that both dipole-dipole interactions and hydrogen bonding could direct potential ligand binding. VPP, however, perturbed a mixture of hydrophobic and hydrophilic residues, suggesting that the added acyl groups could encourage some nonpolar contacts during ligand interaction with Aβ. π -π stacking between VPP and the aromatic side chains of Y10 and/or F20 could also be responsible for the observed shifts. Despite causing shifts in unique residue resonances, both compounds alter residues near the purported metal binding site of Aβ (residues 1-16) suggesting that the compounds preferentially interact with the N-terminus and may be oriented in a manner which promotes interaction with metal ions when present in solution or could alter the conformation of the N-terminus of the peptide, disrupting the peptide's ability to bind metal ions 4,[8][9][10] . Phlorizin (Fig. 5e) caused minor chemical shifts in both the hydrophobic core (F19-A21) and the C-terminus (V36); nonpolar forces may be responsible for any prospective Phlorizin-Aβ 40 interaction. Rutin (Fig. 5f) prompted small shifts in regions similar to those altered by Phlorizin (L17 and M35). These small and non-localized chemical shifts could be indicative of weak interaction and/or non-specific binding to the peptide by the two compounds. Both Phlorizin and R2, which show limited or no influence on amyloidogenesis, display potentially weak interactions with more C-terminal residues while the interaction of both Verbascoside and VPP with Aβ appear to be more centrally or N-terminally located. Due to the modest CSP values observed, these data may suggest a general preference for interaction rather than a conventional, structured binding site for all compounds examined. Our NMR studies, therefore, suggest a possible molecular mechanism responsible for their differing activities towards Aβ species in vitro (vide supra). Ligand association with the N-terminus (close to the metal binding site in Aβ) may be favorable in the development of multifunctional compounds to target metal-free and/or metal-associated Aβ species while interacting with the central hydrophobic residues of Aβ may also allow for general inhibition, as has been previously predicted 4,6,9,10,15 . To gain further molecular level insight into the interactions between these polyphenols and different Aβ structures, saturation transfer difference (STD) NMR was employed to map the regions of both Verbascoside and VPP which bind to preformed Aβ 42 fibrils (Fig. 6) 18,28,49,50 . The intensity of peaks within the STD spectra, relative to the reference spectra, is associated with the proximity of the ligand's protons to the fibril 50 . From these values, a group epitope map can be generated, determining which protons of Verbascoside and VPP are in proximity to and binding with the fibrils. Verbascoside (Fig. 6c) only showed STD signals for protons linked to the glucose and rhamose rings. While most of these protons presented modest to weak signals in the STD spectrum, the protons on the C6 of the glucose ring indicated an extremely strong STD effect. For VPP, weak STD effect was observed around all methyl and ethyl protons of the esters appended to the sugar moieties. The stronger STD effect was observed around both the caffeic acid and hydroxytyrosol moieties, localized especially around the two aromatic protons of the catechol groups (Fig. 6d). This suggests that the esterification of the sugar rings of Verbascoside blocks the binding of the compound to the fibrils through the two sugar rings, redirecting their interaction to a different portion of the compound. It may be that the multiple hydroxyl groups of the sugars promote hydrogen bonds with some of the exposed side chains of the fibril; the esterification of the sugar rings could make these interactions sterically unfavorable and direct VPP to interact through hydrogen bonds and π -π interactions through the catechol moieties of the compound instead.</p><p>The different binding modes identified by STD experiments led us to explore the affinity of these two compounds for fibrillar Aβ. Using a fluorescence blue shift assay, the change in the fluorescence emission wavelength was monitored as fibril was titrated into solution (Fig. S6a and b). Verbascoside and VPP were shown to have an affinity of 7.80 ± 0.75 μ M and 6.98 ± 0.80 μ M, respectively, under the condition employed in this study. Despite very different modes of binding to the fibril, the compounds have nearly identical affinities. It should be noted that the actual affinity for the fibril in solution is likely stronger than those measured; the concentration of fibrils was calculated based on the monomer equivalent concentration. It is not possible to measure the molar concentration of fibrils in solution given their heterogeneous length and subsequently heterogeneous molecular weight. The fibril concentration is likely much lower than the monomer equivalent which would affect the concentration values used in the fit. Because the exact fibril concentration cannot be measured, however, and because the same sample of fibril was used for both titrations, we are confident that these results are internally consistent and suggest that the two compounds have very similar affinities.</p><p>The distinct difference in group epitope but similar affinity led us to investigate the binding site on the fibril for these two compounds; it was expected that two compounds with similar affinities but unique binding moieties would likely bind to separate regions of the fibril structure and therefore do not compete for binding. To probe this, we performed a competition titration experiment in which fibrils were first treated with 1 equiv of a compound (either Verbascoside or VPP) and then titrated with the other compound. The intensity of an STD peak associated with each compound (relative to the intensity of the same peak in the reference spectrum) was then used to monitor the binding of each compound to the fibril. As the competing compound is titrated into solution, if the compounds bind in the same site, it is expected that the relative intensity of the first compound's peak will gradually decrease while the relative intensity of the titrant's peak will increase as it competes the first compound off of the binding site. Conversely, under non-competitive binding conditions, the initial compound would maintain a constant relative intensity while the relative intensity of the titrant would increase as it binds at higher and higher concentrations. When Verbascoside was titrated into a solution already containing VPP and fibrils (Fig. S6c), it was observed that the relative intensity of the peak associated with Verbascoside increased during the titration while the relative intensity of the peak associated with VPP gradually decreased. This suggests that Verbascoside competes with VPP for a similar binding site on the fibril. The reverse of this experiment was performed (VPP titrated into a solution already containing Verbascoside and fibrils; Fig. S6d) and the same trend was observed. The relative intensity of the VPP peak increased while the relative intensity of the Verbascoside peak decreased slightly. This implies that, while the two compounds bind to the fibril using unique portions of their related structure, they are targeting a similar location on the fibril. Additionally, the blue shift assay suggests that they do so with a nearly identical affinity. This is unexpected and suggests that the esterification changes how the two compounds are able to disaggregate fibrillar species (vide supra) but they bind to the fibril in a relatively similar manner though using unique parts of their structure. It could be that the differing orientation of the sugar and catechol moieties of the two compounds is at least partly responsible for this difference in reactivity as STD group epitope maps reveal that the orientation is unique between compounds.</p><!><p>Because oxidative stress is believed to play a role in AD, it would be valuable for multifunctional probes to possess antioxidant activity, on top of the ability to interact simultaneously with Aβ species and various transition metal ions 9,13,14 . The ability of these six compounds to scavenge organic radical cations (i.e., ABTS •+ ) was determined by the Trolox equivalence antioxidant capacity (TEAC) assay using cell lysates (Fig. 7a) 51,52 . Both Verbascoside and Rutin, two known antioxidants 31,32,52 , scavenged ABTS •+ slightly better than Trolox (6-hydroxy-2,5,7,8-tetramethylchroman-2-carboxylic acid) (by a factor of ca. 1.4 and 1.2, respectively). VPP, Phlorizin, and F2 showed a lower ability to scavenge ABTS •+ relative to Trolox, suggesting limited function as antioxidants relative to the other compounds investigated herein. R2 presented no ability to scavenge ABTS •+ . This, again, indicates that esterification of the sugar moiety transforms the function of the polyphenols. In this case, the antioxidant capacity of the all compounds was reduced upon esterification. Verbascoside's activity was reduced by ca. 65% while Rutin lost all activity. The activity of Phlorizin was reduced by ca. 40%. This reduction in antioxidant capacity is surprising given the conservation of the catechol structures between Verbascoside/VPP and Rutin/R2 which are thought to be potentially responsible for ABTS •+ quenching through semi-quinone and quinone formation 28,52 .</p><!><p>Verbascoside and Rutin have suggested the two compounds may alleviate the toxicity of Aβ species in SH-SY5Y and APPswe cells, respectively 22,24 . We probed this relationship further to examine the effect of these two compounds, as well as VPP, on the cytotoxicity of both metal-free Aβ and metal-Aβ in murine Neuro-2a neuroblastoma (N2a) cells. In the absence of Aβ, Verbascoside was relatively nontoxic with and without metal ions (Fig. S7). VPP, however, reduced cell viability (ca. 70%) at high concentrations in the presence of either Cu(II) or Zn(II), but showed minimal toxicity in the absence of metal ions. Rutin had no significant impact on cell viability under any conditions (Fig. S6).</p><p>Cells incubated with Aβ (20 μ M) in either the absence or presence of metal ions (Cu(II) or Zn(II), 20 μ M) indicated viability of ca. 70-80% under all conditions (Fig. 7b). The addition of Verbascoside (20 μ M) improved cell survival (ca. 90-100%) regardless of both Aβ isoforms and the presence or absence of metal ions. VPP (20 μ M), however, was generally unable to regulate Aβ-triggered cytotoxicity. Finally, treatment of cells with Rutin (20 μ M) yielded slight improvements in cell survival under all conditions. Overall, Verbascoside was capable of attenuating the broad toxicity indicated in the presence of both metal-free Aβ and metal-Aβ. Furthermore, esterification of the compound is shown to limit these protective capabilities.</p><!><p>Naturally occurring polyphenolic glycosides (Phlorizin, Verbascoside, and Rutin), along with their esterified derivatives (F2, VPP, and R2), were investigated for their potential to modulate the aggregation and toxicity of metal-free Aβ and metal-Aβ. Both Verbascoside and VPP, bearing multiple catechol moieties, were capable of distinctly redirecting the aggregation of metal-free Aβ (mainly, VPP) and/or metal-Aβ to different extents, as confirmed by both biochemical and TEM studies. The ability of these two compounds to interact with both metal ions and Aβ was confirmed through physical methods, including UV-Vis and 1D/2D/STD NMR. Verbascoside and VPP are able to interact with both Cu(II) and Zn(II) as well as with multiple forms of Aβ. The esterification of Verbascoside to VPP distinctly alters the interactions between the compounds and the components of the metal-Aβ system studied herein, suggesting that the properties (metal binding and Aβ interaction) can be tunable by synthetic modifications. In addition, Verbascoside is shown to have some promising chemical properties for a potential probe (antioxidant capacity, no toxicity, and cytoprotective against both metal-free Aβ and metal-Aβ). Overall, this study points to Verbascoside as a promising starting point for constructing a new multifunctional probe to interrogate the etiology of AD.</p><p>This study also lends insight into the efficacy of unique chemical moieties in design of anti-amyloidogenic compounds. Comparing Phlorizin, Verbascoside, and Rutin to each other, there is a direct correlation between the number of catechol moieties and the efficacy of each compound against metal-free Aβ and metal-Aβ aggregation. This is further supported by the group epitope map generated from STD NMR for VPP which shows strong STD effect around both catechols. This suggests that the catechol moiety specifically, not just polyphenols as a general class of compounds, may be effective at redirecting the protein misfolding. Previous studies have also indicated that including a catechol-like moiety makes a small molecule more effective; 28 two of the most thoroughly investigated anti-amyloidogenic natural products, EGCG and curcumin, contain multiple variations of the catechol moiety (pyrogallol for EGCG, o-methyl-catechol for curcumin), further indicating that the catechol moiety is an effective component of an anti-amyloidogenic probe 20,53 . It should be noted, however, that catechol-containing compounds have been shown to be promiscuous in their action and functionality so their use and functionalization must be carefully considered in compound design for in vivo applications in order to avoid off-target effects 54,55 .</p><p>We have also highlighted the importance of carefully chosen synthetic alterations to efficacious reagents. In this suite of compounds, it is apparent that esterification of the sugar moieties affected the efficacy of the compounds in the various assays to differing extents; esterification increased the ability of VPP to alter the structure of metal-free Aβ aggregates while simultaneously maintaining its affinity for preformed fibrils relative to that of Verbascoside, its parent compound. Esterification also reduced the ability of all compounds to scavenge organic radicals despite not directly modifying the catechol moiety thought to be responsible for the scavenging activity. Overall, this points to the unpredictability inherent in small molecule design for complex biological systems; with many variables all</p><!><p>A A</p><!><p>A A contributing to the disease phenotype, it is challenging, if not impossible, to accurately predict the effects of small structural changes on a compounds efficacy against the suite of potential targets. Thus, the performance of the compounds evaluated here may serve as a benchmark for which compounds are most worthwhile investigating further in more complex and biologically relevant systems. Furthermore, it indicates the care which must be taken when functionalizing known compounds. Even potentially reversible changes, such as esterification, may drastically alter the compound's function, in both beneficial and detrimental ways. Importantly, however, it appears from the data presented herein that Verbascoside possesses positive characteristics for a compound in the investigation of the role of Aβ in AD and, furthermore, is structurally amenable to a variety of future derivatization aimed at further improving its function.</p><!><p>Materials and Procedures. All reagents were purchased from commercial suppliers and used as received unless otherwise noted. The natural products (Phlorizin, Verbascoside, and Rutin) and their previously synthesized, esterified derivatives (F2, VPP, and R2) were prepared following the previously reported methods [30][31][32] . Aβ 40 and Aβ 42 were purchased from Anaspec (Fremont, CA, USA). Transmission electron microscopy (TEM) images were recorded on a Philips CM-100 transmission electron microscope (Microscopy and Image Analysis Laboratory, University of Michigan, Ann Arbor, MI, USA). Optical spectra for metal binding were recorded on an Agilent 8453 UV-Visible (UV-Vis) spectrophotometer. Nuclear magnetic resonance (NMR) spectra for the characterization of Zn(II) binding studies of Verbascoside and VPP were acquired on an Agilent 400 MHz NMR spectrometer. NMR studies of 15 N-labeled Aβ 40 with ligands were carried out on a Bruker 600 MHz NMR spectrometer equipped with a cryogenic probe. Absorbance values for biological assays, including the TEAC assay and cell viability assay, were measured on a SpectraMax M5 microplate reader (Molecular Devices, Sunnyvale, CA, USA).</p><!><p>Aβ experiments were performed according to previously published methods 28,39 . Prior to the sample preparation, Aβ 40 or Aβ 42 was dissolved with ammonium hydroxide (NH 4 OH, 1% v/v, aq), aliquoted, lyophilized, and stored at − 80 °C. Stock solutions (ca. 200 μ M) of Aβ 40 and Aβ 42 were prepared by dissolving the lyophilized peptide in 1% NH 4 OH (10 μ L) and diluting with doubly distilled (dd) H 2 O. The peptide stock solution was diluted to a final concentration of 25 μ M in buffered solution containing HEPES [4-(2-hydroxyethyl)-1-piperazineethanesulfonic acid; 20 μ M, pH 6.6 for Cu(II) samples; pH 7.4 for metal-free and Zn(II) samples] and NaCl (150 μ M). For the inhibition studies 28,39 , compound (50 μ M final concentration, 1% v/v DMSO) was added to the sample of Aβ 40 or Aβ 42 in the absence and presence of a metal chloride salt (CuCl 2 or ZnCl 2 , 25 μ M) followed by incubation for 4 and 24 h at 37 °C with constant agitation. For the disaggregation studies 28,39 , Aβ 40 or Aβ 42 with and without metal ions was first incubated for 24 h at 37 °C with continuous agitation prior to the addition of compound (50 μ M). The resulting samples were incubated for an additional 4 or 24 h at 37 °C with constant agitation.</p><p>Gel Electrophoresis with Western Blotting. The samples from the inhibition and disaggregation experiments were analyzed by gel electrophoresis with Western blot using an anti-Aβ antibody (6E10) 20,28 . Each sample (10 μ L) was separated on a 10-20% Tris-tricine gel (Invitrogen, Grand Island, NY, USA). Following separation, the gel was transferred onto nitrocellulose membrane which was blocked with bovine serum albumin (BSA, 3% w/v, Sigma-Aldrich, St. Louis, MO, USA) in Tris-buffered saline (TBS) containing 0.1% Tween-20 (TBS-T) for 3 h at room temperature. The membrane was treated with antibody (6E10, Covance, Princeton, NJ, USA; 1:2000) in a solution of BSA (2% w/v) in TBS-T overnight at 4 °C. Following washing, the membrane was treated with horseradish peroxidase-conjugated goat antimouse secondary antibody (1:5000; Cayman Chemical, Ann Arbor, MI, USA) in 2% BSA in TBS-T solution for 1 h at room temperature. Protein bands were visualized using ThermoScientific Supersignal West Pico Chemiluminescent Substrate (Thermo Scientific, Rockford, IL, USA).</p><!><p>The samples for TEM were prepared following a previously reported method 28,39 . Glow-discharged grids (Formar/Carbon 300-mesh, Electron Microscopy Sciences, Hatfield, PA, USA) were treated with the samples from the disaggregation experiments (5 μ L, 25 μ M Aβ) for 2 min at room temperature. Excess sample was removed with filter paper and washed with ddH 2 O. Each grid was stained with uranyl acetate (1% w/v, ddH 2 O, 5 μ L, 1 min), blotted to remove excess stain, and dried for 15 min at room temperature. TEM images were taken by a Philips CM-100 transmission electron microscope (80 kV, 25,000× magnification).</p><p>Metal Binding Studies. The interaction of Phlorizin, F2, Verbascoside, VPP, Rutin, and R2 with Cu(II) and Zn(II) was determined by UV-Vis or 1 H NMR, respectively, based on previously reported procedures 20,56 . A solution of ligand (20 μ M, pH 7.4) was prepared, treated with 0.5 to 10 equiv of CuCl 2 , and incubated at room temperature for 2 h (for Phlorizin, F2, Verbascoside, VPP, Rutin, and R2). The optical spectra of the resulting solutions were measured by UV-Vis. The interaction of both Verbascoside and VPP with ZnCl 2 was observed by 1 ). Each spectrum was obtained using 64 complex t 1 points and a 0.1 sec recycle delay on a Bruker Avance 600 MHz spectrometer. The 2D data were processed using TOPSPIN 2.1 (from Bruker). Resonance assignment was performed with SPARKY 3.1134 using published assignments for Aβ 40 as a guide [57][58][59] . Chemical shift perturbation (CSP) was calculated using the following equation (Eq. 1):</p><p>Saturation Transfer Difference (STD) NMR Spectroscopy. For the STD NMR 48 experiments, an 150 μ M solution of fibrillar Aβ 42 was prepared by incubating Aβ 42 for 24 h at 37 °C with constant agitation in 10 mM deuterated Tris-DCl, 95% D 2 O at pD 7.4 (corrected for the isotope effect). The samples for STD experiments were prepared by diluting fiber to 1 μ M (effective monomer concentration) into 10 mM deuterated Tris-DCl to which was added 250 μ M of ligand (0.5% DMSO-d 6 ). STD experiments were acquired with a train of 50 dB Gausian-shaped pulses of 0.049 sec with an interval of 0.001 sec at either − 1.0 ppm (on resonance) or 40 ppm (off resonance) with a total saturation time of 2 sec on a Bruker 600 MHz NMR spectrometer. A total of 1024 scans were recorded for the STD spectrum and 512 scans were recorded for the reference spectrum at 25 °C. An inter-scan delay of 1 sec was used for both the STD and the reference experiments.</p><p>For the competition experiments with Verbascoside and VPP, the above procedure was followed for sample preparation. STD experiments were acquired with a train of 50 dB Gausian-shaped pulses of 0.049 sec with an interval of 0.001 sec at either − 1.0 ppm (on resonance) or 40 ppm (off resonance) with a total saturation time of 2 sec on a Bruker 500 MHz NMR spectrometer. A total of 2048 scans were recorded for the STD spectrum and 1024 scans were recorded for the reference spectrum at 25 °C. An inter-scan delay of 1 sec was used for both the STD and the reference experiments. To a solution already containing either Verbascoside or VPP (250 μ M), the other compound was titrated to 0.5 equiv (125 μ M), 1 equiv (250 μ M), and 3 equiv (750 μ M). The intensity of peaks unique to Verbascoside (3.85 ppm) and VPP (7.78 ppm) in the STD spectra relative to their intensity in the reference spectra were used to monitor the binding of the compounds to the fiber.</p><p>Blue Shift Fluorescence Assay. The change in the fluorescence emission wavelength of Verbascoside or VPP was monitored upon treatment with Aβ 42 fibrils on a Fluoromax-4 Spectrofluorimeter (Horiba Scientific, Edison, NJ, USA). A 25 μ M solution of either Verbascoside or VPP in buffer (20 mM PO 4 , pH 7.4, 50 mM NaCl) was titrated with preformed Aβ 42 fibrils, prepared as described above for STD experiments. The fluorescence emission was monitored between 380 and 550 nm following excitation (350 nm for Verbascoside and 330 nm for VPP) with slits setting for 5 nm bandwidths. The blue shift was calculated by the difference between the emission maximum wavelength of the titration point and the emission maximum wavelength of the compound in absence of fibrils. The data was then fit to a hyperbolic curve to calculate the K d value.</p><p>Trolox Equivalent Antioxidant Capacity (TEAC) Assay. The antioxidant activity of Phlorizin, F2, Verbascoside, VPP, Rutin, and R2 was determined by the TEAC assay employing cell lysate following the protocol of the antioxidant assay kit purchased from Cayman Chemical Company (Ann Arbor, MI, USA) with modifications 18 . Murine Neuro-2a (N2a) cells were used for this assay. This cell line, purchased from the American Type Culture Collection (ATCC, Manassas, VA, USA), was maintained in media containing 50% Dulbecco's modified Eagle's medium (DMEM) and 50% OPTI-MEM (GIBCO), supplemented with 10% fetal bovine serum (FBS, Sigma), 1% Non-essential Amino Acids (NEAA, GIBCO), 2 mM glutamine, 100 U/mL penicillin, and 100 mg/mL streptomycin (GIBCO). The cells were grown and maintained at 37 °C in a humidified atmosphere with 5% CO 2 . For the antioxidant assay using cell lysates, cells were seeded in a 6 well plate and grown to approximately 80-90% confluence. Cell lysates were prepared following the previously reported method with modifications 60 . N2a cells were washed once with cold PBS (pH 7.4, GIBCO) and harvested by gently pipetting off adherent cells with cold PBS. The cell pellet was generated by centrifugation (2,000 × g for 10 min at 4 °C). This cell pellet was sonicated on ice (5 sec pulses, 5 times with 20 sec intervals between each pulse) in 2 mL of cold Assay Buffer (5 mM potassium phosphate, pH 7.4, containing 0.9% NaCl and 0.1% glucose). The cell lysates were centrifuged at 5,000 × g for 10 min at 4 °C. The supernatant was removed and stored on ice until use. To standard and sample wells in a 96 well plate, cell lysates (10 μ L) were delivered; they were followed by addition of compound, metmyoglobin, ABTS, and H 2 O 2 in order. After 5 min incubation at room temperature on a shaker, absorbance values at 750 nm were recorded. The final concentrations (0.045, 0.090, 0.135, 0.180, 0.225, and 0.330 mM) of Trolox (Sigma-Aldrich; dissolved in DMSO) and all polyphenolic glycosides were used. The percent inhibition was calculated according to the measured absorbance [% Inhibition = (A 0 -A)/A 0 , where A 0 is absorbance of the supernatant of cell lysates] and was plotted as a function of compound concentration. The TEAC value of ligands was calculated as a ratio of the slope of the standard curve of the compound to that of Trolox.</p><!><p>Cell viability upon treatment with compounds was determined using the MTT assay (Sigma). N2a cells were seeded in a 96 well plate (15,000 cells in 100 μ L per well). The cells were treated with Aβ (20 μ M) with or without either CuCl 2 or ZnCl 2 (20 μ M), followed by the addition of compound (20 μ M, 1% v/v final DMSO concentration for Verbascoside, VPP, and Rutin) and incubated for 24 h in the cells. After incubation, 25 μ L MTT [5 mg/mL in phosphate buffered saline (PBS), pH 7.4, GIBCO, Grand Island, NY, USA] was added to each well and the plate was incubated for 4 h at 37 °C. Formazan produced by the cells was solubilized using an acidic solution of N,N-dimethylformamide (DMF, 50% v/v, aq) and sodium dodecyl sulfate (SDS, 20% w/v) overnight at room temperature in the dark. The absorbance was measured at 600 nm using a microplate reader. Cell viability was calculated relative to cells treated an equivalent volume of DMSO.</p>

Scientific Reports - Nature

Goal Setting to Promote a Health Lifestyle

The purpose of this parallel-group study was to determine whether a feasibility study based on newsletters and telephone counseling would improve goal-setting constructs; physical activity (PA); and fruit and vegetable (F & V) intake in a sample of older adults. Forty-three older adults (M age = 70 years, >70% Asian, 54% female) living in Honolulu, Hawaii were recruited and randomly assigned to either a PA or F & V intake condition. All participants completed measures of PA, F & V intake, and goal setting mechanisms (i.e., specificity, difficulty, effort, commitment, and persistence) at baseline and 8-weeks. Paired t-tests were used to evaluate changes across time. We found that F & V participants significantly increased F & V intake and mean scores of goal specificity, effort, commitment, and persistence (all p < .05). No statistically significant changes in PA or goal setting mechanisms were observed for participants in the PA condition. Overall, our results show that a short-term intervention using newsletters and motivational calls based on goal-setting theory was effective in improving F & V intake; however, more research is needed to determine whether these strategies are effective for improving PA among a multiethnic sample of older adults.

goal_setting_to_promote_a_health_lifestyle

1. Introduction<!>Study Population<!>Study Design<!>Telephone Counseling<!>Newsletters<!>Study Measures<!>Data Analysis<!>Descriptive Characteristics<!>Change in F & V intake and PA<!>Change in goal-setting mechanisms/moderators<!>4. Discussion

<p>Goal setting is an important element in physical activity (PA) and dietary interventions. Goal-setting techniques are effective in helping individuals initiate and maintain health behavior over time (Shilts, Horowitz, & Townsend, 2004). Meta-analytic results suggest that participants in goal-setting conditions significantly increase fiber intake, report fewer dropouts in physical activity interventions, demonstrate higher exercise adherence, and reduce dietary sodium intake (Shilts et al., 2004). Setting specific and challenging goals leads to improved health behavior. Constructs of goal setting theory such as effort toward goal-related activities, persistence (i.e., time and intensity spent on goal-related tasks), and commitment (i.e. one's attachment or determination to reach a goal), have been shown to mediate and/or moderate goal setting and outcomes (Locke & Latham, 2002, 2006). In particular, the relation between goal setting and performance is strongest when people set specific challenging goals, are committed, expend effort to accomplish goals, persist despite barriers, and have higher levels of confidence.</p><p>Previous interventions based on goal-setting have not assessed whether these interventions are associated with improvements in mechanisms of the goal-setting process. Therefore, the purpose of this parallel-group intervention study was to assess whether participating in a goal-setting intervention is associated with changes in theoretically-related mechanisms that influence behavior change. Specifically, older adults aged55 and older were randomized to a goal-setting intervention for PA or fruit and vegetable (F & V) intake to determine whether interventions produce differential effects in goal specificity (i.e., the degree of precision with which the aim is specified); goal difficulty (i.e., the degree of complexity needed to accomplish a specified task); effort; commitment; persistence; and task-specific (i.e., beliefs in one's ability to perform a specific task) and barrier self-efficacy (i.e., belief in one's ability to overcome barriers to a specific task).</p><!><p>Study participants were recruited in October 2006 from a senior fair in Honolulu, HI. Eligible participants were > 55 years of age and able to participate in physical activity ≥ 10 minutes at a moderate-intensity pace and consume a diet high in fiber. The Institutional Review Board of the University of Hawaii approved the study protocol.</p><!><p>A randomized, parallel-group experimental design was used. Participants were randomized to either a PA or F & V goal-setting condition and were given $10 gift cards as incentives for participation. Participants in both condition received telephone counseling and group tailored newsletters.</p><!><p>Participants received calls at baseline and 4-weeks. During each call, the intervention specialist: (a) helped the participant set/evaluate a short-term goal(s); (b) discussed/reviewed the importance of specificity and difficulty in the goal-setting process; and (c) discussed/evaluated barriers and ways to overcome each barrier. The format for the second call was similar with the additional introduction of the concepts of effort, commitment, and persistence. Call times ranged from 15 to 25 minutes.</p><!><p>Newsletters were mailed out after the baseline assessment and telephone counselling in weeks 5 and 9. The initial newsletters reinforced the telephone counseling calls and provided information on the benefits of PA or F&V consumption. The second provided strategies to overcome barriers to meeting goals to PA or F&V.</p><!><p>All outcome measures and mechanisms of the goal-setting process were assessed by trained staff during the baseline assessment and 8-week follow-up intervention. PA was measured using an adaptation of the Godin Leisure-Time Exercise Questionnaire (GLTEQ), which assessed mild (i.e., takes minimal effort), moderate (i.e., increases your heart and breathing rate a little), and strenuous activity (i.e., caused one's heart to beat rapidly) for at least 30 minutes at a time. The GLTEQ has adequate test-retest reliability and is significantly associated with objective measures of physical activity (Godin, Jobin, & Bouillon, 1986). F & V consumption was assessed by asking two questions: "How many servings of fruits do you eat each day?" and "How many servings of vegetables do you eat each day?" The validity of the two-item instrument was established in a separate study by significant correlations with a longer 19-item inventory and 24-hour dietary recalls (Resnicow et al., 2000).</p><p>The original goal specificity and difficulty measures were designed to take into account the multi-component nature of physical activity by rating items in terms of frequency, type, duration, and intensity of PA or F&V intake per week. Each subscale was rated on a six-point, Likert-type scale. A score of zero was provided for participants who reported that they were not setting goals. The remaining scale was rated from one (strongly disagree) to five (strongly agree). The original goal specificity and difficulty measures demonstrated adequate reliability, factorial validity, and predictive validity in a previous study (Frahm-Templar, Estabrooks, & Gyurcsik, 2003). Author-created subscales used to assess effort, commitment, and persistence were created for each concept based on the conceptual definitions proposed by Locke and Latham (1990). Sample items for effort, commitment, and persistence, respectively, include: "I'm going to put a lot of effort into reaching my goals," "I am committed to pursuing my exercise goals," and "Even if I don't reach my short-term exercise goals, I will continue to pursue them." Each subscale was rated on a six-point, Likert-type scale. A score of zero was provided for participants who reported that they were not setting goals. The remaining scale was rated from one (strongly disagree) to five (strongly agree). The investigator-created scales appeared to have adequate face validity (E. Locke, personal communication, 2006).</p><!><p>Chi-square and t-tests were used to determine whether intervention conditions differed in terms of demographic variables. Paired t-tests were used to determine whether groups improved significantly on study outcomes and goal-setting mediators and moderators. Between-group differences over time were examined with the nonparametric Kruskal-Wallis test. In addition, adjusted (i.e., partial) correlation coefficients were computed to determine whether changes in mechanisms (e.g., specificity and difficulty) of goal-setting theory were associated with changes in PA and F & V consumption. Correlations were adjusted for baseline measures of PA or F & V intake. These data were analyzed using SAS version 9.13 software, and statistical significance (p < 0.05) was determined using two-sided tests.</p><!><p>Twenty-one participants were randomized to the PA condition, and 22 participants were randomized to the F & V condition. No significant differences were observed between study condition with respect to descriptive characteristics (all p > .05; see Table 1).</p><!><p>F & V participants significantly increased the mean servings of fruits and vegetables consumed (p < .05), whereas PA participants did not significantly increase physical activity (p > .05). Interestingly, PA participants significantly improved F&V consumption (p > .05). The treatment-by-time interactions were not significant (p > .05, see Table 2)</p><!><p>F&V participants significantly increased goal mechanisms throughout the study period (all p < 0.05), whereas no significant improvements in goal mechanisms were observed among PA participants (all p > 0.05). The change in goal effort, commitment, and persistence among participants in the F&V condition was significantly greater than that of the PA condition.</p><!><p>In this study, we found that participants randomized to a F & V intervention significantly improved F & V intake and constructs related to the goal-setting process. We are remised to say that the change in F&V intake was not significantly different between conditions. The results from our research suggest that goal setting can be an effective strategy to improve dietary intake and mechanisms of goal-setting theory in an ethnically diverse sample of older adults.</p><p>The results we observed were similar to those observed in other behavioral interventions that demonstrated improvements in dietary behavior but not PA (Resnicow et al., 2004). We believe that the reduction observed in PA could be due to participants reporting at the upper ends of the exercise scale and gradual "regression to the mean" was observed. In addition, maintaining high levels of PA or improving PA may require greater preparation and resources than that needed for increasing F & V consumption.</p><p>Interestingly, the PA participants increased their F & V consumption from baseline to follow-up, which may be due, in part, to alternate goal specification when previously set goals are thwarted.</p><p>The results among the F & V participants were similar to those observed in previous research examining goal-setting theory in health promotion studies (Kyllo & Landers, 1995; Shilts et al., 2004). For example, Gyurcsik and colleagues (2003) found that goal specificity and difficulty (e.g., setting challenging PA goals) were associated with aquatic exercise attendance, whereas Dishman and colleagues (2009) found that setting challenging goals and higher levels of self-efficacy and commitment were significantly associated with greater increases in PA. Although change in F&V intake was not significantly different from that of the PA condition, a 1.85 serving increase is large in magnitude and provides evidence that goal-setting is a consistent determinant of health behavior change.</p><p>Overall, our results indicate that the theoretical constructs of goal-setting theory are associated with changes in health behavior. Although our sample sizes were small and all of our instruments were self-report, our results show that a short-term, inexpensive goal-setting intervention using motivational calls and newsletters can be effective in significantly improving F & V intake.</p>

PubMed Author Manuscript

An Efficient Computational Model to Predict Protonation at the Amide Nitrogen and Reactivity along the C\xe2\x80\x93N Rotational Pathway

N-protonation of amides is critical in numerous biological processes, including amide bonds proteolysis and protein folding, as well as in organic synthesis as a method to activate amide bonds towards unconventional reactivity. A computational model enabling prediction of protonation at the amide bond nitrogen atom along the C\xe2\x80\x93N rotational pathway is reported. Notably, this study provides a blueprint for the rational design and application of amides with a controlled degree of rotation in synthetic chemistry and biology.

an_efficient_computational_model_to_predict_protonation_at_the_amide_nitrogen_and_reactivity_along_t

<p>The amide bond is one of the most important functional groups in chemistry and biology.1 The nN → π*C=O conjugation and the resulting planarity controls the vast majority of chemico-physical properties of amides (Fig. 1).2 In contrast to planar amides, distorted amides have received much less attention3 despite their profound implications for structure,4 reactivity5 and wide significance in biology and medicinal chemistry6 (amide bond proteolysis,6a isomerization of cis-trans peptides,6b protein splicing,6c β-lactam antibiotics,6d conformational preferences of peptides,6e generation of new pharmacophores and lead structures6f). More fundamentally, non-planar amides that are characterized by ground-state distortion provide crucial insight into the amide bond resonance, including the long-standing question of structural factors governing the O- vs. N-protonation switch.7 N-protonation of amides is critical in numerous biological processes, including amide bonds proteolysis8 and protein folding,9 as well as in organic synthesis as a method to activate amide bonds towards unconventional reactivity.10,11 N-protonation results in disruption of the amide bond resonance, which has been used to catalyze isomerization of amides12 and promote reactions of C–N bonds adjacent to the carbonyl group.10,11 At present, the structural factors that govern the N- vs. O-protonation aptitude are poorly understood.</p><p>Herein, we present an efficient computational model that allows to predict the likelihood that a given amide can be protonated at the nitrogen atom, and disclose how one may predict the reactive properties of amides along the C–N rotational pathway from simple calculation of structural properties. This approach makes the amide bond distortion models developed by Greenberg13 generally applicable to accurately predict the reactive properties of amide linkages.1–12 We expect that these findings will enable synthetic and medicinal chemists to utilize amides with a controlled degree of rotation as versatile intermediates and target lead structures in organic synthesis.14,15</p><p>We designed a model series of amides in which the overall sum of carbon atoms forming the core scaffold is between five and ten (Scheme 1). These compounds are based on the highly promising one-carbon bridged 1-azabicyclo scaffold (Fig 1c).10,11 The compounds have been selected on the basis of their synthetic accessibility and distortion range of amide bonds. The conceptual advantage of using one-carbon bridged twisted amides (cf. two-carbon or larger bridge analogues)13a,b stems from their better hydrolytic profile,16 their successful use as a platform for discovery of new reactivity of amide bonds,10,11 and accessibility of diverse ring systems that span the whole spectrum of the amide bond distortion.17 Crucially, our approach determines the relationship between distortion and energetic parameters in readily-accessible amide models that (i) are non-planar in the ground-state conformation;1–5 and (ii) have already been synthesized in a laboratory environment and validated experimentally (cf. two-carbon or larger bridge analogues).13a,b As a key design element, we hypothesized that mapping the amide bond distortion along energetic parameters of amide bonds over sufficiently large distortion range would allow for elucidating the role of strain and the nN → π*C=O delocalization on the N-protonation of amide linkages.</p><p>We first established geometric changes that occur upon the amide bond rotation in the series.18 Total energies as well as selected structural parameters are listed in Table 1. The results validate the use of small- and medium-sized one-carbon bridged twisted amides10,11 as models for a systematic study of geometric changes that occur during the amide bond rotation.1–12 The selected set covers the whole spectrum of the amide bond distortion geometries. The twist angle changes from essentially planar to fully perpendicular (τ = 11.32-90.0°), while the pyramidalization at nitrogen varies from essentially sp2 to fully sp3 hybridized nitrogen (χN = 5.99-71.44°). In contrast, the pyramidalization at carbon remains relatively unchanged (χC = 0.0-15.42°) with the carbonyl carbon essentially planar in the series, which is consistent with previous observations regarding amide bond distortion.3,4 In agreement with the amide bond distortion parameters, the length of the N–C(O) bond varies between 1.474 Å and 1.365 Å, while the length of the C=O bond is between 1.201 Å and 1.233 Å.</p><p>The approach verifies structural changes that occur with the N–C(O) bond rotation as a function of the twist angle and the pyramidalization at nitrogen. Most importantly, a plot of N–C(O) bond length versus the sum of twist angles and nitrogen pyramidalization (Στ+χN) gives an excellent linear correlation over the observed range of amide bond geometries (R2 = 0.97, Fig. 2). Plots of N–C(O) bond length versus twist angle (R2 = 0.87) and nitrogen pyramidalization (R2 = 0.88) give scattered correlations in the series (not shown). In agreement with the classical resonance,1,2,7c disruption of the nN → π*C=O delocalization results in a significant lengthening of the N–C(O) bond, while the C=O bond experiences minor shortening.</p><p>Importantly, the results demonstrate that description of the amide bond distortion by additive twist angle/pyramidalization at nitrogen parameters (i.e. the sum of twist and nitrogen pyramidalization angles (Στ+χN)) provides a more accurate representation of the geometric properties than the twist angle or the pyramidalization at nitrogen alone. Previous studies demonstrated that 1-azabicyclo[3.3.1]nonan-2-one (X-ray of a phenyl analogue: τ = 20.8°; χN = 48.8°; χC = 5.9°; Στ+χN = 69.6°)10c and derivatives of one-carbon bridged amides containing the [4.3.1] ring system (X-ray of an aryl analogue: τ = 42.8°; χN = 34.1°; χC = 16.5°; Στ+χN = 76.9°)10b feature amide bonds that are predominantly protonated at nitrogen.10,11 It was thus proposed that a twist angle of as low as approx. 40-50° may be sufficient to promote N-protonation of amide bonds.10a–c Additionally, earlier reports by Brown demonstrated that extended 2-quinuclidone analogues favor N-alkylation ([3.2.2] amide ring system, X-ray: τ = 33.2°; χN = 52.8°; χC = 11.0°; Στ+χN = 86.0°) or O-alkylation ([3.3.2] ring system, X-ray: τ = 15.3°; χN = 38.6°; χC = 4.3°; Στ+χN = 53.9°);3a detailed experimental results not disclosed. The additive (Στ+χN) parameter normalizes these results and reveals that a (Στ+χN) value of 60-70° appears to be close to a barrier between N- vs. O-protonation of amides. Note that this value is much lower than the distortion that would correspond to a fully perpendicular amide bond (Στ+χN = 150.0°), which has been proposed to be required for N-protonation.11a–c,3 In addition, these structural features indicate that an increase of the amide bond distortion results in a significant lengthening of the N–C(O) bond, while the C=O bond remains relatively unchanged, providing a strong support for the classical resonance.1,2,7c</p><p>Having validated our model, we addressed the key question of structural factors governing the O- vs. N-protonation switch in amides. As predicted by the resonance, planar amides undergo protonation at oxygen (formamide: O- vs. N-protonation is favored by ca. 11.5 kcal/mol).1,2 O-protonation of planar amides shortens the N–C(O) bond and significantly reinforces the double bond character of amides (RE of O-protonated amides of ca. 40 kcal/mol).19</p><p>Table 2 lists proton affinities (PA) and differences between N- and O-protonation affinities (ΔPA) in the series of studied lactams.13a,b Most importantly, there is an excellent linear correlation between proton affinity difference and the sum of twist and pyramidalization at nitrogen angles (R2 = 0.98, Fig. 3), which can be compared with the correlation between proton affinity difference and twist angles (R2 = 0.90) and the correlation between proton affinity difference and pyramidalization at nitrogen angles (R2 = 0.86). This finding validates the use of additive descriptors of the geometric transformations of amide bonds to predict reactive properties of non-planar amides (vide supra). Importantly, this equation provides a very useful tool to predict N- vs. O-protonation sites of amides as only a single determination of the amide bond geometry is required from the X-ray crystallography or calculations to provide a reliable prediction of ΔPA.1,2 Since our calculations have already shown that (Στ+χN) can be correlated with the N–C(O) bond lengths (vide supra), this can be used in conjunction with amide bond distortion parameters to accurately predict ΔPA in one-carbon bridged amides.20 Overall, this finding provides a compelling argument to the long-standing question of structural factors governing the O- vs. N-protonation switch in non-planar lactams, and determines that (Στ+χN) of ca. 50-60° should be sufficient to promote N-protonation of amides.1,3–5,10,11,13 The N- vs. O-cross-over point in the studied series of lactams is located around the geometry region defined by the [5.4.1] ring system.</p><p>In conclusion, we have presented an efficient model enabling to predict the likelihood that a given amide can be protonated at the nitrogen atom. This study provides a compelling answer to the long-standing question of structural factors governing the N- vs. O-protonation switch in non-planar amides.1,2,10,11,13 The (Στ+χN) value of around 50-60° appears to be close to a barrier between N- vs. O-protonation of amides. This is much lower than the distortion that would correspond to a fully perpendicular amide bond (Στ+χN = 150.0°), which has been proposed to be required for efficient N-protonation. Given the availability of distorted amides in moderate distortion range,3–5 our data suggest that a wide range of amide analogues can be readily applied for probing the unconventional reactivity of amide bonds via N-protonation, including in biological contexts. We expect that the understanding provided for the amide bond protonation will enable the rational application of non-planar amides with a controlled-degree of rotation in organic synthesis and medicinal chemistry. Further studies on the effect of functional groups on the N-/O-protonation aptitude of non-planar amides are ongoing and these results will be reported shortly.</p>

PubMed Author Manuscript

C6 \xe2\x80\x93C8 Bridged Epothilones: Consequences of Installing a Conformational Lock at the Edge of the Macrocycle

A series of conformationally restrained epothilone analogs with a short bridge between the methyl groups at C6 and C8 was designed to mimic the binding pose assigned to our recently reported EpoA-microtubule binding model. A versatile synthetic route to these bridged epothilone analogs has been successfully devised and implemented. Biological evaluation of the compounds against A2780 human ovarian cancer and PC3 prostate cancer cell lines suggested that the introduction of a bridge between C6-C8 reduced potency by 25\xe2\x80\x931000 fold in comparison with natural epothilone D. Tubulin assembly measurements indicate these bridged epothilone analogs to be mildly active, but without significant microtubule stabilization capacity. Molecular mechanics and DFT energy evaluations suggest the mild activity of the bridged epo-analogs may be due to internal conformational strain.

c6_\xe2\x80\x93c8_bridged_epothilones:_consequences_of_installing_a_conformational_lock_at_the_edge_

Introduction<!>Synthesis of C4-C8 Bridged Epothilones: Retrosynthetic Analysis<!>Model Study<!>Construction of Building Blocks<!>Assembly of Building Blocks and Synthesis of 12,13-trans-6,8-Bridged Epothilone via Olefin Metathesis<!>Second Generation Synthesis via Suzuki Coupling<!>Construction of Building Partners for Suzuki Coupling<!>Completion of the Synthesis of Bridged Epothilones A-D<!>Bioactivity: Microtubule Assembly and Cytotoxicity<!>Molecular Mechanics and DFT Energy Evaluations<!>Conclusion<!>General<!>Molecular Modeling and Docking<!>trans-12,13-Macrolactone 36. Procedure A<!>trans-2,3-keto lactone 38<!>trans-12,13-Hydroxy Lactone 37<!>Bridged Epothilone C (7)<!>Bridged Epothilone A (5) and (5a)<!>Bridged Epothilone D (8)<!>C6-C8 bridged epothilone B (6)<!>Tubulin Purification<!>Tubulin Polymerization Assay<!>Determination of Cytotoxic Activity

<p>Epothilones are a family of cytotoxic polyketide natural products originally isolated from the bacterium Sorangium cellulosum.[1] Epothilone A (EpoA, 1) and epothilone B (EpoB, 2) (Figure 1), two major representatives, were recognized to be potent inhibitors against breast and colon cancer cells shortly after their initial isolation. [1] The mechanism of action of both EpoA and EpoB was established by the Merck group to be induction of tubulin polymerization in vitro resulting in the stabilization of microtubules under normally destabilizing conditions similar to the clinical anticancer drugs Taxol and docetaxel.[2] While epothilones exert their antiproliferative action in a similar way to Taxol, the two classes of compounds are distinctly different in terms of their potency and ability to inhibit the growth of multidrug-resistant cancer cell lines. [2–4] In contrast to Taxol, the epothilones are more efficacious promoters of cancer cell death with EpoB being the most active. Epothilones have also been proven to be very poor substrates for the phosphoglycoprotein 170 (P-gp) efflux pump. Thus, they retain almost full activity against P-gp-overexpressing, Taxol-resistant cell lines. Furthermore, epothilones are also active against cells with tubulin mutations which induce paclitaxel resistance.[4a] This suggests that epothilone-derived drugs might be useful for treating certain drug resistant tumors. In addition, although EpoA and EpoB were the major products isolated from the myxobacterium, numerous other related structures of the epothilone class have been identified as minor components of the fermentation of myxobacteria, including, for example, epothilone C (EpoC, 3) and D (EpoD, 4). These compounds also exhibit potent anticancer properties (Figure 1).[5]</p><p>These exceptional biological advantages, combined with the ease of synthesis by comparison with paclitaxel have evoked a vast research effort within academic and pharmaceutical research groups.[3] Numerous total and partial epothilone syntheses have been published since the determination of absolute stereochemistry in 1996.[6] During the development of these syntheses, a variety of methodologies have enabled the development of diverse libraries of synthetic analogs. In turn, these have contributed to mapping the extensive structure-activity relationship (SAR) profiles of epothilones and to elucidating interactions between the ligands and microtubules.[7–9] The tremendous efforts exerted to generate epothilone SAR profiles have greatly aided our understanding of the drug pharmacophore and the development of natural/unnatural analogs with improved biological activity and reduced toxicity. Significantly, these efforts have delivered at least seven compounds in advanced clinical trials, one of which recently won FDA approval for clinical use as an anti-cancer drug (Ixabepilone®).[10]</p><p>Since the discovery of the microtubule-stabilizing properties of epothilones in 1995, the details of the binding poses for the structurally diverse taxanes and epothilones have been pursued in order to facilitate the rational design of improved and perhaps structurally simplified analogs.[11–14] A variety of epothilone conformations and tubulin binding modes have been proposed by pharmacophore mapping,[11,12] solution NMR[15,16] and the superposition of epothilones on taxanes in the electron crystallographic tubulin complex.[13,14] By combining NMR spectroscopy, electron crystallography and molecular modeling, our group proposed a unique EpoA conformation and microtubule binding model that offers an alternative to the common pharmacophore model by describing the tubulin binding cavity as promiscuous.[17] According to this model, epothilone and Taxol occupy the same gross binding pocket, but the tubulin-ligand binding is mediated through different sets of hydrogen bonds and hydrophobic interactions for the two compounds. The electron crystallographic structure of epothilone was superposed with that of Taxol bound to tubulin. The overlap suggested that the thiazole moiety of epothilone A and the benzoyloxy phenyl of Taxol do not reside in the same region of the tubulin pocket. In addition, among the five oxygen-containing polar groups in epothilone, only the C7-OH falls near the similar C7-OH moiety in Taxol. In this comparison, the latter is the only common center between the two molecules.</p><p>An unusual feature of the EpoA binding conformer derived by electron crystallography (EC) is the presence of a syn-pentane interaction between the methyl groups at C-6 and C-8. A conceivable test of the latter structural feature introduces a short bridge between the ring carbons to constrain the macrocycle to the EC model geometry. The cyclohexane rings depicted in 5–8 (Figure 2) illustrate one solution to the problem. A potential liability of this strategy is that the small ring, expanding the volume of the epothilone, might introduce steric congestion with the tubulin residues lining the binding pocket. To examine specific geometric details of the corresponding structures, 5 and 6 in the proposed binding form were optimized by molecular mechanics to show that both reside in a stable local minimum. Subsequent docking of the latter into the β-tubulin taxoid site suggested that the additional CH2 in the newly installed cyclohexane ring would not experience undue steric congestion with the protein (Figure 3).</p><p>To our knowledge, modifications within the C6-C8 epothilone sector have received little previous attention (Figure 4).[18–20] In a related study, however, Martin and colleagues introduced a three carbon bridge between C4 and C6 from the pro-R methyl at C4 in the EpoB framework (9, Figure 4).[20] The compound had no effect when exposed to the MCF-7 tumor cell line. Since the EC binding conformer predicts the pro-S attachment to be the appropriate attachment direction, stereochemical inversion may be responsible for the lack of activity. Accordingly, bridged analogs 5–8 were selected as targets suitable for diagnostic tests of the EC epothilone binding model. In addition, ZK-EPO (Sagopilone, 10) in which a vinyl moiety was introduced to the C-6 methyl group and the C16-C21 side chain was displaced by a benzothiazole moiety, is currently undergoing advanced clinical development as the first fully synthetic epothilone candidate.[18]</p><p>Previously, we briefly communicated the synthesis of bridged 5, and reported the compound to be only weakly active against the A2780 ovarian cancer cell line.[19] In the present work, we describe the full synthetic details for 5–8. All four analogs were further subjected to cytotoxicity assessment against human ovarian (A2780) and prostate cancer (PC3) cell lines, tubulin assembly and microtubule cold stabilization.</p><!><p>The first generation synthetic plan for the C6-C8 bridged epothilones, based on ring closure metathesis (RCM) as a key step, is summarized in Scheme 1. Compound 5 is used as an example. Although RCM is known to give both cis and trans isomers during total syntheses of natural epothilones, [21] it was applied as a key step here considering that isolation of both isomers could contribute to the structure activity profile (SAR) for the compound series. Applying a general disconnection strategy to epothilones, the bridged target 5 could be traced back to alcohol 12 and the advanced intermediate keto acid 13 after retrosynthetic epoxidation, RCM and esterification. The preparation of keto acid 13 was conceived as the key step along this route, by which the cyclohexane core structure with three adjacent chiral centers would be constructed. First, the stereochemistry at C7 and C8 in 13 was contemplated by means of sequential substrate directed epoxidation[22] and regiocontrolled epoxide opening from homoallylic alcohol 14. Moving further along the retrosynthetic path, alcohol 14 was envisioned to arise from aldehyde 15 by utilizing Brown's asymmetric cyclohexenylboration stratetgy.[23]</p><!><p>To test the feasibility of the substrate-directed epoxidation and subsequent regio-controlled epoxide opening strategy, a simplified model system was studied as shown in Scheme 2. The model study started from (−)-B-2-cyclohexen-1-yldiisopinocampheylborane 16, prepared by treating cyclohexa-1,3-diene with diisopinocampheylborane derived from (+)-α-pinene at −25 °Cin tetrahydrofuran (THF) as described by Brown.[23] The freshly prepared solution of borane 16 in THF was cooled to −100 °C, and treated with pivalaldehyde. After oxidation with H2O2 in the presence of NaHCO3, homoallylic alcohol 17 was obtained in 70% yield (dr > 95% by 1H NMR).[24]</p><p>The highly stereoselective epoxidation of 17 was first achieved by a homoallylic alcohol-directed vanadium-catalyzed epoxidation strategy to afford hydroxy epoxide 18 in 88% yield as a single isomer. Further study indicated that mCPBA-based epoxidation also delivered the desired epoxide 30 in 84% yield. The relative configuration was confirmed by NOE. Considering the difference between the optical rotation value of 17 and that reported previously ([α]D25= − 3.4, c 1.0, CHCl3, Lit[23b]+ 6.83 l 0.5, neat), the p-nitro-benzoyl derivative 19 was prepared. X-ray crystallography of 19 unambiguously confirmed the absolute configuration of alcohol 17 and hydroxy epoxide 18 (Scheme 2). The site of epoxide ring opening by chloride anion further supports the original plan for regioselective nucleophilic opening of hydroxy epoxide.</p><p>With successful stereoselective epoxidation, the next key step in Scheme 2 is epoxide opening with an alkyl nucleophile in a regioselective manner. Fortunately, this transformation was successfully performed by treatment of 18 with freshly prepared 4-pentenylmaganesium bromide in the presence of CuCN (10 mol%). The desired diol 20 was obtained exclusively in 89% yield. We concur with Flippin and co-workers[25] that the regioselectivity of this metal catalyzed epoxide opening is not only controlled by the Fürst-Plattner rule favoring diaxial orientation, but also is most likely reinforced by a chelation process.</p><p>Following the remarkable success of the two key steps in the model study, we turned our attention to probe the regioselective protection of the two secondary hydroxy groups in 20 and the subsequent oxidation of the sterically hindered secondary alcohol in the model system. Selective silylation of the sterically less hindered OH group in 20 was achieved by slow addition of tert-butyldimethylsilyl triflate (TBSOTf) to a solution of 20 in CH2Cl2 at −78 °C in the presence of 2,6-lutidine to provide the mono-silyl ether 21 in 85% yield. Surprisingly, no silylation of the sterically hindered hydroxyl group was detected even when 1.5 equiv of TBSOTf was added. At this stage, a NOESY analysis for silyl ether 21 supported the previously described regioselectivity of the oxirane opening and selective TBS protection (Scheme 2). Swern oxidation of the sterically hindered alcohol afforded the desired olefinic ketone 22 in quantitative yield.</p><!><p>Encouraged by the results of the model studies, we proceeded to construct carboxylic acid 13. In pursuit of this advanced intermediate, the previously reported silyl ether 23[19] was subjected to ozonolysis, followed by acid catalyzed acetal protection with ethylene glycol and selective desilylation of the primary TBS silyl ether to afford primary alcohol 24 in 56% yield over three steps (Scheme 3). Exposure of 24 to Swern oxidation producedthe desired aldehyde 15 in quantitative yield.</p><p>At this point, we were in a position to probe the feasibility of establishing the C5-C6 bond by Brown's protocol.[23] Unfortunately, when aldehyde 15 was subjected to the standard Brown conditions, no workable amounts of product 14 could be separated (Scheme 3), while over 90% of aldehyde 15 was recovered before the oxidative quench. Attempts to facilitate the reaction by increasing temperature and reaction time did not lead to satisfactory results. We presume the allylboration is disfavored not only by steric hindrance from the α-quaternary carbon of the aldehyde, but also by the coordination between borane and the acetal oxygen atoms, which in turn interrupts the interaction between borane and aldehyde.</p><p>To address this problem, we turned our attention to an alternative aldehyde 25 (Scheme 4) in which a terminal olefin replaces the acetal in 15 and, thereby, avoids the potential coordination described above. Selective desilylation and subsequent Swern oxidation converted silyl ether 23 into the desired aldehyde 25. Upon treatment of the modified aldehyde 25 with freshly prepared borane 16, the desired homoallylic alcohol 26 was obtained in excellent yield and selectivity (96%, dr > 20:1 by 1H NMR) as shown in Scheme 4. Surprisingly, both the C-C bond formation and oxidative cleavage of the B-O bond were unexpectedly sluggish, taking about three weeks. Stereochemistry at C5 and C6 was assigned on the basis of the model study (Scheme 2).</p><p>With this chemistry in hand, the next phase involved a crucial stereoselective epoxidation and subsequent regioselective oxirane ring opening. Following the successful strategy achieved in the model study, alcohol 26 underwent chemoselective epoxidation by vanadium-catalysis to provide hydroxy epoxide 27 in 93% yield (dr > 20:1 by 1H NMR), followed by copper-catalyzed epoxide opening with Grignard reagent to furnish diol 28 in 90% yield as the sole diastereomer (Scheme 4). It is worth noting that an excess of Grignard reagent (8–9 equiv) was required to minimize formation of the bromohydrin side product.[26] Selective silylation of the sterically less hindered OH group in 28 furnished silyl ether 29 (85% yield). The relative stereochemistry of compound 29 was confirmed on the basis of NOESY experiments. In practice, the conversion from 26 to 29 could be completed in 93% yield over three steps without purification of the intermediates 27 and 28. To complete Scheme 4, Swern oxidation converted the secondary alcohol into ketone 30 in quantitative yield. The stereochemical assignment for ketone 30 was confirmed subsequently by comparison with its analog 50 (Scheme 9), the stereochemistry of which was determined by X-ray crystallography of a derivative.[19]</p><p>It is clear that aldehyde 25 has obvious advantages over aldehyde 15 in the context of cyclohexenylboration. However, application of 25 raised a second challenging problem, namely differentiation between the two terminal olefins with high structural similarity in 30. As will be shown, the terminal olefin homoallyic to OTBS could be selectively converted to a carboxylic acid by taking advantage of the masked homoallyic alcohol. Thus, we turned our attention to the hydroxy directed epoxidation to give a carboxylic acid precursor. To pursue this strategy, desilylation of 30 with trifluoroacetic acid afforded diol 31 (78% yield). Subsequently, vanadium catalyzed chemoselective epoxidation of 31, as expected, led to β-hydroxy epoxide 32 in 89% total yield as a mixture of two diastereomeric epoxides (ca. 10:1 by 1H NMR). The stereochemistry of the epoxide has been tentatively assigned in accord with the proposed model by Mihelich and coworkers.[27] In view of the subsequent cleavage of the epoxide, the diastereomeric epoxides were subjected to the next step without separation.</p><p>At this stage, we initially attempted to reinstall the TBS silyl ether onto 32 in order to pursue the original synthetic plan (Scheme 1). However, all attempts with classical conditions failed to give satisfactory results.[28] Fortunately, diacetyl epoxide 33 was cleanly obtained in 93% yield by treatment of alcohol 32 with acetic anhydride and 4-dimethylaminopyridine (DMAP). At this point, it appeared timely to transfer the primary epoxide to the carboxylic acid. The conversion was accomplished with a three step sequence. The epoxide first underwent tetrabutylammonium bisulfate catalyzed hydrolysis,[29] followed by NaIO4 cleavage of the resulting diol to furnish an aldehyde intermediate. Purification of the aldehyde by silica gel flash column chromatography was unmanageable due to the instability of this intermediate. Thus, Pinnick oxidation of the crude aldehyde with NaClO2 in the presence of 2-methyl-2-butene and NaH2PO4 in t-BuOH/H2O provided the desired carboxylic acid 34 in 45% yield over the three-step procedure (Scheme 5).</p><!><p>With the key building block 34 in hand, our attention was directed to the feasibility of the olefin metathesis strategy. Therefore, the coupling between alcohol 12 and carboxylic acid 34 was performed under the influence of 1-ethyl-3-((dimethylamino)propyl)carbodiimide hydrochloride (EDCI) and DMAP to furnish the proposed metathesis precursor 35 in 58% yield as depicted in Scheme 6. In this coupling reaction, serious β-elimination from the β-acetate of the carboxylic acid was responsible for the modest yield. However, β-elimination was completely suppressed by a modified Yonemitsu-Yamaguchi protocol,[30] giving keto ester 35 in 86% yield. Exposure of 35 to metathesis catalysts 39–42 under highly dilute conditions resulted in clean formation of a single trans-macrocyclic olefin 36 (JH12-H13 = 14.4 Hz). Grubbs catalysts 39 and 40 as well as the Hoveyda catalyst 41 gave the trans-product in high yields, while the reaction was unresponsive to Hoveyda catalyst 42 (Entry 5, Scheme 6). It is widely known that the E/Z selectivity of the ring closure metathesis depends on many factors including substrate, solvent, temperature and concentration.[31] In this specific case, attempts to modify the geometric outcome of the reaction by choosing solvents and temperatures as recorded in Table 1 were unsuccessful.</p><p>The disheartening geometric results from olefin metathesis temporarily directed our attention to the 12,13-trans-6,8-bridged epothilone C (37, Scheme 6). Considering previous SAR studies suggesting that the non-natural epothilone analog 12,13-trans-epothilone C is only slightly less active then the natural epothilone C (3),[32] the C6-C8 bridged analog 37 was regarded as potentially providing valuable structural information for the project. With this in mind, we turned to the deacylation of 36 (Scheme 6). Surprisingly, none of the reaction conditions applied was able to accomplish deprotection to produce the dihydroxyl lactone 37. In all cases, either unreacted acetate was recovered or decomposition took place. One of the major side reactions arising from attempted deacylation was β-elimination leading to lactone 38, which could be alternatively prepared from 36 in 96% yield by treatment with 8-diazabicyclo[5.4.0]undec-7-ene (DBU). Such a process has been documented to suggest that the approximate 180° torsion angle around C2-C3 might be responsible for the elimination.[3, 33]</p><p>We presume the unsuccessful deacylation could also arise from the competition between hydrolysis and elimination of the β-acetate. In this specific case, β-elimination might be much faster than hydrolysis of the acetate. To facilitate deacylation, we envisioned introducing a substituent to the acetyl moiety which could increase the acetyl hydrolysis rate, while not significantly altering its leaving group character. With this scenario in mind, the chloroacetyl group was introduced, recognizing that it might be 350–700 fold more quickly hydrolyzed than acetyl depending on the nature of the intermediates.[34] As shown in Scheme 7, the required chloroacetate was prepared from epoxy alcohol 32 in quantitative yield. Subsequent exposure to NaIO4/H5IO6 mediated cleavage of the terminal epoxide to generate an aldehyde which was subsequently subjected to Pinnick oxidation to furnish the carboxylic acid 43 in 72% overall yield. Esterification of keto acid 43 by alcohol 12 was achieved in modest yield with the modified Yonemitsu-Yamaguchi protocol. [35] Without complication, the following olefin metathesis led cleanly to trans-product 45 (JH12-H13 = 14.8 Hz). For example, a 77% yield of 45 was achieved with Hoveyda second generation metathesis catalyst 42. After screening various conditions for the crucial deacylation step, we discovered that it could be successfully performed by careful treatment of 45 with ammonium hydroxide in methanol (1/10, v/v) followed by treatment with ammonia in methanol to afford the 12,13-trans-6,8-bridged EpoC analog 37 in 57% yield (Scheme 7).</p><p>To bring the synthesis to a close, we initially attempted to isomerize the C12-C13 trans double regioselectively to the desired cis geometry. Unfortunately, following attempts such as photoirradiated isomerization,[18] iodine-catalyzed free radical isomerization[36] and Vedejs isomerization,[37] no observable amounts of the isomerized product could be separated (assuming isomerization occurred). Either unreacted trans-olefin was recovered or decomposition took place.</p><!><p>As discussed above, some surprising limitations surfaced in the ring forming olefin metathesis reaction. Although there is still much opportunity to optimize reaction conditions and employ alternative methods such as the molybdenum-based Schrock catalyst,[38] the epothilone literature teaches that the stereochemical outcome of the RCM process is highly substrate dependent.[21a,39] With such an ambiguous precedent, it was imperative to select a reliable method for accessing the desired Z-stereochemistry. An important alternative to introduce the Z-double bond at C12-C13 is Danishefsky's B-alkyl Suzuki coupling strategy.[40] Given the widespread application of this strategy in epothilone synthesis, we elected to explore its utility for our targets.</p><p>The retrosynthesis for bridged epothilones 5 and 6 via the second generation Suzuki coupling strategy is summarized in Scheme 8.[41] The key disconnection for this route is at the C11-C12 bond, leading to vinyl iodides 48/49 and olefin 50 as the Suzuki coupling partners. The keto diene 50 derives from aldehyde 51 following a sequence similar to that for the synthesis of dienyl ketone 30 in Scheme 4. In contrast to keto diene 30, a gem dimethyl moiety was introduced at the right-side terminal olefin of 50 in order to easily differentiate the two olefins.</p><!><p>To pursue this modified route, the preparation of Suzuki coupling precursor 50 is illustrated in Scheme 9.[19] The modified aldehyde 51 was firstly obtained by a four-step sequence from 23 in 58% yield, followed by allylboration of aldehyde 51 with freshly prepared (−)-B-2-cyclohexen-1-yldiisopinocampheylborane 16 to give homoallylic alcohol 52 in 96% yield (dr > 20:1 by 1H NMR). Then, using the previously validated vanadium-catalysis strategy, alcohol 52 was converted to epoxy alcohol 53 in 93% yield (dr > 20:1 by 1H NMR). Reaction of epoxide 53 with allylmagnesium bromide in a copper catalyzed fashion furnished epoxide-opened product 54 (85% yield) along with a trace of the C7-alkylated isomer and bromohydrin. The sterically less hindered hydroxyl group from 54 was selectively converted to TBS silyl ether 55 in 97% yield. At this point, a NOESY analysis was executed to confirm the relative stereochemistry (Scheme 9). Finally, the sterically hindered secondary alcohol in diene 55 was transformed to dienyl ketone 50 by Swern oxidation in 85% yield. The absolute configuration of dienyl ketone 50 has been verified by X-ray crystal structure analysis of a derivative.[19] Preparation of the other Suzuki coupling partners, vinyl iodides 48/49, was accomplished from alcohol 12 in a four-step sequence (Scheme 9).[39c, 42]</p><!><p>With the requisite coupling precursors in hand, the final steps in the synthesis of bridged epothilones 5/6 were carried out as depicted in Scheme 10. After regioselective hydroboration with 9-BBN, olefin 50 was coupled with vinyl iodides 48/49 in accordance with an approach reported by Danishefsky to furnish cis-olefins 46/47 (JH12-H13 = 10.8 Hz, 46) in 92% and 57% yields respectively.[40a] The following crucial regioselective dihydroxylation of trienes 56/57 was performed under Sharpless asymmetric dihydroxylation conditions[43] to selectively convert the gem-dimethyl olefin to diols 56/57 as a mixture of diastereomers (56: 42% yield, 86% BRSM, ca. 5:1 ratio by 1H NMR; 57: 42% yield, 87% BRSM, ca. 4:1 ratio by 1H NMR ). The stereochemistry of the hydroxyl group was undefined. Without separation, the resulting mixture of diols was subjected to NaIO4 mediated glycol cleavage and subsequent Pinnick oxidation to furnish the corresponding carboxylic acids 58 and 59 in 72% and 58% yields after two steps (Scheme 10).</p><p>As shown in scheme 10, final steps in the synthesis of the target compounds 5/6 involved conversion of keto acids 58/59 to dihydroxy lactones 60/61 by employing a procedure utilized by Nicolaou in the total synthesis of epothilones A and B.[44] Selective desilylation with tetra-n-butylammonium fluoride (TBAF), followed by Yamaguchi lactonization (62 in 51% yield, and 63 in 60% yield) and global desilylation in the presence of a freshly prepared trifluoroacetic acid (TFA) solution in CH2Cl2 (v/v, 1/4) gave C6-C8 bridged epothilone C (7, 88% yield) and epothilone D (8, 91% yield). Finally, we were pleased to obtain the C6-C8 bridged epothilone A as a mixture of 5 and its cis-epoxide diastereomer 5a (84% total yield, ca. 2:1 ratio, 1H NMR) by treatment with 3,3-dimethyldioxirane (DMDO) as described by Danishefsky.[40a] Fortunately, these two diastereomers were separable by preparative thin-layer chromatography. In a similar fashion, the C6-C8 bridged epothilone D analog 8 was converted to the bridged epothilone B analog 6 in 52% yield (dr > 20:1 by 1H NMR). The stereochemistry of the epoxide of these bridged epothilones was determined by 1D and 2D NOE analysis (Scheme 10).</p><!><p>The C6-C8 bridged epothilone analogs 5a, 5–8, 36–38 were exposed to A2780 ovarian cancer and PC3 prostate cancer cell lines to evaluate their antiproliferative properties (Table 2). In general, these C6-C8 bridged compounds are significantly less potent than the corresponding open chain analogs. Against the A2780 cell line, compound 38 exhibited the highest potency with an IC50 = 1.1 μM, but it is still 27-fold less potent than EpoD. C6-C8 Bridged Epo B 6 exhibited the highest potency against the PC3 prostate cancer cell line with a 206-fold potency loss by comparison with the activity of EpoD.</p><p>The ability of C6-C8 bridged epothilones to promote tubulin assembly and stabilize the microtubules against cold induced disassembly was also studied using 50 μM of the compounds in 4% (v/v) DMSO. The results were compared to that of 10 μM PTX under identical conditions (Figure 5). Compounds 37 and 5a displayed the least effect on tubulin assembly. Compounds 6 and 8 were the most active, however, less so than PTX. None of the C6-C8 bridged epothilones showed significant ability to stabilize the microtubules against cold induced disassembly.</p><!><p>Obviously, these biological outcomes are not consistent with our having installed a C6-C8 bridge that constrains 5–8 to an efficacious bioactive epothilone pose. There are several possible explanations for this: 1) the C6-C8 bridge has retained the target conformation of the epothilone ring, but altered the binding mode to prevent effective coordination with the protein; 2) incorporation of the C6-C8 bridge has raised the energy of the bound target conformations of 5–8 making them inaccessible to microtubules; 3) the molecules have been constrained as expected (Figure 3), but the target conformation is not the de facto bound form as proposed. Concerning 1), there are precedents for loss of activity resulting from internal bridge-building in taxanes. In two cases, NMR evidence supported the conclusion that the ligand conformations appear to be retained, but the additional tether interferes sterically with the tubulin binding site, lifts the molecule higher in the pocket and thereby reduces ligand binding.[45] In one of these reports involving cyclic constraints in the C-13 side chain, the differences in tubulin binding and MCF7 cell cytotoxicity (5–10 fold and 37–120 fold, respectively) arose from expansion of ring size from 5-members to 6.[50a] Thus, in spite of our initial assessment that a 6-membered ring in 5–8 would cause little steric congestion (Figure 3), the degree of crowding may have been underestimated. With respect to 2), we have examined the conformational landscapes of 1 and 5 to determine the energy differences between the respective bound conformations (Figure 3A) and their corresponding global minima (GM). Thus, individual 25,000 step conformational searches for the two structures were performed with the MMFF/GBSA/H2O protocol in Schrodinger's Macromodel.[46] The GM was then re-optimized with the OPLS-2005, MM3 and AMBER molecular mechanics methods.[46] The EpoA EC structure (1, Figure 3A) and that of 5 based on it were likewise optimized with the same four methods to put the structures (LM or local minimum forms) on the same energy scales. A comparison of the EC-related structures and the optimized variants demonstrated minimal average root mean square deviations (RMSD) of 0.098 to 0.35 Å for all heavy atoms. Superposition of the EC-based structures and the optimized OPLS-2005 conformers (average heavy-atom RMSD values of 0.26 and 0.14 Å, respectively) are provided in the Supporting Information to demonstrate essentially a perfect match between the EC bound and force-field optimized conformers. The energy differences between GM and LM for epoA and 5 for each method (ΔΔE((GM-LM)epoA – (GM-LM)5) were computed to show that epoA falls lower by 3.0 (OPLS-2005), 2.4 (AMBER), 1.1 (MMFF) and 0.8 (MM3) kcal/mol (see SI for details). Thus, in the molecular mechanics regime, the bridged structure 5 is predicted to require an additional 1.8 kcal/mol on average relative to epoA to achieve the bound form on tubulin in the context of the EC structure. In order to avoid issues surrounding the variable parameterization of the force field protocols, we calculated the energies of the OPLS-2005 optimized GM and LM forms for EpoA (1) and 5 with the density functional protocol B3LYP/6-31G*. The ΔΔE = 5.5 kcal/mol reinforces the force field trend and suggests that the weaker binding for compound 5 may be a result of increased internal conformational strain energy relative to EpoA. The possibility remains that 3) is the underlying cause of the bio-data presented here. However, until a definitive structure of the tubulin-epothilone complex is determined, the current EC[17] and NMR[47] structures remain the front-line contenders for the bound conformation of epothilones to β-tubulin.</p><!><p>A series of conformationally restrained epothilone analogs with a short bridge between methyl groups at C6 and C8 was designed to mimic the binding pose derived for our recently reported EpoA-microtubule binding model. A versatile synthetic route to these bridged epothilone analogs has been successfully devised and implemented. The key stereochemistry within the bridged C6-C8 sector was controlled by asymmetric allyboration followed by hydroxy-directed epoxidation and regiocontrolled opening of the resultant epoxide. The cis-C12-C13 double was constructed via Suzuki coupling while the ring closure metathesis exclusively gave trans selectivity.</p><p>The C6-C8 bridged epothilones were evaluated for their biological activity against the A2780 human ovarian cancer and PC3 prostate cancer cell lines. The cytotoxicity data implies that these epothilone analogs are considerably less potent than the natural epothilones. The tubulin assembly and microtubule cold stabilization assay reveal that compounds 6 and 8 inhibit tubulin assembly weakly, while none of the bridged epothilone analogs show significant microtubule stabilization against cold induced disassembly. Possible causes for the poor activity of the bridged epothilones include steric congestion between tubulin and the C6-C8 bridge, torsional strain in the flexible portion of the epothilone ring raising the energy requirement for binding and a mismatch with the empirical binding pose. Insights into the first two explanations are provided.</p><!><p>Unless otherwise noted, commercial reagents and solvents were used as received unless otherwise noted. Flash column chromatography was performed by employing either Sorbent Technologies 200–400 mesh or Waterman 230–400 mesh silica gel 60. Analytical thin-layer chromatography (TLC) was performed on pre-coated silica gel 60 F254 (0.25mm thick) from EM Science. TLC plates were visualized by exposure to ultraviolet light (UV) and/or exposure to phosphomolybdic acid or potassium permanganate TLC stains followed by brief heating on a hot plate. Preparative TLC separation was performed on Analtech preparative plates pre-coated with silica gel 60 UV254 (0.5, 1.0 or 1.5 mm thick). Melting points (mp), determined on a MEL-TEMP Melting Point Apparatus from Laboratory Devices, are uncorrected. Optical rotations were measured on a Perkin Elmer Model 341 digital polarimeter with a sodium lamp at room temperature. Infrared (IR) spectra were recorded on a Nicolet 370 with a diamond probe or an ASI ReactIR 1000 FI-IR Spectrophotometer with a silicone probe (wavenumbers (cm−1)). Where noted "neat", the sample was loaded as a thin film. Proton nuclear magnetic resonance (1H NMR) spectra and carbon nuclear magnetic resonance (13C NMR) spectra were determined on an INOVA400 (1H NMR: 400 MHz, and 13C NMR: 100 MHz) or INOVA600 (1H NMR: 600 MHz, and 13C NMR: 150 MHz) instrument. Chemical shifts for 1H NMR are reported in parts per million (δ scale) with deuterated chloroform (CDCl3) as the internal standard (7.26 ppm) and coupling constants are in hertz (Hz). The following abbreviations are used for spin multiplicity: s = singlet, d = doublet, t = triplet, q = quartet, m = multiplet, bs = broad singlet. Chemical shifts for 13C NMR are reported in parts per million (δ scale) relative to the central line of the triplet at 77.23 ppm for deuterated chloroform (CDCl3). High resolution mass spectra (HRMS) were obtained on a JEOL JMS-SX102/SX102A/E or Thermo Finnigan LTQ-FTMS instrument.</p><p>Experimental details, characterization data for all new compounds, NMR spectra of key intermediates, X-ray crystal structure data for 19 and results for energy calculations for 1 and 5 are available in the Supporting Information.</p><!><p>The 3-D structure of bridged epothilones 5 and 6 were constructed based on the electron crystallographic (EC) pose of EpoA bound to a β-tubulin.[17] The resulting structure of 5 and 6 was then fully optimized with the MMFF/GBSA/H2O force field to provide the nearest local minimum. The latter was flexibly Glide-docked[48] into the electron crystallographic structure of EpoA-tubulin. [17] The best docking pose was chosen on the basis of the Glide scoring function together with visualizationto ensure a reasonable binding mode and match with the EC complex. Conformational analysis for 1 and 5 and energy evaluations are described in the Supporting Information.</p><!><p>To a solution of diene 35 (7.5 mg, 0.0125 mmol, 1.0 equiv) in CH2Cl2 (12.5 mL, 0.001 M) was added Grubbs catalyst I (1.1 mg, 0.00125 mol, 10 mol %), and the reaction mixture was allowed to stir at 25 °C for 12 h. After the completion of the reaction as established by TLC, the solvent was removed under reduced pressure and the crude product was purified by preparative thin-layer chromatography (Hexanes/ethyl acetate, 3/1) to afford the trans-lactone 36 (6.0 mg, 84%) as a white foam. Procedure B: To a solution of diene 35 (15 mg, 0.0249 mmol, 1.0 equiv) in toluene (25 mL) was added with Grubbs catalyst I (2.1 mg, 0.00249 mol, 10 mol %), and the reaction mixture was allowed to stir at 80 °C for 12 h. After the reaction is complete, the mixture was worked up according to the procedure described in procedure A to furnish 36 (10.9 mg, 76%). Procedure C: Diene 35 (13 mg, 0.0216 mmol, 1.0 equiv) was converted to 36 (12.2 mg, 100%) in accordance with the procedure described in procedure A except for the use of Grubbs catalyst II (1.8 mg, 0.0022 mol, 10 mol %). Procedure D: Diene 35 (13 mg, 0.0216 mmol, 1.0 equiv) was converted to 36 (10.1 mg, 100%) in accordance with the procedure described in procedure B except for the use of Grubbs catalyst II (1.8 mg, 0.0022 mol, 10 mol %). Procedure E: Diene 35 (12 mg, 0.02 mmol, 1.0 equiv) was converted to 36 (10.8 mg, 95%) in accordance with the procedure described in procedure A except for the use of Hoveyda-Grubbs catalyst II (0.6 mg, 0.002 mol, 5 mol %). The crude reaction mixtures in procedures A, B, C, D and E were determined to be >20:1 ratio of diastereomeric trans-olefin by 1H NMR spectroscopy. Rf = 0.37 (hexanes/ethyl acetate, 2/1); [α]22 D −47.8 (c 1.0, CHCl3); IR (thin film) νmax 2926, 2862, 1731, 1707, 1504, 1443, 1371, 1239, 1180, 1029, 972, 916, 731 cm−1; 1H NMR (600 MHz, CDCl3) δ ppm 6.97 (s, 1H, SCH=C), 6.50 (s, 1H, CH=CCH3), 5.87 (dd, J = 7.2, 5.2 Hz, 1H, CH2CHOAc), 5.57 (ddd, J =14.4, 7.2, 7.2 Hz, 1H, CH=CH), 5.47 (ddd, J =14.4, 7.2, 7.2 Hz,, 1H, CH=CH), 5.31 (dd, J = 9.4, 1.9 Hz, 1H, CHOC(O)CH2), 5.07 (s, 1H, CHCHOAc), 3.29-3.23 (m, 1H, CHC(O)), 2.69 (s, 3H, N=C(S)CH3), 2.71-2.68 (m, 1H, CH2CO2), 2.62-2.56 (m, 2H), 2.41 (dd, J = 15.3, 4.9 Hz, 1H), 2.21-2.15 (m, 1H), 2.10 (s, 3H, ArCH=CCH3), 2.05 (s, 3H, CH3CO2), 2.03 (s, 3H, CH3CO2), 1.98-1.89 (m, 2H), 1.86-1.76 (m, 1H), 1.72-1.67 (m, 2H), 1.57-1.493 (m, 3H), 1.40-1.35 (m, 2H), 1.33-1.21 (m, 2H), 1.14 (s, 3H, C(CH3)2), 1.07 (s, 3H, C(CH3)2); 13C NMR (100 MHz, CDCl3) δ 211.62, 170.71, 170.27, 169.21, 164.79, 152.88, 137.68, 132.67, 126.91, 119.64, 116.73, 79.62, 71.33, 70.48, 53.62, 41.94, 37.83, 37.71, 36.50, 31.30, 28.82, 26.94, 24.99, 24.09, 21.42, 21.24, 20.00, 19.89, 19.50, 18.87, 15.39; HRMS calcd for C31H44NO7S 574.28385 [M + H]+, found 574.28292.</p><!><p>A mixture of macrolactone 36 (21 mg, 0.0366 mmol, 1.0 equiv) in anhydrous CH2Cl2 (2 mL) was treated with 1,8-Diazabicyclo[5.4.0]undec-7-ene (DBU) (55.7 mg, 0.366 mmol, 10.0 equiv) at room temperature. After being stirred for 3 h, no more 36 was detected from TLC. The solvent was removed under reduced pressure without further workup. The resultant residue was purified by preparative thin-layer chromatography (Hexanes/ethyl acetate, 4/1) to furnish product 38 (18.4 mg, 96%) as a colorless oil: Rf = 0.51 (hexanes/ethyl acetate, 2/1); [α]22D +17.4 (c 1.68, CHCl3); IR (thin film) νmax 2929, 2861, 1713, 1645, 1503, 1444, 1379, 1362, 1294, 1242, 1177, 1048, 1017, 992, 970, 913, 879, 731 cm−1; 1H NMR (600 MHz, CDCl3) δ 7.40 (d, J = 16.0 Hz, 1H, CH=CHC(O)), 6.96 (s, 1H, SCH=C), 6.61 (s, 1H, CH=CCH3), 6.07 (d, J = 16.0 Hz, 1H, CH=CHC(O)), 5.56 (dd, J = 10.3, 2.2 Hz, 1H, CHOC(O)CH2), 5.53-5.37 (m, 2H, CH2CH=CH), 4.86 (s, 1H, CHCHOAc), 3.01-2.98 (m, 1H, CHC(O)), 2.71 (s, 3H, N=C(S)CH3), 2.52-2.49 (m, 1H), 2.44-2.39 (m, 1H), 2.19-2.14 (m, 1H), 2.11 (s, 3H, ArCH=CCH3), 2.00 (s, 3H, CH3CO2), 2.04-1.95 (m, 1H), 1.93-1.85 (m, 1H), 1.69-1.62 (m, 2H), 1.58-1.52 (m 1H), 1.50-1.37 (m, 3H), 1.26-1.12 ( m, 10H); 13C NMR (100 MHz, CDCl3) δ 210.45, 170.59, 165.38, 164.88, 152.72, 152.39, 138.22, 132.55, 127.20, 121.98, 112.00, 116.39, 77.77, 71.34, 51.91, 43.51, 39.24, 36.58, 33.09, 28.54, 26.72, 23.64, 23.29, 23.00, 22.79, 21.38, 19.46, 15.61; HRMS calcd for C29H40NO5S 514.26272 [M + H]+, found 574.26186.</p><!><p>A solution of chlorolactone 45 (60 mg, 0.093 mmol) in methanol (10 mL) at 0 °C was treated with ammonium hydroxide (0.5 mL), and stirred at that temperature until the reaction was complete (ca. 12 h). The solvent was removed under reduced pressure to give white foam. Next, the white foam was dissolved in methanol (10 mL), and treated with amino methanol (1mL, 7N in methanol) at 0 °C. After being stirred for 48 h, 1H NMR suggested the reaction was complete. The solvent was removed under reduced pressure and the residue was purified by preparative thin-layer chromatography (Hexanes/ethyl acetate, 15/4) to afford hydroxy lactone 37 (26 mg, 57%) as a white foam: Rf = 0.38 (CH2Cl2/MeOH, 15:1); [α]22D −11.6 (c 0.85, CHCl3); IR (thin film) νmax 3486 (br), 2930, 2860, 1729, 1679, 1505, 1444, 1405, 1374, 1336, 1297, 1247, 1177, 1123, 1085, 1046, 984, 915, 865, 726, 676 cm−1; 1H NMR (600 MHz, CDCl3) δ 6.96 (s, 1H, SCH=C), 6.52 (s, 1H, CH=CCH3), 5.48 (dd, J = 9.8, 3.7 Hz, 1H, CHOC(O)), 5.45-5.40 (m, 1H, CH=CH), 5.37-5.32 (m, 1H, CH=CH), 4.41 (dd, J = 10.5, 1.8 Hz, 1H, CHOHC(CH3)2), 4.16 (s, 1H, CHCHOH), 3.67 (d, J = 1.5 Hz, 1H, CHOHC(CH3)2), 3.65 (d, J = 2.5 Hz, 1H, CHCHOH), 2.81 (d, J = 10.4 Hz, 1H, CHC(O)), 2.70 (s, 3H, N=C(S)CH3), 2.50-2.42 (m, 3H), 2.24-2.19 (m, 1H), 2.17 (d, J = 17.0 Hz, 1H, CH2CO2), 2.07 (s, 3H, CH=CCH3), 1.93-1.83 (m, 3H), 1.73 (bs, 1H), 1.59-1.52 (m, 3H), 1.51-1.43 (m, 1H), 1.34-1.22 (m, 5H), 1.21-1.12 (m, 2H), 1.02 (s, 3H, C(CH3)2); 13C NMR (100 MHz, CDCl3) δ 221.27, 173.30, 165.04, 152.42, 137.41, 134.02, 126.68, 120.60, 116.80, 79.52, 71.59, 71.53, 53.85, 43.83, 38.41, 37.97, 36.69, 31.70, 28.14, 27.45, 24.84, 24.27, 22.49, 20.64, 19.46, 16.05, 15.12; HRMS calcd for C27H40NO5S 490.26272 [M + H]+, found 490.26064.</p><!><p>A solution of Yamaguchi lactonization product 62 (89 mg, 0.124 mmol) in CH2Cl2 (0.1 mL) was treated with a freshly prepared CF3CO2H/CH2Cl2(0.73 mL, v/v, 1:4) at −20 °C. The reaction mixture was allowed to reach 0 °C in 20 min and was stirred for additional 1 h at that temperature at which time all silyl ether disappeared from TLC plate. The mixture was diluted with CH2Cl2 (5 mL) and carefully neutralized by saturated aqueous NaHCO3. After separation, the aqueous phase was further extracted with CH2Cl2 (2 × 5 mL). The combined organics were dried over MgSO4, filtered and concentrated. The resulting residue was purified by preparative thin-layer chromatography (CH2Cl2/MeOH, 20/1) to afford pure desired epothilone C analog 7 (53.4 mg, 88%) as a colorless oil: Rf = 0.36 (CH2Cl2/MeOH, 15:1); [α]22D −86.8 (c 1.0, CHCl3); IR (thin film) νmax 3478 (br), 2928, 2860, 1736, 1678, 1507, 1443, 1409, 1291, 1248, 1187, 1084, 1046, 982, 913, 731 cm−1; 1H NMR (600 MHz, CDCl3) δ 6.97 (s, 1H, SCH=C), 6.62 (s, 1H, CH=CCH3), 5.50 (ddd, J = 10.5, 10.5, 5.0 Hz, 1H, CH=CH), 5.40 (ddd, J = 10.5, 10.5, 5.0 Hz, 1H, CH=CH), 5.22 (d, J = 8.3 Hz, 1H, CHOC(O)), 4.42 (dd, J = 11.5, 1.8 Hz, 1H, CHOHC(CH3)2), 4.20 (s, 1H, CHCHOH), 3.89 (s, 1H, OH), 3.49 (s, 1H, OH), 2.98 (d, J = 10.8 Hz, 1H, CHC(O)), 2.76-2.63 (m, 4H), 2.50 (dd, J = 14.7, 11.6 Hz, 1H, CH2CO2), 2.32 (dd, J = 14.7, 2.3 Hz, 1H, CH2CO2), 2.30-2.24 (m, 1H), 2.18 (tt, J = 10.7, 7.6 Hz, 1H), 2.08 (s, 3H, CH=CCH3), 2.04-1.85 (m, 3H), 1.84-1.74 (m, 1H), 1.66-1.43 (m, 4H), 1.36-1.29 (m, 2H), 1.28 (s, 3H, C(CH3)2), 1.26-1.16 (m, 2H), 1.06 (s, 3H, C(CH3)2)); 13C NMR (100 MHz, CDCl3) δ 220.74, 170.52, 165.35, 152.04, 139.43, 133.17, 125.23, 119.44, 115.82, 78.76, 73.05, 69.62, 54.22, 43.97, 39.79, 39.54, 31.91, 30.05, 28.83, 28.10, 25.17, 23.92, 23.58, 20.85, 19.25, 17.77, 16.26; HRMS calcd for C27H40NO5S 490.26272 [M + H]+, found 490.26144.</p><!><p>To a solution of bridged epothilone C (7) (23 mg, 0.047 mmol, 1.0 equiv) in dry CH2Cl2 (2mL) at −50 °C was added a freshly prepared dry solution of 3,3-dimethyldioxirane (1.18 mL, ca. 0.094 mmol, 0.08 M in acetone, 2.0 equiv). The resulting solution was allowed to warm to −30 °C for 2 h. A stream of argon was then bubbled through the solution to remove excess dimethyldioxirane. The crude mixture was determined to be a mixture of diastereomeric cis-epoxides (ca. 5:2 ratio by 1H NMR). Preparative thin-layer chromatography (CH2Cl2/MeOH, 20/1) to afford bridged epothilone A (7) (13.0 mg, 55%) as a white foam and the cis-epoxide diastereomer 5a (7.0 mg, 29%) as a white solid. 5: Rf = 0.34 (CH2Cl2/MeOH, 15:1); [α]22 D −26.9 (c 0.87, CHCl3); IR (thin film) νmax 3462 (br), 2930, 2864, 1731, 1679, 1509, 1444, 1413, 1390, 1293, 1258, 1181, 1154, 1085, 1046, 980, 919, 725 cm−1; 1H NMR (600 MHz, CDCl3) δ 6.98 (s, 1H, SCH=C), 6.61 (s, 1H, CH=CCH3), 5.35 (dd, J = 9.9, 1.5 Hz, 1H, CHOC(O)), 4.38 (d, J = 10.3 Hz, 1H, CHOHC(CH3)2), 4.31 (s, 1H, CHCHOH), 4.00 (s, 1H, OH), 3.83 (bs, 1H, CHCHOH), 3.03 (ddd, J = 9.6, 3.0, 3.0 Hz, 1H, CH2CH-O(epoxide)CH ), 2.98 (d, J = 10.6 Hz, 1H, CHC(O)), 2.96-2.93 (ddd, J = 9.6, 3.0, 3.0 Hz, 1H, CH2CH-O(epoxide)CH), 2.68 (s, 3H, N=C(S)CH3), 2.52 (dd, J = 14.5, 11.3 Hz, 1H, CH2CO2), 2.30 (dd, J = 14.5, 2.5 Hz, 1H, CH2CO2), 2.24 -2.18 (m, 1H), 2.09 (s, 3H, CH=CCH3), 2.08-2.02 (m, 1H), 1.95-1.84 (m, 3H), 1.80 (ddd, J = 15.0, 9.9, 9.9 Hz, 1H), 1.68 (ddd, J = 25.4, 12.1, 5.3 Hz, 1H), 1.63-1.56 (m, 2H), 1.56-1.46 (m, 2H), 1.44-1.39 (m, 1H), 1.38-1.22 (m, 6H), 1.08 (s, 3H, C(CH3)2); 13C NMR (100 MHz, CDCl3) δ 220.87, 170.45, 165.54, 151.68, 138.84, 120.16, 116.19, 76.96, 72.93, 68.08, 57.70, 55.75, 54.11, 43.80, 39.82, 39.65, 31.89, 30.90, 27.98, 25.34, 25.26, 24.74, 23.36, 21.04, 19.25, 17.96, 16.07; HRMS calcd for C27H40NO6S 506.25763 [M + H]+, found 506.25654. cis-Epoxide diastereomer 5a: Rf = 0.34 (CH2Cl2/MeOH, 15:1); [α]22D −58.9 (c 1.0, CHCl3); IR (thin film) νmax 3466 (br), 2926, 2860, 1737, 1679, 1556, 1509, 1447, 1413, 1390, 1324, 1297, 1254, 1189, 1150, 1085, 1042, 1004, 984, 953, 919 cm−1; 1H NMR (600 MHz, CDCl3) δ 6.99 (s, 1H, SCH=C), 6.65 (s, 1H, CH=CCH3), 5.60 (t, J =3.9 Hz, 1H, CHOC(O)), 4.36 (dd, J = 11.1, 1.9 Hz, 1H, CHOHC(CH3)2), 4.16 (s, 1H, CHCHOH), 3.99 (s, 1H, OH), 3.80 (bs, 1H, OH), 3.24 (ddd, J = 7.4, 4.4, 4.4 Hz, 1H, CH2CH-O(epoxide)CH), 3.11 (dd, J = 12.3, 1.7 Hz, 1H, CHC(O)), 3.05-3.02 (m, 1H, CH2CH-O(epoxide)CH), 2.69 (s, 3H, N=C(S)CH3), 2.56 (dd, J = 14.7, 11.2 Hz, 1H, CH2CO2), 2.37 (dd, J = 14.6, 2.6 Hz, 1H, CH2CO2), 2.10 (s, 3H, CH=CCH3), 2.09-1.96 (m, 3H), 1.93-1.82 (m, 3H), 1.66-1.46 (m, 6H), 1.39-1.18 (m, 6H), 1.08 (s, 3H, C(CH3)2); 13C NMR (100 MHz, CDCl3) δ 221.01, 169.96, 165.32, 151.91, 136.90, 119.81, 116.12, 75.86, 72.88, 68.25, 57.23, 54.24, 53.79. 44.18, 39.78, 39.59, 30.83, 29.01, 28.28, 25.42, 25.06, 24.51, 22.20, 21.08, 19.27, 18.82, 16.31; HRMS calcd for C27H40NO6S 506.25763 [M + H] +, found 506.25659. The direction of the epoxide was further determined by 1D and 2D NOE.</p><!><p>The desired C6-C8 bridged epothilone D (8) was prepared from bis(silyl ether) macrolactone 63 (205 mg, 0.28 mmol) by treatment with CF3CO2H according to the same procedure described above for the preparation of 7, to obtain pure lactone 10 (137 mg, 91%) as a colorless oil or white foam: Rf = 0.46 (CH2Cl2/MeOH, 15:1); [α]22D −109.9 (c 1.35, CHCl3); IR (thin film) νmax 3435 (br), 2934, 2864, 1733, 1679, 1509, 1447, 1413, 1378, 1336, 1293, 1251, 1185, 1143, 1081, 1046, 984, 938, 914, 849, 714, 683 cm−1; 1H NMR (600 MHz, CDCl3) δ 6.95 (s, 1H, SCH=C), 6.60 (s, 1H, ArCH=CCH3), 5.13 (m, 2H, CHOC(O), CH2CH=CCH3), 4.62-4.39 (m, 1H, CHOHC(CH3)), 4.28 (s, 1H, CHCHOH), 3.87 (s, 1H, OH), 3.69 (d, J = 5.9 Hz, 1H, OH), 2.98 (dd, J = 12.4, 2.2 Hz, 1H, CHC(O)), 2.67 (s, 3H, N=C(S)CH3), 2.62 (ddd, J = 15.2, 10.1, 10.1 Hz, 1H), 2.44 (dd, J = 14.2, 11.4 Hz, 1H, CH2CO2), 2.35-2.30 (m, 1H), 2.26-2.21 (m, 2H), 2.06 (d, J = 1.1 Hz, 3H, ArCH=CCH3), 2.02-1.85 (m, 2H), 1.81-1.74 (m, 2H), 1.67 (s, 3H, CH=CCH3CH2), 1.62-1.44 (m, 5H), 1.38-1.19 (m, 6H), 1.04 (s, 3H, C(CH3)2); 13C NMR (100 MHz, CDCl3) δ 221.06, 170.48, 165.33, 151.94, 140.02, 138.48, 121.17, 119.17, 115.61, 79.13, 73.02, 69.77, 54.28, 43.81, 40.03, 39.82, 32.89, 32.13, 29.93, 27.03, 25.26, 23.91, 23.88, 23.39, 20.89, 19.18, 17.02, 16.28; HRMS calcd for C28H42NO5S 504.27837 [M + H]+, found 504.27762.</p><!><p>To a solution of bridged desoxyepothilone B (8) (0.20 mg, 0.04 mmol, 1.0 equiv) in CH2Cl2 (0.4 mL) at −78 °C was added freshly prepared 3,3-dimethyldioxirane (1.0 mL, ca. 0.08 mmol,ca. 0.08 M in acetone, 2.0 equiv) dropwise. The resulting solution was warmed to −50 °C for 1 h, and another portion of dimethyldioxirane (0.4 mL, 0.032 mmol) was added. After stirring at −50 °C for additional 2.5 h, A stream of argon was then bubbled through the solution at −50 °C to remove excess dimethyldioxirane and solvent. The crude reaction mixture was determined to be >20:1 ratio of diastereomeric cis-epoxides by 1H NMR spectroscopy. The resulting residue was purified by preparative thin-layer chromatography (CH2Cl2/MeOH, 30/1) to afford bridged epothilone B (6) (10.8 mg, 52%) as a white foam: Rf = 0.39 (CH2Cl2/MeOH, 15:1); [α]22D −57.8 (c 1.0, CHCl3); IR (thin film) νmax 3470 (br), 2934, 2864, 1737, 1679, 1505, 1467, 1444, 1413, 1382, 1324, 1293, 1251, 1177,1154, 1073, 1042, 984, 941, 880, 760 cm−1; 1H NMR (600 MHz, CDCl3) δ 6.98 (s, 1H, SCH=C), 6.61 (s, 1H, ArCH=CCH3), 5.33 (dd, J = 8.9, 2.7 Hz, 1H, CHOC(O)), 4.43 (d, J = 10.0 Hz, 1H, CHOHC(CH3)2), 4.37 (s, 1H, CHCHOH), 4.13 (s, 1H, OH), 4.00 (s, 1H, OH), 3.01 (dd, J = 12.1, 1.8 Hz, 1H, CHC(O)), 2.79 (dd, J = 9.1, 2.8 Hz, 1H, CH2CH-O(epoxide)C), 2.68 (s, 3H, N=C(S)CH3), 2.50 (dd, J = 14.2, 10.7 Hz, 1H, CH2CO2), 2.24 (dd, J = 14.3, 2.6 Hz, 1H, CH2CO2), 2.16 (ddd, J = 15.1, 2.8, 2.8 Hz, 1H), 2.08 (s, 3H, CH=CCH3), 2.07-2.00 (m, 1H), 1.97-1.72 (m, 5H), 1.72-1.47 (m, 5H), 1.41-1.20 (m, 9H), 1.07 (s, 3H, C(CH3)2); 13C NMR (100 MHz, CDCl3) δ 221.45, 170.43, 165.59, 151.62, 138.88, 120.07, 116.10, 77.02, 72.99, 68.27, 62.51, 61.72, 53.94, 43.76, 40.21, 39.76, 33.06, 32.68, 30.92, 25.47, 25.33, 24.77, 23.50, 23.00, 21.14, 19.23, 17.70, 16.18; HRMS calcd for C28H42NO6S 520.27328 [M + H]+, found 520.27228.</p><!><p>Tubulin was prepared by two cycles of temperature dependent assembly–disassembly as described by Williams and Lee. [49] Determination of protein concentration was carried out as previously described. [50]</p><!><p>Tubulin assembly was monitored by apparent absorption at 350 nm. Tubulin samples (10 μM) were incubated at 37 °C in PME buffer (100 mM PIPES, 1 mM MgSO4, and 2 mM EGTA) containing 1 mM GTP in the spectrometer until a baseline was obtained. The ligand to be tested in DMSO was added a final concentration of 50 μM (4% DMSO). After the assembly reached steady state, the temperature was dropped to 4 °C. A tubulin sample (10 μM) with 4% DMSO and no promoter was used as the reference.</p><!><p>Cytotoxicity of bridged epothilones in PC3 cells was assessed with the SRB assay. [51] The A2780 ovarian cancer cell line assay was performed at Virginia Polytechnic Institute and State University as previously reported. [52]</p>

PubMed Author Manuscript

Effects of carboxyl-terminal methylation on holoenzyme function

Phosphoprotein Phosphatases (PPPs) are enzymes highly conserved from yeast and human and catalyze the majority of serine and threonine dephosphorylation in cells. To achieve substrate specificity and selectivity, PPPs form multimeric holoenzymes consisting of catalytic, structural/scaffolding, and regulatory subunits. For the Protein Phosphatase 2A (PP2A)-subfamily of PPPs, the holoenzyme assembly is at least in part regulated by an unusual carboxyl-terminal methyl-esterification, commonly referred to as \xe2\x80\x9cmethylation\xe2\x80\x9d. Carboxyl-terminal methylation is catalyzed by leucine carboxyl methyltransferase-1 (LCMT1) that utilizes S-adenosyl-methionine (SAM) as the methyl donor and removed by protein phosphatase methylesterase 1 (PME1). For PP2A, methylation dictates regulatory subunit selection and thereby downstream phosphorylation signaling. Intriguingly, there are four families of PP2A regulatory subunits, each exhibiting different levels of methylation sensitivity. Thus, changes in PP2A methylation stoichiometry alters the complement of PP2A holoenzymes in cells and creates distinct modes of kinase opposition. Importantly, selective inactivation of PP2A signaling through the deregulation of methylation is observed in several diseases, most prominently Alzheimer\xe2\x80\x99s disease (AD). In this review, we focus on how methylation of the carboxyl-terminus of the PP2A subfamily (PP2A, PP4, and PP6) regulates holoenzyme function and thereby phosphorylation signaling, with an emphasis on AD.

effects_of_carboxyl-terminal_methylation_on_holoenzyme_function

Introduction<!>PP2A, PP4, and PP6 holoenzymes<!>Carboxyl-terminal methylation of the catalytic subunits of the PP2A subfamily<!>The regulatory function of carboxyl-terminal methylation<!>Challenges in the detection of carboxyl-terminal methylation<!>Regulation of PP2A carboxyl-terminal methylation<!>The role of PP2A carboxyl-terminal methylation in Alzheimer\xe2\x80\x99s disease<!>Conclusion

<p>Reversible protein phosphorylation regulates most cellular processes. In mammalian cells, ~98% of phosphorylation occurs on serine and threonine residues [1,2]. There are more than 400 protein kinases encoded in the human genome that catalyze the phosphorylation of serine and threonine residues [3]. Conversely, the family of Phosphoprotein Phosphatases (PPPs) carry out the majority of the opposing dephosphorylation reactions [4,5]. The PPP family consists of PP1, PP2A, PP2B, PP4, PP5, PP6, and PP7, with 90% of all serine and threonine dephosphorylation being attributed to PP1 and PP2A [4,5]. Most PPPs achieve substrate specificity and selectivity through the formation of holoenzymes of catalytic and non-catalytic subunits [4,5] (Figure 1A).</p><!><p>The PP2A subfamily of serine/threonine phosphatases includes PP2A, PP4, and PP6, whose catalytic subunits contain a phosphatase domain that is highly conserved with other members of the PPP family from yeast to human [6,7]. The PPP catalytic subunits form heterodimeric or heterotrimeric holoenzyme complexes with scaffolding and regulatory subunits [4,8]. Despite the high sequence similarities between the catalytic subunits, PP2A, PP4, and PP6 have their own unique sets of structural and regulatory subunits. The PP2A holoenzyme is formed by a structural scaffolding 'A' subunit (PPP2R1A), a catalytic 'C' subunit (PPP2CA), and a regulatory B subunit. The A and the C subunits both have two isoforms, α and β, and form a high affinity dimer (AC) that binds to a regulatory B subunit or other interacting proteins. There are 15 human genes encoding regulatory B subunits, which belong to four distinct families: B55 (B/PPP2R2/PR55), B56 (B'/PPP2R5/PR61), B72 (B"/PPP2R3/PR72), and the striatin family (B'"/STRN/PR93/PR110) [9,10] (Figure 1B). Alternative splicing of several regulatory subunit genes generates additional isoforms resulting in the formation of up to 100 different PP2A holoenzymes with defined substrate specificity, subcellular localization, and phosphatase activity. The PP4 catalytic subunit (PPP4C) forms both heterodimers and heterotrimers with five different regulatory subunits: PPP4R1, PPP4R2, PPP4R3A, PPP4R3B, and PPP4R4 [11] (Figure 1C), while the PP6 catalytic subunit (PPP6C) forms heterotrimeric complexes with one of three Sit4-associated protein subunits (SAPS1-3 or PPP6R1-3) and one of three Ankyrin repeat (ANKRD28/44/52) subunits [12,13] (Figure 1D).</p><!><p>The carboxyl-terminal regions of PPP2CA, PPP4C, and PPP6C play an important role in facilitating the interactions with regulatory subunits. The last six amino acids of PPP2CA (TPDYFL-309) and the last three amino acids of PPP2CA, PPP4C, and PPP6C are identical from yeast to human and are post-translationally modified to regulate holoenzyme function (Figure 1E). Unusually, the α-carboxyl group of the carboxyl-terminal leucine (Leu 309 in human PPP2CA, Leu 307 in human PPP4C, and Leu 305 in human PPP6C) of the catalytic subunits undergoes methyl-esterification [14-17], hereafter referred to as "methylation" as is done by convention in the field. In mammalian cells, the majority of PPP2CA is methylated. For example, in NIH3T3 cells more than 90% of PPP2CA is methylated [18] while in transformed human embryonic kidney (HEK) cells [19] about 74% of PPP2CA is methylated [20].</p><p>For all members of the PP2A subfamily, reversible methylation is catalyzed by the enzyme Leucine carboxyl methyltransferase-1 (LCMT1), which utilizes S-adenosyl-methionine (SAM) as the methyl donor [14,15,17,21,22]. The modification is removed by protein phosphatase methylesterase 1 (PME1), at least in the case of PPP2CA and PPP4C [23-25] (Figure 2A). For PPP6C, the role of PME1 in removing the methyl group has not yet been established. The mechanisms of methylation and hydrolysis are best understood for PP2A. The substrate for LCMT1 and PME1 is the PP2A AC heterodimer, not the B subunit-containing heterotrimer [23,26]. Substrate recognition by both enzymes requires structural features beyond the PPP2CA carboxyl-terminus as peptides based on the sequence of the PPP2CA carboxyl-terminus are not modified by either enzyme [15,23]. Additionally, a peptide corresponding to the last eight amino acids of PPP2CA failed to co-crystalize with LCMT1 or bind to the active site when LCMT1 crystals were soaked with it [27]. Indeed, further investigations revealed that interactions of LCMT1 and PME1 with the active site of the PPP2CA catalytic subunit are required to facilitate the reaction. Consequently, mutations that reduce PPP2CA activity (H59Q, H118Q, D85N, and R89A) also reduce carboxyl-terminal methylation of PPP2CA to levels ranging from 3% to 13% depending on the mutations [18]. Consistently, inhibition of phosphatase activity using PPP inhibitors such as okadaic acid inhibits the methylation and demethylation reactions [23,28,29]. Structural analysis of LCMT1 bound to S-adenosyl-homocysteine (SAH), a co-product of the methylation reaction, and PPP2CA revealed that LCMT1 makes contacts with the PPP2CA active site, and the PPP2CA carboxyl-terminus binds to the LCMT1 active site (Figure 2B) [27]. The interaction of LCMT1 with the PPP2CA active site suppresses PP2A activity and promotes a conformation of LCMT1 essential for the methylation reaction [27]. Thus, these observations explain why a peptide composed of the last eight amino acids of PPP2CA is not a substrate of LCMT1. Additional contacts between LCMT1 and PPP2CA are needed to activate LCMT1 and facilitate the methylation reaction. Although the name suggests otherwise, the carboxyl-terminal leucine is not the only amino acid that can be methylated [30]. While deletion of Leu309 abolishes methylation, substitutions of the leucine with alanine or valine are still methylated in vitro [30].</p><p>PME1 also binds to the PPP2CA active site, which results in a structural rearrangement of PME1 active site residues and its activation [31]. This rearrangement places the PME1 active site close to the PPP2CA carboxyl-terminus, allowing the latter to reach into the PME1 active site (Figure 2C). Interestingly, structural analysis of the PME1 and PPP2CA interaction revealed that in vitro, PME1 displaces the two catalytic bivalent cations from the PPP2CA active site, which are required for the catalysis of the dephosphorylation reaction [31]. Thus, over time the phosphatase activity of PPP2CA declines. This is not an immediate process upon binding of PPP2CA to PME1. In cells, the interaction of PME1 and PP2A did not affect PPP2CA activity [32], suggesting that the duration of the PPP2CA-PME1 interaction determines if only demethylation or also inactivation occurs. These observations that LCMT1 and PME1 both require the interaction with the phosphatase active site explain why LCMT1 and PME1 can modify all members of the PP2A subfamily. While PP2A, PP4, and PP6 have unique non-catalytic subunits and hence distinct quaternary structures, their active sites are highly conserved, providing the structural basis for their interactions with LCMT1 and PME1.</p><!><p>The role of the carboxyl-terminus of PPP2CA in PP2A holoenzyme formation and activity was first identified by a mutational analysis deleting the last eight amino acids [33], in which the Thr301stop mutant does not bind to B55 subunits. However, it binds to the polyomavirus middle tumor antigen (MT), a viral protein that can interact with the AC dimer in place of B-type subunits. Single point mutations of L309Q [34] or L309A [35] result in the recovery of mostly AC dimers, indicating disruption of the ABC trimer by lack of methylation. Deletion of Leu 309 (L309Δ) abolished the interaction of PPP2CA with B55, but not B56, B72, and the binding of striatin and S/G2 nuclear antigen are increased [18,30]. Furthermore, MT and other viral tumor antigens such as polyomavirus small (PyST) tumor antigen and the SV40 small tumor antigen (SVST) bound to PPP2CA L309Δ as well as the catalytically inactive, unmethylated PPP2CA mutant D85N [18,20]. Consistently, endogenous B55 isoforms have been shown to exclusively bind to methylated PPP2CA (Figure 2D) [26,30]. However, B56 containing trimers have been reported to contain methylated as well as unmethylated catalytic subunit (Figure 2D) [26,30]. For both B55 and B56 subunits, methylation increases the affinity of the AC dimer. While methylation per se does not reduce PP2A activity, the methylation-dependent recruitment of B subunits does [26]. B72 subunits also bind both methylated and unmethylated PPP2CA (Figure 2D) [30]. The depletion of LCMT1 reduces PPP2CA methylation over time and results in preferential incorporation of the remaining methylated PPP2CA into B55-containing trimers, as compared to B56- or B72-containing holoenzymes [30]. Once methylation decreases below a threshold needed to incorporate available B55 subunits into trimeric holoenzymes, non-complexed B55 is degraded [30]. The role of methylation in holoenzyme assembly is conserved to yeast, which express B55 and B56 regulatory subunits [36,37]. This methylation-dependent selection of regulatory subunits, specifically of B55 and B56, is of major consequence to cellular signaling. PP2A-B55 and PP2A-B56 have unique and non-overlapping substrate preferences [38]. While PP2A-B55 nearly exclusively dephosphorylates proline-directed serine and threonine phosphorylation sites, PP2A-B56 mostly dephosphorylates basophilic motifs, thereby generating unique modes of kinase opposition. In this way, changes in the relative composition of the repertoire of PP2A-B55 and −B56 holoenzymes differentially impacts cellular signaling pathways. For instance, a decrease in PP2A methylation will first affect PP2A-B55 holoenzymes, resulting in a stabilization and a prolonged half-life of cyclin-dependent kinase- and mitogen-activated protein kinase-dependent phosphorylation sites.</p><p>PPP4C and PPP6C are also highly methylated by LCMT1 in mouse embryonic fibroblasts [17]. For PPP4C, loss of methylation reduces PPP4R1 binding, while PP6 holoenzyme components bind PPP6C methylation-independently [17]. Taken together, these observations suggest that methylation is an essential regulatory mechanism in the assembly of PP2A holoenzymes, and to a certain degree of PP4 holoenzymes, that determines the cellular repertoire of PP2A and PP4 trimeric complexes.</p><!><p>PPP2CA methylation is most commonly detected using methyl-specific antibodies, and methylation stoichiometry is determined by hydrolysis of the carboxyl-terminal ester using base followed by detection using demethyl-specific antibodies. However, recently the specificity of several PP2A antibodies has come under scrutiny. Antibodies employed to detect and immunoprecipitate PPP2CA for activity assays have been shown to exhibit differential preferences for methylated and demethylated PPP2CA [39]. Many antibodies against PPP2CA were raised against demethylated peptides from the carboxyl-terminus of PPP2CA and preferentially recognize and precipitate demethylated PPP2CA. Thus, methylation-based biases in the detection of PPP2CA can significantly impact the interpretation of existing data. Because of the high occupancy of PPP2CA methylation and the methylation-dependent regulation of regulatory subunit binding, analyses using these antibodies provides an incomplete presentation of the PP2A holoenzyme. Furthermore, because of the high conservation of the carboxyl-terminus between PPP2CA, PPP4C, and PPP6C, some PP2A antibodies cross-react with PP4 and PP6 [39]. In addition, the carboxyl-terminus of PPP2CA (TPDYFL-309) can be modified by phosphorylation. For PPP2CA, phosphorylation has been reported to occur on threonine (Thr 304 in human PPP2CAα) and tyrosine (Tyr 307 in human PPP2CAα) and was found to impact PPP2CA recognition by demethyl-specific antibodies [30,39-41]. Interestingly, some monoclonal antibodies used to detect Tyr 307 phosphorylation do not discriminate between phosphorylated and unphosphorylated Tyr 307, but are sensitive to phosphorylation of Thr 304 or the methylation state of the carboxyl-terminus [42,43]. Thus, the selection of antibodies for the characterization of PPP2CA expression, its modification states, and immunoprecipitation of PP2A holoenzyme complexes must be carefully considered. Correspondingly, the existing literature on PP2A might require a reinterpretation of results, depending on the antibodies used in the respective study.</p><!><p>The regulation of LCMT1 and PME1 and thereby the methylation status of the PP2A subfamily is less well understood. Some reports suggest that the stoichiometry of PPP2CA methylation changes during the cell cycle and in a subcellular location-specific manner [44,45]. LCMT1 is mainly localized in the cytoplasm, while PME1 is enriched in the nucleus [46]. This distribution correlates with an enrichment of demethylated PPP2CA in the nucleus [46]. In mouse neuroblastoma (N2A) cells, LCMT1, methylated PPP2CA and PP2A-B55 are enriched in membrane rafts and cholesterol- and sphingolipid-enriched microdomains (CEMs), while demethylated PPP2CA and PME1 are present in non-raft membrane microdomains [47]. LCMT1 and PME1 themselves are further regulated by post-translational modifications. PME1 is phosphorylated by Ca2+/calmodulin-dependent kinase 1 (CaMK1) [48] and salt-inducible kinase 1 (SIK1) [49], both of which regulate the binding of PME1 to PP2A.</p><p>Intriguingly, it was recently shown that the availability of the LCMT1 co-substrate SAM is a major regulatory factor of the methylation reaction. SAM is the universal methyl-donor in cells for proteins, nucleic acids, lipids, and secondary metabolites. Upon methyl-transfer, SAM is converted to S-adenosylhomocysteine (SAH), which is recycled into SAM through the methionine cycle (Figure 3). SAH is cleaved in a reversible reaction by the enzyme SAH hydrolase into adenosine and homocysteine (Hcy). Hcy is remethylated to methionine by the betaine-homocysteine S-methyltransferase (BHMT) using betaine as a co-substrate (Figure 3) [50]. Alternatively, 5,10-methylene-tetrahydrofolate (5,10-methylene-THF) is converted to 5-methyl-THF by the methylene-tetrahydrofolate reductase (MTHFR), which functions as a co-substrate in the remethylation of Hcy by methionine synthase (MS) in a vitamin B12-dependent reaction (Figure 3) [50]. Finally, methionine is converted to SAM by the S-adenosylmethionine synthase (Figure 3) [50]. The availability of methionine and thereby SAM dictates cellular methylation levels, including that of PPP2CA [51,52]. Indeed, the treatment of mammalian cells with SAH reduced PPP2CA methylation by inhibiting LCMT1[53]. In yeast, transfer from nutrient-rich to minimal growth media can induce non-nitrogen starvation (NNS)-induced autophagy, which depends on methionine and SAM availability [51]. Methionine starvation activates NNS autophagy and results in the demethylation of PPP2CA, which increases the phosphorylation of proteins involved in NNS autophagy and other growth-related processes [51]. Furthermore, methionine starvation-induced PPP2CA demethylation and inactivation leads to increased phosphorylation and activation of histone H3K36 demethylases and histone demethylation [54]. Thus, PPP2CA acts as an amino acid sensor as its methylation status can be linked to amino acid availability.</p><!><p>The regulation of the methylation cycle and thereby PPP2CA methylation and PP2A holoenzyme formation have been linked to the neuropathological symptoms of Alzheimer's disease (AD). AD is pathologically characterized by amyloid-beta (Aβ) containing senile plaque deposition, accumulation of neurofibrillary tangles (NFTs), neuronal loss, and severe cognitive impairment [55]. Misfolded protein aggregates are a common feature in AD, including hyperphosphorylated Tau protein deposits in the intraneuronal NFTs [56-58]. Tau is a microtubule-associated protein (MAP) that is highly enriched in neurons and plays a crucial role in maintaining microtubule stability. It has been previously shown that Tau protein can be reversibly phosphorylated on over 40 serine/threonine residues by multiple protein kinases, including glycogen synthase kinase 3 beta (GSK3β) and cyclin-dependent kinase 5 (CDK5), among others [59-61].</p><p>PP2A and PP1 phosphatases have been associated with Tau dephosphorylation, with PP2A-B55α containing holoenzymes acting as the major Tau phosphatases [62]. Low levels of Tau phosphorylation contribute to microtubule assembly and stability. However, hyperphosphorylation of Tau leads to its dissociation from microtubules, aggregation, and formation of NFTs. Tau hyperphosphorylation in AD is often due to decreased PP2A-B55α activity [55,63]. Specific inhibition of PP2A alone is sufficient to form hyperphosphorylated Tau aggregates in vitro and in vivo [64]. Inhibition of serine/threonine phosphatases with a selective PPP inhibitor okadaic acid induces Tau hyperphosphorylation and Aβ deposition, leading to neurodegeneration and cognitive impairment. Moreover, fibroblasts and affected brain regions from AD patients have been shown to have decreased PP2A activity with reduced protein expression of PPP2CA as well as B55 regulatory protein [63,65,66]. Deregulation of PP2A-B55 holoenzymes has been associated with enhanced Tau phosphorylation, microtubule stability and neurite outgrowth in neuroblastoma cells [57].</p><p>Reduced PPP2CA methylation has been associated with Tau hyperphosphorylation and correlates with the severity of phospho-Tau pathology in AD brains [67,68]. LCMT1 has been shown to be downregulated in AD neurons with neurofibrillary tangles [68]. Changes in the methionine cycle have been linked to AD though the regulation of PPP2CA methylation. Elevated Hcy plasma levels are a risk factor for AD and patients with high Hcy plasma levels show more rapid neural atrophy than those with lower levels of plasma Hcy [69]. Higher Hcy levels have also been shown to be a global inhibitor of cellular methylation through increased SAH production, affecting PP2A methylation and holoenzyme formation and in turn affecting Tau phosphorylation and microtubule binding ability [70]. As noted earlier, Hcy is converted to methionine by MTHFR using 5-methyl-THF as a co-substrate in a vitamin B12-dependent reaction [50]. Increased levels of folate and vitamin B12 can decrease Hcy plasma levels by increasing methionine and SAM synthesis. This increased PPP2CA methylation can lead to activation of PP2A-B55 preventing hyperphosphorylation of Tau[71]. Conversely, vitamin B12 and folate deficiency decrease methionine and SAM production and reduce cellular methylation through product inhibition by increased SAH levels. Thus, dietary intervention by supplementation of folic acid and vitamin B12 can decrease plasma Hcy levels, which may help reduce AD occurrence or slow its progression via modulating PP2A methylation. Furthermore, a functional polymorphism in MTHFR (Mthfr 677C→T) causes mild hyperhomocysteinemia due to reduced enzymatic activity and is associated with an increased risk of AD in Asian populations [72,73]. Mthfr deficient mice show reduced LCMT1 expression and PPP2CA methylation, and hyperphosphorylation of Tau in the hippocampus and cerebellum, and to a lesser degree in the cortex [58]. Thus, reactivation of PP2A-B55 through increased methylation of PPP2CA represents a therapeutic option in the treatment of AD. Besides dietary interventions using folic acid and vitamin B12, specific inhibitors of PME1 have been developed to inhibit methyl-hydrolysis. ABL127 [74] and AMZ30 [75] inhibit PME1 at nanomolar concentrations by covalently binding to an active site serine. However, only limited cellular effects have been observed upon PME1 inhibition [76], potentially due to the high stoichiometry of carboxyl-terminal methylation of PPP2CA or the lack of inhibition of the second mechanism of PPP2CA inactivation by displacement of the two catalytic bivalent cations from the active site.</p><!><p>The PP2A-subfamily of PPPs, specifically PP2A, are major cellular serine/threonine phosphatases whose substrate selection and specificity are determined by the regulatory subunits to which they bind. Thus, determining the mechanisms controlling holoenzyme assembly is essential for understanding phosphorylation signaling by PPPs. While phosphorylation of the carboxyl-terminus of the catalytic subunit can regulate B subunit binding, the modulation of the affinity of the PP2A AC dimer for different B-type regulatory subunits through methylation adds an orthogonal, phosphorylation-independent mode of regulation. Utilizing SAM as the methyl-donor, the methylation reaction connects phosphorylation signaling with the metabolic state of the cell, making PP2A a sensor of amino acid and energy availability. This is specifically important in the context of AD, where metabolic changes lead to hyperphosphorylated states due to changes in PP2A holoenzyme composition.</p><p>In most mammalian cells and tissues, the majority of PP2A is methylated. The high methylation stoichiometry suggests that rather than dynamically changing PPP holoenzyme composition, methylation establishes a steady-state repertoire of PPP holoenzymes that disconnects B subunit abundance from holoenzyme incorporation. While B55 subunits are degraded when not part of PP2A holoenzymes, other B subunits are stable, and only a portion of them is incorporated into PP2A holoenzymes. Thus, PP2A methylation might serve as a rheostat, fine-tuning the affinity of AC dimers to ensure a sufficient supply of each B regulatory subunit-containing holoenzyme to balance kinase activities. Further investigations into how B regulatory subunits are incorporated into holoenzyme complexes are needed to elucidate this mechanism. PP2A-B55 and −B56 holoenzymes exhibit distinct substrate recruitment mechanisms and phosphorylation site preferences. Thus, understanding the mechanisms governing the cellular abundances of specific B-subunit containing holoenzymes is essential for elucidating kinase opposition in the regulation of phosphorylation signaling.</p>

PubMed Author Manuscript

Micelle Enhanced Fluorimetric and Thin Layer Chromatography Densitometric Methods for the Determination of (±) Citalopram and its S – Enantiomer Escitalopram

Two sensitive and validated methods were developed for determination of a racemic mixture citalopram and its enantiomer S-(+) escitalopram. The first method was based on direct measurement of the intrinsic fluorescence of escitalopram using sodium dodecyl sulfate as micelle enhancer. This was further applied to determine escitalopram in spiked human plasma, as well as in the presence of common and co-administerated drugs. The second method was TLC densitometric based on various chiral selectors was investigated. The optimum TLC conditions were found to be sensitive and selective for identification and quantitative determination of enantiomeric purity of escitalopram in drug substance and drug products. The method can be useful to investigate adulteration of pure isomer with the cheap racemic form.

micelle_enhanced_fluorimetric_and_thin_layer_chromatography_densitometric_methods_for_the_determinat

Introduction<!>Instrumentation<!>Chemicals and reagents<!>Extraction of escitalopram from cipralex tablets<!>Characterization of isolated escitalopram<!>Standard solutions<!>Synthetic mixtures<!>Spectrofluorimetric method<!>TLC method<!>Application to tablets<!>Application to spiked human plasma<!>Results and Discussion<!>Fluorimetric method<!>TLC-densitometric method<!>Method validation<!>Conclusion

<p>Citalopram (Fig. 1) a selective serotonine re-uptake inhibitor (SSRI), has been used for the treatment of depression, social anxiety disorder, panic disorder and obsessive-compulsive disorder.1–3 Citalopram is sold as a racemic mixture, consisting of 50% R-(−)-citalopram and 50% S-(+)-citalopram. As the S-(+) enantiomer has the desired antidepressant effect4 it is now marketed under the generic name of escitalopram. It has been shown that the R-enantiomer present in citalopram counteracts the activity of escitalopram. Citalopram and escitalopram have demonstrated different pharmacological and clinical effects.5</p><p>A number of techniques including spectrophotometric,6,7 fluorimetric,8,9 electrochemical,10 chromatography11,12 and capillary electrophoresis13,14 have been developed for the determination of enantiomeric citalopram. Although several chiral methods including LC15–20 and CE21,22 are available for separation of racemic citalopram, there is no report concerning enantiomeric separation of citalopram using thin layer chromatography (TLC).</p><p>In this study we develop two simple, economic and validated methods for determination of escitalopram and enantioseparation of its racemic mixture in drug substance and drug products. The fluorimetric method was based on the fluorescence spectral behavior of escitalopram in micellar systems, such as Triton® X-100, Cetylpyridinium bromide; and sodium dodecyl sulfate (SDS). The fluorescence intensity of escitalopram and its racemic mixture citalopram was compared under the same experimental conditions. The method was successfully applied to the analysis of escitalopram in drug substances, drug product as well as spiked human plasma. Furthermore, the method was found to tolerate high concentrations of co-administrated and common drugs without potential interference. In addition to the fluorimetric method, TLC densitometry was proposed for stereoselective separation of (±) citalopram and determination of its enantiomer, escitalopram, using the different chiral selectors namely, brucine sulphate, chondroitin sulphate, heparin sodium and hydroxypropyl-β-cyclodextrin (HP-β-CD). The developed TLC method based on chiral mobile phase additives (CMPAs), tend to be cheap and feasible and offer a potential strategy for simultaneous separation of different chiral drugs on the same plate. The method was validated according to ICH guidelines and can become the method of choice compared to other techniques for fast routine enantiomeric analysis.</p><!><p>Waters-2525 LC system, equipped with a dual wavelength absorbance detector 2487, an auto-sampler injector and Mass Lynx v 4.1, was used. The LC column was C18 reverse-phase column (4.6 mm diameter × 100 mm length, 5 μm particles, phenomenex, monolithic).1 H-NMR spectra were recorded on Bruker Avance-400 spectrometer operating at 400 MHz. FT-IR spectrometer Avatar 360 was used. Spectrofluorimetric measurement was carried out using a Shimadzu spectrofluorimeter Model RF-1501 equipped with xenon lamp and 1-cm glass cells. Excitation and emission wavelengths were set at 242 nm and 306 nm respectively. Precoated TLC plates (10 × 10 cm, aluminum plate coated with 0.25 mm silica gel F254) were purchased from Merck Co., Egypt. Samples were applied to the TLC plates with 25 μL Hamilton microsyringe. UV short wavelength lamp (Desaga Germany) and Shimadzu dual wavelength flying spot densitometer, Model CS-9301, PC were used. The experimental conditions of the measurements were as follows: wavelength = 240 nm, photo mode = reflection, scan mode = zigzag, and swing width = 10.</p><!><p>All chemicals used were of analytical grade if not stated otherwise. Escitalopram oxalate (Alkan Pharm Co., Egypt) certified to contain 99.60% was used as the reference standard. Cipralex containing 10 mg escitalopram oxalate per tablet (manufactured by Lund beck Co., Denmark, Batch No 2147226, Mfg. D: 2008, Exp. D: 2011) was purchased from the market. Escitalopram oxalate was extracted and purified from cipralex tablets. Citalopram was kindly supplied by Adwia Co., Egypt. Its purity was found to be 99.80% according to official HPLC method.23 Lecital, containing 40 mg citalopram hydrobromide per tablet (manufactured by Joswe Medical Co., Jordon) was purchased from the market. Human plasma was kindly supplied from Vacsera, Egypt. Sodium dodecyl sulfate (BHD, Egypt), brucine sulphate (BHD, Egypt), chondroitin sulfate, (Eva Co., Egypt), heparin sodium and 2-hydroxypropyl-β-cyclodextrin (Fluka, Egypt) were purchased. Trifluoroacetic acid (Aldrich, U.K), methanol and acetonitrile (Fisher Scientific, U.K.) were LC grade. Ultra pure water (ELGA, U.K.) was used.</p><!><p>Ten cipralex tablets were finely powdered and transferred to a 100 mL conical flask to which 50 mL methanol was added and stirred for 20 min. The solution was filtered through whatman No. 42 filter paper. The residue was washed several times with small volume of methanol for complete recovery. The combined extract was evaporated and the pure sample was obtained by recryslallization from methanol.</p><!><p>The weight of escitalopram oxalate obtained by extraction and recrystalization was the same as the labeled value. Characterization of the extracted escitalopram was done using UV, TLC, LC-MS and NMR.</p><p>Absorbance spectra were recorded in methanol and TLC separation was carried out using tolueneethyl acetate-triethylmine (7:3.5:3 v/v/v) as the mobile phase.11 LC-MS was used for establishing the purity of escitalopram using a reverse phase C18 column at flow rate of 1 ml/min and acetonitrile: water:trifluoroacetic acid (60:40:0.01% v/v/v) mobile phase. Further characterization included FT-IR, 1H-NMR in deuterated methanol (CD3OD).</p><!><p>Standard stock solutions of (±)citalopram hydrobromide, and S-(+)escitalopram oxalate 0.05 mg mL−1 and 4 mg mL−1 were prepared by dissolving appropriate amounts of each in water and methanol for fluorimetric and TLC methods respectively. The stock solutions were subsequently used to prepare working standards in methanol. All solutions were stored in refrigerator at 4 °C.</p><!><p>For TLC method, synthetic mixtures of escitalopram and citalopram in proportions ranging from 10%–90% were analyzed and the percentage recovery of escitalopram was calculated.</p><!><p>1 mL of aqueous stock solution equivalent to (1.25–162.5 μg mL−1) of escitalopram or (1.25–125.0 μg mL−1) citalopram was transferred into a series of 10 mL volumetric flasks followed by 1 mL SDS (5 mmol aqueous solution). The volume was completed to the mark with methanol. The fluorescence was measured at 306 nm using 242 nm as excitation wavelength. To obtain the standard calibration graph, concentrations were plotted against fluorescence intensity and the linear regression equations were computed.</p><!><p>Chromatograms were developed in clean, dry, paper-lined glass chambers (12 × 24 × 24 cm) pre-equilibrated with developer for 10 minutes. The TLC plates were prepared by running the mobile phase of acetonitrile-water (17:3 v/v) containing 1 mmol chiral selector to the lost front in the usual ascending way and were air-dried. For detection and quantification, 10 μL each of citalopram and escitalopram solutions within the quantification range were applied side-by-side as separate compact spots 20 mm apart and 10 mm from the bottom of the TLC plates using a 25 μL Hamilton micro syringe. The chromatograms were developed up to 8 cm in the usual ascending way using the same mobile phase omitting the chiral selectors, and were then air dried. The plates were visualized at 254 nm or by exposure to I2 vapor and scanned for escitalopram at wavelength 240 nm using the instrumental parameters mentioned above.</p><p>For quantitative determination of escitalopram aliquots of standard solution (4 mg mL−1) equivalent to 0.125–4.000 mg were transferred into 10 mL volumetric flasks and made up to volume with methanol. 10 μL of each concentration was applied on the TLC plate, air dried and scanned for escitalopram at 240 nm using the instrumental parameters mentioned above. The average peak areas were calculated and plotted against concentration. The linear relationship was obtained and the regression equation was recorded.</p><!><p>An accurately weighed amount of powdered tablets equivalent to 100 mg of escitalopram and citalopram were dissolved in 50 mL methanol. The solutions were stirred with magnetic stirrer for 20 min. Each solution was transferred quantitatively to a 50 mL volumetric flask, diluted to the volume with methanol, and filtered. For fluorimetric analysis, a portion equivalent to 25 mg was evaporated, transferred quantitatively to a 50 mL volumetric flask and made up to volume with water. The procedure was completed as mentioned above.</p><!><p>Aliquots equivalent to 0.1–0.4 mg mL−1 of escitalopram were sonicated with 1 mL plasma for 5 minutes. Acetonitrile (2 mL) was added and then centrifuged for 30 minutes. One milliliter of supernatant was evaporated and the procedure was completed as described above.</p><!><p>In this work a simple method was used for isolation of escitalopram from its drug product rather than the published procedure.24 The isolated escitalopram was characterized and confirmed by different analytical techniques as mentioned above.</p><!><p>Escitalopram solution was found to exhibit an intense fluorescence at a wavelength of 306 nm on excitation at 242 nm as shown in Figure 2. Different media such as water, methanol and ethanol were attempted. Maximum fluorescence intensity was obtained upon using methanol as diluting solvent, while water decreases the fluorescence intensity.</p><p>The effect of different surfactants on the fluorescence intensity of escitalopram was studied by adding 1 mL of each surfactant to the aqueous drug solution. CPB and Triton X-100 led to peak broadening and no effect on fluorescence intensity, while SDS caused two fold increasing in the intensity. The fluorescence intensity was stable for at least two hours.</p><p>When compared to its racemic form, escitalopram showed a lower fluorescence intensity. This is concordance with published data giving the molar absorbitivity of escitalopram as 13.630 mol−1cm−1 while that of citalopram is 15.630 mol−1cm−1.8 The quantum yield was found to be 0.026 for escitalopram and 0.030 for citalopram according to the following equation.25 Yu=Ys . Fu/Fs . As/Auwhere Yu and Ys referred to fluorescence quantum yield of escitalopram and quinine sulphate, respectively; Fu and Fs represented the integral fluorescence intensity of escitalopram and quinine sulphate, respectively; Au and As referred to the absorbance of escitalopram and quinine sulphate at the excited wavelength respectively.</p><p>The method was validated by testing linearity, specificity, precision and reproducibility as presented in (Table 1).</p><p>Calibration plot was found to hold good over a concentration range of 0.125–16.25 μg mL−1 and 0.125–12.50 μg mL−1 for escitalopram and citalopram respectively. The procedure gave good reproducibility when applied to escitalopram drug substance over three concentration levels; 3.30, 6.60 and 13.30 μg mL−1. Whereas the specificity was proved by quantitate the studied drug in its tablet form, confirming non-interference from excipients and additives.</p><p>The results were comparable to those given by a reported method8 as revealed by statistical analysis adopting Student's t- and F-tests, where no significant difference was noticed between the two methods as presented in (Table 2). The validity of the procedure was further assured by the recovery of the standard addition. The limit of detection (LOD) and the limit of quantification (LOQ) were found to be 0.017 and 0.056 μgmL−1 respectively.</p><p>The high sensitivity attained by the fluorimetric procedure allowed its successful application to the analysis of escitalopram in spiked human plasma. To avoid variation in background fluorescence, a simple deproteination of plasma samples with acetonitrile was performed followed by centrifugation, the clear supernatant containing escitalopram was analyzed. A calibration graph was obtained by spiking plasma samples with escitalopram in the range 3.30–16.25 μg mL−1. Linear regression analysis of the data gave the equation FI=37.27 C +126 r=0.991 (n=6)where FIis the fluorescence intensity, C is the concentration of escitalopram in plasma in μg mL−1 and r is correlation coefficient. The limit of detection and quantification in spiked plasma were found to be 0.17 μg mL−1 and 0.56 μg mL−1. The average recovery was 98.00% ± 2.80% RSD. The results from analysis of 5 spiked plasma samples are presented in Table 2.</p><p>The interference due to co-administrated and common drugs was investigated in mixed solutions containing 5 μg mL−1 escitalopram and different concentrations of an interferant. The resulting fluorescence was compared to those obtained for escitalopram only at the same concentration. Tolerance was defined as the amount of interferant that produced an error not exceeding 5% in determination of the analyte. The method was found to be selective enough to tolerate high concentration of co-administerated and common drugs. Table 3, shows the maximum tolerable weight ratio for these drugs.</p><p>The fluorimetric method offers simplicity, rapid response and the potential to be efficient for bioavailability assessments and therapeutic drug monitoring of patients treated with citalopram or escitalopram.</p><!><p>Compared to other chromatographic techniques, TLC is a simple, economical, rapid and flexible technique allowing sensitive parallel processing of many samples on one plate. For enantiomeric separation, chiral stationary phases and mobile phase additives can be used. Brucine, chondroitin, heparin and HP-β-CD were used as chiral selectors for enantiomeric separation of different pharmaceutical compounds using TLC, LC and CE.26–28</p><p>The literature reveals that chiral recognition may occur due to formation of inclusion complexes, hydrogen-bonding, π–π interaction, hydrophobic interaction or steric repulsion.29 For instance, enantioselectivity using brucine arises due to the formation of two diastereomers through simple ionic interactions between racemate and chiral selector, e.g. (+)-citalopram/brucine and (−)-citalopram/brucine.30 The enantiomeric resolution by HP-β-CD may involve the inclusion of drug within the CD cavity relative to the comparability of sizes, shapes and hydrophibicities. Whereas steric effect derived from the anion of chondroitin sulphate contributes mainly to the interactions with drug enantiomer,31 the chiral discriminating capability of heparin is believed to be due to formation of a helical structure in aqueous solution.</p><p>In this work, TLC methodology was developed for separation of (±) citalopram and determination of escitalopram using different chiral selectors, the method depending on the difference in Rf values of (R)- and (S)- forms of (±) citalopram. The experimental conditions such as mobile phase composition, chiral selector, pH and temperature were optimized to provide accurate, precise, reproducible and robust separation. Various chiral mobile phase additives including brucine sulphate, chondroitin sulphate, heparin sodium and HP-β-CD were tested. The best resolution was achieved by using 1 mM of brucine sulphate in acetonitrile:water (17:3 v/v) as a mobile phase (Table 4). The order of enantioselectivity was found to be brucine sulphate > HP-β-CD > heparin sodium > chondroitin sulphate as shown in Figure 3. The Rf values were 0.17, 0.22, 0.22, 0.29 for escitalopram and 0.71, 0.70, 0.66, 0.77 for (R)-citalopram for the four selected chiral additives respectively as shown in Figure 4. Due to it's lower health risks, HP-β-CD was chosen over brucine sulphate for the determination of escitalopram. We also investigated the effect of pH and temperature on resolution of racemic citalopram as they have been known to affect chiral recognition.26 The best conditions for discrimination of citalopram enantimers were found at pH 8.0 and 25 ± 2 °C.</p><!><p>The method was validated according to ICH regulations by documenting its linearity, accuracy, precision, limit of detection and quantification, specificity and, robustness.30,33 The good linearity was obtained for seven concentrations in the range of 0.5– 40 μg/spot as shown in Figure 5. The accuracy based on the mean percentage of measured concentrations (n = 6) to the actual concentration is stated in (Table 1). The precision of the method was assessed by determining RSD% values of intra-and inter-day analysis (n = 9) of escitalopram over three days. Two different analysts performed intermediate precision experiments with separate mobile phase systems according to the proposed procedure. The RSD% values of the intermediate precision are less than 2% for drug substance and drug product. The LOD and LOQ were found to be 0.014 and 0.076 μg/spot respectively (Table 1). The specificity of the method was assessed by analyzing synthetic mixtures of escitalopram and citalopram in different proportions as shown in (Table 5). The conditions for this method were modified slightly with respect to mobile phase ratio, pH and temperature, the results indicating its ability to remain unaffected by small changes in the method's parameters, thus the method is considered robust.</p><p>The standard addition recoveries were carried out by adding a known amount of escitalopram to the powdered tablets at three different levels (5, 10 and 20 μg) with each level in triplicates (n = 3). The recovery percentage was evaluated by the ratio of the amount found to added. The average recovery was calculated and presented in (Table 6).</p><!><p>The present work makes use of micelle enhanced intrinsic fluorescence of escitalopram for its determination in drug substance, commercial tablets and spiked human plasma. It was found to be selective and tolerate high concentrations of other co-administrated and common drugs. The TLC method developed was effective for enantioseparation and determination of enantiomers of citalopram. A comparative study using different chiral selectors was described with the methods being completely validated, showing satisfactory data for all validation parameters tested. Both methods offer simplicity, rapid response and economy.</p>

PubMed Open Access