Despite substantial progress in single-cell screening techniques, antibody (Ab) repertoires still remain enigmatic. Here we show that Ab sequences can be linked to their functionality by using big data obtained from high-throughput sequencing. Using the expansive SARS-CoV-2 pandemic data, we develop an AI-based method to reveal the neutralization potential of Ab repertoires. We employ machine learning to process public 3D structural data of Ab-RBD complexes and create a comprehensive tool, RBD-AIM (https://rbdaim.2a2i.org/), for high-throughput prediction of structural Ab epitopes based on Ab sequence. Using RBD-AIM, we analyze the local big data sources to evaluate the functional biodiversity of native B cell repertoires raised after vaccination and reconstructed in a yeast display system using single-cell microfluidics. This pipeline allows for rapid isolation of neutralizing Abs that promote the survival of transgenic hACE2+ mice in lethal models of SARS-CoV-2 infection. We believe that the AI-guided sequence-functionality link can be successfully employed for further high-throughput discovery of therapeutic Abs and functional analysis of Ab repertoires.

Study of mechanisms by which antibodies recognize different viral strains is necessary for the development of new drugs and vaccines to treat COVID-19 and other infections. Here, we report 2.5 Å cryo-EM structure of the SARS-CoV-2 Delta trimeric S-protein in complex with Fab of the recombinant analog of REGN10987 neutralizing antibody. S-protein adopts "two RBD-down and one RBD-up" conformation. Fab interacts with RBDs in both conformations, blocking the recognition of angiotensin converting enzyme-2. Three-dimensional variability analysis reveals high mobility of the RBD/Fab regions. Interaction of REGN10987 with Wuhan, Delta, Omicron BA.1, and mutated variants of RBDs is analyzed by microscale thermophoresis, molecular dynamics simulations, and ΔG calculations with umbrella sampling and one-dimensional potential of mean force. Variability in molecular dynamics trajectories results in a large scatter of calculated ΔG values, but Boltzmann weighting provides an acceptable correlation with experiment. REGN10987 evasion of the Omicron variant is found to be due to the additive effect of the N440K and G446S mutations located at the RBD/Fab binding interface with a small effect of Q498R mutation. Our study explains the influence of known-to-date SARS-CoV-2 RBD mutations on REGN10987 recognition and highlights the importance of dynamics data beyond the static structure of the RBD/Fab complex.

Nicotinic acetylcholine receptor of α7 type (α7-nAChR) presented in the nervous and immune systems and epithelium is a promising therapeutic target for cognitive disfunctions and cancer treatment. Weak toxin from Naja kaouthia venom (WTX) is a non-conventional three-finger neurotoxin, targeting α7-nAChR with weak affinity. There are no data on interaction mode of non-conventional neurotoxins with nAChRs. Using α-bungarotoxin (classical three-finger neurotoxin with high affinity to α7-nAChR), we showed applicability of cryo-EM to study complexes of α7-nAChR extracellular ligand-binding domain (α7-ECD) with toxins. Using cryo-EM structure of the α7-ECD/WTX complex, together with NMR data on membrane active site in the WTX molecule and mutagenesis data, we reconstruct the structure of α7-nAChR/WTX complex in the membrane environment. WTX interacts at the entrance to the orthosteric site located at the receptor intersubunit interface and simultaneously forms the contacts with the membrane surface. WTX interaction mode with α7-nAChR significantly differs from α-bungarotoxin’s one, which does not contact the membrane. Our study reveals the important role of the membrane for interaction of non-conventional neurotoxins with the nicotinic receptors.

“Fluorescence-Activating and absorption-Shifting Tag” (FAST) is a well-studied fluorogen-activating protein with high brightness and low size, able to activate a wide range of fluorogens. This makes FAST a promising target for both protein and fluorogen optimization. Here, we describe the structure-based rational design of the enhanced FAST mutants, optimized for the N871b fluorogen. Using the spatial structure of the FAST/ N871b complex, NMR relaxation analysis, and computer simulations, we identify the mobile regions in the complex and suggest mutations that could stabilize both the protein and the ligand. Two of our mutants appear brighter than the wild-type FAST, and these mutants provide up to 35% enhancement for several other fluorogens of similar structure, both in vitro and in vivo. Analysis of the mutants by NMR reveals that brighter mutants demonstrate the highest stability and lowest length of intermolecular H-bonds. Computer simulations provide the structural basis for such stabilization.

Understanding fusion mechanisms employed by SARS-CoV-2 spike protein entails realistic transmembrane domain (TMD) models, while no reliable approaches towards predicting the 3D structure of transmembrane (TM) trimers exist. Here, we propose a comprehensive computational framework to model the spike TMD only based on its primary structure. We performed amino acid sequence pattern matching and compared the molecular hydrophobicity potential (MHP) distribution on the helix surface against TM homotrimers with known 3D structures and selected an appropriate template for homology modeling. We then iteratively built a model of spike TMD, adjusting "dynamic MHP portraits" and residue variability motifs. The stability of this model, with and without palmitoyl modifications downstream of the TMD, and several alternative configurations (including a recent NMR structure), was tested in all-atom molecular dynamics simulations in a POPC bilayer mimicking the viral envelope. Our model demonstrated unique stability under the conditions applied and conforms to known basic principles of TM helix packing. The original computational framework looks promising and could potentially be employed in the construction of 3D models of TM trimers for a wide range of membrane proteins.

The dehydroshikimate dehydratase (DSD) from Corynebacterium glutamicum encoded by the qsuB gene is related to the previously described QuiC1 protein (39.9% identity) from Pseudomonas putida . QuiC1 and QsuB are both two-domain bacterial DSDs. The N-terminal domain provides dehydratase activity, while the C-terminal domain has sequence identity with 4-hydroxyphenylpyruvate dioxygenase. Here, the QsuB protein and its DSD domain (N-QsuB) were expressed in the T7 system, purified and characterized. QsuB was present mainly in octameric form (60%), while N-QsuB had a predominantly monomeric structure (80%) in solution. Both proteins possessed DSD activity with one of the following cofactors (listed in order of decreasing activity): Co2+, Mg2+, Mn2+ or Ca2+. The K m and k cat values for QsuB were two and three times higher, respectively (K m ~ 1 mM, k cat ~ 61 s−1) than those for N-QsuB. Notably, 3,4-DHBA inhibited both enzymes via an uncompetitive mechanism. QsuB and N-QsuB were tested for 3,4-DHBA production from glucose in E. coli . MG1655Δ aroE P lac – qsuB produced at least two times more 3,4-DHBA than MG1655Δ aroE P lac – n - qsuB in the presence of isopropyl β-D-1-thiogalactopyranoside.

Antibiotics (AB) resistance is a major threat to global health, thus the development of novel AB classes is urgently needed. Lantibiotics (i.e. nisin) are natural compounds that effectively control bacterial populations, yet their clinical potential is very limited. Nisin targets membrane-embedded cell wall precursor - lipid II - via capturing its pyrophosphate group (PPi), which is unlikely to evolve, and thus represents a promising pharmaceutical target. Understanding of exact molecular mechanism of initial stages of membrane-bound lipid II recognition by water-soluble nisin is indispensable. Here, using molecular simulations, we demonstrate that the structure of lipid II is determined to a large extent by the surrounding water-lipid milieu. In contrast to the bulk solvent, in the bilayer only two conformational states remain capable of nisin binding. In these states PPi manifests a unique arrangement of hydrogen bond acceptors on the bilayer surface. Such a "pyrophosphate pharmacophore" cannot be formed by phospholipids, which explains high selectivity of nisin/lipid II recognition. Similarly, the "recognition module" of nisin, being rather flexible in water, adopts the only stable conformation in the presence of PPi analogue (which mimics the lipid II molecule). We establish the "energy of the pyrophosphate pharmacophore" approach, which effectively distinguishes nisin conformations that can form a complex with PPi. Finally, we propose a molecular model of nisin recognition module/lipid II complex in the bacterial membrane. These results will be employed for further study of lipid II targeting by antimicrobial (poly)cyclic peptides and for design of novel AB prototypes.

How is a water-soluble globular protein able to spontaneously cross a cellular membrane? It is commonly accepted that it undergoes significant structural rearrangements on the lipid-water interface, thus acquiring membrane binding and penetration ability. In this study molecular dynamics (MD) simulations have been used to explore large-scale conformational changes of the globular viscumin A chain in a complex environment – comprising urea and chloroform/methanol (CHCl3/MeOH) mixture. Being well-packed in aqueous solution, viscumin A undergoes global structural rearrangements in both organic media. In urea, the protein is “swelling” and gradually loses its long-distance contacts, thus resembling the “molten globule” state. In CHCl3/MeOH, viscumin A is in effect turned “inside out”. This is accompanied with strengthening of the secondary structure and surface exposure of hydrophobic epitopes originally buried inside the globule. Resulting solvent-adapted models were further subjected to Monte Carlo simulations with an implicit hydrophobic slab membrane. In contrast to only a few point surface contacts in water and two short regions with weak protein-lipid interactions in urea, MD-derived structures in CHCl3/MeOH reveal multiple determinants of membrane interaction. Consequently it is now possible to propose a specific pathway for the structural adaptation of viscumin A with respect to the cell membrane – a probable first step of its translocation into cytoplasmic targets.

Molecular surfaces are the key players in processes of bimolecular recognition, interaction, and signaling. Nowadays, state of the art methods exist for visualizing molecule surface and surface distributed properties in three-dimensional space. However, such a visual information could only be analyzed by human eye and therefore prompt to be biased and onerous in case of biomacromolecules with a complex structural organization or for large sets of objects. This is especially true for such mesoscopic systems like cell membranes. Therefore, alternative (mostly computational) techniques operating with such "molecular portraits" are indispensable. Here, we present a number of original in silico methods to map and process multivariate and dynamic surface properties of biomolecules and their assemblies - proteins, membranes, carbohydrates and their complexes. These techniques were shown to improve protein-ligand [1] and protein-protein [2] docking results. A new protein surface topography (PST) method [3] is described. It permits pictorial visualization of the whole protein surface in terms of 2D "earth maps". PST can be used to delineate conformational changes, perform group analysis of molecules. It is a useful tool to add to docking experiments. The proposed original tools have been shown to be very efficient in analysis of atomistic details of the recognition processes in cell membranes - they allow easy monitoring of dynamic amphiphilic properties of lipid membranes, proteins, peptides, polysaccharides, etc. Implementation of the "molecular portrait" technologies is capable of rational design of new physiologically important compounds with predefined activities [4]. References. 1. Pyrkov T.V. et al. (2009) Bioinformatics. 25:1201. 2. Polyansky A.A. et al. (2014) Bioinformatics 30:889. 3. Koromyslova A.D. et al. (2014) J. Chem. Inf. Mod. 54:1189. 4. Kasheverov I.E. et al. (2016) Scientific Reports, 6:36848. Acknowledgements Russian Foundation for Basic Research (18-54-15007), Russian Science Foundation (18-14-00375).

Heat-activated transient receptor potential channel TRPV1 is one of the most studied eukaryotic proteins involved in temperature sensation. Upon heating, it exhibits rapid reversible pore gating, which depolarizes neurons and generates action potentials. Underlying molecular details of such effects in the pore region of TRPV1 is of a crucial importance to control temperature responses of the organism. Despite the spatial structure of the channel in both open (O) and closed (C) states is known, microscopic nature of channel gating and mechanism of thermal sensitivity are still poorly understood. In this work, we used unrestrained atomistic molecular dynamics simulations of TRPV1 (without N- and C-terminal cytoplasmic domains) embedded into explicit lipid bilayer in its O- and C-states [1]. We found that the pore domain with its neighboring loops undergoes large temperature-dependent conformational transitions in an asymmetric way, when fragments only one monomer move with large amplitude, freeing the pore upon heating. Such an asymmetrical gating looks rather biologically relevant because it is faster and more reliable than traditionally proposed “iris-like” symmetric scheme of channel opening. Analysis of structural, dynamic, and hydrophobic organization of the pore domain revealed entropy growth upon TRPV1 gating, which is in line with current concepts of thermal sensitivity. Acknowledgements: Authors thank Russian Science Foundation (14-14-00871), Russian Foundation for Basic Research (16-04-00578), RAS MCB Program. Access to computational facilities of the Supercomputer Center “Polytechnical” at the St. Petersburg Polytecnic University is gratefully acknowledged. Reference: [1] A.O. Chugunov et al., Scientific Reports 6 (2016) 33112.

Heat-activated transient receptor potential channel TRPV1 is one of the most studied eukaryotic proteins involved in temperature sensation. Upon heating, it exhibits rapid reversible pore gating, which depolarizes neurons and generates action potentials. Underlying molecular details of such effects in the pore region of TRPV1 is of a crucial importance to control temperature responses of the organism. Despite the spatial structure of the channel in both open (O) and closed (C) states is known, microscopic nature of channel gating and mechanism of thermal sensitivity are still poorly understood. In this work, we used unrestrained atomistic molecular dynamics simulations of TRPV1 (without N- and C-terminal cytoplasmic domains) embedded into explicit lipid bilayer in its O- and C-states. We found that the pore domain with its neighboring loops undergoes large temperature-dependent conformational transitions in an asymmetric way, when fragments of only one monomer move with large amplitude, freeing the pore upon heating. Such an asymmetrical gating looks rather biologically relevant because it is faster and more reliable than traditionally proposed "iris-like" symmetric scheme of channel opening. Analysis of structural, dynamic, and hydrophobic organization of the pore domain revealed entropy growth upon TRPV1 gating, which is in line with current concepts of thermal sensitivity.

SUMMARYHere we present PREDDIMER, a web tool for prediction of dimer structure of transmembrane (TM) helices. PREDDIMER allows (i) reconstruction of a number of dimer structures for given sequence(s) of TM protein fragments, (ii) ranking and filtering of predicted structures according to respective values of a scoring function, (iii) visualization of predicted 3D dimer structures and (iv) visualization of surface hydrophobicity of TM helices and their contacting (interface) regions represented as 2D maps.RESULTSWe implemented online the original PREDDIMER algorithm and benchmarked the server on 11 TM sequences, whose 3D dimer conformations were obtained previously by nuclear magnetic resonance spectroscopy. In the most of tested cases backbone root-mean-square deviations of closest predicted conformations from the experimental reference are below 3 Å. A randomization test displays good anticorrelation (-0.82) between values of the scoring function and statistical significance of the prediction 'by chance'. Going beyond a single dimer conformation, our web tool predicts an ensemble of possible conformations, which may be useful for explanation of a functioning of bitopic membrane proteins, e.g. receptor tyrosine kinases.AVAILABILITY AND IMPLEMENTATIONPREDDIMER can be accessed for free on the web at http://model.nmr.ru/preddimer/CONTACTnewant@gmail.comSUPPLEMENTARY INFORMATIONSupplementary data are available at Bioinformatics online.

Delineation and analysis of lateral clustering of lipids in model bilayers is an important step toward understanding of the physical processes underlying formation of lipid domains and rafts in cell membranes. Computer modeling methods represent a powerful tool to address the problem since they can detect clusters of only few lipid molecules this issue still resists easy characterization with modern experimental techniques. In this work, we propose a computational method to detect and analyze parts of membrane with different packing densities and hydrogen bonding patterns. A series of one- and two-component fluid systems containing lipids with the same polar heads and different acyl chains, dioleoylphosphatidylcholine (18:1) and dipalmitoylphosphatidylcholine (16:0), or with same acyl chains and different polar heads, dioleoylphosphatidylserine (18:1) and dioleoylphosphatidylcholine (18:1), were studied via molecular dynamics simulations. Four criteria of clustering were considered. It was shown that the waterlipid interface of biomembranes represents a highly dynamic and mosaic picture, whose parameters depend on the bilayer composition. Some systems (e.g. with 2030% of the anionic lipid) demonstrate unusual clustering properties and demand further investigation at molecular level. Lateral microheterogeneities in fluid lipid bilayers seem to be among the most important factors determining the nature of the membranewater interface in a cell.

Bacterial cell wall is targeted by many antibiotics. Among them are lantibiotics, which realize their function via interaction with plasma membrane lipid-II molecule — a chemically conserved part of the cell wall synthesis pathway. To investigate structural and dynamic properties of this molecule, we have performed a series of nearly microsecond-long molecular dynamics simulations of lipid-II and some of its analogs in zwitterionic single component and charged mixed simulated phospholipid bilayers (the reference and the mimic of the bacterial plasma membrane, respectively). Extensive analysis revealed that lipid-II forms a unique “amphiphilic pattern” exclusively on the surface of the simulated bacterial membrane (and not in the reference one). We hypothesize that many lantibiotics exploit the conserved features of lipid-II along with characteristic modulation of the bacterial membrane as the “landing site”. This putative recognition mechanism opens new opportunities for studies on lantibiotics action and design of novel armament against resistant bacterial strains.

Investigation of lateral heterogeneities (clusters) in cell membranes is an important step toward understanding the physical processes that lead to the formation of lipid domains and rafts. Computer modeling methods represent a powerful tool to solve the problem, since they can detect clusters containing only a few lipid molecules—the situation that still resists characterization with modern experimental techniques. Parameters of clustering depend on lipid composition of a membrane. In this work, we propose a computational method to detect and analyze parts of membrane with different packing densities. Series of one- and two-component fluid systems containing lipids with the same polar heads and different acyl chains, dioleoylphosphatidylcholine (18 : 1) and dipalmitoylphosphatidylcholine (16 : 0), were chosen as the objects under study. The developed algorithm is based on molecular dynamics simulation of hydrated lipid bilayers in all-atom mode. The method is universal and could be applied to any other membrane system with arbitrary lipid composition. Here, we demonstrated that the studied lipid bilayers reveal small lateral dynamic clusters composed of just several (most often, three) lipid molecules. This seems to be one of the most important reasons determining the “mosaic” nature of the membrane-water interface.

Computational techniques designed to predict the spatial structure of ligand—receptor complexes (molecular docking) are widely used in investigations of molecular details of protein functioning and in drug design. Here, a brief review of docking methods is given and recent advances in improvement of their accuracy and efficiency are discussed. Two acute problems of standard docking algorithms are considered: proper ranking of putative docking solutions and simulation of receptor flexibility. The recent trends to overcome these problems are demonstrated with the results obtained in our Laboratory. Particular attention is paid to protein—ligand hydrophobic and stacking interactions, which are not always adequately represented in scoring criteria of docking applications.