Machine learning predictions of MHC-II specificities reveal alternative binding mode of class II epitopes.

CD4+ T cells orchestrate the adaptive immune response against pathogens and cancer by recognizing epitopes presented on class II major histocompatibility complex (MHC-II) molecules. The high polymorphism of MHC-II genes represents an important hurdle toward accurate prediction and identification of CD4+ T cell epitopes. Here we collected and curated a dataset of 627,013 unique MHC-II ligands identified by mass spectrometry. This enabled us to precisely determine the binding motifs of 88 MHC-II alleles across humans, mice, cattle, and chickens. Analysis of these binding specificities combined with X-ray crystallography refined our understanding of the molecular determinants of MHC-II motifs and revealed a widespread reverse-binding mode in HLA-DP ligands. We then developed a machine-learning framework to accurately predict binding specificities and ligands of any MHC-II allele. This tool improves and expands predictions of CD4+ T cell epitopes and enables us to discover viral and bacterial epitopes following the aforementioned reverse-binding mode.

SwissDock 2024: major enhancements for small-molecule docking with Attracting Cavities and AutoDock Vina.

Drug discovery aims to identify potential therapeutic compounds capable of modulating the activity of specific biological targets. Molecular docking can efficiently support this process by predicting binding interactions between small molecules and macromolecular targets and potentially accelerating screening campaigns. SwissDock is a computational tool released in 2011 as part of the SwissDrugDesign project, providing a free web-based service for small-molecule docking after automatized preparation of ligands and targets. Here, we present the latest version of SwissDock, in which EADock DSS has been replaced by two state-of-the-art docking programs, i.e. Attracting Cavities and AutoDock Vina. AutoDock Vina provides faster docking predictions, while Attracting Cavities offers more accurate results. Ligands can be imported in various ways, including as files, SMILES notation or molecular sketches. Targets can be imported as PDB files or identified by their PDB ID. In addition, advanced search options are available both for ligands and targets, giving users automatized access to widely-used databases. The web interface has been completely redesigned for interactive submission and analysis of docking results. Moreover, we developed a user-friendly command-line access which, in addition to all options of the web site, also enables covalent ligand docking with Attracting Cavities. The new version of SwissDock is freely available at https://www.swissdock.ch/.

SwissParam 2023: A Modern Web-Based Tool for Efficient Small Molecule Parametrization.

Most steps of drug discovery are now routinely supported and accelerated by computer-aided drug design tools. Among them, structure-based approaches use the three-dimensional structure of the targeted biomacromolecule as a major source of information. When it comes to calculating the interactions of small molecules with proteins using the equations of molecular mechanics, topologies, atom typing, and force field parameters are required. However, generating parameters for small molecules remains challenging due to the large number of existing chemical groups. The SwissParam web tool was first released in 2011 with the aim of generating parameters and topologies for small molecules based on the Merck molecular force field (MMFF) while being compatible with the CHARMM22/27 force field. Here, we present an updated version of SwissParam, providing various new features, including the possibility to setup covalent ligands. Molecules can now be imported from different file formats or via a molecular sketcher. The MMFF-based approach has been updated to provide parameters and topologies compatible with the CHARMM36 force field. An option was added to generate small molecule parametrizations following the CHARMM General Force Field via the multipurpose atom-typer for CHARMM (MATCH) approach. Additionally, SwissParam now generates information on probable alternative tautomers and protonation states of the query molecule so that the user can consider all microspecies relevant to its compound. The new version of SwissParam is freely available at www.swissparam.ch and can also be accessed through a newly implemented command-line interface.

The SwissSimilarity 2021 Web Tool: Novel Chemical Libraries and Additional Methods for an Enhanced Ligand-Based Virtual Screening Experience.

Hit finding, scaffold hopping, and structure-activity relationship studies are important tasks in rational drug discovery. Implementation of these tasks strongly depends on the availability of compounds similar to a known bioactive molecule. SwissSimilarity is a web tool for low-to-high-throughput virtual screening of multiple chemical libraries to find molecules similar to a compound of interest. According to the similarity principle, the output list of molecules generated by SwissSimilarity is expected to be enriched in compounds that are likely to share common protein targets with the query molecule and that can, therefore, be acquired and tested experimentally in priority. Compound libraries available for screening using SwissSimilarity include approved drugs, clinical candidates, known bioactive molecules, commercially available and synthetically accessible compounds. The first version of SwissSimilarity launched in 2015 made use of various 2D and 3D molecular descriptors, including path-based FP2 fingerprints and ElectroShape vectors. However, during the last few years, new fingerprinting methods for molecular description have been developed or have become popular. Here we would like to announce the launch of the new version of the SwissSimilarity web tool, which features additional 2D and 3D methods for estimation of molecular similarity: extended-connectivity, MinHash, 2D pharmacophore, extended reduced graph, and extended 3D fingerprints. Moreover, it is now possible to screen for molecular structures having the same scaffold as the query compound. Additionally, all compound libraries available for screening in SwissSimilarity have been updated, and several new ones have been added to the list. Finally, the interface of the website has been comprehensively rebuilt to provide a better user experience. The new version of SwissSimilarity is freely available starting from December 2021.

Genetic Algorithm Based Design and Experimental Characterization of a Highly Thermostable Metalloprotein.

The development of thermostable and solvent-tolerant metalloproteins is a long-sought goal for many applications in synthetic biology and biotechnology. In this work, we were able to engineer a highly thermostable and organic solvent-stable metallo variant of the B1 domain of protein G (GB1) with a tetrahedral zinc binding site reminiscent of the one of thermolysin. Promising candidates were designed computationally by applying a protocol based on classical and first-principles molecular dynamics simulations in combination with genetic algorithm optimization. The most promising of the computationally predicted mutants was expressed and structurally characterized and yielded a highly thermostable protein. The experimental results thus confirm the predictive power of the applied computational protein engineering approach for the de novo design of highly stable metalloproteins.

The Structure of the Protonated Serine Octamer.

The amino acid serine has long been known to form a protonated "magic-number" cluster containing eight monomer units that shows an unusually high abundance in mass spectra and has a remarkable homochiral preference. Despite many experimental and theoretical studies, there is no consensus on a Ser8H+ structure that is in agreement with all experimental observations. Here, we present the structure of Ser8H+ determined by a combination of infrared spectroscopy and ab initio molecular dynamics simulations. The three-dimensional structure that we determine is ∼25 kcal mol-1 more stable than the previous most stable published structure and explains both the homochiral preference and the experimentally observed facile replacement of two serine units.

Genetic-Algorithm-Based Optimization of a Peptidic Scaffold for Sequestration and Hydration of CO2.

Biomimicry is a strategy that makes practical use of evolution to find efficient and sustainable ways to produce chemical compounds or engineer products. Exploring the natural machinery of enzymes for the production of desired compounds is a highly profitable investment, but the design of efficient biomimetic systems remains a considerable challenge. An ideal biomimetic system self-assembles in solution, binds a desired range of substrates and catalyzes reactions with turnover rates similar to the native system. To this end, tailoring catalytic functionality in engineered peptides generally requires site-directed mutagenesis or the insertion of additional amino acids, which entails an intensive search across chemical and sequence space. Here we discuss a novel strategy for the computational design of biomimetic compounds and processes that consists of a) characterization of the wild-type and biomimetic systems; b) identification of key descriptors for optimization; c) an efficient search through sequence and chemical space to tailor the catalytic capabilities of the biomimetic system. Through this proof-of-principle study, we are able to decisively understand and identify whether a given scaffold is useful, appropriate and tailorable for a given, desired task.

Lessons from nature: computational design of biomimetic compounds and processes.

Through millions of years of evolution, Nature has accomplished the development of highly efficient and sustainable processes and the idea to understand and copy natural strategies is therefore very appealing. However, in spite of intense experimental and computational research, it has turned out to be a difficult task to design efficient biomimetic systems. Here we discuss a novel strategy for the computational design of biomimetic compounds and processes that consists of i) target selection; ii) atomistic and electronic characterization of the wild type system and the biomimetic compounds; iii) identification of key descriptors through feature selection iv) choice of biomimetic template and v) efficient search of chemical and sequence space for optimization of the biomimetic system. As a proof-of-principles study, this general approach is illustrated for the computational design of a 'green' catalyst mimicking the action of the zinc metalloenzyme Human Carbonic Anhydrase (HCA). HCA is a natural model for CO2 fixation since the enzyme is able to convert CO2 into bicarbonate. Very recently, a weakly active HCA mimic based on a trihelical peptide bundle was synthetized. We have used quantum mechanical/molecular mechanical (QM/MM) Car-Parrinello simulations to study the mechanisms of action of HCA and its peptidic mimic and employed the obtained information to guide the design of improved biomimetic analogues. Applying a genetic algorithm based optimization procedure, we were able to re-engineer and optimize the biomimetic system towards its natural counter part. In a second example, we discuss a similar strategy for the design of biomimetic sensitizers for use in dye-sensitized solar cells.

Predicting Antigen-Specificities of Orphan T Cell Receptors from Cancer Patients with TCRpcDist.

Approaches to analyze and cluster T-cell receptor (TCR) repertoires to reflect antigen specificity are critical for the diagnosis and prognosis of immune-related diseases and the development of personalized therapies. Sequence-based approaches showed success but remain restrictive, especially when the amount of experimental data used for the training is scarce. Structure-based approaches which represent powerful alternatives, notably to optimize TCRs affinity toward specific epitopes, show limitations for large-scale predictions. To handle these challenges, TCRpcDist is presented, a 3D-based approach that calculates similarities between TCRs using a metric related to the physico-chemical properties of the loop residues predicted to interact with the epitope. By exploiting private and public datasets and comparing TCRpcDist with competing approaches, it is demonstrated that TCRpcDist can accurately identify groups of TCRs that are likely to bind the same epitopes. Importantly, the ability of TCRpcDist is experimentally validated to determine antigen specificities (neoantigens and tumor-associated antigens) of orphan tumor-infiltrating lymphocytes (TILs) in cancer patients. TCRpcDist is thus a promising approach to support TCR repertoire analysis and TCR deorphanization for individualized treatments including cancer immunotherapies.

Identification of clinically relevant T cell receptors for personalized T cell therapy using combinatorial algorithms.

A central challenge in developing personalized cancer cell immunotherapy is the identification of tumor-reactive T cell receptors (TCRs). By exploiting the distinct transcriptomic profile of tumor-reactive T cells relative to bystander cells, we build and benchmark TRTpred, an antigen-agnostic in silico predictor of tumor-reactive TCRs. We integrate TRTpred with an avidity predictor to derive a combinatorial algorithm of clinically relevant TCRs for personalized T cell therapy and benchmark it in patient-derived xenografts.

Neoantigen-specific CD8 T cells with high structural avidity preferentially reside in and eliminate tumors.

The success of cancer immunotherapy depends in part on the strength of antigen recognition by T cells. Here, we characterize the T cell receptor (TCR) functional (antigen sensitivity) and structural (monomeric pMHC-TCR off-rates) avidities of 371 CD8 T cell clones specific for neoantigens, tumor-associated antigens (TAAs) or viral antigens isolated from tumors or blood of patients and healthy donors. T cells from tumors exhibit stronger functional and structural avidity than their blood counterparts. Relative to TAA, neoantigen-specific T cells are of higher structural avidity and, consistently, are preferentially detected in tumors. Effective tumor infiltration in mice models is associated with high structural avidity and CXCR3 expression. Based on TCR biophysicochemical properties, we derive and apply an in silico model predicting TCR structural avidity and validate the enrichment in high avidity T cells in patients' tumors. These observations indicate a direct relationship between neoantigen recognition, T cell functionality and tumor infiltration. These results delineate a rational approach to identify potent T cells for personalized cancer immunotherapy.

Structural Prediction of Peptide-MHC Binding Modes.

The immune system is constantly protecting its host from the invasion of pathogens and the development of cancer cells. The specific CD8+ T-cell immune response against virus-infected cells and tumor cells is based on the T-cell receptor recognition of antigenic peptides bound to class I major histocompatibility complexes (MHC) at the surface of antigen presenting cells. Consequently, the peptide binding specificities of the highly polymorphic MHC have important implications for the design of vaccines, for the treatment of autoimmune diseases, and for personalized cancer immunotherapy. Evidence-based machine-learning approaches have been successfully used for the prediction of peptide binders and are currently being developed for the prediction of peptide immunogenicity. However, understanding and modeling the structural details of peptide/MHC binding is crucial for a better understanding of the molecular mechanisms triggering the immunological processes, estimating peptide/MHC affinity using universal physics-based approaches, and driving the design of novel peptide ligands. Unfortunately, due to the large diversity of MHC allotypes and possible peptides, the growing number of 3D structures of peptide/MHC (pMHC) complexes in the Protein Data Bank only covers a small fraction of the possibilities. Consequently, there is a growing need for rapid and efficient approaches to predict 3D structures of pMHC complexes. Here, we review the key characteristics of the 3D structure of pMHC complexes before listing databases and other sources of information on pMHC structures and MHC specificities. Finally, we discuss some of the most prominent pMHC docking software.

Sensitive identification of neoantigens and cognate TCRs in human solid tumors.

The identification of patient-specific tumor antigens is complicated by the low frequency of T cells specific for each tumor antigen. Here we describe NeoScreen, a method that enables the sensitive identification of rare tumor (neo)antigens and of cognate T cell receptors (TCRs) expressed by tumor-infiltrating lymphocytes. T cells transduced with tumor antigen-specific TCRs identified by NeoScreen mediate regression of established tumors in patient-derived xenograft mice.

Deciphering the Mechanisms of Improved Immunogenicity of Hypochlorous Acid-Treated Antigens in Anti-Cancer Dendritic Cell-Based Vaccines.

Hypochlorous acid (HOCl)-treated whole tumor cell lysates (Ox-L) have been shown to be more immunogenic when used as an antigen source for therapeutic dendritic cell (DC)-based vaccines, improving downstream immune responses both in vitro and in vivo. However, the mechanisms behind the improved immunogenicity are still elusive. To address this question, we conducted a proteomic and immunopeptidomics analyses to map modifications and alterations introduced by HOCl treatment using a human melanoma cell line as a model system. First, we show that one-hour HOCl incubation readily induces extensive protein oxidation, mitochondrial biogenesis, and increased expression of chaperones and antioxidant proteins, all features indicative of an activation of oxidative stress-response pathways. Characterization of the DC proteome after loading with HOCl treated tumor lysate (Ox-L) showed no significant difference compared to loading with untreated whole tumor lysate (FT-L). On the other hand, detailed immunopeptidomic analyses on monocyte-derived DCs (mo-DCs) revealed a great increase in human leukocyte antigen class II (HLA-II) presentation in mo-DCs loaded with Ox-L compared to the FT-L control. Further, 2026 HLA-II ligands uniquely presented on Ox-L-loaded mo-DCs were identified. In comparison, identities and intensities of HLA class I (HLA-I) ligands were overall comparable. We found that HLA-II ligands uniquely presented by DCs loaded with Ox-L were more solvent exposed in the structures of their source proteins, contrary to what has been hypothesized so far. Analyses from a phase I clinical trial showed that vaccinating patients using autologous Ox-L as an antigen source efficiently induces polyfunctional vaccine-specific CD4+ T cell responses. Hence, these results suggest that the increased immunogenicity of Ox-L is, at least in part, due to qualitative and quantitative changes in the HLA-II ligandome, potentially leading to an increased HLA-II dependent stimulation of the T cell compartment (i.e., CD4+ T cell responses). These results further contribute to the development of more effective and immunogenic DC-based vaccines and to the molecular understanding of the mechanism behind HOCl adjuvant properties.

Analysis of Secondary Structure Biases in Naturally Presented HLA-I Ligands.

Recent clinical developments in antitumor immunotherapy involving T-cell related therapeutics have led to a renewed interest for human leukocyte antigen class I (HLA-I) binding peptides, given their potential use as peptide vaccines. Databases of HLA-I binding peptides hold therefore information on therapeutic targets essential for understanding immunity. In this work, we use in depth and accurate HLA-I peptidomics datasets determined by mass-spectrometry (MS) and analyze properties of the HLA-I binding peptides with structure-based computational approaches. HLA-I binding peptides are studied grouping all alleles together or in allotype-specific contexts. We capitalize on the increasing number of structurally determined proteins to (1) map the 3D structure of HLA-I binding peptides into the source proteins for analyzing their secondary structure and solvent accessibility in the protein context, and (2) search for potential differences between these properties in HLA-I binding peptides and in a reference dataset of HLA-I motif-like peptides. This is performed by an in-house developed heuristic search that considers peptides across all the human proteome and converges to a collection of peptides that exhibit exactly the same motif as the HLA-I peptides. Our results, based on 9-mers matched to protein 3D structures, clearly show enriched sampling for HLA-I presentation of helical fragments in the source proteins. This enrichment is significant, as compared to 9-mer HLA-I motif-like peptides, and is not entirely explained by the helical propensity of the preferred residues in the HLA-I motifs. We give possible hypothesis for the secondary structure biases observed in HLA-I peptides. This contribution is of potential interest for researchers working in the field of antigen presentation and proteolysis. This knowledge refines the understanding of the rules governing antigen presentation and could be added to the parameters of the current peptide-MHC class I binding predictors to increase their antigen predictive ability.

Cryogenic Spectroscopy and Quantum Molecular Dynamics Determine the Structure of Cyclic Intermediates Involved in Peptide Sequence Scrambling.

Collision-induced dissociation (CID) is a key technique used in mass spectrometry-based peptide sequencing. Collisionally activated peptides undergo statistical dissociation, forming a series of backbone fragment ions that reflect their amino acid (AA) sequence. Some of these fragments may experience a "head-to-tail" cyclization, which after proton migration, can lead to the cyclic structure opening in a different place than the initially formed bond. This process leads to AA sequence scrambling that may hinder sequencing of the initial peptide. Here we combine cryogenic ion spectroscopy and ab initio molecular dynamics simulations to isolate and characterize the precise structures of key intermediates in the scrambling process. The most stable peptide fragments show intriguing symmetric cyclic structures in which the proton is situated on a C2 symmetry axis and forms exceptionally short H-bonds (1.20 Å) with two backbone oxygens. Other nonsymmetric cyclic structures also exist, one of which is protonated on the amide nitrogen, where ring opening is likely to occur.

Establishing the catalytic mechanism of human pancreatic α-amylase with QM/MM methods.

In this work, we studied the catalytic mechanism of human pancreatic α-amylase (HPA). Our goal was to determine the catalytic mechanism of HPA with atomic detail using computational methods. We demonstrated that the HPA catalytic mechanism consists of two steps, the first of which (glycosylation step) involves breaking the glycosidic bond to culminate in the formation of a covalent intermediate. The second (deglycosylation step) consists of the addition of a water molecule to release the enzyme/substrate covalent intermediate, completing the hydrolysis of the sugar. The active site was very open to the solvent. Our mechanism basically differs from the previously proposed mechanism by having two water molecules instead of only one near the active site that participate in the mechanism. We also demonstrate the relevant role of the three catalytic amino acids, two aspartate residues and a glutamate (D197, E233, and D300), during catalysis. It was also shown that the rate limiting step was glycosylation, and its activation energy was in agreement with experimental values obtained for HPA. The experimental activation energy was 14.4 kcal mol(-1), and the activation energy obtained computationally was 15.1 kcal mol(-1).

Infrared Spectroscopy of Mobility-Selected H+-Gly-Pro-Gly-Gly (GPGG).

We report the first results from a new instrument capable of acquiring infrared spectra of mobility-selected ions. This demonstration involves using ion mobility to first separate the protonated peptide Gly-Pro-Gly-Gly (GPGG) into two conformational families with collisional cross-sections of 93.8 and 96.8 Å(2). After separation, each family is independently analyzed by acquiring the infrared predissociation spectrum of the H(2)-tagged molecules. The ion mobility and spectroscopic data combined with density functional theory (DFT) based molecular dynamics simulations confirm the presence of one major conformer per family, which arises from cis/trans isomerization about the proline residue. We induce isomerization between the two conformers by using collisional activation in the drift tube and monitor the evolution of the ion distribution with ion mobility and infrared spectroscopy. While the cis-proline species is the preferred gas-phase structure, its relative population is smaller than that of the trans-proline species in the initial ion mobility drift distribution. This suggests that a portion of the trans-proline ion population is kinetically trapped as a higher energy conformer and may retain structural elements from solution.

Computational Alanine Scanning Mutagenesis: MM-PBSA vs TI.

Understanding protein-protein association and being able to determine the crucial residues responsible for their association (hot-spots) is a key issue with huge practical applications such as rational drug design and protein engineering. A variety of computational methods exist to detect hot-spots residues, but the development of a fast and accurate quantitative alanine scanning mutagenesis (ASM) continues to be crucial. Using four protein-protein complexes, we have compared a variation of the standard computational ASM protocol developed at our group, based on the Molecular Mechanics/Poisson-Boltzmann Surface Area (MM-PBSA) approach, against Thermodynamic Integration (TI), a well-known and accurate but computationally expensive method. To compare the efficiency and the accuracy of the two methods, we have calculated the protein-protein binding free energy differences upon alanine mutation of interfacial residues (ΔΔGbind). In relation to the experimental ΔΔGbind values, the average error obtained with TI was 1.53 kcal/mol, while the ASM protocol resulted in an average error of 1.18 kcal/mol. The results demonstrate that the much faster ASM protocol gives results at the same level of accuracy as the TI method but at a fraction of the computational time required to run TI. This ASM protocol is therefore a strong and efficient alternative to the systematic evaluation of protein-protein interfaces, involving hundreds of amino acid residues in search of hot-spots.