Orbitrap-Based Mass and Charge Analysis of Single Molecules.

Native mass spectrometry is nowadays widely used for determining the mass of intact proteins and their noncovalent biomolecular assemblies. While this technology performs well in the mass determination of monodisperse protein assemblies, more real-life heterogeneous protein complexes can pose a significant challenge. Factors such as co-occurring stoichiometries, subcomplexes, and/or post-translational modifications, may especially hamper mass analysis by obfuscating the charge state inferencing that is fundamental to the technique. Moreover, these mass analyses typically require measurement of several million molecules to generate an analyzable mass spectrum, limiting its sensitivity. In 2012, we introduced an Orbitrap-based mass analyzer with extended mass range (EMR) and demonstrated that it could be used to obtain not only high-resolution mass spectra of large protein macromolecular assemblies, but we also showed that single ions generated from these assemblies provided sufficient image current to induce a measurable charge-related signal. Based on these observations, we and others further optimized the experimental conditions necessary for single ion measurements, which led in 2020 to the introduction of single-molecule Orbitrap-based charge detection mass spectrometry (Orbitrap-based CDMS). The introduction of these single molecule approaches has led to the fruition of various innovative lines of research. For example, tracking the behavior of individual macromolecular ions inside the Orbitrap mass analyzer provides unique, fundamental insights into mechanisms of ion dephasing and demonstrated the (astonishingly high) stability of high mass ions. Such fundamental information will help to further optimize the Orbitrap mass analyzer. As another example, the circumvention of traditional charge state inferencing enables Orbitrap-based CDMS to extract mass information from even extremely heterogeneous proteins and protein assemblies (e.g., glycoprotein assemblies, cargo-containing nanoparticles) via single molecule detection, reaching beyond the capabilities of earlier approaches. We so far demonstrated the power of Orbitrap-based CDMS applied to a variety of fascinating systems, assessing for instance the cargo load of recombinant AAV-based gene delivery vectors, the buildup of immune-complexes involved in complement activation, and quite accurate masses of highly glycosylated proteins, such as the SARS-CoV-2 spike trimer proteins. With such widespread applications, the next objective is to make Orbitrap-based CDMS more mainstream, whereby we still will seek to further advance the boundaries in sensitivity and mass resolving power.

Long Time Scale Ensemble Methods in Molecular Dynamics: Ligand-Protein Interactions and Allostery in SARS-CoV-2 Targets.

We subject a series of five protein-ligand systems which contain important SARS-CoV-2 targets, 3-chymotrypsin-like protease (3CLPro), papain-like protease, and adenosine ribose phosphatase, to long time scale and adaptive sampling molecular dynamics simulations. By performing ensembles of ten or twelve 10 µs simulations for each system, we accurately and reproducibly determine ligand binding sites, both crystallographically resolved and otherwise, thereby discovering binding sites that can be exploited for drug discovery. We also report robust, ensemble-based observation of conformational changes that occur at the main binding site of 3CLPro due to the presence of another ligand at an allosteric binding site explaining the underlying cascade of events responsible for its inhibitory effect. Using our simulations, we have discovered a novel allosteric mechanism of inhibition for a ligand known to bind only at the substrate binding site. Due to the chaotic nature of molecular dynamics trajectories, regardless of their temporal duration individual trajectories do not allow for accurate or reproducible elucidation of macroscopic expectation values. Unprecedentedly at this time scale, we compare the statistical distribution of protein-ligand contact frequencies for these ten/twelve 10 µs trajectories and find that over 90% of trajectories have significantly different contact frequency distributions. Furthermore, using a direct binding free energy calculation protocol, we determine the ligand binding free energies for each of the identified sites using long time scale simulations. The free energies differ by 0.77 to 7.26 kcal/mol across individual trajectories depending on the binding site and the system. We show that, although this is the standard way such quantities are currently reported at long time scale, individual simulations do not yield reliable free energies. Ensembles of independent trajectories are necessary to overcome the aleatoric uncertainty in order to obtain statistically meaningful and reproducible results. Finally, we compare the application of different free energy methods to these systems and discuss their advantages and disadvantages. Our findings here are generally applicable to all molecular dynamics based applications and not confined to the free energy methods used in this study.

No Substrate Left behind-Mining of Shotgun Proteomics Datasets Rescues Evidence of Proteolysis by SARS-CoV-2 3CLpro Main Protease.

Proteolytic processing is the most ubiquitous post-translational modification and regulator of protein function. To identify protease substrates, and hence the function of proteases, terminomics workflows have been developed to enrich and detect proteolytically generated protein termini from mass spectrometry data. The mining of shotgun proteomics datasets for such 'neo'-termini, to increase the understanding of proteolytic processing, is an underutilized opportunity. However, to date, this approach has been hindered by the lack of software with sufficient speed to make searching for the relatively low numbers of protease-generated semi-tryptic peptides present in non-enriched samples viable. We reanalyzed published shotgun proteomics datasets for evidence of proteolytic processing in COVID-19 using the recently upgraded MSFragger/FragPipe software, which searches data with a speed that is an order of magnitude greater than many equivalent tools. The number of protein termini identified was higher than expected and constituted around half the number of termini detected by two different N-terminomics methods. We identified neo-N- and C-termini generated during SARS-CoV-2 infection that were indicative of proteolysis and were mediated by both viral and host proteases-a number of which had been recently validated by in vitro assays. Thus, re-analyzing existing shotgun proteomics data is a valuable adjunct for terminomics research that can be readily tapped (for example, in the next pandemic where data would be scarce) to increase the understanding of protease function and virus-host interactions, or other diverse biological processes.

Structural Dynamics Predominantly Determine the Adaptability of Proteins to Amino Acid Deletions.

The insertion or deletion (indel) of amino acids has a variety of effects on protein function, ranging from disease-forming changes to gaining new functions. Despite their importance, indels have not been systematically characterized towards protein engineering or modification goals. In the present work, we focus on deletions composed of multiple contiguous amino acids (mAA-dels) and their effects on the protein (mutant) folding ability. Our analysis reveals that the mutant retains the native fold when the mAA-del obeys well-defined structural dynamics properties: localization in intrinsically flexible regions, showing low resistance to mechanical stress, and separation from allosteric signaling paths. Motivated by the possibility of distinguishing the features that underlie the adaptability of proteins to mAA-dels, and by the rapid evaluation of these features using elastic network models, we developed a positive-unlabeled learning-based classifier that can be adopted for protein design purposes. Trained on a consolidated set of features, including those reflecting the intrinsic dynamics of the regions where the mAA-dels occur, the new classifier yields a high recall of 84.3% for identifying mAA-dels that are stably tolerated by the protein. The comparative examination of the relative contribution of different features to the prediction reveals the dominant role of structural dynamics in enabling the adaptation of the mutant to mAA-del without disrupting the native fold.

From Deep Mutational Mapping of Allosteric Protein Landscapes to Deep Learning of Allostery and Hidden Allosteric Sites: Zooming in on "Allosteric Intersection" of Biochemical and Big Data Approaches.

The recent advances in artificial intelligence (AI) and machine learning have driven the design of new expert systems and automated workflows that are able to model complex chemical and biological phenomena. In recent years, machine learning approaches have been developed and actively deployed to facilitate computational and experimental studies of protein dynamics and allosteric mechanisms. In this review, we discuss in detail new developments along two major directions of allosteric research through the lens of data-intensive biochemical approaches and AI-based computational methods. Despite considerable progress in applications of AI methods for protein structure and dynamics studies, the intersection between allosteric regulation, the emerging structural biology technologies and AI approaches remains largely unexplored, calling for the development of AI-augmented integrative structural biology. In this review, we focus on the latest remarkable progress in deep high-throughput mining and comprehensive mapping of allosteric protein landscapes and allosteric regulatory mechanisms as well as on the new developments in AI methods for prediction and characterization of allosteric binding sites on the proteome level. We also discuss new AI-augmented structural biology approaches that expand our knowledge of the universe of protein dynamics and allostery. We conclude with an outlook and highlight the importance of developing an open science infrastructure for machine learning studies of allosteric regulation and validation of computational approaches using integrative studies of allosteric mechanisms. The development of community-accessible tools that uniquely leverage the existing experimental and simulation knowledgebase to enable interrogation of the allosteric functions can provide a much-needed boost to further innovation and integration of experimental and computational technologies empowered by booming AI field.

3DFlex: determining structure and motion of flexible proteins from cryo-EM.

Modeling flexible macromolecules is one of the foremost challenges in single-particle cryogenic-electron microscopy (cryo-EM), with the potential to illuminate fundamental questions in structural biology. We introduce Three-Dimensional Flexible Refinement (3DFlex), a motion-based neural network model for continuous molecular heterogeneity for cryo-EM data. 3DFlex exploits knowledge that conformational variability of a protein is often the result of physical processes that transport density over space and tend to preserve local geometry. From two-dimensional image data, 3DFlex enables the determination of high-resolution 3D density, and provides an explicit model of a flexible protein's motion over its conformational landscape. Experimentally, for large molecular machines (tri-snRNP spliceosome complex, translocating ribosome) and small flexible proteins (TRPV1 ion channel, αVß8 integrin, SARS-CoV-2 spike), 3DFlex learns nonrigid molecular motions while resolving details of moving secondary structure elements. 3DFlex can improve 3D density resolution beyond the limits of existing methods because particle images contribute coherent signal over the conformational landscape.

Reverse tracking from drug-induced transcriptomes through multilayer molecular networks reveals hidden drug targets.

Identifying molecular targets of a drug is an essential process for drug discovery and development. The recent in-silico approaches are usually based on the structure information of chemicals and proteins. However, 3D structure information is hard to obtain and machine-learning methods using 2D structure suffer from data imbalance problem. Here, we present a reverse tracking method from genes to target proteins using drug-perturbed gene transcriptional profiles and multilayer molecular networks. We scored how well the protein explains gene expression changes perturbed by a drug. We validated the protein scores of our method in predicting known targets of drugs. Our method performs better than other methods using the gene transcriptional profiles and shows the ability to suggest the molecular mechanism of drugs. Furthermore, our method has the potential to predict targets for objects that do not have rigid structural information, such as coronavirus.

De novo design of protein interactions with learned surface fingerprints.

Physical interactions between proteins are essential for most biological processes governing life1. However, the molecular determinants of such interactions have been challenging to understand, even as genomic, proteomic and structural data increase. This knowledge gap has been a major obstacle for the comprehensive understanding of cellular protein-protein interaction networks and for the de novo design of protein binders that are crucial for synthetic biology and translational applications2-9. Here we use a geometric deep-learning framework operating on protein surfaces that generates fingerprints to describe geometric and chemical features that are critical to drive protein-protein interactions10. We hypothesized that these fingerprints capture the key aspects of molecular recognition that represent a new paradigm in the computational design of novel protein interactions. As a proof of principle, we computationally designed several de novo protein binders to engage four protein targets: SARS-CoV-2 spike, PD-1, PD-L1 and CTLA-4. Several designs were experimentally optimized, whereas others were generated purely in silico, reaching nanomolar affinity with structural and mutational characterization showing highly accurate predictions. Overall, our surface-centric approach captures the physical and chemical determinants of molecular recognition, enabling an approach for the de novo design of protein interactions and, more broadly, of artificial proteins with function.

Standardizing macromolecular structure files: further efforts are needed.

Investigating large datasets of biological information by automatic procedures may offer chances of progress in knowledge. Recently, tremendous improvements in structural biology have allowed the number of structures in the Protein Data Bank (PDB) archive to increase rapidly, in particular those for severe acute respiratory syndrome coronavirus 2 (SARS-CoV-2)-associated proteins. However, their automatic analysis can be hampered by the nonuniform descriptors used by authors in some records of the PDB and PDBx/mmCIF files. In this opinion article we highlight the difficulties encountered in automating the analysis of hundreds of structures, suggesting that further standardization of the description of these molecular entities and of their attributes, generalized to the macromolecular structures contained in the PDB, might generate files more suitable for automatized analyses of a large number of structures.

Evolution of Mass Spectrometry Instruments and Techniques for Blood Proteomics.

Mass spectrometry (MS)-based blood proteomics is a crucial research focus in identifying disease biomarkers. Blood serum or plasma is the most commonly used sample for such analysis; however, it presents challenges due to the complexity and dynamic range of protein abundance. Despite these difficulties, the development of high-resolution MS instruments has made comprehensive investigation of blood proteomics possible. The evolution of time-of-flight (TOF) or Orbitrap MS instruments has played a significant role in the field of blood proteomics. These instruments are now among the most prominent techniques for blood proteomics due to their sensitivity, selectivity, fast response, and stability. For optimal results, it is necessary to eliminate high-abundance proteins from the blood sample, to maximize the depth coverage of the blood proteomics analysis. This can be achieved through various methods, including commercial kits, chemically synthesized materials, and MS technologies. This paper reviews recent advancements in MS technology and its remarkable applications in biomarker discovery, particularly in the areas of cancer and COVID-19 studies.

Machine learning and protein allostery.

The fundamental biological importance and complexity of allosterically regulated proteins stem from their central role in signal transduction and cellular processes. Recently, machine-learning approaches have been developed and actively deployed to facilitate theoretical and experimental studies of protein dynamics and allosteric mechanisms. In this review, we survey recent developments in applications of machine-learning methods for studies of allosteric mechanisms, prediction of allosteric effects and allostery-related physicochemical properties, and allosteric protein engineering. We also review the applications of machine-learning strategies for characterization of allosteric mechanisms and drug design targeting SARS-CoV-2. Continuous development and task-specific adaptation of machine-learning methods for protein allosteric mechanisms will have an increasingly important role in bridging a wide spectrum of data-intensive experimental and theoretical technologies.

Analysis of protein-protein interface with incorporating low-frequency molecular interactions in molecular dynamics simulation.

Protein-protein interactions are vital for various biological processes such as immune reaction, signal transduction, and viral infection. Molecular Dynamics (MD) simulation is a powerful tool for analyzing non-covalent interactions between two protein molecules. In general, MD simulation studies on the protein-protein interface have focused on the analysis of major and frequent molecular interactions. In this study, we demonstrate that minor interactions with low-frequency need to be incorporated to analyze the molecular interactions in the protein-protein interface more efficiently using the complex of SARS-CoV2-RBD and ACE2 receptor as a model system. It was observed that the dominance of interactions in the MD-simulated structures didn't directly correlate with the interactions in the experimentally determined structure. The interactions from the experimentally determined structure could be reproduced better in the ensemble of MD simulated structures by including the less frequent interactions compared to the norm of choosing only highly frequent interactions. Residue Interaction Networks (RINs) analysis also showed that the critical residues in the protein-protein interface could be more efficiently identified by incorporating low-frequency interactions in MD simulation. It is expected that the approach proposed in this study can be a new way of studying protein-protein interaction through MD simulation.

Exploring and Learning the Universe of Protein Allostery Using Artificial Intelligence Augmented Biophysical and Computational Approaches.

Allosteric mechanisms are commonly employed regulatory tools used by proteins to orchestrate complex biochemical processes and control communications in cells. The quantitative understanding and characterization of allosteric molecular events are among major challenges in modern biology and require integration of innovative computational experimental approaches to obtain atomistic-level knowledge of the allosteric states, interactions, and dynamic conformational landscapes. The growing body of computational and experimental studies empowered by emerging artificial intelligence (AI) technologies has opened up new paradigms for exploring and learning the universe of protein allostery from first principles. In this review we analyze recent developments in high-throughput deep mutational scanning of allosteric protein functions; applications and latest adaptations of Alpha-fold structural prediction methods for studies of protein dynamics and allostery; new frontiers in integrating machine learning and enhanced sampling techniques for characterization of allostery; and recent advances in structural biology approaches for studies of allosteric systems. We also highlight recent computational and experimental studies of the SARS-CoV-2 spike (S) proteins revealing an important and often hidden role of allosteric regulation driving functional conformational changes, binding interactions with the host receptor, and mutational escape mechanisms of S proteins which are critical for viral infection. We conclude with a summary and outlook of future directions suggesting that AI-augmented biophysical and computer simulation approaches are beginning to transform studies of protein allostery toward systematic characterization of allosteric landscapes, hidden allosteric states, and mechanisms which may bring about a new revolution in molecular biology and drug discovery.

Improving Protein-Ligand Interaction Modeling with cryo-EM Data, Templates, and Deep Learning in 2021 Ligand Model Challenge.

Elucidating protein-ligand interaction is crucial for studying the function of proteins and compounds in an organism and critical for drug discovery and design. The problem of protein-ligand interaction is traditionally tackled by molecular docking and simulation, which is based on physical forces and statistical potentials and cannot effectively leverage cryo-EM data and existing protein structural information in the protein-ligand modeling process. In this work, we developed a deep learning bioinformatics pipeline (DeepProLigand) to predict protein-ligand interactions from cryo-EM density maps of proteins and ligands. DeepProLigand first uses a deep learning method to predict the structure of proteins from cryo-EM maps, which is averaged with a reference (template) structure of the proteins to produce a combined structure to add ligands. The ligands are then identified and added into the structure to generate a protein-ligand complex structure, which is further refined. The method based on the deep learning prediction and template-based modeling was blindly tested in the 2021 EMDataResource Ligand Challenge and was ranked first in fitting ligands to cryo-EM density maps. These results demonstrate that the deep learning bioinformatics approach is a promising direction for modeling protein-ligand interactions on cryo-EM data using prior structural information.

Active Learning Guided Drug Design Lead Optimization Based on Relative Binding Free Energy Modeling.

In silico identification of potent protein inhibitors commonly requires prediction of a ligand binding free energy (BFE). Thermodynamics integration (TI) based on molecular dynamics (MD) simulations is a BFE calculation method capable of acquiring accurate BFE, but it is computationally expensive and time-consuming. In this work, we have developed an efficient automated workflow for identifying compounds with the lowest BFE among thousands of congeneric ligands, which requires only hundreds of TI calculations. Automated machine learning (AutoML) orchestrated by active learning (AL) in an AL-AutoML workflow allows unbiased and efficient search for a small set of best-performing molecules. We have applied this workflow to select inhibitors of the SARS-CoV-2 papain-like protease and were able to find 133 compounds with improved binding affinity, including 16 compounds with better than 100-fold binding affinity improvement. We obtained a hit rate that outperforms that expected of traditional expert medicinal chemist-guided campaigns. Thus, we demonstrate that the combination of AL and AutoML with free energy simulations provides at least 20× speedup relative to the naïve brute force approaches.

Improved AlphaFold modeling with implicit experimental information.

Machine-learning prediction algorithms such as AlphaFold and RoseTTAFold can create remarkably accurate protein models, but these models usually have some regions that are predicted with low confidence or poor accuracy. We hypothesized that by implicitly including new experimental information such as a density map, a greater portion of a model could be predicted accurately, and that this might synergistically improve parts of the model that were not fully addressed by either machine learning or experiment alone. An iterative procedure was developed in which AlphaFold models are automatically rebuilt on the basis of experimental density maps and the rebuilt models are used as templates in new AlphaFold predictions. We show that including experimental information improves prediction beyond the improvement obtained with simple rebuilding guided by the experimental data. This procedure for AlphaFold modeling with density has been incorporated into an automated procedure for interpretation of crystallographic and electron cryo-microscopy maps.

Electrostatics in Computational Biophysics and Its Implications for Disease Effects.

This review outlines the role of electrostatics in computational molecular biophysics and its implication in altering wild-type characteristics of biological macromolecules, and thus the contribution of electrostatics to disease mechanisms. The work is not intended to review existing computational approaches or to propose further developments. Instead, it summarizes the outcomes of relevant studies and provides a generalized classification of major mechanisms that involve electrostatic effects in both wild-type and mutant biological macromolecules. It emphasizes the complex role of electrostatics in molecular biophysics, such that the long range of electrostatic interactions causes them to dominate all other forces at distances larger than several Angstroms, while at the same time, the alteration of short-range wild-type electrostatic pairwise interactions can have pronounced effects as well. Because of this dual nature of electrostatic interactions, being dominant at long-range and being very specific at short-range, their implications for wild-type structure and function are quite pronounced. Therefore, any disruption of the complex electrostatic network of interactions may abolish wild-type functionality and could be the dominant factor contributing to pathogenicity. However, we also outline that due to the plasticity of biological macromolecules, the effect of amino acid mutation may be reduced, and thus a charge deletion or insertion may not necessarily be deleterious.

Large-Scale Functionalized Metasurface-Based SARS-CoV-2 Detection and Quantification.

Plasmonic metasurfaces consist of metal-dielectric interfaces that are excitable at background and leakage resonant modes. The sharp and plasmonic excitation profile of metal-free electrons on metasurfaces at the nanoscale can be used for practical applications in diverse fields, including optoelectronics, energy harvesting, and biosensing. Currently, Fano resonant metasurface fabrication processes for biosensor applications are costly, need clean room access, and involve limited small-scale surface areas that are not easy for accurate sample placement. Here, we leverage the large-scale active area with uniform surface patterns present on optical disc-based metasurfaces as a cost-effective method to excite asymmetric plasmonic modes, enabling tunable optical Fano resonance interfacing with a microfluidic channel for multiple target detection in the visible wavelength range. We engineered plasmonic metasurfaces for biosensing through efficient layer-by-layer surface functionalization toward real-time measurement of target binding at the molecular scale. Further, we demonstrated the quantitative detection of antibodies, proteins, and the whole viral particles of SARS-CoV-2 with a high sensitivity and specificity, even distinguishing it from similar RNA viruses such as influenza and MERS. This cost-effective plasmonic metasurface platform offers a small-scale light-manipulation system, presenting considerable potential for fast, real-time detection of SARS-CoV-2 and pathogens in resource-limited settings.

Ligand Binding Prediction Using Protein Structure Graphs and Residual Graph Attention Networks.

Computational prediction of ligand-target interactions is a crucial part of modern drug discovery as it helps to bypass high costs and labor demands of in vitro and in vivo screening. As the wealth of bioactivity data accumulates, it provides opportunities for the development of deep learning (DL) models with increasing predictive powers. Conventionally, such models were either limited to the use of very simplified representations of proteins or ineffective voxelization of their 3D structures. Herein, we present the development of the PSG-BAR (Protein Structure Graph-Binding Affinity Regression) approach that utilizes 3D structural information of the proteins along with 2D graph representations of ligands. The method also introduces attention scores to selectively weight protein regions that are most important for ligand binding. Results: The developed approach demonstrates the state-of-the-art performance on several binding affinity benchmarking datasets. The attention-based pooling of protein graphs enables identification of surface residues as critical residues for protein-ligand binding. Finally, we validate our model predictions against an experimental assay on a viral main protease (Mpro)-the hallmark target of SARS-CoV-2 coronavirus.

The impact of AlphaFold2 on experimental structure solution.

AlphaFold2 is a machine-learning based program that predicts a protein structure based on the amino acid sequence. In this article, we report on the current usages of this new tool and give examples from our work in the Coronavirus Structural Task Force. With its unprecedented accuracy, it can be utilized for the design of expression constructs, de novo protein design and the interpretation of Cryo-EM data with an atomic model. However, these methods are limited by their training data and are of limited use to predict conformational variability and fold flexibility; they also lack co-factors, post-translational modifications and multimeric complexes with oligonucleotides. They also are not always perfect in terms of chemical geometry. Nevertheless, machine learning-based fold prediction is a game changer for structural bioinformatics and experimentalists alike, with exciting developments ahead.