Search

The primary immunological target of COVID-19 vaccines is the SARS-CoV-2 spike (S) protein. S is exposed on the viral surface and mediates viral entry into the host cell. To identify possible antibody binding sites, we performed multi-microsecond molecular dynamics simulations of a 4.1 million atom system containing a patch of viral membrane with four full-length, fully glycosylated and palmitoylated S proteins. By mapping steric accessibility, structural rigidity, sequence conservation, and generic antibody binding signatures, we recover known epitopes on S and reveal promising epitope candidates for structure-based vaccine design. We find that the extensive and inherently flexible glycan coat shields a surface area larger than expected from static structures, highlighting the importance of structural dynamics. The protective glycan shield and the high flexibility of its hinges give the stalk overall low epitope scores. Our computational epitope-mapping procedure is general and should thus prove useful for other viral envelope proteins whose structures have been characterized., Author summary The SARS-CoV-2 virus has caused a global health crisis. The spike protein exposed at its surface is key for infection and the primary antibody target. However, spike is covered by highly mobile glycan molecules that could impair antibody binding. To identify accessible epitopes, we performed molecular dynamics simulations of an atomistic model of glycosylated spike embedded in a membrane. By combining extensive simulations with bioinformatics analyses, we recovered known antibody binding sites and identified several epitope candidates as targets for further vaccine development.

High-throughput B-cell sequencing has opened up new avenues for investigating complex mechanisms underlying our adaptive immune response. These technological advances drive data generation and the need to mine and analyze the information contained in these large datasets, in particular the identification of therapeutic antibodies (Abs) or those associated with disease exposure and protection. Here, we describe our efforts to use artificial intelligence (AI)-based image-analyses for prospective classification of Abs based solely on sequence information. We hypothesized that Abs recognizing the same part of an antigen share a limited set of features at the binding interface, and that the binding site regions of these Abs share share common structure and physicochemical property patterns that can serve as a “fingerprint” to recognize uncharacterized Abs. We combined large-scale sequence-based protein-structure predictions to generate ensembles of 3-D Ab models, reduced the Ab binding interface to a 2-D image (fingerprint), used pre-trained convolutional neural networks to extract features, and trained deep neural networks (DNNs) to classify Abs. We evaluated this approach using Ab sequences derived from human HIV and Ebola viral infections to differentiate between two Abs, Abs belonging to specific B-cell family lineages, and Abs with different epitope preferences. In addition, we explored a different type of DNN method to detect one class of Abs from a larger pool of Abs. Testing on Ab sets that had been kept aside during model training, we achieved average prediction accuracies ranging from 71–96% depending on the complexity of the classification task. The high level of accuracies reached during these classification tests suggests that the DNN models were able to learn a series of structural patterns shared by Abs belonging to the same class. The developed methodology provides a means to apply AI-based image recognition techniques to analyze high-throughput B-cell sequencing datasets (repertoires) for Ab classification., Author summary The ability to take advantage of the rapid progress in AI for biological and medical application oftentimes requires looking at the problem from a non-traditional point-of-view. The adaptive immune system plays a key role in providing long-term immunity against pathogens. The repertoire of circulating B-cells that produce unique pathogen-specific antibodies in an individual contains immense information on both the status of the immune response at particular time and that individual’s immune history. With high-throughput sequencing, we can now obtain Ab sequences for thousands of B cells from a single patient blood sample, but functionally characterizing antibodies on this scale remains on daunting task. Here, we propose to use AI to functionally classify Abs from sequence alone by re-casting this classification problem as an image recognition problem. Just as traditional image recognition involves training AI to distinguish different types of objects, we sought to use AI to distinguish different types of Ab-antigen binding interfaces. Towards that end, we generated ensembles of Ab structures from sequence, and generated 2-D ‘fingerprints’ of each structure that captures the essential molecular and chemical structure of the Ab binding site regions, and trained a Convolution and Deep Neural Network based AI model to classify Ab fingerprints associated with different functional characteristics. We applied this DNN-based approach to accurately predict antibody family lineage and epitope specificity against Ebola and HIV-1 viruses, and to detect sequence-diverse antibodies with similar binding properties as the ones we used for training.

Identifying the epitope of an antibody is a key step in understanding its function and its potential as a therapeutic. Sequence-based clonal clustering can identify antibodies with similar epitope complementarity, however, antibodies from markedly different lineages but with similar structures can engage the same epitope. We describe a novel computational method for epitope profiling based on structural modelling and clustering. Using the method, we demonstrate that sequence dissimilar but functionally similar antibodies can be found across the Coronavirus Antibody Database, with high accuracy (92% of antibodies in multiple-occupancy structural clusters bind to consistent domains). Our approach functionally links antibodies with distinct genetic lineages, species origins, and coronavirus specificities. This indicates greater convergence exists in the immune responses to coronaviruses than is suggested by sequence-based approaches. Our results show that applying structural analytics to large class-specific antibody databases will enable high confidence structure-function relationships to be drawn, yielding new opportunities to identify functional convergence hitherto missed by sequence-only analysis., Author summary Antibodies are a key component of the immune system that combat pathogens by binding to a defined region of their molecular surface (known as an ‘epitope’). The ability to map which antibodies target the same epitopes is crucial when designing non-competing antibody therapeutics or predicting the influence of pathogen mutation on population immunity. While one can use laboratory experiments to deduce when pairs of antibodies engage the same epitope, such experiments are very expensive and time consuming if used to compare on the order of thousands of antibodies. In this work, we report a new computational algorithm (SPACE) that clusters antibodies that target the same epitope based on their predicted 3D structure, as binding site structure is a property often conserved between binders complementary to the same epitope. Unlike existing antibody epitope profiling tools which assume two antibodies must share a high sequence identity/similar genetic basis to engage the same region, our orthogonal method can detect broader patterns of convergent evolution across binders to different pathogen strains, and between antibodies with different genetic and even species origins.

It is difficult to predict how antibodies will behave when mixed together, even after each has been independently characterized. Here, we present a statistical mechanical model for the activity of antibody mixtures that accounts for whether pairs of antibodies bind to distinct or overlapping epitopes. This model requires measuring n individual antibodies and their n(n-1)2 pairwise interactions to predict the 2n potential combinations. We apply this model to epidermal growth factor receptor (EGFR) antibodies and find that the activity of antibody mixtures can be predicted without positing synergy at the molecular level. In addition, we demonstrate how the model can be used in reverse, where straightforward experiments measuring the activity of antibody mixtures can be used to infer the molecular interactions between antibodies. Lastly, we generalize this model to analyze engineered multidomain antibodies, where components of different antibodies are tethered together to form novel amalgams, and characterize how well it predicts recently designed influenza antibodies., Author summary With the rise of new antibody combinations in therapeutic regimens, it is important to understand how antibodies work together as well as individually. Here, we investigate the specific case of monoclonal antibodies targeting a cancer-causing receptor or the influenza virus and develop a statistical mechanical framework that predicts the effectiveness of a mixture of antibodies. The power of this model lies in its ability to make a large number of predictions based on a limited amount of data. For example, once 10 antibodies have been individually characterized and their epitopes have been mapped, our model can predict how any of the 210 = 1024 combinations will behave. This predictive power can aid therapeutic efforts by assessing which combinations of antibodies will elicit the most effective response.

The ongoing Antibody Mediated Prevention (AMP) trials will uncover whether passive infusion of the broadly neutralizing antibody (bNAb) VRC01 can protect against HIV acquisition. Previous statistical simulations indicate these trials may be partially protective. In that case, it will be crucial to identify the mechanism of breakthrough infections. To that end, we developed a mathematical modeling framework to simulate the AMP trials and infer the breakthrough mechanisms using measurable trial outcomes. This framework combines viral dynamics with antibody pharmacokinetics and pharmacodynamics, and will be generally applicable to forthcoming bNAb prevention trials. We fit our model to human viral load data (RV217). Then, we incorporated VRC01 neutralization using serum pharmacokinetics (HVTN 104) and in vitro pharmacodynamics (LANL CATNAP database). We systematically explored trial outcomes by reducing in vivo potency and varying the distribution of sensitivity to VRC01 in circulating strains. We found trial outcomes could be used in a clinical trial regression model (CTRM) to reveal whether partially protective trials were caused by large fractions of VRC01-resistant (IC50>50 μg/mL) circulating strains or rather a global reduction in VRC01 potency against all strains. The former mechanism suggests the need to enhance neutralizing antibody breadth; the latter suggests the need to enhance VRC01 delivery and/or in vivo binding. We will apply the clinical trial regression model to data from the completed trials to help optimize future approaches for passive delivery of anti-HIV neutralizing antibodies., Author summary Infusions of broadly neutralizing antibodies are currently being tested as a novel HIV prevention modality. To help interpret the results of these antibody mediated prevention (AMP) studies we developed a mathematical modeling framework. The approach combines antibody potency and drug levels with models of HIV viral dynamics, which will be generally applicable to future studies. Through simulating these clinical trials, we found trial outcomes can be used in combination to infer whether breakthrough infections are caused by large fractions of antibody-resistant circulating strains or some reduction in potency against all strains. This distinction helps to focus future trials on enhancing neutralizing antibody breadth or antibody delivery and/or in vivo binding.

The SARS-CoV-2 pandemic highlights the need for a detailed molecular understanding of protective antibody responses. This is underscored by the emergence and spread of SARS-CoV-2 variants, including Alpha (B.1.1.7) and Delta (B.1.617.2), some of which appear to be less effectively targeted by current monoclonal antibodies and vaccines. Here we report a high resolution and comprehensive map of antibody recognition of the SARS-CoV-2 spike receptor binding domain (RBD), which is the target of most neutralizing antibodies, using computational structural analysis. With a dataset of nonredundant experimentally determined antibody-RBD structures, we classified antibodies by RBD residue binding determinants using unsupervised clustering. We also identified the energetic and conservation features of epitope residues and assessed the capacity of viral variant mutations to disrupt antibody recognition, revealing sets of antibodies predicted to effectively target recently described viral variants. This detailed structure-based reference of antibody RBD recognition signatures can inform therapeutic and vaccine design strategies., Author summary The ongoing COVID-19 pandemic, and the emergence of SARS-CoV-2 variants that evade antibodies induced by vaccines and natural infection, highlights the need for assessment of key molecular and structural features of immune responses against the SARS-CoV-2 virus. Using a large nonredundant set of structures of monoclonal antibodies in complex with the SARS-CoV-2 spike receptor binding domain, we performed analysis of molecular determinants of antibody recognition of the spike glycoprotein, mapping key residues through analysis of atomic contacts and computational modeling to identify molecular hotspots. Clustering was used to identify four major groups of antibodies based on target residues, and we compared epitope conservation and impact of SARS-CoV-2 variant mutations, showing that certain sets of antibodies predicted to be affected by those variants, while others are capable of targeting escape variants. This analysis can serve as a useful reference for vaccine and immunotherapeutic studies, and we provide updated classifications of antibodies to the research community on our CoV3D site.

Treating macaques with an anti-α4β7 antibody under the umbrella of combination antiretroviral therapy (cART) during early SIV infection can lead to viral remission, with viral loads maintained at < 50 SIV RNA copies/ml after removal of all treatment in a subset of animals. Depletion of CD8+ lymphocytes in controllers resulted in transient recrudescence of viremia, suggesting that the combination of cART and anti-α4β7 antibody treatment led to a state where ongoing immune responses kept the virus undetectable in the absence of treatment. A previous mathematical model of HIV infection and cART incorporates immune effector cell responses and exhibits the property of two different viral load set-points. While the lower set-point could correspond to the attainment of long-term viral remission, attaining the higher set-point may be the result of viral rebound. Here we expand that model to include possible mechanisms of action of an anti-α4β7 antibody operating in these treated animals. We show that the model can fit the longitudinal viral load data from both IgG control and anti-α4β7 antibody treated macaques, suggesting explanations for the viral control associated with cART and an anti-α4β7 antibody treatment. This effective perturbation to the virus-host interaction can also explain observations in other nonhuman primate experiments in which cART and immunotherapy have led to post-treatment control or resetting of the viral load set-point. Interestingly, because the viral kinetics in the various treated animals differed—some animals exhibited large fluctuations in viral load after cART cessation—the model suggests that anti-α4β7 treatment could act by different primary mechanisms in different animals and still lead to post-treatment viral control. This outcome is nonetheless in accordance with a model with two stable viral load set-points, in which therapy can perturb the system from one set-point to a lower one through different biological mechanisms., Author summary Some macaques treated with an anti-α4β7 monoclonal antibody along with antiretroviral therapy during the early stages of simian immunodeficiency virus infection had their viral load become undetectable (below 50 SIV RNA copies/ml) after all treatment was stopped, whereas animals not given the antibody all had their viral loads rebound to high levels. Using a mathematical model, we examined four potential ways in which the antibody could have altered the balance between viral growth and immune control to maintain an undetectable viral load. We show that a shift to controlled infection can occur through multiple biologically reasonable mechanisms of action of the anti-α4β7 antibody.

Developing HIV-1 vaccines that trigger broadly neutralizing antibodies (bnAbs) is a priority as bnAbs are considered key to elicitation of a protective immune response. To investigate whether the breadth of a neutralizing antibody (nAb) depended on the conservation of its epitope among circulating viruses, we examined Antibody:Envelope (Ab:Env) interactions and worldwide Env diversity. We found that sites corresponding to bnAb epitopes were as variable as other accessible, non-hypervariable Env sites (p = 0.50, Mann-Whitney U-test) with no significant relationship between epitope conservation and neutralization breadth (Spearman’s ρ = -0.44, adjusted p = 0.079). However, when accounting for key sites in the Ab:Env interaction, we showed that the broadest bnAbs targeted more conserved epitopes (Spearman’s ρ = -0.70, adjusted p = 5.0e-5). Neutralization breadth did not stem from the overall conservation of Ab epitopes but depended instead on the conservation of key sites of the Ab:Env interaction, revealing a mechanistic basis for neutralization breadth that could be exploited for vaccine design., Author summary So far, no HIV-1 vaccine has elicited broadly neutralizing antibodies (bnAbs) in humans. HIV-1, one of the most rapidly evolving pathogens, is remarkable for its high variability across individuals and adaptability within hosts. We tested the relationship between HIV-1 diversity and neutralization breadth. While bnAbs did not specifically target more conserved regions of HIV-1 Env, we found that the broadest bnAbs relied forcibly more on structural interactions at key sites of the Ab:Env interaction than other Abs. Understanding mechanisms underlying neutralization breadth provides guidelines to design more efficacious vaccines and antibody-based therapeutics.

We present a flexible, open source R package designed to obtain biological and epidemiological insights from serological datasets. Characterising past exposures for multi-strain pathogens poses a specific statistical challenge: observed antibody responses measured in serological assays depend on multiple unobserved prior infections that produce cross-reactive antibody responses. We provide a general modelling framework to jointly infer infection histories and describe immune responses generated by these infections using antibody titres against current and historical strains. We do this by linking latent infection dynamics with a mechanistic model of antibody kinetics that generates expected antibody titres over time. Our aim is to provide a flexible package to identify infection histories that can be applied to a range of pathogens. We present two case studies to illustrate how our model can infer key immunological parameters, such as antibody titre boosting, waning and cross-reaction, as well as latent epidemiological processes such as attack rates and age-stratified infection risk., Author summary Antibody levels can determine previous exposure to a pathogen and how likely individuals are to be infected in the future. However, antibody concentrations change over time, and some pathogens are continually evolving. In such cases, individuals may be infected and vaccinated multiple times when their pre-existing immunity fails, leading to a wide range of antibody profiles. Traditional approaches to analyse such data do not typically account for this. In addition, studies collecting antibody data may be designed differently, but are often underpinned by similar biological processes. We developed a statistical method and accompanying software package to better understand the immunology and epidemiology of these complex systems using serological data. We present two case studies to demonstrate how our software package, serosolver, can be applied to different settings: i) the epidemiology of the 2009 pandemic A/H1N1 influenza virus in Hong Kong and ii) historical patterns of A/H3N2 influenza infection in Guangzhou, China. These results demonstrate how modern analytical methods can reveal additional information from serological data that is otherwise missed using traditional approaches.

Computational protein design has been successful in modeling fixed backbone proteins in a single conformation. However, when modeling large ensembles of flexible proteins, current methods in protein design have been insufficient. Large barriers in the energy landscape are difficult to traverse while redesigning a protein sequence, and as a result current design methods only sample a fraction of available sequence space. We propose a new computational approach that combines traditional structure-based modeling using the Rosetta software suite with machine learning and integer linear programming to overcome limitations in the Rosetta sampling methods. We demonstrate the effectiveness of this method, which we call BROAD, by benchmarking the performance on increasing predicted breadth of anti-HIV antibodies. We use this novel method to increase predicted breadth of naturally-occurring antibody VRC23 against a panel of 180 divergent HIV viral strains and achieve 100% predicted binding against the panel. In addition, we compare the performance of this method to state-of-the-art multistate design in Rosetta and show that we can outperform the existing method significantly. We further demonstrate that sequences recovered by this method recover known binding motifs of broadly neutralizing anti-HIV antibodies. Finally, our approach is general and can be extended easily to other protein systems. Although our modeled antibodies were not tested in vitro, we predict that these variants would have greatly increased breadth compared to the wild-type antibody., Author summary In this article, we report a new approach for protein design, which combines traditional structural modeling with machine learning and integer programming. Using this method, we are able to design antibodies that are predicted to bind large panels of antigenically diverse HIV variants. The combination of methods from these fields allows us to surpass protein design limitations that have been seen up to this point. We predict that if we tested these modified antibodies against HIV variants they would have greater neutralization breadth than any antibodies seen to this point.

Heavy glycosylation of the envelope (Env) surface subunit, gp120, is a key adaptation of HIV-1; however, the precise effects of glycosylation on the folding, conformation and dynamics of this protein are poorly understood. Here we explore the patterns of HIV-1 Env gp120 glycosylation, and particularly the enrichment in glycosylation sites proximal to the disulfide linkages at the base of the surface-exposed variable domains. To dissect the influence of glycans on the conformation these regions, we focused on an antigenic peptide fragment from a disulfide bridge-bounded region spanning the V1 and V2 hyper-variable domains of HIV-1 gp120. We used replica exchange molecular dynamics (MD) simulations to investigate how glycosylation influences its conformation and stability. Simulations were performed with and without N-linked glycosylation at two sites that are highly conserved across HIV-1 isolates (N156 and N160); both are contacts for recognition by V1V2-targeted broadly neutralizing antibodies against HIV-1. Glycosylation stabilized the pre-existing conformations of this peptide construct, reduced its propensity to adopt other secondary structures, and provided resistance against thermal unfolding. Simulations performed in the context of the Env trimer also indicated that glycosylation reduces flexibility of the V1V2 region, and provided insight into glycan-glycan interactions in this region. These stabilizing effects were influenced by a combination of factors, including the presence of a disulfide bond between the Cysteines at 131 and 157, which increased the formation of beta-strands. Together, these results provide a mechanism for conservation of disulfide linkage proximal glycosylation adjacent to the variable domains of gp120 and begin to explain how this could be exploited to enhance the immunogenicity of those regions. These studies suggest that glycopeptide immunogens can be designed to stabilize the most relevant Env conformations to focus the immune response on key neutralizing epitopes., Author Summary Heavy glycosylation of the envelope surface subunit, gp120, is a key adaptation of HIV-1, however, the precise effects of glycosylation on the folding, conformation and dynamics of this protein are poorly understood. The network of glycans on gp120 is of particular interest with regards to vaccine design, because the glycans both serve as targets for many classes of broadly neutralizing antibodies, and contribute to patterns of immune evasion and escape during HIV-1 infection. In this manuscript, we report on how glycosylation influences an immunogenic but disordered region of gp120. Glycosylation stabilizes the pre-existing conformation, and reduces its propensity to form other secondary structures. It also stabilizes preformed conformation against thermal unfolding. These complementary effects originate from a combination of multiple factors, including the observation that having a glycosylation site adjacent to the disulfide bond further promotes the formation of beta-strand structure in this peptide.

The spikes on virus surfaces bind receptors on host cells to propagate infection. High spike densities (SDs) can promote infection, but spikes are also targets of antibody-mediated immune responses. Thus, diverse evolutionary pressures can influence virus SDs. HIV’s SD is about two orders of magnitude lower than that of other viruses, a surprising feature of unknown origin. By modeling antibody evolution through affinity maturation, we find that an intermediate SD maximizes the affinity of generated antibodies. We argue that this leads most viruses to evolve high SDs. T helper cells, which are depleted during early HIV infection, play a key role in antibody evolution. We find that T helper cell depletion results in high affinity antibodies when SD is high, but not if SD is low. This special feature of HIV infection may have led to the evolution of a low SD to avoid potent immune responses early in infection., Author summary The spike protein on the virus surface mediates its entry to the host cell and a high spike density promotes infection. HIV has a spike density that is almost two orders of magnitude lower than other viruses. This unique feature of HIV has defied explanation since it was first observed. By bringing together theory and computation, rooted in statistical mechanics, with immunology we suggest that the effects of dramatic depletion of T helper cells during HIV infection on antibody production provided an evolutionary driving force for HIV to evolve a low spike density in order to avoid potent immune responses. Additionally, we show that an intermediate spike density induces maximally potent antibody production, a result with implications for vaccine design.

Accumulation of somatic mutations in antibody variable regions is critical for antibody affinity maturation, with HIV-1 broadly neutralizing antibodies (bnAbs) generally requiring years to develop. We recently found that the rate at which mutations accumulate decreases over time, but the mechanism governing this slowing is unclear. In this study, we investigated whether natural selection and/or mutability of the antibody variable region contributed significantly to observed decrease in rate. We used longitudinally sampled sequences of immunoglobulin transcripts of single lineages from each of 3 donors, as determined by next generation sequencing. We estimated the evolutionary rates of the complementarity determining regions (CDRs), which are most significant for functional selection, and found they evolved about 1.5- to 2- fold faster than the framework regions. We also analyzed the presence of AID hotspots and coldspots at different points in lineage development and observed an average decrease in mutability of less than 10 percent over time. Altogether, the correlation between Darwinian selection strength and evolutionary rate trended toward significance, especially for CDRs, but cannot fully explain the observed changes in evolutionary rate. The mutability modulated by AID hotspots and coldspots changes correlated only weakly with evolutionary rates. The combined effects of Darwinian selection and mutability contribute substantially to, but do not fully explain, evolutionary rate change for HIV-1-targeting bnAb lineages., Author Summary In an infected individual, the HIV-1 Env gene evolves at a rate of about 0.015 substitutions per site per year. Changes in viral epitopes in turn stimulate the co-evolution of recognizing antibody lineages. We previously showed that young antibody lineages can evolve at a rate ~10-fold faster than observed for HIV-1 and the rate of antibody evolution decreases over time. Here we investigate two factors, Darwinian selection and genetic mutability, which have been shown to influence evolutionary rates in other settings. We quantified both of these factors for three broadly HIV-1-neutralizing antibody lineages, and analyzed the association of these factors with changes in evolutionary rate. We found that Darwinian selection is a major factor in the slowing of evolutionary rate, while genetic mutability modulates antibody evolutionary rate weakly. Moreover, the combined effects of the two factors are unlikely to fully account for the slowing of antibody evolutionary rate.