Functional Deimmunization of Botulinum Neurotoxin Protease Domain via Computationally Driven Library Design and Ultrahigh-Throughput Screening

Botulinum neurotoxin serotype A (BoNT/A) is a widely used cosmetic agent that also has diverse therapeutic applications; however, adverse antidrug immune responses and associated loss of efficacy have been reported in clinical uses. Here, we describe computational design and ultrahigh-throughput screening of a massive BoNT/A light-chain (BoNT/A-LC) library optimized for reduced T cell epitope content and thereby dampened immunogenicity. We developed a functional assay based on bacterial co-expression of BoNT/A-LC library members with a Förster resonance energy transfer (FRET) sensor for BoNT/A-LC enzymatic activity, and we employed high-speed fluorescence-activated cell sorting (FACS) to identify numerous computationally designed variants having wild-type-like enzyme kinetics. Many of these variants exhibited decreased immunogenicity in humanized HLA transgenic mice and manifested in vivo paralytic activity when incorporated into full-length toxin. One variant achieved near-wild-type paralytic potency and a 300% reduction in antidrug antibody response in vivo. Thus, we have achieved a striking level of BoNT/A-LC functional deimmunization by combining computational library design and ultrahigh-throughput screening. This strategy holds promise for deimmunizing other biologics with complex superstructures and mechanisms of action.

Computational epitope binning reveals functional equivalence of sequence-divergent paratopes

•

Epitope binning groups target-specific protein binders recognizing the same binding region.

•

The “Epibin” method utilizes docking models to computationally predict competition and identify bins.

•

Epibin recapitulated binding competition of repebody variants as determined by immunoassays.

•

In addition, Epibin enabled identification of ‘paratope-equivalent’ residues in sequence-dissimilar variants.

•

Computational epitope binning can scale to allow characterization of entire antigen-specific antibody repertoires.

The therapeutic efficacy of a protein binder largely depends on two factors: its binding site and its binding affinity. Advances in in vitro library display screening and next-generation sequencing have enabled accelerated development of strong binders, yet identifying their binding sites still remains a major challenge. The differentiation, or “binning”, of binders into different groups that recognize distinct binding sites on their target is a promising approach that facilitates high-throughput screening of binders that may show different biological activity. Here we study the extent to which the information contained in the amino acid sequences comprising a set of target-specific binders can be leveraged to bin them, inferring functional equivalence of their binding regions, or paratopes, based directly on comparison of the sequences, their modeled structures, or their modeled interactions. Using a leucine-rich repeat binding scaffold known as a “repebody” as the source of diversity in recognition against interleukin-6 (IL-6), we show that the “Epibin” approach introduced here effectively utilized structural modelling and docking to extract specificity information encoded in the repebody amino acid sequences and thereby successfully recapitulate IL-6 binding competition observed in immunoassays. Furthermore, our computational binning provided a basis for designing in vitro mutagenesis experiments to pinpoint specificity-determining residues. Finally, we demonstrate that the Epibin approach can extend to antibodies, retrospectively comparing its predictions to results from antigen-specific antibody competition studies. The study thus demonstrates the utility of modeling structure and binding from the amino acid sequences of different binders against the same target, and paves the way for larger-scale binning and analysis of entire repertoires.

Ensuring scientific reproducibility in bio-macromolecular modeling via extensive, automated benchmarks

Each year vast international resources are wasted on irreproducible research. The scientific community has been slow to adopt standard software engineering practices, despite the increases in high-dimensional data, complexities of workflows, and computational environments. Here we show how scientific software applications can be created in a reproducible manner when simple design goals for reproducibility are met. We describe the implementation of a test server framework and 40 scientific benchmarks, covering numerous applications in Rosetta bio-macromolecular modeling. High performance computing cluster integration allows these benchmarks to run continuously and automatically. Detailed protocol captures are useful for developers and users of Rosetta and other macromolecular modeling tools. The framework and design concepts presented here are valuable for developers and users of any type of scientific software and for the scientific community to create reproducible methods. Specific examples highlight the utility of this framework, and the comprehensive documentation illustrates the ease of adding new tests in a matter of hours.

Diverse antiviral IgG effector activities are predicted by unique biophysical antibody features

BACKGROUND

The critical role of antibody Fc-mediated effector functions in immune defense has been widely reported in various viral infections. These effector functions confer cellular responses through engagement with innate immune cells. The precise mechanism(s) by which immunoglobulin G (IgG) Fc domain and cognate receptors may afford protection are poorly understood, however, in the context of HIV/SHIV infections. Many different in vitro assays have been developed and utilized to measure effector functions, but the extent to which these assays capture distinct antibody activities has not been fully elucidated.

RESULTS

In this study, six Fc-mediated effector function assays and two biophysical antibody profiling assays were performed on a common set of samples from HIV-1 infected and vaccinated subjects. Biophysical antibody profiles supported robust prediction of diverse IgG effector functions across distinct Fc-mediated effector function assays. While a number of assays showed correlated activities, supervised machine learning models indicated unique antibody features as primary contributing factors to the associated effector functions. Additional experiments established the mechanistic relevance of relationships discovered using this unbiased approach.

CONCLUSIONS

In sum, this study provides better resolution on the diversity and complexity of effector function assays, offering a clearer perspective into this family of antibody mechanisms of action to inform future HIV-1 treatment and vaccination strategies.

Comprehensive analysis of lectin-glycan interactions reveals determinants of lectin specificity

Lectin-glycan interactions facilitate inter- and intracellular communication in many processes including protein trafficking, host-pathogen recognition, and tumorigenesis promotion. Specific recognition of glycans by lectins is also the basis for a wide range of applications in areas including glycobiology research, cancer screening, and antiviral therapeutics. To provide a better understanding of the determinants of lectin-glycan interaction specificity and support such applications, this study comprehensively investigates specificity-conferring features of all available lectin-glycan complex structures. Systematic characterization, comparison, and predictive modeling of a set of 221 complementary physicochemical and geometric features representing these interactions highlighted specificity-conferring features with potential mechanistic insight. Univariable comparative analyses with weighted Wilcoxon-Mann-Whitney tests revealed strong statistical associations between binding site features and specificity that are conserved across unrelated lectin binding sites. Multivariable modeling with random forests demonstrated the utility of these features for predicting the identity of bound glycans based on generalized patterns learned from non-homologous lectins. These analyses revealed global determinants of lectin specificity, such as sialic acid glycan recognition in deep, concave binding sites enriched for positively charged residues, in contrast to high mannose glycan recognition in fairly shallow but well-defined pockets enriched for non-polar residues. Focused fine specificity analysis of hemagglutinin interactions with human-like and avian-like glycans uncovered features representing both known and novel mutations related to shifts in influenza tropism from avian to human tissues. As the approach presented here relies on co-crystallized lectin-glycan pairs for studying specificity, it is limited in its inferences by the quantity, quality, and diversity of the structural data available. Regardless, the systematic characterization of lectin binding sites presented here provides a novel approach to studying lectin specificity and is a step towards confidently predicting new lectin-glycan interactions.

Bioinformatics-driven discovery of novel Clostridioides difficile lysins and experimental comparison with highly active benchmarks

Clostridioides difficile is the single most deadly bacterial pathogen in the United States, and its global prevalence and outsized health impacts underscore the need for more effective therapeutic options. Towards this goal, a novel group of modified peptidoglycan hydrolases with significant in vitro bactericidal activity have emerged as potential candidates for treating C. difficile infections (CDI). To date, discovery and development efforts directed at these CDI-specific lysins have been limited, and in particular there has been no systematic comparison of known or newly discovered lysin candidates. Here, we detail bioinformatics-driven discovery of six new anti-C. difficile lysins belonging to the amidase-3 family of enzymes, and we describe experimental comparison of their respective catalytic domains (CATs) with highly active CATs from the literature. Our quantitative analyses include metrics for expression level, inherent antibacterial activity, breadth of strain selectivity, killing of germinating spores, and structural and functional measures of thermal stability. Importantly, prior studies have not examined stability as a performance metric, and our results show that the panel of eight enzymes possess widely variable thermal denaturation temperatures and resistance to heat inactivation, including some enzymes that exhibit marginal stability at body temperature. Ultimately, no single enzyme dominated with respect to all performance measures, suggesting the need for a balanced assessment of lysin properties during efforts to find, engineer, and develop candidates with true clinical potential.

MHCEpitopeEnergy, a Flexible Rosetta-Based Biotherapeutic Deimmunization Platform

As non-"self" macromolecules, biotherapeutics can trigger an immune response that can reduce drug efficacy, require patients to be taken off therapy, or even cause life-threatening reactions. To enable the flexible and facile design of protein biotherapeutics while reducing the prevalence of T-cell epitopes that drive immune recognition, we have integrated into the Rosetta protein design suite a new scoring term that allows design protocols to account for predicted or experimentally identified epitopes in the optimized objective function. This flexible scoring term can be used in any Rosetta design trajectory, can be targeted to specific regions of a protein, and can be readily extended to work with a variety of epitope predictors. By performing extensive design runs with varied design parameter choices for three case study proteins as well as a larger diverse benchmark, we show that the incorporation of this scoring term enables the effective exploration of an alternative, deimmunized sequence space to discover diverse proteins that are potentially highly deimmunized while retaining physical and chemical qualities similar to those yielded by equivalent nondeimmunizing sequence design protocols.

Building blocks and blueprints for bacterial autolysins

Bacteria utilize a wide variety of endogenous cell wall hydrolases, or autolysins, to remodel their cell walls during processes including cell division, biofilm formation, and programmed death. We here systematically investigate the composition of these enzymes in order to gain insights into their associated biological processes, potential ways to disrupt them via chemotherapeutics, and strategies by which they might be leveraged as recombinant antibacterial biotherapies. To do so, we developed LEDGOs (lytic enzyme domains grouped by organism), a pipeline to create and analyze databases of autolytic enzyme sequences, constituent domain annotations, and architectural patterns of multi-domain enzymes that integrate peptidoglycan binding and degrading functions. We applied LEDGOs to eight pathogenic bacteria, gram negatives Acinetobacter baumannii, Klebsiella pneumoniae, Neisseria gonorrhoeae, and Pseudomonas aeruginosa; and gram positives Clostridioides difficile, Enterococcus faecium, Staphylococcus aureus, and Streptococcus pneumoniae. Our analysis of the autolytic enzyme repertoires of these pathogens reveals commonalities and differences in their key domain building blocks and architectures, including correlations and preferred orders among domains in multi-domain enzymes, repetitions of homologous binding domains with potentially complementarity recognition modalities, and sequence similarity patterns indicative of potential divergence of functional specificity among related domains. We have further identified a variety of unannotated sequence regions within the lytic enzymes that may themselves contain new domains with important functions.

Protein Interaction Interface Region Prediction by Geometric Deep Learning

MOTIVATION

Protein-protein interactions drive wide-ranging molecular processes, and characterizing at the atomic level how proteins interact (beyond just the fact that they interact) can provide key insights into understanding and controlling this machinery. Unfortunately, experimental determination of three-dimensional protein complex structures remains difficult and does not scale to the increasingly large sets of proteins whose interactions are of interest. Computational methods are thus required to meet the demands of large-scale, high-throughput prediction of how proteins interact, but unfortunately both physical modeling and machine learning methods suffer from poor precision and/or recall.

RESULTS

In order to improve performance in predicting protein interaction interfaces, we leverage the best properties of both data- and physics-driven methods to develop a unified Geometric Deep Neural Network, "PInet" (Protein Interface Network). PInet consumes pairs of point clouds encoding the structures of two partner proteins, in order to predict their structural regions mediating interaction. To make such predictions, PInet learns and utilizes models capturing both geometrical and physicochemical molecular surface complementarity. In application to a set of benchmarks, PInet simultaneously predicts the interface regions on both interacting proteins, achieving performance equivalent to or even much better than the state-of-the-art predictor for each dataset. Furthermore, since PInet is based on joint segmentation of a representation of a protein surfaces, its predictions are meaningful in terms of the underlying physical complementarity driving molecular recognition.

AVAILABILITY

PInet scripts and models are available at https://github.com/FTD007/PInet.

SUPPLEMENTARY INFORMATION

Supplementary data are available at Bioinformatics online.

Learning context-aware structural representations to predict antigen and antibody binding interfaces

MOTIVATION

Understanding how antibodies specifically interact with their antigens can enable better drug and vaccine design, as well as provide insights into natural immunity. Experimental structural characterization can detail the 'ground truth' of antibody-antigen interactions, but computational methods are required to efficiently scale to large-scale studies. To increase prediction accuracy as well as to provide a means to gain new biological insights into these interactions, we have developed a unified deep learning-based framework to predict binding interfaces on both antibodies and antigens.

RESULTS

Our framework leverages three key aspects of antibody-antigen interactions to learn predictive structural representations: (i) since interfaces are formed from multiple residues in spatial proximity, we employ graph convolutions to aggregate properties across local regions in a protein; (ii) since interactions are specific between antibody-antigen pairs, we employ an attention layer to explicitly encode the context of the partner; (iii) since more data are available for general protein-protein interactions, we employ transfer learning to leverage this data as a prior for the specific case of antibody-antigen interactions. We show that this single framework achieves state-of-the-art performance at predicting binding interfaces on both antibodies and antigens, and that each of its three aspects drives additional improvement in the performance. We further show that the attention layer not only improves performance, but also provides a biologically interpretable perspective into the mode of interaction.

AVAILABILITY AND IMPLEMENTATION

The source code is freely available on github at https://github.com/vamships/PECAN.git.

A Chimeric Antigen Receptor That Binds to a Conserved Site on MICA

The NKG2D ligand MHC class I chain-related protein A (MICA) is expressed on many varieties of malignant cells but is absent from most normal tissues, and thus represents a potential target for chimeric Ag receptor (CAR) T cell-based therapeutics. However, there are more than 100 alleles of MICA, so the ability to target a conserved site is needed for a therapy to be used in most patients. In this study, we describe a fully human anti-MICA CAR created by fusing the single-chain fragment variable B2 to the full length DAP10 protein and the traditional CD3ζ signaling domain. Human T cells expressing the B2 CAR killed MICA-positive tumor cells, produced IFN-γ when in contact with MICA-positive tumor cells or plate-bound MICA protein, and inhibited PANC-1 growth in a mouse xenograft model. To localize B2's epitope on MICA, we used novel computational methods to model potential binding modes and to design mutational variants of MICA testing these hypotheses. Flow cytometry using a commercial anti-MICA/MICB Ab indicated that the variant proteins were expressed at high levels on transduced P815 cell lines. One variant protein (R38S/K40T/K57E) showed reduced staining with a B2-IgG1 fusion protein compared with controls and did not induce IFN-γ production by human T cells expressing the B2 CAR. These results show antitumor activity of MICA-specific CAR T cells and indicate an essential role for a conserved site in the exposed loop involving aa 38-57 of MICA. This study describes a novel MICA-specific CAR and discusses its potential use as a cancer therapeutic.

Combinatorial Resurfacing of Dengue Envelope Protein Domain III Antigens Selectively Ablates Epitopes Associated with Serotype-Specific or Infection-Enhancing Antibody Responses

Mutagenesis of surface-exposed residues, or "resurfacing", is a protein engineering strategy that can be utilized to disrupt antibody recognition or modulate the capacity of a protein to elicit antibody responses. We apply resurfacing to engineer Dengue virus envelope protein domain III (DENV DIII) antigens with the goal of focusing humoral recognition on epitopes of interest by selective ablation of irrelevant and undesired epitopes. Cross-reactive but non-neutralizing antibodies have the potential to enhance Dengue virus (DENV) infection by a process called antibody-dependent enhancement, thought to be associated with severe secondary heterotypic infection. Thus, a focus on epitopes associated with broadly neutralizing antibodies is important both for understanding human antibody responses against DENV and for the development of a successful DENV vaccine. To engineer DENV DIII antigens focusing on the AG strand epitope associated with broadly neutralizing antibody responses, we generated yeast surface display libraries of DENV2 DIII where the AB loop (associated with cross-reactive but non-neutralizing antibody responses) and FG loop (associated with serotype-specific antibody responses) were mutagenized to allow for all possible amino acid substitutions. Loop variants that maintained the AG strand epitope and simultaneously disrupted the AB and FG loop epitopes exhibited high and diverse mutational loads that were amenable to loop exchange and transplantation into a DENV4 DIII background. Thus, several loop variants fulfill this antigenicity criteria regardless of serotype context. The resulting resurfaced DIII antigens may be utilized as AG strand epitope-focusing probes or immunogen candidates.

Globally deimmunized lysostaphin evades human immune surveillance and enables highly efficacious repeat dosing

There is a critical need for novel therapies to treat methicillin-resistant Staphylococcus aureus (MRSA) and other drug-resistant pathogens, and lysins are among the vanguard of innovative antibiotics under development. Unfortunately, lysins' own microbial origins can elicit detrimental antidrug antibodies (ADAs) that undermine efficacy and threaten patient safety. To create an enhanced anti-MRSA lysin, a novel variant of lysostaphin was engineered by T cell epitope deletion. This "deimmunized" lysostaphin dampened human T cell activation, mitigated ADA responses in human HLA transgenic mice, and enabled safe and efficacious repeated dosing during a 6-week longitudinal infection study. Furthermore, the deimmunized lysostaphin evaded established anti-wild-type immunity, thereby providing significant anti-MRSA protection for animals that were immune experienced to the wild-type enzyme. Last, the enzyme synergized with daptomycin to clear a stringent model of MRSA endocarditis. By mitigating T cell-driven antidrug immunity, deimmunized lysostaphin may enable safe, repeated dosing to treat refractory MRSA infections.

Characterizing Epitope Binding Regions of Entire Antibody Panels by Combining Experimental and Computational Analysis of Antibody: Antigen Binding Competition

Vaccines and immunotherapies depend on the ability of antibodies to sensitively and specifically recognize particular antigens and specific epitopes on those antigens. As such, detailed characterization of antibody-antigen binding provides important information to guide development. Due to the time and expense required, high-resolution structural characterization techniques are typically used sparingly and late in a development process. Here, we show that antibody-antigen binding can be characterized early in a process for whole panels of antibodies by combining experimental and computational analyses of competition between monoclonal antibodies for binding to an antigen. Experimental "epitope binning" of monoclonal antibodies uses high-throughput surface plasmon resonance to reveal which antibodies compete, while a new complementary computational analysis that we call "dock binning" evaluates antibody-antigen docking models to identify why and where they might compete, in terms of possible binding sites on the antigen. Experimental and computational characterization of the identified antigenic hotspots then enables the refinement of the competitors and their associated epitope binding regions on the antigen. While not performed at atomic resolution, this approach allows for the group-level identification of functionally related monoclonal antibodies (i.e., communities) and identification of their general binding regions on the antigen. By leveraging extensive epitope characterization data that can be readily generated both experimentally and computationally, researchers can gain broad insights into the basis for antibody-antigen recognition in wide-ranging vaccine and immunotherapy discovery and development programs.

Computer-guided binding mode identification and affinity improvement of an LRR protein binder without structure determination

Precise binding mode identification and subsequent affinity improvement without structure determination remain a challenge in the development of therapeutic proteins. However, relevant experimental techniques are generally quite costly, and purely computational methods have been unreliable. Here, we show that integrated computational and experimental epitope localization followed by full-atom energy minimization can yield an accurate complex model structure which ultimately enables effective affinity improvement and redesign of binding specificity. As proof-of-concept, we used a leucine-rich repeat (LRR) protein binder, called a repebody (Rb), that specifically recognizes human IgG1 (hIgG1). We performed computationally-guided identification of the Rb:hIgG1 binding mode and leveraged the resulting model to reengineer the Rb so as to significantly increase its binding affinity for hIgG1 as well as redesign its specificity toward multiple IgGs from other species. Experimental structure determination verified that our Rb:hIgG1 model closely matched the co-crystal structure. Using a benchmark of other LRR protein complexes, we further demonstrated that the present approach may be broadly applicable to proteins undergoing relatively small conformational changes upon target binding.

Induction of cross-reactive HIV-1 specific antibody responses by engineered V1V2 immunogens with reduced conformational plasticity

Antibodies against the HIV-1 V1V2 loops were the only correlate of reduced infection risk in the RV144 vaccine trial, highlighting the V1V2 loops as promising targets for vaccine design. The V1V2 loops are structurally plastic, exhibiting either an α-helix-coil or β-strand conformation. V1V2-specific antibodies may thus recognize distinct conformations, and an antibody's conformational specificity can be an important determinant of breadth and function. Restricting V1V2 conformational plasticity in an immunogen may thus provide control over the conformational specificity and quality of a vaccine-elicited antibody response. Previously, we identified a V1V2 sequence variant (K155M) that results in enhanced recognition by cross-reactive antibodies recognizing the β-strand conformation. Here, we relate V1V2 antigenicity to immunogenicity by comparing the immunogenicity profiles of wildtype and K155M immunogens in two mouse models. In one model, immunization with gp70 V1V2 K155M but not wildtype elicited antibody responses that were cross-reactive to a panel of heterologous gp120 and gp140 antigens. In a second model, we compared the effect of K155M on immunogenicity in the context of gp70 V1V2, gD V1V2 and gp120, examining the effects of scaffold, epitope-focusing and immunization regimen. K155M variants, especially in the context of a gp120 immunogen, resulted in more robust, durable and cross-reactive antibody responses than wildtype immunogens. Restriction of the β-stranded V1V2 conformation in K155M immunogens may thus be associated with the induction of cross-reactive antibody responses thought to be required of a protective HIV-1 vaccine.

DisruPPI: structure-based computational redesign algorithm for protein binding disruption

Motivation

Disruption of protein–protein interactions can mitigate antibody recognition of therapeutic proteins, yield monomeric forms of oligomeric proteins, and elucidate signaling mechanisms, among other applications. While designing affinity-enhancing mutations remains generally quite challenging, both statistically and physically based computational methods can precisely identify affinity-reducing mutations. In order to leverage this ability to design variants of a target protein with disrupted interactions, we developed the DisruPPI protein design method (DISRUpting Protein–Protein Interactions) to optimize combinations of mutations simultaneously for both disruption and stability, so that incorporated disruptive mutations do not inadvertently affect the target protein adversely.

Results

Two existing methods for predicting mutational effects on binding, FoldX and INT5, were demonstrated to be quite precise in selecting disruptive mutations from the SKEMPI and AB-Bind databases of experimentally determined changes in binding free energy. DisruPPI was implemented to use an INT5-based disruption score integrated with an AMBER-based stability assessment and was applied to disrupt protein interactions in a set of different targets representing diverse applications. In retrospective evaluation with three different case studies, comparison of DisruPPI-designed variants to published experimental data showed that DisruPPI was able to identify more diverse interaction-disrupting and stability-preserving variants more efficiently and effectively than previous approaches. In prospective application to an interaction between enhanced green fluorescent protein (EGFP) and a nanobody, DisruPPI was used to design five EGFP variants, all of which were shown to have significantly reduced nanobody binding while maintaining function and thermostability. This demonstrates that DisruPPI may be readily utilized for effective removal of known epitopes of therapeutically relevant proteins.

Availability and implementation

DisruPPI is implemented in the EpiSweep package, freely available under an academic use license.

Supplementary information

Supplementary data are available at Bioinformatics online.

Pareto Optimization of Combinatorial Mutagenesis Libraries

In order to increase the hit rate of discovering diverse, beneficial protein variants via high-throughput screening, we have developed a computational method to optimize combinatorial mutagenesis libraries for overall enrichment in two distinct properties of interest. Given scoring functions for evaluating individual variants, POCoM (Pareto Optimal Combinatorial Mutagenesis) scores entire libraries in terms of averages over their constituent members, and designs optimal libraries as sets of mutations whose combinations make the best trade-offs between average scores. This represents the first general-purpose method to directly design combinatorial libraries for multiple objectives characterizing their constituent members. Despite being rigorous in mapping out the Pareto frontier, it is also very fast even for very large libraries (e.g., designing 30 mutation, billion-member libraries in only hours). We here instantiate POCoM with scores based on a target's protein structure and its homologs' sequences, enabling the design of libraries containing variants balancing these two important yet quite different types of information. We demonstrate POCoM's generality and power in case study applications to green fluorescent protein, cytochrome P450, and β-lactamase. Analysis of the POCoM library designs provides insights into the trade-offs between structure- and sequence-based scores, as well as the impacts of experimental constraints on library designs. POCoM libraries incorporate mutations that have previously been found favorable experimentally, while diversifying the contexts in which these mutations are situated and maintaining overall variant quality.

Balancing sensitivity and specificity in distinguishing TCR groups by CDR sequence similarity

Background

Repertoire sequencing is enabling deep explorations into the cellular immune response, including the characterization of commonalities and differences among T cell receptor (TCR) repertoires from different individuals, pathologies, and antigen specificities. In seeking to understand the generality of patterns observed in different groups of TCRs, it is necessary to balance how well each pattern represents the diversity among TCRs from one group (sensitivity) vs. how many TCRs from other groups it also represents (specificity). The variable complementarity determining regions (CDRs), particularly the third CDRs (CDR3s) interact with major histocompatibility complex (MHC)-presented epitopes from putative antigens, and thus encode the determinants of recognition.

Results

We here systematically characterize the predictive power that can be obtained from CDR3 sequences, using representative, readily interpretable methods for evaluating CDR sequence similarity and then clustering and classifying sequences based on similarity. An initial analysis of CDR3s of known structure, clustered by structural similarity, helps calibrate the limits of sequence diversity among CDRs that might have a common mode of interaction with presented epitopes. Subsequent analyses demonstrate that this same range of sequence similarity strikes a favorable specificity/sensitivity balance in distinguishing twins from non-twins based on overall CDR3 repertoires, classifying CDR3 repertoires by antigen specificity, and distinguishing general pathologies.

Conclusion

We conclude that within a fairly broad range of sequence similarity, matching CDR3 sequences are likely to share specificities.

Electronic supplementary material

The online version of this article (10.1186/s12859-019-2864-8) contains supplementary material, which is available to authorized users.

Antibody Fab-Fc properties outperform titer in predictive models of SIV vaccine-induced protection

Characterizing the antigen-binding and innate immune-recruiting properties of the humoral response offers the chance to obtain deeper insights into mechanisms of protection than revealed by measuring only overall antibody titer. Here, a high-throughput, multiplexed Fab-Fc Array was employed to profile rhesus macaques vaccinated with a gp120-CD4 fusion protein in combination with different genetically encoded adjuvants, and subsequently subjected to multiple heterologous simian immunodeficiency virus (SIV) challenges. Systems analyses modeling protection and adjuvant differences using Fab-Fc Array measurements revealed a set of correlates yielding strong and robust predictive performance, while models based on measurements of response magnitude alone exhibited significantly inferior performance. At the same time, rendering Fab-Fc measurements mathematically independent of titer had relatively little impact on predictive performance. Similar analyses for a distinct SIV vaccine study also showed that Fab-Fc measurements performed significantly better than titer. These results suggest that predictive modeling with measurements of antibody properties can provide detailed correlates with robust predictive power, suggest directions for vaccine improvement, and potentially enable discovery of mechanistic associations.

Towards conformational fidelity of a quaternary HIV-1 epitope: computational design and directed evolution of a minimal V1V2 antigen

Structure-based approaches to antigen design utilize insights from antibody (Ab):antigen interactions and a refined understanding of protective Ab responses to engineer novel antigens presenting epitopes with conformations relevant to eliciting or discovering protective humoral responses. For human immunodeficiency virus-1 (HIV-1), one model of protection is provided by broadly neutralizing Abs (bnAbs) against epitopes present in the closed prefusion trimeric conformation of HIV-1 envelope glycoprotein, such as the variable loops 1-2 (V1V2) apex. Here, computational design and directed evolution yielded a novel V1V2 sequence variant with potential utility for inclusion in an immunogen for eliciting bnAbs, or as an epitope probe for their detection. The computational design goal was to engineer a minimal single-chain antigen with three copies of the V1V2 loops to support maintenance of closed prefusion V1V2 trimeric conformation and presentation of bnAb epitopes. Via directed evolution of this computationally designed single-chain antigen, we isolated a V1V2 sequence variant that in monomeric form exhibited preferential recognition by quaternary-preferring and conformation-dependent mAbs. Structural context and transferability of this phenotype to V1V2 sequences from all strains of HIV-1 tested suggest a conformation-stabilizing effect. This example demonstrates the potential utility of computational design and directed evolution-based protein engineering strategies to develop minimal, conformation-stabilized epitope-specific antigens.

High-resolution definition of humoral immune response correlates of effective immunity against HIV

Defining correlates of immunity by comprehensively interrogating the extensive biological diversity in naturally or experimentally protected subjects may provide insights critical for guiding the development of effective vaccines and antibody-based therapies. We report advances in a humoral immunoprofiling approach and its application to elucidate hallmarks of effective HIV-1 viral control. Systematic serological analysis for a cohort of HIV-infected subjects with varying viral control was conducted using both a high-resolution, high-throughput biophysical antibody profiling approach, providing unbiased dissection of the humoral response, along with functional antibody assays, characterizing antibody-directed effector functions such as complement fixation and phagocytosis that are central to protective immunity. Profiles of subjects with varying viral control were computationally analyzed and modeled in order to deconvolute relationships among IgG Fab properties, Fc characteristics, and effector functions and to identify humoral correlates of potent antiviral antibody-directed effector activity and effective viral suppression. The resulting models reveal multifaceted and coordinated contributions of polyclonal antibodies to diverse antiviral responses, and suggest key biophysical features predictive of viral control.

Fine epitope signature of antibody neutralization breadth at the HIV-1 envelope CD4-binding site

Major advances in donor identification, antigen probe design, and experimental methods to clone pathogen-specific antibodies have led to an exponential growth in the number of newly characterized broadly neutralizing antibodies (bnAbs) that recognize the HIV-1 envelope glycoprotein. Characterization of these bnAbs has defined new epitopes and novel modes of recognition that can result in potent neutralization of HIV-1. However, the translation of envelope recognition profiles in biophysical assays into an understanding of in vivo activity has lagged behind, and identification of subjects and mAbs with potent antiviral activity has remained reliant on empirical evaluation of neutralization potency and breadth. To begin to address this discrepancy between recombinant protein recognition and virus neutralization, we studied the fine epitope specificity of a panel of CD4-binding site (CD4bs) antibodies to define the molecular recognition features of functionally potent humoral responses targeting the HIV-1 envelope site bound by CD4. Whereas previous studies have used neutralization data and machine-learning methods to provide epitope maps, here, this approach was reversed, demonstrating that simple binding assays of fine epitope specificity can prospectively identify broadly neutralizing CD4bs-specific mAbs. Building on this result, we show that epitope mapping and prediction of neutralization breadth can also be accomplished in the assessment of polyclonal serum responses. Thus, this study identifies a set of CD4bs bnAb signature amino acid residues and demonstrates that sensitivity to mutations at signature positions is sufficient to predict neutralization breadth of polyclonal sera with a high degree of accuracy across cohorts and across clades.

Computationally-driven identification of antibody epitopes

Understanding where antibodies recognize antigens can help define mechanisms of action and provide insights into progression of immune responses. We investigate the extent to which information about binding specificity implicitly encoded in amino acid sequence can be leveraged to identify antibody epitopes. In computationally-driven epitope localization, possible antibody–antigen binding modes are modeled, and targeted panels of antigen variants are designed to experimentally test these hypotheses. Prospective application of this approach to two antibodies enabled epitope localization using five or fewer variants per antibody, or alternatively, a six-variant panel for both simultaneously. Retrospective analysis of a variety of antibodies and antigens demonstrated an almost 90% success rate with an average of three antigen variants, further supporting the observation that the combination of computational modeling and protein design can reveal key determinants of antibody–antigen binding and enable efficient studies of collections of antibodies identified from polyclonal samples or engineered libraries.

Multiplexed Fc array for evaluation of antigen-specific antibody effector profiles

Antibodies are widely considered to be a frequent primary and often mechanistic correlate of protection of approved vaccines; thus evaluating the antibody response is of critical importance in attempting to understand and predict the efficacy of novel vaccine candidates. Historically, antibody responses have been analyzed by determining the titer of the humoral response using measurements such as an ELISA, neutralization, or agglutination assays. In the simplest case, sufficiently high titers of antibody against vaccine antigen(s) are sufficient to predict protection. However, antibody titer provides only a partial measure of antibody function, which is dependent on both the variable region (Fv) to bind the antigen target, and the constant region (Fc) to elicit an effector response from the innate arm of the immune system. In the case of some diseases, such as HIV, for which an effective vaccine has proven elusive, antibody effector function has been shown to be an important driver of monoclonal antibody therapy outcomes, of viral control in infected patients, and of vaccine-mediated protection in preclinical and clinical studies. We sought to establish a platform for the evaluation of the Fc domain characteristics of antigen-specific antibodies present in polyclonal samples in order to better develop insights into Fc receptor-mediated antibody effector activity, more fully understand how antibody responses may differ in association with disease progression and between subject groups, and differentiate protective from non-protective responses. To this end we have developed a high throughput biophysical platform capable of simultaneously evaluating many dimensions of the antibody effector response.

•

High-throughput array-based characterization platform for polyclonal antibodies.

•

Development of a biophysical proxy for antibody effector function.

•

Antigen and Fc receptor recognition characteristics are captured.

•

Enables systematic serologic studies of NHP and human antibody samples.

Diversity of Antiviral IgG Effector Activities Observed in HIV-Infected and Vaccinated Subjects

Diverse Ab effector functions mediated by the Fc domain have been commonly associated with reduced risk of infection in a growing number of nonhuman primate and human clinical studies. This study evaluated the anti-HIV Ab effector activities in polyclonal serum samples from HIV-infected donors, VAX004 vaccine recipients, and healthy HIV-negative subjects using a variety of primary and cell line-based assays, including Ab-dependent cellular cytotoxicity (ADCC), Ab-dependent cell-mediated viral inhibition, and Ab-dependent cellular phagocytosis. Additional assay characterization was performed with a panel of Fc-engineered variants of mAb b12. The goal of this study was to characterize different effector functions in the study samples and identify assays that might most comprehensively and dependably capture Fc-mediated Ab functions mediated by different effector cell types and against different viral targets. Deployment of such assays may facilitate assessment of functionally unique humoral responses and contribute to identification of correlates of protection with potential mechanistic significance in future HIV vaccine studies. Multivariate and correlative comparisons identified a set of Ab-dependent cell-mediated viral inhibition and phagocytosis assays that captured different Ab activities and were distinct from a group of ADCC assays that showed a more similar response profile across polyclonal serum samples. The activities of a panel of b12 monoclonal Fc variants further identified distinctions among the ADCC assays. These results reveal the natural diversity of Fc-mediated Ab effector responses among vaccine recipients in the VAX004 trial and in HIV-infected subjects, and they point to the potential importance of polyfunctional Ab responses.

Computationally driven antibody engineering enables simultaneous humanization and thermostabilization

Humanization reduces the immunogenicity risk of therapeutic antibodies of non-human origin. Thermostabilization can be critical for clinical development and application of therapeutic antibodies. Here, we show that the computational antibody redesign method Computationally Driven Antibody Humanization (CoDAH) enables these two goals to be accomplished simultaneously and seamlessly. A panel of CoDAH designs for the murine parent of cetuximab, a chimeric anti-EGFR antibody, exhibited both substantially improved thermostabilities and substantially higher levels of humanness, while retaining binding activity near the parental level. The consistently high quality of the turnkey CoDAH designs, over a whole panel of variants, suggests that the computationally directed approach encapsulates key determinants of antibody structure and function.

Design and engineering of deimmunized biotherapeutics

Therapeutic proteins are powerful next-generation drugs able to effectively treat diverse and devastating diseases, but the development and use of biotherapeutics entails unique challenges and risks. In particular, protein drugs are subject to immune surveillance in the human body, and ensuing antidrug immune responses can cause a wide range of problems including altered pharmacokinetics, loss of efficacy, and even life-threating complications. Here we review recent progress in technologies for engineering deimmunized biotherapeutics, placing particular emphasis on deletion of immunogenic antibody and T cell epitopes via experimentally or computationally guided mutagenesis.

Depletion of T cell epitopes in lysostaphin mitigates anti-drug antibody response and enhances antibacterial efficacy in vivo

The enzyme lysostaphin possesses potent anti-staphylococcal activity and represents a promising antibacterial drug candidate; however, its immunogenicity poses a barrier to clinical translation. Here, structure-based biomolecular design enabled widespread depletion of lysostaphin DRB1(∗)0401 restricted T cell epitopes, and resulting deimmunized variants exhibited striking reductions in anti-drug antibody responses upon administration to humanized HLA-transgenic mice. This reduced immunogenicity translated into improved efficacy in the form of protection against repeated challenges with methicillin-resistant Staphylococcus aureus (MRSA). In contrast, while wild-type lysostaphin was efficacious against the initial MRSA infection, it failed to clear subsequent bacterial challenges that were coincident with escalating anti-drug antibody titers. These results extend the existing deimmunization literature, in which reduced immunogenicity and retained efficacy are assessed independently of each other. By correlating in vivo efficacy with longitudinal measures of anti-drug antibody development, we provide the first direct evidence that T cell epitope depletion manifests enhanced biotherapeutic efficacy.

Polyfunctional HIV-Specific Antibody Responses Are Associated with Spontaneous HIV Control

Elite controllers (ECs) represent a unique model of a functional cure for HIV-1 infection as these individuals develop HIV-specific immunity able to persistently suppress viremia. Because accumulating evidence suggests that HIV controllers generate antibodies with enhanced capacity to drive antibody-dependent cellular cytotoxicity (ADCC) that may contribute to viral containment, we profiled an array of extra-neutralizing antibody effector functions across HIV-infected populations with varying degrees of viral control to define the characteristics of antibodies associated with spontaneous control. While neither the overall magnitude of antibody titer nor individual effector functions were increased in ECs, a more functionally coordinated innate immune-recruiting response was observed. Specifically, ECs demonstrated polyfunctional humoral immune responses able to coordinately recruit ADCC, other NK functions, monocyte and neutrophil phagocytosis, and complement. This functionally coordinated response was associated with qualitatively superior IgG3/IgG1 responses, whereas HIV-specific IgG2/IgG4 responses, prevalent among viremic subjects, were associated with poorer overall antibody activity. Rather than linking viral control to any single activity, this study highlights the critical nature of functionally coordinated antibodies in HIV control and associates this polyfunctionality with preferential induction of potent antibody subclasses, supporting coordinated antibody activity as a goal in strategies directed at an HIV-1 functional cure.

Immune camouflage: relevance to vaccines and human immunology

High strain sequence variability, interference with innate immune mechanisms, and epitope deletion are all examples of strategies that pathogens have evolved to subvert host defenses. To this list we would add another strategy: immune camouflage. Pathogens whose epitope sequences are cross-conserved with multiple human proteins at the TCR-facing residues may be exploiting “ignorance and tolerance," which are mechanisms by which mature T cells avoid immune responses to self-antigens. By adopting amino acid configurations that may be recognized by autologous regulatory T cells, pathogens may be actively suppressing protective immunity. Using the new JanusMatrix TCR-homology-mapping tool, we have identified several such ‘camouflaged’ tolerizing epitopes that are present in the viral genomes of pathogens such as emerging H7N9 influenza. Thus in addition to the overall low number of T helper epitopes that is present in H7 hemaglutinin (as described previously, see http://dx.doi.org/10.4161/hv.24939), the presence of such tolerizing epitopes in H7N9 could explain why, in recent vaccine trials, whole H7N9-HA was poorly immunogenic and associated with low seroconversion rates (see http://dx.doi.org/10.4161/hv.28135). In this commentary, we provide an overview of the immunoinformatics process leading to the discovery of tolerizing epitopes in pathogen genomic sequences, provide a brief summary of laboratory data that validates the discovery, and point the way forward. Removal of viral, bacterial and parasite tolerizing epitopes may permit researchers to develop more effective vaccines and immunotherapeutics in the future.

Hit-and-run, hit-and-stay, and commensal bacteria present different peptide content when viewed from the perspective of the T cell

Different types of bacteria face different pressures from the immune system, with those that persist ("hit-and-stay") potentially having to adapt more in order to escape than those prone to short-lived infection ("hit-and-run"), and with commensal bacteria potentially different from both due to additional physical mechanisms for avoiding immune detection. The Janus Immunogenicity Score (JIS) was recently developed to assess the likelihood of T cell recognition of an antigen, using an analysis that considers both binding of a peptide within the antigen by major histocompatability complex (MHC) and recognition of the peptide:MHC complex by cognate T cell receptor (TCR). This score was shown to be predictive of T effector vs. T regulatory or null responses in experimental data, as well as to distinguish viruses representative of the hit-and-stay vs. hit-and-run phenotypes. Here, JIS-based analyses were conducted in order to characterize the extent to which the pressure to avoid T cell recognition is manifested in genomic differences among representative hit-and-run, hit-and-stay, and commensal bacteria. Overall, extracellular proteins were found to have different JIS profiles from cytoplasmic ones. Contrasting the bacterial groups, extracellular proteins were shown to be quite different across the groups, much more so than intracellular proteins. The differences were evident even at the level of corresponding peptides in homologous protein pairs from hit-and-run and hit-and-stay bacteria. The multi-level analysis of patterns of immunogenicity across different groups of bacteria provides a new way to approach questions of bacterial immune camouflage or escape, as well as to approach the selection and optimization of candidates for vaccine design.

Development and validation of an epitope prediction tool for swine (PigMatrix) based on the pocket profile method

Background

T cell epitope prediction tools and associated vaccine design algorithms have accelerated the development of vaccines for humans. Predictive tools for swine and other food animals are not as well developed, primarily because the data required to develop the tools are lacking. Here, we overcome a lack of T cell epitope data to construct swine epitope predictors by systematically leveraging available human information. Applying the “pocket profile method”, we use sequence and structural similarities in the binding pockets of human and swine major histocompatibility complex proteins to infer Swine Leukocyte Antigen (SLA) peptide binding preferences.

We developed epitope-prediction matrices (PigMatrices), for three SLA class I alleles (SLA-1*0401, 2*0401 and 3*0401) and one class II allele (SLA-DRB1*0201), based on the binding preferences of the best-matched Human Leukocyte Antigen (HLA) pocket for each SLA pocket. The contact residues involved in the binding pockets were defined for class I based on crystal structures of either SLA (SLA-specific contacts, Ssc) or HLA supertype alleles (HLA contacts, Hc); for class II, only Hc was possible. Different substitution matrices were evaluated (PAM and BLOSUM) for scoring pocket similarity and identifying the best human match. The accuracy of the PigMatrices was compared to available online swine epitope prediction tools such as PickPocket and NetMHCpan.

Results

PigMatrices that used Ssc to define the pocket sequences and PAM30 to score pocket similarity demonstrated the best predictive performance and were able to accurately separate binders from random peptides. For SLA-1*0401 and 2*0401, PigMatrix achieved area under the receiver operating characteristic curves (AUC) of 0.78 and 0.73, respectively, which were equivalent or better than PickPocket (0.76 and 0.54) and NetMHCpan version 2.4 (0.41 and 0.51) and version 2.8 (0.72 and 0.71). In addition, we developed the first predictive SLA class II matrix, obtaining an AUC of 0.73 for existing SLA-DRB1*0201 epitopes. Notably, PigMatrix achieved this level of predictive power without training on SLA binding data.

Conclusion

Overall, the pocket profile method combined with binding preferences from HLA binding data shows significant promise for developing T cell epitope prediction tools for pigs. When combined with existing vaccine design algorithms, PigMatrix will be useful for developing genome-derived vaccines for a range of pig pathogens for which no effective vaccines currently exist (e.g. porcine reproductive and respiratory syndrome, influenza and porcine epidemic diarrhea).

Electronic supplementary material

The online version of this article (doi:10.1186/s12859-015-0724-8) contains supplementary material, which is available to authorized users.

Antibody humanization by structure-based computational protein design

Antibodies derived from non-human sources must be modified for therapeutic use so as to mitigate undesirable immune responses. While complementarity-determining region (CDR) grafting-based humanization techniques have been successfully applied in many cases, it remains challenging to maintain the desired stability and antigen binding affinity upon grafting. We developed an alternative humanization approach called CoDAH ("Computationally-Driven Antibody Humanization") in which computational protein design methods directly select sets of amino acids to incorporate from human germline sequences to increase humanness while maintaining structural stability. Retrospective studies show that CoDAH is able to identify variants deemed beneficial according to both humanness and structural stability criteria, even for targets lacking crystal structures. Prospective application to TZ47, a murine anti-human B7H6 antibody, demonstrates the approach. Four diverse humanized variants were designed, and all possible unique VH/VL combinations were produced as full-length IgG1 antibodies. Soluble and cell surface expressed antigen binding assays showed that 75% (6 of 8) of the computationally designed VH/VL variants were successfully expressed and competed with the murine TZ47 for binding to B7H6 antigen. Furthermore, 4 of the 6 bound with an estimated KD within an order of magnitude of the original TZ47 antibody. In contrast, a traditional CDR-grafted variant could not be expressed. These results suggest that the computational protein design approach described here can be used to efficiently generate functional humanized antibodies and provide humanized templates for further affinity maturation.

Protein deimmunization via structure-based design enables efficient epitope deletion at high mutational loads

Anti-drug immune responses are a unique risk factor for biotherapeutics, and undesired immunogenicity can alter pharmacokinetics, compromise drug efficacy, and in some cases even threaten patient safety. To fully capitalize on the promise of biotherapeutics, more efficient and generally applicable protein deimmunization tools are needed. Mutagenic deletion of a protein's T cell epitopes is one powerful strategy to engineer immunotolerance, but deimmunizing mutations must maintain protein structure and function. Here, EpiSweep, a structure-based protein design and deimmunization algorithm, has been used to produce a panel of seven beta-lactamase drug candidates having 27-47% reductions in predicted epitope content. Despite bearing eight mutations each, all seven engineered enzymes maintained good stability and activity. At the same time, the variants exhibited dramatically reduced interaction with human class II major histocompatibility complex proteins, key regulators of anti-drug immune responses. When compared to 8-mutation designs generated with a sequence-based deimmunization algorithm, the structure-based designs retained greater thermostability and possessed fewer high affinity epitopes, the dominant drivers of anti-biotherapeutic immune responses. These experimental results validate the first structure-based deimmunization algorithm capable of mapping optimal biotherapeutic design space. By designing optimal mutations that reduce immunogenic potential while imparting favorable intramolecular interactions, broadly distributed epitopes may be simultaneously targeted using high mutational loads.

Structure-based redesign of lysostaphin yields potent antistaphylococcal enzymes that evade immune cell surveillance

Staphylococcus aureus infections exert a tremendous burden on the health-care system, and the threat of drug-resistant strains continues to grow. The bacteriolytic enzyme lysostaphin is a potent antistaphylococcal agent with proven efficacy against both drug-sensitive and drug-resistant strains; however, the enzyme's own bacterial origins cause undesirable immunogenicity and pose a barrier to clinical translation. Here, we deimmunized lysostaphin using a computationally guided process that optimizes sets of mutations to delete immunogenic T cell epitopes without disrupting protein function. In vitro analyses showed the methods to be both efficient and effective, producing seven different deimmunized designs exhibiting high function and reduced immunogenic potential. Two deimmunized candidates elicited greatly suppressed proliferative responses in splenocytes from humanized mice, while at the same time the variants maintained wild-type efficacy in a staphylococcal pneumonia model. Overall, the deimmunized enzymes represent promising leads in the battle against S. aureus.

Learning sequence determinants of protein:protein interaction specificity with sparse graphical models

In studying the strength and specificity of interaction between members of two protein families, key questions center on which pairs of possible partners actually interact, how well they interact, and why they interact while others do not. The advent of large-scale experimental studies of interactions between members of a target family and a diverse set of possible interaction partners offers the opportunity to address these questions. We develop here a method, DgSpi (data-driven graphical models of specificity in protein:protein interactions), for learning and using graphical models that explicitly represent the amino acid basis for interaction specificity (why) and extend earlier classification-oriented approaches (which) to predict the ΔG of binding (how well). We demonstrate the effectiveness of our approach in analyzing and predicting interactions between a set of 82 PDZ recognition modules against a panel of 217 possible peptide partners, based on data from MacBeath and colleagues. Our predicted ΔG values are highly predictive of the experimentally measured ones, reaching correlation coefficients of 0.69 in 10-fold cross-validation and 0.63 in leave-one-PDZ-out cross-validation. Furthermore, the model serves as a compact representation of amino acid constraints underlying the interactions, enabling protein-level ΔG predictions to be naturally understood in terms of residue-level constraints. Finally, the model DgSpi readily enables the design of new interacting partners, and we demonstrate that designed ligands are novel and diverse.

Machine learning methods enable predictive modeling of antibody feature:function relationships in RV144 vaccinees

The adaptive immune response to vaccination or infection can lead to the production of specific antibodies to neutralize the pathogen or recruit innate immune effector cells for help. The non-neutralizing role of antibodies in stimulating effector cell responses may have been a key mechanism of the protection observed in the RV144 HIV vaccine trial. In an extensive investigation of a rich set of data collected from RV144 vaccine recipients, we here employ machine learning methods to identify and model associations between antibody features (IgG subclass and antigen specificity) and effector function activities (antibody dependent cellular phagocytosis, cellular cytotoxicity, and cytokine release). We demonstrate via cross-validation that classification and regression approaches can effectively use the antibody features to robustly predict qualitative and quantitative functional outcomes. This integration of antibody feature and function data within a machine learning framework provides a new, objective approach to discovering and assessing multivariate immune correlates.

Structure-based design of combinatorial mutagenesis libraries

The development of protein variants with improved properties (thermostability, binding affinity, catalytic activity, etc.) has greatly benefited from the application of high-throughput screens evaluating large, diverse combinatorial libraries. At the same time, since only a very limited portion of sequence space can be experimentally constructed and tested, an attractive possibility is to use computational protein design to focus libraries on a productive portion of the space. We present a general-purpose method, called "Structure-based Optimization of Combinatorial Mutagenesis" (SOCoM), which can optimize arbitrarily large combinatorial mutagenesis libraries directly based on structural energies of their constituents. SOCoM chooses both positions and substitutions, employing a combinatorial optimization framework based on library-averaged energy potentials in order to avoid explicitly modeling every variant in every possible library. In case study applications to green fluorescent protein, β-lactamase, and lipase A, SOCoM optimizes relatively small, focused libraries whose variants achieve energies comparable to or better than previous library design efforts, as well as larger libraries (previously not designable by structure-based methods) whose variants cover greater diversity while still maintaining substantially better energies than would be achieved by representative random library approaches. By allowing the creation of large-scale combinatorial libraries based on structural calculations, SOCoM promises to increase the scope of applicability of computational protein design and improve the hit rate of discovering beneficial variants. While designs presented here focus on variant stability (predicted by total energy), SOCoM can readily incorporate other structure-based assessments, such as the energy gap between alternative conformational or bound states.

Mapping the Pareto optimal design space for a functionally deimmunized biotherapeutic candidate

The immunogenicity of biotherapeutics can bottleneck development pipelines and poses a barrier to widespread clinical application. As a result, there is a growing need for improved deimmunization technologies. We have recently described algorithms that simultaneously optimize proteins for both reduced T cell epitope content and high-level function. In silico analysis of this dual objective design space reveals that there is no single global optimum with respect to protein deimmunization. Instead, mutagenic epitope deletion yields a spectrum of designs that exhibit tradeoffs between immunogenic potential and molecular function. The leading edge of this design space is the Pareto frontier, i.e. the undominated variants for which no other single design exhibits better performance in both criteria. Here, the Pareto frontier of a therapeutic enzyme has been designed, constructed, and evaluated experimentally. Various measures of protein performance were found to map a functional sequence space that correlated well with computational predictions. These results represent the first systematic and rigorous assessment of the functional penalty that must be paid for pursuing progressively more deimmunized biotherapeutic candidates. Given this capacity to rapidly assess and design for tradeoffs between protein immunogenicity and functionality, these algorithms may prove useful in augmenting, accelerating, and de-risking experimental deimmunization efforts.

Protein therapeutics have created a revolution in disease therapy, providing improved outcomes for prevalent illnesses and conditions while at the same time yielding treatments for diseases that were previously intractable. However, this powerful class of drugs is subject to their own unique challenges and risk factors. In particular, the biological origins of therapeutic proteins predispose them towards eliciting a detrimental immune response from the patient's own body. Therefore, fully capitalizing on the medicinal reservoir of natural and engineered proteins will require efficient, effective, and broadly applicable deimmunization technologies. We have developed deimmunization algorithms that simultaneously optimize therapeutic candidates for both low immunogenicity and high-level activity and stability. Here, we combine computational modeling and experimental analysis to show that the process of protein deimmunization manifests inherent tradeoffs between immunogenic potential and biomolecular function. Our experimental results demonstrate that dual objective optimization allows us to assess and design for these tradeoffs, thereby enabling facile construction of deimmunized variants that span a broad range of immunogenicity and functionality performance parameters. Thus, we can rapidly map the design space for deimmunized drug candidates, and we can use this information to guide selection of engineered proteins that are most likely to meet performance benchmarks for a given clinical application.

HCV epitope, homologous to multiple human protein sequences, induces a regulatory T cell response in infected patients

BACKGROUND & AIMS

Spontaneous resolution of hepatitis C virus (HCV) infection depends upon a broad T cell response to multiple viral epitopes. However, most patients fail to clear infections spontaneously and develop chronic disease. The elevated number and function of CD3(+)CD4(+)CD25(+)FoxP3(+) regulatory T cells (T(reg)) in HCV-infected patients suggest a role of Treg cells in impaired viral clearance. The factors contributing to increased Treg cell activity in chronic hepatitis C cases remain to be delineated.

METHODS

Immunoinformatics tools were used to predict promiscuous, highly-conserved HLA-DRB1-restricted immunogenic consensus sequences (ICS), each composed of multiple T cell epitopes. These sequences were synthesized and added to cultures of peripheral blood mononuclear cells (PBMCs), derived from patients who resolved HCV infection spontaneously, patients with persistent infection, and non-infected individuals. The cells were collected and following 5days incubation, quantified and characterized by flow cytometry.

RESULTS

One immunogenic consensus sequence (ICS), HCV_G1_p7_794, induced a marked increase in Treg cells in PBMC cultures derived from infected patients, but not in patients who spontaneously cleared HCV or in non-infected individuals. An analogous human peptide (p7_794), on the other hand, induced a significant increase in Treg cells among PBMCs derived from both HCV-infected and non-infected individuals. JanusMatrix analyses determined that HCV_G1_p7_794 is comprised of Treg cell epitopes that exhibit extensive cross-reactivity with the human proteome.

CONCLUSIONS

A virus-encoded peptide (HCV_G1_p7_794) with extensive human homology activates cross-reactive CD3(+)CD4(+)CD25(+)FoxP3(+) natural Treg cells, which potentially contributes to immunosuppression and to the development of chronic hepatitis C.

Polyfunctional Fc-effector profiles mediated by IgG subclass selection distinguish RV144 and VAX003 vaccines

The human phase 2B RV144 ALVAC-HIV vCP1521/AIDSVAX B/E vaccine trial, held in Thailand, resulted in an estimated 31.2% efficacy against HIV infection. By contrast, vaccination with VAX003 (consisting of only AIDSVAX B/E) was not protective. Because protection within RV144 was observed in the absence of neutralizing antibody activity or cytotoxic T cell responses, we speculated that the specificity or qualitative differences in Fc-effector profiles of nonneutralizing antibodies may have accounted for the efficacy differences observed between the two trials. We show that the RV144 regimen elicited nonneutralizing antibodies with highly coordinated Fc-mediated effector responses through the selective induction of highly functional immunoglobulin G3 (IgG3). By contrast, VAX003 elicited monofunctional antibody responses influenced by IgG4 selection, which was promoted by repeated AIDSVAX B/E protein boosts. Moreover, only RV144 induced IgG1 and IgG3 antibodies targeting the crown of the HIV envelope V2 loop, albeit with limited coverage of breakthrough viral sequences. These data suggest that subclass selection differences associated with coordinated humoral functional responses targeting strain-specific protective V2 loop epitopes may underlie differences in vaccine efficacy observed between these two vaccine trials.

Smarter vaccine design will circumvent regulatory T cell-mediated evasion in chronic HIV and HCV infection

Despite years of research, vaccines against HIV and HCV are not yet available, due largely to effective viral immunoevasive mechanisms. A novel escape mechanism observed in viruses that cause chronic infection is suppression of viral-specific effector CD4(+) and CD8(+) T cells by stimulating regulatory T cells (Tregs) educated on host sequences during tolerance induction. Viral class II MHC epitopes that share a T cell receptor (TCR)-face with host epitopes may activate Tregs capable of suppressing protective responses. We designed an immunoinformatic algorithm, JanusMatrix, to identify such epitopes and discovered that among human-host viruses, chronic viruses appear more human-like than viruses that cause acute infection. Furthermore, an HCV epitope that activates Tregs in chronically infected patients, but not clearers, shares a TCR-face with numerous human sequences. To boost weak CD4(+) T cell responses associated with persistent infection, vaccines for HIV and HCV must circumvent potential Treg activation that can handicap efficacy. Epitope-driven approaches to vaccine design that involve careful consideration of the T cell subsets primed during immunization will advance HIV and HCV vaccine development.

CHOPPI: a web tool for the analysis of immunogenicity risk from host cell proteins in CHO-based protein production

Despite high quality standards and continual process improvements in manufacturing, host cell protein (HCP) process impurities remain a substantial risk for biological products. Even at low levels, residual HCPs can induce a detrimental immune response compromising the safety and efficacy of a biologic. Consequently, advanced-stage clinical trials have been cancelled due to the identification of antibodies against HCPs. To enable earlier and rapid assessment of the risks in Chinese Hamster Ovary (CHO)-based protein production of residual CHO protein impurities (CHOPs), we have developed a web tool called CHOPPI, for CHO Protein Predicted Immunogenicity. CHOPPI integrates information regarding the possible presence of CHOPs (expression and secretion) with characterizations of their immunogenicity (T cell epitope count and density, and relative conservation with human counterparts). CHOPPI can generate a report for a specified CHO protein (e.g., identified from proteomics or immunoassays) or characterize an entire specified subset of the CHO genome (e.g., filtered based on confidence in transcription and similarity to human proteins). The ability to analyze potential CHOPs at a genomic scale provides a baseline to evaluate relative risk. We show here that CHOPPI can identify clear differences in immunogenicity risk among previously validated CHOPs, as well as identify additional “risky” CHO proteins that may be expressed during production and induce a detrimental immune response upon delivery. We conclude that CHOPPI is a powerful tool that provides a valuable computational complement to existing experimental approaches for CHOP risk assessment and can focus experimental efforts in the most important directions. Biotechnol. Bioeng. 2014;111: 2170–2182.

Stoichiometries and affinities of interacting proteins from concentration series of solution scattering data: decomposition by least squares and quadratic optimization

In studying interacting proteins, complementary insights are provided by analyzing both the association model (the stoichiometry and affinity constants of the intermediate and final complexes) and the quaternary structure of the resulting complexes. Many current methods for analyzing protein interactions either give a binary answer to the question of association and no information about quaternary structure or at best provide only part of the complete picture. Presented here is a method to extract both types of information from X-ray or neutron scattering data for a series of equilibrium mixtures containing the initial components at different concentrations. The method determines the association pathway and constants, along with the scattering curves of the individual members of the mixture, so as to best explain the scattering data for the mixtures. The derived curves then enable reconstruction of the intermediate and final complexes. Using simulated solution scattering data for four hetero-oligomeric complexes with different structures, molecular weights and association models, it is demonstrated that this method accurately determines the simulated association model and scattering profiles for the initial components and complexes. Recognizing that experimental mixtures contain static contaminants and nonspecific complexes with the lowest affinities (inter-particle interference) as well as the desired specific complex(es), a new analytical method is also employed to extend this approach to evaluating the association models and scattering curves in the presence of static contaminants, testing both a nonparticipating monomer and a large homo-oligomeric aggregate. It is demonstrated that the method is robust to both random noise and systematic noise from such contaminants, and the treatment of nonspecific complexes is discussed. Finally, it is shown that this method is applicable over a large range of weak association constants typical of specific but transient protein-protein complexes.

A high throughput MHC II binding assay for quantitative analysis of peptide epitopes

Biochemical assays with recombinant human MHC II molecules can provide rapid, quantitative insights into immunogenic epitope identification, deletion, or design(1,2). Here, a peptide-MHC II binding assay is scaled to 384-well format. The scaled down protocol reduces reagent costs by 75% and is higher throughput than previously described 96-well protocols(1,3-5). Specifically, the experimental design permits robust and reproducible analysis of up to 15 peptides against one MHC II allele per 384-well ELISA plate. Using a single liquid handling robot, this method allows one researcher to analyze approximately ninety test peptides in triplicate over a range of eight concentrations and four MHC II allele types in less than 48 hr. Others working in the fields of protein deimmunization or vaccine design and development may find the protocol to be useful in facilitating their own work. In particular, the step-by-step instructions and the visual format of JoVE should allow other users to quickly and easily establish this methodology in their own labs.

Integrated assessment of predicted MHC binding and cross-conservation with self reveals patterns of viral camouflage

Background

Immune recognition of foreign proteins by T cells hinges on the formation of a ternary complex sandwiching a constituent peptide of the protein between a major histocompatibility complex (MHC) molecule and a T cell receptor (TCR). Viruses have evolved means of "camouflaging" themselves, avoiding immune recognition by reducing the MHC and/or TCR binding of their constituent peptides. Computer-driven T cell epitope mapping tools have been used to evaluate the degree to which particular viruses have used this means of avoiding immune response, but most such analyses focus on MHC-facing 'agretopes'. Here we set out a new means of evaluating the TCR faces of viral peptides in addition to their agretopes, integrating evaluations of both sides of the ternary complex in a single analysis.

Methods

This paper develops what we call the Janus Immunogenicity Score (JIS), bringing together a well-established method for predicting MHC binding, with a novel assessment of the potential for TCR binding based on similarity with self. Intuitively, both good MHC binding and poor self-similarity are required for high immunogenicity (i.e., a robust T effector response).

Results

Focusing on the class II antigen-processing pathway, we show that the JIS of T effector epitopes and null or regulatory epitopes deposited in a large database of epitopes (Immune Epitope Database) are significantly different. We then show that different types of viruses display significantly different patterns of scores over their constituent peptides, with viruses causing chronic infection (Epstein-Barr and cytomegalovirus) strongly shifted to lower scores relative to those causing acute infection (Ebola and Marburg). Similarly we find distinct patterns among influenza proteins in H1N1 (a strain against which human populations rapidly developed immunity) and H5N1 and H7N9 (highly pathogenic avian flu strains, with significantly greater case mortality rates).

Conclusion

The Janus Immunogenicity Score, which integrates MHC binding and TCR cross-reactivity, provides a new tool for studying immunogenicity of pathogens and may improve the selection and optimization of antigenic elements for vaccine design.

Learning Sequence Determinants of Protein:protein Interaction Specificity with Sparse Graphical Models

In studying the strength and specificity of interaction between members of two protein families, key questions center on which pairs of possible partners actually interact, how well they interact, and why they interact while others do not. The advent of large-scale experimental studies of interactions between members of a target family and a diverse set of possible interaction partners offers the opportunity to address these questions. We develop here a method, DgSpi (Data-driven Graphical models of Specificity in Protein:protein Interactions), for learning and using graphical models that explicitly represent the amino acid basis for interaction specificity (why) and extend earlier classification-oriented approaches (which) to predict the ΔG of binding (how well). We demonstrate the effectiveness of our approach in analyzing and predicting interactions between a set of 82 PDZ recognition modules, against a panel of 217 possible peptide partners, based on data from MacBeath and colleagues. Our predicted ΔG values are highly predictive of the experimentally measured ones, reaching correlation coefficients of 0.69 in 10-fold cross-validation and 0.63 in leave-one-PDZ-out cross-validation. Furthermore, the model serves as a compact representation of amino acid constraints underlying the interactions, enabling protein-level ΔG predictions to be naturally understood in terms of residue-level constraints. Finally, as a generative model, DgSpi readily enables the design of new interacting partners, and we demonstrate that designed ligands are novel and diverse.