"Redirecting an anti-IL-1ß antibody to bind a new, unrelated and computationally predicted epitope on hIL-17A".

Antibody engineering technology is at the forefront of therapeutic antibody development. The primary goal for engineering a therapeutic antibody is the generation of an antibody with a desired specificity, affinity, function, and developability profile. Mature antibodies are considered antigen specific, which may preclude their use as a starting point for antibody engineering. Here, we explore the plasticity of mature antibodies by engineering novel specificity and function to a pre-selected antibody template. Using a small, focused library, we engineered AAL160, an anti-IL-1ß antibody, to bind the unrelated antigen IL-17A, with the introduction of seven mutations. The final redesigned antibody, 11.003, retains favorable biophysical properties, binds IL-17A with sub-nanomolar affinity, inhibits IL-17A binding to its cognate receptor and is functional in a cell-based assay. The epitope of the engineered antibody can be computationally predicted based on the sequence of the template antibody, as is confirmed by the crystal structure of the 11.003/IL-17A complex. The structures of the 11.003/IL-17A and the AAL160/IL-1ß complexes highlight the contribution of germline residues to the paratopes of both the template and re-designed antibody. This case study suggests that the inherent plasticity of antibodies allows for re-engineering of mature antibodies to new targets, while maintaining desirable developability profiles.

Accurate profiling of full-length Fv in highly homologous antibody libraries using UMI tagged short reads.

Deep parallel sequencing (NGS) is a viable tool for monitoring scFv and Fab library dynamics in many antibody engineering high-throughput screening efforts. Although very useful, the commonly used Illumina NGS platform cannot handle the entire sequence of scFv or Fab in a single read, usually focusing on specific CDRs or resorting to sequencing VH and VL variable domains separately, thus limiting its utility in comprehensive monitoring of selection dynamics. Here we present a simple and robust method for deep sequencing repertoires of full length scFv, Fab and Fv antibody sequences. This process utilizes standard molecular procedures and unique molecular identifiers (UMI) to pair separately sequenced VH and VL. We show that UMI assisted VH-VL matching allows for a comprehensive and highly accurate mapping of full length Fv clonal dynamics in large highly homologous antibody libraries, as well as identification of rare variants. In addition to its utility in synthetic antibody discovery processes, our method can be instrumental in generating large datasets for machine learning (ML) applications, which in the field of antibody engineering has been hampered by conspicuous paucity of large scale full length Fv data.

DistilProtBert: a distilled protein language model used to distinguish between real proteins and their randomly shuffled counterparts.

SUMMARY: Recently, deep learning models, initially developed in the field of natural language processing (NLP), were applied successfully to analyze protein sequences. A major drawback of these models is their size in terms of the number of parameters needed to be fitted and the amount of computational resources they require. Recently, 'distilled' models using the concept of student and teacher networks have been widely used in NLP. Here, we adapted this concept to the problem of protein sequence analysis, by developing DistilProtBert, a distilled version of the successful ProtBert model. Implementing this approach, we reduced the size of the network and the running time by 50%, and the computational resources needed for pretraining by 98% relative to ProtBert model. Using two published tasks, we showed that the performance of the distilled model approaches that of the full model. We next tested the ability of DistilProtBert to distinguish between real and random protein sequences. The task is highly challenging if the composition is maintained on the level of singlet, doublet and triplet amino acids. Indeed, traditional machine-learning algorithms have difficulties with this task. Here, we show that DistilProtBert preforms very well on singlet, doublet and even triplet-shuffled versions of the human proteome, with AUC of 0.92, 0.91 and 0.87, respectively. Finally, we suggest that by examining the small number of false-positive classifications (i.e. shuffled sequences classified as proteins by DistilProtBert), we may be able to identify de novo potential natural-like proteins based on random shuffling of amino acid sequences. AVAILABILITY AND IMPLEMENTATION: https://github.com/yarongef/DistilProtBert.

Targeting the actin nucleation promoting factor WASp provides a therapeutic approach for hematopoietic malignancies.

Cancer cells depend on actin cytoskeleton rearrangement to carry out hallmark malignant functions including activation, proliferation, migration and invasiveness. Wiskott-Aldrich Syndrome protein (WASp) is an actin nucleation-promoting factor and is a key regulator of actin polymerization in hematopoietic cells. The involvement of WASp in malignancies is incompletely understood. Since WASp is exclusively expressed in hematopoietic cells, we performed in silico screening to identify small molecule compounds (SMCs) that bind WASp and promote its degradation. We describe here one such identified molecule; this WASp-targeting SMC inhibits key WASp-dependent actin processes in several types of hematopoietic malignancies in vitro and in vivo without affecting naïve healthy cells. This small molecule demonstrates limited toxicity and immunogenic effects, and thus, might serve as an effective strategy to treat specific hematopoietic malignancies in a safe and precisely targeted manner.

Computational Design of Epitope-Specific Functional Antibodies.

The ultimate goal of protein design is to introduce new biological activity. We propose a computational approach for designing functional antibodies by focusing on functional epitopes, integrating large-scale statistical analysis with multiple structural models. Machine learning is used to analyze these models and predict specific residue-residue contacts. We use this approach to design a functional antibody to counter the proinflammatory effect of the cytokine interleukin-17A (IL-17A). X-ray crystallography confirms that the designed antibody binds the targeted epitope and the interaction is mediated by the designed contacts. Cell-based assays confirm that the antibody is functional. Importantly, this approach does not rely on a high-quality 3D model of the designed complex or even a solved structure of the target. As demonstrated here, this approach can be used to design biologically active antibodies, removing some of the main hurdles in antibody design and in drug discovery.

Computational design of antibodies.

Antibody design aims to create new antibodies with biological activity that can be used in therapy and research. Traditional methods for antibody discovery, such as animal immunization and large-scale library screening, generate antibodies that bind to the target of interest, but do not necessarily have the desired functional effect. Computational methods can be utilized as a means to guide the search for biologically relevant antibodies, focusing on specificity and affinity determinants to target a particular region of the antigen. Such an approach would allow for the design of epitope-specific antibodies that will have the desired effect on the function of the targeted protein.

Integrating genomic data from high-throughput studies with computational modeling reveals differences in the molecular basis of hyposalivation between type 1 and type 2 diabetes.

OBJECTIVES: Both type 1 and type 2 diabetes are accompanied by a high prevalence of hyposalivation (decreased salivary secretion), resulting in oral tissue damage. However, the molecular basis for the hyposalivation is yet unknown. Identifying genes and proteins that account for diabetes-related hyposalivation will help understanding the basis for this condition and identifying disease biomarkers in saliva. MATERIALS AND METHODS: We integrated genomic data from 110 high-throughput studies with computational modeling, to explore the relationship between diabetes and salivary glands on a genomic scale. RESULTS: A significant overlap exists between genes that are altered in both types of diabetes and genes that are expressed in salivary glands; 87 type 1 diabetes and 34 type 2 diabetes associated genes are also common to salivary glands. However, the overlap between these genes is not significant. CONCLUSIONS: Type 1 and type 2 diabetes associated genes are involved in the salivary secretion process, but mostly at different parts of it. This suggests that type 1 and type 2 diabetes impair salivary secretion by affecting different processes in the salivary tissue. CLINICAL RELEVANCE: The genomic characteristics of Type 1 and type 2 diabetes may explain differences in salivary gland tissues morphology and saliva composition in people with diabetes, and suggest candidate proteins for diabetes salivary biomarkers.

A novel Fer/FerT targeting compound selectively evokes metabolic stress and necrotic death in malignant cells.

Disruption of the reprogrammed energy management system of malignant cells is a prioritized goal of targeted cancer therapy. Two regulators of this system are the Fer kinase, and its cancer cell specific variant, FerT, both residing in subcellular compartments including the mitochondrial electron transport chain. Here, we show that a newly developed inhibitor of Fer and FerT, E260, selectively evokes metabolic stress in cancer cells by imposing mitochondrial dysfunction and deformation, and onset of energy-consuming autophagy which decreases the cellular ATP level. Notably, Fer was also found to associate with PARP-1 and E260 disrupted this association thereby leading to PARP-1 activation. The cooperative intervention with these metabolic pathways leads to energy crisis and necrotic death in malignant, but not in normal human cells, and to the suppression of tumors growth in vivo. Thus, E260 is a new anti-cancer agent which imposes metabolic stress and cellular death in cancer cells.The tyrosine-kinases Fer/FerT associate with the mitochondrial electron transport chain in cancer cells supporting their metabolic reprogramming. Here the authors discover a compound that disrupts Fer /FerT activity and selectively induces cell death of cancer cell lines displaying anti-tumor activity in vivo.

Working toward precision medicine: Predicting phenotypes from exomes in the Critical Assessment of Genome Interpretation (CAGI) challenges.

Precision medicine aims to predict a patient's disease risk and best therapeutic options by using that individual's genetic sequencing data. The Critical Assessment of Genome Interpretation (CAGI) is a community experiment consisting of genotype-phenotype prediction challenges; participants build models, undergo assessment, and share key findings. For CAGI 4, three challenges involved using exome-sequencing data: Crohn's disease, bipolar disorder, and warfarin dosing. Previous CAGI challenges included prior versions of the Crohn's disease challenge. Here, we discuss the range of techniques used for phenotype prediction as well as the methods used for assessing predictive models. Additionally, we outline some of the difficulties associated with making predictions and evaluating them. The lessons learned from the exome challenges can be applied to both research and clinical efforts to improve phenotype prediction from genotype. In addition, these challenges serve as a vehicle for sharing clinical and research exome data in a secure manner with scientists who have a broad range of expertise, contributing to a collaborative effort to advance our understanding of genotype-phenotype relationships.

De-novo protein function prediction using DNA binding and RNA binding proteins as a test case.

Of the currently identified protein sequences, 99.6% have never been observed in the laboratory as proteins and their molecular function has not been established experimentally. Predicting the function of such proteins relies mostly on annotated homologs. However, this has resulted in some erroneous annotations, and many proteins have no annotated homologs. Here we propose a de-novo function prediction approach based on identifying biophysical features that underlie function. Using our approach, we discover DNA and RNA binding proteins that cannot be identified based on homology and validate these predictions experimentally. For example, FGF14, which belongs to a family of secreted growth factors was predicted to bind DNA. We verify this experimentally and also show that FGF14 is localized to the nucleus. Mutating the predicted binding site on FGF14 abrogated DNA binding. These results demonstrate the feasibility of automated de-novo function prediction based on identifying function-related biophysical features.

How far from the SNP may the causative genes be?

While GWAS identify many disease-associated SNPs, using them to decipher disease mechanisms is hindered by the difficulty in mapping SNPs to genes. Most SNPs are in non-coding regions and it is often hard to identify the genes they implicate. To explore how far the SNP may be from the affected genes we used a pathway-based approach. We found that affected genes are often up to 2 Mbps away from the associated SNP, and are not necessarily the closest genes to the SNP. Existing approaches for mapping SNPs to genes leave many SNPs unmapped to genes and reveal only 86 significant phenotype-pathway associations for all known GWAS hits combined. Using the pathway-based approach we propose here allows mapping of virtually all SNPs to genes and reveals 435 statistically significant phenotype-pathway associations. In search for mechanisms that may explain the relationships between SNPs and distant genes, we found that SNPs that are mapped to distant genes have significantly more large insertions/deletions around them than other SNPs, suggesting that these SNPs may sometimes be markers for large insertions/deletions that may affect large genomic regions.

Understanding differences between synthetic and natural antibodies can help improve antibody engineering.

Synthetic libraries are a major source of human-like antibody (Ab) drug leads. To assess the similarity between natural Abs and the products of these libraries, we compared large sets of natural and synthetic Abs using "CDRs Analyzer," a tool we introduce for structural analysis of Ab-antigen (Ag) interactions. Natural Abs, we found, recognize their Ags by combining multiple complementarity-determining regions (CDRs) to create an integrated interface. Synthetic Abs, however, rely dominantly, sometimes even exclusively on CDRH3. The increased contribution of CDRH3 to Ag binding in synthetic Abs comes with a substantial decrease in the involvement of CDRH2 and CDRH1. Furthermore, in natural Abs CDRs specialize in specific types of non-covalent interactions with the Ag. CDRH1 accounts for a significant portion of the cation-pi interactions; CDRH2 is the major source of salt-bridges and CDRH3 accounts for most hydrogen bonds. In synthetic Abs this specialization is lost, and CDRH3 becomes the main sources of all types of contacts. The reliance of synthetic Abs on CDRH3 reduces the complexity of their interaction with the Ag: More Ag residues contact only one CDR and fewer contact 3 CDRs or more. We suggest that the focus of engineering attempts on CDRH3 results in libraries enriched with variants that are not natural-like. This may affect not only Ag binding, but also Ab expression, stability and selectivity. Our findings can help guide library design, creating libraries that can bind more epitopes and Abs that better mimic the natural antigenic interactions.

Antibody specific epitope prediction-emergence of a new paradigm.

The development of accurate tools for predicting B-cell epitopes is important but difficult. Traditional methods have examined which regions in an antigen are likely binding sites of an antibody. However, it is becoming increasingly clear that most antigen surface residues will be able to bind one or more of the myriad of possible antibodies. In recent years, new approaches have emerged for predicting an epitope for a specific antibody, utilizing information encoded in antibody sequence or structure. Applying such antibody-specific predictions to groups of antibodies in combination with easily obtainable experimental data improves the performance of epitope predictions. We expect that further advances of such tools will be possible with the integration of immunoglobulin repertoire sequencing data.

PEASE: predicting B-cell epitopes utilizing antibody sequence.

UNLABELLED: Antibody epitope mapping is a key step in understanding antibody-antigen recognition and is of particular interest for drug development, diagnostics and vaccine design. Most computational methods for epitope prediction are based on properties of the antigen sequence and/or structure, not taking into account the antibody for which the epitope is predicted. Here, we introduce PEASE, a web server predicting antibody-specific epitopes, utilizing the sequence of the antibody. The predictions are provided both at the residue level and as patches on the antigen structure. The tradeoff between recall and precision can be tuned by the user, by changing the default parameters. The results are provided as text and HTML files as well as a graph, and can be viewed on the antigen 3D structure. AVAILABILITY AND IMPLEMENTATION: PEASE is freely available on the web at www.ofranlab.org/PEASE. CONTACT: yanay@ofranlab.org.

Potent neutralization of vaccinia virus by divergent murine antibodies targeting a common site of vulnerability in L1 protein.

UNLABELLED: Vaccinia virus (VACV) L1 is an important target for viral neutralization and has been included in multicomponent DNA or protein vaccines against orthopoxviruses. To further understand the protective mechanism of the anti-L1 antibodies, we generated five murine anti-L1 monoclonal antibodies (MAbs), which clustered into 3 distinct epitope groups. While two groups of anti-L1 failed to neutralize, one group of 3 MAbs potently neutralized VACV in an isotype- and complement-independent manner. This is in contrast to neutralizing antibodies against major VACV envelope proteins, such as H3, D8, or A27, which failed to completely neutralize VACV unless the antibodies are of complement-fixing isotypes and complement is present. Compared to nonneutralizing anti-L1 MAbs, the neutralization antibodies bound to the recombinant L1 protein with a significantly higher affinity and also could bind to virions. By using a variety of techniques, including the isolation of neutralization escape mutants, hydrogen/deuterium exchange mass spectrometry, and X-ray crystallography, the epitope of the neutralizing antibodies was mapped to a conformational epitope with Asp35 as the key residue. This epitope is similar to the epitope of 7D11, a previously described potent VACV neutralizing antibody. The epitope was recognized mainly by CDR1 and CDR2 of the heavy chain, which are highly conserved among antibodies recognizing the epitope. These antibodies, however, had divergent light-chain and heavy-chain CDR3 sequences. Our study demonstrates that the conformational L1 epitope with Asp35 is a common site of vulnerability for potent neutralization by a divergent group of antibodies. IMPORTANCE: Vaccinia virus, the live vaccine for smallpox, is one of the most successful vaccines in human history, but it presents a level of risk that has become unacceptable for the current population. Studying the immune protection mechanism of smallpox vaccine is important for understanding the basic principle of successful vaccines and the development of next-generation, safer vaccines for highly pathogenic orthopoxviruses. We studied antibody targets in smallpox vaccine by developing potent neutralizing antibodies against vaccinia virus and comprehensively characterizing their epitopes. We found a site in vaccinia virus L1 protein as the target of a group of highly potent murine neutralizing antibodies. The analysis of antibody-antigen complex structure and the sequences of the antibody genes shed light on how these potent neutralizing antibodies are elicited from immunized mice.

Large scale analysis of phenotype-pathway relationships based on GWAS results.

The widely used pathway-based approach for interpreting Genome Wide Association Studies (GWAS), assumes that since function is executed through the interactions of multiple genes, different perturbations of the same pathway would result in a similar phenotype. This assumption, however, was not systemically assessed on a large scale. To determine whether SNPs associated with a given complex phenotype affect the same pathways more than expected by chance, we analyzed 368 phenotypes that were studied in >5000 GWAS. We found 216 significant phenotype-pathway associations between 70 of the phenotypes we analyzed and known pathways. We also report 391 strong phenotype-phenotype associations between phenotypes that are affected by the same pathways. While some of these associations confirm previously reported connections, others are new and could shed light on the molecular basis of these diseases. Our findings confirm that phenotype-associated SNPs cluster into pathways much more than expected by chance. However, this is true for <20% (70/368) of the phenotypes. Different types of phenotypes show markedly different tendencies: Virtually all autoimmune phenotypes show strong clustering of SNPs into pathways, while most cancers and metabolic conditions, and all electrophysiological phenotypes, could not be significantly associated with any pathway despite being significantly associated with a large number of SNPs. While this may be due to missing data, it may also suggest that these phenotypes could result only from perturbations of specific genes and not from other perturbations of the same pathway. Further analysis of pathway-associated versus gene-associated phenotypes is, therefore, needed in order to understand disease etiology and in order to promote better drug target selection.

Using a combined computational-experimental approach to predict antibody-specific B cell epitopes.

Antibody epitope mapping is crucial for understanding B cell-mediated immunity and required for characterizing therapeutic antibodies. In contrast to T cell epitope mapping, no computational tools are in widespread use for prediction of B cell epitopes. Here, we show that, utilizing the sequence of an antibody, it is possible to identify discontinuous epitopes on its cognate antigen. The predictions are based on residue-pairing preferences and other interface characteristics. We combined these antibody-specific predictions with results of cross-blocking experiments that identify groups of antibodies with overlapping epitopes to improve the predictions. We validate the high performance of this approach by mapping the epitopes of a set of antibodies against the previously uncharacterized D8 antigen, using complementary techniques to reduce method-specific biases (X-ray crystallography, peptide ELISA, deuterium exchange, and site-directed mutagenesis). These results suggest that antibody-specific computational predictions and simple cross-blocking experiments allow for accurate prediction of residues in conformational B cell epitopes.

Co-expression and co-localization of hub proteins and their partners are encoded in protein sequence.

Spatiotemporal coordination is a critical factor in biological processes. Some hubs in protein-protein interaction networks tend to be co-expressed and co-localized with their partners more strongly than others, a difference which is arguably related to functional differences between the hubs. Based on numerous analyses of yeast hubs, it has been suggested that differences in co-expression and co-localization are reflected in the structural and molecular characteristics of the hubs. We hypothesized that if indeed differences in co-expression and co-localization are encoded in the molecular characteristics of the protein, it may be possible to predict the tendency for co-expression and co-localization of human hubs based on features learned from systematically characterized yeast hubs. Thus, we trained a prediction algorithm on hubs from yeast that were classified as either strongly or weakly co-expressed and co-localized with their partners, and applied the trained model to 800 human hub proteins. We found that the algorithm significantly distinguishes between human hubs that are co-expressed and co-localized with their partners and hubs that are not. The prediction is based on sequence derived features such as "stickiness", i.e. the existence of multiple putative binding sites that enable multiple simultaneous interactions, "plasticity", i.e. the existence of predicted structural disorder which conjecturally allows for multiple consecutive interactions with the same binding site and predicted subcellular localization. These results suggest that spatiotemporal dynamics is encoded, at least in part, in the amino acid sequence of the protein and that this encoding is similar in yeast and in human.

Large-scale analysis of somatic hypermutations in antibodies reveals which structural regions, positions and amino acids are modified to improve affinity.

The principles of affinity maturation of antibodies (Abs), which underlies B cell-mediated immunity, are still under debate. It is unclear whether the antigen (Ag) binding site is a preferred target for mutations, and what the role of activation-induced deaminase (AID) hotspots is in this process. Here we report a structural analysis of 3495 residues that have been replaced through somatic hypermutations (SHMs) in 196 Abs. We show that there is no correlation between the propensity of an amino acid to be in AID hotspot and the probability that it is replaced during the SHM process. Although AID hotspots may be necessary to enable SHMs, they are not a major driving force in determining which residues are mutated. We identified Ab positions that are highly mutated and significantly affect binding. The effect of mutation on binding energy is a major factor in determining which structural regions of the Ab are mutated. There is a clear preference for mutations at the Ag-binding site. However, positions outside this region that also affect binding are often preferred targets for SHMs. As for amino acid preferences, a general trend during SHM is to make Ab-Ag interfaces more similar to protein-protein interfaces in general. In different regions of the Ab, there are different sets of preferences for amino acid substitution. This mapping improves our understanding of Ab affinity maturation and may assist in Ab engineering.

The structural basis of antibody-antigen recognition.

The function of antibodies (Abs) involves specific binding to antigens (Ags) and activation of other components of the immune system to fight pathogens. The six hypervariable loops within the variable domains of Abs, commonly termed complementarity determining regions (CDRs), are widely assumed to be responsible for Ag recognition, while the constant domains are believed to mediate effector activation. Recent studies and analyses of the growing number of available Ab structures, indicate that this clear functional separation between the two regions may be an oversimplification. Some positions within the CDRs have been shown to never participate in Ag binding and some off-CDRs residues often contribute critically to the interaction with the Ag. Moreover, there is now growing evidence for non-local and even allosteric effects in Ab-Ag interaction in which Ag binding affects the constant region and vice versa. This review summarizes and discusses the structural basis of Ag recognition, elaborating on the contribution of different structural determinants of the Ab to Ag binding and recognition. We discuss the CDRs, the different approaches for their identification and their relationship to the Ag interface. We also review what is currently known about the contribution of non-CDRs regions to Ag recognition, namely the framework regions (FRs) and the constant domains. The suggested mechanisms by which these regions contribute to Ag binding are discussed. On the Ag side of the interaction, we discuss attempts to predict B-cell epitopes and the suggested idea to incorporate Ab information into B-cell epitope prediction schemes. Beyond improving the understanding of immunity, characterization of the functional role of different parts of the Ab molecule may help in Ab engineering, design of CDR-derived peptides, and epitope prediction.