Assisted Design of Antibody and Protein Therapeutics (ADAPT)

Enhancing the Affinity of a Novel Selective scFv for Soluble ST2 through Computational Design

Suppression of Tumorigenicity 2 (ST2) is a member of the IL-1 receptor family, which includes transmembrane (ST2L) and soluble (sST2) isoforms. sST2 functions as a decoy receptor for Interleukin-33 (IL-33), thereby blocking the activation of the IL-33/ST2L signaling axis, which is essential for tissue repair and immune regulation. Clinical evidence indicates that elevated sST2 levels are associated with increased disease severity in conditions such as ulcerative colitis (UC), cardiovascular disease, and asthma. However, current antibodies cannot reliably distinguish between sST2 and its membrane-bound isoform ST2L, limiting their effectiveness for diagnostic and therapeutic use. To address this limitation, we developed an antibody that selectively targets sST2. Using a phage display library, we identified a single-chain variable fragment (scFv) with high specificity for a unique five amino acid sequence (SKECF) located at the C-terminus of sST2. Our parental scFv showed high selectivity for sST2 with minimal cross-reactivity to ST2L, as demonstrated by both flow cytometry and immunoprecipitation. Molecular simulations identified key binding residues, allowing the design of four scFv mutants, three of which displayed improved binding in surface plasmon resonance (SPR) analyses. The A183YL2 mutant exhibited a 3.4-fold increase in binding affinity, while G100WH3 demonstrated reduced binding due to unfavorable conformations. This study presents an anti-sST2 scFv with enhanced specificity and affinity, offering a promising tool for the diagnosis and treatment of inflammatory diseases, in which sST2 interferes with IL-33-mediated tissue repair.

AI-augmented physics-based docking for antibody-antigen complex prediction

MOTIVATION

Predicting the structure of antibody-antigen complexes is a challenging task with significant implications for the design of better antibody therapeutics. However, the levels of success have remained dauntingly low, particularly when high standards for model quality are required, a necessity for efficient antibody design. Artificial intelligence (AI) has significantly impacted the landscape of structure prediction for antibodies, both alone and in complex with their antigens.

METHODS

We utilized AI-guided antibody modeling tools to generate ensembles displaying diversity in the complementarity-determining region (CDR) and integrated those into our previously published AlphaFold2-rescored docking pipeline, a strategy called AI-augmented physics-based docking. In this study, we also compare docking performance with AlphaFold and Boltz-1, the new state-of-the-art. We distinguish between two types of success tailored to specific downstream applications: (i) criteria sufficient for epitope mapping, where gross quality is adequate and can complement experimental techniques, and (ii) criteria for producing higher-quality models suitable for engineering purposes.

RESULTS

We highlight that the quality of the ensemble is crucial for docking performance, that including too many models can be detrimental, and that prioritization of models is essential for achieving good performance. In a scenario analogous to docking using a crystallized antigen, our results robustly demonstrate the advantages of AI-augmented docking over AlphaFold2, further accentuated when higher standards in quality are imposed. Docking also shows improvements over Boltz-1, but those are less pronounced. Docking performance is still noticeably lower than AlphaFold3 in both epitope mapping and antibody design use cases. We observe a strong dependence on CDR-H3 loop length for physics-based tools on their ability to successfully predict. This helps define an applicability range where physics-based docking can be competitive to the newer generation of AI tools.

AVAILABILITY AND IMPLEMENTATION

The AF2 rescoring scripts are available at github.com/gaudreaultfnrc/AF2-Rescoring.

Designing and bioengineering of CDRs with higher affinity against receptor-binding domain (RBD) of SARS-CoV-2 Omicron variant

The emergence of the SARS-CoV-2 Omicron variant highlights the need for innovative strategies to address evolving viral threats. This study bioengineered three nanobodies H11-H4, C5, and H3 originally targeting the Wuhan RBD, to bind more effectively to the Omicron RBD. A structure-based in silico affinity maturation pipeline was developed to enhance their binding affinities. The pipeline consists of three key steps: high-throughput in silico mutagenesis of complementarity-determining regions (CDRs), protein-protein docking for screening, and molecular dynamics (MD) simulations for assessment of the complex stability. A total of 741, 551, and 684 mutations were introduced in H11-H4, C5, and H3 nanobodies, respectively. Protein-protein docking and MD simulations shortlisted high-affinity mutants for H11-H4(6), C5(5), and H3(6). Further, recombinant production of H11-H4 mutants and Omicron RBD enabled experimental validation through Isothermal Titration Calorimetry (ITC). The H11-H4 mutants R27E, S57D, S107K, D108W, and A110I exhibited improved binding affinities with dissociation constant (KD) values ranging from ~8.8 to ~27 μM, compared to the H11-H4 nanobody KD of ~32 μM, representing a three-fold enhancement. This study demonstrates the potential of the developed in silico affinity maturation pipeline as a rapid, cost-effective method for repurposing nanobodies, aiding the development of robust prophylactic strategies against evolving SARS-CoV-2 variants and other pathogens.

Structure-guided mutations in CDRs for enhancing the affinity of neutralizing SARS-CoV-2 nanobody

The optimization of antibodies to attain the desired levels of affinity and specificity holds great promise for the development of next generation therapeutics. This study delves into the refinement and engineering of complementarity-determining regions (CDRs) through in silico affinity maturation followed by binding validation using isothermal titration calorimetry (ITC) and pseudovirus-based neutralization assays. Specifically, it focuses on engineering CDRs targeting the epitopes of receptor-binding domain (RBD) of the spike protein of SARS-CoV-2. A structure-guided virtual library of 112 single mutations in CDRs was generated and screened against RBD to select the potential affinity-enhancing mutations. Protein-protein docking analysis identified 32 single mutants of which nine mutants were selected for molecular dynamics (MD) simulations. Subsequently, biophysical ITC studies provided insights into binding affinity, and consistent with in silico findings, six mutations that demonstrated better binding affinity than native nanobody were further tested in vitro for neutralization activity against SARS-CoV-2 pseudovirus. Leu106Thr mutant was found to be most effective in virus-neutralization with IC50 values of ∼0.03 μM, as compared to the native nanobody (IC50 ∼0.77 μM). Thus, in this study, the developed computational pipeline guided by structure-aided interface profiles and thermodynamic analysis holds promise for the streamlined development of antibody-based therapeutic interventions against emerging variants of SARS-CoV-2 and other infectious pathogens.

Structure-Based Optimization of One Neutralizing Antibody against SARS-CoV-2 Variants Bearing the L452R Mutation

Neutralizing antibodies (nAbs) play an important role against SARS-CoV-2 infections. Previously, we have reported one potent receptor binding domain (RBD)-binding nAb Ab08 against the SARS-CoV-2 prototype and a panel of variants, but Ab08 showed much less efficacy against the variants harboring the L452R mutation. To overcome the antibody escape caused by the L452R mutation, we generated several structure-based Ab08 derivatives. One derivative, Ab08-K99E, displayed the mostly enhanced neutralizing potency against the Delta pseudovirus bearing the L452R mutation compared to the Ab08 and other derivatives. Ab08-K99E also showed improved neutralizing effects against the prototype, Omicron BA.1, and Omicron BA.4/5 pseudoviruses. In addition, compared to the original Ab08, Ab08-K99E exhibited high binding properties and affinities to the RBDs of the prototype, Delta, and Omicron BA.4/5 variants. Altogether, our findings report an optimized nAb, Ab08-K99E, against SARS-CoV-2 variants and demonstrate structure-based optimization as an effective way for antibody development against pathogens.

The rise of big data: deep sequencing-driven computational methods are transforming the landscape of synthetic antibody design

Synthetic antibodies (Abs) represent a category of artificial proteins capable of closely emulating the functions of natural Abs. Their in vitro production eliminates the need for an immunological response, streamlining the process of Ab discovery, engineering, and development. These artificially engineered Abs offer novel approaches to antigen recognition, paratope site manipulation, and biochemical/biophysical enhancements. As a result, synthetic Abs are fundamentally reshaping conventional methods of Ab production. This mirrors the revolution observed in molecular biology and genomics as a result of deep sequencing, which allows for the swift and cost-effective sequencing of DNA and RNA molecules at scale. Within this framework, deep sequencing has enabled the exploration of whole genomes and transcriptomes, including particular gene segments of interest. Notably, the fusion of synthetic Ab discovery with advanced deep sequencing technologies is redefining the current approaches to Ab design and development. Such combination offers opportunity to exhaustively explore Ab repertoires, fast-tracking the Ab discovery process, and enhancing synthetic Ab engineering. Moreover, advanced computational algorithms have the capacity to effectively mine big data, helping to identify Ab sequence patterns/features hidden within deep sequencing Ab datasets. In this context, these methods can be utilized to predict novel sequence features thereby enabling the successful generation of de novo Ab molecules. Hence, the merging of synthetic Ab design, deep sequencing technologies, and advanced computational models heralds a new chapter in Ab discovery, broadening our comprehension of immunology and streamlining the advancement of biological therapeutics.

A dual computational and experimental strategy to enhance TSLP antibody affinity for improved asthma treatment

Thymic stromal lymphopoietin is a key cytokine involved in the pathogenesis of asthma and other allergic diseases. Targeting TSLP and its signaling pathways is increasingly recognized as an effective strategy for asthma treatment. This study focused on enhancing the affinity of the T6 antibody, which specifically targets TSLP, by integrating computational and experimental methods. The initial affinity of the T6 antibody for TSLP was lower than the benchmark antibody AMG157. To improve this, we utilized alanine scanning, molecular docking, and computational tools including mCSM-PPI2 and GEO-PPI to identify critical amino acid residues for site-directed mutagenesis. Subsequent mutations and experimental validations resulted in an antibody with significantly enhanced blocking capacity against TSLP. Our findings demonstrate the potential of computer-assisted techniques in expediting antibody affinity maturation, thereby reducing both the time and cost of experiments. The integration of computational methods with experimental approaches holds great promise for the development of targeted therapeutic antibodies for TSLP-related diseases.

Sequence-based engineering of pH-sensitive antibodies for tumor targeting or endosomal recycling applications

The engineering of pH-sensitive therapeutic antibodies, particularly for improving effectiveness and specificity in acidic solid-tumor microenvironments, has recently gained traction. While there is a justified need for pH-dependent immunotherapies, current engineering techniques are tedious and laborious, requiring repeated rounds of experiments under different pH conditions. Inexpensive computational techniques to predict the effectiveness of His pH-switches require antibody-antigen complex structures, but these are lacking in most cases. To circumvent these requirements, we introduce a sequence-based in silico method for predicting His mutations in the variable region of antibodies, which could lead to pH-biased antigen binding. This method, called Sequence-based Identification of pH-sensitive Antibody Binding (SIpHAB), was trained on 3D-structure-based calculations of 3,490 antibody-antigen complexes with solved experimental structures. SIpHAB was parametrized to enhance preferential binding either toward or against the acidic pH, for selective targeting of solid tumors or for antigen release in the endosome, respectively. Applications to nine antibody-antigen systems with previously reported binding preferences at different pHs demonstrated the utility and enrichment capabilities of this high-throughput computational tool. SIpHAB, which only requires knowledge of the antibody primary amino-acid sequence, could enable a more efficient triage of pH-sensitive antibody candidates than could be achieved conventionally. An online webserver for running SipHAB is available freely at https://mm.nrc-cnrc.gc.ca/software/siphab/runner/.

Enhanced antibody-antigen structure prediction from molecular docking using AlphaFold2

Predicting the structure of antibody-antigen complexes has tremendous value in biomedical research but unfortunately suffers from a poor performance in real-life applications. AlphaFold2 (AF2) has provided renewed hope for improvements in the field of protein-protein docking but has shown limited success against antibody-antigen complexes due to the lack of co-evolutionary constraints. In this study, we used physics-based protein docking methods for building decoy sets consisting of low-energy docking solutions that were either geometrically close to the native structure (positives) or not (negatives). The docking models were then fed into AF2 to assess their confidence with a novel composite score based on normalized pLDDT and pTMscore metrics after AF2 structural refinement. We show benefits of the AF2 composite score for rescoring docking poses both in terms of (1) classification of positives/negatives and of (2) success rates with particular emphasis on early enrichment. Docking models of at least medium quality present in the decoy set, but not necessarily highly ranked by docking methods, benefitted most from AF2 rescoring by experiencing large advances towards the top of the reranked list of models. These improvements, obtained without any calibration or novel methodologies, led to a notable level of performance in antibody-antigen unbound docking that was never achieved previously.

Affinity maturation of antibody fragments: A review encompassing the development from random approaches to computational rational optimization

Routinely screened antibody fragments usually require further in vitro maturation to achieve the desired biophysical properties. Blind in vitro strategies can produce improved ligands by introducing random mutations into the original sequences and selecting the resulting clones under more and more stringent conditions. Rational approaches exploit an alternative perspective that aims first at identifying the specific residues potentially involved in the control of biophysical mechanisms, such as affinity or stability, and then to evaluate what mutations could improve those characteristics. The understanding of the antigen-antibody interactions is instrumental to develop this process the reliability of which, consequently, strongly depends on the quality and completeness of the structural information. Recently, methods based on deep learning approaches critically improved the speed and accuracy of model building and are promising tools for accelerating the docking step. Here, we review the features of the available bioinformatic instruments and analyze the reports illustrating the result obtained with their application to optimize antibody fragments, and nanobodies in particular. Finally, the emerging trends and open questions are summarized.

Solvated interaction energy: from small-molecule to antibody drug design

Scoring functions are ubiquitous in structure-based drug design as an aid to predicting binding modes and estimating binding affinities. Ideally, a scoring function should be broadly applicable, obviating the need to recalibrate and refit its parameters for every new target and class of ligands. Traditionally, drugs have been small molecules, but in recent years biologics, particularly antibodies, have become an increasingly important if not dominant class of therapeutics. This makes the goal of having a transferable scoring function, i.e., one that spans the range of small-molecule to protein ligands, even more challenging. One such broadly applicable scoring function is the Solvated Interaction Energy (SIE), which has been developed and applied in our lab for the last 15 years, leading to several important applications. This physics-based method arose from efforts to understand the physics governing binding events, with particular care given to the role played by solvation. SIE has been used by us and many independent labs worldwide for virtual screening and discovery of novel small-molecule binders or optimization of known drugs. Moreover, without any retraining, it is found to be transferrable to predictions of antibody-antigen relative binding affinities and as accurate as functions trained on protein-protein binding affinities. SIE has been incorporated in conjunction with other scoring functions into ADAPT (Assisted Design of Antibody and Protein Therapeutics), our platform for affinity modulation of antibodies. Application of ADAPT resulted in the optimization of several antibodies with 10-to-100-fold improvements in binding affinity. Further applications included broadening the specificity of a single-domain antibody to be cross-reactive with virus variants of both SARS-CoV-1 and SARS-CoV-2, and the design of safer antibodies by engineering of a pH switch to make them more selective towards acidic tumors while sparing normal tissues at physiological pH.

Replica-exchange optimization of antibody fragments

In the framework of the rational design of macromolecules capable of binding to a specific target for biosensing applications, we here further develop an evolutionary protocol designed to optimize the binding affinity of protein binders. In particular we focus on the optimization of the binding portion of small antibody fragments known as nanobodies (or VHH) and choose the hen egg white lysozyme (HEWL) as our target. By implementing a replica exchange scheme for this optimization, we show that an initial hit is not needed and similar solutions can be found by either optimizing an already known anti-HEWL VHH or a randomly selected binder (here a VHH selective towards another macromolecule). While we believe that exhaustive searches of the mutation space are most appropriate when only few key residues have to be optimized, in case a lead binder is not available the proposed evolutionary algorithm should be instead the method of choice.

Epitope mapping of a blood-brain barrier crossing antibody targeting the cysteine-rich region of IGF1R using hydrogen-exchange mass spectrometry enabled by electrochemical reduction

Pathologies of the central nervous system impact a significant portion of our population, and the delivery of therapeutics for effective treatment is challenging. The insulin-like growth factor-1 receptor (IGF1R) has emerged as a target for receptor-mediated transcytosis, a process by which antibodies are shuttled across the blood-brain barrier (BBB). Here, we describe the biophysical characterization of VHH-IR4, a BBB-crossing single-domain antibody (sdAb). Binding was confirmed by isothermal titration calorimetry and an epitope was highlighted by surface plasmon resonance that does not overlap with the IGF-1 binding site or other known BBB-crossing sdAbs. The epitope was mapped with a combination of linear peptide scanning and hydrogen-deuterium exchange mass spectrometry (HDX-MS). IGF1R is large and heavily disulphide bonded, and comprehensive HDX analysis was achieved only through the use of online electrochemical reduction coupled with a multiprotease approach, which identified an epitope for VHH-IR4 within the cysteine-rich region (CRR) of IGF1R spanning residues W244-G265. This is the first report of an sdAb binding the CRR. We show that VHH-IR4 inhibits ligand induced auto-phosphorylation of IGF1R and that this effect is mediated by downstream conformational effects. Our results will guide the selection of antibodies with improved trafficking and optimized IGF1R binding characteristics.

Binding affinity prediction for antibody-protein antigen complexes: A machine learning analysis based on interface and surface areas

Specific antibodies can bind to protein antigens with high affinity and specificity, and this property makes them one of the best protein-based therapeutics. Accurate prediction of antibody‒protein antigen binding affinity is crucial for designing effective antibodies. The current predictive methods for protein‒protein binding affinity usually fail to predict the binding affinity of an antibody‒protein antigen complex with a comparable level of accuracy. Here, new models specific for antibody‒antigen binding affinity prediction are developed according to the different types of interface and surface areas present in antibody‒antigen complex. The contacts-based descriptors are also employed to construct or train different models specific for antibody‒protein antigen binding affinity prediction. The results of this study show that (i) the area-based descriptors are slightly better than the contacts-based descriptors in terms of the predictive power; (ii) the new models specific for antibody‒protein antigen binding affinity prediction are superior to the previously-used general models for predicting the protein‒protein binding affinities; (iii) the performances of the best area-based and contacts-based models developed in this work are better than the performances of a recently-developed graph-based model (i.e., CSM-AB) specific for antibody‒protein antigen binding affinity prediction. The new models developed in this work would not only help understand the mechanisms underlying antibody‒protein antigen interactions, but would also be of some applicable utility in the design and virtual screening of antibody-based therapeutics.

Optimizing Antibody-Antigen Binding Affinities with the ADAPT Platform

The ADAPT (Assisted Design of Antibody and Protein Therapeutics) platform guides the selection of mutants that improve/modulate the affinity of antibodies and other biologics. Predicted affinities are based on a consensus z-score from three scoring functions. Computational predictions are interleaved with experimental validation, significantly enhancing the robustness of the design and selection of mutants. A key step is an initial exhaustive virtual single-mutant scan that identifies hot spots and the mutations predicted to improve affinity. A small number of proposed single mutants are then produced and assayed. Only the validated single mutants (i.e., having improved affinity) are used to design double and higher-order mutants in subsequent rounds of design, avoiding the combinatorial explosion that arises from random mutagenesis. Typically, with a total of about 30-50 designed single, double, and triple mutants, affinity improvements of 10- to 100-fold are obtained.

Design of nanobody-based bispecific constructs by in silico affinity maturation and umbrella sampling simulations

Random mutagenesis is the natural opportunity for proteins to evolve and biotechnologically it has been exploited to create diversity and identify variants with improved characteristics in the mutant pools. Rational mutagenesis based on biophysical assumptions and supported by computational power has been proposed as a faster and more predictable strategy to reach the same aim. In this work we confirm that substantial improvements in terms of both affinity and stability of nanobodies can be obtained by using combinations of algorithms, even for binders with already high affinity and elevated thermal stability. Furthermore, in silico approaches allowed the development of an optimized bispecific construct able to bind simultaneously the two clinically relevant antigens TNF-α and IL-23 and, by means of its enhanced avidity, to inhibit effectively the apoptosis of TNF-α-sensitive L929 cells. The results revealed that salt bridges, hydrogen bonds, aromatic-aromatic and cation-pi interactions had a critical role in increasing affinity. We provided a platform for the construction of high-affinity bispecific constructs based on nanobodies that can have relevant applications for the control of all those biological mechanisms in which more than a single antigen must be targeted to increase the treatment effectiveness and avoid resistance mechanisms.

Prediction of Antibody-Antigen Binding via Machine Learning: Development of Data Sets and Evaluation of Methods

BACKGROUND

The mammalian immune system is able to generate antibodies against a huge variety of antigens, including bacteria, viruses, and toxins. The ultradeep DNA sequencing of rearranged immunoglobulin genes has considerable potential in furthering our understanding of the immune response, but it is limited by the lack of a high-throughput, sequence-based method for predicting the antigen(s) that a given immunoglobulin recognizes.

OBJECTIVE

As a step toward the prediction of antibody-antigen binding from sequence data alone, we aimed to compare a range of machine learning approaches that were applied to a collated data set of antibody-antigen pairs in order to predict antibody-antigen binding from sequence data.

METHODS

Data for training and testing were extracted from the Protein Data Bank and the Coronavirus Antibody Database, and additional antibody-antigen pair data were generated by using a molecular docking protocol. Several machine learning methods, including the weighted nearest neighbor method, the nearest neighbor method with the BLOSUM62 matrix, and the random forest method, were applied to the problem.

RESULTS

The final data set contained 1157 antibodies and 57 antigens that were combined in 5041 antibody-antigen pairs. The best performance for the prediction of interactions was obtained by using the nearest neighbor method with the BLOSUM62 matrix, which resulted in around 82% accuracy on the full data set. These results provide a useful frame of reference, as well as protocols and considerations, for machine learning and data set creation in the prediction of antibody-antigen binding.

CONCLUSIONS

Several machine learning approaches were compared to predict antibody-antigen interaction from protein sequences. Both the data set (in CSV format) and the machine learning program (coded in Python) are freely available for download on GitHub.

Structure-based dual affinity optimization of a SARS-CoV-1/2 cross-reactive single-domain antibody

The SARS coronavirus 2 (SARS-CoV-2) spike (S) protein binding to the human ACE2 receptor is the molecular event that initiates viral entry into host cells and leads to infection and virus replication. There is a need for agents blocking viral entry into host cells that are cross-reactive with emerging virus variants. VHH-72 is an anti-SARS-CoV-1 single-domain antibody that also exhibits cross-specificity with SARS-CoV-2 but with decreased binding affinity. Here we applied a structure-based approach to affinity-mature VHH-72 for the SARS-CoV-2 spike protein while retaining the original affinity for SARS-CoV-1. This was achieved by employing the computational platform ADAPT in a constrained dual-affinity optimization mode as a means of broadening specificity. Select mutants designed by ADAPT were formatted as fusions with a human IgG1-Fc fragment. These mutants demonstrated improved binding to the SARS-CoV-2 spike protein due to decreased dissociation rates. Functional testing for virus neutralization revealed improvements relative to the parental VHH72-Fc up to 10-fold using a SARS-CoV-2 pseudotyped lentivirus and 20-fold against the SARS-CoV-2 authentic live virus (Wuhan variant). Binding and neutralization improvements were maintained for some other SARS-CoV-2 variants currently in circulation. These improved VHH-72 mutants are predicted to establish novel interactions with the S antigen. They will be useful, alone or as fusions with other functional modules, in the global quest for treatments of COVID-19 infections.

Redesigning an antibody H3 loop by virtual screening of a small library of human germline-derived sequences

The design of superior biologic therapeutics, including antibodies and engineered proteins, involves optimizing their specific ability to bind to disease-related molecular targets. Previously, we developed and applied the Assisted Design of Antibody and Protein Therapeutics (ADAPT) platform for virtual affinity maturation of antibodies (Vivcharuk et al. in PLoS One 12(7):e0181490, 10.1371/journal.pone.0181490, 2017). However, ADAPT is limited to point mutations of hot-spot residues in existing CDR loops. In this study, we explore the possibility of wholesale replacement of the entire H3 loop with no restriction to maintain the parental loop length. This complements other currently published studies that sample replacements for the CDR loops L1, L2, L3, H1 and H2. Given the immense sequence space theoretically available to H3, we focused on the virtual grafting of over 5000 human germline-derived H3 sequences from the IGMT/LIGM database increasing the diversity of the sequence space when compared to using crystalized H3 loop sequences. H3 loop conformations are generated and scored to identify optimized H3 sequences. Experimental testing of high-ranking H3 sequences grafted into the framework of the bH1 antibody against human VEGF-A led to the discovery of multiple hits, some of which had similar or better affinities relative to the parental antibody. In over 75% of the tested designs, the re-designed H3 loop contributed favorably to overall binding affinity. The hits also demonstrated good developability attributes such as high thermal stability and no aggregation. Crystal structures of select re-designed H3 variants were solved and indicated that although some deviations from predicted structures were seen in the more solvent accessible regions of the H3 loop, they did not significantly affect predicted affinity scores.

Computational and Rational Design of Single-Chain Antibody against Tick-Borne Encephalitis Virus for Modifying Its Specificity

Tick-borne encephalitis virus (TBEV) causes 5−7 thousand cases of human meningitis and encephalitis annually. The neutralizing and protective antibody ch14D5 is a potential therapeutic agent. This antibody exhibits a high affinity for binding with the D3 domain of the glycoprotein E of the Far Eastern subtype of the virus, but a lower affinity for the D3 domains of the Siberian and European subtypes. In this study, a 2.2-fold increase in the affinity of single-chain antibody sc14D5 to D3 proteins of the Siberian and European subtypes of the virus was achieved using rational design and computational modeling. This improvement can be further enhanced in the case of the bivalent binding of the full-length chimeric antibody containing the identified mutation.

Application of an integrated computational antibody engineering platform to design SARS-CoV-2 neutralizers

As the COVID-19 pandemic continues to spread, hundreds of new initiatives including studies on existing medicines are running to fight the disease. To deliver a potentially immediate and lasting treatment to current and emerging SARS-CoV-2 variants, new collaborations and ways of sharing are required to create as many paths forward as possible. Here, we leverage our expertise in computational antibody engineering to rationally design/engineer three previously reported SARS-CoV neutralizing antibodies and share our proposal towards anti-SARS-CoV-2 biologics therapeutics. SARS-CoV neutralizing antibodies, m396, 80R and CR-3022 were chosen as templates due to their diversified epitopes and confirmed neutralization potency against SARS-CoV (but not SARS-CoV-2 except for CR3022). Structures of variable fragment (Fv) in complex with receptor binding domain (RBD) from SARS-CoV or SARS-CoV-2 were subjected to our established in silico antibody engineering platform to improve their binding affinity to SARS-CoV-2 and developability profiles. The selected top mutations were ensembled into a focused library for each antibody for further screening. In addition, we convert the selected binders with different epitopes into the trispecific format, aiming to increase potency and to prevent mutational escape. Lastly, to avoid antibody-induced virus activation or enhancement, we suggest application of NNAS and DQ mutations to the Fc region to eliminate effector functions and extend half-life.

Progress toward improved understanding of antibody maturation

Upon encountering an antigen, antibodies mature through various rounds of somatic mutations, resulting in higher affinities and specificities to the particular antigen. We review recent progress in four areas of antibody maturation studies. (1) Next-generation and single-cell sequencing have revolutionized the analysis of antibody repertoires by dramatically increasing the sequences available to study the state and evolution of the immune system. Computational methods, including machine learning tools, have been developed for reconstituting antibody clonal lineages and for general repertoire analysis. (2) The availability of X-ray structures, thermodynamic and kinetic data, and molecular dynamics simulations provide information on the biophysical mechanisms responsible for improved affinity. (3) In addition to improved binding to a specific antigen, providing affinity-independent diversity and self/nonself discrimination are fundamental functions of the immune system. Recent studies, including X-ray structures, yield improved understanding of both mechanisms. (4) Results from in vivo maturation help to develop methods of in vitro maturation to improve antibody properties for therapeutic applications, frequently combining computational and experimental approaches.

Structure-based engineering of pH-dependent antibody binding for selective targeting of solid-tumor microenvironment

Recent development of monoclonal antibodies as mainstream anticancer agents demands further optimization of their safety for use in humans. Potent targeting and/or effector activities on normal tissues is an obvious toxicity concern. Optimization of specific tumor targeting could be achieved by taking advantage of the extracellular acidity of solid tumors relative to normal tissues. Here, we applied a structure-based computational approach to engineer anti-human epidermal growth factor receptor 2 (Her2) antibodies with selective binding in the acidic tumor microenvironment. We used an affinity maturation platform in which dual-pH histidine-scanning mutagenesis was implemented for pH selectivity optimization. Testing of a small set of designs for binding to the recombinant Her2 ectodomain led to the identification of antigen-binding fragment (Fab) variants with the desired pH-dependent binding behavior. Binding selectivity toward acidic pH was improved by as much as 25-fold relative to the parental bH1-Fab. In vitro experiments on cells expressing intact Her2 confirmed that designed variants formatted as IgG1/k full-size antibodies have high affinity and inhibit the growth of tumor spheroids at a level comparable to that of the benchmark anti-Her2 antibody trastuzumab (Herceptin®) at acidic pH, whereas these effects were significantly reduced at physiological pH. In contrast, both Herceptin and the parental bH1 antibody exhibited strong cell binding and growth inhibition irrespective of pH. This work demonstrates the feasibility of computational optimization of antibodies for selective targeting of the acidic environment such as that found in many solid tumors.

Predicting antibody affinity changes upon mutations by combining multiple predictors

Antibodies are proteins working in our immune system with high affinity and specificity for target antigens, making them excellent tools for both biotherapeutic and bioengineering applications. The prediction of antibody affinity changes upon mutations (\documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$${{\Delta \Delta {\mathrm{G}}}}_{\mathrm{binding}}$$\end{document}ΔΔGbinding) is important for antibody engineering. Numerous computational methods have been proposed based on different approaches including molecular mechanics and machine learning. However, the accuracy by each individual predictor is not enough for efficient antibody development. In this study, we develop a new prediction method by combining multiple predictors based on machine learning. Our method was tested on the SiPMAB database, evaluating the Pearson’s correlation coefficient between predicted and experimental \documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$${{\Delta \Delta {\mathrm{G}}}}_{\mathrm{binding}}$$\end{document}ΔΔGbinding. Our method achieved higher accuracy (R = 0.69) than previous molecular mechanics or machine-learning based methods (R = 0.59) and the previous method using the average of multiple predictors (R = 0.64). Feature importance analysis indicated that the improved accuracy was obtained by combining predictors with different importance, which have different protocols for calculating energies and for generating mutant and unbound state structures. This study demonstrates that machine learning is a powerful framework for combining different approaches to predict antibody affinity changes.

In vitro affinity maturation to improve the efficacy of a hypoxia-inducible factor 1α single-domain intrabody

Affinity is an important property of therapeutic antibodies, so improving affinity is critical to the biological activity and clinical efficacy. An anti-HIF-1α nanobody, VHH212, was screened via a native ribosome display library with a 26.6 nM of KD value was used as the parent. In this paper, a Venn-intersection of multi-algorithms screening (VIMAS) strategy for computer-aided binding affinity prediction was designed. Homology modeling and protein docking methods were used to substitute the need for a crystal structure. Finally, a mutant with a 17.5-fold enhancement in binding affinity (1.52 nM) was obtained by using the VIMAS strategy. Furthermore, the biological activity of mutants was verified at the cellular level. Targeting HIF-1α can sensitize PDAC (pancreatic ductal adenocarcinoma) tumors to gemcitabine, which is a potential co-treatment method for pancreatic cancer patients. Our results showed that the cytotoxicity of gemcitabine on pancreatic cancer cell lines increased with the enhanced-affinity of an intrabody under combined treatment.

Recombinant expression of nanobodies and nanobody-derived immunoreagents

Antibody fragments for which the sequence is available are suitable for straightforward engineering and expression in both eukaryotic and prokaryotic systems. When produced as fusions with convenient tags, they become reagents which pair their selective binding capacity to an orthogonal function. Several kinds of immunoreagents composed by nanobodies and either large proteins or short sequences have been designed for providing inexpensive ready-to-use biological tools. The possibility to choose among alternative expression strategies is critical because the fusion moieties might require specific conditions for correct folding or post-translational modifications. In the case of nanobody production, the trend is towards simpler but reliable (bacterial) methods that can substitute for more cumbersome processes requiring the use of eukaryotic systems. The use of these will not disappear, but will be restricted to those cases in which the final immunoconstructs must have features that cannot be obtained in prokaryotic cells. At the same time, bacterial expression has evolved from the conventional procedure which considered exclusively the nanobody and nanobody-fusion accumulation in the periplasm. Several reports show the advantage of cytoplasmic expression, surface-display and secretion for at least some applications. Finally, there is an increasing interest to use as a model the short nanobody sequence for the development of in silico methodologies aimed at optimizing the yields, stability and affinity of recombinant antibodies.

•

There is an increasing request for immunoreagents based on nanobodies.

•

The multiplicity of their applications requires constructs with different structural complexity.

•

Alternative expression methods are necessary to achieve such structural requirements.

•

In silico optimization of nanobody biophysical characteristics becomes more and more reliable.

Binding and functional profiling of antibody mutants guides selection of optimal candidates as antibody drug conjugates

An increasingly appreciated conundrum in the discovery of antibody drug conjugates (ADCs) is that an antibody that was selected primarily for strong binding to its cancer target may not serve as an optimal ADC. In this study, we performed mechanistic cell-based experiments to determine the correlation between antibody affinity, avidity, internalization and ADC efficacy. We used structure-guided design to assemble a panel of antibody mutants with predicted Her2 affinities ranging from higher to lower relative to the parent antibody, Herceptin. These antibodies were ranked for binding via SPR and via flow-cytometry on high-Her2 SKOV3 cells and low-Her2 MCF7 cells, the latter acting as a surrogate for low-Her2 normal cells. A subpanel of variants, representative of different Her2-binding affinities (2 strong, 2 moderate and 3 weak), were further screened via high-content imaging for internalization efficacies in high versus low-Her2 cells. Finally, these antibodies were evaluated in ADC cytotoxicity screening assays (using DM1 and MMAE secondary antibodies) and as antibody-drug conjugates (DM1 and PNU159682). Our results identified specific but weak Her2-binding variants as optimal candidates for developing DM1 and PNU ADCs since they exhibited high potencies (low to sub-nM) in high-Her2 SKOV3 cells and low toxicities in low-Her2 cells. The 2 strong-affinity variants were highly potent in SKOV3 cells but also showed significant toxicities in low-Her2 cells and therefore are predicted to be toxic in normal tissues. Our findings show that pharmacological profiling of an antibody library in multiple binding and functional assays allows for selection of optimal ADCs.

Homology Modeling-Based in Silico Affinity Maturation Improves the Affinity of a Nanobody

Affinity maturation and rational design have a raised importance in the application of nanobody (VHH), and its unique structure guaranteed these processes quickly done in vitro. An anti-CD47 nanobody, Nb02, was screened via a synthetic phage display library with 278 nM of K_D value. In this study, a new strategy based on homology modeling and Rational Mutation Hotspots Design Protocol (RMHDP) was presented for building a fast and efficient platform for nanobody affinity maturation. A three-dimensional analytical structural model of Nb02 was constructed and then docked with the antigen, the CD47 extracellular domain (CD47^ext). Mutants with high binding affinity are predicted by the scoring of nanobody-antigen complexes based on molecular dynamics trajectories and simulation. Ultimately, an improved mutant with an 87.4-fold affinity (3.2 nM) and 7.36 °C higher thermal stability was obtained. These findings might contribute to computational affinity maturation of nanobodies via homology modeling using the recent advancements in computational power. The add-in of aromatic residues which formed aromatic-aromatic interaction plays a pivotal role in affinity and thermostability improvement. In a word, the methods used in this study might provide a reference for rapid and efficient in vitro affinity maturation of nanobodies.

Binding symmetry and surface flexibility mediate antibody self-association

Solution stability is an important factor in the optimization of engineered biotherapeutic candidates such as monoclonal antibodies because of its possible effects on manufacturability, pharmacology, efficacy and safety. A detailed atomic understanding of the mechanisms governing self-association of natively folded protein monomers is required to devise predictive tools to guide screening and re-engineering along the drug development pipeline. We investigated pairs of affinity-matured full-size antibodies and observed drastically different propensities to aggregate from variants differing by a single amino-acid. Biophysical testing showed that antigen-binding fragments (Fabs) from the aggregating antibodies also reversibly associated with equilibrium dissociation constants in the low-micromolar range. Crystal structures (PDB accession codes 6MXR, 6MXS, 6MY4, 6MY5) and bottom-up hydrogen-exchange mass spectrometry revealed that Fab self-association occurs in a symmetric mode that involves the antigen complementarity-determining regions. Subtle local conformational changes incurred upon point mutation of monomeric variants foster formation of complementary polar interactions and hydrophobic contacts to generate a dimeric Fab interface. Testing of popular in silico tools generally indicated low reliabilities for predicting the aggregation propensities observed. A structure-aggregation data set is provided here in order to stimulate further improvements of in silico tools for prediction of native aggregation. Incorporation of intermolecular docking, conformational flexibility, and short-range packing interactions may all be necessary features of the ideal algorithm.

Outlier Detection and Smoothing Process for Water Level Data Measured by Ultrasonic Sensor in Stream Flows

Water level data sets acquired by ultrasonic sensors in stream-scale channels exhibit relatively large numbers of outliers that are off the measurement range between the ultrasonic sensor and water surface, as well as data dispersion of approximately 2 cm due to random errors such as water waves. Therefore, this study develops a data processing algorithm for outlier removal and smoothing for water level data measured by ultrasonic sensors to consider these characteristics. The outlier removal process includes an initial cutoff process to remove outliers out of the measurement range and an outlier detection process using modified Z-scores based on the median absolute deviation (MAD) of a robust estimator. In addition, an exponentially weighted moving average (EWMA) method is applied to smooth the processed data. Sensitivity analyses are performed for factors that are subjectively set by the user, including the window size for the MAD outlier detection stage, the rejection criterion for the modified Z-score outlier removal stage, and the smoothing constant for the EWMA smoothing stage, based on four different water level data sets acquired by ultrasonic sensors in stream-scale experiments.

Experimentally guided computational antibody affinity maturation with de novo docking, modelling and rational design

Antibodies are an important class of therapeutics that have significant clinical impact for the treatment of severe diseases. Computational tools to support antibody drug discovery have been developing at an increasing rate over the last decade and typically rely upon a predetermined co-crystal structure of the antibody bound to the antigen for structural predictions. Here, we show an example of successful in silico affinity maturation of a hybridoma derived antibody, AB1, using just a homology model of the antibody fragment variable region and a protein-protein docking model of the AB1 antibody bound to the antigen, murine CCL20 (muCCL20). In silico affinity maturation, together with alanine scanning, has allowed us to fine-tune the protein-protein docking model to subsequently enable the identification of two single-point mutations that increase the affinity of AB1 for muCCL20. To our knowledge, this is one of the first examples of the use of homology modelling and protein docking for affinity maturation and represents an approach that can be widely deployed.

The adaptor protein Ste50 directly modulates yeast MAPK signaling specificity through differential connections of its RA domain

Discriminating among diverse environmental stimuli is critical for organisms to ensure their proper development, homeostasis, and survival. Saccharomyces cerevisiae regulates mating, osmoregulation, and filamentous growth using three different MAPK signaling pathways that share common components and therefore must ensure specificity. The adaptor protein Ste50 activates Ste11p, the MAP3K of all three modules. Its Ras association (RA) domain acts in both hyperosmolar and filamentous growth pathways, but its connection to the mating pathway is unknown. Genetically probing the domain, we found mutants that specifically disrupted mating or HOG-signaling pathways or both. Structurally these residues clustered on the RA domain, forming distinct surfaces with a propensity for protein–protein interactions. GFP fusions of wild-type (WT) and mutant Ste50p show that WT is localized to the shmoo structure and accumulates at the growing shmoo tip. The specifically pheromone response–defective mutants are severely impaired in shmoo formation and fail to localize ste50p, suggesting a failure of association and function of Ste50 mutants in the pheromone-signaling complex. Our results suggest that yeast cells can use differential protein interactions with the Ste50p RA domain to provide specificity of signaling during MAPK pathway activation.

ProPOSE: Direct Exhaustive Protein-Protein Docking with Side Chain Flexibility

Despite decades of development, protein-protein docking remains a largely unsolved problem. The main difficulties are the immense space spanned by the translational and rotational degrees of freedom and the prediction of the conformational changes of proteins upon binding. FFT is generally the preferred method to exhaustively explore the translation-rotation space at a fine grid resolution, albeit with the trade-off of approximating force fields with correlation functions. This work presents a direct search alternative that samples the states in Cartesian space at the same resolution and computational cost as standard FFT methods. Operating in real space allows the use of standard force field functional forms used in typical non-FFT methods as well as the implementation of strategies for focused exploration of conformational flexibility. Currently, a few misplaced side chains can cause docking programs to fail. This work specifically addresses the problem of side chain rearrangements upon complex formation. Based on the observation that most side chains retain their unbound conformation upon binding, each rigidly docked pose is initially scored ignoring up to a limited number of side chain overlaps which are resolved in subsequent repacking and minimization steps. On test systems where side chains are altered and backbones held in their bound state, this implementation provides significantly better native pose recovery and higher quality (lower RMSD) predictions when compared with five of the most popular docking programs. The method is implemented in the software program ProPOSE (Protein Pose Optimization by Systematic Enumeration).

Application of Assisted Design of Antibody and Protein Therapeutics (ADAPT) improves efficacy of a Clostridium difficile toxin A single-domain antibody

Assisted Design of Antibody and Protein Therapeutics (ADAPT) is an affinity maturation platform interleaving predictions and testing that was previously validated on monoclonal antibodies (mAbs). This study expands the applicability of ADAPT to single-domain antibodies (sdAbs), a promising class of recombinant antibody-based biologics. As a test case, we used the camelid sdAb A26.8, a V_HH that binds Clostridium difficile toxin A (TcdA) relatively weakly but displays a reasonable level of TcdA neutralization. ADAPT-guided A26.8 affinity maturation resulted in an improvement of one order of magnitude by point mutations only, reaching an equilibrium dissociation constant (K_D) of 2 nM, with the best binding mutants having similar or improved stabilities relative to the parent sdAb. This affinity improvement generated a 6-fold enhancement of efficacy at the cellular level; the A26.8 double-mutant T56R,T103R neutralizes TcdA cytotoxicity with an IC₅₀ of 12 nM. The designed mutants with increased affinities are predicted to establish novel electrostatic interactions with the antigen. Almost full additivity of mutation effects is observed, except for positively charged residues introduced at adjacent positions. Furthermore, analysis of false-positive predictions points to general directions for improving the ADAPT platform. ADAPT guided the efficacy enhancement of an anti-toxin sdAb, an alternative therapeutic modality for C. difficile.