Removing T-cell epitopes with computational protein design

A suite of designed protein cages using machine learning and protein fragment-based protocols

Designed protein cages and related materials provide unique opportunities for applications in biotechnology and medicine, but their creation remains challenging. Here, we apply computational approaches to design a suite of tetrahedrally symmetric, self-assembling protein cages. For the generation of docked conformations, we emphasize a protein fragment-based approach, while for sequence design of the de novo interface, a comparison of knowledge-based and machine learning protocols highlights the power and increased experimental success achieved using ProteinMPNN. An analysis of design outcomes provides insights for improving interface design protocols, including prioritizing fragment-based motifs, balancing interface hydrophobicity and polarity, and identifying preferred polar contact patterns. In all, we report five structures for seven protein cages, along with two structures of intermediate assemblies, with the highest resolution reaching 2.0 Å using cryo-EM. This set of designed cages adds substantially to the body of available protein nanoparticles, and to methodologies for their creation.

Engineering a Low-Immunogenic Mirror-Image VHH against Vascular Endothelial Growth Factor

Immunogenicity is a major caveat of protein therapeutics. In particular, the long-term administration of protein therapeutic agents leads to the generation of antidrug antibodies (ADAs), which reduce drug efficacy while eliciting adverse events. One promising solution to this issue is the use of mirror-image proteins consisting of d-amino acids, which are resistant to proteolytic degradation in immune cells. We have recently reported the chemical synthesis of the enantiomeric form of the variable domain of the antibody heavy chain (d-VHH). However, identifying mirror-image antibodies capable of binding to natural ligands remains challenging. In this study, we developed a novel screening platform to identify a d-VHH specific for vascular endothelial growth factor A (VEGF-A). We performed mirror-image screening of two newly constructed synthetic VHH libraries displayed on T7 phage and identified VHH sequences that effectively bound to the mirror-image VEGF-A target (d-VEGF-A). We subsequently synthesized a d-VHH candidate that preferentially bound the native VEGF-A (l-VEGF-A) with submicromolar affinity. Furthermore, immunization studies in mice demonstrated that this d-VHH elicited no ADAs, unlike its corresponding l-VHH. Our findings highlight the utility of this novel d-VHH screening platform in the development of protein therapeutics exhibiting both reduced immunogenicity and improved efficacy.

Synthetic transcription factor engineering for cell and gene therapy

Synthetic transcription factors (synTFs) that control beneficial transgene expression are an important method to increase the safety and efficacy of cell and gene therapy. Reliance on synTF components from non-human sources has slowed progress in the field because of concerns about immunogenicity and inducer drug properties. Recent advances in human-derived DNA-binding domains (DBDs) and transcriptional activation domains (TADs) paired with novel control modules responsive to clinically approved small molecules have poised the synTF field to overcome these hurdles. Advances include controllers inducible by autonomous signaling inputs and more complex, multi-input synTF circuits. Demonstrations of advanced control strategies with human-derived transcription factor components in clinically relevant vectors and in vivo models will facilitate progression into the clinic.

Graph-pMHC: graph neural network approach to MHC class II peptide presentation and antibody immunogenicity

Antigen presentation on MHC class II (pMHCII presentation) plays an essential role in the adaptive immune response to extracellular pathogens and cancerous cells. But it can also reduce the efficacy of large-molecule drugs by triggering an anti-drug response. Significant progress has been made in pMHCII presentation modeling due to the collection of large-scale pMHC mass spectrometry datasets (ligandomes) and advances in machine learning. Here, we develop graph-pMHC, a graph neural network approach to predict pMHCII presentation. We derive adjacency matrices for pMHCII using Alphafold2-multimer and address the peptide-MHC binding groove alignment problem with a simple graph enumeration strategy. We demonstrate that graph-pMHC dramatically outperforms methods with suboptimal inductive biases, such as the multilayer-perceptron-based NetMHCIIpan-4.0 (+20.17% absolute average precision). Finally, we create an antibody drug immunogenicity dataset from clinical trial data and develop a method for measuring anti-antibody immunogenicity risk using pMHCII presentation models. Our model increases receiver operating characteristic curve (ROC)-area under the ROC curve (AUC) by 2.57% compared to just filtering peptides by hits in OASis alone for predicting antibody drug immunogenicity.

Mirror-Image Single-Domain Antibody for a Novel Nonimmunogenic Drug Scaffold

Immunogenic responses by protein therapeutics often lead to reduced therapeutic effects and/or adverse effects via the generation of neutralizing antibodies and/or antidrug antibodies (ADA). Mirror-image proteins of the variable domain of the heavy chain of the heavy chain antibody (VHH) are potential novel protein therapeutics with high-affinity binding to target proteins and reduced immunogenicity because these mirror-image VHHs (d-VHHs) are less susceptible to proteolytic degradation in antigen-presenting cells (APCs). In this study, we investigated the preparation protocols of d-VHHs and their biological properties, including stereoselective target binding and immunogenicity. Initially, we established a facile synthetic process of two model VHHs [anti-GFP VHH and PMP12A2h1 (monomeric VHH of caplacizumab)] and their mirror-image proteins by three-step native chemical ligations (NCLs) from four peptide segments. The folded synthetic VHHs (l-anti-GFP VHH and l-PMP12A2h1) bound to the target proteins (EGFP and vWF-A1 domain, respectively), while their mirror-image proteins (d-anti-GFP VHH and d-PMP12A2h1) showed no binding to the native proteins. For biodistribution studies, l-VHH and d-VHH with single radioactive indium diethylenetriamine-pentaacid (111In-DTPA) labeling at the C-terminus were designed and synthesized by the established protocol. The distribution profiles were essentially similar between l-VHH and d-VHH, in which the probes accumulated in the kidney within 15 min after intravenous administration in mice, because of the small molecular size of VHHs. Comparative assessment of the immunogenicity responses revealed that d-VHH-induced levels of ADA generation were significantly lower than those of native VHH, regardless of the peptide sequences and administration routes. The resulting scaffold investigated should be applicable in the design of d-VHHs with various C-terminal CDR3 sequences, which can be identified by screening using display technologies.

Engineering and Expression Strategies for Optimization of L-Asparaginase Development and Production

Genetic engineering for heterologous expression has advanced in recent years. Model systems such as Escherichia coli, Bacillus subtilis and Pichia pastoris are often used as host microorganisms for the enzymatic production of L-asparaginase, an enzyme widely used in the clinic for the treatment of leukemia and in bakeries for the reduction of acrylamide. Newly developed recombinant L-asparaginase (L-ASNase) may have a low affinity for asparagine, reduced catalytic activity, low stability, and increased glutaminase activity or immunogenicity. Some successful commercial preparations of L-ASNase are now available. Therefore, obtaining novel L-ASNases with improved properties suitable for food or clinical applications remains a challenge. The combination of rational design and/or directed evolution and heterologous expression has been used to create enzymes with desired characteristics. Computer design, combined with other methods, could make it possible to generate mutant libraries of novel L-ASNases without costly and time-consuming efforts. In this review, we summarize the strategies and approaches for obtaining and developing L-ASNase with improved properties.

A Suite of Designed Protein Cages Using Machine Learning Algorithms and Protein Fragment-Based Protocols

Designed protein cages and related materials provide unique opportunities for applications in biotechnology and medicine, while methods for their creation remain challenging and unpredictable. In the present study, we apply new computational approaches to design a suite of new tetrahedrally symmetric, self-assembling protein cages. For the generation of docked poses, we emphasize a protein fragment-based approach, while for de novo interface design, a comparison of computational protocols highlights the power and increased experimental success achieved using the machine learning program ProteinMPNN. In relating information from docking and design, we observe that agreement between fragment-based sequence preferences and ProteinMPNN sequence inference correlates with experimental success. Additional insights for designing polar interactions are highlighted by experimentally testing larger and more polar interfaces. In all, using X-ray crystallography and cryo-EM, we report five structures for seven protein cages, with atomic resolution in the best case reaching 2.0 Å. We also report structures of two incompletely assembled protein cages, providing unique insights into one type of assembly failure. The new set of designed cages and their structures add substantially to the body of available protein nanoparticles, and to methodologies for their creation.

Potent and selective eradication of tumor cells by an EpCAM-targeted Ras-degrading enzyme

Despite decades of efforts, an urgent need remains to develop tumor cell-selective rat sarcoma (Ras)-targeting therapies that can treat patients with Ras-driven tumors. Here we report modular engineered proteins that degrade Ras selectively in tumor cells that overexpress the tumor cell marker epithelial cell adhesion molecule (EpCAM) by fusing the Ras degrader Ras-Rap1-specific endopeptidase with the translocation domain of the Pseudomonas aeruginosa exotoxin A (ETA) or diphtheria toxin (DT). Redirection to EpCAM is achieved by a designed ankyrin repeat protein. In two-dimensional tumor cell cultures, complete degradation of Ras proteins after 24 h was observed with EpCAM-targeted Ras degraders fused to ETA or DT in EpCAM-overexpressing MCF7 and HCT116 cells, with median inhibition concentration values at sub-nanomolar levels. The viability of EpCAM-low non-cancerous fibroblasts remained unaffected. In a three-dimensional (3D) tumor-on-a-chip system that mimics the natural tumor microenvironment, effective Ras degradation and selective toxicity toward tumor cells, particularly with the ETA-fused constructs, was determined on-chip. To conclude, we demonstrate the potential of modular engineered proteins to kill tumor cells highly selectively by simultaneously exploiting EpCAM as a tumor-specific cell surface molecule as well as Ras as an intracellular oncotarget in a 3D system mimicking the natural tumor microenvironment.

TransMHCII: a novel MHC-II binding prediction model built using a protein language model and an image classifier

The emergence of deep learning models such as AlphaFold2 has revolutionized the structure prediction of proteins. Nevertheless, much remains unexplored, especially on how we utilize structure models to predict biological properties. Herein, we present a method using features extracted from protein language models (PLMs) to predict the major histocompatibility complex class II (MHC-II) binding affinity of peptides. Specifically, we evaluated a novel transfer learning approach where the backbone of our model was interchanged with architectures designed for image classification tasks. Features extracted from several PLMs (ESM1b, ProtXLNet or ProtT5-XL-UniRef) were passed into image models (EfficientNet v2b0, EfficientNet v2m or ViT-16). The optimal pairing of the PLM and image classifier resulted in the final model TransMHCII, outperforming NetMHCIIpan 3.2 and NetMHCIIpan 4.0-BA on the receiver operating characteristic area under the curve, balanced accuracy and Jaccard scores. The architecture innovation may facilitate the development of other deep learning models for biological problems.

[Approaches for improving L-asparaginase expression in heterologous systems]

L-asparaginase (EC 3.5.1.1) is one of the most demanded enzymes used in the pharmaceutical industry as a drug and in the food industry to prevent the formation of toxic acrylamide. Researchers aimed to improve specific activity and reduce side effects to create safer and more potent enzyme products. However, protein modifications and heterologous expression remain problematic in the production of asparaginases from different species. Heterologous expression in optimized producer strains is rationally organized; therefore, modified and heterologous protein expression is enhanced, which is the main strategy in the production of asparaginase. This strategy solves several problems: incorrect protein folding, metabolic load on the producer strain and codon misreading, which affects translation and final protein domains, leading to a decrease in catalytic activity. The main approaches developed to improve the heterologous expression of L-asparaginases are considered in this paper.

A versatile Halo- and SNAP-tagged BMP/TGFβ receptor library for quantification of cell surface ligand binding

TGFβs, BMPs and Activins regulate numerous developmental and homeostatic processes and signal through hetero-tetrameric receptor complexes composed of two types of serine/threonine kinase receptors. Each of the 33 different ligands possesses unique affinities towards specific receptor types. However, the lack of specific tools hampered simultaneous testing of ligand binding towards all BMP/TGFβ receptors. Here we present a N-terminally Halo- and SNAP-tagged TGFβ/BMP receptor library to visualize receptor complexes in dual color. In combination with fluorescently labeled ligands, we established a Ligand Surface Binding Assay (LSBA) for optical quantification of receptor-dependent ligand binding in a cellular context. We highlight that LSBA is generally applicable to test (i) binding of different ligands such as Activin A, TGFβ1 and BMP9, (ii) for mutant screens and (iii) evolutionary comparisons. This experimental set-up opens opportunities for visualizing ligand-receptor binding dynamics, essential to determine signaling specificity and is easily adaptable for other receptor signaling pathways.

PITHA: A Webtool to Predict Immunogenicity for Humanized and Fully Human Therapeutic Antibodies

Immunogenicity is an important concern to therapeutic antibodies during antibody design and development. Based on the co-crystal structures of idiotypic antibodies and their antibodies, one can see that anti-idiotypic antibodies usually bind the complementarity-determining regions (CDR) of idiotypic antibodies. Sequence and structural features, such as cavity volume at the CDR region and hydrophobicity of CDR-H3 loop region, were identified for distinguishing immunogenic antibodies from non-immunogenic antibodies. These features were integrated together with a machine learning platform to predict immunogenicity for humanized and fully human therapeutic antibodies (PITHA). This method achieved an accuracy of 83% in a leave-one-out experiment for 29 therapeutic antibodies with available crystal structures. The web server of this method is accessible at http://mabmedicine.com/PITHA or http://sysbio.unl.edu/PITHA . This method, as a step of computer-aided antibody design, helps evaluate the safety of new therapeutic antibody, which can save time and money during the therapeutic antibody development.

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.

CD4+ T-Cell Epitope Prediction by Combined Analysis of Antigen Conformational Flexibility and Peptide-MHCII Binding Affinity

Antigen processing in the class II MHC pathway depends on conventional proteolytic enzymes, potentially acting on antigens in native-like conformational states. CD4+ epitope dominance arises from a competition among antigen folding, proteolysis, and MHCII binding. Protease-sensitive sites, linear antibody epitopes, and CD4+ T-cell epitopes were mapped in plague vaccine candidate F1-V to evaluate the various contributions to CD4+ epitope dominance. Using X-ray crystal structures, antigen processing likelihood (APL) predicts CD4+ epitopes with significant accuracy for F1-V without considering peptide-MHCII binding affinity. We also show that APL achieves excellent performance over two benchmark antigen sets. The profiles of conformational flexibility derived from the X-ray crystal structures of the F1-V proteins, Caf1 and LcrV, were similar to the biochemical profiles of linear antibody epitope reactivity and protease sensitivity, suggesting that the role of structure in proteolysis was captured by the analysis of the crystal structures. The patterns of CD4+ T-cell epitope dominance in C57BL/6, CBA, and BALB/c mice were compared to epitope predictions based on APL, MHCII binding, or both. For a sample of 13 diverse antigens, the accuracy of epitope prediction by the combination of APL and I-A^b-MHCII-peptide affinity reached 36%. When MHCII allele specificity was also diverse, such as in human immunity, prediction of dominant epitopes by APL alone reached 42% when using a stringent scoring threshold. Because dominant CD4+ epitopes tend to occur in conformationally stable antigen domains, crystal structures typically are available for analysis by APL, and thus, the requirement for a crystal structure is not a severe limitation.

Key aspects of the past 30 years of protein design

Proteins are the workhorse of life. They are the building infrastructure of living systems; they are the most efficient molecular machines known, and their enzymatic activity is still unmatched in versatility by any artificial system. Perhaps proteins' most remarkable feature is their modularity. The large amount of information required to specify each protein's function is analogically encoded with an alphabet of just ∼20 letters. The protein folding problem is how to encode all such information in a sequence of 20 letters. In this review, we go through the last 30 years of research to summarize the state of the art and highlight some applications related to fundamental problems of protein evolution.

Multiple-Allele MHC Class II Epitope Engineering by a Molecular Dynamics-Based Evolution Protocol

Epitopes that bind simultaneously to all human alleles of Major Histocompatibility Complex class II (MHC II) are considered one of the key factors for the development of improved vaccines and cancer immunotherapies. To engineer MHC II multiple-allele binders, we developed a protocol called PanMHC-PARCE, based on the unsupervised optimization of the epitope sequence by single-point mutations, parallel explicit-solvent molecular dynamics simulations and scoring of the MHC II-epitope complexes. The key idea is accepting mutations that not only improve the affinity but also reduce the affinity gap between the alleles. We applied this methodology to enhance a Plasmodium vivax epitope for multiple-allele binding. In vitro rate-binding assays showed that four engineered peptides were able to bind with improved affinity toward multiple human MHC II alleles. Moreover, we demonstrated that mice immunized with the peptides exhibited interferon-gamma cellular immune response. Overall, the method enables the engineering of peptides with improved binding properties that can be used for the generation of new immunotherapies.

Targeted intracellular delivery of Cas13 and Cas9 nucleases using bacterial toxin-based platforms

Targeted delivery of therapeutic proteins toward specific cells and across cell membranes remains major challenges. Here, we develop protein-based delivery systems utilizing detoxified single-chain bacterial toxins such as diphtheria toxin (DT) and botulinum neurotoxin (BoNT)-like toxin, BoNT/X, as carriers. The system can deliver large protein cargoes including Cas13a, CasRx, Cas9, and Cre recombinase into cells in a receptor-dependent manner, although delivery of ribonucleoproteins containing guide RNAs is not successful. Delivery of Cas13a and CasRx, together with guide RNA expression, reduces mRNAs encoding GFP, SARS-CoV-2 fragments, and endogenous proteins PPIB, KRAS, and CXCR4 in multiple cell lines. Delivery of Cre recombinase modifies the reporter loci in cells. Delivery of Cas9, together with guide RNA expression, generates mutations at the targeted genomic sites in cell lines and induced pluripotent stem cell (iPSC)-derived human neurons. These findings establish modular delivery systems based on single-chain bacterial toxins for delivery of membrane-impermeable therapeutics into targeted cells.

Accelerating therapeutic protein design

Protein structures provide for defined microenvironments that can support complex pharmacological functions, otherwise unachievable by small molecules. The advent of therapeutic proteins has thus greatly broadened the range of manageable disorders. Leveraging the knowledge and recent advances in de novo protein design methods has the prospect of revolutionizing how protein drugs are discovered and developed. This review lays out the main challenges facing therapeutic proteins discovery and development, and how present and future advancements of protein design can accelerate the protein drug pipelines.

Using a Prime-Boost Vaccination Strategy That Proved Effective for High Resolution Epitope Mapping to Characterize the Elusive Immunogenicity of Survivin

Simple Summary

The generation of tumor-specific T cells remains a pillar of modern cancer immunotherapy. Exogenous vaccines often rely on targeting tumor-associated antigens. The anti-apoptotic protein survivin has been deemed a high priority target due to its overexpression in a wide variety of tumor types. To support the analysis of tumor-associated T cell responses, optimization of epitope mapping would be valuable. A heterologous prime-boost vaccination strategy was designed to target survivin to induce anti-tumor immune responses. However, survivin-specific T cell responses could not be detected in mice. Potential mechanisms to explain this failure were explored. To confirm the robustness of the vaccination platform, enhanced green fluorescent protein (eGFP) was targeted since it has been defined as a protein with relatively low immunogenicity. In this context the vaccination strategy uncovered novel T cell epitopes from eGFP in two strains of mice. This research highlighted the utility of the vaccine platform to triage potential target antigens based on their immunogenicity.

Abstract

Survivin is a member of the inhibitor of apoptosis family of proteins and has been reported to be highly expressed in a variety of cancer types, making it a high priority target for cancer vaccination. We previously described a heterologous prime-boost strategy using a replication-deficient adenovirus, followed by an oncolytic rhabdovirus that generates unprecedented antigen-specific T cell responses. We engineered each vector to express a mutated version of full-length murine survivin. We first sought to uncover the complete epitope map for survivin-specific T cell responses in C57BL/6 and BALB/c mice by flow cytometry. However, no T cell responses were detected by intracellular cytokine staining after re-stimulation of T cells. Survivin has been found to be expressed by activated T cells, which could theoretically cause T cell-mediated killing of activated T cells, known as fratricide. We were unable to recapitulate this phenomenon in experiments. Interestingly, the inactivated survivin construct has been previously shown to directly kill tumor cells in vitro. However, there was no evidence in our models of induction of death in antigen-presenting cells due to treatment with a survivin-expressing vector. Using the same recombinant virus-vectored prime-boost strategy targeting the poorly immunogenic enhanced green fluorescent protein proved to be a highly sensitive method for mapping T cell epitopes, particularly in the context of identifying novel epitopes recognized by CD4⁺ T cells. Overall, these results suggested there may be unusually robust tolerance to survivin in commonly used mouse strains that cannot be broken, even when using a particularly potent vaccination platform. However, the vaccination method shows great promise as a strategy for identifying novel and subdominant T cell epitopes.

Polyethylene Glycol Immunogenicity: Theoretical, Clinical, and Practical Aspects of Anti-Polyethylene Glycol Antibodies

Polyethylene glycol (PEG) is a flexible, hydrophilic simple polymer that is physically attached to peptides, proteins, nucleic acids, liposomes, and nanoparticles to reduce renal clearance, block antibody and protein binding sites, and enhance the half-life and efficacy of therapeutic molecules. Some naïve individuals have pre-existing antibodies that can bind to PEG, and some PEG-modified compounds induce additional antibodies against PEG, which can adversely impact drug efficacy and safety. Here we provide a framework to better understand PEG immunogenicity and how antibodies against PEG affect pegylated drug and nanoparticles. Analysis of published studies reveals rules for predicting accelerated blood clearance of pegylated medicine and therapeutic liposomes. Experimental studies of anti-PEG antibody binding to different forms, sizes, and immobilization states of PEG are also provided. The widespread use of SARS-CoV-2 RNA vaccines that incorporate PEG in lipid nanoparticles make understanding possible effects of anti-PEG antibodies on pegylated medicines even more critical.

Guaranteed Diversity and Optimality in Cost Function Network Based Computational Protein Design Methods

Proteins are the main active molecules of life. Although natural proteins play many roles, as enzymes or antibodies for example, there is a need to go beyond the repertoire of natural proteins to produce engineered proteins that precisely meet application requirements, in terms of function, stability, activity or other protein capacities. Computational Protein Design aims at designing new proteins from first principles, using full-atom molecular models. However, the size and complexity of proteins require approximations to make them amenable to energetic optimization queries. These approximations make the design process less reliable, and a provable optimal solution may fail. In practice, expensive libraries of solutions are therefore generated and tested. In this paper, we explore the idea of generating libraries of provably diverse low-energy solutions by extending cost function network algorithms with dedicated automaton-based diversity constraints on a large set of realistic full protein redesign problems. We observe that it is possible to generate provably diverse libraries in reasonable time and that the produced libraries do enhance the Native Sequence Recovery, a traditional measure of design methods reliability.

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.

A human antibody of potent efficacy against SARS-CoV-2 in rhesus macaques showed strong blocking activity to B.1.351

Severe acute respiratory syndrome coronavirus-2 (SARS-CoV-2), which causes coronavirus disease-2019 (COVID-19), interacts with the host cell receptor angiotensin-converting enzyme 2 (hACE2) via its spike 1 protein during infection. After the virus sequence was published, we identified two potent antibodies against the SARS-CoV-2 receptor binding domain (RBD) from antibody libraries using a phage-to-yeast (PtY) display platform in only 10 days. Our lead antibody JMB2002, now in a Phase 1 clinical trial (ChiCTR2100042150), showed broad-spectrum in vitro blocking activity against hACE2 binding to the RBD of multiple SARS-CoV-2 variants, including B.1.351 that was reportedly much more resistant to neutralization by convalescent plasma, vaccine sera and some clinical-stage neutralizing antibodies. Furthermore, JMB2002 has demonstrated complete prophylactic and potent therapeutic efficacy in a rhesus macaque disease model. Prophylactic and therapeutic countermeasure intervention of SARS-CoV-2 using JMB2002 would likely slow down the transmission of currently emerged SARS-CoV-2 variants and result in more efficient control of the COVID-19 pandemic.

Computational design of SARS-CoV-2 spike glycoproteins to increase immunogenicity by T cell epitope engineering

The development of effective and safe vaccines is the ultimate way to efficiently stop the ongoing COVID-19 pandemic, which is caused by severe acute respiratory syndrome coronavirus 2 (SARS-CoV-2). Built on the fact that SARS-CoV-2 utilizes the association of its Spike (S) protein with the human angiotensin-converting enzyme 2 (ACE2) receptor to invade host cells, we computationally redesigned the S protein sequence to improve its immunogenicity and antigenicity. Toward this purpose, we extended an evolutionary protein design algorithm, EvoDesign, to create thousands of stable S protein variants that perturb the core protein sequence but keep the surface conformation and B cell epitopes. The T cell epitope content and similarity scores of the perturbed sequences were calculated and evaluated. Out of 22,914 designs with favorable stability energy, 301 candidates contained at least two pre-existing immunity-related epitopes and had promising immunogenic potential. The benchmark tests showed that, although the epitope restraints were not included in the scoring function of EvoDesign, the top S protein design successfully recovered 31 out of the 32 major histocompatibility complex (MHC)-II T cell promiscuous epitopes in the native S protein, where two epitopes were present in all seven human coronaviruses. Moreover, the newly designed S protein introduced nine new MHC-II T cell promiscuous epitopes that do not exist in the wildtype SARS-CoV-2. These results demonstrated a new and effective avenue to enhance a target protein's immunogenicity using rational protein design, which could be applied for new vaccine design against COVID-19 and other pathogens.

Deimmunization of protein therapeutics - Recent advances in experimental and computational epitope prediction and deletion

Biotherapeutics, and antimicrobial proteins in particular, are of increasing interest for human medicine. An important challenge in the development of such therapeutics is their potential immunogenicity, which can induce production of anti-drug-antibodies, resulting in altered pharmacokinetics, reduced efficacy, and potentially severe anaphylactic or hypersensitivity reactions. For this reason, the development and application of effective deimmunization methods for protein drugs is of utmost importance. Deimmunization may be achieved by unspecific shielding approaches, which include PEGylation, fusion to polypeptides (e.g., XTEN or PAS), reductive methylation, glycosylation, and polysialylation. Alternatively, the identification of epitopes for T cells or B cells and their subsequent deletion through site-directed mutagenesis represent promising deimmunization strategies and can be accomplished through either experimental or computational approaches. This review highlights the most recent advances and current challenges in the deimmunization of protein therapeutics, with a special focus on computational epitope prediction and deletion tools.

Low-resolution structures of modular nanotransporters shed light on their functional activity

Modular nanotransporters (MNTs) are multifunctional chimeric polypeptides for the multistep transport of locally acting cytotoxic agents into the nuclei of cancer target cells. MNTs consist of several polypeptide domains (functional modules) for the recognition of a cell-surface internalizable receptor, pH-dependent endosomal escape and subsequent transport into the nucleus through the nuclear pores. MNTs are a promising means for cancer treatment. As has been shown previously, all of the modules of MNTs retain their functionalities. Despite their importance, there is no structural information available about these chimeric polypeptides, which hampers the creation of new MNT variants. Here, a low-resolution 3D structure of an MNT is presented which was obtained by atomic force microscopy, transmission electron microscopy and small-angle X-ray scattering coupled to size-exclusion chromatography. The data suggest that the MNT can adopt two main conformations, but in both conformations the protein N- and C-termini are distanced and do not influence each other. The change in the MNT conformation during acidification of the medium was also studied. It was shown that the fraction of the elongated conformation increases upon acidification. The results of this work will be useful for the development of MNTs that are suitable for clinical trials and possible therapeutic applications.

Structure-Based Modeling of SARS-CoV-2 Peptide/HLA-A02 Antigens

SARS-CoV-2-specific CD4 and CD8 T cells have been shown to be present in individuals with acute, mild, and asymptomatic Coronavirus disease (COVID-19). Toward the development of diagnostic and therapeutic tools to fight COVID-19, it is important to predict and characterize T cell epitopes expressed by SARS-CoV-2. Here, we use RosettaMHC, a comparative modeling approach which leverages existing structures of peptide/MHC complexes available in the Protein Data Bank, to derive accurate 3D models for putative SARS-CoV-2 CD8 epitopes. We outline an application of our method to model 8-10 residue epitopic peptides predicted to bind to the common allele HLA-A*02:01, and we make our models publicly available through an online database (https://rosettamhc.chemistry.ucsc.edu). We further compare electrostatic surfaces with models of homologous peptide/HLA-A*02:01 complexes from human common cold coronavirus strains to identify epitopes which may be recognized by a shared pool of cross-reactive TCRs. As more detailed studies on antigen-specific T cell recognition become available, RosettaMHC models can be used to understand the link between peptide/HLA complex structure and surface chemistry with immunogenicity, in the context of SARS-CoV-2 infection.

Mitigation of T-cell dependent immunogenicity by reengineering factor VIIa analogue

Vatreptacog alfa (VA), a recombinant activated human factor VII (rFVIIa) variant with 3 amino acid substitutions, was developed to provide increased procoagulant activity in hemophilia patients with inhibitors to factor VIII or factor IX. In phase 3 clinical trials, changes introduced during the bioengineering of VA resulted in the development of undesired anti-drug antibodies in some patients, leading to the termination of a potentially promising therapeutic protein product. Here, we use preclinical biomarkers associated with clinical immunogenicity to validate our deimmunization strategy applied to this bioengineered rFVIIa analog. The reengineered rFVIIa analog variants retained increased intrinsic thrombin generation activity but did not elicit T-cell responses in peripheral blood mononuclear cells isolated from 50 HLA typed subjects representing the human population. Our algorithm, rational immunogenicity determination, offers a broadly applicable deimmunizing strategy for bioengineered proteins.

Prediction of immunogenicity for humanized and full human therapeutic antibodies

Immunogenicity is an important concern for therapeutic antibodies during drug development. By analyzing co-crystal structures of idiotypic antibodies and their antibodies, we found that anti-idiotypic antibodies usually bind the Complementarity Determining Regions (CDR) of idiotypic antibodies. Sequence and structural features were identified for distinguishing immunogenic antibodies from non-immunogenic antibodies. For example, non-immunogenic antibodies have a significantly larger cavity volume at the CDR region and a more hydrophobic CDR-H3 loop than immunogenic antibodies. Antibodies containing no Gly at the turn of CDR-H2 loop are often immunogenic. We integrated these features together with a machine learning platform to Predict Immunogenicity for humanized and full human THerapeutic Antibodies (PITHA). This method achieved an accuracy of 83% in leave-one-out experiment for 29 therapeutic antibodies with available crystal structures. The accuracy decreased to 65% for 23 test antibodies with modeled structures, because their crystal structures were not available, and the prediction was made with modeled structures. The server of this method is accessible at http://mabmedicine.com/PITHA.

Driving CARs with alternative navigation tools - the potential of engineered binding scaffolds

T cells that are genetically engineered to express chimeric antigen receptors (CAR T cells) have shown impressive clinical efficacy against B‐cell malignancies. In contrast to these highly potent CD19‐targeting CAR T cells, many of those directed against other tumor entities and antigens currently suffer from several limitations. For example, it has been demonstrated that many scFvs used as antigen‐binding domains in CARs show some degree of oligomerization, which leads to tonic signaling, T cell exhaustion, and poor performance in vivo. Therefore, in many cases alternatives to scFvs would be beneficial. Fortunately, due to the development of powerful protein engineering technologies, also non‐immunoglobulin‐based scaffolds can be engineered to specifically recognize antigens, thus eliminating the historical dependence on antibody‐based binding domains. Here, we discuss the advantages and disadvantages of such engineered binding scaffolds, in particular with respect to their application in CARs. We review recent studies, collectively showing that there is no functional or biochemical aspect that necessitates the use of scFvs in CARs. Instead, antigen recognition can also be mediated efficiently by engineered binding scaffolds, as well as natural ligands or receptors fused to the CAR backbone. Finally, we critically discuss the risk of immunogenicity and show that the extent of nonhuman amino acid stretches in engineered scaffolds—even in those based on nonhuman proteins—is more similar to humanized scFvs than might be anticipated. Together, we expect that engineered binding scaffolds and natural ligands and receptors will be increasingly used for the design of CAR T cells.

Rational Design of SARS-CoV-2 Spike Glycoproteins To Increase Immunogenicity By T Cell Epitope Engineering

The current COVID-19 pandemic caused by SARS-CoV-2 has resulted in millions of confirmed cases and thousands of deaths globally. Extensive efforts and progress have been made to develop effective and safe vaccines against COVID-19. A primary target of these vaccines is the SARS-CoV-2 spike (S) protein, and many studies utilized structural vaccinology techniques to either stabilize the protein or fix the receptor-binding domain at certain states. In this study, we extended an evolutionary protein design algorithm, EvoDesign, to create thousands of stable S protein variants without perturbing the surface conformation and B cell epitopes of the S protein. We then evaluated the mutated S protein candidates based on predicted MHC-II T cell promiscuous epitopes as well as the epitopes' similarity to human peptides. The presented strategy aims to improve the S protein's immunogenicity and antigenicity by inducing stronger CD4 T cell response while maintaining the protein's native structure and function. The top EvoDesign S protein candidate (Design-10705) recovered 31 out of 32 MHC-II T cell promiscuous epitopes in the native S protein, in which two epitopes were present in all seven human coronaviruses. This newly designed S protein also introduced nine new MHC-II T cell promiscuous epitopes and showed high structural similarity to its native conformation. The proposed structural vaccinology method provides an avenue to rationally design the antigen's structure with increased immunogenicity, which could be applied to the rational design of new COVID-19 vaccine candidates.

Quantifying the nativeness of antibody sequences using long short-term memory networks

Antibodies often undergo substantial engineering en route to the generation of a therapeutic candidate with good developability properties. Characterization of antibody libraries has shown that retaining native-like sequence improves the overall quality of the library. Motivated by recent advances in deep learning, we developed a bi-directional long short-term memory (LSTM) network model to make use of the large amount of available antibody sequence information, and use this model to quantify the nativeness of antibody sequences. The model scores sequences for their similarity to naturally occurring antibodies, which can be used as a consideration during design and engineering of libraries. We demonstrate the performance of this approach by training a model on human antibody sequences and show that our method outperforms other approaches at distinguishing human antibodies from those of other species. We show the applicability of this method for the evaluation of synthesized antibody libraries and humanization of mouse antibodies.

Macromolecular modeling and design in Rosetta: recent methods and frameworks

The Rosetta software for macromolecular modeling, docking and design is extensively used in laboratories worldwide. During two decades of development by a community of laboratories at more than 60 institutions, Rosetta has been continuously refactored and extended. Its advantages are its performance and interoperability between broad modeling capabilities. Here we review tools developed in the last 5 years, including over 80 methods. We discuss improvements to the score function, user interfaces and usability. Rosetta is available at http://www.rosettacommons.org.

Immunogenicity of Immunotoxins Containing Pseudomonas Exotoxin A: Causes, Consequences, and Mitigation

Immunotoxins are cytolytic fusion proteins developed for cancer therapy, composed of an antibody fragment that binds to a cancer cell and a protein toxin fragment that kills the cell. Pseudomonas exotoxin A (PE) is a potent toxin that is used for the killing moiety in many immunotoxins. Moxetumomab Pasudotox (Lumoxiti) contains an anti-CD22 Fv and a 38 kDa portion of PE. Lumoxiti was discovered in the Laboratory of Molecular Biology at the U.S. National Cancer Institute and co-developed with Medimmune/AstraZeneca to treat hairy cell leukemia. In 2018 Lumoxiti was approved by the US Food and Drug Administration for the treatment of drug-resistant Hairy Cell Leukemia. Due to the bacterial origin of the killing moiety, immunotoxins containing PE are highly immunogenic in patients with normal immune systems, but less immunogenic in patients with hematologic malignancies, whose immune systems are often compromised. LMB-100 is a de-immunized variant of the toxin with a humanized antibody that targets mesothelin and a PE toxin that was rationally designed for diminished reactivity with antibodies and B cell receptors. It is now being evaluated in clinical trials for the treatment of mesothelioma and pancreatic cancer and is showing somewhat diminished immunogenicity compared to its un modified parental counterpart. Here we review the immunogenicity of the original and de-immunized PE immunotoxins in mice and patients, the development of anti-drug antibodies (ADAs), their impact on drug availability and their effect on clinical efficacy. Efforts to mitigate the immunogenicity of immunotoxins and its impact on immunogenicity will be described including rational design to identify, remove, or suppress B cell or T cell epitopes, and combination of immunotoxins with immune modulating drugs.

The advent of de novo proteins for cancer immunotherapy

Engineered proteins are revolutionizing immunotherapy, but advances are still needed to harness their full potential. Traditional protein engineering methods use naturally existing proteins as a starting point, and therefore, are intrinsically limited to small alterations of a protein's natural structure and function. Conversely, computational de novo protein design is free of such limitation, and can produce a virtually infinite number of novel protein sequences, folds, and functions. Recently, we used de novo protein engineering to create Neoleukin-2/15 (Neo-2/15), a protein mimetic of the function of both interleukin-2 (IL-2) and interleukin-15 (IL-15). To our knowledge, Neo-2/15 is the first de novo protein with immunotherapeutic activity, and in murine cancer models, it has demonstrated enhanced therapeutic potency and reduced toxicity compared to IL-2. De novo protein design is already showcasing its tremendous potential for driving the next wave of protein-based therapeutics that are explicitly engineered to treat disease.

Structure-based modeling of SARS-CoV-2 peptide/HLA-A02 antigens

As a first step toward the development of diagnostic and therapeutic tools to fight the Coronavirus disease (COVID-19), it is important to characterize CD8+ T cell epitopes in the SARS-CoV-2 peptidome that can trigger adaptive immune responses. Here, we use RosettaMHC, a comparative modeling approach which leverages existing high-resolution X-ray structures from peptide/MHC complexes available in the Protein Data Bank, to derive physically realistic 3D models for high-affinity SARS-CoV-2 epitopes. We outline an application of our method to model 439 9mer and 279 10mer predicted epitopes displayed by the common allele HLA-A*02:01, and we make our models publicly available through an online database ( https://rosettamhc.chemistry.ucsc.edu ). As more detailed studies on antigen-specific T cell recognition become available, RosettaMHC models of antigens from different strains and HLA alleles can be used as a basis to understand the link between peptide/HLA complex structure and surface chemistry with immunogenicity, in the context of SARS-CoV-2 infection.

In silico identification and modification of T cell epitopes in pertussis antigens associated with tolerance

The resurgence of whooping cough since the introduction of acellular (protein) vaccines has led to a renewed interest in the development of improved pertussis vaccines; Outer Membrane Vesicles (OMVs) carrying pertussis antigens have emerged as viable candidates. An in silico immunogenicity screen was carried out on 49 well-known Bordetella pertussis proteins in order to better understand their potential role toward the efficacy of pertussis OMVs for vaccine design; seven proteins were identified as being good candidates for including in optimized cellular and acellular pertussis vaccines. We then screened these antigens for putative tolerance-inducing sequences, as proteins with reduced tolerogenicity have improved vaccine potency in preclinical models. We used specialized homology tools (JanusMatrix) to identify peptides in the proteins that were cross-reactive with human sequences. Four of the 19 identified cross-reactive peptides were detolerized in silico using a separate tool, OptiMatrix, which disrupted the potential of these peptides to bind to human HLA and murine MHC. Four selected cross-reactive peptides and their detolerized variants were synthesized and their binding to a set of eight common HLA class II alleles was assessed in vitro. Reduced binding affinity to HLA class II was observed for the detolerized variants compared to the wild-type peptides, highlighting the potential of this approach for designing more efficacious pertussis vaccines.

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.

Multifunctional CRISPR-Cas9 with engineered immunosilenced human T cell epitopes

The CRISPR-Cas9 system has raised hopes for developing personalized gene therapies for complex diseases. Its application for genetic and epigenetic therapies in humans raises concerns over immunogenicity of the bacterially derived Cas9 protein. Here we detect antibodies to Streptococcus pyogenes Cas9 (SpCas9) in at least 5% of 143 healthy individuals. We also report pre-existing human CD8+T cell immunity in the majority of healthy individuals screened. We identify two immunodominant SpCas9 T cell epitopes for HLA-A*02:01 using an enhanced prediction algorithm that incorporates T cell receptor contact residue hydrophobicity and HLA binding and evaluated them by T cell assays using healthy donor PBMCs. In a proof-of-principle study, we demonstrate that Cas9 protein can be modified to eliminate immunodominant epitopes through targeted mutation while preserving its function and specificity. Our study highlights the problem of pre-existing immunity against CRISPR-associated nucleases and offers a potential solution to mitigate the T cell immune response.

Possible immunogenicity of the Cas9 protein raises concerns about therapeutic applications. Here the authors identify pre-existing CD8+T-cell immunity in healthy individuals and in response modify Cas9 to remove the immunodominant epitopes.

The mKate fluorescent protein expressed by Leishmania mexicana modifies the parasite immunopathogenicity in BALB/c mice

Parasites have been engineered to express fluorescent reporter proteins, yet the impact of red fluorescent proteins on Leishmania infections remains largely unknown. We analysed the infection outcome of Leishmania mexicana parasites engineered for the constitutive expression of mKate protein and evaluated their immunogenicity in BALB/c mice. Infection of BALB/c mice with mKate transfected L. mexicana (LmexmKate ) parasites caused enlarged lesion sizes, leading to ulceration, and containing more parasites, as compared to LmexWT . The mKate protein showed immunogenic properties inducing antibody production against the mKate protein, as well as enhancing antibody production against the parasite. The augmented lesion sizes and ulcers, together with the more elevated antibody production, were related to an enhanced number of TNF-α and IL-1β producing cells in the infected tissues. We conclude that mKate red fluorescent protein is an immunogenic protein, capable of modifying disease evolution of L. mexicana.

GradDock: rapid simulation and tailored ranking functions for peptide-MHC Class I docking

Motivation

The identification of T-cell epitopes has many profound translational applications in the areas of transplantation, disease diagnosis, vaccine/therapeutic protein development and personalized immunotherapy. While data-driven methods have been widely used for the prediction of peptide binders with notable successes, the structural modeling of peptide binding to MHC molecules is crucial for understanding the underlying molecular mechanism of the immunological processes.

Results

We developed GradDock, a structure-based method for the rapid and accurate modeling of peptide binding to MHC Class I (pMHC-I). GradDock explicitly models diverse unbound peptides in vacuo and inserts them into the MHC-I groove through a steered gradient descent with a topological correction process. The simulation process yields diverse structural conformations including native-like peptides. We completely revised the Rosetta score terms and developed a new ranking function specifically for pMHC-I. Using the diverse peptides, a linear programming approach is applied to find the optimal weights for the individual Rosetta score terms. Our examination revealed that a refinement of the dihedral angles and a modification of the repulsion can dramatically improve the modeling quality. GradDock is five-times faster than a Rosetta-based docking approach for pMHC-I. We also demonstrate that the predictive capability of GradDock with the re-weighted Rosetta ranking function is consistently more accurate than the Rosetta-based method with the standard Rosetta score (approximately three-times better for a cross-docking set).

Availability and implementation

GradDock is freely available for academic purposes. The program and the ranking score weights for Rosetta are available at http://bel.kaist.ac.kr/research/GradDock.

Contact

hskim76@kaist.ac.kr.

Supplementary information

Supplementary data are available at Bioinformatics online.

Strategies to Reduce the Immunogenicity of Recombinant Immunotoxins

Recombinant immunotoxins (RITs) are genetically engineered proteins being developed to treat cancer. They are composed of an Fv that targets a cancer antigen and a fragment of a bacterial toxin that kills tumor cells. Because the toxin is a foreign protein, it is immunogenic. The clinical success of RITs in patients with a normal immune system is limited by their immunogenicity. In this review, we discuss our progress in therapeutic protein deimmunization and the balancing act between immunogenicity and therapeutic potency. One approach is to prevent the activation of B cells by mapping and elimination of B-cell epitopes. A second approach is to prevent helper T-cell activation by interfering with major histocompatibility complex II presentation or T-cell recognition. Immunizing mice with RITs that were deimmunized by elimination of the murine B- or T-cell epitopes showed that both approaches are effective. Another approach to control immunogenicity is to modify the host immune system. Nanoparticles containing synthetic vaccine particles encapsulating rapamycin can induce immune tolerance and prevent anti-drug antibody formation. This treatment restores RIT anti-tumor activity that is otherwise neutralized because of immunogenicity.

Anti-Drug Antibodies: Emerging Approaches to Predict, Reduce or Reverse Biotherapeutic Immunogenicity

The development of anti-drug antibodies (ADAs) following administration of biotherapeutics to patients is a vexing problem that is attracting increasing attention from pharmaceutical and biotechnology companies. This serious clinical problem is also spawning creative research into novel approaches to predict, avoid, and in some cases even reverse such deleterious immune responses. CD4⁺ T cells are essential players in the development of most ADAs, while memory B-cell and long-lived plasma cells amplify and maintain these responses. This review summarizes methods to predict and experimentally identify T-cell and B-cell epitopes in therapeutic proteins, with a particular focus on blood coagulation factor VIII (FVIII), whose immunogenicity is clinically significant and is the subject of intensive current research. Methods to phenotype ADA responses in humans are described, including T-cell stimulation assays, and both established and novel approaches to determine the titers, epitopes and isotypes of the ADAs themselves. Although rational protein engineering can reduce the immunogenicity of many biotherapeutics, complementary, novel approaches to induce specific tolerance, especially during initial exposures, are expected to play significant roles in future efforts to reduce or reverse these unwanted immune responses.

Recent advances in automated protein design and its future challenges

INTRODUCTION

Protein function is determined by protein structure which is in turn determined by the corresponding protein sequence. If the rules that cause a protein to adopt a particular structure are understood, it should be possible to refine or even redefine the function of a protein by working backwards from the desired structure to the sequence. Automated protein design attempts to calculate the effects of mutations computationally with the goal of more radical or complex transformations than are accessible by experimental techniques. Areas covered: The authors give a brief overview of the recent methodological advances in computer-aided protein design, showing how methodological choices affect final design and how automated protein design can be used to address problems considered beyond traditional protein engineering, including the creation of novel protein scaffolds for drug development. Also, the authors address specifically the future challenges in the development of automated protein design. Expert opinion: Automated protein design holds potential as a protein engineering technique, particularly in cases where screening by combinatorial mutagenesis is problematic. Considering solubility and immunogenicity issues, automated protein design is initially more likely to make an impact as a research tool for exploring basic biology in drug discovery than in the design of protein biologics.

Population-specific design of de-immunized protein biotherapeutics

Immunogenicity is a major problem during the development of biotherapeutics since it can lead to rapid clearance of the drug and adverse reactions. The challenge for biotherapeutic design is therefore to identify mutants of the protein sequence that minimize immunogenicity in a target population whilst retaining pharmaceutical activity and protein function. Current approaches are moderately successful in designing sequences with reduced immunogenicity, but do not account for the varying frequencies of different human leucocyte antigen alleles in a specific population and in addition, since many designs are non-functional, require costly experimental post-screening. Here, we report a new method for de-immunization design using multi-objective combinatorial optimization. The method simultaneously optimizes the likelihood of a functional protein sequence at the same time as minimizing its immunogenicity tailored to a target population. We bypass the need for three-dimensional protein structure or molecular simulations to identify functional designs by automatically generating sequences using probabilistic models that have been used previously for mutation effect prediction and structure prediction. As proof-of-principle we designed sequences of the C2 domain of Factor VIII and tested them experimentally, resulting in a good correlation with the predicted immunogenicity of our model.