Use of molecular modeling and site-directed mutagenesis to define the structural basis for the immune response to carbohydrate xenoantigens

DeepImmuno: deep learning-empowered prediction and generation of immunogenic peptides for T-cell immunity

Cytolytic T-cells play an essential role in the adaptive immune system by seeking out, binding and killing cells that present foreign antigens on their surface. An improved understanding of T-cell immunity will greatly aid in the development of new cancer immunotherapies and vaccines for life-threatening pathogens. Central to the design of such targeted therapies are computational methods to predict non-native peptides to elicit a T-cell response, however, we currently lack accurate immunogenicity inference methods. Another challenge is the ability to accurately simulate immunogenic peptides for specific human leukocyte antigen alleles, for both synthetic biological applications, and to augment real training datasets. Here, we propose a beta-binomial distribution approach to derive peptide immunogenic potential from sequence alone. We conducted systematic benchmarking of five traditional machine learning (ElasticNet, K-nearest neighbors, support vector machine, Random Forest and AdaBoost) and three deep learning models (convolutional neural network (CNN), Residual Net and graph neural network) using three independent prior validated immunogenic peptide collections (dengue virus, cancer neoantigen and SARS-CoV-2). We chose the CNN as the best prediction model, based on its adaptivity for small and large datasets and performance relative to existing methods. In addition to outperforming two highly used immunogenicity prediction algorithms, DeepImmuno-CNN correctly predicts which residues are most important for T-cell antigen recognition and predicts novel impacts of SARS-CoV-2 variants. Our independent generative adversarial network (GAN) approach, DeepImmuno-GAN, was further able to accurately simulate immunogenic peptides with physicochemical properties and immunogenicity predictions similar to that of real antigens. We provide DeepImmuno-CNN as source code and an easy-to-use web interface.

Broad specificity of immune helminth scFv library to identify monoclonal antibodies targeting Strongyloides

Antibodies have different chemical properties capable of targeting a diverse nature of antigens. Traditionally, immune antibody libraries are perceived to be disease-specific with a skewed repertoire. The complexity during the generation of a combinatorial antibody library allows for a skewed but diverse repertoire to be generated. Strongyloides stercoralis is a parasite that causes strongyloidiasis, a potentially life-threatening disease with a complex diagnosis that impedes effective control and treatment of the disease. This study describes the isolation of monoclonal antibodies against S. stercoralis NIE recombinant protein using an immune antibody phage display library derived from lymphatic filaria-infected individuals. The isolated antibody clones showed both lambda and kappa light chains gene usage, with diverse amino acid distributions. Structural analysis showed that electropositivity and the interface area could determine the binding affinity of the clones with NIE. The successful identification of S. stercoralis antibodies from the filarial immune library highlights the breadth of antibody gene diversification in an immune antibody library that can be applied for closely related infections.

DeepImmuno: Deep learning-empowered prediction and generation of immunogenic peptides for T cell immunity

T-cells play an essential role in the adaptive immune system by seeking out, binding and destroying foreign antigens presented on the cell surface of diseased cells. An improved understanding of T-cell immunity will greatly aid in the development of new cancer immunotherapies and vaccines for life threatening pathogens. Central to the design of such targeted therapies are computational methods to predict non-native epitopes to elicit a T cell response, however, we currently lack accurate immunogenicity inference methods. Another challenge is the ability to accurately simulate immunogenic peptides for specific human leukocyte antigen (HLA) alleles, for both synthetic biological applications and to augment real training datasets. Here, we proposed a beta-binomial distribution approach to derive epitope immunogenic potential from sequence alone. We conducted systematic benchmarking of five traditional machine learning (ElasticNet, KNN, SVM, Random Forest, AdaBoost) and three deep learning models (CNN, ResNet, GNN) using three independent prior validated immunogenic peptide collections (dengue virus, cancer neoantigen and SARS-Cov-2). We chose the CNN model as the best prediction model based on its adaptivity for small and large datasets, and performance relative to existing methods. In addition to outperforming two highly used immunogenicity prediction algorithms, DeepHLApan and IEDB, DeepImmuno-CNN further correctly predicts which residues are most important for T cell antigen recognition. Our independent generative adversarial network (GAN) approach, DeepImmuno-GAN, was further able to accurately simulate immunogenic peptides with physiochemical properties and immunogenicity predictions similar to that of real antigens. We provide DeepImmuno-CNN as source code and an easy-to-use web interface.

DATA AVAILABILITY

DeepImmuno Python3 code is available at https://github.com/frankligy/DeepImmuno . The DeepImmuno web portal is available from https://deepimmuno.herokuapp.com . The data in this article is available in GitHub and supplementary materials.

Carbohydrate-protein interactions: molecular modeling insights

The article reviews the significant contributions to, and the present status of, applications of computational methods for the characterization and prediction of protein-carbohydrate interactions. After a presentation of the specific features of carbohydrate modeling, along with a brief description of the experimental data and general features of carbohydrate-protein interactions, the survey provides a thorough coverage of the available computational methods and tools. At the quantum-mechanical level, the use of both molecular orbitals and density-functional theory is critically assessed. These are followed by a presentation and critical evaluation of the applications of semiempirical and empirical methods: QM/MM, molecular dynamics, free-energy calculations, metadynamics, molecular robotics, and others. The usefulness of molecular docking in structural glycobiology is evaluated by considering recent docking- validation studies on a range of protein targets. The range of applications of these theoretical methods provides insights into the structural, energetic, and mechanistic facets that occur in the course of the recognition processes. Selected examples are provided to exemplify the usefulness and the present limitations of these computational methods in their ability to assist in elucidation of the structural basis underlying the diverse function and biological roles of carbohydrates in their dialogue with proteins. These test cases cover the field of both carbohydrate biosynthesis and glycosyltransferases, as well as glycoside hydrolases. The phenomenon of (macro)molecular recognition is illustrated for the interactions of carbohydrates with such proteins as lectins, monoclonal antibodies, GAG-binding proteins, porins, and viruses.

Xenoantibody response to porcine islet cell transplantation using GTKO, CD55, CD59, and fucosyltransferase multiple transgenic donors

BACKGROUND

Promising developments in porcine islet xenotransplantation could resolve the donor pancreas shortage for patients with type 1 diabetes. Using α1,3-galactosyltransferase gene knockout (GTKO) donor pigs with multiple transgenes should extend xenoislet survival via reducing complement activation, thrombus formation, and the requirement for exogenous immune suppression. Studying the xenoantibody response to GTKO/hCD55/hCD59/hHT islets in the pig-to-baboon model, and comparing it with previously analyzed responses, would allow the development of inhibitory reagents capable of targeting conserved idiotypic regions.

METHODS

We generated IgM heavy and light chain gene libraries from 10 untreated baboons and three baboons at 28 days following transplantation of GTKO/hCD55/hCD59/hHT pig neonatal islet cell clusters with immunosuppression. Flow cytometry was used to confirm the induction of a xenoantibody response. IgM germline gene usage was compared pre- and post-transplant. Homology modeling was used to compare the structure of xenoantibodies elicited after transplantation of GTKO/hCD55/hCD59/hHT pig islets with those induced by GTKO and wild-type pig endothelial cells without further genetic modification.

RESULTS

IgM xenoantibodies that bind to GTKO pig cells and wild-type pig cells were induced after transplantation. These anti-non-Gal antibodies were encoded by the IGHV3-66*02 (Δ28%) and IGKV1-12*02 (Δ25%) alleles, for the immunoglobulin heavy and light chains, respectively. IGHV3-66 is 86.7% similar to IGHV3-21 which was elicited by rhesus monkeys in response to GTKO endothelial cells. Heavy chain genes most similar to IGHV3-66 were found to utilize the IGHJ4 gene in 85% of V-D regions analyzed. However, unlike the wild-type response, a consensus complementary determining region 3 was not identified.

CONCLUSIONS

Additional genetic modifications in transgenic GTKO pigs do not substantially modify the structure of the restricted group of anti-non-Gal xenoantibodies that mediate induced xenoantibody responses with or without immunosuppression. The use of this information to develop new therapeutic agents to target this restricted response will likely be beneficial for long-term islet cell survival and for developing targeted immunosuppressive regimens with less toxicity.

Structure-based prediction of the effects of a missense variant on protein stability

Predicting the effects of amino acid substitutions on protein stability provides invaluable information for protein design, the assignment of biological function, and for understanding disease-associated variations. To understand the effects of substitutions, computational models are preferred to time-consuming and expensive experimental methods. Several methods have been proposed for this task including machine learning-based approaches. However, models trained using limited data have performance problems and many model parameters tend to be over-fitted. To decrease the number of model parameters and to improve the generalization potential, we calculated the amino acid contact energy change for point variations using a structure-based coarse-grained model. Based on the structural properties including contact energy (CE) and further physicochemical properties of the amino acids as input features, we developed two support vector machine classifiers. M47 predicted the stability of variant proteins with an accuracy of 87 % and a Matthews correlation coefficient of 0.68 for a large dataset of 1925 variants, whereas M8 performed better when a relatively small dataset of 388 variants was used for 20-fold cross-validation. The performance of the M47 classifier on all six tested contingency table evaluation parameters is better than that of existing machine learning-based models or energy function-based protein stability classifiers.

Critical epitopes in the nucleocapsid protein of SFTS virus recognized by a panel of SFTS patients derived human monoclonal antibodies

Background

SFTS virus (SFTSV) is a newly discovered pathogen to cause severe fever with thrombocytopenia syndrome (SFTS) in human. Successful control of SFTSV epidemic requires better understanding of the antigen target in humoral immune responses to the new bunyavirus infection.

Methodology/Principal Findings

We have generated a combinatorial Fab antibody phage library from two SFTS patients recovered from SFTSV infection. To date, 94 unique human antibodies have been generated and characterized from over 1200 Fab antibody clones obtained by screening the library with SFTS purified virions. All those monoclonal antibodies (MAbs) recognized the nucleocapsid (N) protein of SFTSV while none of them were reactive to the viral glycoproteins Gn or Gc. Furthermore, over screening 1000 mouse monoclonal antibody clones derived from SFTSV virions immunization, 462 clones reacted with N protein, while only 16 clones were reactive to glycoprotein. Furthermore, epitope mapping of SFTSV N protein was performed through molecular simulation, site mutation and competitive ELISA, and we found that at least 4 distinct antigenic epitopes within N protein were recognized by those human and mouse MAbs, in particular mutation of Glu10 to Ala10 abolished or significantly reduced the binding activity of nearly most SFTS patients derived MAbs.

Conclusions/Significance

The large number of human recombinant MAbs derived from SFTS patients recognized the viral N protein indicated the important role of the N protein in humoral responses to SFTSV infection, and the critical epitopes we defined in this study provided molecular basis for detection and diagnosis of SFTSV infection.

Rahman RNZR, Leow TC, Ali MSM, Basri M, Salleh AB. Improvement of thermal stability via outer-loop ion pair interaction of mutated T1 lipase from Geobacillus zalihae strain T1

Mutant D311E and K344R were constructed using site-directed mutagenesis to introduce an additional ion pair at the inter-loop and the intra-loop, respectively, to determine the effect of ion pairs on the stability of T1 lipase isolated from Geobacillus zalihae. A series of purification steps was applied, and the pure lipases of T1, D311E and K344R were obtained. The wild-type and mutant lipases were analyzed using circular dichroism. The T_m for T1 lipase, D311E lipase and K344R lipase were approximately 68.52 °C, 70.59 °C and 68.54 °C, respectively. Mutation at D311 increases the stability of T1 lipase and exhibited higher T_m as compared to the wild-type and K344R. Based on the above, D311E lipase was chosen for further study. D311E lipase was successfully crystallized using the sitting drop vapor diffusion method. The crystal was diffracted at 2.1 Å using an in-house X-ray beam and belonged to the monoclinic space group C2 with the unit cell parameters a = 117.32 Å, b = 81.16 Å and c = 100.14 Å. Structural analysis showed the existence of an additional ion pair around E311 in the structure of D311E. The additional ion pair in D311E may regulate the stability of this mutant lipase at high temperatures as predicted in silico and spectroscopically.

The anti-nonGal xenoantibody response to alpha1,3-galactosyltransferase gene knockout pig xenografts

PURPOSE OF REVIEW

Anti-nonGal xenoantibodies are a major barrier to the survival of genetically modified porcine xenografts. This review summarizes the contribution of anti-nonGal xenoantibodies to the activation of porcine endothelial cells and graft rejection, and further provides an update on recent advancements in defining the unique features of anti-nonGal xenoantibody structure.

RECENT FINDINGS

Anti-nonGal xenoantibodies pre-exist at low levels in humans and nonhuman primates, and are notably absent in neonates. Exposure of nonhuman primates to alpha1,3-galactosyltransferase gene knockout endothelial cells initiates an induced xenoantibody response that is restricted and encoded by the germline immunoglobulin heavy chain gene IGHV3-21. The target xenoantigen remains undetermined, but several candidate targets have been proposed, including carbohydrate xenoantigens. New advancements in molecular modeling provide insight on the mechanism by which xenoantibodies bind to structurally related carbohydrates.

SUMMARY

Genetic manipulation of porcine donors has significantly prolonged the survival of grafts placed into nonhuman primate recipients, but anti-nonGal xenoantibodies and thrombosis limit the ability of these grafts to function on a long-term basis. Recent developments defining pre-existing anti-nonGal xenoantibody levels, the restriction in the anti-nonGal xenoantibody response and the identification of key sites defining xenoantibody-carbohydrate interactions now provide the information necessary to develop new approaches to preventing xenoantibody-mediated rejection.

In silico analysis of antibody-carbohydrate interactions and its application to xenoreactive antibodies

Antibody-carbohydrate interactions play central roles in stimulating adverse immune reactions. The most familiar example of such a process is the reaction observed in ABO-incompatible blood transfusion and organ transplantation. The ABO blood groups are defined by the presence of specific carbohydrates expressed on the surface of red blood cells. Preformed antibodies in the incompatible recipient (i.e., different blood groups) recognize cells exhibiting host-incompatible ABO system antigens and proceed to initiate lysis of the incompatible cells. Pig-to-human xenotransplantation presents a similar immunological barrier. Antibodies present in humans recognize carbohydrate antigens on the surface of pig organs as foreign and proceed to initiate hyperacute xenograft rejection. The major carbohydrate xenoantigens all bear terminal Gal alpha(1,3)Gal epitopes (or alphaGal). In this study, we have developed and validated a site mapping technique to investigate protein-ligand recognition and applied it to antibody-carbohydrate systems. This site mapping technique involves the use of molecular docking to generate a series of antibody-carbohydrate complexes, followed by analysis of the hydrogen bonding and van der Waals interactions occurring in each complex. The technique was validated by application to a series of antibody-carbohydrate crystal structures. In each case, the majority of interactions made in the crystal structure complex were able to be reproduced. The technique was then applied to investigate xenoantigen recognition by a panel of monoclonal anti-alphaGal antibodies. The results indicate that there is a significant overlap of the antibody regions engaging the xenoantigens across the panel. Likewise, similar regions of the xenoantigens interact with the antibodies.

The anti-non-gal xenoantibody response to xenoantigens on gal knockout pig cells is encoded by a restricted number of germline progenitors

Antibodies directed at non-gal xenoantigens are responsible for acute humoral xenograft rejection when gal knockout (GalTKO) pig organs are transplanted into nonhuman primates. We generated IgM and IgG gene libraries using peripheral blood lymphocytes of rhesus monkeys initiating active xenoantibody responses after immunization with GalTKO pig endothelial cells and used these libraries to identify IgV(H) genes that encode antibody responses to non-gal pig xenoantigens. Immunoglobulin genes derived from the IGHV3-21 germline progenitor encode xenoantibodies directed at non-gal xenoantigens. Transduction of GalTKO cells with lentiviral vectors expressing the porcine alpha1,3 galactosyltransferase gene responsible for gal carbohydrate expression results in a higher level of binding of 'anti-non-gal' xenoantibodies to transduced GalTKO cells expressing the gal carbohydrate, suggesting that anti-non-gal xenoantibodies cross react with carbohydrate xenoantigens. The galactosyltransferase two gene encoding isoglobotriaosylceramide synthase (iGb3 synthase) is not expressed in GalTKO pig cells. Our results demonstrate that anti-non-gal xenoantibodies in primates are encoded by IgV(H) genes that are restricted to IGHV3-21 and bind to an epitope that is structurally related to but distinct from the Gal carbohydrate.

Immunoproteomic identification of bovine pericardium xenoantigens

Bovine pericardium is an important biomaterial with current application in glutaraldehyde-fixed bioprosthetic heart valves and possible future application as an unfixed biological scaffold for tissue engineering. The importance of both humoral and cell-mediated rejection responses toward fixed and unfixed xenogeneic tissues has become increasingly apparent. However, the full scope and specific identities of bovine pericardium proteins that can elicit an immune response remain largely unknown. In this study, an immunoproteomic approach was used to survey bovine pericardium proteins for their ability to elicit a humoral immune response in rabbits. A two-stage protein extraction protocol was used to separate bovine pericardium proteins into water- and lipid-soluble fractions. Two-dimensional (2-D) gel electrophoresis was performed to separate the proteins from each fraction. Western blots were generated from 2-D gels of both bovine pericardium protein fractions. These blots were probed with serum from rabbits immunized with bovine pericardium and a secondary antibody was used to assess for IgG positivity. Western blots were compared to duplicate 2-D gels and proteins in matched spots were identified by tandem mass spectrometry. Thirty-one putative protein antigens were identified, eight of which are known to be antigenic from previous studies. All of the putative antigens demonstrated progressive staining intensity with increasing days of post-exposure serum. Identified antigenic proteins represented a variety of functional and structural protein types, and included both cellular and matrix proteins. The results of this study have implications for the use of bovine pericardium as a biomaterial in bioprostheses and tissue engineering applications, as well as xenotransplantation in general.