Better Epitope Discovery, Precision Immune Engineering, and Accelerated Vaccine Design Using Immunoinformatics Tools

Abstract

Computational vaccinology includes epitope mapping, antigen selection, and immunogen design using computational tools. Tools that facilitate the in silico prediction of immune response to biothreats, emerging infectious diseases, and cancers can accelerate the design of novel and next generation vaccines and their delivery to the clinic. Over the past 20 years, vaccinologists, bioinformatics experts, and advanced programmers based in Providence, Rhode Island, USA have advanced the development of an integrated toolkit for vaccine design called iVAX, that is secure and user-accessible by internet. This integrated set of immunoinformatic tools comprises algorithms for scoring and triaging candidate antigens, selecting immunogenic and conserved T cell epitopes, re-engineering or eliminating regulatory T cell epitopes, and re-designing antigens to induce immunogenicity and protection against disease for humans and livestock. Commercial and academic applications of iVAX have included identifying immunogenic T cell epitopes in the development of a T-cell based human multi-epitope Q fever vaccine, designing novel influenza vaccines, identifying cross-conserved T cell epitopes for a malaria vaccine, and analyzing immune responses in clinical vaccine studies. Animal vaccine applications to date have included viral infections of pigs such as swine influenza A, PCV2, and African Swine Fever. "Rapid-Fire" applications for biodefense have included a demonstration project for Lassa Fever and Q fever. As recent infectious disease outbreaks underscore the significance of vaccine-driven preparedness, the integrated set of tools available on the iVAX toolkit stand ready to help vaccine developers deliver genome-derived, epitope-driven vaccines.

Keywords: ClustiMer; EpiMatrix; JanusMatrix; T cell epitope; Treg epitope; bioinformatics; immunoinformatics; vaccines.

Figure 1

Immune camouflage. T cell epitopes found in the sequences of viruses, bacteria, and parasites and/or in the human microbiome may be conserved with highly prevalent T cell epitopes found in self-proteins from the human genome; cross-conservation may be observed at the TCR face for peptides that bind to the same HLA but do not have identical sequences. This may enable a pathogen to trigger “ignorance” or active tolerance from T cells that have been trained on self-epitopes. Adapted from Moise et al. (5).

Figure 2

Tools comprised in the iVAX Toolkit and laboratory tools used in validation studies. The iVAX toolkit is an on-line, secure access toolkit that provides individualized sites for academic and commercial users. Sequences can be uploaded and analyzed for Class I and Class II T cell epitopes using EpiMatrix and searched for clusters of epitopes (promiscuous epitopes) using ClustiMer. Tools such as Conservatrix define cross-strain conserved epitopes and JanusMatrix identifies T cell epitopes that induce anergy or active tolerance. These algorithms, and others such as VaxCAD and iTEM are integrated into the secure-access, cloud-based toolkit. Beyond the in silico analysis phase, vaccine design usually proceeds to in vitro and in vivo validation.

Figure 3

Recent validation studies: (A) CD8 T cell epitopes. Retrospective analysis of eluted peptide dataset published by Abelin et al. (23) 95% of eluted 9- and 10-mers are predicted to bind to HLA according to EpiMatrix, many of which are predicted strong binders (solid bars). Only ~88% of ligands were accurately recalled by various versions of NetMHCpan. EpiMatrix also identifies the majority of the eluted ligands as high affinity binders (85% of all eluted peptides, solid bars), whereas only up to 58% of ligands were identified as high affinity binders (IC50 ≤ 50 nM) by NetMHCpan. (B) CD4 T cell epitopes. Prospectively validation of EpiMatrix selections in in vitro HLA binding assays: EpiMatrix HLA class II predictions are 74% accurate when tested in in vitro HLA binding assays, with an average observed Positive Predictive Value (PPV) of 81%. The number of peptides tested for each HLA allele is shown in parenthesis. Selected peptides were published in reference 23 and some are unpublished.

Figure 4

EpiMatrix immunogenicity scale, showing vaccine antigens, human proteins that have been reported to be immunogenic, and non-immunogenic antigens alongside the median scores for sets of proteins from the human genome. The vaccine antigen sequences were obtained from UniProt. The EpiMatrix immunogenicity scale is set to zero based on the median HLA DR score of a set of random protein sequences. Normalization of the HLA scoring enables direct comparison of median HLA DR scores to candidate antigens; for example, candidate vaccine antigen A would be preferred over candidate vaccine antigen B for inclusion in a vaccine designed to elicit T helper immune response and to drive humoral response.

Figure 5

EpiMatrix Class II analysis. This analysis of an influenza hemagglutinin sequence shows a 9-mer frame that contains 7 top 1% hits and 2 top 5% hits for HLA DR alleles. This feature is called an EpiBar (Epitope bar) and is a feature of highly immunogenic epitopes. This analysis was performed using HLA DRB1 alleles.

Figure 6

JanusMatrix and self-like epitopes. Each HLA ligand has two faces: the HLA-binding face (agretope, blue residues), and the TCR-interacting face (epitope, green residues). Predicted ligand with identical TCR epitopes and variant HLA-binding agretopes may stimulate cross reactive tolerizing or Treg responses, providing they bind to the same HLA allele.

Figure 7

JanusMatrix Analysis of MageA3-HLA*0101-restricted T cell epitope. Peptides containing identical T cell facing residues include variants of MageA3 (Melanoma Associated Antigen) and three additional human proteins, including human titin, one of the components of cardiac smooth muscle. The figure shows that the HLA-A*0101-restricted T cell epitope EVDPIGHVY is conserved with an epitope that has identical HLA binding restriction but slightly different agretope (XSPDXXVAQY) and identical TCR facing residues (EXXPIXXXX).