Bound to be an Epitope: Determinants of the T-cell response to MHC-I presented peptides

Abstract

Peptides that are produced in cells during the degradation of proteins can be presented on MHC-I molecules. When a cell is infected with a virus, peptides from viral proteins can be presented as well and used by T-cells of the immune system to mount a specific immune response. MHC-I/peptide-complexes (pMHCs) that are used for a specific immune response are called epitopes. In this thesis we approach the question of which pMHCs become epitopes from three different perspectives. First, we investigate which peptides are more likely to be MHC-I presented than others. We show that HLA-A molecules present more pathogen-derived peptides compared to self-derived peptides, by making use of the fact that pathogens have a low genomic G+C content, and therefore an enhanced frequency of the amino acids F, I, N, K and Y (Chapter 2). Furthermore, we investigate the optimal prediction of the MHC-I ligandome using proteolysis, TAP transport and MHC-I binding predictors (Chapter 3). Second, we investigate which pMHCs are recognizable by T-cells, i.e. immunogenic. We show that certain positions and amino acid content of MHC-I presented peptides are associated with immunogenicity. The observed correlates with immunogenicity were summarized in a model that could be applied to successfully predict the immunogenicity of novel pMHCs in large-scale epitope discovery projects (Chapter 4). Third, we investigate which we studied how the similarity to self influences the formation of an immune response to foreign pMHCs. We show that ∼30% of the MHC-I presented foreign peptides is unlikely to be an epitope, as they resemble a self-peptide on the same MHC-I molecule (Chapter 5). We model the constraints that this self/nonself overlap puts on the capacity of the immune system to elicit immune responses without causing auto-immune responses (Addendum to Chapter 5). Finally, we investigate how self-overlaps affect host-versus-graft immune responses after kidney transplantation. We show that a humoral immune response to mismatched HLA class I molecules depends on the presentation of non-overlapping class-I-derived peptides on HLA class II (Chapter 6). The correct prediction of epitopes is important for a better understanding of the T-cell immune response, and can help to develop better vaccines, to enhance transplantation efficacy, and to understand host/pathogen interactions. Based on the studies described in this study as well as older investigations, ~50% of the epitope predictions are correct, and we show which steps can be taken to increase this performance further.