Major histocompatibility complex (MHC) class I molecules form medleys with peptide antigens which are expressed on the cell surface for recognition by CD8+ T cells. Derived from antigens synthesized in the cytoplasm, these peptides are generally 8-10 amino acids in length. For most MHC alleles two of the pockets within the peptide binding groove display a marked preference for one or two amino acids at certain anchor positions within the peptide. This was the breakthrough discovery that enabled more efficient CTL epitope mapping. Dependent on this information, web-based algorithms used to predict CD8+ T cell epitopes were designed to include peptides limited to between 8 and 10 residues. The apparent dominance of MHC class I peptides of 8 to 10 amino acids in length may be misleading and result from this bias of widely used algorithms. Several studies have clearly shown that some longer peptides are naturally processed and presented for recognition by CD8+ T cells. A number of these noncanonical epitopes have been identified in Epstein-Barr virus (EBV). In order to ascertain the role of long peptides in CD8+ T cell responses, studies were required which utilise overlapping peptide screening and not web-based algorithms to predict peptides. The latent herpes virus Epstein-Barr virus is a superb model to determine the relative prevalence of noncanonical T cell epitopes.
A number of highly immunogenic antigens of EBV were the focus of this study. Many novel class I EBV epitopes were identified with a significant proportion arising from the BZLF1 antigen. Overall, 28 latent epitopes from EBNA3A, EBNA3B, EBNA3C and EBNA1 and 13 lytic epitopes from BRLF1, BMLF1 and BMRF1 were identified in addition to novel epitopes from BZLF1, including five noncanonical CD8+ T cell epitopes. The EBV BZLF1 protein demonstrated a propensity for epitope clustering. In total, I have defined 13 novel epitopes from this EBV protein which are restricted to a number of HLA alleles and appear to be clustered and in many instances overlapping. These included two overlapping epitopes of different length that nevertheless conform to the binding motif of the large and abundant HLA-B*44 supertype. While HLA-B*18:01+ individuals responded strongly and exclusively to an octamer peptide 173SELEIKRY180, HLA-B*44:03+ individuals responded to the atypically large dodecamer peptide 169EECDSELEIKRY180 that encompasses the octamer peptide. Moreover, the octamer peptide bound more stably to HLA-B*18:01 than the dodecamer peptide while conversely, HLA-B*44:03 bound only the longer peptide. Furthermore, crystal structures of these viral peptide-HLA complexes showed that the antigen-binding cleft of HLA-B*18:01 was more ideally suited to bind shorter peptides, while HLA-B*44:03 exhibited characteristics that favoured the presentation of longer peptides. Mass spectrometric identification of over a thousand naturally-presented ligands revealed that HLA-B*18:01 was more biased towards presenting shorter peptides than HLA-B*44:03. Collectively, these data highlight a mechanism through which polymorphism within an HLA class I supertype can diversify determinant selection and immune responses by varying peptide length preferences.
Interestingly, a human protein sequence (DELEIKAY) was identified with sequence homology to the octamer EBV epitope. The peptide was shown to bind stably to HLA-B*1801, and peptide elution/mass spectrometric studies showed it is presented by this HLA molecule on the surface of human cells. A significant proportion of T cells raised against the SELEIKRY EBV epitope cross-reacted with this HLA-B*1801-binding self-peptide. Of note, only a limited number of HLA-B*1801+ healthy individuals showing strong IFN-γ responses to SELEIKRY had a detectable response to DELEIKAY. These cross-reactive T cells were shown to express a diverse array of T cell receptors. The potential for self-reactivity by these CTLs is presumably kept under rigorous control by normal self-tolerance mechanisms. However, these EBV/self cross-reactive T cells could pose an autoimmune threat if HLA-B*1801-DELEIKAY levels increased or the T cell activation threshold is reduced as a result of cytokine release during inflammation and tissue damage. These cross-reactive T cell populations should be considered for their potential role in self-reactivity following viral infection.