Double Trouble: HLA Class I And Class II Have Evolved To Target Different Regions On Viral Proteins

The Human Leukocyte Antigen (HLA) system is the most genetically diverse loci in the human genome, with over 28409 class I alleles and 12000 class II alleles. These proteins present short peptides on the surface cells that are surveilled by CD8+ (HLA class I) and CD4+ (HLA class II). Previous work has shown that both HLA class I and II binding occurs preferentially targeted regions termed immunodominant hotspots.
To investigate the targeting of such hotspots, we developed an ensemble model for predicting binding peptides for a set of highly prevalent HLA alleles. Using these predictions, we constructed an epitope hotspot maps for proteins of multiple human viruses including SARS-CoV-2, HIV-1, Influenza A and Dengue. A hotspot was defined as a region that was in the 90% percentile of the aggregate binding score across a set of 27 and 26 class I and class II alleles respectively.
While we found that the epitope maps for HLA class I and II binding peptides were weakly positively correlated (r = 0.23, p <= 0.0001), the locations of the HLA class I and HLA class II immunodominant hotspots were negatively correlated (e.g. SARS-CoV-2 spike: r < -0.65, p <= 0.0001, Dengue E: r < -0.43, p <= 0.0001). This negative correlation was also observed using experimentally confirmed MHC ligands and T-cell epitope data taken from the IEDB and HIV-LANL databases.
To investigate whether this phenomenon arises from inherent properties of the HLA molecules or the viral protein sequence, we conducted simulations in which we considered randomly shuffled protein sequences as well as randomly generated HLA alleles based on amino acid frequencies of a large set of experimentally characterized HLA alleles taken from the SYFPETHI database. While negative correlations were observed for shuffled protein sequences and shuffled HLA alleles, the correlations were weaker only when using random HLA alleles.
Our findings suggest that HLA alleles have evolved to provide broad coverage of viral sequences by presenting different immunodominant regions to CD8+ and CD4+ T-cells. This strategy, reduced the ability to escape T-cell surveillance by viral pathogens.