C > U mutations generate immunogenic peptides in SARS-CoV-2.

The rapid spread of SARS-CoV-2 worldwide has given rise to numerous variants. While the impact of viral mutations on antibody escape has been extensively studied, an unresolved issue concerns how emerging mutations shape HLA-restricted T-cell immune responses. Here, we analyse SARS-CoV-2 genomic variants, showing that 27% of the mutations are C > U transitions, a phenomenon common in human RNA viruses and primarily attributed to APOBEC3 enzyme-driven mutagenesis. We find that this mutation bias generally enhances viral peptide binding to human leukocyte antigen class I (HLA-I) molecules, producing immunogenic epitopes that trigger cytotoxic adaptive immune responses in most individuals across diverse populations. We also identify several HLA-I variants that are especially well-suited for presenting viral epitopes generated by these mutations. Intriguingly, individuals carrying these specific alleles are predominantly located in South and East Asia. Finally, we show that carrying HLA-I molecules that are less likely to bind C > U-induced viral peptides increases risk for severe COVID-19 disease. Our work suggests a link between C > U hypermutation and HLA-I-based presentation of viral epitopes, which may reflect the evolutionary outcome of ancient RNA virus pandemics. More broadly, our findings imply that SARS-CoV-2 diversification leads to ongoing gains of T-cell epitopes despite natural selection favouring immune escape.