MHC class I molecules exploit the low G+C content of pathogen genomes for enhanced presentation

Distinguishing self from nonself and pathogenic from nonpathogenic is a fundamental challenge to the immune system but whether adaptive immune systems use pathogen-specific signatures to achieve this is largely unknown. By investigating the presentation of large sets of viruses and bacteria on MHC class I molecules, we analyze whether MHC-I molecules have a preference for pathogen-derived peptides. The fraction of potential MHC-I binders in different organisms can vary up to eight-fold. We find that this variation can be largely explained by G+C content differences of the organisms, which are reflected in amino acid frequencies. A significant majority of HLA-A, but not HLA-B, molecules has a preference for peptides derived from organisms with a low G+C content. Interestingly, a low G+C content seems to be a universal signature for pathogenicity. Finally, we find the same preferences in chimpanzee and rhesus macaque MHC-I molecules. These results demonstrate that despite the fast evolution of MHC-I alleles and their extreme polymorphism and diversity in peptide-binding preferences, MHC-I molecules can acquire a preference to exploit pathogen-specific signatures.