HPV16 genetic variation provides evidence of positive natural selection driven by HLA class I

Human papillomavirus type 16 (HPV16) causes more cancer than any other virus. However, most HPV16 infections are controlled by the host's immune system and it remains unclear how viral and host genetic variation contribute to infection outcomes. Here, we analyze 4704 HPV16 whole genomes to identify 56 viral codons putatively under positive natural selection to change their amino acids, with evidence including d_N/d_S > 1, evolutionary convergence, and structural importance in the protein. We find that codons under positive selection disproportionately overlap known HPV16 immune epitopes recognized by cytotoxic T lymphocytes, particularly those restricted by the previously reported risk allele HLA-B*07:02 (odds ratio [OR] = 4.9; 95%CI = 2.1-10.3; P_Fisher = 0.00015), exemplified by position 10 of the E6 oncoprotein. Positively selected codons also disproportionately overlap 158 nucleotide sites at which the evolutionary sub/lineages of HPV16 have diverged (OR = 19.1; 95%CI = 10.5-34.7; P_Fisher < 2.2×10^-16), and show more rare variation in cervical precancers/cancers than controls (benign or cleared HPV16 infections) in the E1 protein (OR = 9.34, 95%CI = 1.4-402.5; P_Fisher = 0.0084). Our results suggest that a small subset of HPV16 variants can improve viral persistence through escape of HLA-related immune recognition. The interaction of HPV16 and HLA variation may help to explain how similar or identical viral isolates can have such disparate infection outcomes.