Divergence of HPV16 variants reflects loci undergoing inter-host positive selection, potentially immunologic selection

Human papillomavirus type 16 (HPV16) is the most carcinogenic HPV, causing >50% of cervical cancers. Its unique oncogenicity is unsolved, e.g., the second most carcinogenic type (HPV18, causing ~16% of cancers) is relatively distantly related to HPV16. Mirabello et al. recently reported on viral genome data from 3,215 HPV positive specimens from women undergoing cervical cancer screening at Kaiser Permanente Northern California in the Persistence and Progression (PaP) cohort. Results demonstrated profound differences in disease risk by histologic subtype among the HPV16 sublineages (e.g., A1), with over 100-fold risk differences. Here, we expanded this analysis to examine the genetic and evolutionary underpinnings of HPV16 carcinogenicity through an exhaustive analysis of viral nucleotide distance. The HPV16 genome displays significant evidence of purifying selection ( d N / d S = 0.267; P<0.001). However, within the A1 and A2 sublineages, one of the two oncogenes (E6) does not differ from neutrality, while the other oncogene (E7) is significantly more constrained in cases ( d N / d S = 0.049) than in controls ( d N / d S = 0.27; P<0.001). Thus, benign viral infections exhibit less constraint, implying a nonsynonymous mutational burden. Using an unsupervised sliding window approach, we next identified genomic regions exhibiting strong evidence of inter-host positive selection. Within A1, 14 regions were identified, with d N / d S values up to 59.44. Twelve (12) of these regions overlap phylogenetic sublineage-defining residues, and 9 match known HPV epitope sequences obtained from the Immune Epitope Database and/or the Human Papillomavirus T Cell Antigen Database. In particular, two regions of E6 were identified independently in cases and controls: codons 20-27 and codons 75-90, both of which overlap with experimentally verified HLA epitopes, and the second of which overlaps known antibody epitopes. Moreover, the 75-90 region exhibits a d N / d S ratio of 59.44 in cases and 28.05 in controls, shows ...