HPV E6 Impairs Translesion Synthesis by Blocking POLη Induction

Many DNA tumor viruses inhibit or repurpose host DNA repair pathways to evade viral defense mechanisms or promote their own replication. High risk genus α human papillomaviruses (α-HPVs) express two versatile oncogenes (α-HPV E6 and E7) that use both approaches simultaneously. To identify new interactions with DNA repair pathways, we conducted a computational analysis of gene expression in cervical cancer (CaCx) transcriptomic datasets and identified a frequent upregulation of translesion synthesis (TLS) genes. TLS polymerase genes, particularly the gene for POLη (POLH), did not follow this pattern. Characterization of α-HPV oncogene expressing cell lines and premalignant cervical tissue confirm these data. They also show that the increased TLS protein abundance come in response to nucleoside depletion. Primary and transformed cell lines to demonstrate that α-HPV E6 blocks POLη induction by degrading p53. Lack of POLη dooms the pathway, preventing it from facilitating tolerance of replication stress. Failed TLS results in replication fork stalling and collapse into deleterious double strand breaks in the DNA (DSBs). Consequently, cellular genome fidelity decreases in a manner consistent with the mutations that accumulate during CaCx progression. Alterations in TLS are determinants for the efficacy of the chemotherapeutics most often used to treat CaCx (cisplatin and carboplatin). Exogenous expression of POLη protected CaCx cells from cisplatin-associated toxicity/damage, effectively stabilizing their genome. TLS polymerase expression is also a prognostic factor in CaCx. Analysis of the cancer genome atlas database (TCGA) shows increased expression of these genes correlated with over a decade shorter median survival. Total Samples: 313 from 5 data sets (GSE9750, GSE7803, GSE6791, GSE63514, GSE29570). Data was included based on availability on disease stage information. Data was normalized and merged into a single matrix to allow analysis and comparison of all 5 datasets. data processing/re-analysis step: gene expression data was downladed from GEO2R and normalized via feature scaling normalized datasets were merged into a matrix containing 313 samples and 25203 gene IDs gene IDs were matched with corresponding gene names First column contains row-names (disease stage and gene names) Columns 2-314 contain normalized gene expression data