Introduction: HPV16 is able to integrate into the host genome of oropharyngeal squamous cell carcinomas (OPSCC). Different studies have shown that HPV16-positive OPSCC can be subdivided into tumors harbouring exclusively integrated viral DNA, exclusively extrachromosomal viral episomes, or mixed forms. In a previous study we showed that integration did not affect the levels of the viral genes E2, E6 and E7, nor the levels of transcripts of human genes directly disrupted by viral integration (Olthof et al., PlosOne 2014). Therefore, our aim was to perform a genome-wide screen to identify human genes which expression is influenced by integration of HPV16.

Materials and Methods: Total RNA was collected from 33 fresh-frozen HPV-16 positive OPSCC samples (9 integrated, 4 mixed, 20 episomal), and analysed by mRNA expression profiling using Agilent Whole Human Genome 4644K Microarrays, which represent more than 41,000 unique human transcripts. Bioinformatic analysis was performed using Partek genomic suite (non hierarchical clustering) and Cytoscape with ClueGo plugin (pathway analysis). Viral physical status was determined by integration specific 3’-APOT RACE-PCR and/or ligation mediated DIPS-PCR. AKR1C1 and AKR1C3 expression was confirmed by RT-qPCR on a subset of samples. An additional series of 26 HPV16-positive OPSCC was used to validate gene expression by RT-qPCR and immunohistochemistry. Survival analysis was performed using the Kaplan-Meier method.

Results: Non-hierarchical clustering of mRNA expression profiles resulted in two main groups of expression patterns i.e. one associated with exclusively integrated viral DNA and another associated with the presence of episomal virus DNA. Comparison of these two groups in respect to the gene ontology (GO) terms revealed several deregulated pathways in which the aldo-keto-reductases 1C1 and 1C3 were frequently involved. RT-qPCR and immunohistochemical analysis confirmed upregulation of gene expression in OPSCC with exclusively integrated viral DNA. Furthermore, we showed by survival analysis of an additional group of 26 HPV16-positive OPSCC a worse overall survival for those patients with tumors showing upregulation of AKR1C1 or AKR1C3 (both p < 0.0001).

Conclusion: Our findings show that OPSCC containing exclusively integrated HPV16 DNA harbour a different gene expression profile than tumors containing episomal viral DNA. In particular, upregulation of AKR1C1 and -3 expression was identified in the former group and correlated with a worse outcome. This is in agreement with data on other tumors making these genes promising candidates as indicators of prognosis. In addition, the availability of inhibitors of these gene products may be utilized for drug treatment.