Chromosomal integration of high-risk human papillomavirus (HR-HPV) genomes is believed to represent a significant event in the pathogenesis of cervical cancer associated with progression from preneoplastic lesions to invasive carcinomas. This hypothesis is based on experimental data suggesting that integration-dependent disruption of HR-HPV E2 gene functions is important to achieve neoplastic transformation and on clinical data gathered by analyzing lesions induced by human papillomavirus (HPV) 16 and 18 that revealed integrated viral genome copies in the vast majority of cervical cancer cells. However, a substantial fraction of cervical cancers is associated with other HR-HPV types for which virtually no data concerning their integration status have been reported so far. Here, we compared integration frequencies of the five most common oncogenic HPV types (HPV16, 18, 31, 33, and 45) in a series of 835 cervical samples using a specific mRNA-based PCR assay (Amplification of Papillomavirus Oncogene Transcripts). Most precancerous lesions displayed exclusively episomal viral genomes, whereas 62% of the carcinomas had integrated viral genomes. However, the frequency of integrated HR-HPV genomes showed marked differences for individual HR-HPV types. HPV16, 18, and 45 were found substantially more often in the integrated state compared with HPV types 31 and 33. The analysis of the median age of patients with high-grade precancerous lesions and invasive cancers suggests that precancers induced by HPV types 18, 16, and 45 progress to invasive cervical cancer in substantially less time compared with precancers induced by HPV types 31 and 33. These findings suggest that integration of oncogenic HPV genomes in cervical lesions is a consequence rather than the cause of chromosomal instability induced by deregulated HR-HPV E6-E7 oncogene expression. Distinct HR-HPV types apparently provoke chromosomal instability in their host cells to a different extent than is reflected by their integration frequencies in advanced lesions and the time required for CIN 3 lesions to progress to invasive cancer.

HPV DNA integration into the host genome is a characteristic but not an exclusive step during cervical carcinogenesis. It is still a matter of debate whether viral integration contributes to the transformation process beyond ensuring the constitutive expression of the viral oncogenes. There is mounting evidence for a non-random distribution of integration loci and the direct involvement of cellular cancer-related genes. In this study we addressed this topic by extending the existing data set by an additional 47 HPV16 and HPV18 positive cervical carcinoma. We provide supportive evidence for previously defined integration hotspots and have revealed another cluster of integration sites within the cytogenetic band 3q28. Moreover, in the vicinity of these hotspots numerous microRNAs (miRNAs) are located and may be influenced by the integrated HPV DNA. By compiling our data and published reports 9 genes could be identified which were affected by HPV integration at least twice in independent tumors. In some tumors the viral-cellular fusion transcripts were even identical with respect to the viral donor and cellular acceptor sites used. However, the exact integration sites are likely to differ since none of the integration sites analysed thus far have shown more than a few nucleotides of homology between viral and host sequences. Therefore, DNA recombination involving large stretches of homology at the integration site can be ruled out. It is however intriguing that by sequence alignment several regions of the HPV16 genome were found to have highly homologous stretches of up to 50 nucleotides to the aforementioned genes and the integration hotspots. One common region of homologies with cellular sequences is between the viral gene E5 and L2 (nucleotides positions 4100 to 4240). We speculate that this and other regions of homology are involved in the integration process. Our observations suggest that targeted disruption, possibly also of critical cellular genes, by HPV integration remains an issue to be fully resolved.

Virus integration into the host genome is a characteristic step during cervical carcinogenesis. Experimental data provide evidence that integration could result in increased levels of oncogene (E6/E7) transcripts. This is the first study in which the level of viral transcripts is correlated to the physical state of the viral genome in cervical intraepithelial neoplasia (CIN) and cervical carcinomas (CxCa). Using the APOT-assay integrate-derived transcripts only were detected in 3/28 (11%) CIN and in 28/55 (51%) carcinomas, respectively. The remaining biopsies contained either episome-derived transcripts only or both mRNA species. SybrGreen real time reverse transcriptase-PCR assays were used to quantify viral gene expression for (i) all transcripts initiated from p97, (ii) full-length E6, (iii) E6*I and (iv) E5 transcripts. E6/E7 transcript levels showed a broad distribution but similar median values irrespective of histopathological grading and physical state of the viral genome. Biopsies with integrate-derived transcripts only generally lacked E5-specific mRNA. Our data do not support the hypothesis that HPV integration invariably results in high levels of oncogene transcripts. Instead, constitutive expression of oncogene transcripts rather than the level of expression appears to be decisive for transformation and the maintenance of the malignant phenotype.