The development of cervical cancer is highly associated with human papillomavirus (HPV) infection. Greater than 99% of all cervical tumors contain HPV DNA. Integration of high-risk HPV has been temporally associated with the acquisition of a malignant phenotype. Recent work from our lab has shown that HPV16, the most common high-risk HPV associated with cervical carcinoma, preferentially integrates at loci containing human common fragile sites (CFSs). CFSs are regions of genomic instability that have also been associated with deletions, translocations, and gene amplification during cancer development. The current work shows that HPV18, the second most prevalent high-risk HPV type found in cervical tumors, preferentially targets the CFSs. We identified 27 unique HPV18 integrations in cervical tumors, of which 63% (P<0.001) occur in CFSs. However, the distribution of HPV18 integrations found were profoundly different from those found for HPV16. Specifically, 30% of all HPV18 integrations occurred within the chromosomal band 8q24 near the c-myc proto-oncogene. None of the HPV16 integrations occurred in this region. Previous low-resolution mapping suggested that c-myc may be a target of HPV integration. Our data at nucleotide resolution confirm that in HPV18-positive cervical tumors, the region surrounding c-myc is indeed a hot spot of viral integration. These results demonstrate that CFSs are preferred sites of integration for HPV18 in cervical tumors. In addition, we have identified multiple cellular genes that have been disrupted by HPV18 integration in cervical tumors. Our results suggest that the sites of HPV18 integration are nonrandom and may play an important role in the development of cervical tumors.
Cervical cancer is highly associated with the presence of human papillomavirus (HPV) infection. Greater than 99% of cervical cancers contain HPV sequences (Walboomers et al., 1999; Bosch et al., 2002). HPVs are a family of circular double-stranded DNA viruses that are subdivided into two classes, low and high risk, based on their association with benign lesions or invasive carcinomas, respectively. The two most commonly observed high-risk HPV subtypes found in cervical cancers are HPV16 and HPV18. These viruses infect basal epithelial cells and are frequently found in the episomal state in low- and high-grade intraepithelial lesions (LSIL and HSIL, respectively) (Cho, 1998). Integration of the HPV genome frequently results in the functional elimination of the viral E2 repressor and remodeling of the E6/E7 promoter, thus leading to the continued expression of the E6 and E7 oncoproteins (Ham et al., 1991; McBride et al., 1991; Bechtold et al., 2003). These are potent antagonists to the p53 (Scheffner et al., 1990) and pRB (Boyer et al., 1996) tumor suppressor genes, and their overexpression contributes to the malignant phenotype. However, overexpression of E6 and E7 is only sufficient to immortalize primary epithelial cells; additional alterations such as activating ras mutations are required to transform them (Crook et al., 1988). This suggests that additional human mutations need to take place in order to transform the cells.
We previously described work on the identification of multiple HPV16 integrations in cervical tumors using a rapid PCR-based technique to identify human sequences flanking the sites of viral integration (Thorland et al., 2000, 2003). This work revealed that 48% (P<0.001) of the integrations occurred within common fragile sites (CFSs). CFSs are described as specific loci that form gaps or breaks in metaphase chromosomes from cells that have been challenged with chemicals that induce replicative stress (Glover et al., 1984). There are 89 CFSs present in the human genome, most of which are induced by the DNA polymerase inhibitor aphidicolin. Unlike the rare fragile sites (FRA11B, FRAXA, FRAXE, etc.), CFSs are present in all primates, lower mammals, and possibly yeast (Cha and Kleckner, 2002). The mechanism of CFS fragility is just beginning to be understood. Recent data suggest that the Ataxia-telangiectasia and Rad3-related (ATR) gene products are involved in maintaining DNA integrity at CFS loci (Casper et al., 2002). Loss of ATR expression was shown to produce breakage without the addition of exogenous chemicals, indicating that without proper genome surveillance, CFSs are expressed in otherwise unaltered cells. This expression can promote genomic events such as translocations, sister chromatid exchange, chromosomal deletions and rearrangements, and most pertinent to the current study, the integration of tumor-associated viruses (Glover and Stein, 1987, 1988; Rassool et al., 1991; Thorland et al., 2000).
In this report, we describe our work characterizing HPV18 integrations by RS–PCR. HPV18 is the second most common high-risk HPV subtype, and is found in approximately 20% of cervical tumors (Walboomers et al., 1999), has been associated with more aggressive adenocarcinomas of the uterine cervix (Kurman et al., 1988). Here, we report on the identification of 27 unique HPV18 integrations in cervical tumors. These sites of HPV18 integrations were distinct from those we observed for HPV16 (Thorland et al., 2000, 2003). Specifically, we have identified a cluster of HPV18 integrations surrounding the c-myc proto-oncogene similar to those already reported (Durst et al., 1987; Popescu and DiPaolo, 1989; Couturier et al., 1991).
Our previous analysis of HPV16 integration loci revealed a number of genes implicated in cancer development at or near the sites of viral integration, suggesting that alterations of these genes could play an important role in cervical tumorigenesis. In a similar fashion, we have identified multiple genes that are interrupted by HPV18 integration. These genes include tumor suppressors, proto-oncogenes, as well as genes involved in maintaining DNA integrity. Our results support a model of nonrandom HPV integrations in cervical tumors and indicate that genes at or near the sites of integration may play an important role in tumor development.
Materials and methods
HPV18-positive cervical carcinoma samples were obtained from 32 patients and three cell lines. The 32 cervical tumors were derived from patients from the Mayo Clinic, Leiden University, or the Chinese University of Hong Kong. The three cell lines used were HeLa, SW756, and C4-I (all available from ATCC). Cell culture was performed as recommended by the supplier. DNA was extracted using the IsoQuick kit (ORCA Research). Studies were limited to sensitive DNA-based methodologies.
RS–PCR is a technique that allows for the amplification of unknown sequence that lies adjacent to known sequence (Sarkar et al., 1993). Specific conditions and reagents used, including RS primers and HPV18-specific primers, were previously published (Ferber et al., 2003). Briefly, by using a battery of HPV18-specific primers in conjunction with RS primers, which are semispecific for restriction endonuclease cleavage sites found throughout the human genome, in a two-step heminested PCR system, we are able to amplify HPV18–human junction fragments from cervical tumors. RS–PCR products were electrophoresed on 2.0% agarose gels. Positive bands were selected for sequencing. Additional PCR using HPV18- and human locus-specific primers was performed to confirm the RS–PCR results when a specific junction fragment was identified by sequence analysis (data not shown) (Ferber et al., 2003).
PCR products were prepared for sequencing as previously reported (Ferber et al., 2003). In brief, 5 μl of each positive RS–PCR product was treated with exonuclease-1 and shrimp alkaline phosphatase. Treated aliquots were combined with 3.2 pmol of the proper sequencing primer and submitted to the Mayo Clinic Molecular Core Facility for sequencing.
Computer analysis, BAC clone selection, and FISH
BLASTN searches were performed as previously described (Thorland et al., 2000). Results from these searches were used to identify bacterial artificial chromosome (BAC) clones and genes, if present, which represented the loci of viral integration. BACs were purified and used as probes for FISH analysis as described in Thorland et al. (2003) (Figure 1). These BAC clones were used as FISH probes on normal donor lymphocytes that had been stimulated to express CFSs by adding the DNA polymerase α/δ subunit inhibitor aphidicolin. Metaphase spreads were hybridized with fluorescently labeled BACs representing each integration loci to determine where each integration occurred relative to the CFS from that region (Figure 2a). Fluorescently labeled BACs hybridize to aphidicolin-induced breakage in one of the three possible signal-to-break relationships, proximal, crossing, or distal. Proximal defines a signal-to-break relationship where the labeled BAC hybridizes on the centromeric side of the breakage at the specific fragile site under examination. If breakage spans the region represented by the BAC probe, hybridization signal will be detected on both sides producing a crossing signal pattern. Distal defines a signal-to-break relationship where the labeled BAC hybridizes on the telomeric side of the breakage at the specific fragile site under examination (Figure 2b). A total of 20 clearly visible breaks are counted for each FS under examination. For each BAC, if 19 metaphases have proximal signals while the remaining has a single crossing or distal signal, the BAC is said to be on the fragile site, although at the proximal end of that CFS region. The same is true for a BAC that has one proximal and 19 distal hybridization signals, although that BAC would be localized at the distal end of the CFS region.
HPV18 insertion mapping by RS–PCR and human genome sequence analysis
The RS–PCR technique (Sarkar et al., 1993; Thorland et al., 2000) was used to screen for HPV18 integration in a total of three HPV18-positive cervical carcinoma cell lines and 32 HPV18-positive primary cervical carcinomas. A total of 41 HPV18–human junctions were identified from 22 specimens, representing 30 integration events (Figure 1). Pathology data were available for 18 of the 22 specimens. MC11, MC123, LU3, LU7, HK10, HK13, and HK16 were all squamous cell carcinomas. MC31, MC34, MC360, MC391, MC398, MC415, LU2, LU8, HK9, HK11, and HK15 were all adenocarcinomas. Data for the remaining four samples were not available. Of the 30 events, both 5′ and 3′ junctions were identified in 11 cases, 5′-only junctions were found in five cases, and 3′-only junctions in 14 cases. Each integration event was confirmed by constructing new human loci-specific primers that were used for PCR in conjunction with a battery of HPV18 primers (DNS, Ferber et al., 2003).
As was previously described for HPV16 (Thorland et al., 2003, 2000), many of the HPV18 integration sites were found to occur in human repetitive sequences. Many of the integrations were also associated with deletions and/or complex rearrangements (Figure 1). Of the 41 junction fragments, 18 contain at least a portion of a repetitive element in the human-flanking sequence (nine Alu, seven line 1, and one each of LTR, MIR, and MLT1D). In the 11 cervical tumors and cell lines where both ends of the integration were isolated, we were able to determine the presence of HPV18 and human genomic deletions and/or rearrangements (Figure 1). In every case, the HPV18 genome had lost at least some coding sequences. Human deletions ranged from seven nucleotides (MC398) to over 130 kb (MC34 and MC391). Complex rearrangements were identified in seven of the 11 specimens.
Seven out of 22 cervical carcinoma specimens analysed contained more than a single 5′ and/or a single 3′ integration site (Figure 1). In these multi-integration specimens, five contain multiple hits near the same loci (MC31, MC398, MC415, LU2, and LU8), and may represent complex HPV18 insertion-duplication events. The remaining two specimens (MC123 and MC315) represent cases where single ends were found in different chromosomes, yet did not represent translocations involving the specific chromosomes, as suggested by long-range PCR (DNS).
HPV18 preferentially integrates at loci containing CFS
A total of 30 HPV18 integrations were identified. Several tumors contained three or more junction points within the same BAC, making the total number of unique integrations difficult to ascertain. For example, four junction loci were found in tumor MC398. Two of these were 5′ junctions, two were 3′ junctions, and all occurred within the region spanned by BAC 255B23 (Figure 1). We were unable to determine if cases like this represent two individual integration events or a duplication of a portion of the HPV18 genome at the time of integration. Thus, for the purpose of determining the frequency at which HPV18 integrates into CFS, this pattern of integration can only be counted as a single event, leaving the total number of informative integrations at 27. Using the FISH method outlined earlier, we were able to show that 17/27 (63%, P<0.001) of the HPV18 integrations rescued occurred in CFSs (Table 1), similar to that recently reported for HPV16 (Thorland et al., 2003).
In addition, this work has positioned eight additional CFS with nucleotide resolution for the first time. These include FRA1K (1q31.1, BAC 453E3), FRA2F (2q22.1, BAC 164E7), FRA5D (5q15, BAC 33A7), FRA8C (8q24.1, BACs 468O2, 255B23, and 414A9), FRA9D (9q22.1, BAC 10K22), FRA14C (14q24.1, BAC 723P16), FRA15A (15q23, BAC 170E16), and FRA19A (19q13, COS F13825). This brings the total number of precisely localized CFSs to 17.
HPV18 integrations cluster around 8q24.1/c-myc/FRA8C
Previous studies have shown that several HPV integrations have occurred in the c-myc region (Durst et al., 1987; Popescu et al., 1990; Couturier et al., 1991). In this study, we have identified 7/22 (32%), including the previously described HeLa cell line, specimens with HPV18 DNA integrated at this locus (Figure 1, all samples marked in boldface type, and Figure 3). Our greatly expanded data set, with nucleotide resolution shows that this 8q24.1 hot spot encompasses a 4 Mb region, including a portion of the FRA8C CFS 5′ of c-myc and extending to regions just 3′ of c-myc (Figure 3). The HPV18 integrations in cervical tumors MC391, MC398, MC31, HeLa, HK10, and LU8 were all found to occur in FRA8C (top of Figure 3, box 1, and Table 1) represented by BACs 468O2, 255B23, 414A9, and 382A18, thus spanning a distance just over 4 Mb (based on the UCSC genome browser November 2002 freeze). FISH data for MC391 and BAC 468O2 (Table 1) suggest that this integration is actually much closer to the other integrations in the cluster on FRA8C. In support of this theory, data from the human genome project indicate that there are several gaps that remain in this region and depending on how they are resolved, portions of this contig may be rearranged, bringing the distance across this region down to nearly 2 Mb.
The integrations in boxes 1 and 2 of Figure 3 define a much smaller region encompassing just less than 1 Mb. Box 1 itself represents a region spanning 250 kb, yet holds 5/8 (63%) of all c-myc region HPV18 integrations. Within that group, samples MC31, HeLa, HK10, and LU8 define a subcluster of approximately 88 kb in size, 530 kb 5′ to the c-myc proto-oncogene. An integration cluster of this density and size has never been reported for any HPV in cervical carcinomas.
Box 2 illustrates the two primary cervical tumors, MC31 and MC415, with HPV18 integrations found 19 and 155 kb 3′ to c-myc, respectively. FISH data corresponding to clones in this region show that these integrations do not take place in FRA8C or FRA8D (BAC 80K22 in Table 1 represents BACs 2056O22 and 2267H22 of Figure 3, box 2). The distal end of FRA8C is defined by BAC 382A18 (which overlaps to a large extent with BAC 414A9, but has been illustrated as a minimal overlapping contig in Figure 3) and is located 400 kb 5′ to c-myc, while the proximal end of FRA8D begins roughly 1 Mb 3′ from c-myc (DNS). This shows that the proto-oncogene c-myc is flanked on both sides by the highly unstable FRA8C and FRA8D CFSs.
HPV18 integration leads to genomic alteration of cellular genes
As was previously reported for HPV16 (Thorland et al., 2003), many of the HPV18 integrations mapped in this study interrupted or deleted human genes. The interrupted genes are listed in Table 2. Tumor MC11 had HPV18 integrated in the 5′ promoter region of hTERT, the catalytic subunit of human telomerase (Meyerson et al., 1997). A full report on this integration can be found in Ferber et al. (2003). The entire coding sequence of RNGTT, a vital mRNA capping protein (Pillutla et al., 1998), was deleted in tumor MC34. Tumor MC123 had three separate integrations. The first was in intron 4 of the fragile histidine triad (FHIT) tumor suppressor gene (Boldog et al., 1993; Le Beau et al., 1998). The second occurred in exon 7 of the KIAA1838 gene. The third integration was in exon 8 of the EPHX2 gene, which functions to metabolize xenobiotic epoxides (Fretland and Omiecinski, 2000). Tumor LU2 had HPV18 integrated in the 5′ promoter region of CEACAM5, a transmembrane cell adhesion molecule (Eidelman et al., 1993). Tumor LU7 was interrupted in intron 1 of CD34, the protein expressed on hematopoietic progenitor cells and involved in cell–cell adhesion (Simmons et al., 1992). The candidate tumor suppressor LRP1B was disrupted at intron 41 in tumor HK9 (Liu et al., 2000). Tumor HK11 contained HPV18 DNA in PRKACA, the catalytic subunit of cAMP-dependent protein kinase (Gonzalez and Montminy, 1989). Finally, in HK13, Fanconi's anemia complementation group (FANCC) was altered in intron 7. FANCC is one of the many genes involved in double-strand DNA breaks found in Fanconi's anemia patients (Ahmad et al., 2002).
The loci of HPV16 vs HPV18 integration in cervical carcinoma
Our lab had previously isolated 23 HPV16 integrations from cervical carcinoma specimens (Thorland et al., 2003, 2000), and now report the isolation of 27 HPV18 integrations. When comparing the profile of integrations of the two HPV types, we see little overlap (Figure 4). Only the loci at 3p14, 5p15, and 8q21 have had both an HPV16 and an HPV18 viral integration in their respective cervical tumors. The 3p14 locus contains the FHIT gene and FRA3B (Zimonjic et al., 1997); 5p15 contains the hTERT gene (Meyerson et al., 1997). The 8q21 band does not contain a CFS or a gene that would suggest a reason for integration selection at this locus. HPV16 shows two clusters of integration, each consisting of three separate events. One cluster is located at 13q22, the other at 17q23. The cluster at 13q22 occurs over a tight, 200 kb region and contains several genes that may be important to the development of cervical carcinoma (Thorland et al., 2003). RS–PCR of HPV18-positive tumors revealed a single large cluster at 8q24. Interestingly, RS–PCR has yet to identify a single HPV16 integration at 8q24. However, other groups using various techniques have identified HPV16 integration near 8q24 (Couturier et al., 1991; Brink et al., 2002; Wentzensen et al., 2002).
The development of cervical carcinoma is almost always associated with the presence of high-risk HPV infection (Walboomers et al., 1999). The acquisition of a malignant phenotype is temporally associated with the integration of HPV into the human genome (Cullen et al., 1991). Viral integration generally involves the interruption or deletion of the viral E2 repressor protein, which normally functions to regulate the expression of the E6 and E7 oncoproteins (Howley, 1996; Shah and Howley, 1996; Munger, 2002). This regulation has recently been suggested to also involve the remodeling of the E6/E7 promoter (Bechtold et al., 2003). E6 and E7 interact with the tumor suppressor proteins p53 and pRB, respectively. In this manner, viral integration into the human genome provides a mechanism by which the cell accumulates genomic damage (zur Hausen, 1991). However, transient overexpression of E6 and E7 alone is not sufficient for transformation, suggesting that additional mutations are required (Hawley-Nelson et al., 1989; Woodworth et al., 1989). Given enough time, these mutations may occur without the need for viral integration. An alternative hypothesis is that the integration event itself may contribute to carcinogenesis. Indeed, several groups have reported the occurrence of large deletions or rearrangements at the site of HPV integration (Wilke et al., 1996; Gallego et al., 1997; Thorland et al., 2003). In addition, HPV integration has been shown to occur near cellular proto-oncogenes (Durst et al., 1987; Lazo et al., 1989; Couturier et al., 1991). Together these data suggest that HPV integration could directly influence gene expression by changing the normal human DNA composition. Therefore, we propose that in some cervical cancers, in addition to the resulting E6 and E7 expression, the integration event itself is a nonrandom, irreversible mutation that may contribute to the malignant phenotype. It is also possible that the CFS nature of the integration site is responsible for the acquisition of additional chromosomal aberrations. Indeed, chromosomal translocations are frequently found near HPV integration sites in cervical cancer (Popescu et al., 1990; Koopman et al., 1999; Macville et al., 1999; Brink et al., 2002).
Recent work from our lab has shown that for HPV16, the most common high-risk HPV subtype found in cervical tumors, integration occurs throughout the human genome. However, 48% (P<0.001) of the integrations occurred in the 2% of the genome known as CFSs (Thorland et al., 2003, 2000). In addition, several integrations were shown to disrupt cellular genes that may play a role in tumor development. HPV18, the second most common high-risk HPV type found in cervical tumors, is a distant relative of HPV16, and is more closely related to several low-risk HPVs than it is to HPV16 (Shah and Howley, 1996). However, HPV18 has been associated with more aggressive adenocarcinomas of the endocervix (Schwartz et al., 2001). In the current study, we therefore sought to determine the relationship between HPV18 integration, CFSs, and cancer-related genes.
Using the established RS–PCR technique, we were able to identify 27 unique HPV18 viral integrations in 22 cervical carcinoma specimens. Positioning of these integrations relative to the CFSs revealed that 17/27 (63%, P<0.001) occurred in CFS regions, demonstrating that these unstable genomic regions are also hot spots for HPV18 integrations. Given that non-CFS elements such as the c-myc proto-oncogene may be responsible for the large number of 8q24.1 integrations and thus skewing our data, we repeated the analysis without these data and found that 11/19 (58%, P<0.001) integrations occur in CFSs. These data support the general statement that high-risk HPVs preferentially integrate at CFSs in cervical tumors. Interestingly, while both HPVs preferentially integrated at CFSs, the sites of HPV18 integration were for the most part distinct from those found in HPV16-positive cervical tumors (Figure 4). In addition, we identified a 4 Mb region surrounding c-myc and the FRA8C CFS that contained HPV18 viral integration in 7/22 (32%) of the cervical tumors analysed. While several groups have reported integrated HPV DNA in this region, the current study is of a much larger scale and provides integration data with nucleotide resolution.
The c-myc proto-oncogene has been shown to be deregulated in many tumor types including breast, colon, small-cell lung carcinomas, osteosarcomas, glioblastomas, melanoma, myeloid leukemias, and cervical tumors (reviewed in Pelengaris et al., 2002). In Burkitt's lymphoma, translocations between the region surrounding c-myc and the immunoglobulin heavy chain, κ light, or λ light chains result in overexpression of c-Myc (Boxer and Dang, 2001; Ratsch et al., 2002). These translocations occur over a region both 5′ and 3′ to c-myc, encompass 500 kb, and result in the placement of enhancer regions, from the Ig loci, more than several 100 kb from c-myc, yet result in upregulated c-Myc expression (Boxer and Dang, 2001; Ratsch et al., 2002). In a similar fashion, HPV18 contains viral enhancers that may alter gene expression at this locus. While the cluster of integrations described in the current study represent a much broader region of over 4 Mb (Figure 3), the integrations found in tumors MC415 and the downstream end of MC31 (Figure 3, box 2) occur over a distance in agreement with described BL translocations. The integrations represented in box 1 of Figure 3, however, are over 500 kb 5′ of c-myc, 300 kb farther 5′ than any reported BL translocation. The tightest portion of this subcluster represents ∼88 kb (between MC31 and LU8) and contains no known genes. While it is possible that these integrations affect the expression of c-myc, further studies are underway to determine if the presence of HPV DNA at this subcluster is sufficient to alter the expression of additional genes surrounding this locus.
During the course of this study, we found 10 examples where HPV18 integration disrupted known human genes (Table 2). The function of the KIAA1838 gene is unknown. Three others (EPHX2, SLC1A3, and RNGTT) perform functions that are difficult to tie to tumor development. The remaining six genes each have potential roles in the development or progression of cervical tumorigenesis.
Loss of tumor suppressor function is a hallmark of many cancers. In tumor MC123, HPV18 disrupted intron 4 of the tumor suppressor FHIT gene. The FHIT gene is located within FRA3B, the highest expressed human CFS, and has been shown to harbor HPV16 integrations (Glover and Stein, 1988; Wilke et al., 1996). In addition, loss of FHIT expression has been found in a number of premalignant cervical lesions (Greenspan et al., 1997; Huebner et al., 1998). Together, this suggests that HPV18 integration at this site may promote cervical carcinogenesis.
Another hallmark of carcinogenesis is the acquisition of unrestrained replicative capacity. In many tumors, this is via the activation of telomerase (Kim et al., 1994). hTERT expression is carefully regulated as it is the catalytic subunit of telomerase (Meyerson et al., 1997). The placement of viral DNA containing powerful enhancers in the hTERT promoter region could result in gene activation, thus allowing the cell to divide indefinitely without the catastrophic effect of eroding telomeres. In this study, we identified a single HPV18 integration in tumor MC11, occurring 7 kb upstream from the start site of hTERT transcription. In addition, we recently described two additional HPV16 and two hepatitis B virus (HPV) integrations into the hTERT gene in cervical and hepatocellular carcinomas, respectively. Each of these samples overexpresses hTERT (Ferber et al., 2003). Additional work is underway to determine if the overexpression is due to the presence of integrated viral DNA.
In tumor HK11, HPV18 DNA integrated into exon 10 of the cAMP-dependent protein kinase, catalytic subunit A (PRKACA) gene. PRKACA is one of the two catalytic subunits in the cAMP-dependent protein kinase (PKA) complex (Mayr and Montminy, 2001). Active PKA phosphorylates the transcription factor cyclic AMP response element-binding protein (CREB) (Gonzalez and Montminy, 1989). This phosphorylation leads to the activation of many genes, including BRCA1, retinoblastoma, and Bcl-2. Given the role of these genes in tumor suppression, disrupting PRKACA and thus inhibiting CREB function may have tumor-promoting effects within the cell.
Loss of genomic stability is another hallmark of cancer. In cervical cancer this loss is attributed to the E6 and E7 oncoproteins that inactivate p53 and pRB, respectively. In tumor HK13, HPV18 was integrated into the FANCC gene. Mutations in FANCC result in nonspecific double-stranded DNA breaks and are clinically manifested as Fanconi's anemia (Ahmad et al., 2002). In cervical epithelial cells, integration into this gene could augment the effects of E6 and E7, thus creating an environment ripe for DNA damage and subsequent cellular transformation. Integration into FANCC is not restricted to HPV18. Wentzensen et al. (2002) previously detected a fusion transcript of HPV16 and part of FANCC.
Several additional genes that were disrupted by HPV18 integrations included three different cell–cell adhesion molecules. The first integration to disrupt a cell–cell adhesion molecule occurred in CEACAM5, a molecule that inhibits cellular differentiation (Eidelman et al., 1993). Here, HPV18 DNA integrated into the 5′ promoter region. An in-frame fusion transcript of HPV16 and CEACAM5 was described previously (Wentzensen et al., 2002). The second cell–cell adhesion molecule, CD34 which is normally found on hematopoietic stem cells (Simmons et al., 1992), was disrupted in intron 1 by HPV18 DNA. The third cell–cell adhesion molecule was LRP1B, a putative tumor suppressor that is deleted in non-small-cell lung cancer cell lines (Liu et al., 2000). In tumor HK9, HPV18 was found to interrupt intron 41 of this gene. These three integrations represent mutations that could promote tumor development by altering the way the epithelial cell interacts with neighboring cells or the extracellular matrix.
These data and that from Thorland et al. (2003), (2000), strongly suggest that HPV integration in cervical tumor cells is not random, and that it does target the CFSs. Furthermore, our data shows a large number of interesting genes at the sites of HPV18 integration, suggesting that in many tumors the integration event itself may trigger additional expression changes and thus provide growth advantage to the cell. Our work suggests that CFSs and important oncogenesis-related genes are frequent targets of HPV integration. Identifying the role that HPV insertion plays in gene dysregulation and subsequent cellular transformation will help to determine how viral integration contributes to malignant transformation of cervical epithelial cells. It is imperative that we understand the role of viral integration in the pathophysiology of cervical cancer, if we hope to formulate rational plans for molecular diagnostic and therapeutic interventions.
We thank Gert Jan Fleuren and Ed Schuuring for their contributions to this paper. This study was conducted as a part of MJF's doctoral thesis requirements in the Mayo Graduate School. DIS was supported by NCI Grant CA 48031. AATPB was financially supported by the Vanderes Foundation, Leiden, The Netherlands.
About this article
Nature Communications (2014)