Loss of gene function as a consequence of human papillomavirus DNA integration

Authors


Abstract

Integration of the human papillomavirus (HPV) genome into the host chromatin is a characteristic step in cervical carcinogenesis. Integration ensures constitutive expression of the viral oncogenes E6 and E7 which drive carcinogenesis. However, integration has also an impact on host DNA. There is increasing evidence that integration not only occurs in fragile sites and translocation breakpoints but also in transcriptionally active regions. Indeed, a substantial number of integration sites actually disrupt host genes and may thereby affect gene expression. No doubt, even subtle changes in gene expression may influence the cell phenotype but small fold changes are difficult to quantify reliably in biopsy material. We have, therefore, addressed the question whether a complete loss of gene function that is insertional mutagenesis in combination with deletion or epigenetic modification of the second allele is also a phenomenon pertinent to cervical cancer. Out of the ten preselected squamous cell carcinomas analyzed, all viral integration sites were located within the intron sequences of known genes, giving rise to viral–cellular fusion transcripts of sense orientation. Moreover, for two tumors, we provide evidence for complete functional loss of the gene affected by HPV integration. Of particular note is that one of the genes involved is the recently described novel tumor suppressor gene castor zinc finger 1. Although our study provides no functional proof that any of the genes affected by HPV integration are causally involved in the transformation process, an exhaustive systematic look at the role of insertional mutagenesis in cervical cancer appears to be warranted.

Persistent infection with high-risk human papillomavirus (HPV) is the main risk factor for the development of high-grade precancerous lesions (cervical intraepithelial lesions, CIN2 and 3) and cervical carcinoma,1, 2 with HPV 16 being the most prevalent type, followed by HPV 18, 31, 33 and 45.3 The immortalizing and transforming potential of the virus are ascribed to the oncogenes E6 and E7, underlined by their constitutive expression in HPV-induced cervical carcinomas.4, 5 The E6 and E7 oncoproteins mediate mitogenic and antiapoptotic stimuli by interacting with numerous regulatory proteins of the host cell that control the cell cycle.6, 7 Moreover, the viral oncoproteins induce mitotic defects and genomic instability by uncoupling centrosome duplication from the cell division cycle.8, 9

One factor considered to be of key importance for the progression of cervical intraepithelial neoplasias (CIN) to invasive cervical cancer is the integration of HPV into the host genome.10, 11 Hopman et al.12 have shown by in situ hybridization that 88% of CIN2/3 associated with microinvasive carcinoma contained integrated viral DNA. In contrast, the characteristic punctuate signal of integrated viral DNA was observed only in 29% of solitary CIN2/3. High integration frequencies have also been observed in several further studies in which other techniques for the detection of integrated viral DNA (DIPS, E6/E2 ratios) or integrate-derived viral transcripts (amplification of papillomavirus oncogene transcript, APOT) were used.13–16 Integration seems to provide the cell with a selective growth advantage. However, it remains a matter of debate whether HPV DNA integration precedes E6/E7-induced genetic instability or rather is a consequence thereof.11, 17 In an in vitro study, acquisition of high levels of genomic instability in cervical keratinocytes in monolayer culture occurred after integration of HPV16.18 On the other hand, Melsheimer et al.19 have shown that in CIN and cervical carcinomas aneuploidization precedes integration of HPV DNA in the progression of cervical dysplasia. In this context, another study is of interest which showed highly variable levels of viral oncogene expression in CIN and CxCa.20 Moreover, these levels were independent of histological grading and the physical state of the viral genome, suggesting that HPV integration per se does not necessarily result in high viral transcript levels but primarily ensures their constitutive expression. The minimal levels of viral oncogene expression required to initiate carcinogenesis and to sustain the transformed phenotype are not known.

Besides the effects on viral oncogene expression, a very recent study has shown that HPV integration is responsible for structural alterations of the host genome at the insertion site.21 Peter et al. examined a series of 34 primary cervical carcinomas and eight cervical cancer-derived cell lines using high-resolution genome profiling combined with a systematic identification of the viral integration sites at the molecular level. By this approach, genomic rearrangements were observed in 38% of all cases. Half of these genetic alterations were local amplifications which may have been triggered by activation of the viral origin of replication present within the host genome.22 Moreover, several genomic deletions at the HPV integration site were observed possibly related to DNA fragility induced by the insertion of a heterologous sequence.

To date, more than 250 HPV–host integration sites have been mapped.11, 23–26 All integration loci examined were unique, involving all chromosomes. Fragile sites, translocation break points and transcriptionally active regions were found to be preferred sites for integration.27–29 Interestingly, several are clustered within the cytogenetic bands 4q13.3, 8q24.21, 13q22.1 and 17q21.2.30 The most prominent integration sites are the ones in the location of 8q24.21 near the MYC locus. Worth mentioning are the high levels of MYC expression in these tumors, suggesting that integration may influence transcription of neighboring genes.31–33

Recent data from our laboratory provide evidence that most HPV integration sites involve known or predicted genes, suggesting that the disruption or deregulation of genes through HPV integration could contribute to the process of transformation and carcinogenesis. Indeed, Reuter et al.34 report on the functional inactivation of ZBTB7C, a putative tumor suppressor gene, which had resulted from insertional mutagenesis in combination with loss of heterozygosity. It is postulated that the loss of ZBTB7C function has contributed to the pathogenesis of this cancer cell line. Unfortunately, the tumor from which the cell line ME180 was established was not available for molecular analysis and it cannot be excluded that this genetic aberration has arisen in cell culture.

The aim of our study was to investigate whether HPV integration in combination with further genomic alterations can result in a complete loss of gene expression at the RNA or protein level thereby causing loss of gene function. For this purpose, we performed detailed analyses at the RNA and DNA level for a subset of HPV16- or HPV18-positive squamous cell carcinomas in which cellular genes were directly disrupted by viral integration.

Material and Methods

Clinical samples

All biopsies were taken from patients treated at the Department of Gynecology of Jena University Hospital between 1995 and 2008 and were histopathologically classified as squamous cell carcinoma.

Overall, 121 cervical carcinomas expressing viral–cellular fusion transcripts were identified in our laboratory. In 50 cases, the viral part of the fusion transcript was spliced or fused to cellular sequences of known genes (Kraus et al.30 and unpublished data). For our study, cancers in which the viral-coding sequences were spliced to a regular cellular exon in sense orientation were of particular interest. For these cases (n = 25), it is highly unlikely that the gene affected by integration will encode a wild-type protein because transcripts initiated from the viral promoter will lack upstream cellular exons, whereas transcripts initiated from a cellular promoter will retain viral sequences. Thus, the chance that integration has resulted in insertional mutagenesis is high. Moreover, we postulated that among these cases loss of allele heterozygosity and thus a complete functional loss of the affected gene may have occurred. To test this hypothesis, ten out of the 25 biopsies were available for further characterization. Of these samples, five were positive for HPV 16 and five were positive for HPV 18. The remaining samples were either depleted or not suitable for microdissection.

Nucleic acid isolation of clinical samples

Total RNA was isolated using the NucleoSpin RNA II Kit (Macherey-Nagel, Düren, Germany) according to the protocol for RNA isolation from tissues. In brief, 10 × 10 μm sections of frozen tissue were taken. Samples were homogenized by using injection needles with a diameter of 0.55 mm to disperse the cells. DNA was removed in all samples by DNase treatment for 15 min (room temperature, RT). Total RNA was eluted in 60-μL RNase-free water and stored at −80°C for further analysis.

Reverse transcription

Total RNA (300–500 ng) was reverse transcribed using 200 units of Superscript II reverse transcriptase (Invitrogen, Carlsbad, CA) and an oligo(dT) primer coupled to a linker sequence (5′-AAG CAG TGG TAT CAA CGC AGA GTA CT(30)VN-3′), referred to as CDS-Primer (Clontech, Heidelberg, Germany). The reaction was performed for 70 min at 42°C in a final volume of 20 μL according to the protocol of the Superscript II Kit. Forty units of RNaseOUT (Invitrogen, Carlsbad, CA) were added to inhibit RNase activity.

APOT assay

HPV-derived fusion transcripts were amplified using the APOT assay.14 This assay is based on a 3′-rapid amplification of cDNA ends performed in a nested PCR format. HPV E7 primers are used as forward primers and an adapter primer complementary to the linker sequence in the CDS primer as first reverse primer and a shifted one as second nested primer.

The APOT assay was performed as described previously by Klaes et al. with slight modifications regarding the primers used. The reverse primer for the first PCR comprises the sequence 5′-AAG CAG TGG TAA CAA CGC A-3′, the nested PCR primer comprises the sequence 5′-AAG CAG TGG TAA CAA CGC AGA GTA CT-3′. The reaction mixture, containing 20 mM Tris-HCl, 50 mM KCl, 1.5 mM MgCl2, 200 μM dNTPs, 250 nM primer each and 0.75 units recombinant Taq Polymerase (Invitrogen, Carlsbad, CA), was subjected to an initial denaturation step for 5 min at 94°C, followed by 30 cycles of denaturation at 94°C for 30 sec, primer annealing for 30 sec and elongation at 72°C for 2 min. For HPV16, annealing temperatures of 61 and 66°C for the first and second PCRs, respectively, were used; for HPV18, 61 and 68°C. The reaction was ended by a final elongation step at 72°C for 6 min. Two microliters of the first PCR was used as template for the nested PCR step. Both reactions were performed in a volume of 25 μL.

The amplification products were visualized by 1% agarose gel electrophoresis. Products which differed in their size from the major viral transcript (E6*I-E7-E1vE4-E5) were indicative for the presence of viral–cellular fusion transcripts.

Detection of integrated papillomavirus sequences-PCR

Detection of integrated papillomavirus sequences by ligation-mediated PCR (DIPS-PCR) was performed to determine the HPV integration locus on the DNA level.15 The assay was performed as described previously by Luft et al. with slight modifications concerning the annealing temperature and the sequence of the AP1 primer. Briefly, after digestion of the genomic DNA (0.6 μg) with Sau3AI over night, ligation of the enzyme-specific adapters (50 pmol) to the restricted genomic DNA was performed as described previously by Luft et al. The formation of the ds-adapter, comprising one short oligo (19 mer) and one long oligo (45 mer) is the most critical step and is produced as described previously but stored at −80°C. The primer specific for the adapter sequence used in the second PCR is AP1 which we have shortened at the 5′ end by three nucleotides.

PCR amplification was performed in the first, linear round in a total volume of 25 μL as described previously by Luft et al.15 The second, exponential PCR was performed in a final volume of 50 μL with 2 μL template from the linear PCR. Twenty-five microliters of the PCR product was used to perform a Southern blot to identify HPV-positive bands. The remaining 25 μL were then separated on a 1% agarose gel and HPV-positive bands (as determined by the Southern blot) were extracted using the innuPREP DOUBLEpure Kit (AnalytikJena, Jena, Germany) and send to SeqLab (SeqLab, Göttingen, Germany) for sequencing.

Sequence analysis

The integration locus was determined by database alignments using National Centre for Biotechnology Information (NCBI) Blast tool and the University of California, Santa Cruz (USCS) genome browser, release hg19.

Microdissection

Microdissection of the tumor cells was performed with the PALM® MicroBeam using the Laser Microdissection and Pressure Catapulting (LMPC) technology (P.A.L.M. Microlaser technologies, Carl Zeiss Jena, Germany). Cryosections (15 μm) on MembraneSlides NF (P.A.L.M. Microlaser Technologies, Carl Zeiss Jena, Germany) were stained with 2% cresyl violet, diluted in 96% ethanol and tumor cell islands were microdissected. MembraneSlides NF were pretreated with RNaseZAP (Ambion—Applied Biosystems, Austin, TX) and microdissected tissue was applied directly into the first PCR reaction.

One-step RT-PCR assay

A one-step RT-PCR assay was performed to detect the mRNA of the gene affected by HPV integration. Ideally, this assay should detect all known transcripts initiated from the cellular promoter in a single reaction. However, owing to the large size of the transcripts this was not possible. Therefore, as a compromise the assay amplifies an exon-spanning region at the 5′ end, an internal exon-spanning region corresponding to the counterpart of the viral–cellular splice site, and an exon-spanning region at the 3′ end. For tumors in which the viral sequences were spliced to 3′ cellular sequences, an additional 3′ amplification would span the same exons and was thus omitted.

The assay was designed as a nested PCR, where the first PCR is performed as a one-step reverse transcriptase duplex PCR using the SuperScript III One-Step RT-PCR System (Invitrogen, Carlsbad, CA) with two primer pairs targeting the housekeeping gene (HKG) hydroxymethylbilane synthase and the gene of interest (GOI), performed in an Eppendorf Thermo Cycler. The second PCR is performed as simplex PCR in an ABI-7300 (Applied Biosystems, Carlsbad, CA) using the Fast SYBR Green Master Mix (Applied Biosystems, Carlsbad, CA). A schematic overview of the one-step RT PCR assay is shown in Figure 1.

Figure 1.

Schematic overview of the one-step RT PCR assay. The entire transcript of the GOI cannot be detected with a single primer pair. Therefore, exon-spanning primers were designed for the 5′ end, an internal region corresponding to the counterpart of the viral–cellular splice site, and the 3′ end of the gene. GOI: gene of interest; HKG: housekeeping gene.

Microdissected tumor tissue was placed directly into the tubes in which the RT-PCR was run. The first PCR amplification was performed in a final volume of 25 μL containing 2× reaction buffer [SuperScript III One-Step RT-PCR System (Invitrogen, Carlsbad, CA)], primer 400 nM each, 0.5 mM MgCl2, 8 units RNaseOUT (Invitrogen, Carlsbad, CA), 1 μL enzyme mix containing SuperScript III RT and Platinum Taq polymerase and 2.5 units SuperHot Taq-DNA-Polymerase (Genaxxon BioScience, Ulm, Germany). RT-PCR reactions were performed starting with the reverse transcription at 50°C for 30 min followed by an initial denaturation step at 94°C for 5 min, followed by 15 cycles of denaturation at 94°C for 15 sec, annealing at 60°C for 20 sec and elongation at 72°C for 30 sec. The second PCR amplification was also performed in a final volume of 25 μL containing 2× reaction buffer [Fast SYBR Green Master Mix (Applied Biosystems, Carlsbad, CA)], primer 400 nM each and 1 μL template of the first PCR reaction. qPCR reactions were performed starting with an initial denaturation step at 94°C for 10 min, followed by 45 cycles of denaturation at 94°C for 15 sec, annealing at 58–60°C (depending on the primer pair used) for 20 sec and elongation at 72°C for 30 sec (for primer sequences, see Supporting Information data).

Immunohistochemistry

Immunohistochemistry was performed with cryosections for the castor zinc finger 1 (CASZ1) antibody and with paraffin-embedded tissue sections for the LIPC antibody, as well as for p16INK4a and Ki67 as marker for HPV infection and cell proliferation, respectively. Hematoxylin and eosin staining was performed with cryosections of 7-μm thickness on glass slides (Thermo Fisher Scientific, Waltham, MA) using standard protocols. Immunohistochemistry with cryosections was performed on Superfrost Plus slides (Thermo Fisher Scientific, Waltham, MA) with 7-μm sections. After fixation with 4% paraformaldehyde for 10 min, slides were washed with tris-buffered saline (50 mM Tris, 150 mM NaCl) and 0.1% Tween-20 (TBST) and incubated in 0.6% H2O2 for 7 min. After another washing step, slides were blocked with goat serum (1:5 dilution in TBST) for 20 min and afterward staining was performed over night with rabbit polyclonal LIPC antibody, diluted at a ratio of 1:200 (Santa Cruz Biotechnology, Santa Cruz, CA), rabbit polyclonal antibody against CASZ1, diluted at a ratio of 1:3,000 (WG Thiele, USA; Sigma-Aldrich, St. Louis, MO; Abcam, Cambridge, UK), Ki67 mouse monoclonal antibody clone mib-1 (DAKO, 80 mg/L), diluted at a ratio of 1:100 and CINtek p16INK4a mouse antihuman antibody (Roche mtm Laboratories, Heidelberg, Germany), diluted at a ratio of 1:2. Detection was performed using the DAKO EnVision System according to the protocol. Slides were counterstained with hematoxylin and covered with coverslips in gelatin. For paraffin-embedded tissue, paraffin was removed by incubating the slides in Xylol followed by absolute-, 96, 70 and 50% ethanol. After washing with bidistilled water, slides were boiled in citrate buffer (pH = 6.0) and then treated as described above.

Southern blot analysis

Southern blot analysis was done to identify those PCR products of the DIPS assay which comprise HPV sequences. The PCR products were electrophoretically separated in a 1.5% agarose gel, denatured by incubation three times for 20 minutes in 0.6N NaCl/0.4N NaOH and transferred onto a nylon filter. The filter was washed twice with 2× sodium chloride sodium citrate (SSC) for 10 minutes and then crosslinked using a GS Gene Linker (Bio-Rad, Hercules, California, USA). Prehybridization and hybridization were done in 50% formamide, 5× SSC, 1× Denhardt's solution, 0.1 mg/ml yeast t-RNA for 30 minutes and over night at 42°C respectively. 32P-labelled hybridization probes were generated by random priming (Roche Applied Science, Mannheim, Germany) of the complete HPV16 or HPV18 genome. The labelled probe was purified using Mini Quick Spin Columns (Roche Diagnostics, Pleasanton, USA). The filters were washed twice with 2× SSC, 0.1% SDS at room temperature and twice in 0.1× SSC, 0.1% SDS at 50°C. Autoradiography was performed for 18 hours with a Kodak Scientific Imaging film.

Genomic deletion analysis

A duplex nested PCR was performed to assess the integrity of the second allele not affected by HPV integration. The tumor and controls were microdissected, dissolved in ThinPrep PreservCyt Solution and directly used for the first duplex PCR. Beta Globin (HBB) as a quality control and the region of interest (CASZ1 or LIPC) were amplified. The primers for CASZ1 and LIPC span the genomic integration locus of HPV in the respective tumor, and thus a product is only visible if the second allele is present. The first PCR was performed for 15 cycles. The second PCR was performed as single PCR with separate reactions for HBB and CASZ1/LIPC. Here, 30 cycles were carried out in a SybrGreen real-time format using the ABI-7300 cycler (Applied Biosystems, Carlsbad, CA). Visualization was additionally performed by agarose gel electrophoresis (for primer sequences, see Supporting Information data).

Results

Genomic integration sites

HPV integration sites were determined by the DIPS assay.15 This protocol allowed the identification of the viral–cellular junction downstream of the viral promoter. For each of the ten biopsies analyzed, the integration site from which the previously identified viral–cellular fusion transcript was initiated could be identified by DIPS (Table 1). Interestingly, for sample T3231 two additional integration sites located on Xp11.3 and 1q31.2 were identified. As no APOT correlates were found for these two integration sites, it is likely that these sites are transcriptionally silent.

Table 1. HPV integration sites as determined by the DIPS assay
inline image

Of particular note is that two genes are affected twice by integration. In these cases, LIPC and TMEM49 are disrupted at two different sites which are approximately 24 and 17 kb apart, respectively. Moreover, with one exception the breakpoints within the viral genome for all samples were located within the E1 ORF. In all cases, integration occurred within cellular intron sequences (Supporting Information Table 1). In five cases, short overlapping sequences of two to five nucleotides between viral and cellular sequences could be observed. In seven cases, either none or up to five nonhomologous nucleotides are located between viral and cellular sequences.

Viral–cellular fusion transcripts

Tumors with viral–cellular fusion transcripts were identified by the APOT assay. In the subset of tumors chosen for the detailed analysis of our study, all fusion transcripts were spliced from the viral E1 open reading frame to cellular exons of known genes (Table 2). The viral splice donor site was at nucleotide position 880 in case of HPV16 or 929 in case of HPV18. With one exception, the regular cellular consensus acceptor site NCAG was used. In tumor T182e, the acceptor site YATT was used which is less common. Tumors T2723 and T3231 are of particular interest because their viral–cellular fusion transcripts are identical. However, the integration sites within the gene TMEM49 differ (Table 1) and thus different sized introns were evidently removed by splicing. LIPC is another gene affected by HPV integration in two different tumors. In these cases, the viral sequences of the fusion transcript were spliced to exons 5 and 9, respectively. Seven fusion transcripts were terminated by the use of the regular cellular polyA signal, whereas in the other three cases cDNA synthesis seems to be initiated within an adenine-rich region.

Table 2. Characterization of viral–cellular fusion transcripts
inline image

Loss of gene function at the mRNA level

The key question of our study was to investigate whether HPV integration can contribute to a complete functional loss of cellular genes. For each gene affected by HPV integration, we designed a one-step RT-nested PCR assay (Fig. 1). Sensitivity of the assay was ensured by reliably amplifying mRNA of five SiHa cells with each primer pair.

Amplification of the housekeeping gene was positive for all cases. Out of the ten tumors analyzed, two showed no mRNA expression of the affected gene (Table 3). In case of tumor T2319, a PCR product corresponding to the 5′ region of the CASZ1 transcripts was obtained for one out of the three microdissected tumor cell areas. This was probably owing to the presence of normal cells within the microdissected areas. Moreover, in this tumor no PCR products were obtained with the internal exon-spanning primers. Similarly, tumor T182e in which LIPC was disrupted by integration no PCR products were obtained with internal exon-spanning primers (Table 3). To confirm the reliability of the one-step RT-nested PCR assay, microdissected tumor cell areas from other samples were amplified successfully (data not shown). We then examined transcription of these two genes in ten further cervical carcinomas, 13 cell lines, five CIN and five normal tissues (data not shown). For this purpose, the tissues were not microdissected. CASZ1 transcripts were detectable in all tissues, whereas LIPC was not expressed in two tumors and four cell lines (C4-1, ME180, C33A and HPKII), but in all CIN and normal tissues analyzed.

Table 3. One-step RT duplex PCR assay to determine loss of gene function at the RNA level
inline image

Validation of functional loss at the protein level

To validate the negative mRNA results at the protein level, we performed immunohistochemical staining on cryosections (CASZ1) and paraffin sections (LIPC) of the respective tumors. Unfortunately, none of the antibodies tested was specific for CASZ1. However, we could confirm the loss of LIPC in tumor T182e (Fig. 2). Moreover, tumors T5234 and T4749 which also lacked LIPC transcripts were negative by immunohistochemistry (Fig. 2). The immunohistochemical analysis was then extended by using tissue arrays. Loss of LIPC expression was evident for 1 out of 11 cervical cancers but not in any of the normal epithelia (n = 41), metaplasia (n = 18) or CIN (n = 36) (data not shown).

Figure 2.

Immunohistochemical staining of squamous cell carcinoma biopsies for LIPC (lipase, hepatic). All tumor cell areas typically overexpress p16, a surrogate marker for HPV transformation. Moreover, with exception of case T4749 all tumor cell areas stain for the proliferation marker Ki67. LIPC expression is evident only for tumors T4807 and T4822.

Validation of functional loss at the DNA level

For both CASZ1 and LIPC, primer pairs were designed which span the region of the unaffected allele that is equivalent to the viral–cellular junction. In case of tumor T182e (LIPC), a PCR product was detectable and thus does not provide evidence for deletion (data not shown). By contrast, for tumor T2319 no PCR product for CASZ1 could be amplified from the microdissected tumor cell areas. Stroma cells from the same tumor served as a positive control (Fig. 3). Thus, tumor T2319 is homozygous for the allele affected by HPV integration and cannot express a functional CASZ1 protein.

Figure 3.

Deletion analysis of CASZ1 for tumor T2319. A duplex-nested PCR was performed amplifying HBB (quality control) and CASZ1. Tumor cells from T2319 show no PCR product for CASZ1, leading to the assumption that the second allele not affected by HPV integration is not present. Stroma cells from the same tumor section as well as cells from the tumor T4947 show an amplification product. Further positive controls are DNA from cultured human fibroblasts and keratinocytes. *Microdissected cells directly used for PCR; **20 ng isolated DNA from cultured cells; ***NTC 1 and 2: nontemplate control for first (1) and nested (2) PCR.

Discussion

Deregulation of viral oncogene expression as a consequence of HPV integration is considered to be a crucial step in the development of invasive cervical carcinoma. As integration occurs at random sites, the possible contribution of HPV insertional mutagenesis in the process of carcinogenesis has not received much attention. However, in a recent study we have shown that viral integration frequently occurs in transcriptionally active regions of the host chromatin.30 In that study, 38% of viral–cellular fusion transcripts from individual cervical biopsies contained cellular sequences which were homologous to known genes. These results led us to postulate that at least in some cervical carcinomas HPV integration into a gene involved in cell and tissue homeostasis could alter its expression and thereby influence the cell phenotype. As subtle changes in gene expression are difficult to prove, we addressed the question whether a complete loss of gene function that is insertional mutagenesis in combination with deletion or epigenetic modification of the second allele may also be a scenario pertinent to cervical carcinogenesis.

For our study, we had, therefore, selected ten cervical carcinomas characterized in our laboratory in which the sequences of the viral E1 ORF were spliced in sense orientation to a cellular exon. Subsequent DIPS analysis revealed that HPV integration has always occurred within an intron sequence of the respective gene thereby confirming the phenomenon of insertional mutagenesis. Most importantly, evidence for a complete loss of a functional gene was obtained for two cases. In tumor T2319, the viral–cellular fusion transcript was spliced to the last exon (exon 21) of CASZ1. The viral insertion site is located within an intron region flanked by exons 20 and 21 of CASZ1, 9,963 bp upstream of the cellular acceptor site. RT-PCR of CASZ1 transcripts using primers that span exons 20 and 21 did not yield a PCR product, suggesting that in this tumor CASZ1 of the second allele was not transcriptionally active. Moreover, no PCR products were obtained with primers spanning the viral integration site thereby providing clear evidence that CASZ1 of the second allele is functionally inactivated owing to deletion. As for RT-PCR, PCR for gene loss was performed with microdissected tumor cell areas in which a control gene was amplified in a duplex reaction with CASZ1. As the primer efficiencies of the control gene HBB and CASZ1 were similar, a competitive reaction which could influence the results can be excluded. Moreover, CASZ1 could be amplified from the stromal component of the tumor. Currently, we cannot completely rule out the presence of the recently described short isoform of CASZ1 (Gene bank accession number NM_017766.3) which comprises only the first 16 exons of the full-length CASZ1 (NM_001079843).35, 36 Owing to the poor specificity of the antibodies available for CASZ1 immunohistochemical analysis, loss of gene function could not be validated at the protein level.

The second tumor with evidence for a complete functional loss of the gene affected by viral integration is tumor T182e. In this case, viral E1 sequences were spliced to exon 5 of LIPC. RT-PCR was performed with primers spanning exons 4 and 5 and primers spanning the last two exons, exons 8 and 9. Transcripts comprising exons 8 and 9 could be amplified but not those comprising exons 4 and 5. No evidence for gene loss of the second allele was obtained by PCR. However, as only primers that span the viral integration site are specific for LIPC of the second allele, deletions of other parts of this gene are conceivable. We could, however, confirm the lack of LIPC expression in T182e by immunohistochemical analysis. Of particular note is that two further cervical carcinomas and four cell lines also lack LIPC transcripts comprising exons 4 and 5. Moreover, both carcinomas were also negative for LIPC expression by immunohistochemical staining. By contrast, LIPC was expressed in all of the normal cervical epithelia, metaplasia and CIN examined. LIPC was also disrupted in tumor T1907 but in this case LIPC expression of the second allele was not affected.

To our knowledge, this is the first investigation of primary cervical carcinomas in which genes were discovered that are not only disrupted by HPV integration but also showed altered expression of the second allele, resulting in loss of gene function. The frequency of such an event is difficult to determine because the one-step RT-PCR assay we used to prove functional loss has inherent technical limitations and may have underestimated the number of true cases. By extrapolating the results obtained in our study, about 1 out of 25 tumors would show a complete functional loss of the affected gene. This does not take into account those tumors in which integration leads only to a partial loss of gene function—a phenomenon which cannot be reliably determined. Complete loss of function is a characteristic property of classical tumor suppressors. In this context, the contribution of LIPC to tumor progression is not readily apparent. LIPC is a cytoplasmic protein mainly expressed in liver cells (Swiss-Prot, Source). It is involved in the high-density lipoprotein metabolism and catalyzes hydrolysis of phospholipids, mono-, di- and triglycerides and acyl-CoA thioesters, a biochemical process with no direct link to known oncogenic pathways. LIPC is disrupted by HPV16 in another tumor (T1907) that we have analyzed at RNA and DNA level but gene expression was maintained. Intriguingly, 4 out of 22 cervical carcinomas were negative for LIPC protein by immunohistochemical staining, whereas all normal cervical epithelia (n = 41), metaplasia (n = 18) and CIN (n = 36) were positive. By contrast, the observed loss of function of CASZ1 in tumor T2319 is in line with a very recent publication in which CASZ1 is reported to be a candidate tumor suppressor gene implicated in the development of neuroblastoma.37 CASZ1 localizes to Chr1p36.22 which is one of the commonly deleted regions in neuroblastoma. Moreover, several publications show the loss of heterozygosity in this particular chromosomal region in cervical carcinoma.38–40 It is a highly evolutionarily conserved gene which encodes a zinc-finger transcription factor which regulates the expression of genes crucial for cell growth and neural and muscle developmental processes as proposed in Drosophila and Xenopus models.37, 41–44 Putative target genes as defined by structural in silico analysis using databases and protein motif predicting software are c-myc, CREB, Pax-6, AP-1 and p53.35 Low CASZ1 expression in neuroblastoma correlated significantly with 1p loss of heterozygosity, MYCN amplification and decreased survival. In neuroblastoma with differentiated histopathology, CASZ1 was more highly expressed. Specific restoration of CASZ1 expression in neuroblastoma cell lines induced cell differentiation, enhanced cell adhesion and inhibited migration. Moreover, CASZ1 expression suppressed tumorigenicity, providing further functional evidence for its tumor suppressor properties.37 Based on the novel findings for CASZ1 in neuroblastoma, we envisage that the loss of CASZ1 function may indeed have contributed to the development of the squamous cell carcinoma T2319 (pT2aC4 pN1C4 M0C2, grade 2) analyzed in our study.

The data described in this article provide evidence that HPV-mediated loss of gene function does occur and may be more frequent in primary cancers than generally assumed. It is plausible that host gene disruption may be of biological relevance in individual cases but proof remains lacking. These results should encourage further research on the topic of HPV-mediated insertional mutagenesis.

Acknowledgements

This work was supported in part by funds of the Deutsche Forschungsgemeinschaft granted to Corina Driesch and Matthias Dürst.

Ancillary