|Home | About | Journals | Submit | Contact Us | Français|
HPV-related HNSCC generally have a better prognosis than HPV-negative HNSCC. However, a subgroup of HPV-positive tumors with poor prognosis has been recognized, particularly related to smoking, EGFR overexpression and chromosomal instability. Viral integration into the host genome might contribute to carcinogenesis, as is shown for cervical carcinomas. Therefore, all HPV16-positive HNSCC cell lines currently available have been carefully analysed for viral and host genome parameters.
The viral integration status, viral load, viral gene expression and the presence of aneusomies was evaluated in the cell lines UD-SCC-2, UM-SCC-047, UM-SCC-104, UPCI:SCC090, UPCI:SCC152, UPCI:SCC154 and 93VU147T. HPV integration was examined using FISH, APOT-PCR and DIPS-PCR. Viral load and the expression of the viral genes E2, E6 and E7 were determined via quantitative PCR.
All cell lines showed integration-specific staining patterns and signals indicating transcriptional activity using FISH. APOT- and DIPS-PCR identified integration-derived fusion products in six cell lines, and only episomal products for UM-SCC-104. Despite the observed differences in viral load and the number of viral integration sites, this did not relate to the identified viral oncogene expression. Furthermore, cell lines exhibited EGFR expression, and aneusomy (except UPCI:SCC154).
In conclusion, all HPV16-positive HNSCC cell lines showed integrated and/or episomal viral DNA that is transcriptionally active, although viral oncogene expression was independent of viral copy number and the number of viral integration sites. Because these cell lines also contain EGFR expression and aneusomy, which are parameters of poor prognosis, they should be considered suitable model systems for the development of new antiviral therapies.
Infection with high-risk human papillomavirus (HPV) is a causative factor for the development of several types of neoplasms, including cervical, anogenital and a subset of head and neck cancers. Although most head and neck squamous cell carcinomas (HNSCC) can be attributed to the exposure to environmental factors, such as excessive tobacco- and alcohol use, approximately 20-25% of these cancers are caused by infection with HPV16. Notably, HPV-induced tumors are preferentially found in the oropharynx, where up to 90% of the malignancies are associated with HPV 1,2,3.
Patients with HPV-induced carcinomas display clinical and molecular characteristics that are distinct from those in patients with tobacco- and alcohol-induced tumors. It is now well accepted that patients with HPV-derived tumors have a favourable prognosis, independent of the chosen treatment strategy 4-6.
As first described by Ang et al. 7, HNSCC can be divided into three risk groups, depending on HPV status, the number of pack-years (< 10 versus > 10) of tobacco smoking and nodal stage (N0-N2a versus N2b-N3). In addition, it was shown that EGFR overexpression negatively affects overall survival both in HPV-positive as well as in HPV-negative tumors 4. Chromosomal instability was also reported to have a negative influence on prognosis, especially in HPV-positive tumors 8. These data indicate that smoking, advanced nodal stage, EGFR overexpression and chromosomal instability are risk parameters for poor prognosis in HPV-associated HNSCC. From uterine cervical (pre)malignancy models it is hypothesized that integration of HPV into the host genome plays an important role in the carcinogenic process. It correlates with the severity and progression of these lesions and is considered a risk factor for the development of uterine cervical squamous cell carcinoma 9,10. In addition, a higher viral load is associated with higher grade lesions. It is, however, not possible to predict tumor progression based on the integration status of HPV or the viral load 11.
HNSCC, in particular oropharyngeal squamous cell carcinomas, are mostly discovered as metastatic disease and data on premalignant lesions are scarce. Therefore limited information is available on the role of viral integration in the development of these tumors. It is hypothesized, however, that viral integration also promotes carcinogenesis in HNSCC, but this is not substantiated by clinical studies on premalignant lesions, as is the case for cervical lesions.
Furthermore, the reported integration percentages in HNSCC vary considerably, ranging from 0% to almost 100% in different publications 1,12-15. This extreme variation might be explained by the different patient populations studied, the different methods applied to study viral integration and a lack of correlation between these methods.
However, the integration status of the virus can be of significance in relation to treatment. Chang et al. 16 and Herdman et al. 17 have shown a clear difference in the response of keratinocytes with episomal versus integrated HPV. Using these cell line models it could be shown that interferon therapy can eradicate episomal HPV infection, but leads to a growth advantage for cells containing integrated HPV. This indicates that antiviral therapies might be influenced by the viral integration status. Thus, it is imperative to have a well-characterized model for HPV-associated HNSCC, for which the physical status of the virus is known.
This study presents a detailed analysis of the integration status in 7 HPV-positive HNSCC cell lines, established from HPV-induced tumors. The integration status has been assessed using a set of independent techniques, i.e. DNA and RNA FISH, APOT- and DIPS-PCR. In addition, the expression levels of the viral genes E2, E6 and E7 and the expression levels of EGFR, as well as the viral load were analyzed.
Human HPV16-positive HNSCC cell lines were obtained from different sources. Cell lines UPCI:SCC090, UPCI:SCC152 and UPCI:SCC154 were kindly provided by Dr. Susanne Gollin from the University of Pittsburgh, Pittsburgh, USA. The UPCI:SCC152 cell line was established from a recurrence in the patient from which the cell line UPCI:SCC090 was derived 18. Cell line UD-SCC-2 was a kind gift from Dr. Thomas Hoffmann; present address University of Ulm, Ulm, Germany 19. The cell lines UM-SCC-047 and UM-SCC-104 were a kind gift from Dr. Thomas Carey, University of Michigan, Michigan, USA. The cell line 93VU147T was provided by Dr. J.P. de Winter, Free University Medical Centre, Amsterdam, the Netherlands 20. Clinicopathological data of the patients and tumors from which the cell lines were derived are given in table 1.
Cell lines UD-SCC-2 and 93VU147T were cultured in MEM (Invitrogen, Carlsbad, CA, USA) containing 10% FCS (Sigma, St Louis, MO, USA), 2 mM glutamine (Invitrogen), 0.1 mM MEM non-essential amino acids (Invitrogen) and 0.005% gentamycin (Eurovet Animal Health BV, Bladel, the Netherlands). The remaining cell lines were cultured in DMEM (Invitrogen) containing the same additives as described above. Cell lines UM-SCC-047, UPCI:SCC090 and UD-SCC-2 were received in 2008 and UM-SCC-104, UPCI:SCC152, UPCI:SCC154 and 93VU147T in 2011 and were frozen immediately upon arrival. All cell lines were confirmed to have unique genotypes, as tested using the ProfilerPlus assay, except the cell lines UPCI:SCC090 and UPCI:SCC152, which shared the same genotype as they are derived from the same patient (see supplementary table 1). Cell lines in culture were regularly tested for infection with mycoplasma.
The uterine cervical cancer cell lines SiHa and CaSki were used as HPV16- and p16INK4A–positive controls in p16INK4A immunostainings and HPV16-specific PCR analysis. The uterine cervical cancer cell line HeLa, the HPV-negative HNSCC cell lines UPCI:SCC003, and the osteosarcoma cell line U2OS containing the empty pJ4Ω-vector 21 were used as negative controls in these assays. The latter cell line harbouring HPV16-E2-containing pJ4Ω-vector was used as a positive control for HPV16-E2 RT-PCR and immunohistochemistry.
DNA was isolated from cultured cells using QIAamp DNA Blood Mini Kit (Qiagen, Hilden, Germany) as per the manufacturers' instruction. Total RNA was extracted using the RNeasy mini kit (Qiagen) according to the manufacturer's instructions, including DNase treatment. RNA concentration and quality were determined by RNA standard sense chips on a BioRad Experion system (BioRad, Munich, Germany).
The presence of HPV DNA was detected by PCR using the consensus primer set GP5+/6+ 22, followed by type-specific primers for HPV16 and HPV18. The following primers were used:
Primers were synthesized by Biolegio BV, Nijmegen, the Netherlands.
Five μl of PCR product were separated on a 1.5 % agarose gel and visualized using GelStar Nucleic Acid Gel Stain (Cambrex Bio Science Rockland Inc., Rockland, USA).
Viral load of HPV16 was determined using real-time fluorescence PCR with type-specific primers and probes as described earlier 23. Briefly, viral load was expressed as the number of HPV16 DNA copies per β-globin-gene copy. Gene copy numbers of β-globin were determined using the LightCycler-Control Kit DNA (Roche Molecular Biochemicals) according to the manufacturer's instructions as previously described 24. Calculation of initial copy numbers in samples was performed by the LightCycler 480 software (Version 1.5) using a standard curve generated with exactly quantified HPV DNA standards (ten-fold dilution series of full length HPV16 plasmid) that were amplified in the same PCR run 23,24. The analytical sensitivity of the assay was ten copies of HPV16 standard DNA. A negative control (water or DNA extracted from RTS3B cells that are negative for HPV) was included in each run and never yielded fluorescence signals above the background 31.
HPV16-specific probes were purchased from PanPath, Amsterdam, the Netherlands. BAC-clones, used for colocalization experiments, were grown according to the manufacturer's instructions (BACPAC Resources Centre, Childrens Hospital Oakland Research Institute, Oakland, USA). DNA was isolated using the Nucleobond BAC-100 kit (BioKé, Leiden, the Netherlands). Probes for centromeres (CEPs) 1, 3 and 9 were available in our lab, previously described in Hopman et al. 25. Probes and clones were labelled using either the Dig-nick translation kit or the Biotin-nick translation kit (Roche, Basel, Switzerland), according to the manufacturer's instructions. Labelled CEP probes for CEP17 and CEPX were kindly provided by the Department of Clinical Genetics, Maastricht University Medical Centre, Maastricht, the Netherlands. To exclude possible hybridization to RNA transcripts cells were treated with 10 ng/ml ribonuclease A (Qiagen) in a control experiment.
Fluorescence in situ hybridisation (FISH) was performed as described earlier 12,26. Briefly, cells were fixed in 70% ethanol, incubated with 0.01% pepsin (800-1200 U/mg protein from porcine stomach mucosa; Sigma) in 0.01N HCl, and postfixed in 1% formaldehyde in PBS. Probe and target DNA were denatured simultaneously for 3 minutes at 80°C prior to hybridization overnight at 37°C in a humid chamber. After hybridization the preparations were washed stringently. Biotin-labelled probe was detected using FITC-conjugated-avidine (Vector laboratories, Burlingame, CA, USA), biotin-conjugated goat-anti-avidine (Vector), and again FITC-conjugated-avidine. In cases where two hybridisation probes were used simultaneously, biotin-labelled probes was detected as described above. The digoxigenin-labelled probe was detected using mouse-anti-digoxigenin, TRITC-conjugated rabbit-anti-mouse and finally TRITC-conjugated swine-anti-rabbit (all from Dako, Glostrup, Denmark),. Slides were mounted in Vectashield (Vector) containing 4′,6-diamidino-2-phenylindole (DAPI; Sigma).
The HPV16 RNA specific FISH was performed on CytoRichRed fixed paraffin embedded slides of the different cell lines according to Smedts at al. 27, or on 70% ethanol suspensions additionally fixed in CytoRichRed. In contrast to DNA FISH, cells were rehydrated in an ethanol series using RNAse block (Biogenex) in the 50% ethanol and demi-water solutions, digested with proteinase K (DAKO PNA ISH detection kit) for 20 min at RT, and hybridized with denatured digoxigenin-labeled HPV16- DNA-probes. Cells were not denatured to allow hybridization to viral RNA and to prevent hybridization to viral DNA. After overnight hybridization the slides were processed according to the FISH protocol described above. As a negative control, slides were treated with RNAse.
Images were acquired using a Leica DMRXA microscope (Leica, Wetzlar, Germany) equipped with optical filters for DAPI, fluorescein and TRITC and a 63× Plan Apo (NA 1.32) objective. The microscope was connected to a digital black and white CCD camera (Metasystems Image Pro System, Sandhausen, Germany) for image recording. Staining patterns were scored as described earlier 28.
Metaphase spreads were made as described previously 29, with minor adjustments. Briefly, cells were cultured until approximately 50% confluency. Medium was then replaced by medium containing 0.1-0.5 μg/ml colcemid (Sigma) and 0-0.01 μg/ml ethidium bromide and incubated for 2-4 hours at 37°C. Optimal conditions varied per cell line. After incubation, cells were trypsynized and subsequently treated with 0.075 M KCl for 10 minutes at 37°C. Cells were washed twice and fixed in methanol/acetic acid (3:1, v/v). FISH probes were used for hybridisation as outlined above, with minor adjustments. Slides were only incubated for 8 minutes with 0.01% pepsin (800-1200 U/mg protein from porcine stomach mucosa; Sigma) in 0.01N HCl at 37°C. After washing, slides were dehydrated in an ascending ethanol series, and airdried. Slides were then baked for 10 minutes at 56°C. Postfixation was performed in 4% formaldehyde in PBS for 10 minutes at RT. After postfixation, the protocol was followed as described above.
HPV oncogene transcripts were amplified as described before 10,26. Briefly, reverse transcription was performed using 25 μM oligo-(dT)17 primer coupled to a linker sequence (dT)17-p3, 10 mM dNTPs each, 0.1 M DTT, 5× RT-buffer and SuperScript reverse transcriptase (Invitrogen, Karlsruhe, Germany). Quality of cDNA was determined by a standard GAPDH gene PCR. First-strand cDNAs containing viral oncogene sequences were subsequently amplified with semi-nested PCR using HPV-E7 specific 5′-primers and oligo(dT) and adaptor primers (3′). PCR products were separated on a 1.2% agarose gel. All bands were cut out, purified (QIAGEN Gel extraction kit, QIAGEN, Hilden, Germany) and sequenced (GATC Biotech, Konstanz, Germany). Sequence results were analyzed using the BLASTN program 30 and further mapped using map viewer, both provided by the National Center for Biotechnology Information (NCBI) 31.
Integrated papillomavirus sequences were detected using the Detection of Integrated Papillomavirus Sequences-PCR (DIPS-PCR) assay, as described earlier 26,32. Shortly, genomic DNA was digested using the Sau3AI restriction enzyme and an enzyme-specific adapter was ligated to the restriction-digested DNA using T4 DNA ligase (Roche Diagnostics, Mannheim, Germany). Linear PCR was performed using 5 HPV16 specific forward primers in independent setups. All independent PCRs were followed by individual exponential PCRs using further virus-specific forward primers and the AP1 reverse primer. PCR products were separated on a 1.2% agarose gel and products of interest were excised and subsequently sequenced (GATC Biotech). Sequence results were analyzed using the BLASTN program 30 and further mapped using map viewer, both provided by the NCBI 31.
RNA was reverse transcribed using the iScript cDNA Synthesis Kit (BioRad Laboratories, Hercules, CA, USA). Real-time PCR reactions were performed using SensiMix SYBR & Fluorescein (GC Biotech, Alphen a/d Rijn, the Netherlands). The detection of the housekeeping gene hypoxanthine phosphoribosyltransferase (HPRT) was used for normalization of mRNA levels. Primer sequences for HPRT were kindly provided by prof. dr. L. Hofland from Erasmus Medical Centre, Rotterdam, the Netherlands. The following primers, synthesized by Biolegio BV, were used:
3 μM CytoRichRed fixed paraffin embedded slides were used for immunohistochemistry. The slides were deparaffinised using xylol, rehydrated in a descending ethanol series and pretreated with citrate buffer pH 6.1 (DAKO, EnVision FLEX Target Retrieval Solution, Low pH) for 10 min at 97°C. Primary mouse monoclonal anti-HPV16-E2 antibody (Clone TVG 261; 1:100, Merck Millipore Corporation, Amsterdam, The Netherlands) was added to the slides and incubated for 1 hr at RT, followed by incubation with FLEX+ Mouse LINKER (DAKO) for 20 min at RT and EnVision FLEX/HRP (DAKO) for 30 min at RT. HRP was visualized using DAB+ Chromogen (DAKO) substrate. Between all the incubation steps the slides were washed with wash buffer (DAKO). Finally, slides were counterstained with hematoxylin dehydrated and mounted in Vectashield (Vector). Images were recorded in the brightfield mode using the microscope described above.
All cell lines have been tested for active HPV16 infection, using the algorithm suggested previously 4. AgarCyto blocks of the cell lines showed overexpression of CDK-inhibitor p16INK4A using immunocytochemistry, which is routinely used as a surrogate marker for HPV infection (see figure 1A). Furthermore, the presence of HPV16 DNA has been shown by PCR, first using the consensus primers GP5+/6+ and subsequently by using type-specific primers for HPV16- and -18 (see figure 1B). RT-PCR analysis of the viral oncogenes E6 and E7 showed viral gene expression in all cell lines (see below).
HPV16 load in all cell lines was estimated using PCR, and was expressed as the number of HPV16 copies per β-globin copy. Viral load varied from 1 (UM-SCC-104) up to 739 (UPCI:SCC090) HPV16 copies per β-globin gene copy (see table 2).
All cell lines were tested for the presence of HPV16 DNA using FISH on 70% ethanol suspensions and they all showed a reproducible punctate nuclear signal, which indicates viral integration. Each cell line showed a specific number of FISH signals per nucleus, ranging from 2 (UD-SCC-2 and UPCI:SCC154) to 7 (93VU147T) (figure 1C). UPCI:SCC090 and -152, derived from the same patient, showed identical spot numbers. UM-SCC-104, however, showed a variable number of signals per nucleus, ranging from 1 to 4 (table 2). Notably, cell lines with a high viral load exhibited a higher number and/or more intense FISH signals (see figure 1C and table 2) and also showed stronger HPV16-specific PCR bands on agarose gel (see figure 1B). Possible hybridization to RNA transcripts was excluded by treating the cells with RNAse in a control experiment.
APOT-PCR, a PCR-based method that can amplify both viral-human fusion transcripts as well as episome-derived viral transcript, was used to identify the site of viral integration in the cellular genome. A single integration site was identified in 5 cell lines (UM-SCC-047, UPCI:SCC090, UPCI:SCC152, UPCI:SCC154 and 93VU147T)(see table 3). Sequencing analysis of the obtained viral-human fusion transcripts revealed viral integration in 3 genes and in 2 intergenic regions. The protein-coding regions identified were DIAPH2, exon 21 (UD-SCC-2), TP63 exon 7 (3q28, UM-SCC-047) and NAP1 exon 4 (9q22.33, UPCISCC:090 and UPCI:SCC152; both derived from two subsequently diagnosed recurrent tumors from the same patient). The intergenic regions identified were located on chromosomes 17q12 (93VU147T) and 21q11.2 (UPCI:SCC154).
All fusion transcripts detected by APOT-PCR showed type A splicing as described by Klaes et al. 10 from viral nucleotide 880 directly into the human sequence in sense orientation with the identified gene. Interestingly, only episomal transcripts could be identified in the UM-SCC-104 cell line, suggesting the presence of complete episomes, either integrated or extrachromosomal. DIPS-PCR was performed to verify the identified integration sites and to search for possible viral-human fusion products in the cell line where no fusion transcript could be identified using APOT-PCR (UM-SCC-104). No fusion product could be amplified in this cell line, although the presence of episomal viral copies was confirmed. In the remaining cell lines the integration sequences identified using APOT-PCR were confirmed by DIPS-PCR. DIPS-PCR identified the integration site in the UPCI:SCC154 cell line to be located within the NRIP1 gene. Identified integration sites are summarized in table 3.
To confirm that the identified human sequences were localized adjacent to the viral genome, double-target FISH analysis was performed using both the HPV16 probe and BAC-probes complementary to the putative integration sequence (see figure 1D). BAC-probes were shown to colocalize with all HPV16 signals in two cell lines (UM-SCC-047 and UPCI:SCC154). In three additional cell lines, HPV16 signals were also detected without colocalization with the BAC-probe (UPCI:SCC090, UPCI:SCC152 and 93VU147T).
For the cell lines in which the integration loci did not colocalize to a BAC-probe (UM-SCC-104, UPCI:SCC090 and 93VU147T), HPV16 FISH analysis in combination with karyotyping based on inverted DAPI staining was performed on metaphase preparations. All data are summarized in table 2. In the UPCI:SCC090 cell line additional integration events were mapped to chromosomes 3 and 6, which is in agreement with published data 33,34. Interestingly, in the metaphase preparations of the UM-SCC-104 cell line, metaphase chromosomes did not show HPV16 FISH signals, whereas interphase cells, also present in the preparation, did show a nuclear FISH staining pattern similar to the staining on 70% ethanol suspensions (see figure 1C). This was shown in repeated experiments with different observers (JvLA, E-JS, NO). Although this might be a sensitivity issue, another explanation might be that the virus is extrachromosomal, which is supported by the detection of only episomal transcripts by APOT- and DIPS-PCR (see above).
To determine whether multiple colocalizing FISH-signals indicated duplication or translocation of the involved chromosome, double-target FISH was performed on metaphase preparations of the cell lines, using both a CEP-probe as well as an HPV probe. This showed that integration can be followed by deletion of the sister chromosome and duplication (UD-SCC-2) or multiplication (UM-SCC-047) of the chromosome containing HPV integration, with all respective chromosomes showing signals for both the CEP-probe as well as the HPV16 probe. Duplication of chromosomes with concurrent duplication of the sister chromosome was also seen (UPCI:SCC090, UPCI:SCC152), shown by the presence of chromosomes containing only a signal for the CEP-probe as well as chromosomes containing signals for both CEP- and HPV16 probes. Lastly, duplication could also be accompanied by translocation of a small portion of the involved chromosome to a different chromosome (UM-SCC-047).
The detected multiplication of chromosomes is in line with the ploidy status of the different cell lines, which in 6 of the 7 cell lines showed copy number variations for chromosomes 1 and 7, a hallmark for chromosomal instability and aneusomy 8 (table 2).
To determine whether integration affects the expression of the viral genes, the expression levels of the viral genes E2, E6 and E7 were determined using RT-PCR (see figure 2). Expression of the viral gene E2, which is known to regulate gene transcription of the oncogenes E6 and E7, was present at variable levels in six of the seven cell lines. To verify the presence of HPV16-E2 protein expression, we carried out immunohistochemistry on tissue sections of five CytoRichRed fixed and paraffin embedded cell lines, a positive control (U2OS cell line transfected with E2) and a negative control (U2OS cell line transfected with empty pJ4Ω-vector) (figure 1 F). U2OS E2 cells showed a strong diffuse nuclear immunostaining in all cells. U2OS vector cells and the UM-SCC104 cell line were negative, and the cell lines UPCI:SCC090, 93VU147T, UD-SCC-2, UM-SCC47 showed E2 positive foci in the nuclei at a frequency of 43%, 9%, 2% and 0.5%, respectively. mRNA expression of the viral oncogenes E6 and E7 could be detected in all cell lines at variable levels, but no correlation could be found with the levels of E2 mRNA expression.
No correlation was found between the number of FISH HPV16 signals, viral load and viral mRNA gene expression. To verify if all DNA FISH detected HPV16 integration sites are transcriptionally active, we performed RNA FISH (see supplementary table 1). These data showed that there is a variation of transcriptionally active sites within different cells of each cell line ranging from the maximum number of viral integration signals up to zero. To ensure that this variation was not the result of incomplete cells due to the cutting of tissue sections, we also performed RNA FISH on 70% ethanol suspensions leading to comparable results (data not shown).
In addition, the level of EGFR expression was examined for each of the seven cell lines. This showed a variable expression of 0.4 (UPCI:SCC090) up to 5.6-fold (UM-SCC-104) expression to HPRT. As the median EGFR expression in 61 primary HPV-associated HNSCC is 0.3 fold to HPRT (range 0.001-18.6-fold to HPRT, unpublished results), this indicates an increased expression of EGFR in all cell lines.
HR-HPV infection is a well-established risk factor contributing to the development of HNSCC, particularly OSCC. HPV-positive tumors have different clinical and patho-biological characteristics, including a better prognosis as compared to HPV-negative tumors 4. It is still under debate which factors, in addition to the presence of HPV DNA and the resulting viral gene expression in the host cell, are contributing to tumor development and progression. For example, integration of the viral DNA into the host genome, and the resulting disruption of invaded genes might play an important role in the process of HPV carcinogenesis, as seen in HPV-associated cervical cancer. In the case of HNSCC, few studies have addressed this highly relevant issue 26,34-37. Since in vitro testing is still an important step in the development of new therapeutic options, it is imperative to be able to study well-established cell line models for HPV-positive HNSCC. In the underlying study, HPV16-associated parameters, such as viral load and integration, have been assessed in seven established HNSCC cell lines, shown to be HPV-induced by p16INK4A immunostaining and HPV-type-specific PCR.
Integration of HPV DNA was detected in 6 of the 7 cell lines using three independent methods, i.e. APOT- and DIPS-PCR and FISH, which all gave concurrent results with respect to the integration locus. Furthermore, for three cell lines (UD-SCC-2, UM-SCC-47 and UPCI:SCC090) our results confirmed previously reported integration sites 33,34. Two cell lines that were derived from the same patient (UPCI:SCC090 and UPCI:SCC152), showed identical integration sites despite the fact that they were established from biopsies taken from different locations and at different time points, indicating clonal expansion of the tumor. One cell line (UM-SCC-104) did not show integration, but rather the presence of extrachromosomal HPV DNA, as assessed by APOT- and DIPS-PCR. This is in contrast to a recent study by Akagi et al., suggesting viral integration in this cell line 34. However, in our opinion their presented FISH and Northern blotting data on UM-SCC-104 did not provide sufficient evidence for viral integration. We were unable to detect the integrated virus DNA on metaphase chromosomes of the cell line, and it is very unlikely that we missed this integration site by our DIPS- and APOT-PCR assay in the here presented study. Maybe the detected deletion in HPV16-E2 by Akagi et al. using whole genome sequencing is present in episomal DNA, or alternatively viral integration or loss of the integration site in UM-SCC-104 could be a result of culturing 34.
We detected viral loads varying from 1 up to 739 HPV16 copies per β-globin gene copy. In contrast to results obtained from cervical malignancies 38, viral load did not seem to be correlated to integration status. This can be concluded from the two cell lines showing a viral load of 1 HPV DNA copy per β-globin copy, one of which showed HPV DNA integration and the other showed only episomal HPV.
Cell lines with a higher viral load did exhibit strong punctate FISH signals, indicating that the HPV genome may be present in integrated tandem repeats. This assumption is supported by the finding that these strong signals are not derived from RNA transcripts present at these loci, since RNAse treatment before the DNA FISH analysis did not lead to a substantial decrease in signal intensity (data not shown). Moreover, the cell lines showed 2 up to 7 HPV integration sites by FISH, whereas in primary tumors generally only 1 signal per nucleus is detected 8. To determine whether this is due to either multiplication of the integration locus, or a result of multiple independent integration events, co-hybridization of HPV16 probes with a BAC-probe complementary to the identified integration sequences was performed. In the cell lines UD-SCC-2, UM-SCC-047 and UPCI:SCC154 all HPV16 signals were shown to colocalize with the BAC-probe, while in UPCI:SCC090, UPCI:SCC152 and 93VU147T the BAC-probe colocalized with part of the HPV16 FISH signals. The presence of more than one colocalizing signal indicates multiplication of the integration locus. Multiplication of an existing integration site was furthermore shown by co-hybridization of the HPV-probe with CEP-probes for the corresponding chromosome. For the UD-SCC-2 cell line, derived from a male patient, the two HPV16 FISH signals were located on two different CEPX-containing chromosomes, indicating that a duplication event has occurred. In the UPCI:SCC090 four chromosome copies containing CEP9 were detected, of which two harboured HPV16 signals, indicating duplication of both copies of chromosome 9. In the UM-SCC-047 cell line five HPV16 signals were detected on five separate chromosomes, of which four also contained CEP3, indicating multiplication of the initial integration site. The fifth chromosome contained very strong HPV- and BAC-signals but no CEP3 signal (see figure 1E), which might indicate that this chromosome evolved by a different mechanism, for example involving both translocation of part of chromosome 3 and amplification of the integration site. Figure 3 provides a schematic representation of the multiplication and translocation events identified in the HNSCC cell lines indicated. This finding further supports the results of Akagi et al. that show that HPV integrants may promote genomic instability 34
Besides HPV16 signals colocalizing with the BAC-probe in UPCI:SCC090, UPCI:SCC152 and 93VU147T cell lines, additional HPV16 signals were detected. The non-colocalizing HPV signals might indicate multiple independent integration events, which were not detected by APOT- or DIPS-PCR. A possible explanation for the fact that the APOT-PCR assay is unable to detect all the different integration sequences may be that (several of) these sequences are not actively transcribed and therefore are not clinically relevant. On the other hand, the inability of the DIPS-PCR assay to detect these integrated sequences may come from an alternative integration mechanism that is not detected by our PCR, which is based on E1 and E2 HPV sequences. Using DIPS-PCR with additional primers covering the entire HPV genome, Li et al. 39 described a mechanism for HPV16 integration in exfoliated cervical cells based on disruption of the L1 or L2 gene. These authors indicated that integration at these viral sites would normally not result in progression to (pre)malignant lesions. It is therefore likely that such integration events have occurred in our cell lines, before or after the clinically relevant integration event took place.
From studies in cervical cancer it is known that viral load does not necessarily correlate to expression levels of E6 and E7 40. Similarly, in the HNSCC cell lines, as well as in primary HNSCC, the variation in viral load was not reflected in the expression of the viral genes E2, E6 and E7. Furthermore, the current concept of cervical carcinogenesis suggests that HPV integration leads to disruption of the viral genome in the E1 or E2 open reading frame (ORF), followed by loss of the E2 inhibitory gene function on E6 and E7 expression. Subsequently expression of both oncogenes is thought to be enhanced 41-43. However, in the different HNSCC cell lines, such a correlation could not be found. E2 mRNA and protein levels were highly variable in UD-SCC-2, UM-SCC-47, UPCI:SCC090, UPCI:SCC152 and 93VU147T, even though integration was identified. A recent publication suggests that HPV integration without concurrent loss of the E2 gene can also occur in HNSCC resulting in viral concatenates in the human genome 35. The high viral load in combination with a limited number of strong FISH signals indicates that integration of such stretches of multiple HPV-copies could have occurred in our cell lines as well. E6 and E7 mRNA levels were equally variable in all cell lines, and no correlation was seen with the expression levels of E2. This is in agreement with the results obtained in primary HNSCC, in which the expression levels of E2, E6 and E7 mRNA transcripts were highly variable and did not show correlation to the physical status of the virus 8,26.
In conclusion, we have characterized seven established HPV16-positive HNSCC cell lines, which showed integrated and/or episomal viral DNA that is transcriptionally active. However, viral E2, E6 and E7 gene expression proved to be independent of viral load and the number of viral integration sites, suggesting that integration does not affect viral gene expression. The analyzed cell lines, furthermore, exhibit characteristics of primary HPV-positive OSCC with a poor prognosis, such as EGFR overexpression and aneusomy 8,44,45, although it cannot be ruled out completely that molecular changes may have accumulated due to cell culturing. We believe that these cell lines represent suitable model systems of HPV-positive HNSCC that respond poorly to current treatment modalities, and that they may be beneficial in future studies for the analysis of new treatment options.
Human papillomavirus (HPV) infection and genomic integration is an important risk factor for HNSCC development. In this study using seven HPV16-positive HNSCC cell lines available to date, we could show viral integration sites that are transcriptionally active, and both variable viral load as well as expression of viral early genes E2, E6 and E7. These cell lines reflect tumors with a poor prognosis and provide good model systems for studies testing new treatment options.
Potential conflicts of interest: none