Human Papillomavirus Type 16 E6 and E7 Cause Polyploidy in Human Keratinocytes and Up-Regulation of G₂-M-Phase Proteins

INTRODUCTION

Human papillomaviruses (HPVs) are small DNA virus, which are implicated in intraepithelial neoplasia of the lower genital tract, in particular, premalignant and malignant disease of the cervix as well as some oral and laryngeal cancers (1, 2, 3) . The genital HPV types are subdivided into a low-risk subgroup that are predominantly found in benign lesions (4) and a high-risk subgroup, including the most common types HPV-16 and HPV-18, that are associated with >90% of all premalignant and malignant disease of the cervix (4) . HPV-16 and HPV-18 have been shown to immortalize and inhibit differentiation of human keratinocytes in vitro (5, 6, 7) , and two HPV-encoded oncogenes, E6 and E7, play a major role in stimulating G₁-S-phase progression in stratified epithelial cells that have been programmed to terminally differentiate. E7 from high-risk HPV types binds to and abrogates the repressive function of the retinoblastoma protein family (Rb), which are important regulators of cell proliferation (8 , 9) . pRbp105, the prototypical member, and the other family members, p107 and p130, are involved in various stages of the cell cycle and in a complex with the transcription factor family, E2F, repress transcription of genes required for G₁-S-phase entry (10) . E6 has been shown to bind and degrade p53 protein by forming a tripartite complex with the E3 ligase, E6AP (11, 12, 13) . p53 is essential for cell arrest at checkpoints in G₁-S when tetraploidy occurs (14 , 15) , as well as G₂ phase under DNA damaging conditions (16 , 17) . Thus, both pRb and p53 are tumor suppressors, especially important in cell cycle progression checkpoints and their functions, can be abrogated by E6 and E7 from the high-risk HPV.

As well as deregulating the functions of p53 and the Rb family, E6 and E7 proteins also have affects on other cellular proteins. For instance, E6 up-regulates the telomerase enzyme through activation of the transcription factor c-myc (18) , and E7 can associate with a histone deacetylase activity through binding to Mi2β (19) . Through deregulation of multiple cellular proteins involved in cell cycle progression, E6 and E7 can disrupt various checkpoints necessary for faithful DNA replication and chromosome segregation during cell division. Failure to activate the checkpoints that govern accurate cell division can result in abnormal segregation of chromosomes, leading to genomic instability. Chromosomal instability, a hallmark of a majority of cancers, is increasingly thought to be the cause rather than consequence of malignancy (20) . E6 and E7 proteins from oncogenic HPV types have been shown to induce chromosomal instability during infection, leading to polyploidy and aneuploidy (21 , 22) . Previous work by Thomas and Laimins (22) has demonstrated that HPV-16 E6 and E7 can independently bypass mitotic spindle checkpoint, resulting in accumulation of polyploid cells. Recent work has shown that E6 and E7 expression causes centrosome amplifications and lagging chromosomes during mitosis, leading to genomic instability (23 , 24) . It has been suggested that the centrosomal abnormalities seen in the presence of E6 and E7 are most likely caused by disruption of checkpoint control machinery; however, the mechanism involved has not been elucidated.

To try and understand which functions or domains of E6 and E7 are required for production of the polyploidy state, we used a panel of E6 and E7 mutants with known functions. Here, we report genetic evidence that suggests that the ability of E7 to disrupt the functions of Rb family members, in cooperation with p53 degradation by E6, are both required to cause chromosomal instability. Also, we provide preliminary evidence that deregulation of functions p53 and Rb family members by E6 and E7, respectively, correlates with up-regulation of polo-like kinase (Plk1), a mitotic kinase that is involved in entry and exit of mitosis (25 , 26) .

MATERIALS AND METHODS

Tissue Culture and Cell Lines.

Primary human keratinocytes derived from neonatal foreskins [human foreskin keratinocytes (HFKs)] were maintained in Epi-Life media supplemented with growth factors (Cascade) at 37°C and 5% CO₂. φNX retroviral packaging cells (American Type Culture Collection) were cultured in DMEM (Life Technologies, Inc.) supplemented with 10% fetal clone 1 (Hyclone), 1% penicillin-streptomycin (Life Technologies, Inc.), and 1% sodium pyruvate (Life Technologies, Inc.).

Retroviral Constructs and Infections.

The pBabe Puro (pBabe) retroviral system was used to generate HPV-16 E6 and E7 wild-type and mutant cell lines (27) . The pBabe backbone was a gift from Hartmut Land. Mutations in E6 and E7 previously characterized are listed as follows: E6/E7.C24G mutation in the LXCXE Rb-binding domain of E7 (28) ; E6/E7.L67R, E6/E7.LL82/83RR, and E6/E7.R77G all in the COOH-terminal zinc finger region of E7 (19 , 28 , 29) ; and E7/E6.F125L, E7/E6.G134V, and E7/E6.Delta123-127 mutations in the second zinc finger of E6 (30 , 31) . All E6/E7 constructs were cloned into pBabe using BamHI and EcoRI.

φNX cells were transfected by calcium phosphate method following manufacturers protocol (BD Biosciences) with 15 μg of the vesicular stomatitis virus envelope gene and 15 μg of pBabe retroviral construct mentioned above. Retrovirus was concentrated 48 h after transfection by centrifugation at 15,000 rpm, 4°C, for 90 min. Pelleted viruses were resuspended in 1 ml of Epi-Life media and incubated at 4°C overnight. HFKs were infected the following day with 1 ml of virus in 8 μg/ml Polybrene. Media were changed after 6 h of infection, and cells were allowed to recover for 24 h. Cells were selected with 1.25 μg/ml puromycin for 3–4 days. As indicated in the text, some cells were treated with 100 ng/ml nocodazole (Sigma) or DMSO for 20 h.

Western Blot Analysis.

HFKs were harvested, pelleted, and stored at −80°C until lysis. Cell pellets were lysed in radioimmunoprecipitation assay buffer [1× PBS, 1% NP40, 0.5% sodium deoxycholate, 0.1% SDS, 1:100 protease inhibitor mixture (Sigma)]. Lysates were centrifuged, and protein was quantitated by Bradford Assay. Twenty-five μg of protein was boiled in 2× sample buffer, and samples were run on 10% nondenaturing SDS-polyacrylamide gel. Protein was transferred to 0.2 μm of nitrocellulose membrane and stained with Ponceau S (Sigma) to verify equal loading of lanes. Membranes were blocked overnight at 4°C in 1× PBS containing 5% milk/0.1% Tween 20 (PBST). Plk1 mouse monoclonal antibody (0.5 μg/μl; Zymed) was used at 1:2000 dilution in blocking buffer. Nek2 mouse monoclonal was used at 1:1000 dilution (BD Biosciences), Aurora-A polyclonal antibody and Cdk1 monoclonal antibody (Cell Signaling) were used at 1:1000 dilution. Antimouse IgG-horseradish peroxidase secondary antibody (0.2 mg/ml; Santa Cruz Biotechnology) was used at 1:2000 dilution in milk/PBST. Blots were visualized by using enhanced chemiluminescence reagent (Pierce).

Flow Cytometry.

For analysis of the DNA content by flow cytometry, cells were harvested, fixed in 70% ethanol, treated with Rnase A (1 mg/ml), and stained with propidium iodide (20 μg/ml). Cells were examined by fluorescence-activated cell sorting (FACS) and data analyzed using FACSCaliber (ELITE) multicycle software. For analysis of DNA synthesis, media were supplemented with 10 μm bromodeoxyuridine (BrdUrd) for 45 min, cells were harvested, resuspended in 70% ice-cold ethanol, and stored at 4°C for at least 12 h before analysis. The day of analysis, cells were incubated in 0.2 mg/ml pepsin (Sigma)/2 n HCl in 1× PBS at 37°C for 20 min to remove the cytoplasmic membrane, followed by two subsequent washes in 1× PBS/0.5% fetal clone 1 (HyClone)/0.5% Tween 20 and a 20-min room temperature incubation in 1× PBS/2% fetal clone 1. The resulting nuclei were treated with 25 μg/ml anti-BrdUrd conjugated to fluorescein (Boehringer Mannheim) for 45 min, washed in 1× PBS/0.5% fetal clone 1/0.5% Tween 20, and resuspended in 1 ml of 1 mg/ml RNase A (Sigma) for 30 min. Cells were pelleted and resuspended in 1 ml of 20 μg/ml propidium iodide (Sigma) at least 10 min before analysis. Ten to 30,000 events were collected on the FACSCalibur (ELITE) flow cytometer, and data were analyzed by Multicycle software.

Immunofluorescence Staining.

Cells were grown to 30–40% confluency on glass coverslips in defined Epi-Life media supplemented with growth factors (Cascade). Cells were washed and fixed with ice-cold 4% paraformaldehyde at room temperature for 20 min. The fixed cells were washed and permeabilized with 0.5% Triton X-100 in PBST (PBS plus 0.2% Tween 20) at room temperature for 20 min, followed by incubation in blocking solution (5% FCS in PBST) at 37°C for 15 min. Polyclonal antibody to nuclear protein, ataxia telangiectsia locus (NPAT; a gift from Jiyong Zhao, University of Rochester) diluted 1:1000 in blocking solution was added, and the cells were incubated at 37°C for 1 h. Cells were washed three times for 5 min each with PBST and incubated with Alexa Fluor 488 goat antirabbit IgG (Molecular Probes, Inc.) in blocking solution at 37°C for 1 h. Cells were washed with PBST three times, and imaging was performed using inverted-fluorescence microscopy (Olympus CK40), and images were captured with a Quality Imaging camera and software. Exposure times were kept constant.

Microarray Analysis.

pBabeE6/E7 infected and pBabe vector alone control keratinocytes were suspended in 1.6% methylcellulose (Sigma) in DMEM plus 10% FCS and incubated for 14 h. The cells were washed, pelleted, and total RNA isolated using RNeasy mini kit (Qiagen). From each sample, 10 μg of total RNA were used to generate high-fidelity cDNA by the University of Rochester Microarray Core facility for hybridization to Affymetrix U133A chips, which contained, at the time, 22,000 potential genes. The data were statistically analyzed using significance analysis of microarrays, and two-class unpaired analysis was used to test for significance (32) . Microarrays were carried out on three independent sets of keratinocytes.

Chromosome Spreads.

The two cell lines of E6/E7 and control pBabe were grown on 22 × 22-mm coverslips in 35 × 10-mm Petri dishes with 3 ml of defined Epi-Life media supplemented with growth factors (Cascade). To harvest cells for chromosome analysis, the cells at 60–70% confluency were arrested at metaphase with Colcemid (0.05 μg/ml; Life Technologies, Inc.) for 1 h, hypotonically treated by an initial incubation at 37 C for 10 min with the addition of 2 ml of the hypotonic solution [0.4%(w/v) KCl:0.4%(w/v) sodium citrate (1:1, v/v)], followed by a 20-min exposure to the hypotonic solution at room temperature. The cells were then fixed in methanol:acetic (3:1, v/v) and air dried. The chromosomes were G-banded after trypsin treatment and stained with Giemsa stain.

RESULTS

HPV-16 E6 and E7 Cause Polypoidy in Primary Human Keratinocytes.

Altered expression of mitotic genes has been implicated in deregulation of mitosis and cytokinesis, leading to polyploid and aneuploid cells. Because HPV-16 E6 and E7 genes are known to cause polyploidy and aneuploidy in HFKs, we initially decided to analyze cycling and M-phase-arrested cells for DNA content by FACS analysis. It should be noted that the cells used in these experiments are from pooled populations of neonate foreskin keratinocytes, and each batch of cells varies in the population of cells that have >4N DNA content (2–13% in our studies); therefore, it is important to use the same pools of cells for control and test samples. Keratinocytes were infected with a retrovirus expressing E6 and E7 in tandem, selected for 3–4 days in puromycin, and then used in the following experiments. A microtubulin polymerization inhibitor, nocodazole, was used to study the effects of an arrest in mitosis. Cycling and nocodazole arrested E6/E7, and pBabe control cells were stained with propidium iodide and analyzed by FACS for DNA content. As expected, cycling E6/E7 cells had a high percentage of cells with >4N DNA content [15% compared with 2% in pBabe vector control cells (Fig. 1A) ⇓] . Interestingly, cell cycle arrest by nocodazole treatment results in an even larger population of polyploidy in E6/E7 cells (28% in E6/E7 compared with 3.2% in pBabe). One explanation for this observed phenotype would be that E6/E7 cells bypass the nocodazole arrest and continue to replicate. To determine that the E6/E7-expressing cells were capable of reentering the cell cycle and synthesizing DNA, E6/E7-expressing and pBabe control nocodazole-treated cells were labeled for 1 h with BrdUrd, fixed and stained. FACS analysis of these cells revealed that indeed, a proportion of the E6/E7 cells bypass the spindle checkpoint arrest initiated by nocodazole and maintain DNA synthesis (Fig. 1B) ⇓ . Under the spindle checkpoint arrest signal, a high percentage of E6/E7 cells continue to go through another round of DNA synthesis compared with pBabe vector control cells resulting in a >4N population.

E6 and E7 Are Both Required for Polyploid Phenotype.

Next, we wanted to delineate which of the two viral proteins, E6 or E7, were required for the aneuploid phenotype. HFKs were infected with a retrovirus containing either pBabeE6 or pBabeE7 as well as vector control. Cells were grown to 30% confluency and treated with nocodazole for 18 h. FACS analysis for DNA content was carried out on propidium iodide-stained cells (Fig. 2) ⇓ . Surprisingly, neither E6 nor E7 seemed to have elevated levels of >4N cells, suggesting these cells were arrested and could not bypass the mitotic spindle checkpoint. This result was unexpected because previous work has reported that expression of E6 and E7 can independently bypass a mitotic checkpoint, resulting in polyploid cells (22) . However, these experiments were carried out with HFKs grown in the presence of fibroblast feeder cells in serum containing media, whereas our experiments were carried out in defined nonserum-containing medium, without feeders. Subsequently, to check if culture conditions were important, we grew keratinocytes in serum-containing medium and feeders, as previously described (21) and found that E6 and E7 could independently cause polyploidy upon G₂-M arrest with nocodazole (data not shown). It is likely that in the absence of secondary signals mediated through feeder cells and serum, E6 and E7 independently, cannot bypass spindle checkpoint.

The Functions of E6 and E7 Required for Polyploidy.

To additionally establish the requirement for both genes and to determine which functions of E6 and E7 are required for the polyploid phenotype, we used E6 mutations in the context of wild-type E7 and E7 mutants in the context of wild-type E6. HFKs were infected with retroviruses carrying the individual mutants plus E7 and selected with puromycin. Presence of E6 mRNA and E7 protein in these cells was confirmed by reverse transcription-PCR and Western blot analysis, respectively (data not shown). Although the E6 mutant E6.F125L has been shown to target p53 for degradation, mutant E6.Δ123-127 does not degrade p53 (30 , 33) .

We carried out FACS analysis of these mutants to determine whether p53 is required for maintenance of chromosomal stability. The >4N cell population in E6.Δ123-127-expressing HFKs is the same as control keratinocytes in cycling cells and slightly higher in G₂-M arrested cells (Fig. 3) ⇓ . However, the E6 mutant, F125L, still caused polyploidy, suggesting a role for p53 in maintenance of chromosomal stability. We also tested E6.G134V, a mutant of E6 which has been shown to degrade p53 but unable to activate telomerase activity (18) . FACS analysis showed that this mutant, E6.G134V, was able to bypass M-phase arrest and cause polyploidy, suggesting that the ability to activate telomerase is not required for the bypass of this checkpoint.

We also tested E7 genes that were mutated in the LXCXE motif, which is the binding site for the Rb family and E7 mutations in the zinc finger region contained in the COOH terminus. The E7 mutant C24G (in the LXCXE motif) has been described to not bind or derepress pRbp105, although it has been reported to bind to p107 and reduce the level by 60% (34) . However, other studies show no binding to p107 (35 , 36) . One consistent finding is that this mutant is unable to bypass arrest signals compared with wild-type E7. The binding of this mutant to p130 has not been comprehensively tested. Another mutant E7.R77G (in the zinc finger domain) binds and derepresses pRbp105 such as wild-type E7 (29) . We also included mutants E7.L67R and E7.LL82.83RR, which have been shown to bind pRbp105 but do not derepress its function (19) . These E7 mutants were expressed in the context of wild-type E6. The polyploidy levels seen in E7.C24G mutant are the same as pBabe control cells (Fig. 4) ⇓ . However, zinc finger mutants E7.R77G as well as E7.L67R and E7.LL82.83RR can cause polyploidy, suggesting that derepression of pRbp105 is not necessary to cause chromosomal instability (Fig. 4) ⇓ . Although mutants E7.L67R and E7.LL82.83RR have been shown not to disrupt the repressive function of pRbp105, little is known about their effect on the functions of the other Rb family members, p107 and p130. It is possible that these E7 mutants bind and deregulate the normal function of these proteins. A summary of some of the biological functions of E6 and E7 mutants and their ability to cause polyploidy is shown in Table 1 ⇓ .

Polyploidy Is Not Caused by Endoreduplication or Multinucleation.

Polyploidy can result from a number of disruptions in mitosis and cytokinesis, resulting in different cellular phenotypes. For instance, (a) failure of mitosis results in one larger nuclei, (b) a failure to undergo cytokinesis results in two normal sized nuclei being present, and (c) multiple small nuclei are caused by a failure somewhere in the process of karyokinesis. We have evidence from our BrdUrd FACS analysis (Fig. 1B) ⇓ that the polyploidy is not due to multinucleation because the BrdUrd staining protocol requires the isolation of nuclei. Therefore, the single nuclei isolated exhibit >4N chromosome complement. To substantiate this observation, we stained the cells with an antibody against NPAT, a transcription factor that is known to bind to the histone loci on chromosomes 1 and 6. Normally NPAT appears as two spots in nuclei of cells in G₀-G₁ phase and as four spots in nuclei of cells going through S phase, and it disappears during M phase (37) . This makes it a good marker for cells that have >2N complement of chromosomes (37) . In E6/E7-expressing cells, 30% contain four or more spots/nuclei, indicating a polyploidy state, whereas most control cells contain mostly two or on occasion four spots (Fig. 5A) ⇓ . In addition, we did not observe a significant number of multinucleated cells (<2%).

Because cytokinesis appears to occur normally, we attempted to determine whether polyploidy was caused by a failure in mitosis, which could result in endoreduplicated chromosomes, due to a failure in disjunction of centromeres. This would cause the presence of four chromatids instead of the usual two. We carried out chromosome spreads on the keratinocytes expressing E6 and E7 (Fig. 5B) ⇓ . From 50 chromosome spreads, none showed evidence of four chromatids, and therefore, we assumed that the chromosomes have not endoreduplicated. This preliminary evidence would suggest that polyploidy is caused by a failure at some stage in mitosis, after disjunction, rather than during cytokinesis (38) .

Microarray Analysis.

In a microarray analysis carried out in the laboratory to identify genes differentially regulated in differentiating E6/E7 cells compared with normal HFKs, a number of genes involved in the G₂-M transition of the cell cycle were identified as being up-regulated. In view of our observations with the polyploid phenotype, we reasoned that the E6/E7 differential expression of some of these genes may have a role in the underlying mechanism causing polyploidy in E6/E7 cells. Table 2 ⇓ shows the fold activation or repression of selected genes in E6/E7-expressing human keratinocytes from three independent experiments, using different pools of human keratinocytes for each experiment.

Because we were mainly interested in genes known to contribute to chromosomal instability when deregulated in their expression, we selected Plk1, Aurora-A, Nek2, and cdk1 for verification of the microarray data. These genes have previously been implicated in abnormal chromosomal segregation when overexpressed (39 , 40) . Additional analysis showed that these genes were regulated, at the protein level, in a similar fashion to that observed in the microarray analysis (Fig. 6) ⇓ . Because all these genes are involved in the G₂-M transition and M phase of the cell cycle, we wanted to determine whether the expression of these genes is altered in cells that were arrested at that phase. In control cells, expression of the selected genes, namely Plk1, cdk1, and Aurora-A, were up-regulated, whereas Nek2 was reduced during mitotic arrest (Fig. 6) ⇓ . However, in E6/E7 cells, these genes are significantly increased in cycling as well as M-phase-arrested cells compared with the controls. These results confirm the microarray data in that E6/E7 up-regulates the G₂-M genes and, in addition, shows that the levels are maintained during a M-phase arrest in E6/E7-expressing HFKs.

Up-Regulation of Plk1 Is Dependent on p53 Degradation and Repression of pRb Functions.

Plk1 is a serine/threonine kinase that was up-regulated at the transcriptional and protein level in E6/E7-expressing cells as shown above. Plk1 regulates several events during the G₂-M transition and M phase of the cell cycle (25 , 41) and has been shown to contribute to activation of cyclinB/cdk1 as well as centrosome maturation (41 , 42) . Also, Plk1 regulates the anaphase-promoting complex required for successful mitotic exit as well as in events leading to cytokinesis (43) . Considering that this kinase activity is pivotal in the G₂-M transition and during M-phase stage of the cell cycle, we decided to determine relative Plk1 levels in the various E6 and E7 mutants. A portion of nocodazole-treated cells that were used to carry out the FACS analysis in previous experiments were used for protein extraction and Western blot analysis. Plk1 expression correlates with the polyploid phenotype observed in E6/E7 cells and in the studies with the E6 and E7 mutants (Fig. 7) ⇓ . In E6 mutants that do not degrade p53 (E6.Δ123-127), Plk1 is at the same level as control cells, whereas mutants of E6 that do degrade p53 show up-regulation of Plk1 just as wild-type E6/E7 cells (E6.F125L and E6.G134V, Fig. 7 ⇓ ). Up-regulation of Plk1 is observed in E7 mutants that bind and repress Rb family members functions (E7.R77G), whereas a mutant that fails to bind and repress the Rb family (E7.C24G) shows no increase in Plk1 protein, both in cycling and M-phase-arrested cells. However, mutants E7.LL82.83RR and E7.L67R, which bind to but do not derepress pRbp105, have the same level of Plk1 as wild-type E6/E7.

DISCUSSION

HPV-16 E6 and E7 oncoproteins are known to induce chromosomal abnormalities during infection as abnormal mitotic figures are observed in carcinoma in situ, a stage just before invasive cancer (44 , 45) . In addition, these viral genes have been shown in tissue culture to cause centrosome amplification, chromosome lagging, polyploidy, and aneuploidy (22 , 24 , 46) . However, it is not clear from the previous studies if the cells that exhibit chromosome lagging or centrosome amplifications are viable and will go on to new rounds of DNA synthesis. Also, apart from the possible involvement of pRb, no mechanism has been elucidated and no mitotic targets have been determined.

In an attempt to answer some of the questions raised above, we carried out FACS analysis on keratinocytes expressing HPV-16 E6 and E7 and compared these to matching vector controls, both in cycling and M-phase-arrested cells. We observed a high percentage of E6/E7-expressing keratinocytes with a population of >4N compared with the controls. Interestingly, upon a mitotic arrest signal, this population of >4N cells was additionally elevated. Although we observed polyploidy (even numbers of chromosomes, 4N and 8N) in the majority of cells, it is possible that a small percentage exhibit aneuploidy because late passage cells containing the whole HPV-16 have been shown to be aneuploid (21) . We showed that polyploidy required both E6 and E7 functions, which is at variance with an earlier report, suggesting that E6 and E7 could independently produce >4N cells. However, we demonstrated that the difference observed was due to cell culture conditions because the other investigators used serum containing media along with feeder cells (22) . It has been shown that when E6- and E7-expressing HFKs are grown in serum plus feeder cells, the viral proteins can cooperate with factors (possibly growth factors) in the serum or secreted by feeder cells to exhibit phenotypes not observed when the same cells are cultured in defined media (22 , 47) . These growth conditions may induce mitogenic signals that might alter the normal functions of checkpoint pathways in cooperation with either E6 or E7. Cells cultured in defined synthetic media presumably lack these cellular or serum factors and therefore need both E6 and E7 to bypass cellular checkpoints.

To understand the possible mechanisms by which E6 and E7 caused polyploidy, we used well-defined mutants of each of the proteins. The mutational analysis used to define the functions of E6 and E7 required to cause polyploidy showed that the E7 mutant E7.C24G that cannot bind nor abrogate the functions of any of the Rb family members (8 , 9 , 35) is the only E7 mutant that is no longer able to cause polyploidy. Interestingly, the zinc finger E7 mutants, E7.L67R and E7.LL82.83RR, which can bind to pRbp105 but do not abrogate its repressive function (19) , can still cause polyploidy. We speculate that one of the other Rb family members, namely p107 or p130, is likely involved in causing the polyploidy phenotype. None of these zinc finger mutants of E7 have been adequately tested for their ability to abrogate p107 or p130 functions, which can be different from those of pRbp105. For instance, it has been shown that p130 and p107 are required for p16ink4a mediated G₁ arrest (48) and that p130 and p107, in a complex with E2F4, can repress G₂-M genes (17 , 49) . A possible alternative explanation would be that there is a still unidentified cellular protein(s) that can bind to E7 in the same region as Rb members and the mutation in the LXCXE motif of E7, which would include E7.C24G, also abrogates the binding of this factor. These alternative possibilities are presently being tested. Also, the p53 binding and degradation function, but not the telomerase activation function of E6, is necessary for disruption in chromosomal stability. Importantly, both E6 and E7 functions are required to cause significant changes in the polyploid phenotype under the culture conditions described.

The polyploidy observed in this study is probably caused by a defect during mitosis, rather than at cytokinesis. If there was a failure in cytokinesis, the cells would show multinucleation, and our results suggest that this is not the case. Firstly, the BrdUrd protocol required the isolation of nuclei. Because these nuclei showed >4N chromosome content, it suggests polyploidy cannot be due to multinucleation, although a small percentage of multinucleated cells (<2%) was observed, consistent with previous studies (23) . Secondly, using antibodies against NPAT, we show that E6/E7-expressing keratinocytes contain >2N copies of chromosome 1 and 6, whereas the control cells contain the normal two copies. A perturbation during mitosis would be the most straightforward suggestion, although this needs to be studied further. Endoreduplication has not been observed as each chromosome has two chromatids and not four chromatids, the expected number if endoreduplication had taken place.

The cause of the mitotic failure, resulting in polyploidy, may be caused by the deregulation, by E6 and E7, of genes in the G₂-M phase of the cycle that are normally regulated by the Rb family and p53. The microarray analysis revealed differential expression of a group of genes involved in the G₂-M-phase transition in keratinocytes expressing E6 and E7. Deregulated expression of some of these genes, namely the kinases Plk1, Aurora-A, and cdk1, which have functions at various stages of mitosis, have been shown to cause chromosomal instability in a variety of systems, including mammalian cells (39 , 40 , 50) . For example, overexpression of Plk1 and Aurora-A has been shown to cause centrosome amplifications presumably through deregulated mitosis and cytokinesis (39) . Also, in p53 null cells, this centrosome duplication and appearance of tetraploid cells is exacerbated most likely due to failure of a p53-mediated G₁ checkpoint (39) . Aurora-A and Plk1 are overexpressed in many tumors, and overexpression of Aurora-A in NIH 3T3 cells can lead to transformation in nude mice (51 , 52) . Therefore, we reasoned that the abnormal chromosomal segregation described in E6- and E7-expressing cells could, in part, be due to the deregulation of these mitotic kinases. In addition, these cells also lack intact spindle and DNA damage checkpoint mechanisms due to the presence of E6 and E7 and would be able to bypass arrest signals and continue to divide. One of the targets that is deregulated in the presence of E6/E7 is Plk1, a serine threonine kinase, which we examined in more detail. We demonstrated a correlation between up-regulation and maintenance of Plk1 expression and the polyploid phenotype observed in cycling and M-phase- arrested cells in E6/E7-expressing cells. Whether this deregulation alone or in combination with other G₂-M kinases is required for polyploidy is under investigation.

In summary, we have shown that HPV-16 E6 and E7 are required to act cooperatively to disrupt the spindle checkpoint and cause polyploidy. The ability of E7 to derepress member(s) of the Rb family and of E6 to degrade p53 are required for the observed chromosomal instability.