The binding of the papillomavirus E6 protein to E6AP and the induction of p53 degradation are common features of high-risk genital human papillomaviruses (HPV); cutaneous HPVs, on the other hand, lack these capacities. Nevertheless, several cutaneous HPV types of the β-genus, such as HPV38 are associated with tumor formation when combined with genetic predisposition, immunosuppression, or UV exposure. In an animal model system, the cottontail rabbit papillomavirus (CRPV) rapidly induces skin cancer without additional cofactors, and CRPVE6 and E7 immortalize rabbit keratinocytes in vitro. However, CRPVE6 neither interacts with E6AP and p53 nor does it induce p53 degradation. In this study, we show that the interaction of CRPVE6, or HPV38E6, with the histone acetyltransferase p300 is crucial to inhibit the ability of p53 to induce apoptosis. Strikingly, E6 mutants deficient for p300 binding are incapable of preventing p53 acetylation, p53-dependent transcription, and apoptosis induction. Moreover, E6 mutants deficient for p300 binding cannot contribute to HPV38-induced immortalization of human keratinocytes or CRPV-induced tumor formation. Our findings highlight changes in the p53 acetylation status mediated by the viral E6 protein as a crucial requirement in the ability of high-risk cutaneous papillomaviruses to immortalize primary keratinocytes and induce tumors. Cancer Res; 70(17); 6913–24. ©2010 AACR.
Although more than 100 different human papillomaviruses (HPV) have been characterized based on sequence homologies, only a limited number were shown to be associated with cancer development. Thirteen so-called high-risk types of the α genus HPVs (α-HPV) play a critical role in the development of cervical cancer and have been implicated in other anogenital cancers and a subset of head and neck carcinomas (1, 2). In addition, β-HPVs have been linked to the development of nonmelanoma skin cancer, which represents the most common malignancy in Caucasians worldwide (3). β-Papillomaviruses such as HPV5 and HPV8 have been described to be associated with skin carcinogenesis in patients with the rare genetic disorder epidermodysplasia verruciformis. Patients with epidermodysplasia verruciformis develop warts, which progress in up to 60% of individuals mainly into primary squamous cell carcinoma (4). Furthermore, up to 70% of long-term immunosuppressed patients develop skin cancer, of which more than 80% contain HPV-DNA (5–7). A causative relationship between β-HPV infection and nonmelanoma skin cancer development in immunocompetent individuals is suggested by epidemiologic data and by the fact that the associated risk factors (age, UV-mediated local immunosuppression, and transplantation-related immunosuppression) point to an infectious agent, most probably belonging to the β2 group, which includes HPV38 (8, 9).
An excellent animal model to study papillomavirus-induced skin cancer formation is the New Zealand White rabbit, in which infections with cottontail rabbit papillomavirus (CRPV) causes the development of tumors within 3 to 6 weeks postinfection, progressing within 6 to 12 months into invasive cancer without additional cofactors (10, 11). Recently, we reported that the CRPV oncoproteins E6 and E7 are able to immortalize primary rabbit keratinocytes (12), a common feature of all carcinogenic HPV types (13), which is believed to be a key requirement for cancer formation.
High-risk HPVs disrupt cell cycle control by the interaction of E7 with pRB family members, activating telomerase, and inhibiting p53 via E6 (14). Comparable with high-risk HPV16, CRPVE7 alone immortalizes primary rabbit keratinocytes and coexpression of E6 increases the efficiency of immortalization (12, 15, 16). Like HPV16E7, CRPVE7 binds to and induces the degradation of pRB and thus induces cell cycle progression (12, 17, 18). However, in contrast with HPV16-immortalized cells, no reduction of p53 levels in CRPVE6/E7–immortalized cells was observed (12, 13). Comparable findings were described for β-HPV, in which E6 interacts with the ubiquitin ligase E6AP but is unable to degrade p53. Consistent with UV as a major risk factor, β-HPV types such as HPV38 were shown to prevent UVB-induced apoptosis by degradation of Bak (19, 20). This activity is surprisingly shared with high-risk genital HPV18, causing cervical cancer, in which UV is not a known cofactor. In contrast, CRPVE6 interacts neither with p53 nor with E6AP, and therefore, is not able to degrade p53 and probably not Bak as well (12, 20, 21). Currently, it is not understood why increased p53 levels in CRPVE6/E7-immortalized cells do not interfere with immortalization or tumor development (12).
The HPV38E6/E7 genes immortalize normal human keratinocytes (22) and induce hyperproliferation and neoplasia in a transgenic mouse model (23). This was proposed to occur by a p53-dependent transcriptional mechanism through ΔNp73 accumulation (24). Comparable with CRPV-positive cells, p53 accumulates but does not induce growth arrest or apoptosis in HPV38E6/E7–immortalized keratinocytes (20, 22). ΔNp73 represses the expression of p53-dependent target genes by binding site competition with p53 (24, 25). However, the mechanism of ΔNp73 induction bypassing the p53-ΔNp73 feedback loop and the individual contributions of HPV38E6 and E7 have not been addressed.
Thus, it seems possible that β-HPVs and CRPV block p53-dependent apoptosis independent of proteasomal degradation. Besides ΔNp73 activation, E6 could affect p53 functions by inhibition of p53-binding to DNA, mislocalization, or posttranslational modifications of p53 (26). An interesting mechanism is the inhibition of p53-regulated genes via interaction of E6 with the histone acetylase p300/CBP (27–29). p300/CBP modulates p53 activity via regulation of degradation by mdm2 (30, 31), coactivation of p53-regulated genes (32), and the acetylation of p53 (33). Prevention of p53 acetylation via interaction of E6 with p300 has been shown for high-risk and low-risk genital types, as well as for bovine papillomavirus type 1, but not for β-papillomaviruses or CRPV. Inhibition of p53 acetylation prevents p21 induction, which has been reported to alter the growth of keratinocytes (34) or to be antiapoptotic in human mammary epithelial cells (35). What remains to be determined are the functional consequences of the prevention of p53 acetylation mediated by p300 with regard to carcinogenesis and whether skin cancer–associated papillomaviruses manipulate this pathway.
Here, we show for the first time that CRPVE6 binds efficiently to p300 and prevents p300-mediated p53-induced apoptosis in vitro and in vivo, which is required for tumor formation by CRPV in vivo. HPV38E6 also binds p300 via a motif present in CRPVE6, but not in HPV16E6, and this interaction is necessary for keratinocyte immortalization by HPV38E6/E7. Taken together, our results imply a novel mechanism for p53 inactivation by β-HPV types and CRPV that is important for skin carcinogenesis.
Materials and Methods
Vectors and plasmids
Constructs were generated with PCR using the primers in Supplementary Fig. S1. pLA2-CRPV(10) was used as a template to generate plasmids expressing CRPVE6 and the respective mutants. Alanine mutants were generated by overlap extension PCR or by insertion of synthetic AvrII-BamHI oligonucleotides. Mutant CRPV genomes were constructed by replacing the BspEI fragment in pLA2-CRPV with fragments from the respective pMalc2x-CRPVSE6 plasmids. pMal, pcDNA, or pMSCVpuro expression vectors for HPV38E6 and 38E6Ala80–84 were generated by overlap extension PCR. The expression vector Rc/CMV-p53 was kindly provided by M. Scheffner (Department of Biology, University of Konstanz, Konstanz, Germany).
Cell culture and transient short interfering RNA cotransfection
Primary keratinocytes were isolated from normal skin tissue (12). C33A, HEK293T, and Saos-2 cell lines were provided by the American Type Culture Collection and authenticated using standard methods. Cells were aliquoted, immediately frozen, replated for the purposes of the experiments, and used within 6 months. Cells were cultivated as described previously (12). Saos-2 cells (4.5 × 105) were transfected 24 hours after seeding with 150 ng of short interfering RNA (siRNA; sip300-2, GGACUACCCUAUCAAGUAA) and 12 μL of HiPerFect reagent (Qiagen). The next day, cells were cotransfected with 3.4 μg of p53 expression vector or empty vector by calcium phosphate precipitation. RNA was isolated 30 hours posttransfection.
Quantitative real-time PCR
RNA was isolated with the RNeasy kit (Qiagen), and cDNA was synthesized from 1 μg of total RNA using the QuantiTect Reverse Transcription Kit (Qiagen). PCR reactions (20 μL) consisted of 10 μL of LightCycler 480 SYBR Green I Master (Roche), 50 ng of cDNA, and 3 μmol/L of forward and reverse primers (for sequences, see Supplementary Fig. S1). Relative transcript levels were calculated using phosphoglycerate kinase 1 transcripts as a reference (36).
Generation and analysis of stable keratinocyte cell lines
The generation of amphotropic recombinant retroviruses and infection of NHK was performed as previously described (37). Transduced cells were selected with puromycin (0.4 μg/mL) or G418 (50 μg/mL) for 14 days. To determine generation times, cells were trypsinized at 80% confluence and counted. The total cell number was then extrapolated.
p53 stability assays
CRPVE6/E7–transformed rabbit keratinocytes, HPV16E6/E7–, or HPV38E6/E7–immortalized human keratinocytes were treated with DMSO alone, 30 μg/mL of cycloheximide (Roth; dissolved in DMSO), or 20 μg/mL of MG132 (Merck; dissolved in DMSO). Treated and untreated cells were harvested at the indicated time points by scraping into cold PBS, pelleted by centrifugation, and lysed by boiling in 4× SDS sample buffer. Aliquots were then analyzed by immunoblotting.
Maltose-binding protein pull-down and coimmunoprecipitation assays
Maltose-binding protein (MBP) pull-down assays were performed as described previously (12). C33A cells were cultivated in DMEM supplemented with 10% FCS and gentamicin (0.5 mg/mL). Cells (8 × 106) were transfected with pcDNA-E6-HA or pcDNA-E6Ala200–204-HA using FuGENE HD transfection reagent (Roche). Forty-eight hours later, cells were lysed on ice in 700 μL of immunoprecipitation buffer [100 mmol/L NaCl, 50 mmol/L Tris (pH 8.0), 1 mmol/L EDTA, 0.1% Igepal CA-630, and 0.1% β-mercaptoethanol] and insoluble proteins were removed by centrifugation. An aliquot of the cell lysates was removed (input control), and the rest of the lysates were mixed with 50 μL of prewashed μMACS anti-HA Micro Beads (Miltenyi Biotec) and incubated for 3 hours at 4°C. Beads were collected in μColumns (Miltenyi Biotec) and washed five times with immunoprecipitation buffer. Bound proteins were eluted with 4× SDS sample buffer (95°C) and analyzed by immunoblotting.
Infections of New Zealand White rabbits with CRPV DNA were performed as previously described (10). Tumor numbers and the maximum diameter of each tumor were determined over a period of 5 months. Biopsies were removed 3 months postinfection, embedded in Tissue-Tek optimal cutting temperature compound (Sakura Finetek), snap-frozen, and stored at −80°C, or DNA was extracted using a BioRobot EZ1 workstation with the EZ1 DNA Tissue Kit (Qiagen). Viral DNA was amplified using primers CRPV444-463F and CRPV1124-1104R and sequenced.
Immunofluorescence microscopy and terminal deoxynucleotidyl transferase–mediated dUTP biotin nick-end labeling assay staining
Cells were grown on MatTek glass-bottomed culture dishes (MatTek Corporation), fixed for 10 minutes in 2% paraformaldehyde and incubated with HA antibody (sc-805; Santa Cruz Biotechnology) in PBS/3% bovine serum albumin for 1 hour at room temperature, washed, and then incubated with FITC-conjugated anti-rabbit IgG antibodies. Cells were counterstained with 4′,6-diamidino-2-phenylindole. For terminal deoxynucleotidyl transferase–mediated dUTP biotin nick-end labeling assay (TUNEL) staining, cells were fixed 48 hours posttransfection with 4% neutral-buffered formaldehyde for 5 minutes and permeabilized with 20 μg/mL of proteinase K (Merck) for 5 minutes at room temperature. TUNEL was performed using the QIA39 FragEL In situ Apoptosis Detection Kit (Merck). Fluorescence signals were visualized with a Zeiss Axiovert 200M microscope (Carl Zeiss MicroImaging GmbH). For each sample, at least five different fields were examined and counted to determine the rate of apoptosis.
Cells were lysed in radioimmunoprecipitation assay buffer (Sigma). Normalized amounts of protein (Micro BCA Protein Assay Reagent Kit; Thermo Scientific) were separated by 10% SDS-PAGE and transferred to nitrocellulose. Primary antibodies were α-p53 (DO-1, Santa Cruz Biotechnology), α-p53-Lys382 (2525S, New England Biolabs), α-tubulin (DM1A, Merck), or α-GFP (FL 32, Santa Cruz Biotechnology). Secondary antibodies conjugated to horseradish peroxidase (Dako) were detected using SuperSignal West substrates (Thermo Scientific) and Fluor-S Max MultiImager (Bio-Rad). MBP pull-down and coimmunoprecipitation eluates were run on 5% to 12% SDS-polyacrylamide gels and transferred to nitrocellulose. The membranes were probed with antibodies detecting p300 [C-20 (sc-585; Santa Cruz Biotechnology), ab37142 and ab10485 (both from Abcam)], and MBP (E8038; New England Biolabs).
Blots were developed with SuperSignal West (Thermo Scientific) substrate and visualized by Fluor-S Max MultiImager (Bio-Rad). Band intensities were quantified using the Quantity One quantification software package (version 4; Bio-Rad).
p53 is not degraded by E6 proteins of cutaneous papillomaviruses
The addition of the proteasome inhibitor MG132 to CRPV and HPV38E6/E7–immortalized keratinocytes indicated that the respective E6 proteins do not induce p53 degradation (12, 24). To further confirm these findings, HPV16, HPV38, or CRPVE6/E7–immortalized keratinocytes were treated with cycloheximide, an inhibitor of protein biosynthesis (Fig. 1A). In HPV16E6/E7–positive cells, cycloheximide treatment resulted in a rapid decrease of detectable p53 levels, whereas p53 was unaffected in CRPV or HPV38E6/E7–immortalized cells. Consistent with this, MG132 treatment only increased p53 in HPV16E6/E7 cells, but not in HPV38E6/E7 cells. These data clearly show that p53 is not degraded by CRPV or HPV38E6 proteins.
CRPVE6 represses the transcriptional activity of p53
To evaluate if rabbit p53 (rp53) is similar to human p53 (hp53) with regard to transcriptional activation, we expressed rp53 and hp53 in p53-deficient Saos-2 cells and tested for p21 induction. After 48 hours, a 4-fold induction of p21 was observed by rp53 and hp53. p21 induction of rp53 was strongly impaired by cotransfection of CRPVE6 (Fig. 1B), although it neither interacts with nor degrades p53 (12, 20, 21).
CRPVE6 interacts with p300
Immobilized E6 proteins fused to MBP were incubated with HEK293T cell extracts containing high amounts of p300 protein. Similar to high-risk HPV, CRPV encodes two forms of E6. Short E6 (SE6) is generated by the second methionine (M98) within the full-length E6 (long E6, LE6). Both CRPVE6 proteins revealed a much stronger interaction with p300 than the E6 proteins of HPV11, HPV16, or HPV18 (Fig. 1C). As the smaller SE6 protein also bound p300, this indicated that the binding motif is located within a region common to both proteins.
The interaction of HPV16E6 with p300 was described to rely on glycine 134 (G134), which is conserved among genital HPVE6 proteins (28, 38). The corresponding glycine at positions 135 (CRPVLE6) and 38 (CRPVSE6) were replaced by valine (MBP-SE6 G38V; Table 1; HPV16-G134 and corresponding glycines underlined) and tested for interaction with p300 by MBP pull-down assays (Fig. 1D). Surprisingly, both wild-type CRPVSE6 and the mutant SE6 G38V bound comparable amounts of p300, suggesting that the CRPVE6 interaction domain with p300 is different from HPV16E6.
Identification of the p300-binding region within CRPVE6
As both LE6 and SE6 interact with p300, we generated subfragments of SE6 to map the interaction domain by MBP pull-down assays (Fig. 2A). The amino acid positions are based on the respective amino acids in the full-length CRPVE6 protein (LE6). The minimal interaction domain consisted of amino acids 166 to 214 (Fig. 2B). Further COOH- or NH2-terminal truncations (amino acids 178–214 or 166–197) resulted in a complete loss of p300 binding. To further narrow down the binding motif, we performed an alanine scanning mutagenesis of amino acids 166 to 178 and amino acids 197 to 214 by replacing four to five consecutive residues with alanines (Fig. 2B). Quantification of the binding affinities derived from two to five individual MBP pull-down experiments revealed no effect by mutation of amino acids 210 to 214 or by amino acids 166 to 169. Mutation of amino acids 170 to 173, 174 to 177, 178 to 181, or 205 to 209 revealed an intermediate loss of p300 binding (25–75% compared with wild-type protein). In contrast, mutation of amino acids 195 to 199 or amino acids 200 to 204 resulted in a reduction in p300-binding affinity to 15% to 20% of the wild-type SE6 protein. This is similar to the background p300-binding activity of MBP (9% of wild-type SE6).
To verify this, we performed coimmunoprecipitation with an HA-tagged CRPV full-length E6 protein (Fig. 2C) in C33A cells, an HPV-negative keratinocyte cell line. To avoid expression of SE6 from the LE6 expression vectors, the methionine at residue 98 was exchanged with serine in the wild-type (LE6M98S) and both mutant proteins (LE6M98SAla195–199 and LE6M98SAla200–204). CRPVE6M98S, HA-tagged CRPVE6M98SHA, E6M98SHAAla195–199, or E6M98SHAAla200–204 mutants were transfected into C33A cells. Cell lysates were precipitated with HA antibody and the immunoprecipitates were analyzed for p300 by immunoblotting. Only the HA-tagged wild-type E6 protein was able to precipitate p300. The mutant E6 proteins were present even at higher levels than the wild-type, but did not immunoprecipitate p300 (Fig. 2C). These results show that CRPVE6 interacts with p300 and that residues 195 to 204 are crucial for this interaction and do not impair protein stability. Immunofluorescence analyses revealed that wild-type E6 and both mutants (E6Ala195–199 and E6Ala200–204) were present in the nucleus and, to a minor extent, in the cytosol (Fig. 2D), which is consistent with previous reports (21). Localization of E6M98SHAAla195–199 and E6M98SHAAla200–204 were indistinguishable from E6M98SHA. From these data, we conclude that these mutations in LE6 specifically prevent interaction with p300 but do not change subcellular localization.
Interaction of CRPVE6 with p300 inhibits p53 acetylation
Four different siRNAs against p300 were constructed and tested for their protein knockdown efficiency (Fig. 3A, top). The most effective siRNA (sip300-2) was further tested for p300 mRNA reduction (Fig. 3A, middle) and used for functional assays. The knockdown of p300 reduced p21 induction by rabbit and human p53 by 2- to 3-fold (Fig. 3A, bottom). It is known that p300 acetylates p53 at lysine-382, which stabilizes binding to the p21 promoter (39). Therefore, we exchanged the corresponding lysine-380 in rp53 to arginine (rp53K380R). In contrast with wild-type rp53, rp53K380R no longer induced p21 transcription in Saos-2 cells (Fig. 3B). This indicated that rp53, like hp53, depends on p300 as a coactivator. We then transfected Saos-2 cells with hp53 and CRPVE6, and analyzed the amount of p53 and p53 acetylated at lysine-382 (Fig. 3C). Cells transfected with p53 alone or p53 together with CRPVE6 revealed similar amounts of total p53. However, coexpression of CRPVE6 led to a loss of p53 acetylation at K382. When we cotransfected the p300-binding–deficient CRPVE6 (E6M98SAla200–204), acetylated hp53 was clearly visible, indicating that the interaction of wild-type CRPVE6 with p300 is responsible for inhibition of p53 acetylation.
CRPVE6 inhibits p53-induced apoptosis
As E6 prevents acetylation of p53 by p300, and therefore, p21 induction, we next investigated the effect of CRPVE6 on p53-mediated apoptosis. Saos-2 cells were cotransfected with expression vectors for hp53, rp53, rp53K380R, and wild-type CRPVE6 or E6M98SAla200–204. rp53 and hp53 induced similar numbers of apoptotic cells (19%) as determined by TUNEL staining, whereas rp53K380R did not (9.7 ± 1.3%; Fig. 3D). Interestingly, coexpression of CRPVE6 reduced apoptosis induction by rp53 to 10.8 ± 2.6%, whereas E6M98SAla200–204 did not (20.4 ± 1.3%). This suggests that the ability of CRPVE6 to abrogate p300-mediated acetylation of p53 is crucial for cell survival.
E6-p300 binding is important for carcinogenesis
To analyze the importance of the ability of E6 to bind p300 and inhibit the acetylation of p53 in vivo, we created p300-binding–deficient (pLA2-CRPVE6Ala195–199 and pLA2-CRPVE6Ala200–204) and competent (pLA2-CRPVE6Ala210–214) E6 mutant proteins in the context of the CRPV genome. Wild-type and mutant CRPV genomes were each injected at six sites in the skin of the back of two New Zealand white rabbits. Tumor development and growth were monitored for 5 months. Wild-type CRPV and pLA2-CRPVE6Ala210–214 produced tumors comparable in size and growth at all injection sites in both animals (Fig. 4A; Table 2). Multiple small tumors became visible around 25 days postinfection and grew up to 1 to 2 cm in diameter at all injection sites in both animals within 55 days. In contrast, pLA2-CRPVE6Ala195–199 and pLA2-CRPVE6Ala200–204 produced single, isolated, very small tumors, that never exceeded 3 mm in diameter (Fig. 4A; Table 2). The tumors induced by the mutant viruses retained their slower growth over the course of the experiment (Fig. 4A, Table 2). To examine if the reduced growth rate of the tumors was due to increased apoptosis, biopsies were taken 85 days postinfection. Cryosections of the biopsies were TUNEL-stained for apoptotic cells and analyzed by fluorescence microscopy. Whereas the tumors from both wild-type and pLA2-CRPVE6Ala210–214 genomes showed no signs of apoptotic cells, the tumors induced by p300-binding–deficient genomes showed increased numbers of apoptotic cells (Fig. 4B). CRPVE6/E7 genes were subjected to sequence analysis, which revealed that only the introduced mutations but no other mutations in the E6/E7 region were present in the tumor cells. This strongly suggests that the retarded tumor growth is a consequence of increased apoptosis due to a lack of interaction between CRPVE6 and p300.
The novel p300-binding motif of CRPVE6 and the ability to repress p53 functions are conserved in HPV38E6
A search for homologies of the CRPVE6 p300-binding motif (amino acids 195–204) in cutaneous HPVE6 proteins revealed the highest similarity with the E6 protein of nonmelanoma skin cancer–associated HPV38, in which amino acids 81 to 83 are identical to CRPVE6 amino acids 201 to 203. To investigate if HPV38E6 is able to interact with p300 and whether this interaction is mediated by amino acids 81 to 83, wild-type and mutant HPV38E6 (HPV38E6Ala80–84) were expressed as MBP fusion proteins. Pull-down experiments revealed that p300 bound to HPV38E6, but not to the mutant (Fig. 5A). In contrast, both the wild-type and the mutant HPV38E6 protein bound to E6AP (40), suggesting that the mutation is specific for p300 (Fig. 5A). Consistent with the observations for CRPVE6, only wild-type HPV38E6, but not 38E6Ala80–84, was able to repress p21 induction by p53 and acetylation of p53K382 in a p300-binding dependent manner (Fig. 5B and C).
p300 binding to HPV38E6 is important for immortalization
Comparable with high-risk genital HPV types, the E6/E7 proteins of HPV38 are able to immortalize primary human foreskin keratinocytes, which normally undergo senescence within approximately 20 days after isolation (22). To address if the E6-p300 interaction influences the immortalization capacity, HPV38E6 (wild-type or 38E6Ala80–84) and HPV38E7 were expressed from separate retroviral vectors (pLXSN and pMSCVpuro). Normal human foreskin keratinocytes from two different donors were infected in two independent experiments with E7 alone, wild-type E6 and E7, or 38E6Ala80–84 and E7 (Fig. 5D). For the first 20 days, all transfected cells displayed a slow proliferation rate, comparable with normal keratinocytes, followed by a period of 30 to 50 days without proliferation. Then, a phase of exponential cell proliferation only in cells infected with wild-type HPV38E6 and HPV38E7 continued for more than 15 passages. In contrast, cells transduced with HPV38E7 alone or HPV38E7 together with 38E6Ala80–84 did not proliferate, but continuously detached from the culture dish until all cells were lost. From these data, we conclude that p300 binding is essential for immortalization of primary human keratinocytes by HPV38E6/E7.
Although epidemiologic studies hint to a possible role for HPV types of the β2 genus in the development of skin cancer (8), the underlying molecular mechanisms have not been understood in great detail. Recent work has suggested that the E7 proteins of CRPV and the skin cancer–associated β2 subgenus type HPV38 interfere with pRB function similar to HPV16 and HPV18 (12, 22). In contrast, very little is known about how these viruses prevent p53-mediated apoptosis provoked by E7-activated E2F proteins (41). The E6 proteins of both viruses lack the ability to degrade p53 (21, 22), indicating that they rely on alternative mechanisms to inactivate p53.
To address differences or mutations in the rabbit p53, which shares 85% amino acid identity with human p53, we analyzed rp53 DNA sequences from the immortalized rabbit keratinocytes and tumors of different stages and found only silent mutations in comparison to the reference (NM_001082404; data not shown). Furthermore, in this report, we show that rp53 is very similar to hp53 with regard to p21 and apoptosis induction and dependence on acetylation by p300.
Both E6 of CRPV and HPV38 were able to prevent p300-mediated acetylation of p53 at lysine-382, probably through- a direct interaction with p300. Interestingly, we were able to detect weak p53 acetylation in Saos-2 cells transfected with p53 without using histone deacetylase inhibitors, which are commonly used to detect p53 acetylation (39).
Recently, it has been shown that loss of acetylation completely abolishes p53-dependent growth arrest and apoptosis (33). Acetylation of p53 abrogates Mdm2-mediated repression by blocking the recruitment of Mdm2 and Mdmx to p53-responsive promoters (42–44). The six carboxyl-terminal lysines of human p53, including K382, are acetylated by p300/CBP (45). In addition, lysine-120 is acetylated by Tip60/hMOF and lysine-164 by p300/CBP (33). Although it seems that some acetylation defects in human p53 could be compensated by the modification of other sites, loss of acetylation at all major sites completely abolishes the ability of human p53 to activate p21 (33). In contrast, only the rp53K380R mutation showed a major effect on p21 and apoptosis induction in Saos-2 cells, which could indicate the central role of this lysine for rabbit p53-induced apoptosis. Because seven of eight possible lysines of human p53 could be acetylated by p300/CBP (33, 45), and as we have shown that E6 prevents p300-mediated acetylation, it is conceivable that the acetylation of other lysines by p300 is also impeded. This needs to be tested in additional experiments. In that case, however, lack of acetylation at the COOH terminus might allow increased ubiquitination of p53 by Mdm2 and thereby affect p53 stability, which is clearly not the case in CRPV- or HPV38-immortalized keratinocytes. The ubiquitin ligase activity of Mdm2 could, however, at least in CRPV immortalized cells, be inhibited by high rp19Arf levels (12). p53 hypoacetylation mediated by E6 could then allow binding of p53 together with Mdm2 and Mdmx to p53-responsive promoters and thereby prevent the transcription of proapoptotic genes, which would explain the effects we observed in this work. In addition, it has been recently shown that acetylation at K320, K373, and K382 is required for the transcription-independent proapoptotic functions of p53 at the mitochondrion (46). Whether this also contributes to the antiapoptotic effects of CRPVE6 remains to be determined.
Two known CRPV strains with low or high tumor progression potential show amino acid exchanges (Q225L or G200D) in the carboxyl-terminal part of E6 (47), which is the newly identified p300 interaction domain. The reduced carcinogenicity of the E6G200D genome probably results from a decreased E6-p300 interaction as indicated by our data, and as a consequence, increased expression of p53-dependent proapoptotic target genes.
The highest similarity to the p300-binding domain of CRPV among cutaneous HPVE6 proteins was found with HPV38E6. This was notable as HPV38 and CRPV are the only cutaneous papillomaviruses that are able to immortalize normal keratinocytes. Consistent with the results obtained with CRPV, immortalization of primary keratinocytes by HPV38E6/E7 is dependent on the E6-p300 interaction. It was reported previously that p53 accumulates in HPV38E6/E7–immortalized keratinocytes and induces ΔNp73, but is unable to displace ΔNp73 from p53-responsive promoters (24). This is consistent with our findings, as the binding of p53 to its responsive promoters is dependent on the acetylation state (45, 48). The observed lack of immortalization of keratinocytes by a p300-binding–deficient HPV38E6 could be due to acetylation of p53 by p300, which could then overcome the ΔNp73 blockage and induce senescence or apoptosis. Interestingly, the cutaneous papillomavirus types behave similarly to the adenoviral E1A protein, which binds to the TAZ2 domain of p300/CBP and competes with p53 for p300 binding (49). Although the binding domain of E1A shows only little overall homology to the CRPV or HPV38E6 p300-binding domain, all three proteins contain a hydrophobic residue followed by phenylalanine. One difference between CRPVE6 and the cutaneous HPVE6 proteins is the ability of the latter to bind E6AP, which makes them capable of degrading the UV-induced Bak protein to escape the induction of apoptosis. For CRPV, such a function is not required as the fur of the rabbit protects the skin from UV irradiation.
It has been described that genital HPV types, which inactivate p53 via E6AP-dependent degradation, also prevent p300-mediated acetylation of p53. However, these E6 proteins differ from CRPV and HPV38E6 because they can bind to p53 and form trimeric complexes with p53 and p300 (28). The HPV16E6 protein interacts with the NH2-terminal CH1 and with COOH-terminal regions of p300/CBP, including the TAZ2 domain (27, 29). The mechanism of p53 inactivation obviously differs from E1A as HPV16E6 still requires p53 binding to repress its transcriptional activity. Furthermore, the interaction domain described for HPV16E6 with p300 is different from the one we identified for CRPV and HPV38E6. For HPV16E6 residue, G134 within the second zinc finger domain was described to be important (28). However, this mutation has been described to be impaired not only for p300 binding but also for E6AP and p53 binding and degradation (38). The benefit of p300-mediated p53 inactivation for HPV16 and HPV18 is puzzling as they, in contrast to CRPV and HPV38E6, predominantly inactivate p53 via degradation, suggesting that repression of p300-mediated p53 acetylation might be necessary for its ubiquitination by E6AP. This was reported for Mdm2-mediated p53 degradation (50). Alternatively, p300 inhibition might be necessary to maintain residual undegraded p53 in an inactive state before its degradation.
In summary, this is the first report demonstrating that the E6 protein of skin cancer–associated papillomaviruses mainly targets p53 acetylation and not degradation of p53 via binding to p300 to immortalize primary keratinocytes and to induce tumors in an in vivo situation.
Disclosure of Potential Conflicts of Interest
No potential conflicts of interest were disclosed.
We thank Martin Scheffner (Department of Biology, University of Konstanz, Konstanz, Germany) for providing reagents.
Grant Support: Deutsche Forschungsgemeinschaft SFB 773 B4 (T. Iftner) and funding under the Sixth Research Framework Programme of the European Union, Project INCA (LSHC-CT-2005-018704).
The costs of publication of this article were defrayed in part by the payment of page charges. This article must therefore be hereby marked advertisement in accordance with 18 U.S.C. Section 1734 solely to indicate this fact.
- Received April 15, 2010.
- Revision received June 15, 2010.
- Accepted July 2, 2010.
- ©2010 American Association for Cancer Research.