|Home | About | Journals | Submit | Contact Us | Français|
High-risk human papillomaviruses are the causative agents of cervical and other anogenital cancers. In these cancers, two viral oncogenes, E6 and E7, are expressed. E6 is best known for its ability to inactivate the tumor suppressor p53, which is thought to arise through ubiquitin-mediated degradation of p53 and involve a ternary complex between E6, p53 and the E3 ligase, E6AP. In mice transgenic for wild-type HPV16 E6, its expression leads to epithelial hyperplasia and an abrogation of normal cellular responses to DNA damage. Whereas only the latter phenotype is dependent upon E6's inactivation of p53, both are reduced in transgenic mice expressing an E6 mutant severely reduced in its binding to E6AP and other cellular proteins that bind E6 through a shared α-helix motif. Here, we investigated whether E6AP is required for the induction of the above phenotypes through the use of both E6AP-mutant and E6AP-null mice. E6, in the absence of E6AP retains an ability to induce epithelial hyperplasia, abrogate DNA damage responses and inhibit the induction of p53 protein following exposure to ionizing radiation. We conclude that E6 is able to induce both p53-dependent and p53-independent phenotypes through E6AP-independent pathways in the mouse.
Human papillomaviruses (HPVs) are small DNA tumor viruses that infect epithelial cells resulting in the development of warts. A subset of mucosotropic HPVs, the ‘high-risk’ HPVs including HPV16, are etiologically associated with nearly all cases of cervical cancer (zur Hausen, 1996; Walboomers et al., 1999) and approximately 25% of head and neck cancers (Gillison and Shah, 2001). Two viral genes, E6 and E7, that are consistently upregulated in human cervical cancer cells possess oncogenic activities including the abilities to immortalize/transform various types of cells (Munger et al., 1989; Band et al., 1991) and to induce cancers in animal models (Pan and Griep, 1994; Arbeit et al., 1996; Herber et al., 1996; Song et al., 1999, 2000; Riley et al., 2003). High-risk HPV E6 and E7 are most notable for their capacities to inactivate the tumor suppressors, p53 and pRb, respectively (Dyson et al., 1989; Werness et al., 1990; Huibregtse et al., 1991; Chellappan et al., 1992; Hubbert et al., 1992).
HPV16 E6 is a multifunctional protein that can associate with many different cellular factors, some of which have been subcategorized by how they bind E6. Among these subcategories are partners that bind the α-helix and the postsynaptic density-95/discs-large protein/zona occludens (PDZ) domains. Of these, the α-helix partner, E6AP, or E6-associated protein, is probably the most well known. E6AP or UBE3A, is associated with the human neurological disorder, Angelman syndrome (Nakao et al., 1994; Kishino et al., 1997; Matsuura et al., 1997) and belongs to the homologous to the E6-AP carboxyl terminus (HECT) family of E3 ubiquitin ligases (Huibregtse et al., 1995). High but not low-risk HPV E6 proteins are thought to inactivate p53 by forming a ternary complex with E6AP, thereby targeting p53 for ubiquitin-mediated degradation (Scheffner et al., 1993). E6's degradation of p53 is thought to contribute to the oncogenic potential of high-risk HPVs because p53 is inactivated as a consequence of mutation in many human cancers, but is not commonly mutated in HPV-associated cancers. E6AP has also been implicated in E6-mediated degradation of additional cellular partners including hScribble, a PDZ domain partner (Nakagawa and Huibregtse, 2000), hMCM7 (Kuhne and Banks, 1998), E6TP1 (Gao et al., 2002) and factors such as Myc (Gross-Mesilaty et al., 1998) which is involved in E6's activation of the TERT gene (Gewin and Galloway, 2001; Liu et al., 2005); however, it remains controversial whether E6AP mediates E6-induced degradation of other cellular partners of E6 such as Dlg, another PDZ domain partner (Pim et al., 2000; Grm and Banks, 2004; Matsumoto et al., 2006).
K14E6WT transgenic mice expressing the HPV16 E6 oncoprotein in stratified squamous epithelia have provided insight into the acute and long-term carcinogenic properties of E6 in vivo. These K14E6WT transgenic mice develop epithelial hyperplasia, spontaneous or chemically induced skin tumors and have their DNA damage responses abrogated in the epidermis (Song et al., 1998, 1999, 2000). The inactivation of p53 by E6 does not contribute to the development of epithelial hyperplasia (Nguyen et al., 2003a; Simonson et al., 2005), but does account largely for the abrogation of DNA damage responses (Song et al., 1998). Epithelial hyperplasia was severely reduced or eliminated in K14E6I128T and K14E6Δ146–151 transgenic mice (Nguyen et al., 2002, 2003a; Simonson et al., 2005), which express mutant forms of E6 unable to bind either the α-helix or the PDZ domain partners, respectively. Importantly, the degree of epithelial hyperplasia in K14E6WT, K14E6I128T and K14E6Δ146–151 transgenic mice correlates with the susceptibility of these mice to develop cancers in the skin (Nguyen et al., 2002, 2003a; Simonson et al., 2005) and in the cervix (Shai et al., paper submitted). E6's PDZ partners include the mammalian homologs of Dlg and Scribble, two genes that when disrupted in Drosophila result in the development of epithelial hyperplasia. The reduction or elimination of epithelial hyperplasia in K14E6I128T and K14E6Δ146–151 transgenic mice, respectively, supports the hypothesis that one or more α-helix partner(s) and one or more PDZ partner(s) contribute to E6-induced epithelial hyperplasia. Interestingly, E6AP has been argued to mediate E6's destabilization of hScribble (Nakagawa and Huibregtse, 2000) providing a potential interplay between α-helix and PDZ partners. The K14E6I128T transgenic mice also failed to display any inhibition of normal DNA damage responses, indicating that this E6I128T protein fails to inactivate p53 (Nguyen et al., 2002). This finding was consistent with the hypothesized role of E6AP in mediating E6's inactivation of p53.
We have now determined that the above-mentioned acute phenotypes of epithelial hyperplasia and the inhibition of DNA damage responses are retained in K14E6WT transgenic mice when placed on either an E6AP mutant or an E6AP-null background. In mice expressing an E6AP that is defective for its E3 ubiquitin ligase activity, or in mice completely devoid of E6AP, E6 retained its abilities to induce epithelial hyperplasia and to abrogate the DNA damage response, which was scored both by monitoring the inhibition of DNA synthesis and the induction of apoptosis in the epidermis. Remarkably, in K14E6WT transgenic mice on an E6AP−/− background, E6 was also able to prevent the accumulation of mouse p53 protein following treatment with ionizing radiation. These results indicate that E6 retains the ability to inactivate mouse p53 in vivo in an E6AP-independent manner.
To determine whether E6AP is required for E6-induced epithelial hyperplasia, we crossed the K14E6WT transgenic mice (FVB) (Taketo et al., 1991) onto an E6AP−/− (C57Bl6/129) background and monitored proliferation in 6-week-old adult mice by measuring incorporation of the nucleotide analog 5-bromo-2′-deoxyuridine (BrdU) into nascently synthesized DNA. In these and other experiments reported in this study, expression of the E7 oncogene, which is commonly co-expressed with E6 in HPV-associated cancers, was purposefully excluded, because E7's capacity to induce similar phenotypes would have masked our ability to monitor E6-dependent activities. As reported previously, K14E6WT transgenic mice on an E6AP-sufficient (K14E6WT/E6AP+/+) background are able to induce epithelial hyperplasia as indicated by higher levels of suprabasal DNA synthesis relative to non-transgenic (NTG) mice (8.9 vs 1.1%, respectively, P = 0.03) ((Song et al., 1999), Figure 1). K14E6WT transgenic mice on the E6AP-null (K14E6WT/E6AP−/−) background retained this phenotype: there was no difference in epithelial hyperplasia compared to that observed with the K14E6WT mice on the E6AP-sufficient background (P = 0.57). NTG, E6AP-null mice (NTG/E6AP−/−) had similar low levels of suprabasal DNA synthesis as seen in NTG mice that were E6AP-sufficient (NTG/E6AP+/+) (1.1 vs 1.0%, respectively, P = 0.3) (Figure 1). Thus, E6AP is not the α-helix partner required for E6-induced epithelial hyperplasia.
The primary response in the epidermis to DNA damage induced by ionizing radiation is a transient cessation of DNA synthesis owing to the upregulation of p53 (Kuerbitz et al., 1992). K14E6WT transgenic mice can inhibit this DNA damage response largely through its inactivation and degradation of p53 (Song et al., 1998). When E6 is hindered in binding to its α-helix partner E6AP, as in K14E6I128T mice, E6 loses its ability to inhibit the DNA damage response (Nguyen et al., 2002). This latter result supports the prevailing dogma that E6AP mediates E6's inactivation and degradation of p53. Therefore, we hypothesized that K14E6WT transgenic mice expressing a mutant E6AP defective in its ubiquitin ligase activity would fail to abrogate DNA damage responses in the epidermis because of a failure of E6 to inactivate p53. To test this hypothesis, we used E6APmut/mut mice carrying a mutant form of the E6AP allele (Miura et al., 2002). The E6AP encoded by these E6APmut/mut mice is defective in its ubiquitin ligase activity owing to a C-terminal mutation in the E6AP gene that eliminates a required C-terminal cysteine in addition to the last four amino acids, all of which are necessary for successful ubiquitin transfer (Talis et al., 1998; Salvat et al., 2004). K14E6WT transgenic mice were crossed onto the E6APmut/mut background and the resulting mice were subjected to ionizing radiation. The DNA damage response was measured by scoring the frequency of cells supporting DNA synthesis 24 h post-irradiation. NTG mice, regardless of E6AP status displayed a significant (P-values <0.05) reduction in DNA synthesis after exposure to ionizing radiation, reflecting their expected DNA damage response. Consistent with prior studies (Song et al., 1998; Nguyen et al., 2002, 2003a), this DNA damage response was abrogated in the K14E6WT/E6AP+/+ mice (Figure 2a), as evidenced by the retention of normal levels of DNA synthesis following exposure to ionizing radiation (P = 0.07). Surprisingly, the same property was observed in the K14E6WT/E6APmut/mut mice (Figure 2a); that is, E6 retained the ability to abrogate DNA damage responses in the absence of a fully functional E6AP ubiquitin ligase (P = 0.06).
A caveat with the interpretation of the results obtained in Figure 2a is that the mutant E6AP allele encodes a truncated E6AP gene product that, although expressed at reduced levels such that it is unable to ubiquitinate target proteins, could possibly still form a complex with E6 and p53, and thereby potentially interfere with p53 function. To address this caveat, we repeated the above analysis using mice homozygous for an E6AP-null (E6AP−/−) allele. NTG and K14E6WT transgenic mice on either the E6AP+/+ or the E6AP−/− genetic background were subjected to the irradiation studies as described above. NTG mice, regardless of E6AP status, once again displayed a significant reduction in DNA synthesis upon irradiation (P-values <0.05). As seen with K14E6WT/E6APmut/mut transgenic mice, E6 retained the ability to abrogate the DNA damage response in K14E6WT/E6AP−/− mice (P = 0.02) (Figure 2b). Therefore, E6 is able to inhibit the DNA damage response even in the complete absence of E6AP.
In response to DNA damage, such as that induced by ionizing radiation, cells with intact wild-type p53 not only can be inhibited in their DNA synthesis but also can undergo apoptosis (reviewed by Fei and El-Deiry, 2003). We therefore performed terminal deoxynucleotidyl transferase-mediated dUTP nick end labeling (TUNEL) assays on skin sections from both unirradiated and irradiated mice to assess the apoptotic responses. No apoptotic cells were detected within the epidermis of unirradiated mice in any of the genotypes tested (Figure 3a). Skin from irradiated NTG mice showed an induction of apoptotic cells in the epidermis, predominantly in the stratum basale for both dorsal skin and ear and also in the stratum spinosum for the ear (data not shown for ear) indicating that on this mixed (FVB/C57Bl6/129) genetic background both responses to DNA damage, the inhibition of DNA synthesis and the induction of apoptosis, can arise within the epidermis. Apoptosis was also observed in the hair follicles of both irradiated NTG and K14E6WT transgenic mice regardless of E6AP status (Figure 3a), consistent with prior studies indicating an absence of E6 effects in this epithelial compartment (Song et al., 1998). However, no apoptosis was observed in the irradiated epidermis of the stratum basale or the stratum spinosum of the dorsal skin (Figure 3a) or ear (data not shown) of either K14E6WT/E6AP+/+ or K14E6WT/E6AP−/− mice. Thus, we conclude that: (1) E6 inhibits a second DNA damage response in the epidermis and (2) E6AP is not required for E6 to inhibit either the DNA damage response such as the inhibition of DNA synthesis or the induction of apoptosis.
In K14E6WT mice, E6 inhibits the accumulation of p53 protein in the epidermis that normally arises following exposure to ionizing radiation (Song et al., 1998). This property is thought to reflect E6's ability to destabilize p53 protein (Scheffner et al., 1990). In human cells, downregulation of E6AP expression with antisense RNAs, or inhibition of its function through use of a dominant-negative form of E6AP, reduced E6's capacity to degrade p53 (Talis et al., 1998; Traidej et al., 2000). Therefore, we hypothesized that in the context of our K14E6WT transgenic mice, E6 would be unable to destabilize mouse p53 in the absence of E6AP. To test the hypothesis that E6AP mediates the destabilization of mouse p53 by E6, mouse p53 protein levels were monitored via immunohistochemistry on sections taken from either NTG or K14E6WT transgenic mice on either the E6AP+/+ or the E6AP−/− genetic background after irradiation or mock treatment. As expected, the levels of mouse p53 protein increased in NTG mice exposed to ionizing radiation (Figure 3b). NTG/E6AP−/− mice responded similarly to their NTG/E6AP+/+ littermates upon irradiation, demonstrating that mouse p53, like human p53 (Scheffner et al., 1993), is not a natural target of E6AP in the absence of E6. As previously observed (Song et al., 1998), irradiated K14E6/E6AP+/+ mice failed to show any detectable levels of mouse p53 in the epidermis (Figure 3). Remarkably, irradiated K14E6WT/E6AP−/− mice also displayed little to no detectable mouse p53 protein in the epidermis (Figure 3b). This surprising result reveals that E6 possesses alternative means for inducing the degradation of mouse p53.
In our studies using E6APmut/mut and E6AP−/− mice, we found that E6 can inhibit the DNA damage response in the absence of the E3 ubiquitin ligase, E6AP, and this inhibition correlates with an ability to prevent the induction of p53. Furthermore, E6-induced epithelial hyperplasia observed in K14E6WT mice is retained on an E6AP−/− background. Thus, E6AP is neither required for p53-dependent nor p53-independent phenotypes conferred by HPV16 E6 in vivo in mice.
Our prior studies indicated that both α-helix and PDZ domain partners of E6, but not p53, contribute to E6's induction of epithelial hyperplasia in the mouse epidermis (Song et al., 1999; Nguyen et al., 2002, 2003a; Simonson et al., 2005). Furthermore, the absence of complementation between K14E6I128T and K14E6Δ146–151 transgenic mice supported the hypothesis that the same molecule of E6 must interact with both an α-helix and a PDZ partner to induce epithelial hyperplasia (Nguyen et al., 2003a). E6AP became an intriguing possibility as the relevant α-helix partner because it had been implicated in mediating E6's degradation of the PDZ partner Scribble (Nakagawa and Huibregtse, 2000). Our finding that E6AP is not the relevant α-helix partner raises questions about the relevance of E6AP-mediated degradation of Scribble. One possible interpretation is that degradation of Scribble, per se, is not critical in mediating epithelial hyperplasia. Some investigators have observed decreased steady-state levels of Scribble in E6-expressing cells and tissues (Nakagawa and Huibregtse, 2000; Nguyen et al., 2003b; Massimi et al., 2004), others have not (Lee and Laimins, 2004; Simonson et al., 2005). Alternatively, one can hypothesize that some other cellular factor is responsible for mediating the degradation of Scribble or other PDZ partners in the absence of E6AP. Consistent with this possibility, Banks and coworkers recently observed the destabilization of multiple PDZ partners of E6 in E6-expressing E6AP−/− cells (P Massimi and L Banks, personal communication). It is possible that this other cellular factor which is causing the destabilization of these PDZ cellular factors in the absence of E6AP, is also contributing to E6-induced epithelial hyperplasia, a phenotype that correlates with E6's oncogenic properties in our transgenic mice (Nguyen et al., 2002, 2003a; Simonson et al., 2005; Shai et al., paper submitted).
E6AP can mediate E6's inactivation and degradation of human p53 (Huibregtse et al., 1991, 1993). E6's inactivation of human p53 has been argued to be E6AP dependent based upon the use of antisense oligomers (Traidej et al., 2000), a dominant-negative E6AP (Talis et al., 1998) and small interfering RNAs (siRNAs) (Hengstermann et al., 2001, 2005; Kelley et al., 2005). In particular, siRNA-mediated knockdown of E6AP in HPV16-positive cervical cancer-derived cell lines resulted in the same upregulation of p53-responsive genes as seen with E6 siRNAs (Kelley et al., 2005; Nomine et al., 2006). Thus, we originally hypothesized that E6's inactivation and degradation of mouse p53 would also be E6AP dependent, as in the case for human p53. However, our analysis of K14E6WT transgenic mice expressing either a defective E6AP ubiquitin ligase or no E6AP at all demonstrates that E6 retains its ability to abrogate the DNA damage responses and prevents accumulation of mouse p53 in the epidermis. Our result showing that the ubiquitin ligase activity of E6AP is dispensable for E6's inactivation of mouse p53 is consistent with a recent report by Nomine et al. (2006) investigating E6's inactivation of human p53, but differs in that we provide evidence that the protein E6AP itself is also not required to degrade mouse p53. Perhaps the difference in the dependence on E6AP for E6-mediated degradation of p53 reflects the species type of p53 being investigated. In our in vivo study, we looked at endogenous mouse p53. Tissue culture studies, on the hand, have investigated human p53. Consistent with this possibility, studies using mouse embryo fibroblasts isolated from E6AP−/− mice indicated that E6's degradation of human p53 was dependent upon E6AP (Cooper et al., 2003). Another possible difference may be that the levels of E6 protein expressed in the biological systems studied are indeed not physiological. Higher than physiological levels of E6 expressed in these systems could lead to novel, but biologically irrelevant activities. In this regard, we have recently analysed E6 protein levels in tissues from our K14E6WT mice using a newly acquired, highly sensitive HPV16 E6 antibody. The results indicate that the levels of E6 expressed in our mouse model are within the range seen in human cervical cancer-derived cell lines and early passage cells derived from cervical intraepithelial neoplasia (CIN) lesions containing integrated viral genomes, which is commonly observed in cervical cancer (Shai et al., paper submitted).
Other hypotheses have been raised to explain how E6 inactivates p53 independently of E6AP, specifically, through E6's binding to the p53 transcriptional coactivators p300/CBP (Patel et al., 1999; Thomas and Chiang, 2005) or ADA3 (Kumar et al., 2002), thereby blocking p53-dependent transcription. It is difficult, however, to reconcile our results with these alternative hypotheses, as we observed a continued ability of E6 to inhibit the induction of p53 on the E6AP-deficient/defective backgrounds, an observation that is not predicted by these alternative hypotheses. Thus, from our results, we predict that E6 can recruit another factor besides E6AP to induce the degradation of mouse p53. Because K14E6I128T mice fail to display any inhibition of p53, we hypothesize further that this factor is likely another α-helix binding partner, bringing further speculation that this other α-helix binding partner may be causing the destabilization of multiple PDZ substrates in E6AP−/− epithelial cells. Whether this other factor is the preferred choice by which E6 inactivates mouse p53, is used in tandem with E6AP, or is only active in the absence of E6AP remains unclear.
The K14E6WT transgenic mouse line (FVB) has been previously described and characterized (Song et al., 1999). Mice (C57Bl/6J) that contain a defective E6AP ubiquitin ligase (E6APmut/mut) were obtained from the Wagstaff laboratory (Miura et al., 2002). E6AP knockout (E6AP−/−) mice (129/SvEv) were obtained from Yong-hui Jiang and have also been previously characterized (Jiang et al., 1998). For all experiments described, homozygous K14E6WT transgenic mice were crossed either to E6APmut/mut or E6AP−/− mice to generate a heterozygous F1 generation. The F1 generation was then backcrossed to either to E6APmut/mut or E6AP−/− mice to generate all necessary genotypes (NTG, K14E6WT, K14E6WT/E6APmut/mut and K14E6WT/E6AP−/−) on a mixed FVB/C57/129 genetic background. Siblings of the different genotypes were used for all studies to discount genetic background effects. All mice were bred and maintained in the American Association for Accreditation of Laboratory Animal Care-approved McArdle Laboratory Cancer Center Animal Care Facility and studies carried out according to an Institutional Animal Care and Use Committee-approved animal protocol.
Eight-day-old mice were exposed to 4.41 Gy of ionizing radiation from a 137Cs source or mock-treated and killed 24 h later. The mice were injected intraperitonatally with the nucleotide analog, BrdU 1 h before killing. BrdU (12.5 mg/ml in phosphate-buffered saline (PBS)) was injected at 1 μl/10 g body weight. Dorsal skin samples were harvested and fixed in 10% buffered formalin for histological sections.
To quantify epithelial hyperplasia, the total number of suprabasal BrdU-positive cells were counted and divided by the total number of cells and multiplied by 100 to determine the percentage. To quantify the DNA damage response, the total number of BrdU-positive cells were counted and divided by the total number of cells in the epidermis and multiplied by 100. BrdU was counted in 10, × 40 microscopic fields per mouse (number of cells range from 1200 to 2000), with a total of at least three or more mice per genotype group. Determination of statistical significance was performed using the two-sided Wilcoxon's rank sum test using the MSTAT software program (http://mcardle.oncology.wisc.edu/mstat).
Torso skin and ear tissues were fixed overnight in 10% phosphate-buffered formalin at 4°C. Tissues were paraffin embedded and cut into 5-μm serial sections. Sections were used to analyse BrdU incorporation immunohistochemically and immunofluorescence for TUNEL.
Sections were deparaffinized in xylenes and dehydrated in a sequence of graded alcohol/water mixtures, followed by a 10 min peroxidase quenching step in 3% H2O2 diluted in methanol. The slides were then washed with PBS and unmasked by boiling for 20 min in 10 mM citrate buffer, pH 6. For BrdU analysis, the slides were immersed in 2 N HCl for 20 min to unmask further. Slides were then blocked in a 5% mixture of powdered milk and horse serum diluted in PBS for 30 min. Primary antibody was applied to the sections at 1:100 for BrdU (Calbiochem, Darmstadt, Germany) or 1:500–1:1000 in blocking solution for p53 (CM5, Novocastra Laboratories, UK) overnight at 4°C. Sections were subjected to several PBS washes. A universal biotinylated secondary antibody and streptavidin-peroxidase were applied to the sections as per instructions from the Vectastain ABC kit (PK-6200, catalog #, Vector Labs, CA, USA). The sections were developed with 3,3′-diaminobenzidine solution (SK-4100, Vector Labs) and stopped with H2O. The sections were then counterstained with hematoxylin, rehydrated in a sequence of graded alcohol/water mixtures and xylenes and then cover-slipped.
For TUNEL analysis, slides were deparaffinized in xylenes and dehydrated in a sequence of graded alcohol/water mixtures and the ApopTag Fluorescein kit (Intergen, Norcross, GA, USA) was used per manufacturer's instructions.
We thank Drs Drinkwater and Sugden for critical reading of the paper, members of the Lambert lab for helpful discussions and Lawrence Banks for the communication of unpublished observations. This study was supported by Grants CA098428, CA022443, CA014520 and CA009135 from the NIH.