Oncogene (2010) 29, 5061–5070 & 2010 Macmillan Publishers Limited All rights reserved 0950-9232/10 www.nature.com/onc ORIGINAL ARTICLE The E2F5 repressor is an activator of E6/E7 transcription and of the S-phase entry in HPV18-associated cells

S Teissier1, CL Pang2 and F Thierry1

1Institute of Medical Biology, A-STAR, Singapore and 2Graduate School for Integrative Sciences and Engineering, National University of Singapore, Singapore

High-risk papillomavirus type 18 (HPV18) is one of the crucial role in carcinogenic progression. The pattern of less represented HPV types in low-grade lesions of the the viral DNA integration implies conservation of the anogenital tract, whereas it occupies the second place in viral oncogenes, E6 and E7, which, respectively, induce cervical cancer, where it can be found in 16% of the cases degradation through the ubiquitin-proteasome worldwide, after HPV16 present in 54% of them. These pathway (Scheffner et al., 1993) and pRb inactivation epidemiological data indicate that HPV18 infection is (Dyson et al., 1989) and the disruption of the E2 open more prone to carcinogenic progression. The main reading frame encoding a potent repressor of E6/E7 oncogenic , E6 and E7 of HPV18, are functionally transcription (Thierry and Yaniv, 1987; Bernard et al., comparable to the homologous proteins of the other high- 1989; Thierry and Howley, 1991). For high-risk HPV risk viruses, including HPV16. In this work, we investi- types HPV16 and 18 represent 70–80% of all cervical gated the possibility that the higher oncogenic potential of cancers worldwide (Castellsague et al., 2007). Prevalence HPV18 might be due to transcriptional regulation of the of the two high-risk papillomavirus types, HPV16 and E6/E7 oncogenes. By comparing the E6/E7 promoter and HPV18, in low-grade and high-grade lesions of the enhancer sequences of the mucosal HPV genomes, we cervix, indicates that HPV16 is the most abundant identified binding sites specific for HPV18. The E2F genotype in all types of lesions (20.3% in low grade: first family of transcription factors contains activators (– position and 54.4% in cervical cancer: first position), 3) and repressors (–8) that regulate the transcription but the prevalence of HPV18 increases with the grade of of S-phase and mitotic and thereby have a crucial the lesion (6.1% in low-grade lesions: 10th position and role in cell-cycle progression. Surprisingly, we identified 15.9% in cervical cancer: second position) (Castellsague E2F5 as a direct activator of HPV18 E6/E7 transcription et al., 2007). This observation was supported by the by sequential silencing of E2F members in HeLa cells. In finding that the transforming activity of HPV18 in vitro addition, we could show that E2F5 positively regulates is better than that of HPV16 (Villa and Schlegel, 1991). S-phase entry in HeLa cells and that this activation of the As the known properties of the E6 and E7 oncogenes are cell cycle by a member of the E2F repressor family is well conserved among high-risk HPV types, a way to specific for HPV18-expressing cells. Diverting the func- explain the differences in transforming efficiencies is to tion of E2F5 from a cell-cycle repressor into an activator examine their transcriptional regulation. In a previous might contribute to the higher oncogenic potential of study, this difference correlated with the fact that HPV18 HPV18 when compared with other high-risk HPV types. transcription of the oncogenes seemed to be approximately Oncogene (2010) 29, 5061–5070; doi:10.1038/onc.2010.246; 10- to 50-fold higher than transcription of the HPV16 published online 19 July 2010 oncogenes in keratinocytes (Villa and Schlegel, 1991). In HPV18, the E6 and E7 proteins are translated from Keywords: papillomavirus; transcription; HPV; E2F; differential splicing of a unique mRNA, the transcrip- cell cycle tion of which is controlled by the P105 promoter and an upstream tissue-specific enhancer. The HPV18 enhancer has been shown to bind several transcription factors, Introduction such as AP1, HMG-I(Y) or the nucleolin (Bouallaga et al., 2000, 2003; Grinstein et al., 2002; reviewed in High-risk human papillomaviruses (HPVs) are the Thierry, 2009). It can form a three-dimensional struc- causal agents of cervical cancer, and their genome is ture, called enhanceosome, through efficient recruiting found in more than 99% of these cancers (Walboomers of coactivators such as CBP/P300, and can induce high et al., 1999), where integration of the viral genome has a transcriptional levels (Bouallaga et al., 2003). In this study, we describe the role of a new cellular in the HPV18 enhancer, which is a Correspondence: Dr F Thierry, Institute of Medical Biology, Immunos member of the E2F family. These transcription factors #06-05, 8A Biomedical Grove, Singapore 138648, Singapore. E-mail: [email protected] are involved in the G1/S transition of the cell cycle and, Received 28 December 2009; revised 14 May 2010; accepted 20 May as discovered more recently, also in the G2/M transition 2010; published online 19 July 2010 (Ishida et al., 2001; Ren et al., 2002; Polager and E2F5 activates HPV18 E6/E7 transcription S Teissier et al 5062 Ginsberg, 2003; Thierry et al., 2004; Zhu et al., 2004). that this sequence was previously found to be essential for There are eight E2F members; E2F1–3 activate the HPV transcription in sequential deletion experiments transcription of S-phase genes, whereas E2F4–8 repress (Figure 1a). We reasoned that if the E2F family members their transcription. E2F5 (Sardet et al., 1995) had been were involved in the HPV18 E6/E7 transcription, their shown to repress the cell cycle and to induce G0 in expression should vary concomitantly during the cell keratinocytes, depending on its sub-cellular localization cycle. Analyses of the mRNA levels of six E2F family and therefore probably favoring differentiation, members by reverse transcriptase–PCR (RT–PCR) in although the mechanisms remain unclear (Apostolova synchronized HeLa cells indicated that mRNA levels of et al., 2002; Reed et al., 2007). the , 4 and 6 were not modulated, whereas those of In this work, we characterized an unexpected activat- the E2F1 and 2 activators were decreased in G2 and, on ing role of E2F5 in cell-cycle progression of HPV18- the contrary, mRNA levels of the repressor, E2F5, were associated cells. This activation depends on the HPV18 activated (Figure 1b). transcription and involves the binding of E2F5 to the We then investigated whether the endogenous E7 HPV18 enhancer. This observation suggests a new transcription in HeLa cells was also cell-cycle modu- regulatory function of HPV18 in overcoming G0 arrest lated. We used HeLa cells synchronized by thymidine by changing the cell-cycle repressor E2F5 into an block and released at different time points, as shown by activator, through direct transcriptional activation of flow cytometry analyses (Figure 1c). We then studied E7 and subsequent S-phase entry. E6/E7 mRNA levels by RT–PCR and E7 expression by western blots in these cells at different phases of the cell cycle. E6/E7 mRNA levels were found Results to be higher in the late G1 and S phases compared with early G1 and G2/M (Figure 1d, upper panel). Modula- E6/E7 transcription varies during the cell cycle together tion of the quantity of E7 protein was also detected, with the E2F transcription factors although with a clear delay compared with the To identify new potential transcription factors involved in transcription (Figure 1d, lower panel). The higher E6/ cell-cycle-dependent HPV18 transcription, we analyzed E7 mRNA levels in the early S phase are therefore the sequences of HPV18 regulatory region using the consistent with regulation of the HPV18 oncogene TRANSFAC software. Two potential E2F binding sites transcription by the E2F transcription factors (Mudryj were found in the core of the enhancer as defined et al., 1991; Johnson et al., 1993). previously (Bouallaga et al., 2000) (Figure 1a). Sequences of the two sites were close to the consensus sequence for The HPV18 enhancer binds and E2F5 in vitro E2F binding, but the upstream site A marked the limit of We then examined which of the three E2F factors is the previously defined core enhancer region, indicating involved in HPV18 E6/E7 transcription by studying

Figure 1 Cell-cycle modulation of E2F and E7 transcription in HeLa cells. (a) Sequence of the core HPV18 enhancer containing two potential E2F binding sites A and B as indicated. (b) Detection of the levels of transcription of six members of the E2F family as shown by RT–PCR in synchronized HeLa cells as described in panel (c). (c) Fluorescence-activated cell sorting analyses of the phases of the cell cycle in HeLa cells after thymidine block at 0, 6 and 12 h release as indicated. (d) E6/E7 transcription varies during the cell cycle and is higher in late G1, decreasing progressively during S phase. These variations of mRNA levels are consistent with variation of the E7 protein levels observed by immunoblot (lower panel).

Oncogene E2F5 activates HPV18 E6/E7 transcription S Teissier et al 5063

Figure 2 Gel-shift assays to detect E2F binding from HeLa nuclear extracts and purified E2F1. (a) Three different probes containing the E2F consensus sequence, site A and site B from HPV18, as indicated at the top, were incubated with no protein (–), HeLa nuclear extract and the purified E2F1 as indicated. Free probes are indicated by arrows, specific shifts are indicated by square brackets and a non-specific band by an asterisk. (b) The E2F consensus probe was incubated in HeLa cell extract followed by pre-incubation with the anti-E2F1, anti-E2F2 or anti-E2F5 antibodies as indicated. Supershifts are shown by arrowheads and the non-specific band by an asterisk as in (a). their binding to the two E2F putative binding sites A on the silencing of the E2F transcription factors by and B in gel-shift assays (Figure 2a). We compared the pSUPER shRNA-expressing vectors. Silencing of the binding profiles of an E2F consensus binding site with three E2F activators separately in HeLa cells did not two probes containing the HPV18 A and B sequences in reduce their respective mRNA levels, thus indicating gel-shift assays using HeLa nuclear extracts or purified that they could compensate each other (data not shown) E2F1 protein fused to GST. Only the upstream putative as previously reported (Kong et al., 2007). We, there- E2F binding site A of the HPV18 core enhancer bound fore, transfected the three pSUPER vectors together and efficiently to both the purified E2F1 and factors present could obtain silencing of the transcription of E2F in the HeLa nuclear extract. Supershift experiments, activators (Figure 3a). As for the E2F5 repressor, the using specific antibodies against the E2F family silencing of the three activators did not modulate its members, indicated that from a HeLa cell extract mRNA levels, which were, however, efficiently silenced (Figure 2b) both E2F2 and E2F5 were able to bind to by the use of the specific E2F5 shRNA sequence the E2F binding site, whereas no supershift was found (Figure 3a). Intriguingly, silencing of the E2F activators with the anti-E2F1 antibody, which was shown to did not modulate E6/E7 transcription, whereas silencing efficiently shift the purified protein (data not shown). of E2F5 strongly reduced the E6/E7 transcription by These experiments indicated that the HPV18 core 70%, indicating a positive control of the endogenous enhancer sequence is able to bind to E2F and that this HPV18 transcription by the E2F5 repressor in HeLa binding is presumably responsible for cell-cycle-depen- cells (Figure 3b). The effect of E2F5 silencing on E7 dent transcription of the HPV18 E6 and E7 genes in expression was also observed at protein level in western HeLa cells. blot experiments (Figure 3b, lower panel). We have previously shown that the HPV18 E7 expressed in HeLa cells is a strong activator of many E2F5 activates S-phase genes through activation of E7 E2F targets (Thierry et al., 2004; Teissier et al., 2007). If transcription E2F5 is an activator of E7, its inactivation should The putative involvement of the E2F5 repressor in the reduce the transcription of S-phase target genes in regulation of HPV18 transcription was intriguing and, HeLa. To verify this hypothesis, we transfected HeLa to determine its role, we used an approach that relied cells with the E2F shRNA expression plasmids and

Oncogene E2F5 activates HPV18 E6/E7 transcription S Teissier et al 5064

Figure 3 Endogenous HPV18 E7 expression in HeLa cells is dependent on E2F5. (a) Real-time RT–PCR analyses of the transcriptional silencing with the pSUPER shRNA E2F1, 2, 3 and 5 expressing vectors. The mixture of shRNA of E2F1, 2 and 3 activators efficiently silenced their transcription, whereas shE2F5 did not significantly affect it. On the contrary, shRNA E2F5 silenced its own transcription to about 50%, whereas it did not affect the transcription of E2F1, 2 and 3. (b) Real-time PCR analyses of endogenous HPV18 E6/E7 and cyclin E1 mRNA in transfected HeLa cells. HPV18 E7 silencing in HeLa cells decreases E7 mRNA and has a repressive effect on cyclin E1 transcription. E2F5 silencing alone induces a repression of cyclin E transcription, which is equivalent to that obtained by the silencing of E2F activators (1, 2 and 3) or silencing of E2F activators and E2F5. Silencing of E2F5 decreases the endogenous HPV18 E7 protein level (lower panel). (c) Real-time PCR analysis of HPV18 E6/E7 mRNA level in C-41 cells transfected with the pSUPER control and E2F5 as indicated and in nTERT cells cotransfected with the HPV18 genome. (d) ChIP with an anti-GFP in GFP–E2F5-transfected HeLa cells. E2F5 binds directly to the HPV18 LCR and polA promoter and not to the GADPH promoter (negative control). (e) Luciferase assay in C33-A cells transfected with the HPV18 LCR wild type (LCR18), the mutant for the E2F binding site A (LCR18 mA), the mutant for the E2F binding site B (LCR18 mB) and the double mutant (LCR18 mAmB) in the presence of the pSUPER control or the pSUPER E2F5 as indicated.

found that, indeed, the transcription of the S-phase E1 remains controlled by the E2F activators. Activation cyclin E1 was repressed both by the silencing of E2F of S-phase genes in HeLa cells would rather come from activators and by silencing of E2F5 (Figure 3b). an indirect activation of the E7 transcription followed However, it is interesting that although silencing of the by activation of the E2F activators through modulation E2F activators repressed cyclin E1 mRNA levels, pre- of the negative regulator pRb. An assumption from this sumably by a direct effect, it had no effect on HPV18 hypothesis is that E2F5 can function as an activator of transcription as already mentioned. Taken together, these the cell cycle only when it is able to modulate E7 data indicate that E2F5 might have an activating role for transcription. the E2F target genes in HeLa cells, although transcription We verified that the silencing of E2F5 did not affect of these genes remains under the control of the classical the transcription of the cyclin E1 (S-phase gene) in E2F activators (E2F1–3). When cells are transfected with HPV-negative C33-A cells (data not shown), thus pSUPER-expressing shRNA for E2F1, 2, 3 and 5 all implying that the regulation in HeLa cells was likely together, there is no additive effect on cyclin E1 repression to be due to regulation of the E7 transcription. and the repression of E6/E7 transcription observed is To test whether E2F5 activation of the HPV18 E6/E7 comparable to that obtained using pSUPER-E2F5 alone transcription could be extended to other HPV18- (Figure 3b). From these experiments we could deduce that associated cells, we transfected the C-41 cervical E2F5 is not a direct activator of cellular S-phase genes, carcinoma cell line with pSUPER-E2F5 and could show as it is expected from previously published data, and that the endogenous E6/E7 transcription was repressed that, for example, an S-phase gene promoter such as cyclin concomitantly with E2F5 silencing (Figure 3c). We also

Oncogene E2F5 activates HPV18 E6/E7 transcription S Teissier et al 5065 detected repression of E6/E7 transcription of episomal inhibition)). The same experiment was done with DNA in nTERT (primary keratinocytes immortalized E2F1, 2 and 3 silencing and showed a G1/S arrest (data with telomerase hTERT) cells cotransfected with the not shown). These results are consistent with the HPV18 genome and the pSUPER E2F5 (Figure 3c), transcriptional analyses and indicate that E2F5 indeed thus confirming that E2F5 is an activator of HPV18 E6/ has a role in the induction of the S-phase of the cell cycle E7 transcription from both integrated and episomal in HPV18-positive cells. The effects observed on the cell HPV18 sequences. cycle were validated by cell-growth experiments with We were intrigued by the fact that the transcriptional HeLa cells transfected with various pSUPER plasmids repressor E2F5 protein could modulate HPV18 tran- to induce silencing of the E2F activators and E2F5 scription, and decided to check whether it could bind its (Figure 4b). Cell growth was monitored for 88 h, during regulatory region in vivo by chromatin immuno-pre- which the control cells continuously grow, while in the cipitation (ChIP) experiments. We used a transfected presence of pSUPER-E7 the cells stopped growing expression vector expressing E2F5 fused to the green rapidly around 38 h post-transfection and did not start fluorescent protein (GFP) to perform ChIP experiments growing again. Cells transfected with either pSUPER- on the endogenous HPV sequences integrated in the cellular E2F1, 2 or 3 (all the activator members) or pSUPER genome in HeLa cells. We also checked whether E2F5 will E2F5 show a decrease in their growth speed after 43 and be recruited to the promoter of PolA, an S-phase gene used 52 h post-transfection, then eventually stop growing in this study as a typical E2F target. In both cases E2F5 was completely after 52 h. In all cases, the growth arrest was recruited, although it was not recruited by the negative sustained for more than 30 h and none of these cells control GADPH, indicating a specific direct E2F5 binding showed any relapse during this time frame, thus to the HPV18 enhancer (Figure 3d). confirming a sustained anti-proliferative function of To further confirm the direct involvement of E2F5 in the E2F5 silencing in HeLa cells. To check whether the transcriptional activation of the HPV18 early promoter, cell-cycle modulation is dependent on HPV18 E7 we transfected C33-A (HPV negative) cells with different transcriptional modulation, we performed similar ana- constructs containing either the wild-type HPV18 LCR lyses in the HPV-negative C33-A cells. As expected, the or mutants for E2F binding sites A and/or B (the same shRNA against E7 did not affect their cell cycle, point mutations that impair E2F binding in Figure 2 whereas E2F5 silencing showed neither positive nor were used), controlling luciferase transcription. Muta- negative effect on the G1/S transition (Figure 4b). These tion of either site A or site B significantly decreased the results confirmed that the positive role of E2F5 on the LCR activity, with a stronger effect of site A mutation cell cycle depends on the presence of the HPV18 and and no additive effect of the two mutations together expression of the E6/E7 oncoproteins. (Figure 3e). Cotransfection of pSUPER-E2F5 induces a decrease of the luciferase activity, although not as strong and reproducible as the decrease of the E6/E7 mRNA E2F5 is not a cell-cycle activator of HPV16-associated observed in HeLa, C-41 and HPV18-transfected cells nTERT. Site A seems to be more sensitive to E2F5 Using the TRANSFAC software we looked for E2F silencing, which might reflect its greater affinity for E2F binding sites in the regulatory region of other HPV binding as shown earlier (Figure 2). genomes, concentrating our efforts on mucosal HPV types. We could not find equivalent sequences in any of the HPV regulatory regions, even when looking at the E2F5 activates the cell cycle in HPV18-associated cells HPV18 closest type, HPV45. We therefore decided to One important assumption of the modulation of check whether the E2F5-dependent positive regulation S-phase genes by E2F5 is that silencing of E2F5 should of the cell cycle, which we observed in HPV18-positive modulate the cell cycle in HeLa cells. To test this cells, was also seen in the HPV16-positive Caski cells. hypothesis, we analyzed the effects of the silencing We first studied the effect of the silencing of E7 and of E2F5 compared with the silencing of E2F activators E2F5 on HPV16 E6/E7 transcription. Although E2F5 and E7 in three cell lines: HeLa (HPV18), C33-A (HPV and E7 siRNAs very efficiently induced silencing of their negative) and Caski (HPV16) cells by flow cytometry respective targets, silencing of E2F5 did not change the (Figures 4 and 5). In the presence of the siE7, HeLa level of endogenous E7 expression in Caski (Figure 5a). cells did not release from G1/S block compared with Cell cycle analyses confirmed that E2F5 did not function control cells (Figure 4a). Inactivation of E2F5 in similar as a positive factor for S-phase entry in this cell line, as experiments also showed a block in G1, although the silencing of E2F5 did not change the cell-cycle release less efficient than with E7 silencing, as only 54% of (Figure 5b), contrary to that seen in HeLa cells (Figure 4a). the cells were blocked in G1 after 6 h of release, In this experiment, the behavior of HPV16-associated compared with 85% for shE7 and 32% for the control. Caski cells was, therefore, comparable to that of HPV- This observation is consistent with the results obtained negative C33 (compare Figures 5b and 4b). These data by RT–PCR, indicating that E2F5 inactivation has a thus confirmed that E2F5 could function as an activator weaker effect on the S-phase genes than E7 silencing of the cell cycle and entry in S phase only in HPV18- itself (Figure 3b, compare the transcription of cyclin E associated cervical carcinoma and, therefore, correlate in the presence of pSUPER-E7-HPV18 (60% inhibition) with the specifically aggressive behavior of this viral type and that in the presence of pSUPER-E2F5 (30% in natural infection.

Oncogene E2F5 activates HPV18 E6/E7 transcription S Teissier et al 5066

Figure 4 E2F5 modulates the cell cycle in HeLa but not in C33 cells. Flow cytometry analyses in synchronized transfected cells that were arrested by a thymidine block and released for 6 h. (a) HeLa cells transfected with siRNA E7 do not release and remain arrested in G1 as expected, whereas siRNA E2F5 show a delay in the release when compared with the control siRNA: fluorescence-activated cell sorting analyses (left panel) are quantified in the right panel (b). Growth curves of HeLa cells transfected with different pSUPER constructs as indicated on the graph. (c) No effect of the E7 and E2F5 siRNA on the cell cycle of C33-A cells. Analyses were carried out as described in (a).

A model of regulation for S-phase genes by E2F5 is responsible for the cell-cycle arrest preceding differ- is depicted in Figure 6, showing the HPV18-specific entiation (Apostolova et al., 2002). In HPV18-asso- E2F5-mediated activation of S-phase genes through an ciated cells, we show that the E2F5 repressor does not amplification loop of E7 activation and re-routing of the function as a repressor of the cell cycle but rather as an E2F5 from a repressor to a transcriptional activator. activator, and furthermore, that this is specific for the cervical carcinoma associated with HPV18 and not with HPV16, for instance. Activation of S-phase genes by E2F5 in HeLa cells, such as cyclin E, was also clearly Discussion shown in silencing experiments. However, these experi- ments also indicated that cyclin E remained under the E2F5 is one of the members of the E2F family and has control of the E2F activators and that there were no been shown to have a crucial role in keratinocytes, as it additive effects on its repression when the E2F5 shRNA

Oncogene E2F5 activates HPV18 E6/E7 transcription S Teissier et al 5067

Figure 5 E2F5 does not modulate the cell cycle in Caski cells. (a) Real-time RT–PCR in the HPV16-associated Caski cells indicate that the silencing of E2F5 has no effect on E6/E7 transcription. (b) Flow cytometry analyses were carried out as described in Figure 4.

the HPV18 oncogenes. The effects of the silencing of E7 transcription by E2F5 were extended to C-41 cells and to nTERT transfected with the HPV18 episomal genome. We could deduce from our experiments that E2F5 activates the cell cycle in HPV18-positive cells by directly activating transcription of E6 and E7, which in turn activate E2F activity by degrading p53 and pRb. This point is reinforced by the cell proliferation assays indicating that silencing of E2F5 arrests cell growth with a delay of few hours compared with silencing of the E2F activators (E2F1–3). The specificity of the transcriptional activation of E6/ E7 by E2F5 in HPV18-positive cells is correlated with the presence of E2F binding sites only in the HPV18 LCR and their efficient binding to E2F in vitro and in vivo. In addition, mutation of these sites in a luciferase reporter assay clearly shows their functional importance in the regulation of HPV18 transcription. In contrast, no E2F binding sites could be found in the HPV16 regulatory region and consequently, E2F5 silencing in Figure 6 Model for the E2F-mediated control of the cell cycle in Caski cells associated with HPV16 affected neither the keratinocytes expressing HPV18. In normal keratinocytes, the E6/E7 transcription nor the cell cycle. As E2F5 has been S-phase genes are under the control of the E2F pathway and they shown to have a key role in keratinocyte differentiation are activated by the E2F activators, whereas they are repressed by (Apostolova et al., 2002), we can speculate that in E2F5. In HPV18-associated keratinocytes, the S-phase genes are controlled by E7 through negative regulation of pRB inducing the HPV18-infected keratinocytes, activation of E6/E7 E2F activators and through direct activation of E7 transcription by transcription would be concomitant with induction of E2F5, which, by both activating E7 and diverting the E2F the differentiation through E2F5, therefore inducing a repressor, leads to amplification of the S-phase activating signal. delay in cell differentiation. The mechanism by which E2F5, classified as a repressor, leads to activation of the HPV18 transcription was added to the shRNA activators. This observation remains to be elucidated. Nevertheless, we can hypo- indicated that probably the S-phase genes are not thesize that this property is carried by the enhancer of directly activated by E2F5, and this hypothesis is HPV18 itself. The localization of the E2F binding sites sustained by our finding that E2F5 is an activator of in an enhanceosome structure is unusual, as they are

Oncogene E2F5 activates HPV18 E6/E7 transcription S Teissier et al 5068 more often found associated with promoter structures Materials and methods (Elkon et al., 2003; Linhart et al., 2005), and it might be a clue to explain the switch of a transcriptional Cell culture and transfections repressor to an activator. The regulation of E2F5 is HeLa, C33-A, C-41 and Caski cells were grown using standard 1 linked to the factors with which it interacts. Indeed, procedures in a 37 C humidified incubator with 5% CO2. All E2F5 contains a nuclear export signal but no nuclear cell lines were grown in high-glucose Dulbecco’s modified Eagle’s medium (Invitrogen, Singapore) with 10% heat-inactivated fetal localization signal and, therefore, needs to bind to bovine serum. nTERT cells were cultivated in keratinocyte-SFM another factor to enter the nucleus. One of the best medium (Invitrogen, 13944) complemented with Bovine pituitary known interactors of E2F5 is the pocket protein p130, extract (25 mg/ml), epithelial growth factor (0.2 ng/ml) and which has been shown to be crucial for its function CaCl2 (0.3 mM). as a transcriptional repressor although, and despite the The cells were prepared the day before the transfection at fact that p130 contains a nuclear localization signal, 500 000 cells per 10-cm dish. For plasmid transfection, we used it does not seem responsible for E2F5 nuclear localization Lipofectamin LTX (Invitrogen) according to the manufacturer’s (Caracciolo et al., 2007). instructions. shRNA-expressing vectors were pSUPER plasmids In an attempt to study direct activation by E2F5, (Oligoengine, Seattle, WA, USA). In HeLa, C-41 and C33 cells, we overexpressed untagged or GFP–E2F5 in HeLa or the control pSUPER plasmid used was pSUPER-E7-HPV16. The shRNA sequences cloned in pSUPER vectors are: in cotransfected C33-A cells with various luciferase pSUPER-E7-16: 50-GATCCCCGAGCTGCAAACAACT reporters, but could not detect any functional activity. ATACTTCAAGAGAGAGCUGCAAACAACUAUACTTT In addition, E2F5 overexpression did not modulate TTGGAAA-30; pSUPER-E7-18: 50-GATCCCCGATGGAGT E2F target gene transcription, cell cycle or cell growth TAATCATCAACTTCAAGAGAGTTGATGATTAACTCC (data not shown). Cellular localization of GFP–E2F5 ATCTTTTTGGAAA-30; pSUPER-E2F1: 50-GATCCCCGCA was mainly cytoplasmic, explaining, at least in part, its GAAGCGGCGCATCTATTTCAAGAGAATAGATGCGCC transcriptional defect (data not shown). To overcome GCTTCTGCTTTTTGGAAA-30; pSUPER-E2F2: 50-GATCC this problem, we overexpressed an E2F5 containing an CCGGGCCCATCGAAGTCTACCTTCAAGAGAGGTAGA CTTCGATGGGCCCTTTTTGGAAA-30; pSUPER-E2F3: SV40 nuclear localization signal to increase its nuclear 0 concentration, but that did not increase its functionality. 5 -GATCCCCGCGATTGCTCAGTTTCTATTTCAAGAGAA- TAGAAACTGAGCAATCGCTTTTTGGAAA-30; pSUPER- We deduce from these data that to observe an increase E2F5: 50-GATCCCCGCACCTTCTGGTACACAACTTCA in its functions, we probably should overexpress E2F5 AGAGAGTTGTGTACCAGAAGGTGCTTTTTGGAAA-30. together with its regulators, the difficulty being that they Caski cells were transfected with siRNA using Dharmafect are not all known, including the one responsible for (Thermo Scientific, Research Instruments, Singapore) accord- nuclear localization. ing to the manufacturer’s instructions. The control used in Another E2F repressor, E2F6, has recently been these HPV16-positive cells was si-E7-HPV18. shown to interact with HPV16-E7 to induce a ‘stem-cell- siE7-HPV18: 50-GAUGGAGUUAAUCAUCAACdTdT-30; like state’ of infected cells (McLaughlin-Drubin et al., siE7-HPV16: 50-GAGCUGCAAACAACUAUACdTdT-30; siE2F1: 50-GCAGAAGCGGCGCAUCUAUdTdT-30; siE2F2: 2008), although we could not show such direct interac- 0 0 0 tion between HPV18-E7 and E2F5. In addition, we were 5 -GGGCCCAUCGAAGUCUACCdTdT-3 ; siE2F3: 5 -GCGA UUGCUCAGUUUCUAUdTdT-30; siE2F5: 50-GCACCUUCU unable to show any direct effect of E7 expression on the GGUACACAACdTdT-30. E2F5 transcriptional regulation, indicating that E7 itself The nTERT cells had been electroporated with NEON does not induce the transcriptional switch. It has been system (Invitrogen) as per the manufacturer’s instructions. shown that the high- and low-risk HPV-E7 oncogenes The RT–PCR experiments performed in HeLa with can target p130 for degradation, thus leading to relief of pSUPER-E2F5 were also carried out using an siE2F5 pool the E2F5 transcriptional repression (Zhang et al., 2006). provided by ON-TARGETplus SMARTpool from Thermo As this repression is required to arrest the cell cycle Scientific (50-GGUGCUGGCUGUAAUACUA-30;50-UAAA before keratinocyte differentiation, it seems that in both UAAAGAGUCGAGUUC-30;50-GAGGUACCCAUUCCAG 0 0 0 high- and low-risk HPV infection, this delay in cell AAA-3 ;5-UCAGCAGGAUCUAUUAGUG-3 ) to ensure that differentiation is required for the viral cycle to proceed. the results obtained were not because of a potential off-target effect. Luciferase assay was carried out by transfecting C33 cells with We have found no obvious implication of either p130 5 mg of HPV18 LCR full-length cloned in pTL vector (Panomics, or E7 itself in the HPV18 transcriptional regulation Fremont, CA, USA). The luciferase activity was normalized by by E2F5 (not shown), which is consistent with the the Renilla activity carried by 0.5 mg of control vector pRL-TK observation that this mechanism is specific for HPV18, (Promega Pte Limited, Singapore). Point mutations are 50-ATT thus not involving common regulatory mechanisms. GATGC-30 for site A and 50-TTTGATGC-30 for site B. This would imply that in HPV18-infected lesions, the regulation of E2F5 could occur by two different Luciferase assay pathways, one common pathway involving degrada- Luciferase assay was carried out using a Promega kit (Cat no. tion of p130 by E7 and one specific pathway involving E1960) according to the manufacturer’s instructions. the use of E2F5 as a transcriptional activator of the Gel-shift assay experiments E7 transcription. We believe that this original mecha- 32 DNA probes were labeled with 50 nM alpha P-dATP for 30 min nism of re-routing a transcription factor has a role in with Klenow enzyme (Roche, Singapore) in the provided buffer the high level of transcription of the viral oncogenes containing 50 nM of dCTP, dTTP and dGTP. Klenow enzyme in HPV18 and possibly in its higher transforming was inactivated by heating for 10 min at 70 1C and the labeled capabilities. probes were purified on columns (Amersham Bioscience, Uppsala,

Oncogene E2F5 activates HPV18 E6/E7 transcription S Teissier et al 5069 Sweden). Sequences of the E2F binding sites in the probes are Chromatin immunoprecipitation (underlined sequences are overhang for Klenow filling) as follows: Chromatin was purified and digested using the ChIP-IT kit E2F consensus probe: 50-CTAGATGTTTAGGCGCGAA (Active Motif, Carlsbad, CA, USA) according to the manu- AACTTCTAG-30; HPV18 E2F site A: 50-CTAGAAACAATT facturer’s instructions. Pre-clearing was carried out for 1 h GGCGCGCCTTCTAG-30; HPV18 E2F site B: 50-CTAGACC on a rotating wheel using protein A/G agarose (Santa Cruz TCTTTGGCGCATATTCTAG-30. Biotechnology). The chromatin was then incubated over- Nuclear extracts were prepared as previously described night at 4 1C with the rabbit polyclonal anti-GFP antibody (Bouallaga et al., 2003) and were incubated for 30 min at room (Torrey Pines Biolabs, East Orange, NJ, USA, TP401). The temperature in 12 mM Hepes, 0.5 mM EDTA, 4 mM MgCl2, protein A/G agarose mixture was added and incubated during 60 mM KCl, 0.1% NP40, 10% glycerol and 4 mM DTT, 3 mM 1 h at 4 1C. Beads were centrifuged and washed three times spermidine, 0.5 mg/ml bovine serum albumin and 50 mg/ml with Mix1 and were resuspended in 200 ml of Mix2 and Salmon Sperm DNA. The labeled probes were then added and incubated overnight at 65 1C. The mixture was then digested further incubated for 10 min before migration in a 5% native by 0.5 mg/ml proteinase K at 42 1C for 2 h. Immunoprecipi- polyacrylamide 29/1 gel in 0.5% TBE. When antibodies were tated DNA was purified by phenol–chloroform (50–50%) used, 1 ml of anti-E2F1, E2F2 or E2F5 rabbit polyclonal anti- extraction and ethanol precipitation. bodies from Santa Cruz Biotechnology (Santa Cruz, CA, USA) (SC-193, SC-633 and SC-999) was added to the mixture. Western blot After migration, the gels were dried and revealed using a Immunoblotting experiments for the detection of HPV18 E7 Phosphor-Imager. GST–E2F1 had been purified from bacteria were performed with an anti-HPV18 E7 antibody provided by as described previously (Heino et al., 2000). Dr Zentgraf and Dr Hoppe-Seyler as described (Kuner et al., 2007); anti-actin rabbit antibody was from Sigma (St Louis, Real-time RT–PCR MO, USA) (A-2066). The RNA was extracted using RNeasy Mini Kit (Qiagen, Singapore) according to the manufacturer’s instructions. A Flow cytometry volume of 2.5 mg of total RNA measured with nanodrop ND- The cell cycle of HeLa, Caski and C33 cells was arrested 24 h 1000 was reverse transcribed with superscript II (Invitrogen) after transfection with the various shRNAs by adding 2.5 mM according to the manufacturer’s instructions. Real-time PCR thymidine to the medium for 24 h. T cells were then harvested was carried out using the Stratagene MX3005p machine 0 by trypsinization at the release, whereas T cells were harvested (Stratagene, Singapore) in 25 ml of mixture containing 25 ng of 6 6 h after changing the medium to release the thymidine block. cDNA, 2.5 Â 10À5 mM of each primer and 12.5 mlofSybrGreen Cells were fixed in 70% ethanol and DNA was labeled with PCR master mixture (Applied Biosystems Asia Pte Ltd, 20 mg/ml propidium iodide and 10 mg/ml RNase. Cell-cycle Singapore). The thermal profile used was 50 1Cfor2min,951C analyses were carried out using an LSR-II flow cytometer (BD for 10 min, 95 1C for 15 s, 60 1C for 1 min, over 40 cycles, and the Biosciences, San Jose, CA, USA). dissociation curves were obtained by measuring the fluorescence from 60 to 95 1C. Primers used were as follows: Cell proliferation analyses E2F1: 50-AAGGCCCGATCGATGTTTTC-30;50-TGATC HeLa cells were reverse transfected and seeded on E-plates 96 CCACCTACGGTCTCCT-30 E2F2: 50-GAGGACAAGGCCA (Roche). Cell growth was measured with the xCELLigence ACAAGAGG-30;50-TTGCCAACAGCACGGATATC-30 E2F3: system (Roche); the cell density is given by the cell index, an 50-GAATATCCCTAAACCCGCTTCC-30;50-TGTCCTGAGT arbitrary unit indicating the percentage of the well occupied by TGGTTGAAGCC-30;E2F4:50-AGCACAGCCAGTTTCAAC- cells every hour during 88 h. The growth curves were norma- GAC-30;50-TGATAGCCCCCAAACCCAG-30; E2F5: 50-ACCA lized 27 h after the transfection. AGTTCGTGTCGCTGC-30;50-TGAGATCCAGAACGCCGT C-30;E2F6:50-ATGTTCCAGCTCCCAGAGAAGAC-30;50-TG 0 0 GTGCTCCTTATGTGCACTG-3 16E7: 5 -GCTCAGAGGAG Conflict of interest GAGGATGAAATAG-30;50-TCCGGTTCTGCTTGTCCAG- 0 0 0 0 3 ;18E7:5-CCCCAAAATGAAATTCCGGT-3 ;5-GTCGC The authors declare no conflict of interest. TTAATTGCTCGTGACATA-30; Cyclin E1: 50-TGACCGTTT TTTTGCAGGATC-30;50-TCCTGTCGATTTTGGCCATT-30; LCR-HPV18: 50-TCGGTTGCCTTTGGCTTATG-30;50-GCCA GCGTACTGTATTGTGCAG-30; PolA promoter: 50-TCCTCC Acknowledgements GGCTTCCCCTC-30;50-GCCCTGTGATGACCCCTTAG-30; GADPH promoter: 50-GTCTGCCCTAATTATCAGGTCCAG-30; We thank Dr Felix Hoppe-Seyler and Professor Hanswalter 50-AGGTCACGATGTCCTGCAGC-30. Zentgraf, who provided us with the HPV18 E7 antibody. The data were analyzed using the MxPro v.4.00 software We thank Dr Sardet for providing us the E2F5-expressing (Stratagene). vector.

References

Apostolova MD, Ivanova IA, Dagnino C, D’Souza SJ, Dagnino L. Bouallaga I, Massicard S, Yaniv M, Thierry F. (2000). An enhanceo- (2002). Active nuclear import and export pathways regulate E2F-5 some containing the JunB/Fra2 heterodimer and the HMGI(Y) subcellular localization. J Biol Chem 277: 34471–34479. architectural protein controls HPV18 transcription. EMBO Rep 1: Bernard BA, Bailly C, Lenoir MC, Darmon M, Thierry F, Yaniv M. 422–427. (1989). The human papillomavirus type 18 (HPV18) E2 gene Bouallaga I, Teissier S, Yaniv M, Thierry F. (2003). HMGI(Y) and the product is a repressor of the HPV18 regulatory region in human CBP/P300 coactivator are essential for the HPV18 enhanceosome keratinocytes. J Virol 63: 4317–4324. transcriptional activity. Mol Cell Biol 23: 2329–2340.

Oncogene E2F5 activates HPV18 E6/E7 transcription S Teissier et al 5070 Caracciolo V, Reiss K, Crozier-Fitzgerald C, De Pascali F, Macaluso Reed SA, Ouellette SE, Liu X, Allen RE, Johnson SE. (2007). M, Khalili K et al. (2007). Interplay between the retinoblastoma E2F5 and LEK1 translocation to the nucleus is an early related pRb2/p130 and E2F-4 and -5 in relation to JCV-TAg. J Cell event demarcating myoblast quiescence. J Cell Biochem 101: Physiol 212: 96–104. 1394–1408. Castellsague X, De Sanjose S, Aguado T, Louie KS, Bruni L, Munoz J et al. Ren B, Cam H, Takahashi Y, Volkert T, Terragni J, Young RA (2007). HPV and cervical cancer in the World, 2007 report. Vaccine 25. et al. (2002). E2F integrates cell cycle progression with DNA Dyson N, Howley PM, Mu¨nger K, Harlow E. (1989). The human repair, replication, and G(2)/M checkpoints. Genes Dev 16: papillomavirus-16 E7 oncoprotein is able to bind to the retino- 245–256. blastoma gene product. Science 243: 934–937. Sardet C, Vidal M, Cobrinik D, Geng Y, Onufryk C, Chen A et al. Elkon R, Linhart C, Sharan R, Shamir R, Shiloh Y. (2003). Genome- (1995). E2F-4 and E2F-5, two members of the E2F family, are wide in silico identification of transcriptional regulators controlling expressed in the early phases of the cell cycle. Proc Natl Acad Sci the cell cycle in human cells. Genome Res 13: 773–780. USA 92: 2403–2407. Grinstein E, Wernet P, Snijders PJ, Rosl F, Weinert I, Jia W et al. (2002). Scheffner M, Huibregtse JM, Vierstra RD, Howley PM. (1993). The Nucleolin as activator of human papillomavirus type 18 oncogene HPV16 E6 and E6-AP complex functions as ubiquitin-protein ligase transcription in cervical cancer. J Exp Med 196: 1067–1078. in the ubiquitination of p53. Cell 75: 495–505. Heino P, Zhou J, Lambert P. (2000). Interaction of the papillomavirus Teissier S, Ben Khalifa Y, Mori M, Pautier P, Desaintes C, Thierry F. transcription/replication factor E2 and the viral capsid protein L2. (2007). A new E6/P63 pathway, together with a strong E7/E2F Virology 276: 304–314. mitotic pathway, modulates the transcriptome in cervical cancer Ishida S, Huang E, Zuzan H, Spang R, Leone G, West M et al. (2001). cells. J Virol 81: 9368–9376. Role for E2F in control of both DNA replication and mitotic Thierry F. (2009). Transcriptional regulation of the papillomavirus functions as revealed from DNA microarray analysis. Mol Cell Biol oncogenes by cellular and viral transcription factors in cervical 21: 4684–4699. carcinoma. Virology 384: 375–379. Johnson DG, Schwarz JK, Cress WD, Nevins JR. (1993). Expression Thierry F, Benotmane MA, Demeret C, Mori M, Teissier S, Desaintes of transcription factor E2F1 induces quiescent cells to enter S phase. C. (2004). A genomic approach reveals a novel mitotic pathway in Nature 365: 349–352. papillomavirus carcinogenesis. Cancer Res 64: 895–903. Kong LJ, Chang JT, Bild AH, Nevins JR. (2007). Compensation and Thierry F, Howley PM. (1991). Functional analysis of E2-mediated specificity of function within the E2F family. Oncogene 26: 321–327. repression of the HPV18 P105 promoter. New Biol 3: 90–100. Kuner R, Vogt M, Sultmann H, Buness A, Dymalla S, Bulkescher J Thierry F, Yaniv M. (1987). The BPV1-E2 trans-acting protein can be et al. (2007). Identification of cellular targets for the human either an activator or a repressor of the HPV18 regulatory region. papillomavirus E6 and E7 oncogenes by RNA interference and EMBO J 6: 3391–3397. transcriptome analyses. J Mol Med 85: 1253–1262. Villa LL, Schlegel R. (1991). Differences in transformation activity Linhart C, Elkon R, Shiloh Y, Shamir R. (2005). Deciphering between HPV-18 and HPV-16 map to the viral LCR-E6-E7 region. transcriptional regulatory elements that encode specific cell cycle Virology 181: 374–377. phasing by comparative genomics analysis. Cell Cycle 4: 1788–1797. Walboomers JM, Jacobs MV, Manos MM, Bosch FX, McLaughlin-Drubin ME, Huh KW, Munger K. (2008). Human Kummer JA, Shah KV et al. (1999). Human papillomavirus is papillomavirus type 16 E7 oncoprotein associates with E2F6. a necessary cause of invasive cervical cancer worldwide. J Pathol J Virol 82: 8695–8705. 189: 12–19. Mudryj M, Devoto SH, Hiebert SW, Hunter T, Pines J, Nevins JR. Zhang B, Chen W, Roman A. (2006). The E7 proteins of low- and (1991). Cell cycle regulation of the E2F transcription factor involves high-risk human papillomaviruses share the ability to target the an interaction with cyclin A. Cell 65: 1243–1253. pRB family member p130 for degradation. Proc Natl Acad Sci USA Polager S, Ginsberg D. (2003). E2F mediates sustained G2 arrest and 103: 437–442. down-regulation of Stathmin and AIM-1 expression in response to Zhu W, Giangrande PH, Nevins JR. (2004). E2Fs link the control of genotoxic stress. J Biol Chem 278: 1443–1449. G1/S and G2/M transcription. EMBO J 23: 4615–4626.

Oncogene