Keratinocyte sensitization to tumour necrosis factor-induced nuclear factor kappa B activation by the E2 regulatory protein of human papillomaviruses

Human papillomavirus (HPV) infect squamous epithelia of the skin (cutaneous HPV) or mucosa of the genital or oral tractus (mucosal HPV). They are responsible for asymptomatic infections or induce benign proliferative lesions, which possibly evolve toward malignant progression only for ‘high risk’ HPV (HR-HPV).

Infection initiates in the basal proliferative cells where the E6 and E7 early proteins interfere with cell cycle controls, thereby inducing cell proliferation in suprabasal layers. High levels of viral DNA replication, synthesis of capsid proteins and virus particles assembly are confined to suprabasal and terminally differentiated keratinocytes (Doorbar, 2006). The early E2 protein binds to specific sites within the viral genome to ensure its mitotic segregation in basal-infected cells, as well as the regulation of viral transcription and replication (Desaintes & Demeret, 1996; McBride et al., 2006). In addition, the E2 proteins of mucosal HR-HPV directly impact on cell proliferation by inducing a G2/M arrest (Bellanger et al., 2005) and apoptosis (Thierry & Demeret, 2008).

Upon viral infection, the host initiates a cell response mediated by nuclear factor kappa B (NF-κB). This transcription factor leads to the expression of a large panel of cytokines as well as cell cycle and survival genes. Accordingly, many viral pathogens developed strategies to modulate NF-κB signalling in order to enhance viral replication, host cell survival and evasion to immune responses (Hiscott et al., 2001). Surprisingly, little is known about the regulation of NF-κB by HPV. Viral oncoproteins have been reported to induce either attenuation or enhancement of NF-κB signalling depending on the cellular context (James et al., 2006; Nair et al., 2003; Nees et al., 2001; Spitkovsky et al., 2002), and a clear picture of the role of NF-κB in the natural history of HPV infection and HR-HPV-associated carcinogenic progression is lacking.

To address this issue, we analysed the impact on NF-κB signalling of the regulatory E2 proteins from cutaneous HPV of α (HPV3), β (HPV5, 8 and 9) or μ (HPV1) types, and mucosal α types low-risk (LR-HPV6 and 11) or HR (HPV16, 18, 33 and 39). NF-κB activation was monitored through the activity of a luciferase reporter gene under the control of NF-κB in HaCaT keratinocytes, with an RNA polIII-directed Renilla reporter used as internal control. NF-κB activity was measured either without or with tumour necrosis factor (TNF) treatment (20 ng ml⁻¹) for 20 h, as it plays a central role in response to viral infection. The E2 proteins were expressed as GFP-fusion proteins that have been shown to keep functions, stability and subcellular distribution of the untagged E2 polypeptides (Bellanger et al., 2001; Blachon et al., 2005). All GFP–E2 proteins enhanced TNF-induced NF-κB activation when compared with control GFP-expressing cells (Fig. 1a). Of note, the E2 proteins from cutaneous β or μ-HPV (HPV1, 5, 8 and 9) activated NF-κB reporter to the highest levels. In contrast, E2 proteins had no significant effect on NF-κB reporter activity without TNF, indicating that E2 did not induce constitutive NF-κB signalling (Supplementary Fig. S1, available in JGV Online).

The untagged version of E2 from HPV5, 11 and 18 enhanced TNF-induced activation of the NF-κB reporter similarly to their GFP fusion counterparts, indicating that the GFP tag does not exaggerate the observed NF-κB activation (Fig. 1b, left panel). The expression of untagged E2 proteins was monitored through activation of a synthetic E2-responsive promoter (Thierry et al., 1990) (Fig. 1b, right panel). Overall, these results establish that the E2 proteins of various HPV potentiate TNF-induced NF-κB activation.

We next determined whether NF-κB activation could be assigned to a specific domain of the E2 protein. The N-terminal domain, central Hinge and C-terminal domains of HPV5, 11 and 18 E2 proteins were expressed separately in fusion with GFP, and their effect on the activation of NF-κB reporter was analysed. For the three E2 proteins, expression of the N-terminal domain enhanced TNF-induced NF-κB activation, whereas the Hinge and C-terminal domains had no effect (Fig. 1c). The transactivation-dead point mutant I77A of HPV18 E2 (Desaintes et al., 1999) enhanced TNF-induced NF-κB activation with an efficiency similar to that of the wild-type protein, indicating that the N-terminal domain of E2 mediates NF-κB activation independently of its transcriptional activity (Fig. 1d).

The E2 proteins from HPV16 and 18 have been shown to interact with Tax1BP1 (Wang et al., 2009b), which is a known regulator of NF-κB signalling (Journo et al., 2009). Despite this, interaction could be detected with a whole range of E2 proteins (see below), silencing of Tax1BP1 through siRNA did not relieve HPV5 or 18 E2-mediated NF-κB activation (Fig. 2a). Tax1BP1-directed siRNA reduced the level of ectopically expressed Tax1BP1 (Fig. 2b), and lowered E2-dependent transcription for HPV16 or 18 E2, but not for HPV5 E2 (Supplementary Fig. S2, available in JGV Online). Together, these results suggest that enhancement of TNF-induced NF-κB activation by E2 in HaCaT keratinocytes is not mediated by E2 binding to endogenous Tax1BP1. However, this interaction affects transcriptional activation by some of the E2 proteins, as previously described for bovine papillomavirus 1 (BPV1) E2 (Wang et al., 2009b).

We then analysed whether E2 could stimulate other NF-κB signalling pathways, as those triggered by interleukin (IL)-1β. Expression of E2 had no effect on NF-κB activation upon IL-1β treatment (20 ng ml⁻¹ for 20 h), in contrast to TNF-induced NF-κB activation (Fig. 2c). Our results suggest that the E2 protein specifically activates TNF-induced NF-κB signalling.

NF-κB signalling involves different intermediates, such as TRAF6 for IL-1-induced pathways and TRAF2 or TRAF5 for TNF-induced pathways (Lomaga et al., 1999; Tada et al., 2001). The TRAF proteins are E3 ubiquitin ligases catalysing the non-degradative K63-linked polyubiquitination of NF-κB signalling factors, which plays a crucial role in recruiting and activating downstream factors (Chen & Sun, 2009). We examined TRAF ubiquitination patterns in the presence or absence of HPV18 E2 using HA-tagged ubiquitin with only one lysine available for polyubiquitin chain formation at position 29 (K29 ubiquitin), 48 (K48 ubiquitin) or 63 (K63 ubiquitin). As shown in Fig. 3(a), the polyubiquitination of TRAF5 and 6 was enhanced upon HPV18 E2 co-expression, whereas changes were minimal for TRAF2. The activating K63-linked ubiquitination of TRAF6 was detected in the absence of E2, in accordance with a previous report (Lamothe et al., 2007). It was not further increased upon E2 expression, in line with the lack of activation of TRAF6-dependent NF-κB pathways by E2. In the presence of E2, TRAF6 exhibited enhanced K29- and K48-polyubiquitination, which essentially target proteins for degradation (Xu et al., 2009). In contrast, for TRAF5, E2 expression enhanced the activating K63-polyubiquitination, concurrently with K29-polyubiquitination (Fig. 3), suggesting that E2 enhanced TRAF5 activation.

We next examined whether E2 could act through direct binding to the TRAF proteins by using the SPICA assay (Sensing of Protein–protein Interactions by Complementation Assay). This assay is based on the reconstitution of Gaussia princeps luciferase activity upon co-expression of interacting proteins in fusion with two inactive fragments of the enzyme (P. Cassonnet and others, unpublished data) Briefly, reconstituted Gaussia activity is expressed as a ratio of luminescence (RL). RL is given by the Gaussia luciferase activity obtained upon co-expression of protein pairs divided by the sum of luciferase activity of each single protein expressed with matched empty vector. Using this assay, the binding to Tax1BP1 was detected with all E2 proteins. In addition, the E2 proteins were found to interact with both TRAF5 and 6, but not with TRAF2 (Fig. 3b). Such binding potentially underlies the effect of E2 on the ubiquitination pattern of TRAF5 and 6. Together, our findings suggest that E2 potentiates a TNF-induced NF-κB signalling pathway mediated by TRAF5 through direct interaction.

In this study, we present evidence that the HPV E2 protein stimulates NF-κB activity upon TNF induction through its N-terminal domain, but independently of its transcriptional functions. The E2 protein functions primarily as a regulator of viral genome transcription, replication and segregation through binding to the viral genome. Our study indicates that E2 activates the transcription of NF-κB target genes upon TNF induction independently of its transcriptional functions. In addition, this activity is independent of the Tax1BP1 protein, although this factor is involved in the regulation of NF-κB signalling (Journo et al., 2009; Shembade et al., 2007) and binds to E2 (Wang et al., 2009b; and this study). This interaction has been shown to take part in the transcriptional activation properties of BPV1 E2 (Wang et al., 2009b), which was observed here for HPV16 and 18 E2 proteins.

NF-κB activation is conserved throughout 11 HPV genotypes representative of diverse HPV tropism and pathogenicity, indicating that it is a basic process of HPV infection. The activation of NF-κB by TNF has a unique role in the control of epithelial differentiation. TNF induces both an arrest in cell proliferation and the activation of keratinocyte differentiation markers (Basile et al., 2001, 2003), thereby being an essential factor in inducing epithelial differentiation. In the course of the viral cycle, the sensitization to TNF-induced NF-κB activation by E2 may be required to activate the differentiation programme of infected keratinocytes, which is necessary for the late phase of the viral cycle. The E2 protein is known to alter the transcription of differentiation markers through co-operation with cellular transcription factors as C/EBP (Hadaschik et al., 2003) or through direct binding to cellular promoter (Oldak et al., 2004). Our study indicates that E2 possibly stimulates keratinocyte differentiation through activation of NF-κB target genes. During the viral cycle, E6 and E7 expression prevent cell cycle exit and differentiation of infected keratinocytes. The E2 protein would overcome these effects by inducing the transcription of keratinocyte differentiation genes through multiple mechanisms, including sensitization to TNF-induced NF-κB activation.

Since TNF induces a cell cycle arrest in keratinocytes, E2 activity would also provide adequate conditions for high levels of viral DNA replication, which has been shown to occur in pseudo G2-arrested cells in the suprabasal layer of infected epithelia (Wang et al., 2009a). Of note, the β and μ-skin genotypes produce lesions with high virus content when compared with lesions associated with α genital HPV. The high level of NF-κB activation by E2 of cutaneous HPV would thus correlate with viral yield, in line with its role in viral DNA replication.

Several lines of evidence point to a role of E2 in the carcinogenic conversion associated with HR-HPV. In case of HR-mucosal HPV, E2 induces a G₂/M arrest through inactivation of the anaphase promoting complex (Bellanger et al., 2005), inducing genomic instability, thus favouring cell immortalization. In addition, E2 from HR cutaneous HPV8 exhibits intrinsic oncogenic potential when expressed in the skin of transgenic mice (Pfefferle et al., 2008). E2-induced skin-tumours are associated with impaired wound-healing and chronic inflammation (Pfefferle et al., 2008). Since NF-κB signalling is a major mediator of inflammation, sustained activation of NF-κB by E2 could have a role in such chronic inflammation and thus participate in HPV-associated carcinogenesis. This could add to E6 and E7 immortalizing properties during natural infection.

The TRAF2, 5 and 6 proteins are essential mediators of NF-κB signalling. We found that E2 stimulates one or both of the K29- or K48-forms of polyubiquitination of TRAF5 and 6, which target proteins to degradation (Xu et al., 2009). E2 concurrently stimulates the activating K63-ubiquitination of TRAF5, potentially underlying the enhancement of TNF-induced NF-κB activation by E2. In addition, E2 binds to both TRAF5 and 6, providing evidence that E2 could alter their ubiquitination pattern through direct interaction. It is unclear at present how E2 stimulates the polyubiquitination of TRAF5 and 6. It can be envisaged that E2 binding to the TRAF proteins could facilitate association with ubiquitin-activating enzymes or ubiquitin ligases, or could increase accessibility of the target lysines through conformational changes. Our data nevertheless suggest that E2 potentiates a signalling pathway mediated by TRAF5 through direct interaction.

In conclusion, we provide evidence that the E2 regulatory protein of HPVs manipulates specific cell signalling, as shown here with the activation of a TNF-induced NF-κB pathway mediated by TRAF5. This effect is probably involved in the commitment of infected keratinocytes to arrest in proliferation and differentiation, both required for the implementation of the productive viral cycle within stratified epithelia. Given that cancers arise from initial pre-cancerous productive lesions associated with HR-HPV, such activities are relevant for the carcinogenic progression of HR-HPV-associated lesions.