The E7 protein from human papillomavirus type 16 enhances keratinocyte migration in an Akt-dependent manner

Abstract

Cyclin-dependent kinase inhibitor p27^kip1 (p27) has recently been implicated as a positive regulator of cellular motility and is a marker of poor prognosis in several forms of cancer when localized to the cytoplasm. Cytoplasmic p27 exerts its effect on migration by binding to and inhibiting the activation of the small GTPase and cytoskeletal organizer RhoA, consequentially loosening cell substrate grip and enhancing movement. Using DNA damage as a p27 nuclear import signal, we found that the E7 oncoprotein from human papillomavirus type 16 (HPV-16), the etiological agent of cervical cancer, enhanced both the cytoplasmic retention of p27 and the migration of human foreskin keratinocytes (HFKs) in a phosphoinositide-3 kinase (PI3K)/Akt-dependent manner using a standard wound assay. Increased migration in E7-expressing HFKs correlated with an Akt-regulated down-regulation of RhoA activity through p27 binding under conditions where a p27 nuclear import signal is given (that is, DNA damage). Under these conditions, inhibition of the downstream RhoA effector ROCK enhanced control cell migration, whereas relatively unaffecting E7-expressing cells, further implicating that the inhibitory effect of E7 on RhoA positively regulates migration. We believe that the E7 protein from HPV-16 can modulate the cytoplasmic localization of p27 and may in turn regulate tumor metastasis/aggressiveness through the PI3K/Akt pathway.

Human papillomavirus (HPV) is a double-stranded DNA tumor virus associated with 99% of cervical cancer cases worldwide (Walboomers et al., 1999) and is the second leading cause of cancer in women worldwide (Baseman and Koutsky, 2005). Oncogenic transformation is believed to be triggered in part by the integration of the viral genome into a host chromosome, leading to unregulated expression of the oncogenes E6 and E7 (Hudelist et al., 2004). E6 degrades p53, whereas E7 is involved in deregulation of the Rb/E2F pathway (reviewed by Doorbar, 2006). Current literature indicates that the E7 protein exerts a more dominant role in transformation than E6 (Fiedler et al., 2004), although both oncoproteins contribute to malignant progression.

Recently, there has been a growing body of evidence supporting the tumor suppressor cyclin-dependent kinase inhibitor p27^kip1 (p27) as a positive regulator of migration and tumor invasion when located in the cytoplasm of cells (Watanabe et al., 2002; McAllister et al., 2003; Besson et al., 2004; Vasko et al., 2004; Li et al., 2006). When mislocalized to the cytoplasm, p27 binds to the small GTPase RhoA and inhibits its association with Rho-GEFs, leading to a loss of RhoA activity and inhibition of its downstream effector Rho-associated coiled-coil domain containing protein kinase 1 (ROCK) (Besson et al., 2004). Attenuation of the RhoA pathway leads to a decrease in both actin stress fiber and focal adhesion formation, allowing for an increase in migration (Nobes and Hall, 1999). Reduction of p27 expression has recently been directly linked to decreased invasiveness of tumor cells injected into mice (Wu et al., 2006a), implicating p27 directly as a modulator of invasive potential.

Interestingly, high levels of p27 have been shown to be a marker of poor prognosis in HPV-positive cervical cancers (Dellas et al., 1998; Shiozawa et al., 2001). The activity of the serine/threonine kinase Akt, a known modulator of p27 localization (Shin et al., 2002), is also increased in HPV-positive cervical lesions when compared to normal surrounding tissue and in E7-expressing human foreskin keratinocytes (HFKs) differentiated in organotypic raft culture (Menges et al., 2006). We have also demonstrated that E7 abrogates Raf-induced arrest through heightened Akt activity resulting in mislocalization of p21 upon arrest. Taken together with recent data implicating E7 as a direct modulator of invasiveness (Wu et al., 2006b), we postulate that E7 may increase the proportion of cytoplasmic p27 through enhancement of Akt and mislocalization of p27 could enhance keratinocyte migration by inhibition of the RhoA pathway.

E7 was previously implicated in p27 mislocalization upon transforming growth factor-β (TGF-β) treatment in mink lung epithelial cells (Mv1Lu; Westbrook et al., 2002), although the underlying mechanism was not established. First, we wished to determine if E7 can influence the cellular localization of p27 upon a nuclear import signal through its recently reported ability to maintain Akt activity (Menges et al., 2006). Mv1Lu cells expressing HPV-16 E7 in an internal ribosome entry site (IRES) with green fluorescent protein (GFP) under a doxycycline (DOX)- inducible promoter were serum-starved to induce p27 expression in the presence or absence of E7. Cells expressing E7 (DOX-treated, GFP-positive) under serum starvation conditions maintained a diffuse distribution of p27, similar to cells in serum, whereas untreated serum-starved cells (GFP-negative) contained mainly nuclear p27 (Figure 1a). Interestingly, LY294002 treatment negated the ability of E7 to mislocalize p27 upon serum withdrawal, indicating E7 may be regulating the phosphoinositide-3 kinase (PI3K) pathway to mislocalize p27. Although LY294002 treatment does lead to stabilization of p27 (Figure 1b), as reported previously (Motti et al., 2005), localization does appear to be mainly nuclear. A similar trend was also observed upon TGF-β treatment (data not shown).

Figure 1.

E7 from HPV-16 mislocalizes p27 in a PI3K-dependent manner in Mv1Lu cells. (a) Mv1lu cells expressing HPV-16 E7 under a doxycycline (DOX) inducible promoter in an IRES with GFP, were maintained as described by Westbrook et al. (2002). Cells were seeded on glass coverslips at equal density and were serum-starved for 24 h with or without 2 μg/ml DOX (Sigma-Aldrich, St Louis, MO, USA) and PI3K inhibitor 10 μm LY294002 (Cell Signaling Technologies, Inc., Danvers, MA, USA). Cells were fixed and stained with anti-p27 (Lab Vision Products, Fremont, CA, USA) and fluorophore-labeled secondary (Molecular Probes, Invitrogen Corporation, Carlsbad, CA, USA). (b) Western blot analysis of total p27 levels (Santa Cruz Biotechnology, Inc., Santa Cruz, CA, USA), HPV-16 E7 (Zymed, S San Francisco, CA, USA) and actin (Santa Cruz Biotechnology) expression under conditions described above. Lysates were prepared and analysed as described by Westbrook et al. (2002). GFP, green fluorescent protein; HPV-16, human papillomavirus type 16; p27, cyclin-dependent kinase inhibitor p27^kip1; PI3K, phosphoinositide-3 kinase.

Next we asked if this phenomenon is also observed in primary HFKs, cells tropic for HPV infection. E7 was expressed in HFKs using a retroviral packaging system in an integration competent vector (pBabe) with E6 and E7 in tandem, as in the viral genome, with a stop codon in E6 preventing its expression; this construct is referred hereafter as E6 × E7. HFKs stably expressing HPV-16 E7 (Figure 2a) were treated with the DNA damaging agent doxorubicin (DXRB) to increase nuclear p27 levels in the presence or absence of the PI3K inhibitor LY294002 or Akt inhibitor 4 (AI4) and localization of p27 was assessed by immunoflourescence analysis (Figure 2b). E7 expression correlated with greater cytoplasmic retention of p27 under DXRB treatment, whereas vector control cells exhibited mostly nuclear p27 (Figure 2b). Treatment with LY294002 or AI4 in E7-expressing HFKs redistributed p27 back to the nucleus, indicating involvement of the PI3K/Akt pathway in mislocalization. Akt activity, assessed by T308 and S473 phosphorylation, was higher in DXRB-treated E7-expressing cells when compared to controls (Figure 2c and d), consistent with previous observations in our laboratory (Westbrook et al., 2002; Menges et al., 2006). Localization of p27 was also assessed in cervical cancer biopsies previously shown to contain increased levels of active Akt (Menges et al., 2006). Normal tissue contained punctate p27 in the remaining nuclei of upper layers, whereas localization in high-grade lesions was mostly cytoplasmic (Supplementary Figure S1).

Figure 2.

E7 mislocalizes p27 to the cytoplasm in a PI3K/Akt-dependent manner in HFKs. (a) Primary HFKs were maintained as described by Incassati et al. (2006) and were infected with retrovirus containing pBabe vector alone or pBabe-E6 × E7. Cells were selected with 1.25 μg/ml puromycin (Sigma) for 2 days and expression of HPV-16 E7 was verified via western blot analysis. (b) HFKs expressing E7 or pBabe were treated with 0.1μg/ml DXRB-HCl (Sigma) for 18 h in the presence or absence of 10 μm LY294002 or 1 μM InSolution Akt Inhibitor IV (Calbiochem, EMD Chemicals, Inc., San Diego, CA, USA) and localization of endogenous p27 was assessed by IFA as in Figure 1. (c) Active Akt levels were assessed by western blot analysis using extracts from HFKs treated as described in (b) with phospho-S473 and total Akt antibodies (Cell Signaling). (d) Treatment of E7-expressing HFKs with AI4 blocks activation of Akt, as determined by reduction in phospho-T308 (Cell Signaling) levels measured by western blot. HFK, human foreskin keratinocyte; HPV-16, human papillomavirus type 16; IFA, immunofluorescence analysis; p27, cyclin-dependent kinase inhibitor p27^kip1; PI3K, phosphoinositide-3 kinase.

Akt has previously been implicated in the cytoplasmic retention of p27 by phosphorylating the Akt consensus site T157 located within the nuclear localization signal (NLS) of p27 (Shin et al., 2002). Therefore, we sought to determine if there were higher levels of phosphor-p27 in E7-expressing cells and whether the PI3K/Akt pathway may be involved. Using an antibody that specifically detects the T157 site of p27 when phosphorylated; we show that E7-expressing HFKs contain a higher level of T157 phosphorylation after DXRB treatment (Figure 3a). To implicate further the PI3K/Akt pathway, p27 was immunoprecipitated under DXRB treatment and the subsequent western blot was probed with the Akt substrate antibody, which recognizes the RXXRXT/S Akt consensus phosphorylation motif when phosphorylated (Figure 3b). Although treatment of E7-expressing cells with LY294002 or AI4 did stabilize total p27 levels, the ratio of phosphorylated to total p27 was similar to that in DXRB-treated pBabe cells (Figure 3c).

Figure 3.

E7 enhances the level of active Akt and phosphorylation of p27 within its Akt consensus phosphorylation site. (a) Primary HFKs expressing E6 × E7 or vector control were treated with DXRB as described earlier and the level of phosphorylated T157 p27 was assessed by western blot using a TI57 phospho-specific antibody (R&D Systems, Inc., Minneapolis, MN, USA) in the presence and absence of the Akt inhibitor AI4. (b) HFKs in the presence or absence of E7 were treated with DXRB alone or in combination with LY294002 for 18 h. p27 was immunoprecipitated (IP) with mouse anti-p27 (Lab Vision) and the subsequent western blot was probed with the Akt substrate recognizing antibody (Cell Signaling). IPs were performed as described by Westbrook et al. (2002). (c) Levels of p27 from (b) recognized by the Akt substrate antibody were measured by densitometry. The total levels of p27 were normalized and the amount of phosphorylated p27 in E7-expressing HFKs treated with DXRB, in the presence or absence of LY294002, was compared to DXRB-treated vector control cells. DXRB, doxorubicin; HFK, human foreskin keratinocyte; p27, cyclin-dependent kinase inhibitor p27^kip1.

Cytoplasmic p27 has been demonstrated to increase the migration of cancer cells by an Akt-dependent mechanism and is associated with inhibition of the RhoA pathway (Besson et al., 2004; Vasko et al., 2004; Wu et al., 2006a). To evaluate whether RhoA activity itself was perturbed by E7 under conditions where cytoplasmic retention of p27 is observed, control and E7-expressing keratinocytes were treated with DXRB and RhoA activity was evaluated in a pulldown assay (Figure 4a). Active, or GTP bound, RhoA was pulled down using a glutathione S-transferase fusion protein containing the Rho-binding domain of Rhotekin that only binds Rho in the GTP-bound state. In the presence of E7, RhoA activity was attenuated in comparison to vector control cells and was partially derepressed by Akt inhibition with AI4 (Figure 4a). Next, we assessed whether p27 was binding to and consequentially inhibiting RhoA in the presence of E7. Interestingly, when RhoA was immunoprecipitated, a greater amount of p27 was bound in E7-expressing HFKs and inhibition of the Akt pathway reduced the level of binding (Figure 4b). Consistent with our model, under conditions where p27 is observed to be mislocalized in E7-expressing HFKs, the RhoA pathway is attenuated by cytoplasmic p27 in an Akt-dependent manner.

Figure 4.

E7 enhances the migration of primary HFKs and lowers RhoA activity in an Akt-dependent manner. (a) RhoA activity was assessed in using the Rho activation assay kit (Upstate Biotechnology, Charlottesville, VA, USA) from HFKs expressing E7 or not treated with DXRB with or without AI4. See Supplementary data for a detailed protocol of RhoA-GTP measurement. (b) Co-immunoprecipitation of RhoA and p27. Primary HFKs expressing E7 or not were treated with DXRB with or without AI4 for 18 h. IPs were performed using mouse anti-RhoA (Santa Cruz Biotechnology) and rabbit anti-p27 (Santa Cruz Biotechnology) to detect p27 bound to RhoA. (c) HFK migration was assessed by scratch assay under conditions where p27 was observed to be mislocalized. Migration was measured by wound closure using Image-pro Plus software, with * corresponding to P<0.05 and n = 13. Graph shows the relative increase or decrease in migration when normalized to DXRB-treated pBabe cells for the given condition. For a detailed wound assay protocol see Supplementary Figure S2. DXRB, doxorubicin; HFK, human foreskin keratinocyte; IP, immunoprecipitation; p27, cyclin-dependent kinase inhibitor p27^kip1.

To evaluate whether the cytoplasmic p27 retention and RhoA attenuation observed in E7-expressing cells contribute to increased migration, we used a standard scratch assay to determine relative migration. Wound closing was assessed 12 h after the initial wounding to reduce the possibility of mistaking migration for proliferation. Keratinocytes expressing E7 closed the wound more rapidly than vector control cells under DXRB treatment (Figure 4c, Supplelmentary Figure S2), a condition where p27 mislocalization is observed. Interestingly, migration was dependent on Akt, as treatment with AI4 blocked migration (Figure 4c, Supplementary Figure S2). Inhibition of the RhoA downstream effector ROCK with Y-27632 enhanced DXRB vector control migration, although the migration in E7-expressing cells is left relatively unaltered by this treatment (Figure 4c, Supplementary Figure S2). This demonstrates that inhibition of the RhoA pathway in control HFKs recapitulates the migrational phenotype of E7-expressing cells, further implicating that RhoA inhibition by E7 positively regulates migration.

We show here for the first time that primary keratinocytes expressing the E7 protein from HPV-16 migrated more rapidly in a scratch assay than vector controls in an Akt-dependent manner. Enhanced migration was observed under conditions where the cyclin-dependent kinase inhibitor p27 was localized in the cytoplasm of E7-expressing cells and active RhoA levels were attenuated, as observed by several other groups. Interestingly, ROCK inhibition enhances control cell migration, whereas leaving E7-expressing cell migration relatively unaltered; indicating that RhoA attenuation by E7 enhances migration. Migration was reduced when the Akt pathway was perturbed, suggesting that this pathway is required for efficient motility in keratinocytes. Inhibition of Akt also negated the inhibition of RhoA in E7-expressing cells and correlated with a reduced amount of p27 bound RhoA in parallel with relocalization of p27 to the nucleus upon DNA damage. Cytoplasmic p27 expression has been demonstrated to be marker of poor prognosis in several forms of cancer and has been directly linked to migrational enhancement and tumor invasion through RhoA inhibition. We postulate that E7 enhances cytoplasmic retention of p27 through the PI3K/Akt pathway and not only removes constrictions on cell division, but enhances cellular migration by inhibiting the small GTPase RhoA through direct p27 binding. Although further exploration needs to be done, unraveling the mechanism behind E7-mediated migration may provide insight into pathways that could be chemically targeted for treatment of HPV-positive invasive cervical carcinomas.