Activated Notch1 Inhibits p53-Induced Apoptosis and Sustains Transformation by Human Papillomavirus Type 16 E6 and E7 Oncogenes through a PI3K-PKB/Akt-Dependent Pathway

Human cervical tumors represent a major subset of female cancers in the developing world. These tumors are characterized by the persistent expression of the E6 and E7 oncogenes of the high-risk type 16 and 18 human papillomaviruses (HPV-16 and -18) (34). A range of observations have revealed the central roles of these two proteins in immortalization of human epithelial cells and their continuing requirement for persistent tumor progression in vivo (18, 21, 29). Invasive cervical tumors also consistently reveal the intracellular expression of Notch1 and Notch2 proteins, suggestive of deregulated Notch signaling (5, 33). Further, ligands of Notch are expressed in human cervical cancers (10). A functional role for deregulated Notch signaling has been suggested in the context of cervical neoplasias, as activated alleles of the Notch1 gene synergize with E6 and E7 oncogenes in in vitro transformation assays (24), analogous to the synergy with adenovirus E1A (4). Suggested roles for Notch1 in regulating epithelial tumor progression are induction of proliferation through transcriptional upregulation of cyclin D1 (25) and inhibition of apoptosis by activation of the phosphatidylinositol 3-kinase (PI3K)-PKB/Akt pathway (24).

One of the well-documented functions of the E6 protein is to bind and degrade p53 (34). This is believed to inhibit the function of the p53 pathway, as HPV-associated tumors have a low frequency of mutant p53 alleles. The major consequence of inhibiting p53 function is to inhibit a key inducer of apoptosis in transformed cells. Paradoxically, many cervical tumors often reveal detectable levels of p53 (22, 14), possibly reflecting incomplete degradation by E6. Alternative possibilities to explain the detection of the p53 protein include the presence of variants of E6 incapable of degrading p53 (16) and the presence of feedback loops that lead to high p53 levels consequent to E6 expression. In addition, there are reports suggesting that the E7 protein may have a role in stabilizing p53 levels (7).

In addition to being inhibited by viral oncogenes, p53 function is negatively regulated by cellular gene products such as Hdm2 (15). This process is regulated in part by a feedback loop since Hdm2 is a transcriptional target for p53. Activation of PI3K signaling through PKB/Akt can also activate Hdm2 and thus can negatively regulate p53 function (1, 9, 20), apart from modulating apoptosis by phosphorylating proteins on serine or threonine residues to potentiate cell survival proteins such as Bad, caspase 9, GSK3, etc. (3, 6). In this study, we have examined a potential link between activated Notch1 (AcN1) and p53 function mediated through the PI3K pathway.

The human Notch1 was originally cloned from a chromosomal rearrangement generated in a human T-cell acute lymphoblastic leukemia cell line (8). The truncated allele generates a constitutively active intracellular form of Notch1 protein (2). We have shown that expression of AcN1 in HaCaT cells induces resistance to anoikis (apoptosis induced on matrix withdrawal) and sustains colony formation by HPV-16 E6 and E7 oncogenes in soft-agar assays (24). HaCaT is an HPV-negative immortalized line with mutant p53 alleles and stable expression of the mutant protein (17). Consistent with previous reports (23), we find that expression of wild-type p53 in HaCaT cells induced apoptosis (Fig. 1A). Transfection of wild-type p53 in HaCaT-Neo cells induced apoptosis in 40 to 45% of transfected cells by microscopy (Hoechst-stained cells were analyzed for chromatin condensation and nuclear fragmentation as in reference 24). However in HaCaT-AcN1 cells, no apoptosis induction was detected following expression of p53 (Fig. 1A). Similarly, HaCaT-Neo cells cotransfected with both AcN1 (AcN1-green fluorescent protein [GFP]) and p53 showed resistance to p53-induced apoptosis (Fig. 1C).

Expression of the E6 protein prevented the induction of apoptosis caused by p53 (Fig. 1B), as suggested from the well-documented link between these two proteins (26). There are conflicting reports on the role of E7 in regulating the function of p53. Consistent with the suggestion of Hickman et al. (12), we find that, on expression of both E7 and E6, there is p53-induced apoptosis of around 30% (Fig. 1B). However, AcN1 blocks the induction of p53-dependent apoptosis even in the additional presence of E7 (Fig. 1B).

Induction of apoptosis by p53 can be through the transcriptional upregulation of proapoptotic proteins such as Bax or downregulation of prosurvival proteins such as Bcl2 and Bcl-xl (28). Transactivation-independent mechanisms that have been suggested include a direct interaction of the p53 protein in the mitochondria (11, 19, 27). We evaluated the activation of a p53-dependent reporter and find that AcN1 reduces the induction by sevenfold (Fig. 2A). Correspondingly, in the presence of p53, we find a 1.5-fold overexpression of the Bax protein in HaCaT-Neo cells and no increase in HaCaT-AcN1 cells (Fig. 2B). We evaluated the expression of Hdm2, another protein that is transcriptionally upregulated by p53, and find a similar phenomenon. There are 2- and 1.2-fold inductions of Hdm2 by p53 in the absence and presence of AcN1, respectively (Fig. 2B).

Constitutively activated forms of PKB/Akt cooperated with HPV-16 E6 and E7 oncogenes in in vitro transformation assays to the same extent as AcN1 (24). Further, we have earlier reported the detection of phosphorylated forms of PKB/Akt in detached cells in the presence of AcN1 (24). Addition of chemical inhibitors to PI3K induces anoikis and eliminates the phosphorylation of PKB/Akt (24). Here, we extended this analysis and find that, even in adherent cells, there is a markedly higher level of phosphorylated PKB/Akt consequent to AcN1 expression (Fig. 3A). We also evaluated levels of Bad, a downstream target of phosphorylated PKB/Akt. When phosphorylated by Akt, Bad is unable to interact with Bcl-xl (6), restoring Bcl-xl's antiapoptotic function. In the AcN1-expressing cells, levels of Bad phosphorylated at serine 136 were significantly higher (twofold) than those in Neo cells (Fig. 3A). However, there was no difference in levels of the antiapoptotic Bcl-xl protein and total Bad protein between Neo- and AcN1-expressing cells (Fig. 3A). Addition of LY294002 to HaCaT-AcN1 cells grown on coverslips and transfected with p53-GFP eliminated resistance to p53-induced apoptosis. Results of the assay are shown in Fig. 3B, and representative images are shown in Fig. 3C. Apoptosis was induced in around 40% of the p53-GFP-transfected AcN1 cells when treated with LY294002 (Fig. 3C). Similar results were obtained with wortmannin, another chemical inhibitor of PI3K (data not shown).

PKB/Akt has recently been shown to phosphorylate Hdm2 and thus inhibit p53 activation. By confocal imaging we show that HaCaT-Neo cells express endogenous Hdm2 predominantly in the cytoplasm (Fig. 4A), in keeping with very low levels of phosphorylated PKB/Akt. HaCaT-AcN1 cells exhibit the Hdm2 protein in both the nucleus and cytosol (Fig. 4B). Addition of LY294002 restores the predominantly cytosolic localization of Hdm2 (Fig. 4C). This suggests that activation of Hdm2 by PKB/Akt leads to nuclear localization. Hdm2 levels per se in HaCaT cells are fairly low, in keeping with the absence of wild-type p53 protein. Expression of wild-type p53-GFP shows colocalization with Hdm2 in the nucleus in all HaCaT-AcN1 p53-GFP-positive cells. On addition of LY294002, cytosolic Hdm2 can be detected in around 30% of p53-GFP-positive HaCaT-AcN1 cells. It should be noted that on overexpression of wild-type p53 in HaCaT, Hdm2 can be detected in the nucleus even in the absence of detectable phosphorylated PKB/Akt, possibly due to low levels of phosphorylated PKB/Akt in these cells.

The results obtained so far show that AcN1 can inhibit p53-induced apoptosis and that this is dependent on a PI3K-PKB/Akt-dependent pathway. We suggest that there is a bifurcation downstream of PKB/Akt, with two components of a prosurvival mechanism that includes both the upregulation of genes such as the Bad gene and subcellular relocalization of Hdm2. The importance of these observations lies in the potential inability of p53 to block transformation by HPV-16 E6 and E7 oncogenes in the presence of AcN1. Expression of AcN1 along with E6 and E7 typically yields a three- to fivefold increase in soft-agar colonies (24) of HaCaT cells. Expression of p53 does not block this increase (Fig. 5).

To extend the above observations with cell lines that express endogenous wild-type p53, we used both Caski and U2OS cells (Fig. 6) and induced p53-dependent apoptosis with doxorubicin (Adriamycin) (31). Caski and U2OS are cell lines derived from an HPV-16-positive cervical tumor and osteosarcoma, respectively. Caski cells have been shown to accumulate p53 on treatment with actinomycin D and leptomycin B, resulting in apoptotic death of these cells (13). Here, apoptosis dependent on p53 expression was induced by doxorubicin. AcN1 or control (Mock-) transfected cells were identified by coexpressing GFP. Doxorubicin at 0.2 and 0.4 μg/ml in Caski cells led to 20 and 25 to 30% apoptosis, respectively (Fig. 6A). Expression of AcN1 in these cells reduces apoptosis to 5 and 10% with 0.2 and 0.4 μg of doxorubicin/ml, respectively (Fig. 6A). U2OS cells were insensitive to the lower doxorubicin concentration of 0.2 μg/ml. However, around 20% of the transfected cells showed distinct apoptotic morphology at 0.4 μg of doxorubicin/ml, and AcN1 reduced apoptosis in these cells to 5% (Fig. 6B). Insets in Fig. 6A and B show induction of p53 in these cells after addition of doxorubicin, and representative pictures are shown in Fig. 6C. The results in Fig. 6 show that expression of activated Notch1 blocks p53-dependent apoptosis induced by doxorubicin in transformed cell lines irrespective of the expression of HPV oncogenes.

Features of deregulated Notch signaling are consistently observed in the progression of high-grade cervical tumor precursors to invasive tumors (5). In addition to the cooperative transformation that we have noted in HaCaT cells, we have recently analyzed d1DR cells, which are immortalized by using the HPV-16 genome. On coexpression of AcN1 and HPV 16 E6 and E7 oncogenes, these cells generate progressive tumors in vivo (O. Chakrabarti et al., unpublished data). Consistent with these observations, the use of an antisense construct against Notch1 in Caski cells, which are derived from a cervical tumor, inhibited growth in vivo (32). In contrast, recent observations from Talora et al. suggest that Notch1, in contrast to Notch2, may be downregulated in late-stage cervical tumors (30). They suggest that adenovirus Notch1 may negatively regulate the HPV-16 promoter, with consequent growth arrest (30), in cervical tumor-derived lines including Caski. Thus the role of the Notch pathway may involve a complex interplay of signals that both sustain and inhibit aspects of tumor progression.

In this study, we have shown that, in an immortalized human epithelial cell line, activation of Notch1 signaling prevents apoptosis induced on p53 overexpression. Our results extend the link between activated Notch signaling and consequent induction of the PI3K-PKB/Akt pathway as a mediator of apoptosis resistance. Further, in transformed cells, we show that activated Notch1 signaling prevents cell death induced by a DNA-damaging agent.

Expression of HPV-16 E6 and E7 oncogenes with activated alleles of the Notch1 gene sustains transformation even in the presence of p53. This is consistent with the inhibition of p53 function by activated Notch1 in the process of transformation. In addition, the introduction of activated Notch1 alleles inhibits the induction of apoptosis induced by DNA damage in an HPV-16-positive cervical tumor line. Collectively, the results in this study strengthen the evidence of a potential role for Notch signaling in the context of neoplasias such as cancers of the cervix. It would be informative to analyze the link between the various Notch genes and the p53 pathway in diverse cellular contexts.