ABSTRACT
Infection with human papillomavirus (HPV) is a critical factor in the pathogenesis of most cervical cancers and some aerodigestive cancers. The HPV E6 oncoprotein from high-risk HPV types contributes to the immortalization and transformation of cells by multiple mechanisms, including degradation of p53, transcriptional activation of human telomerase reverse transcriptase (hTERT), and degradation of several proteins containing PDZ domains. The ability of E6 to bind PDZ domain-containing proteins is independent of p53 degradation or hTERT activation but does correlate with oncogenic potential (R. A. Watson, M. Thomas, L. Banks, and S. Roberts, J. Cell Sci. 116:4925-4934, 2003) and is essential for induction of epithelial hyperplasia in vivo (M. L. Nguyen, M. M. Nguyen, D. Lee, A. E. Griep, and P. F. Lambert, J. Virol. 77:6957-6964, 2003). In this study, we found that HPV type 16 E6 was able to activate NF-κB in airway epithelial cells through the induction of nuclear binding activity of p52-containing NF-κB complexes in a PDZ binding motif-dependent manner. Transcript accumulation for the NF-κB-responsive antiapoptotic gene encoding cIAP-2 and binding of nuclear factors to the proximal NF-κB binding site of the cIAP-2 gene promoter are induced by E6 expression. Furthermore, E6 is able to protect cells from TNF-induced apoptosis. All of these E6-dependent phenotypes are dependent on the presence of the PDZ binding motif of E6. Our results imply a role for targeting of PDZ proteins by E6 in NF-κB activation and protection from apoptosis in airway epithelial cells.
Persistent infection with high-risk types of human papillomavirus (HPV) plays a critical role in the pathogenesis of cervical cancer, as well as some head and neck cancers. It is estimated from multiple studies that more than 99% of all cervical cancers worldwide are positive for high-risk HPV (34). HPV type 16 (HPV-16) is the most common subtype associated with cervical cancer and is found in approximately 50% of all cases (4). An association of HPV infection with head and neck cancers has more recently been revealed. A portion of cancers of the oral cavity, pharynx, and larynx is associated with HPV infection. Again, HPV-16 is the most prevalent type implicated in aerodigestive cancers, especially of the oropharynx and tonsils (17).
The two most relevant viral gene products contributing to the immortalization and transformation of HPV-infected cells are E6 and E7 (reviewed in reference 24). These two proteins, which are expressed throughout the viral life cycle, are necessary and, in some cells, sufficient for immortalization and oncogenic transformation. E7 binds and inactivates the retinoblastoma tumor suppressor protein (Rb), which is an inhibitor of the E2F family of transcription factors. As a result, the E2F factors are allowed to drive transcription of their target genes, leading to progression of the cell cycle. E6 binds to and directs the degradation of the tumor suppressor protein p53, which normally acts to prevent cell cycle progression or to induce apoptosis under cell stress or DNA damage. The degradation of p53 is accomplished through formation of a complex among p53, E6, and the E6-associated cellular ubiquitin ligase E6-associated protein (E6AP). Polyubiquitination of p53 by E6AP directs the proteasomal degradation of the complex. Additionally, E6 is responsible for transcriptional activation of the telomerase reverse transcriptase (TERT) gene, which is the catalytic subunit of the enzyme telomerase (20). In HPV-infected cells, increased telomerase activity due to TERT transcription is believed to play a role in maintaining the telomeric repeats at chromosomal termini, allowing cells to avoid replicative senescence and become immortal. E6 is unable to degrade p53 or activate telomerase with the loss of E6AP binding, as with the mutant E6Δ9-13 (11).
E6 also targets several other cellular proteins for degradation in a manner similar to that of p53 degradation. Many of these proteins contain an interaction domain termed the PSD95/Dlg/ZO-1 (PDZ) domain, named for postsynaptic density protein, discs large tumor suppressor, and the epithelial tight junction protein ZO-1, all of which are among the first identified PDZ domain-containing proteins. PDZ proteins have also been found to be targeted by oncoproteins from other tumor viruses, such as human T-cell lymphotropic virus Tax (30) and adenovirus E4 orf1 (13, 21). The PDZ domain itself is composed of 80 to 100 amino acid residues and mediates binding to a C-terminal motif or other PDZ domains within its interaction partners. The majority of these binding partners are transmembrane proteins located at the plasma membrane including transmembrane receptors, and channel proteins (18). Biological functions of PDZ proteins include signaling, cell adhesion, ion exchange, and tight-junction integrity. E6 interacts with several PDZ proteins, including MAGUKs (Dlg, Scribble, MAGI-1, MUPP1) (23) and TIP-1 (8), via a C-terminal PDZ binding motif on the E6 protein. The ability of E6 to bind PDZ domain-containing proteins correlates with oncogenic potential (35) and is essential for induction of epithelial hyperplasia in in vivo models (28). An E6Δ146-151 mutant that lacks the PDZ binding motif and is therefore unable to bind PDZ domain targets (19, 21, 25) fails to produce epithelial hyperplasia and squamous cell carcinoma in mice as the wild type does, despite its ability to activate telomerase and degrade p53 (28). Molecular and cellular changes induced by the targeting of PDZ domain-containing proteins which may be related to hyperplasia and oncogenic transformation are thus far undetermined.
NF-κB activation has been implicated in the pathogenesis of many cancers, playing roles in many processes related to transformation and oncogenesis, including proliferation, migration, angiogenesis, and prevention of apoptosis. Upon stimulation, NF-κB factors are translocated to the nucleus after proteasomal processing, where they are allowed to bind and activate their target genes (26). One of two NF-κB heterodimers, p65/p50 or p52/RelB, is typically involved, the latter of which is activated by some tumor necrosis factor (TNF) family receptors (6). While TNF stimulation can lead to either apoptosis or proliferation, NF-κB tips the scale in favor of proliferation. The primary antiapoptotic pressure arising from TNF and other cytokine and growth factor stimulation is the activation of NF-κB and the transcription of its antiapoptotic target genes, including the inhibitor of apoptosis proteins XIAP, cIAP-1, and cIAP-2 (12). cIAP-2 has been shown to be a critical and potent antiapoptotic factor in cells expressing HPV-16 oncoproteins (38). Knockdown of cIAP-2 is sufficient to induce apoptosis in either HeLa cells or human oral keratinocytes immortalized with HPV-16 E6 and E7. NF-κB also activates proliferation-related genes such as cyclins and growth factors (6). NF-κB activation has been implicated in several types of solid tumors (32), including breast cancers (36).
Certain studies have suggested that HPV may modulate the effects of TNF signaling and possibly regulate NF-κB activity. HPV E6 has been found in some cases to protect cells from TNF-induced apoptosis (3, 9) or, at high doses of E6, sensitize cells to TNF-induced apoptosis (10). NF-κB activation by HPV oncoproteins has been previously implied, partially by microarray data suggesting induction of expression of NF-κB-responsive genes such as NF-κB pathway components, nerve growth factor, nerve growth factor receptor, and inflammatory cytokines (15, 16, 27). However, a definitive activation of particular NF-κB components by HPV oncoproteins has not been shown. Moreover, an association of NF-κB activation with a known function of E6 has not been made. In fact, it is not yet established whether it is E6 or E7 that is more important for regulation of NF-κB-responsive genes.
Here we demonstrate that E6 induces the binding of nuclear p52-containing NF-κB complexes to an NF-κB binding element, induces NF-κB-responsive promoter activity, and induces expression of the NF-κB-responsive cIAP-2-encoding gene involved in protection from apoptosis, all in a PDZ binding motif-dependent manner. Furthermore, we show that E6 is able to confer resistance to apoptosis induced by TNF, also dependent on the presence of the PDZ binding motif.
MATERIALS AND METHODS
Cells and culture.Stably transduced primary human airway epithelial cells (AECs) were cultured at early passage in keratinocyte serum-free medium (KSFM) with epidermal growth factor (EGF) and bovine pituitary extract supplements in accordance with the manufacturer's recommendation (Gibco). Cells were transduced with either the retroviral vector LXSN alone or that expressing wild-type E6, E6Δ146-151, or E6Δ9-13. Retroviral plasmids were a gift from Denise Galloway, Fred Hutchinson Cancer Research Center, Seattle, WA. Retroviruses were packaged by transfection of Phoenix cells and collection of viral supernatants. One milliliter of viral supernatant and 8 μg/ml Polybrene was applied to 40% confluent AECs on 100-mm tissue culture dishes. Medium was changed after 24 h, and cells were split 1:4 after 48 h. Cultures were then selected for 7 days on neomycin (50 μg/ml) in KSFM. Surviving cells were pooled.
Reverse transcription-PCR (RT-PCR).RNA was collected from the above-mentioned stably transduced AECs by using Tri-Reagent (MRC, Inc.) in accordance with the manufacturer's recommendation. RNA concentrations were normalized based on spectrophotometric absorbance. RNA was reverse-transcribed into cDNA using RetroScript reagents and protocols (Ambion). PCR was performed with AmpliTaq Gold and accompanying buffers (Roche) for the following targets with the following primers: cIAP-2, 5′-CCTCCTGGGTTGAAGCA-3′ and 5′-GACTCAGTTCTTGTGTGGA-3′, and Actin, 5′-GGGGGAAATCGTGCGTGACATT-3′ and 5′-GATGGAGTTGAAGGTAGTTTCGTG-3′. PCR products were separated on 2% agarose gels at 110 V for 50 min and visualized by staining with ethidium bromide. Reaction mixtures contained 2.5 mM MgCl2, 0.25 mM deoxynucleoside triphosphates, and 2 μM primers. Reactions were carried out at a number of cycles predetermined to give linearly quantitative results. Quantitative real-time PCR for cIAP-2 was performed with Syber-Green PCR Master Mix (Applied Biosystems) on an ABI Prism 7000 instrument with primers 5′-CCATTTTGAACCTGGATA-3′ and 5′-CATTTCCACGGCAGCA-3′. No-RT and no-template controls were included, as well as 18S RNA amplification controls for normalization.
Luciferase reporter assays.The above-mentioned stably transduced AECs were infected with an adenovirus vector containing the firefly luciferase gene driven by four consecutive NF-κB binding sites (31) (gift from Paul McCray, University of Iowa). Adenovirus infection efficiency was normalized for by parallel infection of cells with a vector constitutively expressing firefly luciferase. Infections were carried out in KSFM without supplements for 4 h before replacing the infection medium with fresh KSFM with supplements. Luciferase assays were performed with reagents and protocols from the Steady-Glo kit (Promega) and measured at 48 h postinfection. Human tert (hTERT) promoter luciferase assays were performed by transfection of cells with a pGL3 luciferase reporter vector (Promega) driven by the core 255-bp region of the hTERT promoter. Transfections were carried out with GenePorter reagents (GTS) in KSFM without supplements for 5 h. Transfection medium was replaced with fresh KSFM with supplements, and cultures were incubated for 24 h at 37°C. Transfection efficiency was normalized by cotransfection with a constitutive Renilla luciferase vector (pRL-null) and dual measurement of Renilla and firefly luciferase activity with the dual-luciferase kit (Promega).
Electrophoretic mobility shift assay (EMSA).Gel shift assays were carried out and visualized with Light-Shift reagents and protocols (Pierce). Biotin-labeled and unlabeled probes were annealed by heating to 100°C and cooling to room temperature overnight. Probe oligonucleotide sequences were as follows: wild-type, 5′-GTAGGGGACTTTCCGAGCTCGAGATCCTATG-3′ and 5′-CATCCCCTGAAAGGCTCGAGCTCTAGGATAC-3′; NF-κB mutant, 5′-GTAGGCGACTATCCGAGCTCGAGATCCTATG.-3′ and 5′-CATCCGCTGATAGGCTCGAGCTCTAGGATAC-3′. Probe oligonucleotide sequences for the NF-κB binding site and surrounding sequence in the cIAP-2 promoter were as follows: 5′-TTATTACCGCTGGAGTTCCCCTAAGTC-3′ and 5′-GACTTAGGGGAACTCCAGCGGTAATAA-3′. (The NF-κB binding site is in bold.) Reactions were carried out in accordance with the manufacturer's instructions and included 2 μl of poly(dI-dC) (Sigma), 0.01 to 1 μM probe, 5 mM MgCl2, 3 μl of nuclear extract, 50 μM competitor probe where applicable, 50 μg/ml supershift antibody (mouse monoclonal anti-p-52 [sc-7386X; Santa Cruz], mouse monoclonal anti-p65 [sc-8008X; Santa Cruz], rabbit polyclonal anti-c-Rel [sc272X; Santa Cruz], rabbit polyclonal anti-RelB [sc-226X; Santa Cruz] or rabbit polyclonal anti-p50 [sc-7178X; Santa Cruz]) where applicable, and deionized water to 20 μl. Nuclear extracts were obtained with NE-PER reagents (Pierce) and dialyzed with buffer D (20 mM HEPES, 20% glycerol, 100 mM KCl, 2 μM EDTA, 5 μM dithiothreitol, 600 μM phenylmethylsulfonyl fluoride, 0.6 μg/ml aprotinin, 0.1 μg/ml pepstatin) overnight with 0.1 to 0.5 ml Slide-a-lyzer cassettes (Pierce). Binding products were separated on 6% Tris-borate-EDTA-6% polyacrylamide minigels prerun for 30 min at 100 V and run for 45 min at 100 V. Separated products were transferred to Hybond-N+ nitrocellulose membranes (Amersham) with a Bio-Rad Mini-Protean 3 apparatus for 1 h at 100 V and cross-linked on a Fisher Transilluminator at 80% power for 15 min. Products were detected in accordance with the manufacturer's instructions.
Western blot assays.Western blot analysis was performed as previously described (11), with the following commercially available antibodies: goat polyclonal anti-actin (sc-1616; Santa Cruz) and mouse monoclonal anti-p53 (Ab-6; Oncogene Research Products).
Caspase 3 activity assay.Caspase 3 activity was measured with the Enzolyte Rh110 caspase 3 assay kit (Anaspec). Cells were treated for 24 h with 3 μl of dimethyl sulfoxide (DMSO) or cycloheximide (Sigma) in DMSO (final concentration, 15 μg/ml) along with 50 ng/ml TNF-α in 2 ml of KSFM in six-well tissue culture dishes. Adherent and detached cells were collected by scraping and centrifuging cells and then resuspending cells in lysis buffer. Lysates were incubated with substrate on black 96-well microtiter plates for 3 h to overnight. Samples were read at an excitation wavelength of 485 nm and an emission wavelength of 520 nm on a Tecan fluorimeter.
RESULTS
E6 induces activation of NF-κB-responsive transcription in a PDZ binding motif-dependent manner independent of p53 degradation and hTERT activation.Given evidence that NF-κB-responsive genes may be induced by E6 (27), we investigated whether E6 actually induces NF-κB-responsive transcription and whether the PDZ domain binding motif of E6 is necessary for this induction. To address this, we infected primary human AECs stably expressing E6, E6Δ146-151, E6Δ9-13, or cells containing the LXSN vector alone with an adenovirus vector encoding an NF-κB-responsive reporter gene construct. E6 activated transcription an average of 3.6-fold from this reporter in multiple independent replicate experiments, whereas the non-PDZ binding mutant and the E6AP binding mutant failed to activate transcription (Fig. 1). These results demonstrate a PDZ binding motif-dependent activation of NF-κB-responsive gene transcription. Failure of the E6AP binding mutant E6Δ9-13 to activate transcription suggests that it may be necessary for E6 to direct the proteasomal degradation of a substrate in order to exert its effect on an NF-κB-responsive promoter. The functionality of wild-type E6 and E6Δ146-151 was checked by examining the ability of each to activate hTERT transcription and degrade p53. Both E6 and E6Δ146-151 induced an increase in hTERT transcripts, as well as a 3.5-fold induction of hTERT promoter activity by reporter assay (Fig. 2A and B). Both were also able to degrade p53 despite the loss of the C-terminal PDZ binding motif (Fig. 2C). This shows that both p53 degradation and telomerase activation are independent of the PDZ binding motif of E6, as previously described (20). Separation of these functions of E6 from PDZ binding indicates that PDZ-dependent NF-κB activation by E6 is also separable from hTERT activation and p53 degradation.
E6 induces NF-κB-responsive promoter activity in a PDZ binding motif- and E6AP binding-dependent manner. AECs stably transduced with the LXSN retroviral vector alone or that expressing either wild-type E6, the PDZ binding motif mutant E6Δ146-151, or the E6AP binding motif mutant E6Δ9-13 were infected with an adenovirus (Ad) NF-κB-luciferase reporter vector or a vector constitutively expressing luciferase as an infection control. Luciferase activity with the LXSN vector alone was given a value of 1, and activities in extracts of cells expressing E6, E6Δ146-151, and E6Δ9-13 were expressed as n-fold changes over LXSN alone. E6-expressing cells exhibit a 3.6-fold increase in NF-κB element-regulated luciferase activity over LXSN alone, whereas E6Δ146-151- and E6Δ9-13-expressing cells show no significant change in luciferase activity over LXSN alone. Error bars represent 1 standard deviation based on 15 independent experiments. * = based on triplicate experiments including LXSN and E6 and E6Δ9-13 experimental groups.
AECs stably transduced with either E6 or E6Δ146-151 both exhibit hTERT activation and p53 degradation, showing that PDZ-dependent functions of E6 are separable from hTERT activation and p53 degradation. (A) RT-PCR for hTERT shows transcript accumulation in AECs expressing both E6 and E6Δ146-151, but not with LXSN alone. (B) Similarly, hTERT promoter activity is increased 3.5-fold over that in cells expressing both E6 and E6Δ146-151 over LXSN. (C) Both E6 and E6Δ146-151 are also able to degrade p53, as indicated by a dramatic decrease in accumulation of p53 protein in cells expressing either E6 or E6Δ146-151.
E6 induces the presence of nuclear p52-containing, κB element binding NF-κB complexes in a PDZ binding motif-dependent manner.NF-κB activation definitively occurs when its cytoplasmic components are proteasomally processed and allowed to translocate into the nucleus, where they can bind κB-responsive elements. Typical heterodimeric NF-κB components that translocate to the nucleus are either p50/p65 or p52/RelB, depending on the mechanism of activation. EMSA was employed to determine the activation status and nuclear localization of specific NF-κB components by E6 and to determine if any effect was dependent on the presence of the PDZ binding motif. We found that E6 induces the presence of NF-κB element binding proteins in the nuclei of AECs in a PDZ binding motif-dependent manner (Fig. 3A) and that those complexes contain p52 (Fig. 3B), consistent with activation of the noncanonical pathway of NF-κB activation. Gel shift data did not indicate a nuclear presence of canonical p50 or p65 NF-κB components. Competition for binding with unlabeled probes showed specificity for the NF-κB binding site, as evidenced by the loss of competition with a mutated competitor probe (Fig. 3C). Some diffuse lower-molecular-weight shifted bands appear in the presence of nuclear extracts from cells expressing the mutant forms of E6, which may indicate the existence of a secondary, non-PDZ-dependent mechanism of inducing the binding activity of certain NF-κB factors by E6 (Fig. 3A).
E6 induces NF-κB-responsive element binding of nuclear NF-κB complexes. (A) EMSAs with nuclear extracts from AECs exhibit a strong, specific shift in probe (prb) mobility with extracts from E6-expressing cells but not significantly with extracts from cells with E6Δ146-151, E6Δ9-13, or the vector alone. (B) Supershifting experiments with antibodies that recognize either p50, p65, p52, RelB, or c-Rel indicate a presence of p52 in shifted bands with extracts from E6-expressing cells. Lanes contain the probe (Prb) alone without extract, no antibody, p50 antibody, p65 antibody, p52 antibody, RelB antibody, c-Rel antibody, and Fos nonspecific antibody, respectively. Supershifted bands are demarcated with asterisks. (C) Specificity of this shifting to the NF-κB-responsive site is indicated by a loss of competition when a competitor (Comp) is used that is mutated at the NF-κB-responsive site. Lanes contain the wild-type (WT) competitor and the mutant (MT) competitor. The NF-κB binding site is in bold. Mutations are underlined.
E6 induces NF-κB element binding at the cIAP-2 promoter and an increase in cIAP-2 transcript level, both in a PDZ binding motif-dependent manner.There are several NF-κB-induced genes that have potential roles in proliferation, promotion of hyperplasia, and protection from apoptosis. The implication of NF-κB-responsive genes in these processes, as well as oncogenesis and tumor survival, led us to investigate the regulation of their transcript levels by E6. By RT-PCR, we determined transcript levels of the NF-κB-responsive, antiapoptotic cIAP-2-encoding gene in AECs stably expressing E6, E6Δ146-151, or E6Δ9-13 or containing the LXSN vector alone. With the wild type, but not with E6Δ146-151 or E6Δ9-13, we found a 2.5-fold increase in transcript levels of the gene for cIAP-2 by real-time and conventional RT-PCR (Fig. 4A and B). To determine whether E6 expression induces binding of NF-κB transcription factors to the cIAP-2 promoter, an EMSA was performed with a probe mimicking the proximal NF-κB binding site in the cIAP-2 promoter. Extracts from wild-type E6-expressing cells induced a robust shift in the mobility of the cIAP-2 promoter probe similar to the shifting pattern seen with the non-promoter-specific NF-κB probe used in Fig. 3, whereas extracts from cells expressing the vector alone, E6Δ146-151, or E6Δ9-13 did not (Fig. 4C). Supershift experiments indicate that p52-containing complexes bind to the cIAP-2 promoter NF-κB sequence in E6-expressing cells (Fig. 4D). These results indicate that E6-induced expression of cIAP-2 is directly associated with the ability of E6 to induce NF-κB binding to the cIAP-2 promoter and that this is dependent on the E6 PDZ binding motif.
E6 induces accumulation of cIAP-2 transcript and NF-κB binding to the cIAP-2 promoter in a PDZ binding motif-dependent manner. (A, B) cIAP2 transcript accumulates in cells expressing E6 but not E6Δ146-151, E6Δ9-13, or the vector alone, as shown by quantitative real-time RT-PCR and conventional RT-PCR. RT lanes serve as negative controls for genomic DNA contamination. (C) EMSA with a cIAP-2 promoter-specific NF-κB binding site probe (Prb) demonstrates an induction of binding activity by E6 dependent on the PDZ binding motif and E6AP binding. The probe represents the most proximal of three such binding sites and the surrounding sequence within the cIAP-2 promoter. (D) EMSA supershifts of a cIAP-2 promoter probe with NF-κB antibodies reveal the presence of p52, similar to that seen with the NF-κB probe used in Fig. 3. Supershifted bands are demarcated with asterisks.
E6 protects against TNF-induced apoptotic caspase 3 activity in a PDZ binding motif-dependent manner.TNF is known to induce both proliferation and apoptosis under different conditions and in different cell types. All members of the TNF superfamily of receptors can activate the NF-κB pathway (12). Through activation of antiapoptotic proteins, particularly IAPs, NF-κB can protect against TNF-induced apoptosis (12). Since we found that E6 can activate NF-κB and transcription of its antiapoptotic target gene cIAP-2 in a PDZ binding motif-dependent manner, we next determined the PDZ binding motif-dependent effect of E6 on TNF-α-induced apoptosis. Apoptotic caspase 3 activity was induced ninefold in AECs containing the vector alone by treatment with cycloheximide and TNF-α. However, treated cells expressing wild-type E6 exhibited only approximately 40% of the apoptotic activity of cells expressing the vector alone. No significant protection was conferred by the non-PDZ binding mutant form of E6 (Fig. 5).
E6 confers partial resistance to TNF-α-induced apoptosis. Caspase 3 activity assay demonstrates a ninefold induction of apoptotic caspase 3 activity in AECs containing the vector alone when treated with 15 μg/ml cycloheximide (CHX) and 50 ng/ml TNF-α. Expressed n-fold change is over those cells treated with the solvent DMSO alone. Induction of apoptosis is partially protected against by E6 in a PDZ binding motif-dependent manner, exhibiting only approximately 40% of the caspase 3 activity of AECs with the vector alone. No significant protection is conferred by E6Δ146-151. Error bars indicate a range of values from duplicated experiments. Caspase 3 activity was measured in ambiguous units (amb. units) derived from fluorescence output.
DISCUSSION
NF-κB activation has been implicated in a variety of processes related to transformation and oncogenesis, including proliferation, migration, angiogenesis, and prevention of apoptosis. Although NF-κB function is necessary for proper immune function, its constitutive expression in nonimmune tissues may mediate inappropriate inflammation and even promote tumorigenesis (1). In fact, activation of NF-κB in response to chemotherapeutics and radiation (32) has posed an obstacle to effective treatment of cancer. NF-κB is also involved in invasiveness of tumors, as coculture of macrophages with tumor cells has been found to enhance invasiveness through TNF-α and NF-κB activity (14). Increased NF-κB activity is associated with many cancers, including those associated with viral infections and cancers of the head and neck (6). Inhibition of NF-κB in these head and neck cancers has been shown to decrease cell survival and growth of tumors.
As with other cancers and tumor virus infections, NF-κB-dependent proliferation and protection from apoptosis are likely to be important in the pathogenesis of HPV-associated cancers. HPV E6- and E7-positive cells have been shown to be sensitized to IL-1β-induced NF-κB activation and exhibit elevated levels of NF-κB components (15). E6 rather than E7 expression was found to be associated with nuclear localization of these components (16). Microarray data suggest that NF-κB-responsive genes are induced by E6 expression in cervical keratinocytes (27). Our data show that nuclear binding activity of p52-containing NF-κB complexes and NF-κB-responsive transcription are induced by HPV-16 E6 expression in AECs. EMSA results indicate that complexes involved in the noncanonical NF-κB pathways are activated in the presence of E6. Furthermore, we have established that the PDZ binding motif of E6 is required for activation of NF-κB. We have also shown here that the non-PDZ binding mutant form of E6 that is unable to activate NF-κB retains the ability to activate hTERT transcription and p53 degradation. Failure of the E6Δ9-13 mutant protein to activate NF-κB also suggests that the degradation of a PDZ domain-containing protein(s) by E6 may be necessary for NF-κB activation. At least one study has shown that although E6Δ9-13 cannot bind E6AP or degrade its substrates, the mutant protein retains other functions, such as binding to p53 (5). Delineation of which NF-κB-related pathways are affected by E6 will be important to the understanding of PDZ-dependent mechanisms by which E6 contributes to the transformation of cells.
E6 is known to bind PDZ proteins that affect signaling from several transmembrane receptors. For example, E6 can bind to TIP-2 (8), which plays a role in TGF-β signaling. It has also been indicated that E6 and E6AP can associate with and direct the degradation of PDZ protein PTPH1 (S. B. Vande Pol, personal communication). PTPH1 targets and destabilizes the TNF-α convertase, which is a matrix metalloproteinase involved in the ectodomain shedding of membrane-associated receptor signaling molecules such as TNF and the EGF family receptor ErbB4 (29, 39). A number of PDZ proteins that can affect cell-cell junction signaling are bound and degraded by E6, including hDlg and Scribble (23). These findings may suggest possible mechanisms by which PDZ binding by E6 could induce signaling from membrane-bound receptors upstream of NF-κB; however, further studies are required to elucidate which specific pathway is playing a role.
The finding that E6 induces NF-κB activity is consistent with the hypothesis that E6 may also induce an antiapoptotic phenotype. It was recently shown that cIAP-2 expression is induced in E6- and E7-expressing oral keratinocytes, as well as HPV-positive cancer cell lines, and that its depletion in these cells leads to apoptosis (38). Induction of cIAP-2 in E6- and E7-expressing cells was found to require NF-κB activity. cIAP-2 directly binds and inhibits caspases 3, 7, and 9 and may also inhibit TNF receptor-associated factors 1 and 2 (TRAF1 and TRAF2) (38). Our results indicate that cIAP-2 is regulated by HPV-16 E6 in a PDZ binding motif-dependent manner. We have demonstrated that E6 induces accumulation of transcript for the cIAP-2 antiapoptotic gene, also in a PDZ binding motif-dependent manner. Furthermore, we have shown E6 to induce the binding of nuclear factors to the proximal NF-κB binding site of the cIAP-2 promoter, also dependent on the PDZ binding motif. Regulation of cIAP-2 expression by E6 is consistent with an NF-κB-responsive, antiapoptotic phenotype. In fact, apoptotic caspase 3 activity, which is downstream of the effects of cIAP-2, is diminished in wild-type E6-expressing AECs upon TNF treatment, but not in cells expressing the E6 mutant that is unable to bind PDZ domains.
In the literature, effects of E6 expression on apoptosis, especially that induced by TNF, has been controversial. There are reports of both protection from (17, 37) and sensitization to (22, 33) TNF-induced apoptosis. One complication of studying modulation of TNF-induced apoptosis by HPV oncoproteins is that E7 can also sensitize cells to apoptosis induced by TNF-α (2). The apparent conflict in reports may also be explained by differences in cell types or differences in expression levels of E6. There is now even evidence that cells with low levels of E6 expression protect from TNF-induced apoptosis while cells with high levels are sensitized (9). Studies that have reported sensitization to TNF-induced cytotoxicity by E6 have been done with human ovarian cancer, human colon cancer, and murine fibrosarcoma lines, whereas studies reporting protection from TNF-induced apoptosis have been done with primary fibroblasts or keratinocytes. In mouse fibroblasts, E6 expression is able to protect from TNF-induced inhibition of proliferation and induction of apoptosis (7). This protection provided by E6 was found to be independent of p53, as the tumor suppressor protein did not accumulate during the apoptotic response to TNF and a non-p53 binding E6 mutant retained the ability to protect from TNF-induced apoptosis (7). Our results indicate that E6 is indeed able to protect against apoptosis, as measured by caspase 3 activity, and that this protection relies on the presence of the PDZ binding motif of E6.
In summary, our data obtained with human AECs demonstrate that E6 can protect cells from TNF-induced apoptosis and that this protection is associated with the presence of the E6 PDZ binding motif and activation of NF-κB. Prevention of DNA damage or cytokine-induced apoptosis by activation of NF-κB may be a major obstacle in the treatment of HPV-associated cancers, as well as other cancers that exhibit increased NF-κB activity. NF-κB activation in HPV-infected cells likely plays a role in the proliferative capacity of the cells in addition to protection from apoptosis, which is reflected in the diminished ability of E6 mutants that are not able to bind PDZ domains or activate NF-κB to promote oncogenesis (35) and epithelial hyperplasia (28). Investigation of the potential role of NF-κB activation in the pathogenesis of HPV-associated cancers and hyperplasias is of great interest in regard to the treatment and prevention of such viral cancers and the understanding of the pathogenesis of cancer in general.
ACKNOWLEDGMENTS
We thank Mike Lace and Wendy Maury for discussion regarding EMSA protocols, Paul McCray for adenovirus luciferase reporter vectors, the Gail Bishop laboratory for discussion and reagents, and the other members of the Klingelhutz laboratory for support.
This work was supported by NIH R01 AG18265 (A.J.K.) and a grant from the VAMC (J.H.L.).
FOOTNOTES
- Received 13 September 2005.
- Accepted 13 March 2006.
- Copyright © 2006 American Society for Microbiology
