The human papillomavirus 16 E6 protein binds to Fas-associated death domain and protects cells from Fas-triggered apoptosis.

High risk strains of human papillomavirus (HPV), such as HPV 16, cause human cervical carcinoma. The E6 protein of HPV 16 mediates the rapid degradation of the tumor suppressor p53, although this is not the only function of E6 and cannot completely explain its transforming potential. Previous work in our laboratory has demonstrated that E6 can protect cells from tumor necrosis factor-induced apoptosis by binding to the C-terminal end of tumor necrosis factor R1, thus blocking apoptotic signal transduction. In this study, E6 was shown to also protect cells from apoptosis induced via the Fas pathway. Furthermore, use of an inducible E6 expression system demonstrated that this protection is dose-dependent, with higher levels of E6 leading to greater protection. Although E6 suppresses activation of both caspase 3 and caspase 8, it does not affect apoptotic signaling through the mitochondrial pathway. Mammalian two-hybrid and in vitro pull-down assays were then used to demonstrate that E6 binds directly to the death effector domain of Fas-associated death domain (FADD), with deletion and site-directed mutants enabling the localization of the E6-binding site to the N-terminal end of the FADD death effector domain. E6 is produced in two forms as follows: a full-length version of approximately 16 kDa and a smaller version of about half that size corresponding to the N-terminal half of the full-length protein. Pull-down and functional assays demonstrated that the full-length version, but not the small version of E6, was able to bind to FADD and to protect cells from Fas-induced apoptosis. In addition, binding to E6 leads to degradation of FADD, with the loss of cellular FADD proportional to the amount of E6 expressed. These results support a model in which E6-mediated degradation of FADD prevents transmission of apoptotic signals via the Fas pathway.

In previous work, our laboratory found that E6 could protect cells from TNF (37) and that it does so by binding to the C-terminal end of TNF R1, thus blocking transmission of the apoptotic signal from TNF R1 through the TNF R1-associated death domain (TRADD) (23). These observations led us to ask whether E6 might also modulate apoptotic signaling triggered by related mechanisms, such as the Fas pathway. Apoptosis triggered through the Fas pathway begins with engagement of Fas, either by Fas ligand or with appropriate anti-Fas antibodies, allowing the formation of a death-inducing signaling complex (DISC). Fas first recruits Fas-associated death domain (FADD) to bind to its own death domain (DD) by way of the DD present on FADD. FADD contains both a DD and a death effector domain (DED) and, therefore, functions as an adaptor molecule by also binding to the DED of pro-caspase 8 via a DED/DED interaction. Aggregation of the pro-caspase 8 molecules leads to their mutual activation and the consequent engagement of the caspase cascade, resulting in apoptosis (see Refs. 38 -45 for reviews). FADD may be required for the function of additional signaling pathways as well (46), including the mediation of TRAIL-induced apoptosis (47)(48)(49)(50) and the induction of caspase-independent, necrotic cell death (51).
In this study, we found that HPV 16 E6 protects cells from Fas-triggered apoptosis and that it does so by binding to FADD and accelerating its degradation. Because FADD is known or speculated to participate in a number of apoptotic pathways, these observations suggest that E6 may function to protect expressing cells from apoptosis triggered by a variety of ligands and may therefore play an important role in helping cells infected with HPV 16 avoid elimination by host defense mechanisms and thus contribute to the survival and propagation of the virus.

EXPERIMENTAL PROCEDURES
Reagents-Fas monoclonal antibodies (clone CH-11) (Medical and Biological Laboratories Co., Ltd. (Nagoya, Japan)) were dissolved into phosphate-buffered saline (PBS) to yield a 500 g/ml stock. Mouse monoclonal and rat polyclonal peroxidase-coupled anti-HA antibodies (obtained from Roche Applied Science) were dissolved in PBS to form a 400 g/ml stock. 3-[4,5-Dimethylthiazol-2-yl]-2,5-diphenyltetrazolium bromide (MTT) (Sigma) was dissolved in PBS to yield a 5 mg/ml stock, and cycloheximide (Sigma) was prepared as a 5 mg/ml stock. C 2 -Ceramide (Biomol, Plymouth Meeting, PA) was dissolved in dimethyl sulfoxide (Me 2 SO) to yield a 25 mM stock solution; N-methyl-N-nitro-Nnitrosoguanidine (MNNG) was prepared as a 10 mg/ml stock in Me 2 SO, and mitomycin C (Roche Applied Science) was prepared as a 5 mg/ml stock. Doxycycline (Clontech, Palo Alto, CA) was dissolved in PBS to a 1 mg/ml stock, and MG-132 (Calbiochem) was dissolved in Me 2 SO to yield a 50 mM stock. Each of the above reagents was aliquoted and stored at 20°C prior to use.
Plasmids-The pHA-E6 S and pHA-E6 AS plasmids have been described previously (23), and respectively contain either the sense or the antisense versions of epitope-tagged E6 (HA-E6) under the control of the CMV promoter.
The pTet-Off plasmid coding for tetracycline activator, pTRE-luc, coding for luciferase under the control of the tetracycline activator, pTK-Hyg, coding for hygromycin resistance, as well as the cloning plasmid pTRE2, were obtained from and described in the Tet-Off kit from Clontech. pTRE-HA-E6 was obtained by cloning the HindIII blunt end-BamHI fragment from pHA-E6 S into the EcoRI blunt end-BamHI sites of the pTRE2 vector.
pcDNA3-FADD was kindly provided by Dr. Carl Ware (La Jolla Institute for Allergy and Immunology, La Jolla, CA) and served as the basis for all our additional FADD-expressing constructs. The pM-FADD and pVP-FADD plasmids needed to perform the mammalian two-hybrid analysis (Clontech) were obtained by PCR amplification of FADD using primers 5Ј-GACCCGTTCCTGGTGCTG-3Ј (forward) and 5Ј-AT-GCCTGTGGTCCACCAGC-3Ј (reverse). This PCR product was then cloned into pCR-Blunt II-Topo using the Zero Blunt Topo PCR cloning kit (Invitrogen), and the EcoRI-EcoRI fragment subcloned into the pM DNA-BD and pVP AD EcoRI sites. Positive and negative control plasmids for this system were either provided in the kit (Clontech) or described earlier (23). pFLAG FADD was obtained by cloning the FADD HindIII-XbaI fragment from pTopoFADD into the pFLAG-myc-CMV-22 vector (Sigma).
To express FADD and its variants in both the Escherichia coli system and in the in vitro transcription/translation system (T7 reticulocyte) (Promega), the EcoRI-blunt end-XhoI fragment from pM-FADD, coding for FADD, was cloned in-frame with the His 6 epitope of pTriEx-4 (Novagen) by using the SmaI-XhoI sites of its multiple cloning site to produce the plasmid pHis-FADD. A similar approach was used to produce the two deletion fragments. To produce deletion D1 (lacking amino acids 23-62), the SacI-SacI fragment was removed from the FADD coding sequence. To produce deletion D2 (lacking amino acids 1-79), the SalI-blunt end-XhoI fragment from pM-FADD was cloned into the SmaI-XhoI sites of pTriEx-4. The point mutations were created in a D1 background using the QuickChange TM site-directed mutagenesis kit (Stratagene) according to the manufacturer's protocol. The sequences of the primers used for mutagenesis are presented in Table I.
To express the GST-tagged version of full-length version of E6 in E. coli, we created a plasmid coding for the fusion protein GST-E6 by cloning E6 (EcL136II-EcoRI blunt end fragment) into the SmaI site of pGEX-2T (Amersham Biosciences). For the small version, the EcoRI-BamHI fragment was cloned into the EcoRI-BamHI sites of the pGEX-2T vector. pM-E6, the plasmid coding for E6 in the context of the mammalian two-hybrid system, and pCMV-SEAP, coding for the reporter gene secreted alkaline phosphatase, have been described previously (23).
pHA-E6 large produces a transcript (ϳ480 bp) that cannot be spliced to yield the short processed message, and thus codes only for the large, full-length version of HA-E6. The construct was produced using pHA-E6 S as a template and then mutagenizing the donor splice site from AGGT to AGAT (52) by using the QuickChange TM mutagenesis kit (Stratagene). In contrast, pHA-E6 small produces only the short transcript and thus codes only for the short, half-length version of HA-E6. It was produced by isolating RNA from the stable cell line U2OSE612 (23) and then performing RT-PCR by using 5Ј-ATGATATCCTATCCATAC-GATGTT-3Ј and 5Ј-TGGGTTTCTCTACGTGTTCTTGAT-3Ј, which correspond to the beginning and end of both the short and long versions of E6. The PCR product of 0.2 kb was then cloned into pCR2.1 (Invitrogen), and its EcoRI-EcoRI fragment was then subcloned into the EcoRI site of the pE-CMV-1 vector (23).
Transfections-Transfections were carried out using FuGENE 6 (Roche Applied Science), as directed by the manufacturer. For transient transfections, cells were analyzed 40 -48 h post-transfection. For stable transfections, clones were passaged into selection medium containing G418 (500 g/ml) or hygromycin (100 g/ml). Following 3 weeks of selection, individual clones were grown and analyzed for protein expression by immunoblotting and/or RT-PCR. To normalize the transfection efficiency for the FADD degradation experiment, pCMV-SEAP was co-transfected with pcDNA3-FADD. The media were changed 48 h post-transfection. After an additional 3 h of incubation, during which SEAP was secreted into this fresh media, 20 l of the conditioned media was analyzed for SEAP activity by using the Great EscAPe chemiluminescence detection kit (Clontech) according to the manufacturer's protocol. A normalized amount of lysate, corresponding to 13 relative light units as measured on a ML3000 microtiter plate luminometer (Dynatech Laboratories), was then loaded per lane for separation by SDS-PAGE.
Immunoblotting-Cells (10 6 ) were lysed in 100 l of lysis buffer (50 mM Tris-HCl, pH 7.5, 150 mM NaCl, 1% Triton X-100, 1 mM EDTA, 5% glycerol, 1 mM dithiothreitol, 1 mM phenylmethylsulfonyl fluoride) for 10 min on ice. One tablet of protease inhibitor mixture (Roche Applied Science) per 10 ml of buffer was added just prior to use. The protein concentration in cleared lysates was measured using the Bio-Rad Protein Assay (Bio-Rad). Lysates (40 g total protein/lane) were then subjected to 12% SDS-PAGE and transferred to Immobilon P membranes (Millipore Corp.). After treating membranes with 5% milk in TBST (50 mM Tris-Cl, pH 7.5, 150 mM NaCl, 0.1% Tween 20), rat anti-HA peroxidase-conjugated antibodies (Roche Applied Science), anti-FADD antibodies (Santa Cruz Biotechnology), anti-GST monoclonal antibodies (Santa Cruz Biotechnology), or anti-␤-actin monoclonal antibodies (Sigma) were applied at a 1:1000 dilution (for the anti-␤-actin antibodies, a 1:5000 dilution was used) and allowed to incubate for 2 h, with rocking, at room temperature. After washing with TBST, peroxidase-coupled secondary antibodies were added for the anti-FADD, anti-GST, or anti-␤-actin detections. In the case of anti-FADD, anti-rabbit antibodies (Sigma) were used, whereas in the case of anti-GST and anti-␤-actin, anti-mouse antibodies (Pierce) were applied (1:2000 or 1:5000 dilution) and allowed to incubate with rocking at room temperature for 30 min. Membranes were then washed again with TBST. Detection of the protein was performed by using the chemiluminescent SuperSignal West Femto or Pico Maximum Sensitivity substrate (Pierce).
Immunoprecipitation-Cells (2-5 ϫ 10 6 ) were lysed in 500 l of the immunoblotting lysis buffer, and cleared lysates were incubated with 2 g of monoclonal anti-HA antibody (Roche Applied Science) at 4°C for 1 h with rotation. Protein A-agarose slurry (50 l) (Santa Cruz Biotechnology) was added to each lysate, and lysates were incubated for 3 h at 4°C. The protein A slurry was then washed three times with lysis buffer, followed by one wash with high salt buffer (50 mM Tris-HCl, pH 7.5, 0.5 M NaCl, 0.1% Nonidet P-40, 1 mM EDTA, 1 M dithiothreitol) and one wash with the same buffer lacking NaCl. The precipitates were fractionated by 12% SDS-PAGE, and immunoblotting was performed as described above.
Treatment of Cells with Anti-Fas, Mitomycin C, MNNG, Ceramide, MG-132, and Cycloheximide-To measure cell survival following anti-Fas treatment, cells were seeded into 96-well plates (1 ϫ 10 4 cells/well) and allowed to adhere overnight. Anti-Fas (at the concentrations noted in text and figures) was then added in the presence of cycloheximide (5 g/ml), and the cells were incubated for 16 h prior to measuring the number of viable cells by the MTT assay (described below).
To determine the ability of cells to accumulate increased levels of p53 following DNA damage, cells were treated with mitomycin C and assayed for the resulting p53 levels. Cells were seeded into 24-well plates (1 ϫ 10 5 cells/well) and allowed to adhere for ϳ24 h, treated with mitomycin C for 16 h, harvested, and the lysates analyzed for p53 by ELISA (described below).
To determine cell survival following MNNG or ceramide treatment, cells were seeded into 96-well plates (2 ϫ 10 4 cell/well, 50 l total volume) and allowed to adhere overnight. For the ceramide experiments, the medium was then removed and replaced with medium supplemented to contain 2% serum. Ceramide was then added to the indicated final concentration, and the cells were incubated for 24 h prior to measurement of the number of viable cells by the MTT assay. For the MNNG experiments, MNNG was added to a final concentration of 25 g/ml for 4 h prior to harvest and MTT assay.
To examine the amount of FADD in cells expressing E6, U2OSE6tet24 cells were seeded into 6-well plates (2 ϫ 10 5 cells/well) in the presence of the indicated concentrations of doxycycline. Following an overnight incubation, the cells were transfected with pcDNA3-FADD, and 40 h post-transfection were treated with MG-132 (50 M) or cycloheximide (5 g/ml) for the indicated time. Cells were harvested, and the lysates were analyzed by immunoblot using ␣-FADD antibodies.
Cell Viability Assay-Following treatment of cells as described above, the incubation medium was removed and exchanged with 80 l of fresh medium. Twenty microliters of MTT was then added (5 mg/ml stock), and cells were incubated at 37°C for 3 h. The medium was removed, and 150 l of Me 2 SO was added and allowed to incubate for 10 min. The solution was mixed by pipetting, and the absorbance of each well was measured at 490 nm.
Development and Use of the Tet-off System-U2OS cells capable of expressing variable amounts of HA-E6, regulated by the dose of drug present in the media, were created using the Tet-Off system (Clontech) following the manufacturer's protocol with some modifications. Cultures of these cells were grown in the indicated concentrations of doxycycline for 2 days prior to use, and the indicated concentrations of doxycycline were maintained in the media throughout each experiment.
p53 ELISA-Cells were assayed for p53 by ELISA essentially as described previously (23,53). The monoclonal antibody pAB122 (hybridoma obtained from ATCC; antibodies were purified from the culture medium using protein A-Sepharose) was used as the primary or capture antibody, and biotinylated anti-p53 (Roche Applied Science) was used as the detection antibody, and glutathione S-transferase-p53 (Santa Cruz Biotechnology) was used as a standard.
Reverse Transcriptase-PCR-The reverse transcriptase-PCR was used to analyze E6 transcripts in cell lines stably transfected with E6 AS, E6 S, E6 large , and E6 small . Three and a half micrograms of total RNA was isolated from each cell line using the TRIzol reagent (Invitrogen) and used as a template. cDNA was synthesized by using SuperScript TM II reverse transcriptase (Invitrogen) and an oligo(dT) primer (Amersham Biosciences). The primers 5Ј-ATGATATCCTATCCATACGAT-GTT-3Ј and 5Ј-TGGGTTTCTCTACGTGTTCTTGAT-3Ј were then used to amplify the PCR products from the cDNA, using one-twentieth of the total cDNA reaction mixture. To control for possible contamination by genomic DNA, parallel reactions were run using 0.175 g of total RNA in the absence of the reverse transcriptase enzyme. Reaction mixtures were separated on a 4.5% NuSieve GTG-agarose gel (FMC BioProducts).
Caspase 3 and 8 Assays-Cells were plated onto 100-mm plates at a density of 2-5 ϫ 10 6 cells/plate and incubated overnight. Anti-Fas (50 ng/ml) was added, along with cycloheximide (5 g/ml). Cells were harvested by scraping with a rubber policeman at designated intervals and were lysed in 160 l of Caspase 3 Lysis Buffer (Sigma). Protein concentrations were measured using the Bio-Rad Protein Assay (Bio-Rad).
Caspase 3 activity was measured using the Caspase 3 Colorimetric Assay kit (Sigma) as directed by the manufacturer. Cell lysates (25 l) were incubated with substrate (Ac-DEVD-pNA) in the presence or absence of the caspase 3 inhibitor Ac-DEVD-CHO. Absorbance at 405 nm was determined ϳ3 h following initiation of the reaction. The activity in wells treated with inhibitor was subtracted from the activity in untreated wells, and the activity was normalized to the amount of protein present in each sample. Caspase 3 activity in treated samples was expressed as a percentage of caspase 3 activity in the untreated parental cells. Three plates of treated and untreated cells were measured for each time point.
Caspase 8 activity was measured using the Caspase 8 Assay kit (Calbiochem) according to the manufacturer's instructions, using 50 l of cell lysate per well and using IETD-pNA as the colorimetric substrate, and in the presence or absence of the caspase 8 inhibitor Ac-IETD-CHO. Absorbance at 405 nm was determined ϳ6 h following initiation of the reaction, and calculations were performed as described for the caspase 3 assay.
Mammalian Two-hybrid Assay-The mammalian two-hybrid binding assay was performed according to the manufacturer's instructions (Clontech). The indicated combinations of vectors were transfected into U2OS cells (5 ϫ 10 5 /well, 6-well plates) along with the chloramphenicol acetyltransferase (CAT)-expressing reporter plasmid using FuGENE 6 (Roche Applied Science) as directed by the manufacturer. Forty eight h following transfection, CAT activity was measured colorimetrically using a commercially available CAT-ELISA kit (Roche Applied Science) as directed by the manufacturer.
The His-FADD, His-D1, His-D2, His-M3, and His-QQQ proteins, as well as those resulting from the other point mutations, were purified from E. coli lysates (from a 500-ml culture) using the B-PER His 6 Fusion Protein Purification kit (Pierce) according to the manufacturer's protocol. All proteins were soluble in the original elution buffer (which contains 250 mM imidazole). Following dialysis against a buffer of 20 mM HEPES, pH 7.4, 137 mM NaCl, 2 mM KCl, 5% glycerol, and 1 mM DTT, the His-D1, His-D2, His-M3, and other point mutant proteins remained soluble, although the full-length His-FADD protein precipitated. The expression of full-length His-FADD was achieved using the T7 TNT Quick Coupled Transcription/Translation System (Promega) according to the manufacturer's protocol. All proteins were separated by SDS-PAGE and stained with Coomassie Blue. The protein concentration for each preparation was estimated by using the Bio-Rad Protein Assay.
In Vitro Pull-down Assays-An in vitro pull-down approach was used to measure the ability of bead-bound GST or GST-E6 proteins (glutathione beads obtained from Sigma) to bind to either FADD or FLAG-FADD in mammalian cell lysates or to the transcription/translation product of full-length His-FADD. Cells (5 ϫ 10 6 ) transfected 48 h previously with the desired expression plasmid were lysed in 500 l of lysis buffer (50 mM Tris-HCl, pH 7.5, 150 mM NaCl, 1% Triton X-100, 1 mM EDTA, 1 M DTT, 1 mM phenylmethylsulfonyl fluoride, 1 protease inhibitor tablet/10 ml buffer (Roche Applied Science)), and the cleared cell lysate (or, in the case of the transcription and translation product, 10 l of reticulocyte lysate) was incubated with 30 l of either GST or GST-E6 beads (in PBS plus 2 mM DTT) on a rotator for 2 h at 4°C. Three washes were then done with lysis buffer, followed by one wash in high salt buffer and a final wash in buffer lacking NaCl as described under "Immunoprecipitation." Bound proteins were eluted in 30 l of 2.5ϫ SDS loading buffer and fractionated on 12% PAGE. Following transfer to polyvinylidene fluoride membranes (Millipore), FADD was detected using anti-FADD polyclonal antibodies (Santa Cruz Biotechnology).
To analyze the ability of bead-bound GST, GST-E6, and GST-E6 small to bind to purified His-FADD D1, His-FADD D2, His-FADD M3, His-FADD QQQ, and the other point mutants, the slurry (40 l) was incubated with 6 g of the indicated version of FADD in 500 l of 20 mM HEPES, pH 7.4, 137 mM NaCl, 2 mM KCl, 2 mM DTT, and 5% glycerol for 2 h at 4°C with rotation. Beads were then washed three times with the above buffer with the addition of 0.1% Nonidet P-40 (20 min, 4°C, rotation). Bound proteins were eluted in 30 l of 2.5ϫ SDS loading buffer and fractionated on 12% PAGE. Following transfer to polyvinylidene fluoride membranes (Millipore), FADD was detected using anti-FADD polyclonal antibodies (Santa Cruz Biotechnology). The same membrane was then stripped and reblotted with anti-GST monoclonal antibodies to demonstrate the presence of the GST and GST-E6 proteins. In addition, 0.6 g of the FADD D1 and D2 proteins was separated by SDS-PAGE and stained by Coomassie Blue to show the loaded protein.

Transfection of HPV 16 E6 into Human Cells Provides Protection from Fas-mediated Cell Death-Treatment of U2OS
cells with anti-Fas causes them to undergo apoptosis mediated by the Fas pathway. To determine whether E6 could protect cells from apoptosis triggered by Fas, as well as that induced by TNF, these cells were transfected with a plasmid coding for either the sense version of epitope-tagged E6 (HA-E6) (pHA-E6 S) or the antisense version (pHA-E6 AS). The stable pools of expressing cells were then treated with anti-Fas (Fig. 1A). The results demonstrate that expression of the sense version of E6, but not the antisense version, protected the cells from apoptosis induced by anti-Fas and that this protection could be seen at the 10, 50, and 200 ng/ml levels of anti-Fas. To examine this phenomenon in more detail, we then selected and expanded several stable clones. The clones were then screened for their expression of HA-E6 by immunoblot, and E6-expressing clones were tested for their response to treatment with anti-Fas. The results (Fig. 1B) confirmed that E6 has the ability to protect expressing cells from Fas-mediated apoptosis and demon-strated that there were significant differences in the amount of protection experienced by the different clones. Most interesting, the clone experiencing the greatest protection (U2OSE612) expressed a moderate amount of HA-E6, with less protected clones expressing both greater (for example, U2OSE66) and lesser (U2OSE62) amounts of the protein. One possible explanation was that moderate amounts of E6 do, in fact, provide optimal protection. One difficulty in interpreting these results is that each clone represents a separate integration event, and the variable sites of integration, with the accompanying disruption of the genes in these different regions, could complicate the observance of any dose dependence of this protection. Fig.  1C shows the morphology of untransfected U2OS cells, as well as of clones U2OSE6AS (transfected with the antisense version of HA-E6) and U2OSE612 (transfected with and expressing the sense version of HA-E6), both before and after treatment with anti-Fas. Together, these results indicate that HPV 16 E6 can protect cells from apoptosis triggered through the Fas pathway.
The Effect of E6 on Fas-induced Apoptosis Is Dose-dependent-To examine more carefully any dose dependence of this protection, we developed a regulated system in which the amount of E6, following one particular integration event, could be modulated by controlling the amount of tetracycline or doxycycline added to the culture media (TetOff System, Clontech). The results for two clones selected, U2OSE6tet24 and U2OSE6tet26, are shown in Fig. 2A. For clone U2OSE6tet26, the immunoblot shows a large amount of E6 produced in the absence of doxycycline that gradually decreases as the amount of doxycycline present in the culture medium rises, such that HA-E6 becomes undetectable by a level of 4 ng/ml dox ( Fig. 2A,  top panel). In the case of U2OSE6tet24, the immunoblot shows a more rapid drop in the level of HA-E6 present as the level of dox increases ( Fig. 2A, 2nd panel). Meanwhile, an immunoblot using antibodies directed against ␤-actin, using amounts of lysate corresponding to the input of the top two panels in Fig.  2A, showed no change. To verify that decreases in HA-E6 were continuing to occur, even at the higher levels of doxycycline, RT-PCR was used to detect the E6 transcript in these cells under these conditions. The results are shown in Fig. 2A, 3rd panel, and demonstrate that decreases in the amount of message continue to occur at the higher levels of doxycycline. They also demonstrate that even at a relatively high level of doxycycline (200 ng/ml), some HA-E6 message is still being made. An RT-PCR analysis of the actin message, used for normalization, is shown in Fig. 2A, bottom panel.
The p53 ELISA, which measures the ability of HA-E6 to cause p53 degradation, was then used to verify these results and to show that this particular activity of HA-E6 could be modulated by the amount of dox present in the culture media (Fig. 2B). Clearly, as the amount of dox decreases (Fig. 2B, rightmost 5 sets of bars), the amount of HA-E6 increases and the amount of cellular p53 present following genotoxic damage decreases. These results provide evidence that both the protein itself, as detected by its epitope tag, and a known biological activity of that protein, the ability to mediate degradation of cellular p53, can be modulated by the amount of dox present in the culture medium.
The next step was to test the sensitivity of the U2OSE6tet24 and U2OSE6tet26 cells to anti-Fas when grown in the presence of different concentrations of doxycycline. As shown in Fig. 2C, as the level of dox decreased from 100 to 0.4 ng/ml and the level of HA-E6 rose, the viability of cells exposed to anti-Fas (50 ng/ml) increased from about 30 (U2OSE6tet24) or 43% (U2OSE6tet26) to ϳ60%. There was no significant difference between the response of the two clones. These data provide strong evidence that not only does E6 expression protect cells from Fas-mediated apoptosis but that this protection is dosedependent and rises with increasing amounts of E6.
HPV 16 E6 Inhibits Activation of Caspase 3 and Caspase 8 -If HPV 16 E6 does protect cells by inhibiting one or more of the upstream steps in the Fas pathway, it should interfere with the activation of caspases 3 and 8. To test this prediction, U2OS cells transfected with either the sense (U2OSE612) or the antisense (U2OSE6AS) version of HPV 16 E6 were treated with ␣-Fas and then analyzed for the activation of the two caspases using colorimetric assays that respond to cleavage of the appropriate substrate. Fig. 3A shows that whereas cells expressing the antisense version of E6 respond to anti-Fas treatment with a robust increase in caspase 3 activation within 4.5 h, cells transfected with the sense version of E6 (U2OSE612) do not. The results from the caspase 8 experiment (Fig. 3B) were similar with respect to the trend but to a lesser degree. These HA-E6 protein levels were analyzed by immunoprecipitation followed by immunoblot using polyclonal anti-HA (rat, peroxidase-conjugated) (top and 2nd panels). The input amounts for immunoprecipitation were equivalent as shown by the immunoblot of ␤-actin (15 g/lane). Message levels were analyzed by RT-PCR (4th and 5th panels). B, the ability of E6 to mediate p53 degradation is dose-dependent. Untransfected U2OS cells or cells stably transfected with HA-E6 under the control of the tet-responsive element were cultured in the presence of the indicated concentrations of doxycycline and treated with mitomycin C (2 g/ml) for 16 h. Lysates were made, and the level of p53 was measured by ELISA. The level of p53 was determined by dividing the calculated value by the total protein concentration and then subtracting the amount of p53 in the corresponding untreated samples. Each measurement was done in triplicate, and error bars indicate means Ϯ S.D. C, protection from anti-Fas provided by HA-E6 is dose-dependent. Untransfected U2OS cells or cells stably transfected with HA-E6 under the control of the tet-responsive element were cultured in the presence of the indicated concentrations of doxycycline and treated with anti-Fas (50 ng/ml) for 16 h. Cell viability was measured using the MTT assay. Each measurement was done in triplicate, and error bars indicate means Ϯ S.D. results indicate that E6 acts on the Fas-mediated pathway at or prior to activation of these two caspases and suggest that it may function by interacting with the DISC.
In addition to its actions on the DISC, E6 could also be acting on the downstream parts of the apoptotic pathway. Apoptosis mediated by the mitochondria shares downstream steps with receptor-mediated apoptosis and can be triggered by such factors as the addition of lipid mediators (for example, ceramide) and genotoxic damage (by way of p53 induction). Fig. 4A shows the results following the addition of ceramide, which demonstrate that treated cells experience significant cell death, independent of the amount of E6 expressed. In the case of the tet-regulated cells, the amount of p53 available to mediate apoptotic signals from DNA-damaging agents can be regulated by the amount of dox added to the media. Therefore, one would expect this particular pathway to be most available in cells expressing none or low amounts of E6. The results (Fig. 4B) show that treating cells with the DNA-damaging agent MNNG results in significant cell death, independent of the amount of E6 expressed. Therefore, E6 does not act at the level of the mitochondria or the downstream apoptotic steps.
HPV 16 E6 Binds to FADD-The experiments described above, along with our previous work with the TNF pathway, suggested that E6 might bind to one or more upstream members of the apoptotic pathways. Therefore, its ability to interact with FADD was examined. cDNAs encoding E6 and the death effector domain of FADD or death domain of TNF R1 were therefore cloned into the bait and prey plasmids of a mammalian two-hybrid system. Each set of test plasmids was transfected into U2OS cells along with a reporter, CAT-expressing plasmid, and expression of the CAT gene was measured colorimetrically using a CAT-ELISA kit. Expression of CAT under these conditions indicates an interaction between the two test proteins. The binding of E6 to TNF R1 death domain served as a positive control. The results (Fig. 5A) show that E6 can indeed bind to FADD, although the signal is not as strong as that seen for the binding of E6 to TNF R1. In these experiments, neither plasmid alone was capable of inducing expression of CAT. Most interesting, this and other assays (not shown) indicated that any binding of E6 to TRADD was not statistically significant (the 90% confidence interval included the value of 0), demonstrating some specificity of E6 with regard to the apoptotic signaling molecules to which it binds.
To verify these results, in vitro pull-down assays were performed. E6 was expressed as a GST fusion (GST-E6), and both this fusion product and GST alone were purified using glutathione beads. Lysates known to contain FADD were prepared from U2OS cells transfected with the plasmid pcDNA3-FADD. This lysate was then incubated with glutathione beads complexed with either GST alone (Fig. 5B, left, top panel, lane 1) or with GST-E6 (lane 2). Beads were then washed and bound proteins eluted with SDS prior to separation by SDS-PAGE. FADD was detected by Western blot using polyclonal anti-FADD. The results show that FADD bound to the beads com-

FIG. 3. Caspase 3 and 8 activation is suppressed in HA-E6-expressing cells.
A, U2OSE6AS (coding for the antisense version of E6) and U2OSE612 (coding for the sense version of E6) cells were treated with anti-Fas (50 ng/ml) in the presence of cycloheximide (5 g/ml) for the times indicated and then lysed. Lysates were analyzed for caspase 3 activity using Ac-DEVD-pNA as the substrate in the presence and absence of the caspase 3 inhibitor Ac-DEVD-CHO. The activity in wells containing the inhibitor was subtracted from that in wells lacking the inhibitor and then normalized for the amount of protein added to each well. Activity is expressed as the percentage of caspase activity in untreated U2OSE6AS cells. Each time point was measured in triplicate, and error bars represent the means Ϯ S.D. The Student's one-tailed t test was used to determine statistical significance, with * (B) representing a Ͼ0.95 level of confidence and ** (A) representing a Ͼ0.99 level of confidence. B, U2OSE6AS and U2OSE612 cells were treated and lysed as described in A and then analyzed for caspase 8 activity using IETD-pNA as the substrate in the presence and absence of the caspase 8 inhibitor Ac-IETD-CHO. Data were analyzed as described in A.
plexed with GST-E6 but not to those complexed with GST alone. Fig. 5B, bottom panel, shows the same membrane stripped and reblotted with anti-GST. To confirm these results, a similar experiment using FLAG-tagged FADD, shown in the two right panels of Fig. 5B, was performed. In this experiment, FLAG-tagged FADD (FLAG-FADD) was transfected into the cells, and the bound protein was detected using anti-FLAG. These two pull-down assays, as well as the mammalian twohybrid assay, were all performed either inside intact cells or in the presence of cellular lysate. We were also able to demonstrate binding using His-FADD generated by in vitro transcription and translation (Fig. 5C). Together, these results provide strong evidence that E6 binds to FADD.
The E6 Gene Produces Three Messages and Two Proteins-The E6 gene produces at least three messages because of the differential splicing of mRNA (52, 54) (Fig. 6A). The large transcript (ϳ0.48 kb) codes for the full-length protein of ϳ16 kDa, whereas the two smaller transcripts (of 0.3 and 0.2 kb) code for very similar proteins of approximately half the length of the longer E6. To determine whether the regions of E6 retained in the small version were sufficient for binding to FADD, constructs that coded exclusively for either the short or the long version were created and used to analyze the FADD binding and protective ability of the two E6 proteins. To determine which transcripts were actually generated in expressing cells, the various versions of E6 were stably transfected into cells and the transcripts examined using RT-PCR (Fig. 6B). As anticipated, cells transfected with the intact gene express both the large and small transcripts, whereas cells given only the large or small gene express only the expected transcript. To determine whether the protein associated with the two shorter transcripts of 0.3 and 0.2 kb (approximately a 7-8-kDa protein) was actually made, the vector expressing the HA-tagged version of the small protein was transiently transfected into U2OS cells, and the lysates were analyzed by immunoblot (Fig. 6C). In addition to the expected band at ϳ7 kDa, some faster migrating species were also observed and may represent degradation products.  5. HPV 16 E6 binds to FADD. A, mammalian two-hybrid assay. Sequences encoding E6 and the death domain of TNF R1, FADD, and TRADD were cloned into the bait or prey plasmids of a mammalian two-hybrid assay kit (Clontech). The indicated combinations of plasmids were then transfected into U2OS cells, along with a reporter plasmid coding for CAT, and expression of the CAT gene was measured colorimetrically by using a CAT-ELISA kit. Each of the three experimental values was measured in three independent experiments, with two measurements taken for each experiment. The error bars show the 90% confidence intervals. B, in vitro pull-down assays using cell lysates. GST-E6 beads were used to pull down FADD, which was detected with anti-FADD antibodies (left set) or FLAG-FADD, which was detected with anti-FLAG antibodies (right set). Top panels, lysates were prepared from U2OS cells transiently transfected with either pcDNA3 FADD (left) or pFLAG-FADD (right) and then incubated with glutathione beads complexed with either GST (lane 1) or GST-E6 (lane 2). Beads were washed, and bound proteins were eluted with SDS prior to separation by SDS-PAGE. FADD was detected by immunoblot using polyclonal anti-FADD and FLAG-FADD with monoclonal anti-FLAG. Bottom panels, the same membrane shown in the top panel was reblotted using monoclonal anti-GST. C, in vitro pull-down assay using transcription/translation-produced FADD. The experimental design was as described above, with the exception that the source of FADD was full-length FADD protein expressed using the T7 TNT Quick Coupled Transcription/Translation System (10 l of reticulocyte lysate) rather than cell lysates.
The Large, but Not the Small, E6 Protein Binds to FADD and Protects Cells from Fas-mediated Apoptosis-To determine whether the small protein is capable of binding to FADD, an in vitro pull-down assay using GST-E6 small coupled to beads was performed. The results (Fig. 6D) demonstrate that although the larger version of E6 is able to bind to the D1 version of the  1 and 5), the unmodified full-length gene (lanes 2 and 6), only the large version of E6 (lanes 3 and 7), or only the small E6 version (lanes 4 and 8) and the messages produced detected by PCR. Primers corresponded to the 5Ј and 3Ј ends of E6, and no reverse transcriptase was included in the reactions corresponding to lanes 1-4. C, the small E6 protein is produced in transfected cells. Lysates were prepared from U2OS cells (lane 1) or from U2OS cells transiently transfected with a construct coding for the small version of E6 (HA-E6 small ) and immunoprecipitated using beads cross-linked to antibodies directed against HA. Following elution from the beads, separation by SDS-PAGE, and transfer to a membrane, HA-E6 small was visualized by immunoblotting. D, HPV 16 E6 small does not bind to FADD. Bacterially expressed and purified D1 FADD, lacking amino acids 24 -62, was incubated with glutathione beads bound to either GST E6 small or GST E6 large , as indicated. Following elution from the beads, separation by SDS-PAGE, and transfer to a membrane, the membrane was blotted with anti-FADD. The bottom panel shows the same membrane reblotted with anti-GST antibodies. E, HPV 16 E6 large , but not HPV 16 E6 small , protects U2OS cells from apoptosis induced by Fas. U2OS cells were stably transfected with either HA-E6AS, HA-E6 S, HA-E6 large , or HA-E6 small . Pools of stably transfected cells were then treated with 100 ng/ml anti-Fas in the presence of cycloheximide (5 g/ml). Following a 16-h incubation, cell viability was measured by the MTT assay. The error bars represent the means Ϯ S.D. of the average of five repeats. FADD DED (see scheme in Fig. 7 and description below), the small version of E6 is unable to do so. To answer the question of whether the small version of E6 is able to protect cells from Fas-mediated apoptosis, U2OS cells were transfected with plasmids coding for the antisense version of HA-E6 (HA-E6 AS), the sense version of HA-E6 (wild type) (HA-E6 S), HA-E6 coding for only the large protein (HA-E6 large ), or HA-E6 coding for only the small protein (HA-E6 small ). Pools of stable transfectants were then treated with anti-Fas (Fig. 6E), and the results indicate that although the wild-type version of HA-E6, which yields all three transcripts and both proteins, provides the most effective protection for expressing cells from anti-Fas, protection is also achieved by the plasmid coding for HA-E6 large . However, the level of cell death resulting from anti-Fas treatment is nearly the same in cells transfected with HA-E6 small as it is for cells transfected with the antisense version of HA-E6. These results indicate that although the large, fulllength version of HA-E6 can both bind to FADD and provide protection from Fas, the smaller, truncated version cannot do either, implicating the C terminus of E6 in these functions.
The N terminus of the FADD DED Is Important in Binding to E6 -To begin mapping the region of FADD required for binding to E6, several mutant FADD proteins were bacterially expressed, purified, and then analyzed for their ability to bind to E6. Fig. 7 shows a schematic of the mutants used as well as a summary of our results. Each of these mutants was based on the FADD DED. The first two mutants tested were deletion versions of the FADD DED. The first, D1 (deletion 1), lacks the coding sequences for amino acids 23-62, and the second, D2, lacks amino acids 2-79. The results (Fig. 8A) demonstrate that although binding was achieved with the D1 deletion, it did not occur with the D2 deletion mutant. For this experiment, the purified FADD (Fig. 8A, D1 in lane 2, D2 in lane 3, a point mutant called QQQ (described below) in lanes 1 and 4) proteins were incubated with glutathione beads bound to either GST (lane 1) or GST-E6 (lanes 2-4). In Fig. 8A, the right-most three lanes were loaded with the input amounts of D1, D2, and QQQ. Immunoblotting of the bound proteins demonstrated the presence of a band corresponding to D1 FADD in Fig. 8A, lane 2, showing that GST-tagged E6 can pull down a FADD protein lacking amino acids 23-62 in the absence of other cellular factors. The absence of a band corresponding to D2 FADD in Fig. 8A, lane 3, suggests that amino acids 1-22 and/or 63-79 are required for this binding. The membrane was then reblotted with anti-GST (Fig. 8A, 2nd panel), providing evidence that approximately equal amounts of GST-E6 were present in the samples.
Whereas these results suggest that the important sites for binding are located within amino acids 1-22 and/or 63-79, they do not provide proof of this conclusion, as it is difficult to discriminate between an actual loss of binding domains and a more global alteration in three-dimensional structure when using deletion mutations. The binding residues were therefore further defined by site-directed mutagenesis. Each point mutation (summarized in Fig. 7) was based on D1, as this construct binds strongly to E6 and therefore contains the relevant residues. The first two mutations, FADD D1 (D74A) and FADD D1 (E61A and E65A), were designed to alter an abbreviated version of a putative E6-binding motif called the L2G box. The motif itself is defined as (E/D)L(L/V)G, and although the classical form does not exist in the FADD DED, an abbreviated version of the motif ((E/D)LL) appears in three places (amino acids 61-63, 65-67, and 74 -76). Unexpectedly, however, each of these two mutants bound to E6 in pull-down assays with an affinity equivalent to the parental D1 protein (data not shown). The next two mutants, FADD D1 (L5Q, V6A, L7Q, L8Q) and FADD D1 (L66Q, L75Q, L76Q) (QQQ) were designed to disrupt hydrophobic sequences in either the 1-23 or the 63-79 regions, based on the possibility that such stretches were important in binding. However, neither of these two mutants demonstrated impaired binding. The results for one of these mutants, QQQ, are included in Fig. 8A (lanes 1, 4, and 7) and are representative of the results obtained for each of these four mutants. Finally, a fifth mutant was prepared (FADD D1 (S10A, S14G, S16A, S18G, and E19Q)), also known as M3, that does demon-strate significantly reduced binding (Fig. 8B), although binding is not eliminated. This experiment was structured as described above for Fig. 8A, with the exception that the incubation loaded into lane 1 of the upper panel contained a mixture of the D1, D2, and M3 FADD proteins incubated with GST-E6 beads. The input of the His FADD proteins is shown in lanes 5-7. In Fig.  8B, 3rd panel, a Coomassie stain of one-tenth of the amount of the FADD proteins included in each incubation demonstrates the purity of the proteins used in this assay, and Fig. 8B, bottom-right panel, demonstrates that both D1 and D2, as well as the intact His-FADD, can be detected by these anti-FADD polyclonal antibodies. A final deletion mutation, in which amino acids 79 -208 were deleted (thus containing the DED but  2 and 5), D2 (lanes 3 and 6), or QQQ (lanes 1, 4, and 7) versions of FADD were incubated with glutathione beads bound to either GST (lane 1) or GST-E6 (lanes 2-4). Following elution from the beads, separation by SDS-PAGE, and transfer to a membrane, the membrane was blotted with anti-FADD.  2-4). Following elution from the beads, separation by SDS-PAGE, and transfer to a membrane, the membrane was blotted with anti-FADD. Lanes 5-7 show that approximately equivalent amounts of the indicated FADD variants were used for each incubation. 2nd panel, the same filter as described above was reblotted with antibodies directed against GST. 3rd panel, approximately 10 g of the indicated purified proteins were separated by SDS-PAGE and stained with Coomassie Blue. 4th panel, cells were transiently transfected with the pTriEX-4 vector coding for either the intact protein (Full) or one of the two deletions (D1 or D2) and then lysed. The lysates were separated by SDS-PAGE, transferred to a membrane, and then immunoblotted with antibodies directed against FADD. not the DD of FADD), also was shown to bind to E6 (data not shown).
Together, these data demonstrate that E6 binds to FADD without the requirement of additional cellular factors, that amino acids 23-62 and all those from 79 to 208 are dispensable for this binding, and that a major contribution to this binding is made by serines 10, 14, 16, and 18 and/or glutamic acid 19. These results therefore provide a molecular explanation for the protection from Fas-mediated apoptosis provided to cells by E6.
HPV 16 E6 Mediates the Rapid Degradation of FADD-HPV 16 E6 mediates the rapid degradation of some, but not all, of the proteins to which it binds, leading to a decrease in the steady-state level of those proteins. In previous work, our laboratory found that whereas HPV 16 E6 bound to TNF R1, it did not affect its level or accelerate its degradation (23). To determine whether E6 affected the steady-state level of FADD, U2OS and U2OSE6tet24 cells were transiently transfected with pcDNA-FADD, coding for FADD, and the U2OSE6tet24 cells were incubated in the presence of increasing amounts of doxycycline, thus experiencing the expression of decreasing amounts of HA-E6. Normalized amounts of lysates were then analyzed by Western blot using antibodies directed against FADD. As shown in Fig. 9A, when little or no E6 was expressed (lanes 2 and 7), FADD was readily detected. However, as the level of E6 was increased by lowering the amount of doxycycline present in the media (Fig. 9A, lanes 6, 5, 4, and 3), the level of FADD decreased. In principle, these results could be due either to an E6-mediated increase in FADD degradation or to a decrease in FADD transcription. If an increase in proteosome-mediated degradation were the mechanism, one prediction is that the inclusion of a protease inhibitor, such as MG 132, would attenuate this loss. Performing this experiment demonstrated that when MG 132 was added to U2OSE6tet24 cells in the absence of dox (hence expressing high levels of E6), the amount of cellular FADD did indeed increase (Fig. 9B). A second prediction is that if protein synthesis were blocked with cycloheximide, the levels of FADD would decrease more rapidly in the presence of E6 than in its absence. The results shown in Fig. 9C demonstrate that this is, in fact, the case. In the presence of 500 ng/ml dox (and absence of E6), very little FADD is lost within a 90-min incubation following cycloheximide addition, whereas in the presence of 0.4 ng/ml dox (and hence the presence of E6), the level of FADD decreases by ϳ50%. Together, these results provide strong evidence that HPV 16 E6 can indeed lower cellular levels of FADD and that it does so by accelerating the degradation of FADD. DISCUSSION To survive and propagate, viruses have developed numerous ways to avoid destruction by the host immune system. The means by which they have done so are now known to encompass a wide ranging and diverse set of molecular mechanisms that can target several different steps of multiple immune pathways. Some viruses encode proteins that disable only one or a few of these mechanisms, whereas others encode one or several proteins that can systematically target cellular defenses at several levels (reviewed in Refs. 55 and 56). Several viruses have developed strategies to block Fas-mediated host responses (57). For example, adenovirus codes for two proteins (10.4 and 14.5 kDa) that form a complex called RID, which mediates the rapid internalization and degradation of Fas from the cell surface (58). Virus proteins also act at the level of caspase activation. For example, the bovine herpesvirus-4 BORFE2 protein inhibits both Fas and TNF-induced apoptosis. This protein contains death effector domains and interacts with caspase 8 (59), a similar mechanism as that employed by the equine herpesvirus E8 protein, which also interacts with FADD (60,61), and by the molluscum contagiosum MC159 protein (60 -63). The human cytomegalovirus vICA (64) and the R1 subunit of herpes simplex virus ribonucleotide reductase (65) proteins also prevent caspase-8 activation induced by Fas.
The ability of human papillomaviruses to modulate host apoptotic responses has recently been gaining recognition (reviewed in Ref. 66). For example, some studies have found that cervical carcinoma cells positive for HPV 16 and HPV 16immortalized nonmalignant human keratinocytes are relatively resistant to apoptosis induced by Fas L (67,68), although these studies did not identify the particular viral genes responsible. Other studies have focused on the roles specific genes may play in immune evasion. For example, the E5 protein can reduce the amount of the CD95 receptor at the cell surface (69), and both the E6 and E7 genes have been reported to modulate the cellular response to various apoptotic signals, although the reported effects differ, with some groups reporting sensitization (70 -73) and others reporting protection (23,37,74). It may well be that the effect of E6 and E7 on a given cell will depend on the cellular environment as well as on the nature of the apoptotic signal. Our previous finding that HPV 16 E6 could protect cells from TNF-mediated apoptosis (23) led to the speculation that it could also block apoptotic signaling through the closely related Fas pathway. Treatment of E6-expressing stable transfectants with anti-Fas led to the conclusion that E6 could in fact protect cells from Fas-mediated apoptosis (Fig. 1). The mechanism for this protection was then examined and led to the finding that HPV 16 E6 binds to FADD, accelerates its degradation, and thus makes it unavailable for the transduction of apoptotic signals.
The development of an inducible system for E6 expression, in which the amount of E6 could be regulated by varying the amount of doxycycline present in the media, allowed confirmation of this protective effect of E6 and the determination that it was dose-dependent, with higher levels of E6 providing greater levels of protection, at least up to some point (Fig. 2). This is consistent with a model in which HPV 16 E6 binds to FADD and mediates its degradation (Fig. 9); as the level of E6 increases, correspondingly increasing numbers of FADD molecules can be targeted for degradation. The data from the stable clones (Fig. 1B) suggest that at very high levels of E6, likely higher than those achieved with the tet-regulated system, the protective effect of E6 may be lost, perhaps because these higher amounts of E6 participate in additional interactions that cancel out its protection. Another possibility, of course, is that these clones with very high levels of E6 may not be entirely representative, due to additional and uncharacterized events. Most interesting, even at high (200 g/ml) levels of doxycycline, some HA-E6 message was being produced, presumably leading to protein production, although the level is too low to allow detection by immunoblot ( Fig. 2A). This may be due to the regulatory environment of the genomic sites into which the constructs integrated and may explain why, even at high levels of doxycycline, the U2OSE6tet24 and U2OSE6tet26 cells experience some protection from anti-Fas as compared with the U2OS cells (Fig. 2C, far left data point).
Both genotoxic damage (initiated by MNNG) and ceramide lead to apoptosis through the mitochondrial pathway, which joins the receptor-mediated pathways at the level of caspase 3 activation. Each of these agents was able to efficiently initiate cell death in U2OSE6tet24 cells, regardless of the concentration of doxycycline in which they were cultured (Fig. 4). This meant that E6 targets the apoptosis signal at or upstream of caspase 3 activation in the Fas pathway. This is consistent with the findings that both caspase 3 and caspase 8 activities were diminished in cells expressing E6 (Fig. 3) and supports a model in which E6 interacts with the DISC. It is unclear why greater increases of caspase 3 than of caspase 8 were observed following treatment with anti-Fas, as both caspases are expected to function in the death pathway. One factor may be that in our hands the caspase 3 assay appears to be more sensitive than the caspase 8 assay. Alternatively or in addition, these findings suggest the possibility that players in this pathway may have connections and interactions that are not accounted for in a simple, unidirectional model (75,76).
A molecular explanation for this decreased apoptotic signal was provided by the finding that E6 binds to FADD. This binding was demonstrated by the mammalian two-hybrid system and three types of in vitro pull-down assays (Figs. 5 and 8). Using purified proteins (GST-E6 and His-tagged versions of FADD) (Fig. 8) allowed the conclusion that the binding of E6 to FADD was direct and did not require additional factors. In addition, the use of the different variants of FADD (D1, D2, and M3) allowed us to localize the region(s) of FADD required for binding to E6 to the N terminus of FADD and implicated amino acids 10, 14, 16, 18, and/or 19. It has been reported previously that both Fas-and TNF-mediated apoptosis use the same binding surface of FADD to trigger signal transduction and that binding surface includes helices ␣2 and ␣3 of the FADD death domain (77). However, the implicated stretch required for binding to E6 is N-terminal to these two helices, suggesting that the E6-binding site is distinct from the sites required for interac-tions with their apoptotic signaling partners. Although E6 binds to both TNF R1 and to FADD, it does not indiscriminately interact with all apoptotic signaling molecules. For example, it binds only weakly to TRADD (Fig. 5A and data not shown), demonstrating specificity for its binding partners.
Including FADD, the number of identified E6-binding cellular proteins has now reached nearly two dozen, and the molecular basis for these protein-protein interactions is beginning to be understood. Two E6-binding motifs have thus far been identified. One occurs in the PDZ domain of proteins such as hDlg, hScrib, MAGI-1, and MUPP1 and binds to the ETZV sequence at the far C terminus of E6 (reviewed in Ref. 12). The presence of this PDZ ligand domain is likely important in the targeting of these proteins for degradation by E6 (78). The second motif is known as the L2G box and was initially identified by Elston and co-workers (79), who used a yeast two-hybrid system to screen a 16-mer peptide library for peptides that specifically interact with HPV 16 E6. From this work, they identified an E6-binding motif of (E/D)L(L/V)G, which was also found in the known E6-binding proteins E6-AP and E6-BP. Further work (80) has established that the actual consensus sequence is LhXLsh (where h is an amino acid residue capable of accepting hydrogen bonds (Asp, Glu, Gln, or Asn), X refers to any amino acid, is hydrophobic, and s denotes a small amino acid residue such as G or A) and that this region must be helical to bind to E6. It is found in a number of cellular targets for E6, including E6AP, hMcm7, E6BP, paxillin, and TNF R1 (16,23,79,81). However, it is worth noting that the classical L2G box motif is not present in the death effector domain of FADD, which does bind to E6, and mutagenesis of the three copies of an abbreviated version ((E/D)LL) did not diminish the binding activity. Our identification of the four serines and/or one glutamic acid as important for binding affinity (Figs. 7 and 8) suggests the presence of a novel E6-binding motif present in FADD, possibly interacting with a different region or face of the E6 protein. Further work will be required to further define and establish a consensus for this new motif.
The fact that more than one transcript is produced from the E6 genes of oncogenic strains of HPV has been known for some time (52,54,82,83). Initially, it was thought that the alternatively spliced versions of the E6 message might function to enhance the translation efficiency of E7 (52), but more recent work suggests that the shorter E6 protein is actually produced and may function by binding to and inhibiting the biological activities of the full-length protein. For example, the truncated version of HPV 18 E6 can bind to the full-length versions of HPV 18 E6, HPV 16 E6, and E6AP, inhibits (full-length) E6mediated degradation of p53, and increases E6-mediated transcription and growth arrest in E6-expressing cells (82). It has also been shown that the ratio of the full-length and small versions of E6 may vary in cytologically abnormal cells (84), and this observation may be linked to the ability of the short E6 to interact with proteins such as E6 and E6-AP (66). We have been able to demonstrate production in cells of both the message and protein for the small E6 protein, and we have shown that the large and small versions differ in function, in that only the large version is capable of binding to FADD and protecting cells from Fas (Fig. 6).
It is intriguing that E6 accelerates the degradation of FADD (Fig. 9) although not of TNF R1, reflecting the variation seen in the fate of many of the proteins to which E6 binds. In this respect, it may be relevant to note that whereas E6 binds to the DD of TNF R1, it binds to the DED of FADD even though FADD contains both a DD and a DED. Whereas these two regions do share some homology, they differ sufficiently to be considered distinct motifs. Understanding the molecular basis for this difference in the fate of bound proteins will be an important area for future investigation.
This degradation of FADD is likely to have biological consequences. It has been reported, for example, that loss of FADD expression is correlated with the development of tumoral status in thyroid follicular cells (85) and that inhibition of FADD expression using antisense technology reduced keratinocyte apoptosis triggered by UV irradiation (86) as well as a failure of carboplatin-induced cell death in human tongue carcinoma cells (87). In our system, of course, the E6-mediated degradation of FADD may also contribute to the resistance of cells to TNF, as FADD participates in both pathways.
Papillomaviruses are persistent viruses that remain in their hosts for long periods. The papillomavirus-specific immune response is either weak or undetectable, and little or no inflammatory response is elicited by papillomavirus infection. The capacity of E6 to protect cells from Fas and from TNF may be an important factor in this lack of a host inflammatory response and thus contribute both to the persistence of this virus and to its oncogenic potential. Moreover, the findings by our laboratory and others that many important signaling and regulatory proteins in cells, such as FADD, can bind to E6 suggest additional approaches for the development of therapeutic agents for cervical cancer. For example, it has been reported recently that when RNA i was applied to HeLa cells, which express HPV 18 E6 and E7 genes, the cells underwent senescence (88) and that apoptosis could be induced in HPV-positive cells by peptide aptamers targeting E6 (89). Other groups (90) have focused downstream of E6 and found that restoring p53 expression in HPV 16-immortalized keratinocytes can sensitize them to Fas-triggered apoptosis. Such results suggest that many of the deleterious effects of HPV can be blocked by abrogating the expression and activities of E6 and E7. For this reason, it seems possible that small molecules capable of interfering with the binding of E6 and E7 to their cellular partners might be therapeutically useful. In particular, such molecules that decrease the binding of E6 to FADD could increase the sensitivity of cervical cancer cells to cells bearing FasL or to therapeutically applied agents that trigger its signaling pathway.