The HPV E7 oncoprotein inhibits tumor necrosis factor a ... - Nature

Oncogene (2001) 20, 3629 ± 3640 ã 2001 Nature Publishing Group All rights reserved 0950 ± 9232/01 $15.00 www.nature.com/onc

The HPV E7 oncoprotein inhibits tumor necrosis factor a-mediated apoptosis in normal human ®broblasts David Alan Thompson1,2,4,7, Valerie Zacny1,7, Glenn Scott Belinsky2, Marie Classon3, Dana Leanne Jones1,5, Robert Schlegel2,6 and Karl MuÈnger*,1 1

Department of Pathology and Harvard Center for Cancer Biology, Harvard Medical School, 200 Longwood Avenue, Boston, Massachusetts, MA 02115, USA; 2Department of Cancer Cell Biology, Harvard School of Public Health, 665 Huntington Avenue, Boston, Massachusetts, MA 02115, USA; 3Massachusetts General Hospital Cancer Center, Charlestown, Massachusetts, MA 02129, USA

Tumor necrosis factor-a (TNF) is a cytokine that induces programmed cell death, apoptosis, in a number of cell types and is employed by cytotoxic T cells to eliminate virus infected cells. Consequently, many viruses have acquired mechanisms to undermine these host cell defense mechanisms and cause resistance to TNFmediated apoptosis. Here we show that normal human diploid ®broblasts that express the human papillomavirus type 16 E7 oncoprotein have a decreased propensity to undergo apoptosis in response to TNF treatment. The ability of E7 to undermine TNF-mediated apoptosis correlates with cellular transformation. While E7 does not generally subvert signaling by tumor necrosis factor receptor 1, pro-caspase 8 activation is decreased in E7expressing cells. E7 also provides some protection from apoptosis caused by stimulation of the TNF receptor 1related cytokine receptor Fas, where induction of apoptosis occurs much slower in this cell type. Hence, E7-expressing normal human ®broblasts exhibit a speci®c defect that obstructs cytokine-mediated activation of pro-caspase 8 and apoptosis. Oncogene (2001) 20, 3629 ± 3640. Keywords: apoptosis; tumor necrosis factor-a; caspase 8; human papillomavirus Introduction Tumor necrosis factor-a (TNF) and related cytokines play important roles as mediators of antiviral immunity (reviewed in Ramshaw et al., 1997). TNF preferentially kills virally-infected cells, although TNF is also able to protect cells from viral infection in the absence of a cytocidal eect (reviewed in Wong et al., *Correspondence: K MuÈnger Current addresses: 4Department of Entomology, University of California, Davis, CA 95616, USA; 5Department of Developmental Biology, Stanford University, Stanford, CA 94035, USA; 6 Millennium Predictive Medicine, Cambridge, MA 02139, USA 7 These two authors contributed equally to this work Received 13 November 2000; revised 20 March 2001; accepted 26 March 2001

1992). TNF stimulates apoptosis of many cell types in vitro, although this often requires inhibition of protein synthesis (Kirstein et al., 1986; Reid et al., 1989). The cellular activities of TNF are mediated by its interactions with TNF receptor 1 (TNF R1) and TNF R2, members of a larger receptor superfamily characterized by a conserved extracellular cysteine-rich motif. TNF R1, a 55 ± 60 kDa death-domain-containing TNF receptor, directly transduces the TNFinitiated apoptotic signal, most likely through ligandinduced receptor trimerization (Banner et al., 1993; D'Arcy et al., 1993; Eck and Sprang, 1989; Vandevoorde et al., 1997). TNF R2, a 70 ± 80 kDa TNF receptor lacking a death domain, may play an important role in modulating the strength of the TNF signal, possibly through a ligand-passing mechanism (Tartaglia et al., 1993). For TNF R1-mediated apoptosis signaling, binding of TNF to the receptor facilitates association of the receptor cytoplasmic tail with the intracellular death domain protein, TRADD. TRADD interacts with FADD, a death domain protein also capable of direct interaction with the death domain of the TNF R1-related cytokine receptor Fas. The interaction of TRADD and FADD is mediated by death domains located within the carboxyl termini of TRADD and FADD. Caspases are cysteine proteases activated through precise proteolytic cleavage of latent inactive procaspases at internal aspartate residues and are characterized by caspase-speci®c substrate speci®city (reviewed in Cryns and Yuan, 1998). Caspase 8 acts as the apical caspase in the TNF and Fas pro-apoptotic signal transduction pathway. FADD associates with caspase 8 through interaction of the FADD death eector domain with the death eector domain of caspase 8 (Boldin et al., 1996; Muzio et al., 1996). This interaction may be sucient to promote caspase 8 selfassociation and, in turn, caspase 8 activation (Muzio et al., 1998). Caspase 8 cleaves important substrates including BID (Li et al., 1998; Luo et al., 1998) and RIP (Lin et al., 1999) and initiates a caspase cascade that culminates in the activation of a nuclease, DFF or CAD, that induces DNA fragmentation (Enari et al., 1998; Liu et al., 1997).

HPV-16 E7 and TNF-induced apoptosis DA Thompson et al

3630

Here we show that HPV-16 E7 restrains TNFinduced apoptosis in normal diploid human ®broblasts. This activity of E7 correlates with its ability to transform cells. E7 also confers partial protection against Fas-mediated killing. The HPV-16 E7-induced signaling defect occurs at a TNF R1-proximal signaling step at or upstream of caspase 8 activation.

Results HPV-16 E7 attenuates TNF-induced apoptosis Evasion of apoptotic stimuli constitutes a signi®cant step in tumor progression and is an important means by which viruses evade the immune system (reviewed in Roulston et al., 1999). Hence, we evaluated the capacity of a high risk HPV oncoprotein, HPV-16 E7, to modulate TNF-mediated apoptosis. Ectopic expression of HPV-16 E7 in IMR-90 normal human diploid ®broblasts was achieved using recombinant LXSN-based retroviruses as previously described (Thompson et al., 1997). Cell populations were treated with several TNF doses in conjunction with cycloheximide. After 7 h cells were stained with the DNA dye Hoechst 33258 and their nuclear morphology evaluated by ¯uorescence microscopy. At doses ranging from 1 ± 30 ng/ml TNF, cells expressing HPV-16 E7 showed a more than 50% decreased propensity to undergo apoptosis in response to TNF relative to cells infected with control retrovirus (Figure 1a). Similar results were obtained when the ability of HPV-16 E7 to modulate TNF-mediated apoptosis was evaluated in WI38 cells. Like IMR-90, WI38 are normal diploid human embryonic lung ®broblasts. While these cells were generally less sensitive to TNF, expression of E7 suppressed TNF mediated apoptosis to a similar extent as in IMR-90 cells (Figure 1b). CD95 (Fas) and TNF R1 share a region of intracellular homology termed the `death domain'. Binding of ligand to either receptor can result in death-domain-dependent activation of a caspase 8initiated proteolytic cascade, leading ultimately to apoptosis. We reasoned that, if HPV-16 E7 suppresses TNF-induced apoptosis by targeting a signal transduction component shared with the Fas signaling pathway, HPV-16 E7 would also suppress Fas-induced apoptosis. Using an agonistic anti-Fas antibody to initiate Fas signaling in conjunction with cycloheximide, we found that HPV-16 E7 expressing IMR-90 cells showed a decreased apoptotic response 8 h after stimulation (Figure 1c). However, the overall level of Fas apoptosis was much lower at 8 h compared to TNF and we performed a time course to compare TNF and Fas apoptosis over an interval of 24 h (Figure 1d). Expression of the HPV-16 E7 protein conferred resistance against both Fas and TNF mediated apoptosis. However, the protective eect against Fas was much reduced compared to TNF, and nonapparent by 24 h (Figure 1d). We believe that this is at least in part, due to the slower kinetics of Fas mediated Oncogene

apoptosis in these cells. The protective eect of E7 may be gradually lost due to the inhibition of new protein synthesis in the presence of cycloheximide. Given the more pronounced protective eect of E7 toward TNFmediated apoptosis, we examined this pathway in more detail. HPV-16 E7 interferes with TNF receptor 1-mediated apoptosis and this process involves caspase activation Unlike Fas, TNF R1 also stimulates protective functions which may account for the dierence in E7mediated protection seen between TNF R1 and Fas. Treatment of cells with TNF triggers two receptors, TNF R1 and TNF R2. TNF R1 contains death domains and is thought to be mainly responsible for TNF-mediated apoptosis. In contrast TNF R2 does not contain death domains, but has been reported to activate NF-kB and, therefore, may have anti-apoptotic activities (reviewed in Baker and Reddy, 1996). Hence, E7 may interfere with TNF-mediated apoptosis signaling by undermining pro-apoptotic signals emanating from TNF R1 or enhance anti-apoptotic signals emanating from TNF R1 or TNF R2. To distinguish between these two possibilities we treated E7 or control IMR-90 cells with a soluble form of TNF speci®c for TNF R1. Apoptosis was decreased in E7-expressing cells demonstrating that E7 can interfere with TNF R1-mediated apoptosis (Figure 1e). In addition, treatment of cells with the caspase inhibitor zVADFMK interfered with TNF-and Fas-mediated apoptosis, indicating that caspase activation is crucial to cell death in response to treatment with either cytokine (Figure 1e). TNF receptor 1 remains functional in HPV-16 E7-expressing cells HPV-16 E7-expressing cells may contain lower levels of TNF R1, hence resulting in a decreased cellular response to TNF. However, examination of steadystate TNF R1 levels in cell populations infected with control retrovirus (LXSN) or with retrovirus encoding HPV-16 E7 revealed similar amounts of TNF R1 in both cell populations (Figure 2a). To directly address whether the TNF R1 population in E7-expressing cells was competent for signal transduction we took advantage of the observation that there is a quantitative relationship between extent of receptor/ligand interaction and cellular responses in TNF R1-related signal transduction pathways (P®zenmaier et al., 1987; Wakabayashi et al., 1990). Thus, we examined the activation of p38 MAP kinase (p38) and c-jun amino terminal kinase (JNK) in normal and E7expressing IMR-90 cells. Activation of p38 and JNK in response to TNF is mediated by phosphorylation at Thr 180 and Tyr 182 and Thr 183 and Tyr 185, respectively (Raingeaud et al., 1995). TNF-induced p38 Thr 180/Tyr 182 phosphorylation occurred rapidly in both the E7-expressing and control IMR-90 cell population (Figure 2b). Similarly, TNF-induced phos-


3631

Figure 1 (a) Expression of HPV-16 E7 decreases TNF-induced apoptosis in normal human diploid IMR-90 ®broblasts. Cell populations were treated with the indicated concentration of TNF and with 30 mg/ml cycloheximide for 7 h. Methanol-®xed cells were stained with Hoechst 33258 and apoptotic nuclei identi®ed by ¯uorescence microscopy. Data points represent averages of duplicate plates; error bars represent population standard deviations. (b) Expression of HPV-16 E7 decreases TNF-induced apoptosis in normal human diploid WI38 ®broblasts. Cell populations were treated with the indicated concentrations of TNF and/ or with 30 mg/ml cycloheximide for 7 h. Methanol-®xed cells were stained with Hoechst 33258 and apoptotic nuclei identi®ed by ¯uorescence microscopy. Data points represent averages of duplicate plates; error bars represent population standard deviations. (c) HPV-16 E7 expression inhibits Fas-mediated apoptosis. Cell populations were treated with the indicated concentrations of agonistic anti-Fas antibody and 30 mg/ml cycloheximide for 8 h. Methanol-®xed cells were stained with Hoechst 33258 and apoptotic nuclei identi®ed by ¯uorescence microscopy. Data points represent the average of two experiments; error bars represent population standard deviations. (d) The protective eect of E7 extends over 24 h. Cell populations were treated with the indicated concentrations of TNF or Fas and/or with 30 mg/ml cycloheximide for the indicated times. Hoechst staining was performed as described above. Data points represent averages of duplicate plates; error bars represent population standard deviations. (e) E7mediated decrease of TNF-induced apoptosis involves signaling through TNF receptor 1 and depends on caspase activation. IMR90 cells were treated for 8 h with 30 mg/ml cycloheximide (Chx), cycloheximide plus 5 ng/ml TNF (TNF), cycloheximide plus 10 ng/ ml of a soluble form of TNF that binds TNF R1 but not TNF R2 (TNF (1)), cycloheximide plus 1 mg/ml agonistic anti-Fas antibody for 24 h (Fas), cycloheximide plus 5 ng/ml TNF plus 20 mM pan caspase inhibitor zVAD for 8 h (TNF+zVAD), cycloheximide plus 1 mg/ml agonistic anti Fas antibody plus 20 mM pan caspase inhibitor zVAD for 24 h (Fas+zVAD), or were left untreated (c). Methanol-®xed cells were stained with Hoechst 33258 and apoptotic nuclei identi®ed by ¯uorescence microscopy. Data points represent averages of duplicate plates; error bars represent population standard deviations

phorylation of JNK at Thr 183 and Tyr 185 also occurred swiftly in both cell types (Figure 2c). These results show that TNF R1 remains functional in the E7-expressing cells. Moreover, HPV-16 E7 does not in¯uence TNF-induced p38 or JNK activation in normal human ®broblasts. This observation is signi®cant given that these kinases have been implicated in TNF-mediated cytotoxicity in some cells (Roulston et al., 1998).

TNF-induced IkB inactivation and NF-kB target gene induction occur efficiently irrespective of E7 expression Another hallmark of TNF R1-mediated signaling is the activation of NF-kB. Activation of NF-kB triggers an anti-apoptotic transcriptional response (Antwerp et al., 1996; Liu et al., 1996). Although new protein synthesis is suppressed by treatment with cycloheximide in all of our experiments, we wanted to determine whether NFOncogene


3632

Figure 2 (a) Abundance of TNF R1 is not decreased in IMR-90 cells ectopically expressing HPV-16 E7. Protein extracts were analysed by SDS ± PAGE and immunoblotting. Each pair of lanes represents a matched set of independently infected cell populations. TNF-stimulated phosphorylation of p38 MAP kinase (b), c-jun amino terminal kinase (JNK) (c), and IkBa (d), is maintained in cell populations expressing HPV-16 E7. Cell populations were treated with TNF and cycloheximide for the indicated periods of time and processed for immunoblot analysis using phospho (Thr180/Tyr182)-p38, phospho (Thr183/Tyr185)-JNK, and phospho (ser32)-IkBa-speci®c antibodies. (e) Degradation of IkBa in response to TNF treatment. Cell populations were treated with TNF and cycloheximide for 20 min and processed for immunoblot analysis using an IkBa-speci®c antibody. (f) TNF-stimulated induction of A20 mRNA expression is maintained in cell populations expressing HPV-16 E7. Cell populations were treated with TNF and cycloheximide (+) or with 30 mg/ml cycloheximide alone (7) for 3 h 45 min and analysed for A20 mRNA expression by Northern blotting. Ethidium bromide-stained rRNAs are pictured below as a loading control

kB may be constitutively activated in E7-expressing cells prior to TNF treatment, which might contribute to the dampened apoptotic response of E7-expressing cells in reaction to TNF treatment. In addition, we wanted to compare NF-kB activation in normal and E7-expressing cells to further support our contention that TNF R1-mediated signal transduction was fully functional in E7-expressing cells. TNF receptor-induced activation of the transcription factor NF-kB is mediated through activation of IkB kinases which induces the phosphorylation and subsequent degradation of the cytoplasmic inhibitor IkB (reviewed in Baeuerle, 1998). NF-kB then translocates to the nucleus where it activates transcription of target genes (reviewed in May and Ghosh, 1997). In both control and E7-expressing cell populations we observed rapid transient increases of phospho-IkBa (Ser32) (Figure 2d) and decreases in IkB steady state levels (Figure 2e) upon TNF treatment. Consistent with NFkB activation, mRNA expression of the NF-kBregulated anti-apoptotic gene A20 was eciently Oncogene

induced in response to TNF treatment in both E7expressing and control cells (Figure 2f). Importantly, however, A20 RNA was not expressed prior to TNF treatment in the E7-expressing cell population, consistent with the notion that E7 expression does not result in the constitutive activation of NF-kB. Together with the ecient induction of p38 and JNK (Figure 2b,c), these results demonstrate that TNF R1-mediated signal transduction per se is not defective, nor is NFkB activity constitutively enhanced in E7-expressing cells. Defective TNF-mediated activation of pro-caspase 8 in HPV-16 E7-expressing cells To further analyse the point of action of HPV-16 E7, we compared the activation of caspases in HPV-16 E7expressing cells in response to TNF. Pro-caspase 3 is an eector caspase required for the execution of apoptotic events including DNA degradation (Woo et al., 1998). In response to TNF treatment, pro-caspase 3


is activated through a caspase cascade that is triggered by activation of pro-caspase 8 by TNF R1. TNF treatment induced pro-caspase 3 proteolysis in control retrovirus-infected cells within 6 h (Figure 3a). Consistent with the decrease of TNF-mediated apoptosis in E7-expressing ®broblasts (Figure 1a,b,d,e), TNF-induced pro-caspase 3 cleavage was delayed in E7expressing cells (Figure 3a). The attenuation of procaspase 3 proteolysis in these cells is dependent on TNF treatment since the stability of pro-caspase 3 is not signi®cantly altered by 12 h in unperturbed cells expressing E7 (Figure 3a). Similarly, we compared the proteolytic activation of pro-caspase 8, the apical caspase in the TNF and Fas signal transduction pathway, in HPV-16 E7-expressing versus control cells in response to TNF. Pro-caspase 8 was rapidly cleaved post treatment with TNF (Figure 3b) but cleavage was similarly delayed in E7-expressing cells. The kinetics of activation of caspases 8 and 3 were very similar in response to TNF and cycloheximide. Similar results were obtained in multiple independent experiments. To directly ascertain that the rapid TNF-mediated pro-caspase 8 cleavage corresponds to enzymatic activation, we analysed the death domain containing serine-threonine kinase RIP (Boldin et al., 1995; Hsu et al., 1996; Stanger et al., 1995). Upon recruitment to TNF R1, RIP is cleaved by caspase 8, and a 42 kDa carboxyl terminal cleavage fragment, RIP-C, is produced (Lin et al., 1999). In control cells, RIP cleavage was visible 6 h post TNF treatment as evidenced by the disappearance of the full length RIP protein (Figure 3c) and the appearance of RIP-C at 6, 8, and 12 h after TNF treatment (Figure 3d). Consistent with the delay of pro-caspase 8 cleavage, the disappearance of RIP and the appearance of the cleaved fragment were delayed in E7-expressing cells (Figure 3c,d). Interestingly, while in E7-expressing cells the carboxyl terminal RIP fragment was observed at 8 h, it was no longer detectable at 12 h post TNF treatment and, hence, may have a shorter half life in these cells. To independently con®rm that caspase 8 activation occurred less eciently in E7-expressing cells upon TNF treatment, we also performed caspase 8 activity assays using a ¯uorogenic substrate. Following TNF treatment, enzymatic activity of caspase 8 was enhanced in control cells, and this elevation was decreased in a HPV-16 E7-expressing cell population (Figure 3e). These results show that there is a defect in the rapid TNF-induced proteolysis and activation of caspase 8 in HPV-16 E7-expressing cell populations. Analysis of TNF receptor 1-associated adaptor molecules in E7-expressing cells Expression of the HPV-16 E7 protein alters the half lives of multiple cellular proteins. The half life of the tumor suppressor pRb and the related pocket proteins p107 and p130 are decreased (Boyer et al., 1996; Jones and MuÈnger, 1997), and the half life of p53 and p21cip1 are extended (Jones et al., 1999). Thus, we considered

the possibility that the expression levels of certain TNF R1-associated signal transduction molecules such as TRADD, FADD or RIP might be altered in E7expressing cells. Moreover, it has recently been reported that the half life of TRAF2 is decreased in cells that ectopically express E2F-1 (Phillips et al., 1999). Since E7 expression leads to E2F activation (Phelps et al., 1988) we also analysed the half life of TRAF2 in E7-expressing cells. Expression of the caspase 8 inhibitor cFLIP (Irmler et al., 1997) was also analysed. Control and E7-expressing cells were treated with protein synthesis inhibitor cycloheximide, and the steady state levels of these proteins were visualized. This analysis did not reveal major dierences in steady state levels of turnover of these proteins after 12 h (Figure 4a). In particular TRAF2 was expressed at similar levels and remained stable for 12 h both in normal and E7-expressing ®broblasts. The steady state levels of TRADD appeared slightly decreased in E7-expressing cells; however, this was not consistently observed (see Figure 5a). In contrast FADD levels were somewhat increased in E7-expressing cells. Steady state levels of the caspase 8 inhibitor, cellular FLICE inhibitory protein (cFLIP), were similar in both populations and the half life was shorter than 4 h in control as well as E7-expressing cells (Figure 4a). Similarly, consistent with a published report (Smotkin and Wettstein, 1987), the HPV-16 E7 protein had a short half life of less than 2 h (Figure 4a). Next, we analysed the disappearance of TRADD, FADD and TRAF2 in response to TNF activation in conjunction with cycloheximide (Figure 4b,c). Interestingly, steady state levels of TRADD, FADD and TRAF2 decreased more rapidly upon triggering TNF R1 than in cells treated with cycloheximide alone (Figure 4a,b,c). It also became apparent that upon TNF treatment TRADD and FADD were less rapidly degraded in E7-expressing cells than in control cells. Hence, the observed destabilization of FADD and TRADD in cells upon TNF treatment parallels increased cell death (see Figure 1), and suggests that these adapter molecules may be destabilized upon induction of cell death.

3633

HPV-16 E7 does not bind to TRADD, FADD, or caspase 8 and does not intefere with the TRADD/FADD interaction Next we tested whether E7 could directly interact with TRADD, FADD or caspase 8 and/or whether it could interfere with the interaction between TRADD and FADD. First we performed co-immunoprecipitation experiments in control and E7-expressing IMR-90 cell populations. Both populations expressed similar levels of TRADD and the E7-expressing population contained somewhat increased levels of FADD (Figure 5a). Endogenous TRADD and FADD were coimmunoprecipitated both in normal and E7-expressing cells at similar eciencies. Immunoprecipitation with an E7 antibody did not coprecipitate either FADD or TRADD (Figure 5a). Similarly, immunoprecipitation Oncogene


3634

Figure 3 Delayed TNF-induced cleavage of pro-caspase 3 (a), and pro-caspase 8 (b) in HPV-16 E7-expressing IMR-90 cells (16E7) compared to control vector expressing cells (LXSN). Cells were treated with cycloheximide and TNF (TNF+Chx) or with cycloheximide alone (Chx) for the indicated periods of time, lysed and processed for immunoblot analyses using the corresponding caspase speci®c antibodies. (c,d) Delayed caspase 8-mediated cleavage of RIP in HPV-16 E7-expressing IMR-90 cells (16E7) compared to control vector expressing cells (LX). Cells were treated with 5 ng/ml TNF and 30 mg/ml cycloheximide (TNF+Chx) for the indicated periods of time, lysed and processed for immunoblot analyses using a RIP speci®c antibody. (c) Full length RIP (RIP) and (d) a 42 kDa carboxyl terminal cleavage fragment (RIP-C) are shown. (e) Expression of E7 interferes with TNF-mediated caspase 8 activation. Cells were treated with TNF and cycloheximide (+) or cycloheximide alone (7) for 6.5 h. Extracts of equal cell numbers were analysed for caspase 8 activity using a ¯uorogenic substrate. AU=arbitrary units. Data points represent averages from two experiments; error bars represent population standard deviations

with a TRADD antibody did not coprecipitate E7 (Figure 5a). To determine whether E7 could interact with caspase 8 after stimulation of TNF R1, we performed co-immunoprecipitation experiments in control and E7-expressing IMR-90 cell populations that were treated for up to 30 min with TNF and Oncogene

cycloheximide, corresponding to reported times for DISC activation (Kischkel et al., 1995; Medema et al., 1997). Immunoprecipitation with a caspase 8-speci®c antibody did not coprecipitate E7 (Figure 5b), nor was interaction between TRADD, FADD, and E7 seen at any of these times (data not shown).


3635

Figure 4 (a) Expression of HPV-16 E7 does not decrease stability of TNF apoptosis pathway signal transduction molecules. Cells were treated for the indicated times with 30 mg/ml cycloheximide (Chx), lysed and processed for immunoblot analyses using TRADD, FADD, TRAF2, RIP, cFLIP, HPV-16 E7, and actin-speci®c antibodies. (b,c) Expression of HPV-16 E7 does not decrease stability of TNF apoptosis pathway signal transduction molecules after stimulation. Control vector (LXSN) (b) or HPV-16 E7 (16E7) (c) expressing cells were treated with cycloheximide and TNF (TNF+Chx) for the indicated periods of time, lysed and processed for immunoblot analyses using TRADD, FADD, TRAF2 and actin speci®c antibodies. The corresponding RIP blot is shown as Figure 3c

In addition we performed transient transfection experiments using the highly transfectable human osteosarcoma cell line U2OS. Increasing amounts of an E7 expression plasmid were co-transfected with constant amounts of TRADD and FADD (Figure 5c). Similar to what we observed in IMR-90 cells with endogenous proteins, immunoprecipitation with a TRADD speci®c antibody co-precipitated FADD. Importantly, the amount of FADD in complex with TRADD was constant regardless of the amount of cotransfected E7 suggesting that E7 does not directly interfere with TRADD/FADD interaction (Figure 5d). Moreover, there was no evidence for co-immunoprecipitation of E7 with TRADD or FADD (Figure 5d). Hence, the eect of HPV-16 E7 on TNF-mediated apoptosis and caspase 8 activation is likely to be indirect. Suppression of TNF-induced apoptosis by HPV-16 E7 correlates with cellular transformation Next we determined the structural domains of the E7 protein that contribute to this activity. We utilized a panel of previously characterized HPV-16 E7 mutants (Demers et al., 1996; Jones et al., 1997). This analysis

showed that the ability of E7 to attenuate TNFinduced apoptosis correlates with cellular transformation. Two transformation-de®cient HPV-16 E7 mutants (DP6TLHE10 and DD21LYC24) that lack the capacity to destabilize pRB fail to suppress TNF-induced apoptosis (Figure 6). An HPV-16 E7 mutant at the casein kinase II site (SS31/32 ± GG) that is able to transform cells, albeit less eciently than wild-type HPV-16 E7, suppressed TNF-induced apoptosis, but to a lesser extent than HPV-16 E7 (Figure 6). A transformation competent control mutant (S71 ± I) retained the capacity to inhibit TNF-induced apoptosis (Figure 6). Hence, HPV-16 E7-mediated inhibition of TNF-induced apoptosis correlates with cellular transformation. In agreement with this notion, expression of a transformation de®cient, `low risk' HPV-16 E7 protein did not interfere with TNF sensitivity of normal human ®broblasts (Figure 6). Discussion The elimination of virally infected cells by cytotoxic T lymphocytes (CTL) constitutes an important host defense mechanism that is mediated at least in part Oncogene


3636

Figure 5 HPV-16 E7 does not interact with TRADD, FADD, or caspase 8, and does not interfere with TRADD/FADD interaction. (a) Lysates of control vector (LX) or HPV-16 E7 (E7) expressing IMR-90 cells were immunoprecipitated with TRADD, or E7 speci®c antibodies and analysed by immunoblot analysis using TRADD, FADD or E7 speci®c antibodies. A band corresponding to Ig light chain that migrates similarly as FADD is indicated (*). 100 mg of total lysate was also analysed to document expression of the corresponding proteins. (b) Lysates of control vector (LX) or HPV-16 E7 (E7) expressing IMR-90 cells treated with TNF and CHX for various times were immunoprecipitated with caspase 8 speci®c antibodies and analysed by immunoblot analysis using caspase 8 or E7 speci®c antibodies. A band corresponding to Ig heavy chain is indicated (*) to document similar eciencies of immune complex recovery in each lane. 100 mg of total lysate was also analysed to document expression of the corresponding proteins. (c) The human osteosarcoma cell line U2OS was transiently transfected with the indicated amounts of TRADD, FADD and/or E7 containing expression plasmids. At 48 h post transfection cells were lysed and analysed for expression of the corresponding proteins by immunoblot analysis using TRADD, FADD or E7-speci®c antibodies. (d) Equal amounts of the above lysates were immunoprecipitated with a TRADD speci®c antibody and analysed by immunoblot analysis using TRADD, FADD or E7-speci®c antibodies. A band corresponding to Ig heavy chain is indicated (*) to document similar eciencies of immune complex recovery in each lane, ** Denotes a background band

by the engagement of death domain-receptors such as Fas or TNF R1 on the surface of the infected host cell. Therefore, it is not surprising that many viruses have developed strategies to thwart detection and/or elimination by death domain receptor-mediated apoptosis. Adenoviruses (Ad), for example employ multiple strategies to short-circuit TNF and Fas mediated apoptosis which are relevant to the virus's ability to maintain a persistent infection of cells (reviewed in Wold et al., 1999). Here we report that the HPV-16 E7 oncoprotein impedes TNF-mediated apoptosis and caspase activation in normal human diploid ®broblasts (Figures 1 and 3). Our previously published experiments have demonstrated that E7-expressing IMR-90 cells are highly predisposed to undergo apoptosis in response to other stimuli such as con¯uence or growth factor deprivation (Jones et al., 1997). In fact, the initial experiments described in this report were performed Oncogene

with identical cell populations and in parallel with the growth factor deprivation experiments that we have published (Jones et al., 1997). Hence, HPV-16 E7 can enhance or dampen an apoptotic response of a given cell depending on the stimulus. The ability of HPV-16 E7 to modulate apoptotic responses correlates with cellular transformation (Figure 6) (Jones et al., 1997), and cells expressing the transformation de®cient low risk HPV-6 E7 protein undergo TNF-induced apoptosis as eciently as control cells (Figure 6). Death-domain receptor-mediated activation of caspases and induction of apoptosis has been most thoroughly studied for the TNF R1-related cytokine receptor Fas. Upon stimulation of Fas a death inducing signaling complex (DISC) is formed at the receptor (Kischkel et al., 1995). This Fas-associated complex contains FADD, as well as pro-caspase 8 (Medema et al., 1997). It has been proposed that a similar DISC may also be formed associated with TNF


Figure 6 The ability of HPV E7 to interfere with TNF-mediated apoptosis correlates with cellular transformation. IMR-90 cells expressing control vector (LXSN), wild type HPV-16 E7 (16E7), two transformation de®cient mutants in the amino terminal domain of HPV-16 E7 that interfere with pRB degradation (DP6TLHE10; DD21LYC24), a mutation in the casein kinase II site that shows diminished cellular transformation (SS31/32 ± GG), a mutation in the carboxyl terminal domain that transforms similar to wild type (S71-I), or a transformation de®cient `low risk' HPV6 derived E7 (6E7) were treated with 3 ng/ml TNF and 30 mg/ml cycloheximide for 7 h. Methanol-®xed cells were stained with Hoechst 33258 and apoptotic nuclei identi®ed by ¯uorescence microscopy. Data points represent averages from two experiments, error bars represent population standard deviations

R1 in response to TNF treatment, and it has been detected in some cell types (Bulfone-Pau et al., 1999). Our results documenting decreased caspase 8 activation in E7-expressing cells in response to TNF (Figure 3b) predict that the TNF R1-associated DISC function and pro-caspase 8 activation is suppressed in these cells. This contention is supported by the delay in procaspase 3, pro-caspase 8 and RIP cleavage that we have observed in E7-expressing cells (Figure 3). RIP is a death domain containing protein kinase that interacts directly with TRADD (Boldin et al., 1995; Stanger et al., 1995). It is a component of the TNF R1 signaling complex, and is itself a substrate for caspase 8 (Lin et al., 1999). Upon cleavage by caspase 8 a stable carboxyl terminal RIP fragment, RIP-C, is produced and remains associated with TRADD. It enhances TRADD/FADD interactions, and hence increases TNF-mediated cell death (Lin et al., 1999). In E7expressing cells, we observed a delay in RIP cleavage, consistent with a lag in caspase 8 activation relative to control cells. Moreover, the pro-apoptotic RIP-C fragment appeared to have a shorter half life in E7 expressing cells (Figure 3d). Hence, in addition to a delay in pro-caspase 8 activation, E7 expressing cells may have a defect in sustained TNF R1 mediated proapoptotic signaling. Despite the inhibition of pro-caspase 8 activation, generic TNF R1 signaling is not noticeably perturbed in E7-expressing cells. The steady state levels of TNF R1 are not reduced (Figure 2a), and there is rapid and ecient activation of p38 (Figure 2b), JNK (Figure 2c),

and NF-kB (Figure 2d ± f) in E7-expressing cells. This demonstrates that TNF R1-mediated signaling through TRADD, RIP and TRAF2 is functional in E7expressing cells. The main signaling molecules that have been implicated in TNF R1-mediated apoptosis, TRADD, FADD, TRAF2, RIP, cFLIP and procaspase 8, are expressed at similar levels in control and E7-expressing cells (Figures 4 and 5). In addition, our results indicate that E7 does not stably interact with TRADD, FADD, or caspase 8, and moreover, that E7 does not interfere with the formation of TRADD/FADD complexes which we detected even in the absence of TNF stimulus (Figure 5a,d). Some viruses encode FLICE-inhibitory proteins (vFLIPs) to interfere with receptor-mediated activation of pro-caspase 8. These molecules contain two death-eector domains that interact with FADD, thereby inhibiting productive DISC formation and pro-caspase 8 activation (Thome et al., 1997). It is unlikely that E7 acts as a prototypical FLIP since it lacks any recognizable homologies to death eector domains. Furthermore, given that the half life of HPV-16 E7 is less than 2 h (Figure 4a; Smotkin and Wettstein, 1987) it is unlikely that E7 directly suppresses caspase 8 activation since inhibition of pro-caspase 8 persists to 8 ± 12 h post TNF treatment. We postulate that E7 modulates the expression of an apoptosis regulator with a relatively long half life, either at the level of gene expression or protein stability. One attractive candidate that we investigated was cFLIP, a molecule that is related to viral FLIPs and encodes a catalytically inert caspase 8-like molecule (Irmler et al., 1997; Scadi et al., 1999). However, our experiments showed that cFLIP is expressed at similar levels and has a short half life of less than 4 h both in normal and E7 expressing cells (Figure 4a). The ability of HPV-16 E7 to interfere with TNFinduced apoptosis and caspase 8 activation is reminiscent of the ability of the adenovirus E1B 19K protein to inhibit Fas-mediated apoptosis. While E1B 19K does not directly interact with FADD or pro-caspase 8, it interferes with DISC formation (Perez and White, 1998). The rapid decay of the intracellular pool of E7 after cycloheximide treatment, makes it unlikely that E7 inhibits TNF-induced cell death over extended periods of time via a direct mechanism. Interestingly, HPV-16 E7-expressing cells had only a minor reduction in apoptosis mediated by Fas, a cytokine receptor that signals cell death similar to TNF R1. We found that induction of apoptosis occurred generally much slower and less eciently in response to stimulation of Fas than through TNF R1 (Figure 1d). Importantly, however, cell death triggered by Fas and TNF R1 were both strictly dependent on caspase activation (Figure 1e) and, in conjunction with the typical changes in nuclear morphology observed upon Hoechst staining, likely represent apoptosis and not necrosis. This suggests that there may be a fundamental mechanistic dierence in Fas- and TNF R1mediated-apoptosis in IMR-90 cells.

3637

Oncogene


3638

Cell lines have been classi®ed as type I or type II with respect to Fas-mediated apoptosis (Scadi et al., 1998). In type I cells, caspase 8 is activated rapidly at the DISC, while in type II cells, activation of caspase 8 occurs less rapidly, and is less dependent on DISC formation (Scadi et al., 1998). It is unclear from the published literature whether TNF R1-mediated apoptosis signaling in type I and type II cells mirrors that induced through Fas but our results suggest that this might not be the case. The cleavage of RIP (Figure 3c) indicates that TNF-mediated caspase 8 activation likely involves formation of a DISC, and that E7 protein can interfere with this process. Apoptosis is induced more rapidly and eciently by TNF R1 than Fas antibody in IMR-90 cells (Figure 1d). This may also explain the dierences between this study and a published report by Richard Schlegel's laboratory that provided evidence for increased TNF-mediated apoptosis in E7expressing keratinocytes (Stoppler et al., 1998). We have obtained similar results in keratinocytes: TNF/ cycloheximide treatment induces a relatively low level of apoptosis and the apoptotic response is much delayed compared to ®broblasts (J Basile et al., 2001, manuscript in preparation). Despite the fact that HPVs infect keratinocytes and not ®broblasts, we believe that the ability of E7 to interfere with the rapid activation of pro-caspase 8 and apoptotic response in the background of a normal human diploid cell is relevant and reveals a potentially novel mechanism for the subversion of cytokinemediated apoptosis signaling. The signal transduction pathways disrupted by E7 are disregulated in many dierent tumor types and, hence, these defects may cause decreased sensitivity to cytokine-mediated cell death in such neoplasms. Moreover, other TNF-related cytokines exist and we are currently investigating whether they can induce apoptosis in keratinocytes through similar pathways as TNF RI in IMR-90 ®broblasts. It can then be tested whether E7 can interfere with the apoptotic response to such cytokines in human keratinocytes.

Apoptosis assays Approximately 7 ± 86104 cells were plated per 35 mm dish; 2 days later, cells were exposed to recombinant human TNF (R&D Systems Inc.) or a soluble form of TNF that binds TNF R1 but not TNF R2 (#522-008-C050-Alexis Corporation), agonistic anti-human CD95 (Fas) antibody (DX-2.1, R&D Systems Inc.) and 30 mg/ml cycloheximide, or 30 mg/ml cycloheximide alone. At the times indicated, cells were ®xed by exposure to methanol vapor for 10 min followed by immersion in methanol for at least 10 min. After staining with bisbenzimide (Hoechst 33258, Sigma) (1 mg/ml in water clari®ed with nonfat dry milk) for 7 min and rinsing in water, apoptotic nuclei were quantitated by ¯uorescence microscopy. Typically, 300 ± 500 nuclei were counted per sample. The caspase inhibitor zVAD-FMK (#FK009, Enzyme Systems Products) was used at a concentration of 20 mM and was added to the cells at the same time as TNF or anti Fas antibody. Immunological methods

IMR-90, WI38, PA317, PG13 cells, and the human osteosarcoma cell line U2OS were grown at 378C with 5% CO2 in Dulbecco's modi®ed Eagle medium (DMEM) supplemented with 10% iron-supplemented bovine calf serum (HyClone Laboratories, Inc.), 3.2 mM glutamine and 1% penicillin/streptomycin.

The following commercial antibodies were used: actin (Chemicon), caspase 3 (#19; Transduction Laboratories), caspase 8 (B9-2; Pharmingen), FADD (#1; Transduction Laboratories), HPV-16 E7 (ED17; Santa Cruz Biotechnology, Inc; 8C9; Zymed), IkBa (C21; Santa Cruz Biotechnology, Inc), RIP (G322-2; Pharmingen), TRADD (C20; Santa Cruz Biotechnology, Inc), TRAF2 (#2361; Immunotech), TNF R1 (MAB225; R&D Systems Inc.). Speci®c antibodies for phospho-IkBa (Ser32) (#9241), phospho-p38 (Thr180/ Tyr182) (#9211), and phospho-SAPK/JNK (Thr183/Tyr185) (#9251) were purchased from New England Biolabs, Inc. The cFLIP antibody (#NF6) (Scadi et al., 1998) was a generous gift from Peter Krammer (DKFZ, Heidelberg, Germany). Cells were lysed in 0.1% NP-40 buer (250 mM NaCl, 50 mM Tris-HCl (pH 7.5) 5 mM EDTA, 0.1% Nonidet P-40 (NP-40), 1 mM dithiothreitol (DTT), 0.5 mM PMSF, 1 mg/ml aprotinin and leupeptin, 2 mM NaF, 0.5 mM Na2VO3) on ice for 30 min and centrifuged 15 min at 15 000 g. Immunoprecipitations were performed for 2 h at 48C and antigen/ antibody complexes were puri®ed with protein G-agarose (Gibco ± BRL) for 2 h at 48C. Protein samples resolved by SDS ± PAGE were transferred to a PVDF membrane (Immobilon-P, Millipore). After blocking nonspeci®c binding sites by incubation in 5% Carnation nonfat dry milk in 10 mM Tris HCl, 2.5 mM EDTA, 50 mM NaCl, 0.1% Tween20, pH 7.5, the membrane was hybridized with primary antibody, washed, hybridized with horseradish peroxidaseconjugated secondary antibody (Amersham), and analysed by enhanced chemiluminescence (Amersham).

Retroviral infections

Caspase activity assay

PA317 or PG13 packaging cells transfected with the control vector, LXSN, or with LXSN-based vectors containing HPV type 6 or 16 E7 sequences (Foster et al., 1994; Halbert et al., 1991) were kindly provided by D Galloway (Fred Hutchinson Cancer Center, Seattle, WA, USA). PA317 packaging cells expressing the LXSN-based retroviral construct were prepared

Caspase 8 activity was assessed using the `ApoAlert FLICE/ Caspase-8 Fluorescent Assay Kit' (Clontech Laboratories, Inc.) according to manufacturer's instructions using IETDAFC as a substrate. For ¯uorometric detection of protease activity, sample ¯uorescence was evaluated using a 380 nm excitation ®lter and a 530 nm emission ®lter.

Materials and methods Cell culture

Oncogene

as described (Miller and Rosman, 1989). Supernatant was harvested from packaging cells and IMR-90 cells infected as previously described (Thompson et al., 1997). Cells exposed to retrovirus were selected in 300 mg/ml G418 for at least 6 days and pooled prior to use. Expression of E7 was ascertained by immunoblot analysis in each individual pool of cells.


Northern analysis Total RNA was prepared from subcon¯uent cells according to manufacturer's instructions (RNeasy-Qiagen). Ten mg total RNA was resolved on a MOPS/formaldehyde 1% agarose gel, transferred to a nylon membrane (Nytran+, Schleicher and Schuell), and UV crosslinked. The membrane was hybridized with a 32P-labeled full-length A20 cDNA probe (Opipari et al., 1990) (kindly provided by V Dixit) in QuikHyb (Stratagene), washed at 658C in 16SSC/0.05% SDS and exposed to Kodak X-AR ®lm.

Acknowledgments We are grateful to Arthur Lee and Philip Hinds for critical comments concerning this manuscript. We also thank D Galloway, W Krek, V Dixit, M Oren R Mulligan, E Kie, J Daniels, R Ed Marlborough and TH-T Chang for providing important reagents and support. This work was supported by Public Health Service grant R01 CA81135 (K MuÈnger) from the National Cancer Institute. V Zacny was supported by T32 CA72320. K MuÈnger is a Ludwig Scholar; DA Thompson is a Ryan fellow.

3639

References Antwerp D, Martin S, Kafri T, Green D and Verma I. (1996). Science, 274, 787 ± 789. Baeuerle PA. (1998). Cell, 95, 729 ± 731. Baker SJ and Reddy EP. (1996). Oncogene, 12, 1 ± 9. Banner DW, D'Arcy A, Janes W, Gentz R, Schoenfeld HJ, Broger C, Loetscher H and Lesslauer W. (1993). Cell, 73, 431 ± 445. Boldin MP, Goncharov TM, Goltsev YV and Wallach D. (1996). Cell, 85, 803 ± 815. Boldin MP, Varfolomeev EE, Pancer Z, Mett IL, Camonis JH and Wallach D. (1995). J. Biol. Chem., 270, 7795 ± 7798. Boyer SN, Wazer DE and Band V. (1996). Cancer Res., 56, 4620 ± 4624. Bulfone-Pau SS, Bulanova E, Pohl T, Budagian V, Durkop H, Ruckert R, Kunzendorf U, Paus R and Krause H. (1999). FASEB J., 13, 1575 ± 1585. Cryns V and Yuan J. (1998). Genes Dev., 12, 1551 ± 1570. D'Arcy A, Banner DW, Janes W, Winkler FK, Loetscher H, Schonfeld HJ, Zulauf M, Gentz R and Lesslauer W. (1993). J. Mol. Biol., 229, 555 ± 557. Demers GW, Espling E, Harry JB, Etscheid BG and Galloway DA. (1996). J. Virol., 70, 6862 ± 6869. Eck MJ and Sprang SR. (1989). J. Biol. Chem., 264, 17595 ± 17605. Enari M, Sakahira H, Yokoyama H, Okawa K, Iwamatsu A and Nagata S. (1998). Nature, 391, 43 ± 50. Foster SA, Demers GW, Etscheid BG and Galloway DA. (1994). J. Virol., 68, 5698 ± 5705. Halbert CL, Demers W and Galloway DA. (1991). J. Virol., 65, 473 ± 478. Hsu H, Huang J, Shu HB, Baichwal V and Goeddel DV. (1996). Immunity, 4, 387 ± 396. Irmler M, Thome M, Hahne M, Schneider P, Hofmann K, Steiner V, Bodmer JL, Schroter M, Burns K, Mattmann C, Rimoldi D, French LE and Tschopp J. (1997). Nature, 388, 190 ± 195. Jones DL and MuÈnger K. (1997). J. Virol., 71, 2905 ± 2912. Jones DL, Thompson DA and MuÈnger K. (1997). Virology, 239, 97 ± 107. Jones DL, Thompson DA, Suh-Burgmann E, Grace M and Munger K. (1999). Virology, 258, 406 ± 414. Kirstein M, Fiers W and Baglioni C. (1986). J. Immunol., 137, 2277 ± 2280. Kischkel FC, Hellbart S, Behrmann I, Germer M, Pawlita M, Krammer PH and Peter ME. (1995). EMBO J., 14, 5579 ± 5588. Li H, Zhu H, Xu CJ and Yuan J. (1998). Cell, 94, 491 ± 501. Lin Y, Devin A, Rodriguez Y and Liu ZG. (1999). Genes Dev., 13, 2514 ± 2526. Liu X, Zou H, Slaughter C and Wang X. (1997). Cell, 89, 175 ± 184.

Liu ZG, Hus H, Goeddel DV and Karin M. (1996). Cell, 87, 565 ± 576. Luo X, Budihardjo I, Zou H, Slaughter C and Wang X. (1998). Cell, 94, 481 ± 490. May MJ and Ghosh S. (1997). Semin. Cancer Biol., 8, 63 ± 73. Medema JP, Scadi C, Kischkel FC, Shevchenko A, Mann M, Krammer PH and Peter ME. (1997). EMBO J., 16, 2794 ± 2804. Miller AD and Rosman GJ. (1989). Bio. Techniques, 7, 980 ± 982. Muzio M, Chinnaiyan AM, Kischkel FC, Orourke K, Shevchenko A, Ni J, Scadi C, Bretz JD, Zhang M, Gentz R, Mann M, Krammer PH, Peter ME and Dixit VM. (1996). Cell, 85, 817 ± 827. Muzio M, Stockwell BR, Stennicke HR, Salvesen GS and Dixit VM. (1998). J. Biol. Chem., 273, 2926 ± 2930. Opipari Jr, AW, Boguski MS and Dixit VM. (1990). J. Biol. Chem., 265, 14705 ± 14708. Perez D and White E. (1998). J. Cell. Biol., 141, 1255 ± 1266. P®zenmaier K, Kronke M, Scheurich P and Nagel GA. (1987). Blut, 55, 1 ± 10. Phelps WC, Yee CL, MuÈnger K and Howley PM. (1988). Cell, 53, 539 ± 547. Phillips AC, Ernst MK, Bates S, Rice NR and Vousden KH. (1999). Mol. Cell., 4, 771 ± 781. Raingeaud J, Gupta S, Rogers JS, Dickens M, Han J, Ulevitch RJ and Davis RJ. (1995). J. Biol. Chem., 270, 7420 ± 7426. Ramshaw IA, Ramsay AJ, Karupiah G, Rolph MS, Mahalingam S and Ruby JC. (1997). Immunol. Rev., 159, 119 ± 135. Reid TR, Torti FM and Ringold GM. (1989). J. Biol. Chem., 264, 4583 ± 4589. Roulston A, Marcellus RC and Branton PE. (1999). Annu. Rev. Microbiol., 53, 577 ± 628. Roulston A, Reinhard C, Amiri P and Williams LT. (1998). J. Biol. Chem., 273, 10232 ± 10239. Scadi C, Fulda S, Srinivasan A, Friesen C, Li F, Tomaselli KJ, Debatin KM, Krammer PH and Peter ME. (1998). EMBO J., 17, 1675 ± 1687. Scadi C, Schmitz I, Krammer PH and Peter ME. (1999). J. Biol. Chem., 274, 1541 ± 1548. Smotkin D and Wettstein OF. (1987). J. Virol., 61, 1686 ± 1689. Stanger BZ, Leder P, Lee TH, Kim E and Seed B. (1995). Cell, 81, 513 ± 523. Stoppler H, Stoppler MC, Johnson E, Simbulan-Rosenthal CM, Smulson ME, Iyer S, Rosenthal DS and Schlegel R. (1998). Oncogene, 17, 1207 ± 1214. Tartaglia LA, Rothe M, Hu YF and Goeddel DV. (1993). Cell, 73, 213 ± 216. Oncogene


3640

Oncogene

Thome M, Schneider P, Hofmann K, Fickenscher H, Meinl E, Neipel F, Mattmann C, Burns K, Bodmer JL, Schroter M, Scadi C, Krammer PH, Peter ME and Tschopp J. (1997). Nature, 386, 517 ± 521. Thompson DA, Belinsky G, Chang TH-T, Jones DL, Schlegel R and MuÈnger K. (1997). Oncogene, 15, 3025 ± 3036. Vandevoorde V, Haegeman G and Fiers W. (1997). J. Cell. Biol., 137, 1627 ± 1638. Wakabayashi T, Asada M, Nagasu T, Iijima A, Hasegawa Y, Shikata Y and Kitoh K. (1990). J. Biol. Chem., 265, 7604 ± 7609.

Wold WS, Doronin K, Toth K, Kuppuswamy M, Lichtenstein DL and Tollefson AE. (1999). Curr. Opin. Immunol., 11, 380 ± 386. Wong GH, Kamb A and Goeddel DV. (1992). Tumor Necrosis Factors: the Molecules and Their Emerging Role in Medicine. Beutler B. (ed.). Raven Press: New York, pp. 371 ± 381. Woo M, Hakem R, Soengas MS, Duncan GS, Shahinian A, Kagi D, Hakem A, McCurrach M, Khoo W, Kaufman SA, Senaldi G, Howard T, Lowe SW and Mak TW. (1998). Genes Dev., 12, 806 ± 819.