FADD/MORT1 is a common mediator of CD95 (Fas/APO-1) and tumor necrosis factor receptor-induced apoptosis.

CD95 (Fas/APO-1) and tumor necrosis factor receptor-1 (TNFR-1) are related molecules that signal apoptosis. Recently, a number of novel binding proteins have been proposed to mediate the signaling of these death receptors. Here we report that an N-terminal truncation of one of these candidate signal transducers, FADD/MORT1, abrogates CD95-induced apoptosis, ceramide generation, and activation of the cell death protease Yama/CPP32. In addition, this dominant-negative derivative of FADD (FADD-DN) blocked TNF-induced apoptosis while not affecting NF- kappaB activation. FADD-DN bound both receptors, and in the case of CD95, it disrupted the assembly of a signaling complex. Taken together, our results functionally establish FADD as the apoptotic trigger of CD95 and TNFR-1.

Recently, major advances have been made in understanding the signal transduction of the tumor necrosis factor (TNF) 1 / nerve growth factor receptor family. Activation of these receptors is caused by aggregation mediated by the respective ligands or agonist antibodies (1)(2)(3)(4). There are no identifiable catalytic motifs (e.g. kinase or phosphatase) in the cytoplasmic domains of these cell surface proteins. Instead, it is becoming apparent that signal transduction is accomplished via association with an emerging class of novel and diverse signaling molecules. For example, a dominant-negative mutant of TRAF2 was shown to block TNFR-2-mediated NF-B activation (5), while an analogous mutant of CD40bp (also known as TRAF3, CRAF1, or LAP1 (6 -8)) was shown to inhibit CD40mediated up-regulation of CD23 (7). Numerous candidate sig-nal transducers have been identified for the two death receptors, TNFR-1 and CD95, including FADD/MORT1, TRADD, RIP, FAP-1, FAF1, and TRAP1/2 (9 -13). Both death receptors share a region of homology of about 80 amino acids in their cytoplasmic domain required to signal apoptosis (14,15). This shared "death domain" suggests that both receptors engage a common component of the apoptotic machinery. Here, we investigated the role of FADD in the proximal signal transduction of CD95 and TNFR-1.
Cell Lines and Culture-The B lymphoma cell line BJAB and the breast carcinoma cell line MCF7 were grown in RPMI 1640 complete medium (10% heat-inactivated fetal bovine serum (Hyclone), L-glutamine, penicillin/streptomycin, and non-essential amino acids). MCF7-Fas cells, as described previously (18), were grown in RPMI 1640 complete media supplemented with 0.5 mg/ml G418 (Life Technologies, Inc.). BJAB and MCF7 stable cell lines were grown in complete media supplemented with 3 and 0.5 mg/ml G418, respectively.
Stable and Transient Transfections-To generate the pooled stable cell lines, BJAB-FADD-DN and MCF7-FADD-DN, cells were transfected with pcDNA3-FADD-DN using a protocol described previously (19). From the pooled populations, individual clones were obtained and plated in duplicate on 48-well Costar plates. One set of cells was treated with anti-Fas IgM (100 ng/ml), and clones resistant to Fas-induced cell death were identified. The untreated, resistant clones were then pooled to obtain BJAB-sFADD-DN (which represents a pool of seven resistant clones) and MCF7-sFADD-DN (which represents a pool of nine resistant clones). All stable lines generated were assessed for expression of FADD-DN by immunoblotting. MCF7 cells were transiently transfected with lipofectamine as described previously (9).
Cell Death and Viability Assays-To assess nuclear morphology, fluorescent DNA-staining dyes were utilized as described previously (19). DNA fragmentation (TUNEL staining) was determined using the in situ cell death detection kit (Boehringer Mannheim). BJAB cells were air-dried onto Colorfrost/Plus microscope slides (Fisher) using 4% paraformaldehyde, and the manufacturer's protocol was followed. The TUNEL-stained cells were then counterstained with propidium iodide (10 g/ml) and visualized by fluorescence microscopy. To assess cell viability, the 3-(4,5-dimethyl thiazol-2-yl)-2,5-diphenyl tetrazolium bromide conversion assay was done on the BJAB cell lines as described previously (20).  (23). The TUNEL assay is shown in the graphical inset, and at least 250 cells were counted in three independent experiments (mean Ϯ S.D.). C, anti-CD95-induced ceramide generation is abrogated by FADD-DN. The indicated BJAB cell lines were treated with anti-Fas IgM (1 g/ml) for the various times listed and ceramide levels subsequently assessed (mean Ϯ S.D.; n ϭ 3). Significant levels of ceramide could not be detected at 5-, 10-, 30-and 60-min time points (inset). D, C 2-ceramide (C2), but not C 2 -dihydroceramide (DHC2), can bypass the dominant negative effect of the FADD derivative. The cells characterized include BJAB vector, BJAB-FADD-DN, and BJAB-sFADD-DN. As a control, cells were also exposed to the structurally related inactive analog, C 2 -dihydroceramide (33). Viabilities were not decreased significantly, Ceramide Assays-Ceramide levels were determined by a modified diacylglycerol kinase assay (21,22) as described previously (20).
Immunoprecipitations, Western Blotting, and Two-dimensional Gels-Immunoblotting of cell lysates for PARP was carried out as described previously (17). Yama processing was assessed using 1 ϫ 10 7 BJAB vector and BJAB-sFADD-DN cells untreated or treated with 100 ng/ml anti-Fas IgM for 18 h. Cells were then lysed in 60 l of 0.1% Nonidet P-40, free-thawed three times, and centrifuged at 14,000 rpm for 20 min. Cytoplasmic extracts were carefully added to sample buffer and run on a 15% gel, transferred to a nitrocellulose membrane, and immunoblotted with antibodies directed against the p17 and p12 subunits of Yama. To show expression of FADD-DN in BJAB and MCF7 cells, cells were immunoprecipitated in phosphate-buffered saline-TDS (TDS, 1% Triton X-100, 0.5% deoxycholic acid, 0.1% SDS) with anti-AU1 antibody (1:100) and Western blotted with anti-FADD polyclonal antisera (1:1000). Immunoprecipitation of the DISC and analysis on two-dimensional gels was done as described previously (23). Alternatively, cell lysates were immunoprecipitated with anti-AU1 antibody coupled to protein A-Sepharose beads as described previously (23).
NF-B Assay-MCF7 vector and MCF7-sFADD-DN cells were transfected with the NF-B-dependent E-selectin reporter construct and luciferase activity assessed as described previously (5).

RESULTS AND DISCUSSION
Overexpression of FADD causes apoptosis (9), resulting in cleavage of the death substrate poly(ADP-ribose) polymerase (PARP) to signature apoptotic fragments (data not shown). A previously reported deletion mutant of FADD, NFD4 (hereon referred to as FADD-DN), was able to interact with CD95, but failed to initiate apoptosis (9) (Fig. 1A), suggesting that it may have a dominant negative effect on CD95 signaling. FADD-DN lacks 80 N-terminal amino acids, but contains the death domain responsible for association with the related death domain of CD95 (9). The B lymphoma cell line BJAB was transfected with either the expression vector pcDNA3 alone or as a FADD-DN expression construct. Stable transfectants were generated by neomycin (G418) selection and pooled populations assessed for FADD-DN expression and sensitivity to anti-Fasinduced apoptosis (BJAB-FADD-DN, Fig. 1B). Expression of FADD-DN in both a pooled population and in a mixture of selected clones (BJAB-sFADD-DN) dramatically abrogated CD95-induced cell death (Fig. 1B). The apoptotic nature of the cell death was confirmed by the TUNEL assay which detects 3Ј-OH DNA strand breaks. The FADD-DN expressing BJAB cells were not inherently resistant to apoptotic cell death, since the protein kinase inhibitor staurosporine and the calcium ionophore A23187 equally killed the three cell lines (data not shown). CD95 surface expression was equivalent in the vector and FADD-DN cell lines as assessed by flow cytometry (data not shown). The possibility that clonal variation in the stable lines was responsible for the observed resistance to CD95 killing was discounted by the observation that transient overexpression of the FADD derivative ablated CD95-induced cell death in BJAB and Jurkat cells (data not shown).
The sphingolipid ceramide has been implicated as a signaling intermediate in the CD95 pathway (20, 24 -26). To determine whether this signaling event was blocked by the FADD derivative, BJAB cells expressing FADD-DN were treated with anti-Fas IgM and, subsequently, ceramide levels assessed. Consistent with a proximal role of FADD in CD95 signaling,

FIG. 2. FADD mediates TNF-induced cell death, but not TNFinduced NF-B activation.
A, the stable cell lines utilized include MCF7-sFADD-DN, which represents a pool of nine resistant clones and a corresponding MCF7 vector control. Expression of FADD-DN is shown in the left panels. FADD-DN migrates as a doublet around 18 kDa, possibly due to post-translational modification. The indicated cell lines were either left untreated (UnRx) or treated with anti-Fas IgM (250 ng/ml) plus cycloheximide (CHX, 10 g/ml) or 100 ng/ml TNF for 24 h and stained with propidium iodide. Similar results were obtained using anti-APO-1 antibody plus soluble protein A in the absence of cycloheximide (data not shown). Phase contrast micrographs are shown with corresponding confocal micrographs (insets) depicting nuclear morphology. B, MCF7 vector or MCF7-sFADD-DN cells were transfected with an NF-B-dependent E-selectin-luciferase reporter construct (5) and were either untreated or treated with TNF for 9 h. Luciferase activities were assessed as described previously (5), and values shown are mean Ϯ S.D. of three independent experiments.

TABLE I Expression of FADD-DN in MCF7 cells blocks CD95-and
TNF-induced apoptosis MCF7 stable cell lines were generated using the method described for the BJAB stable lines. MCF7-FADD-DN is a pooled population of transfectants while the MCF7-sFADD-DN is a mixture of selected clones. Expression was confirmed by Western blotting for FADD-DN ( Fig. 2A). The indicated MCF7 cell lines were transiently transfected with pCMV-␤-gal (0.25 g), pcDNA3-Fas (0.1 g), and control vector DNA (1.0 g) or pcDNA3-AU1-hICE (1.0 g). Twenty-four hours after transfection, cells were treated with anti-Fas IgM (250 ng/ml), TNF (100 ng/ml), or 1 M staurosporine (Sigma). Transfection and 5-bromo-4-chloro-3-indoyl ␤-galactoside staining were done as previously described (10). The data (mean Ϯ S.D.; n of at least 300 cells) represents the percentage of nonapoptotic cells. Round cells were confirmed to be apoptotic by propidium iodide staining of nuclei (data not shown). FADD-DN inhibited CD95-mediated ceramide generation (Fig.  1C). Additionally, vector and FADD-DN transfected BJAB cells were equally susceptible to cell death induced by the cellpermeable, active ceramide analogue, C 2 -ceramide, confirming that the block in the death pathway was prior to ceramide generation (Fig. 1D). Mammalian homologs of the Caenorhabditis elegans cell death protease CED-3 are thought to be distal effectors of the CD95 cell death pathway. Here we show that the apoptotic protease Yama/CPP32 (17,27,28) is activated by CD95 engagement (Fig. 1E). Endogenous Yama is expressed as a 32-kDa pro-enzyme and upon activation is proteolytically processed into active p17 and p12 subunits (17,27). One of the proposed substrates of Yama is the nuclear enzyme PARP (17,27). As expected, CD95-mediated activation of Yama and resulting PARP cleavage was blocked by FADD-DN (Fig. 1E).
In the yeast two-hybrid assay, FADD had a weak but specific interaction with TNFR-1 (9). To determine whether FADD has a role in TNFR-1-induced cell death, the FADD derivative was transfected into TNF-sensitive MCF7 breast carcinoma cells and stable cell lines generated. Interestingly, FADD-DN expressing MCF7 cells were equally resistant to both CD95-and TNF-induced cell death ( Fig. 2A, Table I), suggesting a proximal convergence of the cytokine-mediated cell death pathway. Additionally, 293T cells transiently overexpressing TNFR-1 were protected from cell death by co-transfecting FADD-DN (data not shown). As with the BJAB cell lines, the FADD-DN MCF7 cell lines were not resistant to staurosporine-induced apoptosis ( Table I). Overexpression of interleukin-1␤ converting enzyme (ICE) could "bypass" the dominant-negative effect of the FADD derivative, suggesting that the death pathway was being blocked upstream of the ICE-like proteases implicated in the apoptotic pathway.
While the main activity of CD95 is to trigger apoptosis, TNFR-1 can signal an array of diverse pro-inflammatory and immunoregulatory activities (29). Distinct from CD95, TNFR-1 is an inducer of nuclear factor B (NF-B) (30). We therefore investigated whether expression of FADD-DN modulated TNFinduced NF-B activation. MCF7 vector and MCF7-sFADD-DN cells were transfected with an NF-B-dependent reporter gene (5) and relative NF-B activity assessed (Fig. 2B). In both cell lines, NF-B was activated equally well, suggesting that FIG. 3. A mechanism for the inhibitory action of FADD-DN. A, 293T cells were co-transfected with AU1 epitope tagged FADD constructs and FLAG-tagged constructs encoding CD95, TNFR-1, and B94. The cells were lysed and FADD, or FADD-DN was immunoprecipitated with anti-FADD polyclonal antibody, run on a 15% SDS-polyacrylamide gel, and subsequently transferred to a nitrocellulose membrane. Coprecipitating FLAG-Fas and FLAG-TNFR-1 were identified by immunoblotting with anti-FLAG antibody. B94, a 73-kDa protein (31), did not co-precipitate with FADD, verifying the specificity of the protein-protein interaction. All transfected components were assessed for expression by immunoblotting cell lysates (data not shown). B, a truncated derivative of FADD exerts a dominant-negative effect by displacing endogenous FADD from activated CD95 and thereby inhibits DISC formation. BJAB vector (upper panels) and BJAB-sFADD-DN cells (middle panels) were metabolically labeled (with [ 35 S]cysteine and [ 35 S]methionine), lysed with Triton X-100, immunoprecipitated with anti-APO-1/PA-Sepharose, and subsequently analyzed by two-dimensional isoelectric focusing, 12% SDS-polyacrylamide gel electrophoresis. TNFR-1 utilizes FADD to transduce the death signal and activates NF-B by a different mechanism.
To determine the mechanism by which the FADD derivative exerts its dominant negative effect, co-immunoprecipitation of FADD and FADD-DN with CD95 or TNFR-1 was assessed. 293T cells were co-transfected with AU1 epitope-tagged FADD constructs and FLAG epitope-tagged CD95, FLAG-TNFR-1, or FLAG-B94 (31). Cell lysates were immunoprecipitated with anti-FADD antibody and subsequently immunoblotted with anti-FLAG antibody. TNFR-1 and CD95, but not a control cytoplasmic protein, B94, co-immunoprecipitated with FADD-DN (Fig. 3A). The association of FADD and FADD-DN with CD95 was 10-fold greater than with TNFR-1, correlating with the relative apoptotic potential of the two death receptors (32). Thus, our data suggest that FADD-DN exerts its inhibitory action by directly or indirectly forming a complex with the death receptors, preventing recruitment of endogenous FADD. It remains formally possible, however, that FADD-DN functions by binding and sequestering other death signaling molecules.
In the case of CD95, the endogenous signaling machinery was studied. Four proteins, termed CAPs (for cytotoxicity-dependent APO-1-associated proteins), associate with CD95 in a ligand-dependent fashion (23). CAP1 and CAP2 were shown to be FADD, while CAP3 and CAP4 remain unidentified (23). The oligomerized receptor, along with the associated CAPs, has been designated the DISC (23). CAP3 or CAP4 are not the recently described candidate signaling molecules FAP-1 or RIP, as antipeptide antibodies capable of detecting endogenous FAP-1 or RIP, respectively, were unable to detect either protein associated with activated or unactivated CD95. 2 As expected, in vector transfected BJAB cells, the DISC formed upon anti-APO-1 treatment (Fig. 3B). By contrast, anti-APO-1-stimulated sFADD-DN cells did not form a complete DISC. Instead, FADD-DN complexed with CD95 in a ligand-dependent fashion and inhibited the recruitment of FADD (CAP1 and CAP2), CAP3 and CAP4, thereby disrupting the DISC (Fig. 3B). Similar results were obtained using FADD-DN expressing MCF7 cells (data not shown). Thus, our results suggest that the N terminus of FADD, which is missing in FADD-DN, is required for the recruitment and assembly of the downstream DISC components, CAP3 and CAP4. Studies are underway to identify an analagous TNFR-1 DISC.
In conclusion, this is the first report to functionally establish FADD, one of numerous candidate signaling molecules, in the proximal signal transduction of CD95. Many of the contenders, including RIP and FAP1, were not found associated with the active or inactive receptor, nor are dominant negative inhibitors likely to exist. Of paramount importance is the fact that the dominant-negative version of FADD potently abrogates TNF-induced apoptosis but not TNF-induced NF-B activation. This suggests that FADD is the common conduit of the cytokine-mediated death signal and also demonstrates that the signaling pathways for TNF-induced apoptosis and NF-B activation are distinct (Fig. 3C). Taken together, our results demonstrate that FADD is both a necessary and sufficient mediator of CD95 and TNFR-1-induced apoptosis as overexpression of FADD engages the cell death machinery (9), while a truncated derivative acts as a potent dominant-negative regulator. Future studies will hopefully elucidate downstream components of the pathway, thus linking FADD to the apoptotic proteases of the ICE/ced-3 family.