TNF Receptor Death Domain-associated Proteins TRADD and FADD Signal Activation of Acid Sphingomyelinase*

Sphingomyelinase (SMase) activation and ceramide generation have emerged as an important signaling pathway transducing diverse biological effects of cytokine receptors like p55 tumor necrosis factor (TNF) receptor or Fas. Here we describe the TNF-dependent activation of acid SMase (A-SMase) through the p55 TNF receptor-associated proteins TRADD and FADD. Overexpression of TRADD and FADD in 293 cells did not change basal activity of A-SMase but enhanced TNF-induced stimulation of A-SMase. Other TNF R55-associated proteins like TRAF2 and RIP, which were reported to mediate TNF R55-mediated activation of nuclear factor κB, did not affect activation of A-SMase. Caspase inhibitors markedly reduced A-SMase activity, suggesting the involvement of an ICE-like protease in TRADD/FADD-mediated activation of A-SMase. Overexpression of caspase-8/a (FLICE/MACH) or caspase-10/b (FLICE2) did not change A-SMase activity, suggesting that TRADD/FADD-mediated activation of A-SMase involves a yet to be defined caspase-like protease distinct from caspase-8/a or -10/b.

Sphingomyelinase (SMase) activation and ceramide generation have emerged as an important signaling pathway transducing diverse biological effects of cytokine receptors like p55 tumor necrosis factor (TNF) receptor or Fas. Here we describe the TNF-dependent activation of acid SMase (A-SMase) through the p55 TNF receptor-associated proteins TRADD and FADD. Overexpression of TRADD and FADD in 293 cells did not change basal activity of A-SMase but enhanced TNFinduced stimulation of A-SMase. Other TNF R55-associated proteins like TRAF2 and RIP, which were reported to mediate TNF R55-mediated activation of nuclear factor B, did not affect activation of A-SMase.

Caspase inhibitors markedly reduced A-SMase activity, suggesting the involvement of an ICE-like protease in TRADD/ FADD-mediated activation of A-SMase. Overexpression of caspase-8/a (FLICE/MACH) or caspase-10/b (FLICE2) did not change A-SMase activity, suggesting that TRADD/FADD-mediated activation of A-SMase involves a yet to be defined caspase-like protease distinct from caspase-8/a or -10/b.
Ceramide produced by sphingomyelinases (SMases) 1 has been recognized as an important second messenger in membrane receptor signaling. Through binding to the p55 TNF receptor (TNF-R55), TNF rapidly activates two distinct forms of SMases, a membrane-associated neutral (N-) SMase and an acid (A-) SMase (1), which reside in caveolae (2) and the endosomal-lysosomal compartment. Each type of SMase hydrolyzes the phosphodiester bond of sphingomyelin to yield the neutral lipid second messenger ceramide and phosphorylcholine. Ceramide generated by N-SMase at the plasma membrane directs the activation of ceramide-activated protein kinase (CAPK) (3) that phosphorylates and activates Raf-1 kinase (4,5). Activated Raf-1 in turn phosphorylates and stimulates a dual specificity mitogen-activated kinase (MEK-1), which eventually phosphorylates and activates extracellular signal-regulated kinases (ERKs). The downstream targets of endosomal-derived ceramide produced by A-SMase are not yet defined but may include protein kinase C (PKC), JNK, and caspases (6 -8). N-SMase and A-SMase are activated independently by distinct cytoplasmic domains of TNF-R55 (1). N-SMase activation is mediated by neutral sphingomyelinase domain (NSD), spanning an 11amino acid motif N-terminally adjacent to the death domain of TNF-R55 (9,10). The NSD binds a WD-repeat protein, FAN, that mediates activation of N-SMase.
The domain of TNF-R55 activating the A-SMase pathway strikingly corresponds to the death domain signaling the cytotoxic effects of TNF (1,11). The cytoplasmic protein TRADD was recently identified to associate with the death domain of TNF-R55 in a TNF-dependent process (12,13). TRADD serves as an adapter protein that recruits other proteins to the cytoplasmic TNF receptor complex (14,15). Overexpression of TRADD potently activates distinct signaling cascades leading to activation of NF-B and induction of cell death (13). FADD, which directly binds to TRADD, has been shown to mediate activation of a pro-apoptotic protease, caspase-8/a (FLICE/ MACH), eventually leading to apoptosis (16,17).
Here we show that TNF-induced activation of A-SMase is mediated through TRADD and FADD. Neither caspase-8/a nor a dominant negative mutant of caspase-8/a alter the TNFinduced activation of A-SMase. However, several tetrapeptide caspase inhibitors as well as cytokine response modifier A (crmA) attenuated TNF-induced activation of A-SMase activity, suggesting that FADD mediates enhancement of A-SMase activity through activation of a protease distinct from FLICE.

MATERIALS AND METHODS
Cell Culture and Biological Reagents-The human embryonic kidney cell line HEK 293 was kindly provided by Dr. M. Schmidt, Essen, FRG. Hela and COS7 cells were obtained from ATCC. Cells were maintained in high glucose Dulbecco's modified Eagle's medium (ICN) supplemented with 10% fetal calf serum, 10 mM glutamine, and 50 g/ml each of streptomycin and penicillin in a humidified incubator at 5% CO 2 . Highly purified recombinant human TNF (3 ϫ 10 7 units/mg) was kindly provided by Dr. G. Adolf, Boehringer Research Institute, Vienna, Austria. Tetrapeptide caspase inhibitors were obtained from Bachem, Heidelberg, FRG.
Expression Vectors-Mammalian cell expression vectors encoding TRADD, TRAF2, RIP, crmA, caspase-8/a and dominant negative deletion mutants of this protein, as well as the control expression plasmids pRK5 and pCMV-␤-gal were kindly provided by Drs. D. V. Goeddel and M. Rothe, Tularik Inc., S. San Francisco, and have been described previously (12,13,18). FADD and caspase-10/b cDNA were kindly provided by Dr. V. Dixit, Genentech, San Francisco. The expression vector pRK-FADD was generated by inserting a SalI-HindIII fragment of FADD cDNA into the pRK5 vector. Deletion mutants of FADD were generated by inserting corresponding sequences in-frame into the expression vector pRK5.
For transient expression experiments, 1.5 ϫ 10 6 HEK 293 cells or 1.0 ϫ 10 6 COS7 or Hela cells were seeded on 100-mm dishes (Falcon, 3803). Cells were transfected the following day by the calcium phosphate precipitation method (19). After 9 h of incubation, cells were harvested, and enzymatic SMase assays were performed.
Assays for Neutral and Acidic SMases-HEK 293 or COS7 cells were transfected with different TRADD expression constructs, or with iden-* The costs of publication of this article were defrayed in part by the payment of page charges. This article must therefore be hereby marked "advertisement" in accordance with 18 U.S.C. Section 1734 solely to indicate this fact.
Hypodiploidy Apoptosis Assay-Cells were washed twice with cold PBS/EDTA, 5 mM, and 1 ϫ 10 6 cells were resuspended in 0.5 ml of PBS/EDTA and fixed by adding 0.5 ml of ethanol. After 30 min of incubation at room temperature, cells were centrifuged and resuspended in 0.5 ml of PBS/EDTA. 20 l of RNase A (1 g/ml) was added, and after incubation for 30 min at room temperature, 0.5 ml staining solution (500 g/ml propidium iodide in PBS/EDTA, 5 mM) was added. After 1 h of incubation at 4°C, cell cycle analyses were performed by flow cytometry using a Becton Dickinson Calibur FACS Analyzer.

NF-B-Electrophoretic Mobility Shift
Analysis-293 cells were transfected with 1.5 g of expression constructs and vector control. Cells were harvested the next day and left untreated or treated with 10 ng/ml TNF and incubated for 20 min at 37°C. Cells were washed two times with ice-cold PBS, and nuclear extracts were prepared as described previously (20). Electrophoretic mobility shift assays were performed as described (21).

RESULTS
Recent studies have identified a C-terminal domain, corresponding to the death domain of TNF-R55, that is responsible for activation of A-SMase and generation of the lipid second messenger ceramide (1,22). To examine whether one of the so far identified TNF-R55 death domain-associated proteins mediates the activation of A-SMase, 293 cells were first transfected with an expression vector coding for FLAG-tagged TRADD.
Expression of TRADD protein was monitored by Western blotting using a specific FLAG antibody (Fig. 1B). However, at 9 h after transfection, cells did not show any signs of apoptosis, which could be readily observed 24 -36 h post-transfection (data not shown). Overexpression of TRADD enhances TNFinduced A-SMase activity in a dose-dependent manner (Fig.  1A). Notably, basal levels of A-SMase activity remained unaffected, suggesting that the effects of TRADD on A-SMase were dependent on oligomerization of TNF receptors. Similar results were obtained in different cell lines. For example, TRADD also enhanced TNF-induced activation of A-SMase in COS7 cells (Fig. 1C). Basal activity of N-SMase as well as TNF-induced activation of N-SMase remained unaltered in 293 cells transfected with TRADD expression constructs (data not shown).
We next evaluated the effects of individual TRADD interacting proteins TRAF2, RIP, and FADD on the activation of A- SMase. TRAF2 belongs to a family of signal transducers sharing a highly conserved C-terminal TRAF domain, which binds to a N-terminal region of TRADD distinct from the death domain (13,23). TRAF2 was shown to mediate TNF-R55-induced activation of nuclear transcription factor B (NFB) (13,15). As shown in Fig. 2A, TRAF2 had no effect on A-SMase activity. Similarly, overexpression of a dominant negative mutant of TRAF2 (amino acids 87-501), lacking the N-terminal RING finger motif essential for NFB activation, had no inhibitory effect on A-SMase activity ( Fig. 2A). Finally, co-transfection with TRADD cDNA did not reveal TRAF2 action on A-SMase (Fig. 2D). The expression and biological function of TRAF2 cDNA were ensured by Western blot analysis and by NFBactivation in 293 cells, respectively (Fig. 2, B and C). These findings strongly suggest that TRAF2 is not involved in TNFinduced activation of A-SMase.
The receptor interacting protein kinase RIP has been shown to directly interact with TRAF proteins and the death domain of TRADD (13) and to be capable of inducing both the activation of NFB and apoptosis (24). As shown in Fig. 3A, neither overexpression of RIP nor a dominant negative deletion mutant RIP-DN, encoding only the death domain of RIP, altered TNFinduced activation of A-SMase. Notably, both expression constructs induced apoptosis when transfected into 293 cells (Fig.  3B). The lack of the ability of RIP-DN to signal for the activation of NFB may contribute to the greater extent of hypodiploid nuclei in HeLa cells when compared with the full-length RIP expression construct. Co-transfection of TRADD and RIP did not result in enhanced activity of A-SMase (Fig. 3C). These results suggest that RIP, like TRAF2, is not involved in TNFinduced activation of A-SMase.
It has been recently reported that overexpression of FADD causes the activation of caspase-8/a (FLICE/MACH) and eventually apoptosis (16,17). As shown in Fig. 4A, expression of FADD cDNA in 293 cells led to enhanced A-SMase activity in a dose-and TNF-dependent manner. Basal activity of A-SMase remained unchanged. A dominant negative mutant of FADD, lacking the N-terminal death effector domain (DED), FADD-DN, was found to decrease TNF-induced activation of A-SMase in a dose-dependent manner (Fig. 4B). Co-expression of FADD along with a constant amount of TRADD cDNA (Fig. 4C) enhanced the A-SMase activity slightly over the values obtained with FADD expression constructs alone. These findings indicate, that FADD, like TRADD, can enhance TNF-induced A-SMase activation.
To obtain a more detailed estimation for TRADD and FADDmediated enhancement of A-SMase activity, we performed enzyme kinetic analysis from cells transfected with expression constructs for TRADD and FADD. Apparent V max and K m values were determined for the substrate [N-methyl- 14 C]sphingomyelin by fitting initial rates of SM hydrolysis at various substrate concentrations to the Michaelis-Menten equation. As shown in Fig. 5, A-SMase has a basal activity of V max ϭ 21 nmol/h/mg and a K m of 11 M for sphingomyelin in 293 cells. The maximal velocity V max is enhanced upon TNF treatment to V max ϭ 33 nmol/h/mg, whereas K m remained unchanged. Overexpression of TRADD further enhanced the activity of A-SMase, yielding a maximal velocity V max ϭ 45 nmol/h/mg. The highest maximal velocity was observed with TNF-treated cells previously co-transfected with expression vectors for TRADD and FADD (V max ϭ 83 nmol/h/mg). In all instances, K m values remained unchanged.
The kinetic profile of A-SMase activation after TNF treatment is shown in Fig. 6. After stimulation of 293 cells with TNF, the activity of A-SMase is transiently increased, peaking at 3 min, and declined over a period of 10 min to reach basal levels again. Cells transfected with expression vectors for TRADD and FADD showed enhanced A-SMase activity at any time after TNF treatment, while basal levels of A-SMase activity remained unaffected.
Caspase-8/a (FLICE/MACH) is a downstream signaling molecule of FADD that transduces the TNF and FAS-induced death signal. When 293 cells were transfected with various doses of expression constructs for caspase-8/a the basal activity of A-SMase remained unchanged (Fig. 7A). Furthermore, overexpression of caspase-8/a (FLICE/MACH) did not alter the magnitude of TNF-induced A-SMase activation. However, caspase-8/a strongly induced apoptosis in these cells, within 15-24 h post-transfection, which was blocked by the coexpression of a 2-fold excess of crmA expression construct (Fig. 7B).  These findings ensured the expression of a functional caspase-8/a zymogen. A dominant negative mutant of caspase-8/a lacking 43 amino acids of the C terminus was shown to prevent TNF-induced cell death (25). However, overexpression of FLICE-DN did not alter activity of A-SMase (data not shown). Similarly, overexpression of caspase-10/b (FLICE2) did not alter basal or TNF-induced A-SMase activation in 293 cells (data not shown). These findings suggest that caspase-8/a or caspase-10/b are not involved in TNF-induced A-SMase activation in 293 cells.
To investigate a possible involvement of caspases other than FLICE/MACH in A-SMase activation, we finally employed a set of potent caspase inhibitors. When cow pox virus crmA was overexpressed in 293 cells, TNF-induced activation of A-SMase was markedly reduced (Fig. 8A). As shown in Fig. 8D, crmA also inhibited enhancing effects of TRADD and FADD on TNFinduced A-SMase activity. Again, the basal A-SMase activity in these cells remained unaltered. Similar results were obtained with tetrapeptide inhibitors of caspases, z-VAD.fmk, Ac-DEVD.CHO, and Ac-YVAD.cmk. Treatment of cells with Ac-DEVD.CHO, which preferentially inhibits caspase-3-like death proteases, resulted in reduced TNF activation of A-SMase (Fig.  8B). A more pronounced inhibitory effect was obtained with Ac-YVAD.cmk, an inhibitor of caspase-1 (Fig. 8C). In contrast to Ac-DEVD.CHO, Ac-YVAD.cmk also slightly reduced basal activity of A-SMase. The broad range caspase inhibitor z-VAD.fmk also reduced the enhanced activation of A-SMase in TRADD-and FADD-transfected 293 cells. These results suggest that FADD-mediated activation of A-SMase involves an ICE-like caspase activity distinct from caspase-8/a (FLICE/ MACH) and caspase-10/b (FLICE2). DISCUSSION Here we show that acid sphingomyelinase is activated through p55 TNF receptor-associated proteins TRADD and FADD. Overexpression of TRADD and FADD in 293 cells enhanced the TNF-induced activity of A-SMase. Activation of A-SMase could be blocked in cells expressing a dominant negative mutant of FADD. Basal velocity of A-SMase in cells transfected with TRADD/FADD expression plasmids was not altered, suggesting that cross-linking of the TNF-R55⅐TRADD⅐ FADD complex is essential for the activation of A-SMase. Enzyme kinetic analysis revealed that the maximal velocity V max of A-SMase is enhanced in TNF-treated 293 cells overexpressing TRADD or FADD. The highest maximal velocity of A-SMase was obtained in 293 cells transfected with both TRADD and FADD expression plasmids. Notably, the K m remained unchanged, indicating that the enzyme activity of A-SMase was enhanced rather than the accessibility of the substrate.
In contrast, overexpression of neither TRAF2 nor a dominant negative deletion mutant TRAF2-DN altered TNF-induced activity of A-SMase in 293 cells, suggesting that TRAF2 is not required for the activation of A-SMase by TNF. TRAF2 has been shown to mediate TNF-induced activation of SAP/JNK (26) and NFB (15), which involves a MAP3 kinase-related protein kinase, NIK, and the IB-kinase IKK␣ (27)(28)(29). In 293 cells, TRAF2 is apparently not involved in activation of a A-SMase, suggesting that the A-SMase activation pathway is distinct from that causing induction of NFB and SAPK/JNK. These data confirm a recent report by Hannun et al. (30), who showed that ceramide is not a direct member of the NFB activation pathway and rather may antagonize NFB activation in some systems. The previous notion, that A-SMase and ceramide activate NFB may be explained by tissue-specific activation and function of A-SMase or as a result of a nonspecific stress response caused by exogenous ceramide or SMase.
Overexpression of the receptor interacting protein RIP has been shown to be capable of inducing both activation of NFB and apoptosis. (26,31). Notably, RIP was found to be important for TNF-R55-mediated activation of NFB but not for Fas/Apo-1-initiated induction of apoptosis (32). The results of our study indicate that RIP does not alter TNF-induced activation of A-SMase, suggesting that RIP is not a component of TNF R55-triggered A-SMase activation pathway.
It is well established that FADD, through its death effector domain (DED), recruits caspase-8 (FLICE/MACH) to the DISC where it is activated (16,17,33,34). However, overexpression of caspase-8/a or of its dominant negative, did not affect TNFinduced activation of A-SMase. Thus, the apoptotic signaling of caspase-8 pathway to segregate from the A-SMase activation pathway at the level of FADD. While caspase-8/a did not seem to be involved in activation of A-SMase, tetrapeptide caspase inhibitors markedly inhibited TNF-induced A-SMase activation. In addition, overexpression of crmA, a potent inhibitor of ICE-like proteases, was able to block the TRADD and/or FADD-mediated enhancement of TNF-induced activation of A-SMase. Together, these results suggest that ICE-like protease distinct from caspase-8/a are involved in A-SMase activation. Hannun and co-workers (36) reported that ceramide signaled apoptosis in vincristine-treated ALL-697 leukemia cells via a caspase-3 (CPP32)-like protease and that this effect was inhibitable by Bcl-2 (35,36). Cells overexpressing Bcl-2, like wild-type cells, still responded to vincristine treatment with ceramide generation but failed to show PARP cleavage and apoptosis in response to a C6-ceramide analog. Further support for the notion that ceramide may be downstream of caspases is derived from experiments which showed that REAPER-induced ceramide generation was blocked by caspase inhibitor z-VAD.fmk (37). Furthermore, crmA, a pox virus serpin that is an inhibitor of ICE-like proteases but does not block caspase-3 (CPP32), abolished ceramide generation in TNF-treated cells (38). These studies suggested that ceramide generation occurs downstream of caspases. However, the source of ceramide was not addressed in these studies. The data of this report indicate that overexpression of TRADD and FADD enhances TNF-induced activation of A-SMase, which favors A-SMase to be downstream of the death domain adapter complex rather than ceramide synthase and other pathways of ceramide generation.
The question arises whether A-SMase is able to confer apoptosis in some instance of a specific cell type/stimulus combination. This idea is consistent with previous reports demonstrating that deletions of the death domain region of the TNF-R55 and CD95 (1, 39 -41), as well as overexpression of dominant negative FADD-blocked ligand-induced ceramide generation and apoptosis (42). Further, Dixit and co-workers (42) reported that treatment with ceramide analogs can bypass the anti-apoptotic effect of dominant negative FADD and restore apoptosis, indicating that apoptotic elements downstream of ceramide generation are intact in these cells.
The results of the present study clearly indicate that A-SMase activation can be dissociated from caspase-8-induced apoptosis. Overexpression of caspase-8 readily induced apoptosis of 293 cells in the absence of changes of A-SMase activity levels. Thus, A-SMase does not seem to be necessary for caspase-8-driven apoptosis. A-SMase seems neither sufficient for apoptosis of 293 cells because overexpression of A-SMase did not cause cell death. 2 The data do not rule out, though, that A-SMase constitutes an accessory pathway that can enhance the apoptotic response of 293 cells. The role of A-SMase in apoptosis induced by various stimuli in diverse cell types remains an open issue (for review see Ref. 43), and its precise delineation will require further extensive investigation. FIG. 7. FLICE/MACH (caspase-8/a) did not mediate TNF-induced activation of A-SMase. A, 293 cells were transfected with various amounts of caspase-8/a expression constructs or 10 g of control vector pRK5. Cells were stimulated with 100 ng/ml TNF for 3 min or were left untreated. A-SMase activities were determined and are expressed relative to untreated control vector cells. B, 293 cells were transfected with control vector pRK5, 2.5 g of expression caspase-8/a expression constructs, or caspase-8/a expression construct, and a 2-fold excess of crmA expression construct. Cells were harvested the next day, and the extent of hypodiploid nuclei in 293 cells were determined by propidium iodide cell cycle analysis.