Selective Involvement of Ceramide in Cytokine-induced Apoptosis

Among its diverse biologic effects, the cytokine tumor necrosis factor α causes the rapid nuclear translocation of the transcription factor, nuclear factor κB (NF-κB). The p55 tumor necrosis factor (TNF) receptor shares with the related APO-1/Fas antigen the ability to initiate apoptosis. We investigated the role of the sphingolipid mediator ceramide in the cytokine-induced signaling mechanisms leading to NF-κB activation and cell death. Several lines of evidence presented here suggest that ceramide generated in response to TNFα or Fas activation is not involved in NF-κB activation. (i) Cell-permeable ceramides and exogenous sphingomyelinase failed to induce either nuclear translocation of NF-κB or degradation of its cytosolic inhibitor, I-κB, in Jurkat T cells. (ii) Ceramide treatment of cells inhibited phorbol ester-induced activation of NF-κB. (iii) TNFα potently activated NF-κB in a cell line deficient in acid sphingomyelinase. (iv) TNFα activated NF-κB within minutes without altering ceramide levels. (v) Treatment of Jurkat cells with cross-linking antibodies to APO-1/Fas induced large scale increases in ceramide and apoptosis without affecting NF-κB. (vi) Ceramide generation in response to Fas activation was inhibited byN-acetyltyrosinylvalinylalanylaspartyl chloromethyl ketone, a peptide inhibitor of interleukin-1β-converting enzyme-like proteases, whereas TNFα-induced NF-κB activation was unaffected by the inhibitor. These results show that ceramide accumulation belongs selectively to the apoptotic pathway(s) induced by cytokines, and, if anything, ceramide may participate in negative feedback regulation of NF-κB.

TNF␣ is a pleiotropic cytokine, which has a central role in mediating immune regulation and inflammatory response via binding to its 55-and 75-kDa membrane receptors, termed TNFR-1 and TNFR-2, respectively (for review, see Refs. 15 and 16). The APO-1/Fas antigen is a related member of the TNF receptor superfamily, which shares the ability to induce apoptosis in a number of hematopoietic cell lines (for review, see Ref. 17). Recent studies have shed some light on the upstream events that may mediate a common death signaling pathway for both TNF and Fas involving recruitment of the death domain-associated protein FADD (18) and the sequential activation of members of the interleukin-1␤-converting enzyme (ICE)-like protease family (19,20,45). TNF␣ and Fas also both induce sphingomyelinase activation and the generation of ceramide, which can induce apoptosis and may play a role in apoptotic signaling by these cytokines (7)(8)(9).
TNF␣ is additionally known to activate the transcription factor NF-B (21,22) which is thought to mediate the TNF␣induced expression of a variety of genes including the IL-2 receptor. NF-B belongs to the Rel family of transcription factors and in its inactive state exists in the cytosol as a heterodimer bound to the inhibitory complex I-B (for review, see Refs. 23 and 24). Stimulation at the cell surface by cytokines such as TNF␣ and interleukin-1␤ (IL-1␤) or by lipopolysaccharide initiates a poorly understood set of signaling events, which result in the phosphorylation and degradation of I-B, thus allowing the free NF-B dimer to translocate to the nucleus and initiate transcription of B-responsive elements (25,26). The TNF receptor-associated proteins TRADD and TRAF-2 have been implicated in signaling to NF-B by TNFR-1 (27,28). It is unclear whether ceramide generated in response to TNF␣ is involved in NF-B activation. Some studies have suggested an essential role for this lipid second messenger in NF-B activation and a dependence on ceramide generated by acid sphingomyelinase activity in particular (29 -31). However, other studies have provided evidence against a role for ceramide in this signaling pathway (12,(32)(33)(34)(35). Therefore, in the current study we sought to clarify the potential role of ceramide in the TNF␣ and Fas mechanisms of NF-B activation.
In this study, we demonstrate that cell-permeable analogs of ceramide were unable to induce either I-B degradation or nuclear translocation of NF-B in intact Jurkat T cells. Likewise, treatment of cells with bacterial sphingomyelinase, which has been shown to increase intracellular ceramide levels via cleavage of membrane sphingomyelin (36), failed to activate NF-B. TNF␣ remained a potent activator of NF-B in cells from a patient with Niemann-Pick disease type A (NPA), which lack acid sphingomyelinase activity (37,38).
Treatment of Jurkat T cells with cross-linking antibodies to APO-1/Fas caused both a marked increase in intracellular ceramide levels and apoptosis. However, Fas was unable to signal nuclear translocation of NF-B at early or late time points. Pretreatment of Jurkat cells with YVAD.CMK, a site-specific inhibitor of ICE-like proteases (39), inhibited both Fas-induced ceramide generation and apoptosis, but not TNF␣-induced NF-B activation. Furthermore, in Jurkat cells treated with TNF␣, we observed no increase in intracellular ceramide formation in the time course needed for activation of NF-B (1-10 min). Thus, Fas induced intracellular ceramide increases in an ICE-like protease-dependent manner without activating NF-B, whereas TNF␣ activated NF-B within minutes independent of ceramide and ICE-like proteases. Finally, we show that ceramide inhibits PMA-induced activation of NF-B. Taken together, these lines of evidence strongly suggest that ceramide is a specific component of apoptotic signaling pathways and not of the pathways leading to NF-B activation.

EXPERIMENTAL PROCEDURES
Materials-Jurkat T cells were obtained from ATCC, Rockville, MD. Niemann-Pick A and normal skin fibroblasts were obtained from the Coriell Institute (National Institute on Aging). TNF␣ was a kind gift from Dr. Phil Pekala (East Carolina University). Anti-Fas monoclonal antibody was purchased from Upstate Biotechnology, Inc. Staphylococcus aureus sphingomyelinase was purchased from Sigma. C 2 -and C 6ceramide were synthesized as described (2). [␥-32 P]ATP was from NEN Life Science Products. Poly(dI⅐dC) and poly(dN 6 ) were from Pharmacia Biotech Inc. Anti-NF-B monoclonal antibodies were purchased from Santa Cruz Biotechnology, Inc. Anti-I-B monoclonal antibody was purchased from Rockland, Inc. YVAD.CMK (Bachem Bioscience, King of Prussia, PA) was dissolved in Me 2 SO before addition to medium (final Me 2 SO concentration 0.2%, v/v), and appropriate solvent controls were used.
Nuclear Extracts-The nuclear extraction procedure was modified from Dignam (40) and Osborn (21). After treatments, medium was removed and approximately 10 7 cells were washed once in ice-cold PBS. The cell pellet was rapidly frozen in dry ice and ethanol and then thawed by resuspending in 100 l of ice-cold Buffer A (10 mM Hepes (pH 7.9), 10 mM KCl, 1.5 mM MgCl 2 , 1 mM dithiothreitol) resulting in approximately 100% lysis. The nuclei were pelleted by microcentrifugation at 3500 rpm for 10 min at 4°C. The supernatant was discarded, and the nuclei were suspended in 15 l of Buffer C (20 mM Hepes (pH 7.9), 0.4 M NaCl, 1.5 mM MgCl 2 , 25% (v/v) glycerol, 0.2 mM EDTA, 1 mM dithiothreitol, 0.5 mM phenylmethylsulfonyl fluoride). The suspension was mixed gently for 20 min at 4°C, and then microcentrifuged at 14,000 rpm for 20 min at 4°C. The supernatant was diluted with 40 -70 l of Buffer D (20 mM Hepes (pH 7.9), 50 mM KCl, 25% (v/v) glycerol, 0.2 mM EDTA, 1 mM dithiothreitol, 0.5 mM phenylmethylsulfonyl fluoride), and aliquots were stored at Ϫ80°C. Protein concentrations were determined using the Bio-Rad assay.
Electrophoretic Mobility Shift Assay-Reactions were performed in a 20-l volume, using 8 -10 g of nuclear extract in the presence of 1 g of poly(dI⅐dC), 1 g of poly(dN) 6 , and 10 g of bovine serum albumin. Incubations were in the presence of HDKE buffer with the following final concentrations: 20 mM Hepes, pH 7.9, 50 mM KCl, 1 mM EDTA, and 5 mM dithiothreitol. 1 l of radiolabeled oligonucleotide probe (20,000 -50,000 cpm) was added to each reaction. After incubation for 20 min, the reaction was terminated by adding 6 l of 15% Ficoll solution containing indicator dyes. For supershift experiments, 1 l of antibody was added to appropriate samples, which were then incubated for 1 h on ice prior to addition of Ficoll solution. Equal amounts of the reaction mixture were loaded on a 5% nondenaturing polyacrylamide gel in 1 ϫ TBE and were run at 200 V. Gels were transferred to Whatman filter paper, dried at 80°C for 2 h, and exposed to film at Ϫ80°C for 4 -12 h.
Oligonucleotides-The probe utilized was a synthetic NF-B consensus oligonucleotide with the following sequence: 5Ј-AGTTGAGGGGA-CTTTCCCAGGC-3Ј. It was end-labeled using T4 kinase and [␥-32 P]A-TP. The mutant oligonucleotide used in competition experiments had the following sequence: 5Ј-AGTTGAGGCGACTTTCCCAGGC-3Ј.
Western Blot Analysis-After treatments were carried out, cytosolic extracts from Jurkat cells were prepared by washing 10 7 cells in ice-cold PBS and resuspending pellet in ice-cold homogenizing buffer (20 mM Tris-HCl (pH 7.5), 250 mM sucrose, 10 mM EGTA (pH 7.4), 2 mM EDTA (pH 7.4), 1 mM phenylmethylsulfonyl fluoride, 0.02% leupeptin, and 0.1% Triton X-100). The cells were then lysed by sonication and ultracentrifuged at 40,000 rpm for 40 min at 4°C to separate cytosolic from nuclear and membrane components. An aliquot of the supernatant was removed for protein determination, and the remainder of the supernatant was mixed (1:1) with 2 ϫ sodium dodecyl sulfate sample buffer and boiled for 5 min. Samples containing equivalent amounts of protein were then analyzed by Western blot analysis using enhanced chemiluminescence (ECL) by Amersham.
Lipid Extraction and Ceramide Quantitation-Jurkat T cells were seeded at 5 ϫ 10 5 cells/ml and treated for the indicated times as described. Cells were harvested, and lipids were extracted by the Bligh and Dyer method (41); lipids were dried and resuspended in 1 ml of chloroform. Duplicate aliquots of 100 l were set aside for phosphate measurements (42), and 100 l was utilized in the Escherichia coli diacylglycerol kinase assay as modified for ceramide (43,44). Ceramide was quantitated by using external standards and was normalized to phosphate.
Sphingomyelinase Assay-Niemann-Pick fibroblasts were washed with PBS and resuspended in cold lysis buffer (25 mM Tris-HCl, pH 7.4, 5 mM EDTA, 1 mM ATP, 20 g/ml chymostatin, 20 g/ml leupeptin, 20 g/ml antipain, 20 g/ml pepstatin, 1 mM phenylmethylsulfonyl fluoride) to attain a final concentration of 5 ϫ 10 7 cells/ml. Cells were lysed via three cycles of freezing and thawing. Homogenate was obtained by centrifuging the total cell lysate for 10 min at 1000 ϫ g at 4°C. Sphingomyelinase activity was assayed in all fractions using 14 C-labeled sphingomyelin as described (45).

Treatment of Jurkat Cells with Exogenous Ceramide Analogs and Bacterial Sphingomyelinase Does Not Induce Nuclear
Translocation of NF-B-To investigate whether TNF␣-induced ceramide generation is a sufficient signal for NF-B activation, Jurkat T cells were treated with varying concentrations of synthetic cell-permeable ceramide analogs and were then assayed for nuclear translocation of NF-B (Fig. 1, A and B). These cell-permeable analogs have been shown to mimic the cytotoxic (apoptotic) effects of TNF␣ at micromolar concentrations (7). However, neither C 2 -nor C 6 -ceramide was able to induce nuclear translocation and activation of NF-B as compared with untreated and TNF␣-treated controls over both short and extended time courses. To evaluate the possibility that endogenously generated ceramide may provide a signal that the synthetic analogs lack, cells were treated with bacterial sphingomyelinase (Fig. 1C). Incubation of leukemic cell lines with bacterial sphingomyelinase has been shown to result in the hydrolysis of membrane sphingomyelin and the generation of intracellular ceramide in a dose-and time-dependent fashion (36). However, this treatment likewise failed to signal nuclear translocation of NF-B in Jurkat T cells.
Activation of NF-B was also studied by using Western blot analysis of its cytosolic inhibitor, I-B␣. Upon treatment of cells with a variety of inducers of NF-B, I-B␣ is phosphorylated and proteolyzed, thereby releasing NF-B and allowing the free heterodimer to enter the nucleus and bind to target gene promoter regions (23,24). I-B␣ proteolysis has been shown to be a necessary regulated step in the activation of NF-B by TNF␣ (25,26). Treatment of Jurkat T cells with TNF␣ resulted in proteolysis of I-B␣ within 10 min (data not shown), and after 30 min near-complete proteolysis of the band was observed (Fig. 2). Treatment with varying concentrations of cell-permeable ceramide and bacterial sphingomyelinase did not induce I-B␣ proteolysis as compared with untreated control (Fig. 2), further suggesting the divergence of ceramidemediated pathways from the signaling events leading to NF-B activation.
TNF␣ Activates NF-B in the Absence of Acid Sphingomyelinase-TNF␣ signaling through the 55-kDa receptor has been shown to result in activation of sphingomyelinase with both neutral and acidic pH optima, and there has been evidence that a ceramide signal generated by the acid sphingomyelinase in particular is a necessary and sufficient signal for NF-B activation (29). To more closely evaluate the possible role of acid sphingomyelinase in this pathway, we studied skin fibroblasts from a patient with Niemann-Pick disease type A, a lysosomal storage disease characterized at the cellular level by complete lack of acid sphingomyelinase activity and clinically by pathological sphingomyelin accumulation (37,38). To confirm the phenotype of the cell line, post-nuclear extracts from the Niemann-Pick fibroblasts and from age-matched normal skin fibroblasts were assayed for sphingomyelinase activity in neutral and acidic pH ranges (Table I). The Niemann-Pick fibroblast line completely lacked acid sphingomyelinase activity and displayed only a small amount of neutral sphingomyelinase activity, while the normal skin fibroblast line displayed acid and neutral sphingomyelinase activities of 49.25 and 2.50 nmol/mg protein/h, respectively. Despite a complete lack of acid sphingomyelinase activity in the Niemann-Pick fibroblasts, the TNF␣-induced nuclear translocation of NF-B remained unimpaired as compared with controls (Fig. 3A). Nuclear translocation was evident within 15 min of treatment with TNF␣. Therefore, the kinetics of NF-B activation in the enzymatically deficient fibroblast line are identical to those seen in the skin fibroblast controls and in Jurkat and other hematopoietic cell lines. The specificity of NF-B activation by TNF␣ in the Niemann-Pick fibroblasts is shown in Fig. 3B. A monoclonal antibody specific for the p65 subunit of the NF-B dimer caused retardation of the TNF␣-induced band, resulting in a characteristic "supershift," whereas an anti-c-Rel monoclonal antibody did not cause a supershift (lanes 3 and 4). Specificity was further demonstrated by competitive washout of the band by addition of excess cold B consensus oligonucleotide and the inability of mutant B oligonucleotide to compete with the band (Fig. 3B, lanes 5 and 6). This evidence suggests that activation of acid sphingomyelinase is not a required step in the signaling pathway linking TNF␣ binding at the cell surface to the activation of NF-B.
Fas Induces Ceramide Generation but Not NF-B Activation in Jurkat T Cells-Binding of the APO1/Fas cell surface antigen by either its ligand or cross-linking antibodies initiates a poorly understood set of signaling events resulting in programmed cell death in a number of hematopoietic cell lines (17). Recent work has begun to define the upstream mediators of Fas signaling, including the death domain-associated protein FADD (18). Other evidence has implicated ceramide as a downstream effector of the apoptotic pathway; treatment of SKW6.4 cells with a monoclonal cross-linking antibody has been shown to result in activation of membrane bound neutral sphingomyelinase and a subsequent 2-3-fold increase in intracellular ceramide levels within 16 h of treatment (13). In the present study, we treated Jurkat T cells with anti-Fas antibodies and  observed a greater than 5-fold increase in ceramide levels over control within 12 h and a greater than 7-fold increase over control within 20 h as assessed by diacylglycerol kinase assay (Fig. 4A). At the concentrations studied, anti-Fas antibody induced 80 -90% cell death after 24 h (data not shown). TNF␣ induces a well characterized activation of NF-B in Jurkat and other hematopoietic cell lines (21,22). However, unlike TNF␣, Fas did not signal nuclear translocation of NF-B (Fig. 4B). Whereas 2 nM TNF␣ caused nuclear translocation of NF-B within minutes (lane 2), treatment with anti-Fas antibodies failed to activate NF-B at either early or late time points (lanes 3-6). Thus, Fas signaling resulted in the generation of a ceramide signal that was associated specifically with cell death but not with NF-B activation.
YVAD.CMK Inhibits Fas-induced Ceramide Elevation and Apoptosis-Induction of apoptosis via APO1/Fas and TNFR-1 has been shown to involve recruitment of the death domainassociated protein FADD and activation of members of the ICE-like family of proteases (19,20,46). Sphingomyelinase activation and ceramide generation have also been shown to occur after treatment with Fas ligand and TNF␣ and have been suggested to play a role in inducing apoptosis (9,13). To determine whether sphingomyelinase activation and ceramide generation are downstream of the activity of ICE-like proteases, Jurkat T cells were treated with anti-Fas cross-linking antibodies after 30 min of pretreatment with 100 M YVAD.CMK, a site-specific tetrapeptide inhibitor of ICE-like protease activity (39) (Fig. 5A). Whereas Fas antibody alone induced a greater than 2.5-fold increase in intracellular ceramide over vehicle control after 4 h, Fas antibody following YVAD.CMK pretreatment induced a less than 1.5-fold increase in ceramide. After 12 h, Fas antibody alone induced a 6-fold increase in ceramide, whereas YVAD.CMK plus Fas antibody induced only a 2.5-fold increase. Thus pretreatment with the ICE-like protease inhibitor YVAD inhibited Fas-dependent ceramide generation by 67% and 70% at 4 and 12 h, respectively.
We also sought to determine if inhibition of ICE-like protease activity and ceramide generation would have an effect on Fasinduced apoptosis. At the time of lipid extraction, cell viability was determined by trypan blue exclusion (Fig. 5B). In samples treated with anti-Fas antibody alone, the increase in ceramide levels observed was accompanied by the induction of 70% cell death after 12 h. In samples pretreated with YVAD.CMK, Fas-induced apoptosis was reduced to 10% after 12 h, a 7-fold reduction in cell death.
TNF␣ Activates NF-B in a Ceramide-and ICE-like Protease-independent Manner-TNF␣ potently induces the nuclear translocation of NF-B in Jurkat T cells within minutes of binding to its cell surface receptor (21,22). To further examine whether TNF␣-induced activation of NF-B is dependent on ceramide as a second messenger, lipids were extracted and ceramide levels quantitated from Jurkat T cells after treatment of intact cells with TNF␣ (Fig. 6A). Within the time course required for NF-B activation in intact Jurkat cells (1-10 min), TNF␣ induced no appreciable changes in intracellular ceramide levels as compared with untreated controls, suggesting that ceramide is not a component of the signaling pathway leading to NF-B activation.
Members of the ICE-like protease family have been shown to be involved in APO-1/Fas-induced apoptosis (19,20), and appear to be upstream of sphingomyelinase activation and ceramide generation induced by this cytokine as described above.
To clarify whether ICE-like protease activity was additionally an upstream modulator of TNF␣-induced activation of NF-B, we treated Jurkat T cells with TNF␣ after 30 min of preincubation with YVAD.CMK (Fig. 6B). Pretreatment with 20 and 100 M YVAD.CMK had no effect on the ability of TNF␣ to potently induce nuclear translocation of NF-B within 10 min, suggesting that ICE-like protease activity is not involved in NF-B activation by TNF␣.
Ceramide Inhibits PMA Activation of NF-B-Additional studies examining the interactions of ceramide with other inducers of NF-B led to an investigation of the effects of C 2 -and C 6 -ceramide on activation of NF-B by PMA, an activator of protein kinase C that is known to activate NF-B. In these studies, cells were treated with 50 nM PMA alone or in the presence of 10 M C 2 -or C 6 -ceramide. PMA alone caused significant activation of NF-B as shown in Fig. 7A. Both C 2 -and C 6 -ceramide inhibited activation of NF-B in response to PMA (Fig. 7A). Interestingly, C 6 -and C 2 -ceramide at 5-20 M did not inhibit TNF␣-induced activation of NF-B (Fig. 7B), concentrations that were sufficient to inhibit PMA-induced activation, thus demonstrating that the effects of ceramide are not a result of nonspecific interruption of the NF-B complex. Importantly, these results demonstrate that ceramide is capable of inhibiting NF-B, probably through inhibition of the PKC pathway (which does not appear to participate in TNF␣ action; Ref. 24). DISCUSSION TNF␣ is known to induce a number of diverse biologic effects (the nature of which vary depending on target cell type) including cytotoxicity, cell differentiation, and antiviral activity (47). TNF␣ is one of a group of ligands, including certain cytokines, hormones, and growth factors, which cause the activation of sphingomyelinases resulting in the generation of the lipid mediator ceramide (for review, see Ref. 10). With the discoveries that TNF␣ could both induce sphingomyelin hydrolysis and ceramide generation (3) and cause nuclear translocation of NF-B through an undefined signaling mechanism, it was logical to pursue the hypothesis that ceramide may be the second messenger responsible for the TNF␣-induced activation of NF-B. However, despite the recent attention this hypothesis has received and the high degree of interest in delineating the exact mechanisms by which ligand binding at the cell surface results in nuclear translocation of NF-B, the potential role of ceramide in this process has remained unresolved.
Some evidence has suggested an essential role for SMase activity and ceramide generation in NF-B activation by TNF␣, the strongest of which has come from studies of permeabilized cells (29,31). Schü tze et al. showed that in permeabilized Jurkat T cells, treatment with exogenous SMase and with low (2.5-50 nM) concentrations of ceramide caused enhanced NF-B binding activity as assessed by EMSA (31). A subsequent study by this group using nuclei-free lysates of Jurkat cells showed in vitro induction of I-B␣ proteolysis by SMase and ceramide within 5 and 1 min, respectively (48). In addition to implicating ceramide in this pathway, these studies have also suggested that the TNF␣-induced activation of NF-B depends specifically on activation of acidic (endosomal) SMase rather than neutral, Mg 2ϩ -dependent (membrane-associated) SMase (29,31). A truncated form of the p55 TNF receptor lacking the ability to activate acid sphingomyelinase was unable to signal NF-B activation in response to TNF␣, suggesting an essential role for sphingomyelinase with acid pH optima (29). Two studies utilizing intact cell systems have suggested a role for ceramide in this process. Yang et al. (30) described enhanced NF-B binding activity on EMSA in response to direct treatment of HL-60 cells with exogenous bacterial SMase and N-octanoylsphingosine (C 8 -ceramide), although to a significantly lesser degree than that observed with TNF␣ treatment. Johns et al. (34) observed minimal activation of NF-B by exogenous ceramide on EMSA, and activation of an NF-Bdependent reporter to an extent comparable to TNF␣ only at very high ceramide concentrations (500 M).
Additional work, however, has provided conflicting evidence and has suggested that the sphingomyelin cycle and ceramide generation are not involved in the TNF␣-induced activation of NF-B (12,32,33,35,49). Betts et al. (32) observed potent NF-B activation in response to TNF␣ treatment of HL-60 cells, despite observing no appreciable changes in intracellular ceramide levels. Several studies demonstrated that cell-permeable analogs of ceramide could mimic the growth-inhibitory and apoptotic effects of endogenous ceramide but could not induce NF-B activation (12,32,49). Likewise, a study by our group and others recently demonstrated that, while C 2 -ceramide and bacterial SMase activated Jun nuclear kinase (JNK, also known as stress-activated protein kinase), these treatments failed to cause nuclear translocation of NF-B in HL-60 cells (49). A recent study by Higuchi et al. (35) using the myelogenous leukemia line ML-1a found that the addition of exogenous cell-permeable ceramide analogs was not sufficient to induce either NF-B activation or DNA fragmentation (in contrast to TNF␣, which was shown to induce both in this cell line). Thus there exists contradictory evidence as to whether or not a ceramide second messenger comprises an essential component of the signaling machinery that allows NF-B to dissociate from I-B and to translocate to the nucleus of target cells.
In contrast to the studies with TNF␣, little is known concerning the relationship of Fas activators, ceramide, and NF-B activation. Activation of Fas in several cell types results in ceramide accumulation, which has been associated with the apoptotic response of these cells (13,14). On the other hand, the relationship between Fas and NF-B appears to be more complex. Activation of Fas in U937 cells did not result in activation of NF-B (50), whereas activation of Fas in SV80 fibroblasts (transfected with Fas) and in T24 cells caused activation of NF-B (51,52). Since Fas shares with TNF␣ the apoptotic and ceramide responses, it potentially provides for a useful model system to investigate the relationship of ceramide to activation of NF-B.
Several observations in the present study argue against a role for ceramide in the TNF␣-induced signaling cascade leading to NF-B activation. First, cell-permeable C 2 -and C 6 -ceramide analogs, which mimic the apoptotic effects of endoge- nous ceramide, did not enhance NF-B binding in nuclear extracts of treated cells, nor were they observed to have an effect on the cytosolic inhibitor of NF-B, I-B. Likewise, direct addition of exogenous bacterial SMase did not cause I-B proteolysis or NF-B nuclear translocation ( Figs. 1 and 2). We also examined whether acid SMase activation in particular was a required component of TNF␣-induced NF-B activation (as has been proposed previously; Ref. 29) using Niemann-Pick A fibroblasts. We have shown here that cells completely deficient in acid sphingomyelinase retained the ability to normally activate NF-B in response to TNF␣ (Fig. 3). While this work was in progress, a study by Kuno et al. (33) likewise described the ability of Niemann-Pick fibroblasts to activate NF-B in response to TNF␣ and IL-1 despite lacking acid sphingomyelinase. Additional evidence against a role for acid SMase in cytokine activation of NF-B comes from work using the acid SMase inhibitor SR33557 (35). Inhibition of acid SMase prevented TNF␣-induced DNA fragmentation, but had no effect on TNF␣-induced NF-B activation. These results are in contrast to findings by Wiegmann et al. (29) and suggest that acid SMase activity is not necessary for NF-B activation. Also, we demonstrate here that TNF␣ is able to activate NF-B in Jurkat cells within minutes without affecting ceramide levels (Fig. 6A). The studies in the Fas-activated system also provide a clear-cut dichotomy in the activation of NF-B and ceramide accumulation. Whereas APO-1/Fas induced significant cell death and ceramide accumulation in the Jurkat cells, it was unable to induce NF-B activation at early or late time points (Fig. 4).
Taken together, these results suggest that ceramide is nei-ther a sufficient nor necessary signal to induce nuclear translocation of NF-B in intact cells. Indeed, the studies with PMAinduced activation of NF-B (Fig. 7) show that ceramide has inhibitory rather than stimulatory effects on activation of NF-B. The question remains as to why such discrepancies exist in the evidence put forth for the potential role of acid sphingomyelinase and ceramide in this signaling pathway. One possibility is that an accurate comparison of studies performed using permeabilized cells with those performed on membrane-intact cells is not possible with respect to this signaling cascade. Treatment with permeabilizing agents may fundamentally alter the intracellular environment in ways not easily quantified, for example, by inducing higher levels of proteolysis, such that the signaling events observed in this system may not be physiologically relevant. Moreover, those studies relied heavily on D609 as an inhibitor of phosphatidylcholine-specific phospholipase C, which has been proposed as an upstream activator of the acid sphingomyelinase. This putative phosphatidylcholinespecific phospholipase C is a poorly characterized enzyme, the specificity of the inhibitor has not been determined, and, at best, the studies with the inhibitor may implicate the phospholipase C and not the sphingomyelinase. The studies with the Niemann-Pick fibroblasts are more persuasive in ruling out a role for the acid sphingomyelinase in activation of NF-B.
Our evidence suggests a model in which TNF␣ and APO-1/ Fas are each able to initiate an apoptotic signaling pathway, which involves downstream activation of ICE-like proteases and ceramide generation, while TNF␣ additionally induces a distinct set of events, independent of ICE-like proteases and ceramide, resulting in immune modulation via NF-B activation (Fig. 8).
Recent studies have shown that trimerization of the TNFR-1 upon binding its ligand results in recruitment of the TNFR-1 death domain-associated protein TRADD (27). This complex, in turn, has been shown to directly interact with FADD. Thus, it appears that FADD is a point of convergence between the signaling cascades of TNF and APO-1/Fas (53). The common result is the FADD-dependent activation of a series of ICE-like proteases (including MACH/FLICE), which are beginning to be elucidated and which are thought to play a crucial role in the induction of apoptosis (19,20). Here we have shown that treatment with antibodies to APO-1/Fas causes up to a 6-fold elevation in ceramide levels and that ceramide generation precedes the induction of cell death in this line of Jurkat cells (Figs. 4 and 5). Additionally, using the ICE-like protease inhibitor YVAD.CMK, we show that ceramide generation appears to be downstream of the activity of ICE-like proteases in this pathway; pretreatment with YVAD.CMK inhibited ceramide generation and subsequent cell death (Fig. 5). These results concur with a recent study of the Drosophila protein Reaper, which showed that Reaper-induced ceramide generation and apoptosis were largely inhibited in a Drosophila cell line by a peptide inhibitor of ICE-like protease activity (54). Finally, we show here that TNF␣ is able to potently activate NF-B despite inhibition of ICE-like proteases by YVAD.CMK (Fig. 6B). These lines of evidence strongly support the hypothesis that ceramide generation is a downstream component of an apoptotic pathway which involves ICE-like protease activation and which is distinct from mechanisms leading to NF-B activation (Fig. 8).

FIG. 8. Schematic representation of proposed TNF␣-and APO-1/Fas-induced signaling events.
Solid arrows indicate direct interactions, and dashed arrows indicate indirect or incompletely defined interactions. Ceramide accumulation is downstream of ICE-like proteases inhibited by YVAD.CMK, but appears to be upstream of prICE, a distinct member of the ICE family that acts on poly(ADP-ribose) polymerase (55). See ''Discussion'' for additional discussion and references.