Glutathione Regulation of Neutral Sphingomyelinase in Tumor Necrosis Factor-α-induced Cell Death*

Tumor necrosis factor-α (TNFα)-induced cell death involves a diverse array of mediators and regulators including proteases, reactive oxygen species, the sphingolipid ceramide, and Bcl-2. It is not known, however, if and how these components are connected. We have previously reported that GSH inhibits, in vitro, the neutral magnesium-dependent sphingomyelinase (N-SMase) from Molt-4 leukemia cells. In this study, GSH was found to reversibly inhibit the N-SMase from human mammary carcinoma MCF7 cells. Treatment of MCF7 cells with TNFα induced a marked decrease in the level of cellular GSH, which was accompanied by hydrolysis of sphingomyelin and generation of ceramide. Pretreatment of cells with GSH, GSH-methylester, or N-acetylcysteine, a precursor of GSH biosynthesis, inhibited the TNFα-induced sphingomyelin hydrolysis and ceramide generation as well as cell death. Furthermore, no significant changes in GSH levels were observed in MCF7 cells treated with either bacterial SMase or ceramide, and GSH did not protect cells from death induced by ceramide. Taken together, these results show that GSH depletion occurs upstream of activation of N-SMase in the TNFα signaling pathway. TNFα has been shown to activate at least two groups of caspases involved in the initiation and “execution” phases of apoptosis. Therefore, additional studies were conducted to determine the relationship of GSH and the death proteases. Evidence is provided to demonstrate that depletion of GSH is dependent on activity of interleukin-1β-converting enzyme-like proteases but is upstream of the site of action of Bcl-2 and of the execution phase caspases. Taken together, these studies demonstrate a critical role for GSH in TNFα action and in connecting major components in the pathways leading to cell death.

Apoptosis, also known as programmed cell death, is an essential and closely regulated process important in the development and maintenance of multicellular organisms (1). Inducers of apoptosis include cytokines, chemotherapeutic agents, and stress conditions. Intracellular mediators of apoptosis include proteases of the interleukin-1␤-converting enzyme/Ced-3 family (2). The cytokine tumor necrosis factor-␣ (TNF␣) 1 induces apoptosis in target cells through binding to its cell surface receptor, TNF␣ receptor 1. Recently, it has been proposed that TNF␣ exerts its death signal through a series of protein domain-domain interactions; the cytoplasmic segment of TNF␣ receptor 1 binds to TRADD, which in turn associates with FADD, and the latter is thought to interact with MACH/FLICE (caspase-8) (3,4).
Generation of oxidative stress has been proposed as a critical event for TNF␣ as well as other death-inducing agents in the process of initiating their cytotoxic activity (5)(6)(7)(8)(9). Specifically, depletion of glutathione (GSH), the most abundant intracellular thiol-containing small molecule, has recently been found to either precede the onset of apoptotic cell death induced by various agents (10 -17) or render the cells more sensitive to apoptotic agents (18 -20), and it has been suggested that depletion of GSH is a relatively early event in the commitment to apoptosis (10). However, how this transmits cytotoxic signals remains to be answered.
Ceramide, the product of cytokine-activated and sphingomyelinase (SMase)-catalyzed hydrolysis of sphingomyelin (SM), is an important regulator of apoptosis (21). SM is one of the most abundant sphingolipid species in cell membrane with important structural and functional properties (22,23). Activation of SMases is believed to be involved in cell growth, differentiation, and apoptosis induced by cytokines, chemotherapeutic agents, and ionizing radiation (24). To date, five types of SMase have been described, and they differ in subcellular location, pH optimum, cation dependence, and role in cell regulation (25). The lysosomal acid SMase has been cloned, and this enzyme is activated in cells exposed to radiation (26), FAS (27), and TNF␣ (28,29). The neutral and magnesium-dependent SMase, although not yet cloned, has been implicated in mediating apoptosis in cells exposed to serum starvation, FAS, TNF␣, and cytosine arabinoside (29 -33). Relatively little is known about the biological relevance of the other three SMases, namely the zinc-dependent and lysosomal acid SMase-derived acid enzyme (34,35), the magnesium-independent N-SMase (36), and the alkaline SMase (37).
We have observed that the N-SMase from the human acute lymphoblastic leukemic Molt-4 cells is inhibited in vitro by GSH (38). This suggested to us that N-SMase may be a direct target for transmitting at least some of the effects of GSH depletion. In the current study, we report that GSH inhibits the TNF␣-induced activation of N-SMase in MCF7 cells, and depletion of GSH induced by TNF␣ involves activity of a cow pox virus cytokine modifier protein (cytokine response modifier A; CrmA)-sensitive interleukin-1␤-converting enzyme-like protease but not Bcl-2. The implications of our results in bridging the fields of oxidative stress and sphingolipid signaling are discussed.

Materials
MCF7 cells (39) were cultured in RPMI 1640 (Life Technologies, Inc.) containing 10% fetal bovine serum (Life Technologies, Inc.; complete medium) at 37°C in 5% CO 2 . For cells transfected by Bcl-2 and its vector, hygromycin (150 g/ml, Calbiochem) was included in the medium, and for cells transfected by CrmA and its vector, Geneticin (500 g/ml, Life Technologies, Inc.) was used. For treatment, cells were normally seeded at 5 ϫ 10 5 cells/10-cm culture dish (Falcon) in 10 ml of complete medium and grown for 48 h to 50 -75% confluence. Prior to the initiation of treatment, cells were rested in fresh complete medium without the selection antibiotic but with 25 mM HEPES, pH 7.4. GSH was made as a 200 mM stock solution in RPMI and added to cells 2 h before the addition of TNF␣. Unless otherwise indicated, the vector FIG. 1. GSH inhibits N-SMase from MCF7 cells. A, GSH, but not DTT or ␤-mercaptoethanol (␤-ME), inhibits N-SMase. N-SMase, partially purified from membranes of MCF7 cells, was preincubated for 5 min at 37°C in 20 mM Tris-HCl, pH 7.5, with the indicated concentrations of GSH, DTT, or ␤-mercaptoethanol followed by the addition of [ 14 C]sphingomyelin (100,000 dpm, 5 nmol) in 100 mM Tris-HCl, pH 7.5, containing 0.1% Triton X-100 and 5 mM MgCl 2 (total volume, 100 l). The reaction proceeded for 30 min at 37°C and was stopped by the addition of 1.5 ml of chloroform/methanol (2:1, v/v) and 0.2 ml of water. After phase separation, a portion of the upper phase containing the product [ 14 C]choline phosphate was counted for radioactivity by liquid scintillation. B, co-incubation with DTT does not affect inhibition of N-SMase by GSH. N-SMase was preincubated for 5 min at 37°C with the indicated concentrations of GSH and 0, 5, or 20 mM DTT. C, reversibility of inhibition of N-SMase by GSH. N-SMase was preincubated with 4 mM GSH for 5 min, diluted by 3-fold (1:3) or 5-fold (1:5) with 20 mM Tris-HCl, pH 7.5, and then assayed for activity. Results were corrected based on equal amounts of enzyme preparations used for each condition. Results shown are averages of duplicate determination and are representative of three separate experiments.

Methods
Partial Purification of N-SMase-MCF7 cells grown to near confluence in 175-cm 2 flasks were harvested by trypsinization, washed with ice-cold phosphate-buffered saline (PBS), pelleted by centrifugation, frozen in a methanol-dry ice bath, and stored at Ϫ80°C until use. To obtain N-SMase, detergent-solubilized membrane proteins were prepared from homogenate from pooled cell pellets (2 ϫ 10 9 cells) and resolved on a DEAE-Sepharose column (1 ϫ 10 cm) connected to an Amersham Pharmacia Biotech FPLC system essentially as described (38), except that both buffers A and B contained Triton X-100 (0.005%, w/v). N-SMase was efficiently resolved from A-SMase in MCF7 cells (25) by the detergent extraction step and DEAE column, such that the final N-SMase preparation contained Ͻ1% of A-SMase activity under the assay conditions. N-SMase Activity Assay-The activity of N-SMase was determined using a mixed micelle assay system as described (38). The reaction mixture contained enzyme preparation in 100 mM Tris-HCl, pH 7.4, 10 nmol of [ 14 C]sphingomyelin (100,000 dpm), 0.1% Triton X-100, and 5 mM magnesium chloride in a final volume of 100 l.
Measurement of Ceramide--Cells grown in 10-cm Petri dishes were rinsed with ice-cold PBS and scraped into methanol. Cell lipids were extracted by the method of Bligh and Dyer (31). Ceramide content was determined using a modified diacylglycerol kinase assay as described previously (39).
SM Level--Cells were seeded at 2 ϫ 10 5 cells/10-cm culture dish and grown for 48 h. Then cells were switched to fresh complete medium containing [ 3 H]choline (1 Ci/ml). After 48 h, cells were switched again to fresh medium and chased for 2 h before treatment with GSH and/or TNF␣ as described above. The level of SM was determined following a protocol essentially as described (40).
Measurement of GSH Level-Cells were seeded at 2 ϫ 10 5 in 6-cm Petri dishes in 4 ml of complete medium. Two days later, cells were treated with the desired agents as described above. Treated cells were detached by trypsinization, washed (three times) with ice-cold PBS, and solubilized in 150 l of water. 5-Sulfosalicylic acid was added to a final concentration of 2%, and the supernatant was separated from the acid-precipitated proteins by centrifugation. GSH content in the supernatant was determined by the Griffith (41) modification of the Tietze's enzymatic procedure (42) as described (38). Protein content was determined by the dye binding assay using bovine serum albumin as standard (43).
Western Blot for PARP-Cells were scraped into medium, pelleted by centrifugation, and washed (one time) with ice-cold PBS containing 1  ). C, TNF␣-induced ceramide accumulation is inhibited by GSH-ME and NAC. MCF7 cells were pretreated with 10 mM GSH-ME or NAC for 2 h followed by 3 nM TNF␣ for 16 or 24 h. For B and C, ceramide content was determined as described under "Experimental Procedures," and results are expressed as percentages of time-matched controls and are means Ϯ S.D. of duplicate determinations of three separate experiments. D, exogenous GSH could only partially restore the GSH levels of cells treated with TNF␣. Cells were pretreated for 2 h with 20 mM GSH, GSH-ME, or NAC followed by treating with 3 nM TNF␣ for 16 h. After extensive washing with ice-cold PBS, cells were lysed, and GSH content was determined as described under "Experimental Procedures." Results are means Ϯ S.D. of triplicate determinations of two experiments. mM phenylmethylsulfonyl fluoride. The cell pellet was resuspended in 50 l of PBS-phenylmethylsulfonyl fluoride and solubilized in 2ϫ SDS sample buffer. Western blot for PARP was performed as described (39,44).

FIG. 3. GSH inhibits SM hydrolysis and ceramide accumulation induced by TNF␣. A, GSH inhibits TNF␣-induced
Cell Viability-The viability of cells was determined by their ability to exclude trypan blue. The survival of cells was determined with the WST-1 cell proliferation reagent from Boehringer Mannheim. Cells were seeded at 10 3 cells/well/200 l of complete medium in a 96-well culture plate. Two days later, cells were treated in quadruplicate with TNF␣ as described above. At the end of treatment, WST-1 reagent was added, and after a 3-h incubation period the absorbance was measured at 450 nm with a multiwell plate reader as recommended by the manufacturer.

RESULTS
GSH Reversibly Inhibits Neutral SMase in Vitro-Previously, we found that GSH inhibited in vitro the N-SMase from Molt-4 leukemic cells (38). To study the role of GSH in TNF␣ signaling, we chose the human mammary carcinoma cell line MCF7, which is very sensitive to TNF␣ (39). We partially purified N-SMase from MCF7 cells following the procedure described for rat brain (38) and tested the effects of GSH on N-SMase in vitro. When the enzyme was preincubated for 5 min at 37°C with 1-20 mM GSH followed by incubation with substrate for 30 min, a dose-dependent inhibition of N-SMase by GSH was observed with a greater than 95% inhibition observed with 3 mM of GSH (Fig. 1A). Preliminary experiments established that with the minimum preincubation time examined (1 min), GSH (3 mM) inhibited the enzyme activity by Ͼ80% (data not shown). The inhibition was specific for GSH, since two other small thiol-containing molecules, dithiothreitol (DTT) and ␤-mercaptoethanol, at concentrations up to 20 mM, were ineffective (Fig. 1A), and co-incubation of GSH with 5 or 20 mM DTT did not alter the inhibitory profile for GSH (Fig.  1B). When the N-SMase/GSH (4 mM) mixture was diluted by 3or 5-fold, enzyme activity was recovered by 70 and 95%, respectively, suggesting that the inhibition was reversible (Fig. 1C). GSH did not inhibit the acidic SMase (38), although both the N-SMase and acidic SMase have been suggested to be activated in cells treated with TNF␣ (29,33). These results demonstrate that physiologic levels of GSH (3-10 mM) totally inhibit N-SMase. The results also suggest that the sharp drop in cellular levels of GSH may relieve this inhibition and cause activation of N-SMase.
To investigate the effect of GSH on N-SMase in TNF␣ signaling, MCF7 cells were treated with TNF␣ (3 nM) for 2-24 h, and GSH levels were measured. As shown in Fig. 2A, TNF␣ treatment resulted in an initial sharp drop in the level of GSH followed by a steady further decrease, with the first significant decrease observed at 8 h post-treatment. The most dramatic change in the level of GSH occurred between 8 and 10 h after TNF␣ treatment, where the GSH level plunged to nearly a third of that of control cells at the 10-h time point. GSH levels then steadily decreased to 7% of control cells by 24 h (Fig. 2A). At the 10-h time point, cells treated with TNF␣ did not manifest any detectable sign of apoptosis. In particular, proteolysis of the "cell death substrate" poly(ADP-ribose) polymerase (PARP) (45) was only detected at 12-24 h (39). The TNF␣induced decrease in cellular GSH level was accompanied by an increase in the level of ceramide, with an initial significant rise observed at the 8 -12-h time points. Ceramide levels at 12 h after TNF␣ treatment were twice control levels and reached more than 6-fold the control levels at 24 h (Fig. 2B). When the effect of TNF␣ on cellular SM level was examined, significant SM hydrolysis was observed between 10 and 16 h, and a 30% hydrolysis of SM was detected at 14 h (Fig. 2C). These kinetics raise the possibility that GSH may be involved in the regulation of SM hydrolysis and ceramide generation in cells treated with TNF␣.
Since GSH inhibited N-SMase in vitro at physiologically relevant concentrations and changes of GSH levels induced by TNF␣ preceded ceramide accumulation and SM hydrolysis, we next investigated the effects of manipulating intracellular GSH levels on TNF␣-induced SM hydrolysis and ceramide generation. First, replenishment of intracellular GSH by the addition of 10 mM GSH to the culture medium prior to TNF␣ treatment (3 nM; 14 h) completely prevented the TNF␣-induced SM hydrolysis (Fig. 3A). Treatment of cells with GSH (10 mM) alone had no effect on SM levels. Second, GSH pretreatment significantly inhibited the accumulation of ceramide induced by TNF␣ (Fig. 3B). A nearly complete inhibition of ceramide increase was observed at 12 h after TNF␣ treatment (Fig. 3B). At 16 and 24 h, ceramide accumulation induced by TNF␣ was inhibited by about 50%. Third, in addition to GSH, significant inhibition of TNF␣-induced ceramide accumulation was also achieved by treatment of cells with GSH methylester and Nacetylcysteine (NAC), which is converted intracellularly to cysteine, a precursor of GSH biosynthesis (Fig. 3C). These results further support the notion that GSH levels regulate the activity of N-SMase in these cells. The partial protection observed with exogenous GSH may be explained by the fact that the addition of GSH, GSH-ME, or NAC could not completely restore the intracellular GSH levels in cells treated with TNF␣. As shown in Fig. 3D, pretreatment of cells with 20 mM GSH, GSH-ME, or NAC prior to a treatment with 3 nM TNF␣ for 16 h only brought the intracellular GSH levels from 14.5% to 52.8, 64.7, or 57.9% of that of control cells, respectively.
Next, the biological consequences of changes in GSH levels in response to TNF␣ were studied by examination of cell death and survival. Whereas TNF␣ induced significant cell death, as determined by the ability of cells to exclude trypan blue dye, partial rescue of cells was achieved by replenishment of intra-cellular GSH with GSH added to the cell culture medium. Preincubation of cells with 10 and 15 mM GSH for 2 h prior to treatment with TNF␣ lowered TNF␣-induced cell death from 58.5% to 35.5 and 26.5%, respectively (Fig. 4A). The effect of TNF␣ on the viability of MCF7 cells was also determined using the WST-1 assay, which measures the activity of the mitochondrial respiratory chain in viable cells. TNF␣ significantly reduced the survival of MCF7 cells, and a 50% reduction in survival was observed when cells were treated with 1 nM TNF␣ for 24 h (Fig. 4B). Inclusion of 10 mM GSH enhanced the survival of TNF␣-treated cells to 75-95% of control level (Fig. 4B).
Mechanistically, the effects of TNF␣ on PARP cleavage, a close marker of the apoptotic response, were evaluated. TNF␣induced cleavage of PARP was partially inhibited by GSH and NAC in a dose-dependent manner (Fig. 4C). Since PARP is a substrate for "execution" phase proteases such as CPP32/ caspase-3 (46, 47), these results place GSH upstream of these proteases in the pathways leading to cell death.
The hypothesis that GSH levels regulate the activity of N-SMase in response to TNF␣ suggests that the product ceramide should function downstream of GSH and that it probably would not alter the cellular level of GSH. Bacterial SMases are known to induce a fast and significant elevation of intracellular ceramide by hydrolysis of membrane sphingomyelin (44,48). However, when MCF7 cells were treated with 300 milliunits/ml of bacterial SMase from Staphylococcus aureus or Bacillus cereus for up to 24 h, no significant alteration in GSH levels was observed (Fig. 5A). Similarly, treatment of cells with 2.5-10 M of the cell-permeable short chain C 6 -ceramide for up to 48 h did not induce significant reduction in the GSH level (Fig. 5B), although exogenous ceramide at these concentrations caused apoptosis in these cells (Fig. 5C, Ref. 39). Conversely, pretreatment of cells with 10 mM GSH prior to ceramide treatment did not protect cells from death induced by ceramide (Fig. 5C), in sharp contrast to the ability of GSH to inhibit the hydrolysis of SM and ceramide accumulation induced by TNF␣. Finally, preincubation of cells with the inhibitor of GSH biosynthesis, L-buthionine-(S,R)-sulfoximine (BSO), for 24 h did not render the cells more sensitive to death induced by C 6 -ceramide (Fig.  5D). The GSH level in the cells treated with 100 M BSO for 24 h was 14.1 Ϯ 1.3% of that of the control cells. These results place GSH upstream of N-SMase activation in the TNF␣ signaling pathway.
TNF␣-induced accumulation of ceramide has been shown to be inhibited by CrmA but not by Bcl-2 (39), placing activation of SMases downstream of CrmA-inhibitable proteases and upstream of Bcl-2 inhibitable proteases. To investigate the relationship between CrmA and GSH, MCF7 cells transfected with CrmA or empty vector were treated with TNF␣, and GSH levels were determined. No significant changes in GSH levels were observed in CrmA-transfected cells treated with 3 nM TNF␣ over a time period ranging from 1 to 24 h, whereas a time-dependent depletion of GSH was observed in the vector-transfected cells (Fig. 6A). The TNF␣-induced depletion of GSH was also inhibited in a dose-dependent manner by the substratebased tetrapeptide inhibitor of interleukin-1␤-converting enzyme-like proteases, YVAD (49). Pretreatment of MCF7 cells with 50 M YVAD prior to TNF␣ (3 nM) brought the GSH level from 20% of control to 80% of control (Fig. 6B). These results clearly demonstrate that the drop in GSH in response to TNF␣ is dependent upon activation of interleukin-1␤-converting enzyme-like proteases (such as caspase-8/FLICE).
Next, the effects of TNF␣ on the GSH levels in Bcl-2-transfected MCF7 cells were examined following treatment with 1-10 nM TNF␣ for 14 h. In both the Bcl-2-and vector-transfected cells, TNF␣, at a dose as low as 1 nM, caused a drop in GSH levels to 45% of that of untreated control cells (Fig. 7). Concentrations of TNF␣ greater than 1 nM and up to 20 nM further decreased the GSH level to 20 -30% of the control value. These results show that the drop in GSH levels is not downstream of the site of action of Bcl-2.
Finally, the interrelation of GSH, Bcl-2, CrmA, and TNF␣induced hydrolysis of SM was investigated. When SM levels were analyzed in CrmA-and Bcl-2-transfected cells, Bcl-2 had no effect on TNF␣-induced SM hydrolysis, and this SM hydrolysis in Bcl-2 cells was inhibited by pretreatment with GSH (Fig. 8). Cells with CrmA, however, were completely resistant to TNF␣-induced SM hydrolysis (Fig. 8). Thus, similar to the drop in GSH, activation of N-SMase is downstream of the site of action of CrmA and most probably upstream of the site of action of Bcl-2.

DISCUSSION
Our current study implicates GSH in regulation of a number of TNF␣-induced processes. Specifically, we show that TNF␣ causes a dramatic depletion of GSH, which is closely related to regulation of neutral sphingomyelinase and activation of proteases. We show that GSH inhibits N-SMase in vitro and provide evidence for regulation of N-SMase in cells. Replenishment of GSH prevents hydrolysis of SM and offers partial protection against TNF␣-induced generation of ceramide, PARP proteolysis, and cell death. Importantly, activation of interleukin-1␤-converting enzyme-like proteases, those that are inhibited by CrmA or YVAD, is implicated in the depletion of GSH. On the other hand, Bcl-2 does not modulate GSH depletion. Since Bcl-2 inhibits activation of the CPP32 family of proteases (Group II; Ref. 68) and since GSH plays a role in regulating proteolysis of PARP, the best studied substrate of this group, these results suggest that GSH depletion, SM hydrolysis, and ceramide accumulation (39,50) occur at a point upstream of activation of this group of proteases and upstream of the site of action of Bcl-2.
Depletion of GSH has been observed in response to many inducers of apoptosis, including TNF␣ (51-53), FAS (11,14), chemotherapeutic agents (54 -56), viral infections (7), and glutamate (20,57). An important implication of this study is the prediction that those agents that cause GSH depletion will also cause activation of N-SMase as a direct consequence of this depletion, which would relieve the enzyme from inhibition by GSH. Ceramide generated through this mechanism would connect GSH depletion to a number of downstream targets known to be modulated by ceramide, such as caspase 3/CPP32 (58), protein kinase C␣ (59), and phospholipase D (60 -64).
These observations also raise a number of important implications and questions. First, how does the activation of CrmAand YVAD-inhibitable proteases result in depletion of GSH? Second, although this study identifies N-SMase as a direct target regulated by GSH, it is conceivable that the drop in GSH will affect other direct targets that may "communicate" additional messages in response to this drop. Third, although the results suggest an important role for GSH in regulating N-SMase with complete inhibition of SM hydrolysis by GSH, the effects on ceramide levels are incomplete. This raises the possibility that ceramide is regulated by additional pathways not involving N-SMase, such as the acidic SMase or ceramide synthase, each of which has been implicated in ceramide formation (33,(65)(66)(67). Finally, GSH repletion did not reverse in full the effects of TNF␣ on PARP proteolysis or on cell death, suggesting that additional mechanisms are involved.
In conclusion, these results seem to connect important regulators in TNF␣-induced cell death. They provide for a direct target, N-SMase, that detects changes in the level of GSH, which may be of generalized significance in connecting oxidative damage with sphingolipid signaling.