The cytotoxicity of tumor necrosis factor depends on induction of the mitochondrial permeability transition.

Complete prevention of the killing of L929 fibroblasts by tumor necrosis factor α (TNF) in the presence of 0.5 μg/ml actinomycin D (ActD) was obtained with cyclosporin A (CyA), an inhibitor of the mitochondrial permeability transition (MPT), and aristolochic acid (ArA), a phospholipase A2 inhibitor. Peripheral benzodiazepine receptor (PBzR) agonists (PK11195, FGIN 1-27, or chlorodiazepam), agents known to potentiate induction of the MPT, potentiated the cytotoxicity of TNF in the absence of ActD, an effect prevented by CyA plus ArA. The MPT was demonstrated independently of its effect on viability as the CyA-sensitive loss of rhodamine 123 fluorescence from cells preloaded with the dye. Treatment with TNF and ActD resulted in the loss of 80% of rhodamine fluorescence within 6 h, a time prior to any loss of viability. CyA plus ArA completely prevented this effect of TNF. Potentiation of the cytotoxicity of TNF by PBzR agonists was associated with induction of the MPT, as assessed by the loss of rhodamine fluorescence. CyA plus ArA completely prevented the loss of rhodamine 123. Ceramide replaced TNF in killing L929 fibroblasts, an effect also prevented by CyA plus ArA. Ceramide in the presence of ActD resulted in the loss of rhodamine fluorescence, an effect that was again prevented by CyA plus ArA. In addition, CyA plus ArA prevented the ability of PBzR agonists to potentiate the cytotoxicity of ceramide. In the presence of each PBzR agonist, ceramide caused the loss of rhodamine fluorescence, an effect completely prevented by CyA plus ArA. D609, an inhibitor of phosphatidylcholine-specific phospholipase C, completely prevented the killing by TNF, but not by ceramide, in the presence of ActD. D609 prevented induction of the MPT occurring with TNF, but not with ceramide. Inhibitors of endocytosis, as well as lysosomotropic amines, prevented the cytotoxicity of TNF, but not that of ceramide. It is concluded that the MPT is causally linked to the genesis of irreversible cell injury with TNF. In the face of an inhibition of protein synthesis, the MPT occurs as a consequence of the formation of ceramide.

Tumor necrosis factor ␣ (TNF) 1 is a pleiotropic cytokine, which is produced primarily by activated macrophages and lymphocytes (1). Known to induce physiologic effects in a variety of cells and tissues, TNF is also implicated in the pathogenesis of certain diseases, most notably septic shock. In addition, TNF kills cancer cells in intact animals and a variety of cell lines in vitro. Although attributed to both apoptosis and necrosis, the biochemical basis of the cytotoxic action of TNF is still largely unknown. The difficulty in unraveling the mechanism of cell killing lies in the fact that TNF activates many signaling molecules and second messengers, including phospholipases, kinases, phosphatases, oxygen radicals, and transcription factors (2).
The L929 line of mouse fibroblasts has been widely used to explore the mechanism of the cytotoxicity of TNF. In these cells, the signaling pathways initiated by TNF lead to a death that is better characterized as necrosis rather than apoptosis (3,4). An alteration in mitochondrial structure and function with the resultant formation of reactive oxygen species seems to be an important step in the cytotoxic mechanism of TNF (5)(6)(7)(8)(9). The reported ability of antioxidants to protect against TNF cytotoxicity supports this hypothesis (10 -13). Nevertheless, the specific nature of the mitochondrial alteration induced by TNF and how it relates, in turn, to the initial signal transduction events remain to be defined.
The mitochondrial permeability transition (MPT) is the regulatable opening of a large, nonspecific pore in the inner mitochondrial membrane (reviewed in Refs. 14 -17). Although the molecular elements that form this pore have not been definitively established, they are presumed to derive from well known membrane constituents, including the adenine nucleotide translocator, porin molecules, and the complex forming the peripheral benzodiazepine receptor (14 -17). The MPT is a critical event in the killing of cultured hepatocytes that follows the inhibition of electron transport by anoxia, rotenone, cyanide, or N-methyl-4-phenylpyridinium (18,19). Cyclosporin A (CyA) prevents induction of the MPT in isolated mitochondria (20,21). Similarly, CyA prevented the MPT observed in intact cells made anoxic or treated with rotenone (22). In turn, CyA prevented the killing of hepatocytes by anoxia, rotenone, or cyanide (18,22).
In the present study, we have utilized the L929 line of mouse fibroblasts to document that the MPT is an essential event in the pathogenesis of the lethal cell injury induced by TNF. In addition, we provide an account that ceramide is an important part of the signal transduction pathway by which TNF leads to induction of the MPT.

MATERIALS AND METHODS
The L929 line of mouse fibroblasts (ATCC-CCL-1, American Type Culture Collections) was maintained in 25-cm 2 polystyrene flasks (Corning Costar Corp., Oneonta, NY) with 5 ml of Dulbecco's modified Eagle's medium (DMEM) (high glucose; without pyruvate) (Life Technologies, Inc.), containing 100 units/ml penicillin, 0.1 mg/ml streptomycin, and 10% heat-inactivated fetal bovine serum and incubated under an atmosphere of 95% air, 5% CO 2 . All experiments were performed 2 days after plating 1.0 ϫ 10 5 cells in 500 l of the above medium into 1.88 cm 2 wells of a 24-well microtiter plate (Corning Costar). By the second day, the cells were growing exponentially and had achieved a density of 2.5-3.0 ϫ 10 5 cells/well. Prior to treatment, fibroblasts were washed twice with Ca 2ϩ /Mg 2ϩ -free phosphate-buffered saline (PBS), after which 500 l of DMEM without serum was added to the wells.
Cells were pretreated for 30 min with one or more of the following chemicals. D609 (Biomol Inc.) and cyclosporin A (Sandoz, 50 mg/ml in cremophore) were dissolved in deionized, pyrogen-free water and added to wells in a 0.2% volume for a final concentration of 50 g/ml and 5 M, respectively; aristolochic acid (Biomol) was dissolved in PBS and added in a 0.2% volume for a final concentration 50 M; rotenone, antimycin A, and colchicine (Sigma) were dissolved in dimethyl sulfoxide and added to wells in a 0.5% volume for a final concentration of 25 M, 50 M, and 1 mM, respectively. Oligomycin and monensin (Sigma) were dissolved in Me 2 SO and added in a 0.2% volume for a final concentration of 0.1 g/ml and 10 M, respectively. Ammonium chloride (Sigma) was dissolved in deionized pyrogen-free water and added in a 0.2% volume for a final concentration of 10 mM.
Thirty minutes following treatment with the above chemicals, TNF Ϯ ActD or ceramide Ϯ ActD was added. TNF (Sigma; 22 units/ng) was dissolved in PBS and added to the wells in 0.2% volume at the final concentration indicated in the text. C 2 -ceramide or C 2 -dihydroceramide (Biomol) was dissolved in Me 2 SO and added in a 0.2% volume to give a final concentration of 6 M. ActD (Sigma) was dissolved in Me 2 SO, further diluted in PBS, and added to wells in 0.2% volume to a final concentration of 0.5 g/ml. Where indicated in the text, the cells were treated at the same time they were given either TNF or ceramide with one of the following peripheral benzodiazepine receptor (PBzR) agonists. PK11195, FGIN 1-27, and chlorodiazepam (Research Biochemical International) were dissolved in ethanol and added to the wells in a 0.2% volume for a final concentration of 10, 25, and 50 M, respectively. Clonazepam (Sigma) was dissolved in Me 2 SO and added in a 1% volume for a final concentration of 100 M.
Cell viability was determined at the times indicated in the text by the release of lactate dehydrogenase into the culture medium as described previously (23). Cell viability measured in this manner correlated closely with that determined by the uptake of propidium iodide or trypan blue (data not shown). The data from triplicate wells for each experimental point were averaged to obtain a single value for each point in each experiment. All experiments were repeated three times. Protein synthesis was determined by the incorporation of [ 3 H]leucine into an acid-insoluble precipitate as described previously (24).
The MPT was demonstrated in intact cells as a CyA-sensitive loss of rhodamine 123 fluorescence. Cells in a 24-well microtiter plate were washed twice with PBS and returned to DMEM without serum. Rhodamine 123 from a 500 M stock solution in water was added to give a final concentration of 5 M. The cells were incubated at 37°C for 1 h, then washed twice with PBS and placed in fresh DMEM without serum. The cells were treated as described in the text below, after which the medium was removed by aspiration and the cells washed again twice with PBS. Digitonin (Sigma) was dissolved in H 2 O and added in a 0.2% volume to give a final concentration of 7.5 M. CCCP (Sigma) was dissolved in Me 2 SO and added in a 0.2% volume to give a final concentration of 10 M. Phenylarsene oxide (PhAsO) (Sigma) was dissolved in Me 2 SO and added in a 0.2% volume to give a final concentration of 100 M. The cells were covered with a solution of 0.05% trypsin in PBS. Following aspiration of the trypsin, the cells were collected in 500 l of PBS, transferred to a microcentrifuge tube, and centrifuged at 700 ϫ g for 5 min. Following aspiration of the supernatant, the cell pellet was resuspended in 600 l of PBS. The cell suspension was transferred to a quartz cuvette, and the fluorescence of cell-associated rhodamine 123 was read in a Perkin-Elmer spectrofluorimeter at 505 nm (excitation) and 534 nm (emission).

RESULTS
The Killing of L929 Fibroblasts by TNF-Conditions for killing L929 fibroblasts by TNF are illustrated in Fig. 1. With these cells, the cytotoxicity of TNF does not depend on RNA or protein synthesis; rather, it is enhanced by inhibitors of transcription or translation (25). Accordingly, doses of TNF from 0.1 to 1.5 ng/ml did not lethally injure the fibroblasts for at least 14 h (Fig. 1A). However, in the presence of 0.5 g/ml ActD, the same doses of TNF killed an increasing proportion of the fibroblasts. With 1.5 ng/ml TNF, almost 80% of the cells died within 14 h. The time course of the killing of L929 fibroblasts by 1 ng/ml TNF in the presence of ActD is shown in Fig. 1. Dead cells were first detected between 6 and 8 h, and their number increased steadily between 8 and 14 h. Cycloheximide (1 M) similarly sensitized L929 fibroblasts to the cytotoxicity of TNF (data not shown).
Prevention by Cyclosporin A of the Cytotoxicity of TNF-The cell killing by TNF (in the presence of ActD or cycloheximide) depends on induction of the MPT. Three different criteria were used to document the participation of the MPT in the cytotoxicity of TNF: 1) prevention of the cell killing by CyA, an inhibitor of the MPT; 2) potentiation of the cytotoxicity of TNF by Inhibition by CyA of the MPT induced in isolated mitochondria de-energized by cyanide required the additional presence of a phospholipase A 2 inhibitor (18). Likewise, preservation by CyA of the viability of cultured hepatocytes treated with cyanide also required the presence of a phospholipase A 2 inhibitor (18). Similarly, complete protection against the cell killing by TNF was obtained in the presence of CyA plus aristolochic acid, a phospholipase A 2 inhibitor (26) (Fig. 2). CyA alone decreased the cytotoxicity of TNF by 45% and ArA alone by 35%. Other phospholipase inhibitors in combination with CyA also effectively prevented the cell killing by TNF (in the presence of ActD), including CDP-choline (27), ON-RS-082 (28), 3-(4-octadecyl)benzoylacrylic acid (29), and trifluoperazine (30) (data not shown).
The ability of CyA to prevent the cytotoxicity of TNF is not a consequence of the its binding to and, thus, inhibition of calcineurin, a calcium-dependent phosphatase. The immunosuppressive drug FK506 similarly inhibits calcineurin, but is inactive against the MPT (16). FK506 alone or in combination with a phospholipase A 2 inhibitor was without effect on the cytotoxicity of TNF (Table I). In addition, cypermethrin, another potent inhibitor of calcineurin, was unable to prevent the cytotoxicity of TNF (Table I).
Potentiation of TNF Cytotoxicity by Peripheral Benzodiazepine Receptor Agonists-Participation of the MPT was also evident in the ability of PBzR agonists to potentiate the cytotoxicity of TNF. The PBzR is associated with the outer mitochondrial membrane (31)(32)(33), and the PBzR agonist PK11195 (34,35) potentiates induction of the MPT in isolated mitochondria as well as in intact hepatocytes (36). Table II (37), or chlorodiazepam (38). In this experiment, the cells were not treated with ActD. Whereas TNF alone was again not toxic, almost 90% of the cells died in the combined presence of PK11195 and TNF. CyA plus ArA prevented the cell killing in the presence of TNF and PK11195. Similarly, FGIN 1-27 and chlorodiazepam potentiated the cytotoxicity of TNF (Table II), effects prevented by CyA plus ArA. By contrast, the central benzodiazepine receptor agonist clonazepam was without effect (Table II).
Cycloheximide potentiates the cytotoxicity of TNF as a consequence of an inhibition of protein synthesis. However, the potentiation of the cytotoxicity of TNF by PBzR agonists is not a consequence of a similar inhibition of protein synthesis. Table  III shows that neither PK11195, FGIN 1-27, nor chlorodiazepam had an effect on the rate of the incorporation of [ 3 H]leucine into protein at either 6 or 18 h after treatment with each agent. By contrast, treatment of the cells with 0.5 g/ml ActD resulted in a 90% inhibition of protein synthesis after 6 or 18 h.
Demonstration of the MPT in L929 Fibroblasts Treated with TNF-An assay was developed to demonstrate the MPT in intact L929 fibroblasts independently of the effect of the transition on cell viability. The assay is based on the ability of the fluorescent dye rhodamine 123 to accumulate in the mitochondria as a consequence of the mitochondrial membrane potential. The MPT causes the loss of the mitochondrial membrane potential, resulting in the release of the accumulated rhodamine. PhAsO is a potent inducer of the permeability transition in isolated mitochondria (39). Intact cells, however, are impermeable to PhAsO, but respond to it following their permeabilization with digitonin. Fig. 3 shows that 7.5 M digitonin had no effect on the rhodamine 123 fluorescence accumulated by cells preloaded with the dye. However, following permeabilization with digitonin, treatment with PhAsO resulted in the loss of 80% of the rhodamine fluorescence, an effect that was completely prevented by CyA (with or without ArA). Digitonin and CyA in the absence of PhAsO increased the fluorescence yield 25% compared with cells treated with digitonin alone. The basis for this increase is not known, but it may reflect inhibition by CyA of transient, physiological pore openings that reduce rhodamine retention. In any case, these data show that the CyA-inhibitable marked loss of rhodamine fluorescence can document the MPT under conditions where the loss of the mitochondrial membrane potential is a consequence rather than a cause of the transition.  completely prevented the TNF-induced loss at 2, 4, and 6 h. It deserves emphasis that there is no evident loss of viability prior to 8 h (Fig. 1). Thus, prevention of the loss of rhodamine fluorescence by CyA cannot be attributed to a nonspecific consequence of the preservation of cell viability. Fig. 4 also shows that treatment of the cells with CCCP, a proton ionophore that dissipates the mitochondrial membrane potential, similarly produced a time-dependent loss of rhodamine fluorescence from the cells. However, CyA (alone or plus ArA) had no effect on the rate or extent of the loss of rhodamine fluorescence. Thus, no conclusion can be drawn as to whether there is induction of the MPT with CCCP. In other words, the CyA-sensitive loss of rhodamine 123 fluorescence cannot be used to document the MPT under those conditions where the permeability transition is a consequence of the loss of the membrane potential. With either PhAsO or TNF plus ActD, however, the collapse of the mitochondrial membrane potential and the resultant loss of rhodamine 123 is clearly the consequence of the MPT.
Potentiation of the cytotoxicity of TNF by PBzR agonists was associated with induction of the MPT. Fig. 5 shows that treatment of the fibroblasts with TNF and PK11195 (in the absence of ActD) resulted in the time-dependent loss of rhodamine fluorescence. Within 6 h, 90% of the dye had been lost from the cells. The presence of CyA plus ArA completely prevented the loss of rhodamine fluorescence at 2, 4, and 6 h. PK11195 alone had no effect on the rhodamine content of the cells. Interestingly, there was also no loss of rhodamine fluorescence in response to TNF alone (in the absence of ActD) (Fig. 5), a result indicating that the inhibition of transcription promoted the cytotoxicity of TNF at the level of the induction of the MPT. In other words, the induction of the MPT in response to TNF requires the inhibition of protein synthesis.
In the presence of TNF (without ActD), the PBzR agonists FGIN 1-27 and chlorodiazepam also induced the MPT (Fig. 6). A time-dependent loss of rhodamine fluorescence, which was greater than 80% after 6 h, occurred with either FGIN 1-27 or chlorodiazepam in the presence of TNF, a result that parallels the ability of these two agonists to potentiate the cytotoxicity of TNF (Table II). Again, CyA plus ArA completely prevented the loss of rhodamine fluorescence (Fig. 6). There was no effect on the rhodamine content of the cells by either FGIN 1-27 or chlorodiazepam in the absence of TNF.

Ceramide Substitutes for TNF in Killing L929 Fibroblasts-
The lipid ceramide has been implicated as a second messenger in various pathways of TNF signal transduction (40,41). In the presence of ActD, 6 M ceramide killed L929 fibroblasts (see Fig. 2). In the absence of ActD, this dose of ceramide was not toxic. An inactive analogue of ceramide, dihydroceramide, did not kill fibroblasts in the presence of ActD (9 Ϯ 2% cells died over 18 h). The cell killing by ceramide depended on induction of the MPT, as assessed by the same criteria used in the case of TNF.
CyA plus ArA prevented the cell killing by ceramide (see Fig.  2). Treatment with ceramide and ActD resulted in a time-dependent loss of rhodamine fluorescence from the fibroblasts (Fig. 7). More than 80% of the rhodamine was lost from the cells within 6 h. Again, CyA plus ArA completely prevented the loss of rhodamine fluorescence caused by ceramide and ActD (Fig. 7).  PBzR agonists potentiated the cytotoxicity of ceramide. PK11195, FGIN 1-27, and chlorodiazepam potentiated the cytotoxicity of ceramide in the absence of ActD (Table IV). In each case, CyA plus ArA prevented the loss of viability with ceramide in the presence of a PBzR agonist (Table IV). Finally, with each PBzR agonist, ceramide caused the loss of rhodamine 123 fluorescence from the fibroblasts, an effect that was completely prevented by CyA plus ArA (Table IV).
Ceramide is generated as a consequence of the hydrolysis of sphingomyelin by either a neutral or an acidic sphingomyelinase (42). In the case of the acidic sphingomyelinase, binding of TNF to its 55-kDa cell surface receptor activates a phosphatidylcholine-specific phospholipase C, a plasma membrane enzyme that hydrolyzes phosphatidylcholine to yield phosphorylcholine and 1,2-diacylglycerol (40,(42)(43)(44). Following binding of TNF to the cell surface receptor and the activation of PC-PLC, the receptor complex is internalized within an endosomal vesicle (45) by an energy-dependent mechanism, which also acidifies the vesicle (46). Once activated by DAG and in the acidic milieu required for optimal activity, acidic sphingomyelinase liberates ceramide (42,43).
The xanthate D609 specifically inhibits PC-PLC (42,47). Fig.  8 shows that D609 completely prevents the killing of L929 fibroblasts by TNF in the presence of ActD. Whereas over 90% of the cells died within 18 h of exposure to TNF and ActD, only 15% of the cells died over the same time course in the presence of 50 g/ml D609. By contrast, D609 did not prevent the cell killing by ceramide in the presence of ActD (Fig. 8). In the absence of ActD, ceramide and D609 were not toxic.
D609 prevented the induction of the MPT in fibroblasts treated with TNF and ActD, as shown by the ability of D609 to prevent the loss of rhodamine fluorescence (Fig. 8). By contrast, D609 did not prevent the loss of rhodamine fluorescence that occurred in fibroblasts treated with ceramide and ActD. Finally, Fig. 8 indicates that ceramide alone (in the absence of ActD) did not induce the MPT, a result indicating that with ceramide, as with TNF, the effect of ActD is to promote the permeability transition. Importantly, CyA plus ArA still prevented both the loss of viability and the MPT in cells treated with ceramide and D609 (data not shown). In other words, D609 did not change the mechanism of cell killing by ceramide.
Metabolic inhibitors that deplete the fibroblasts of ATP can interfere with receptor-mediated endocytosis, as well as prevent the acidification of endosomal vesicles. Accordingly, rotenone, oligomycin, or antimycin A substantially reduced the cytotoxicity of TNF (Table V). As receptor-mediated endocytosis also depends upon intact microtubules (46), the depolymerization of microtubules by colchicine similarly prevented TNFinduced cell killing (Table V). Lysosomotropic agents such as ammonium chloride and monensin prevent endosomal acidification and protected against the cytotoxicity of TNF (Table V). Importantly, the cell killing by ceramide was not affected by rotenone, oligomycin, antimycin A, colchicine, monensin, or ammonium chloride (Table V). DISCUSSION The data presented above document that the MPT is an essential feature of the mechanism of the cytotoxicity of TNF. Evidence from three different studies supports this conclusion. CyA plus ArA, which inhibits the MPT, prevented the cytotoxicity of TNF (Fig. 2). Peripheral benzodiazepine receptor agonists, agents that potentiate induction of the MPT, potentiated the cytotoxicity of TNF, an effect that was prevented by CyA plus ArA (Table II). Finally, the MPT in response to TNF was assessed in intact fibroblasts as the CyA-sensitive loss of rhodamine 123 fluorescence (Fig. 4). In all cases, TNF-induced cell death was correlated with MPT occurrence.
Alternative explanations, other than an effect on the induction of the MPT for the protection afforded by CyA and the potentiation occurring with PBzR agonists, were ruled out. An interpretation of the protective effect of CyA as a consequence of an inhibition of calcineurin was excluded by the inability of two other inhibitors of this enzyme, FK506 and cypermethrin, to prevent the cell killing by TNF. Similarly, an interpretation

TABLE V Prevention of the cytotoxicity of TNF but not that of ceramide by inhibitors of endocytosis and lysosomotropic amines
The cells were pretreated with the either rotenone, oligomycin, antimycin A, colchicine, monensin, or ammonium chloride as described under "Materials and Methods." Thirty min later, the cells were treated with 2 ng/ml TNF and 0.5 g/ml ActD or with 6 M ceramide and ActD. Cell killing was determined 18 h after addition of TNF or ceramide. The data are the mean Ϯ S.D. from the average of duplicate determinations for each point from three separate experiments. Whereas the specific mechanism whereby the MPT develops in response to TNF is not known, the data presented here suggest that the generation of ceramide plays a role. Ceramide replaced TNF in both inducing the MPT (Fig. 7) and in killing L929 fibroblasts (Fig. 2). As with TNF, the cytotoxicity of ceramide was potentiated by PBzR agonists (Table II) and prevented by CyA plus ArA (Fig. 2). Finally, the data presented here are consistent with the mechanism whereby ceramide is generated by the activation of an acidic syphingomyelinase by DAG, which was formed, in turn, as a result of activation of PC-PLC. An inhibitor of PC-PLC prevented the cytotoxicity of TNF, but not that of ceramide (Fig. 8). Inhibitors of either receptor-mediated endocytosis or endosomal acidification prevented the cytotoxicity of TNF, but again not that of ceramide (Table V). These data are consistent with the scenario whereby the binding of TNF to its 55-kDa surface receptor activates PC-PLC, with the resultant formation of DAG. After the TNFreceptor complexes are actively internalized within acidic endosomal vesicles, DAG activates acidic sphingomyelinase. As a consequence, sphingomyelin is hydrolyzed, releasing ceramide, an event that leads, in turn, to induction of the MPT.
The mechanism of the cytotoxicity of TNF proposed here is consistent with previous observations utilizing L929 mouse fibroblasts. In particular, we confirm that mitochondrial inhibitors modulate the cytotoxicity of TNF (7). However, our data imply that the mechanism whereby such agents act is different from that proposed previously (7). Rotenone, oligomycin, and antimycin A prevent the cell killing by TNF (Table V). By depleting the cells of ATP, these agents prevent receptor-mediated endocytosis and, thus, prevent the formation of ceramide, the metabolite responsible for promoting induction of the MPT. Importantly, the inability of these same inhibitors to prevent the cell killing by ceramide (Table I) argues that a perturbation of mitochondrial function that leads to the formation of reactive oxygen intermediates (7) is not the primary mechanism of their protective effect. The alternative conclusion that ATP depletion is the relevant consequence of the action of these inhibitors also accounts for the recent observation that glutamine starvation protects against the cytotoxicity of TNF (48). In L929 fibroblasts, glutamine is the major energy source that drives ATP formation. Nevertheless, it is noteworthy that oxidants are potent inducers of the MPT (25), and antioxidants protect against the cytotoxicity of TNF (10 -13). Thus, it remains to be determined whether ceramide promotes the MPT by increasing the flux of activated oxygen species, or alternatively, whether such species form as a consequence of the MPT induced by ceramide by a mechanism unrelated to these reactive oxygen intermediates.
Finally, it deserves emphasis that the cytotoxicity of ceramide, like that of TNF, depends on the presence of either ActD or cycloheximide. According to our hypothesis, ceramide acts to promote induction of the MPT, an event linked to the loss of cell viability by a mechanism that remains to be defined. The MPT does not occur with TNF or ceramide alone. Accordingly, the action of ActD or cycloheximide must be to modulate the induction of the MPT by ceramide. ActD or cycloheximide most likely prevents the synthesis of a protective protein constitutively present in L929 fibroblasts that turns over rapidly or of a protein that is induced by TNF. Clearly, the induction of such a protein by TNF would allow cells to react to this cytokine without loss of viability, and the loss of such a response could readily account for the sensitivity of cancer cells to the cytotoxicity of TNF.