Alteration of the sphingomyelin/ceramide pathway is associated with resistance of human breast carcinoma MCF7 cells to tumor necrosis factor-alpha-mediated cytotoxicity.

The interference of tumor necrosis factor-α (TNF) signaling processes with the acquisition of tumor resistance to TNF was investigated using the TNF-sensitive human breast carcinoma MCF7 cell line and its established TNF-resistant variant (R-A1). The resistance of R-A1 cells to TNF correlated with a low level of p55 TNF receptor expression and an absence of TNF signaling through TNF receptors. Stable transfection of wild-type p55 receptor in R-A1 resulted in enhancement of p55 expression and in partial restoration of TNF signaling, including nuclear factor-κB (NF-κB) activation. However, the transfected cells remained resistant to TNF-induced apoptosis. Northern blot analysis revealed a comparable induction of manganous superoxide dismutase and A20 mRNA expression in p55-transfected cells and in sensitive MCF7 cells, making it unlikely that these genes are involved in the resistance to TNF-mediated cytotoxicity. While TNF significantly stimulated both neutral and acidic sphingomyelinase (SMase) activities with concomitant sphingomyelin (SM) hydrolysis and ceramide generation in MCF7, it failed to trigger these events in TNF-resistant p55-transfected cells. In addition, the basal SM content was significantly higher in sensitive MCF7 as compared to the resistant counterparts. Furthermore, the TNF-resistant cells tested could be induced to undergo cell death after exposure to exogenous SMase or cell-permeable C6-ceramide. This study also shows that TNF failed to induce arachidonic acid release in p55-transfected resistant cells, suggesting that an alteration of phospholipase A2 activation may be associated with MCF7 cell resistance to TNF. Our findings strongly suggest a role of ceramide in the mechanism of cell resistance to TNF-mediated cell death and may be relevant in elucidating the biochemical nature of intracellular messengers leading to such resistance.

The interference of tumor necrosis factor-␣ (TNF) signaling processes with the acquisition of tumor resistance to TNF was investigated using the TNF-sensitive human breast carcinoma MCF7 cell line and its established TNF-resistant variant (R-A1). The resistance of R-A1 cells to TNF correlated with a low level of p55 TNF receptor expression and an absence of TNF signaling through TNF receptors. Stable transfection of wild-type p55 receptor in R-A1 resulted in enhancement of p55 expression and in partial restoration of TNF signaling, including nuclear factor-B (NF-B) activation. However, the transfected cells remained resistant to TNFinduced apoptosis. Northern blot analysis revealed a comparable induction of manganous superoxide dismutase and A20 mRNA expression in p55-transfected cells and in sensitive MCF7 cells, making it unlikely that these genes are involved in the resistance to TNF-mediated cytotoxicity. While TNF significantly stimulated both neutral and acidic sphingomyelinase (SMase) activities with concomitant sphingomyelin (SM) hydrolysis and ceramide generation in MCF7, it failed to trigger these events in TNF-resistant p55-transfected cells. In addition, the basal SM content was significantly higher in sensitive MCF7 as compared to the resistant counterparts. Furthermore, the TNF-resistant cells tested could be induced to undergo cell death after exposure to exogenous SMase or cell-permeable C 6 -ceramide. This study also shows that TNF failed to induce arachidonic acid release in p55-transfected resistant cells, suggesting that an alteration of phospholipase A 2 activation may be associated with MCF7 cell resistance to TNF. Our findings strongly suggest a role of ceramide in the mechanism of cell resistance to TNF-mediated cell death and may be relevant in elucidating the biochemical nature of intracellular messengers leading to such resistance.
Tumor necrosis factor-␣ (TNF), 1 originally described for its antitumor activity, is now recognized as one of the most pleiotropic cytokine to act as a host defense factor in immunological and inflammatory responses (1)(2)(3). Like other biological activities of TNF, its antitumor activity is exerted through binding to two distinct but structurally related cell surface receptors, p55 (TNF-R1) and p75 (TNF-R2). Gene knockout experiments and the use of receptor-specific agonistic antibodies confirmed that the two TNF receptors generate nonoverlapping signals (4 -7). Although there is now considerable evidence that p55 is the TNF receptor that directly mediates cytotoxicity in a wide variety of cell types (8), recent studies have indicated that p75-mediated signals may cooperate with p55 to facilite cell death in some cell types (9). Structure function analysis of p55 TNF-R signaling demonstrated that an 80-amino acid region within the cytoplasmic domain is required for initiation of apoptosis and NF-B activation (10).
The cytotoxic effect of TNF toward tumor cells can be affected by both intrinsic and acquired cell resistance. However, the current understanding of the molecular mechanisms critical for tumor resistance to TNF and for subsequent tumor progression remains limited. Cell surface expression of TNF receptors is necessary but not sufficient to induce a biological response, and post-receptor mechanisms are important in controlling the susceptibility to the cytotoxic action of TNF. The elucidation of the TNF signaling transduction pathway is particularly challenging because of the extremely wide variety of TNF responses. It is established that TNF regulates the transcription of several genes, many of which are regulated by NF-B (11). It is also clear that the activation of this transcription factor is a pivotal and integral event for the transfer of the TNF signal to the nucleus. Several mechanisms have been reported to contribute to cellular resistance to TNF-induced cell killing, including the constitutive expression of several protective proteins in resistant tumor cells, such as MnSOD, endogenous TNF, major heat shock protein hsp70, A20 zinc finger protein (12)(13)(14)(15). However, these proteins confer only partial protection against TNF cytotoxicity, suggesting that additional resistance mechanisms exist.
Recently, ceramide was reported to be an important lipid messenger in various pathways of TNF action (16,17). Ceramide can be generated from sphingomyelin (SM) hydrolysis by two types of early TNF-responsive sphingomyelinases (SMases), a membrane-associated neutral (N-)SMase and an endosomal acidic (A-)SMase (18,19). Ceramide targets may include a membrane-associated ceramide-activated protein kinase (20), a cytosolic ceramide-activated protein phosphatase (21), the mitogen-activated protein kinase cascade (22), and the stress-activated protein kinases (23). Recent studies established that ceramide generated by N-SMase directed the activation of proline-directed serine/threonine protein kinase(s) and phospholipase A 2 (PLA 2 ), while ceramide generated by A-SMase triggers the activation of NF-B, suggesting that the two SMases control important yet dissociable and nonoverlapping pathways of TNF receptor signaling (24). Because of the potential role of ceramide in mediating the cytotoxic effect of TNF, we examined its possible involvement in cell resistance to TNF. Our findings indicate that stable transfection of p55 TNF receptor restores TNF signaling including NF-B activation in TNF-resistant MCF7 variant R-A1, but does not restore the susceptibility of these cells to TNF cytotoxicity. The data presented in this study further support the notion that the apoptotic effect of TNF is probably dissociated from NF-B activation, and suggest that an alteration of sphingomyelinase activation and subsequent ceramide generation may represent an important additional mechanism by which human tumor cells may escape TNF-mediated apoptosis.
Cell Cultures and Stable Transfection of p55 TNF-R cDNA-TNF-resistant R-A1 cells were derived from a TNF-sensitive human breast carcinoma MCF7 cell line after continuous exposure to increasing doses of recombinant TNF-␣ (25). The wild-type human p55 TNF-R cDNA cloned in a mammalian expression vector pMPSVEH (26) was used to transfect R-A1 cells by the calcium phosphate precipitation method (27). Briefly, 1000 cells/10-cm tissue culture plate were plated. After 10 -14 days of selection in growth medium containing 200 g/ml G418 (Sigma), 4 -5 resistant colonies were isolated from each plate and examined for human p55 TNF-R expression by fluorescence-activated cell sorting. The positive clones were subsequently maintained in medium with 100 g/ml G418 for more than 2 months. The sensitivity of clones to TNF was tested every 2 weeks during culture. All cell lines were routinely cultured in RPMI 1640 medium containing 5% FCS, 1% penicillin-streptomycin, 1% L-glutamine at 37°C in a humidified atmosphere with 5% CO 2 .
Determination of Cell Viability and DNA Fragmentation-Cell viability was determined using the crystal violet staining method as described previously (25). Absorbance (A), which was proportional to cell viability, was measured at 540 nm. Cell viability (%) ϭ 100 ϫ (A 1 /A 0 ), cell lysis (%) ϭ 1 Ϫ cell viability (%), where A 1 and A 0 were the absorbance obtained from treated and untreated cells, respectively. The mean value of quadruplicate was used for analysis. Quantitative DNA fragmentation was determined as described previously (28). TNF-or ceramide-treated and untreated cells (1 ϫ 10 6 ) were pelleted and washed in PBS. Cells were then resuspended in lysis buffer (0.5% v/v Triton X-100, 20 mM EDTA and 5 mM Tris-HCl, pH 8.0) and centrifuged at 27,000 ϫ g for 20 min to separate the chromatin pellet from fragmented DNA. Both the pellet (resuspended in 1 mM EDTA and 10 mM Tris-HCl, pH 8.0) and supernatant were assayed to determine DNA by the spectrofluorometric DAPI procedure (29).
Flow Cytometric Analysis-Indirect immunofluorescence was performed by incubating of 1 ϫ 10 6 cells with TNF receptor antibodies (htr-9 and utr-1) for 1 h on ice in PBS, 1% FCS. Cells were then washed and stained with 1:50 dilution of affinity-purified biotinylated goat anti-mouse IgG for 30 min. After three washes with PBS, 1% FCS, cells were incubated with 50 l of streptavidin-phycoerythrin solution for 30 min. After additional washing with PBS, stained cells were analyzed using an EPICS profile II Coulter (Coultronic, Margency, France). Fluorescence data were collected on 5 ϫ 10 3 viable cells, as determined by forward light scatter intensity. Background fluorescence was deter-mined using PBS, 1% FCS instead of the TNF-R monoclonal antibody.
Measurement of Internalization and Degradation of Cell-bound TNF-TNF ligand internalization and degradation were estimated, as described previously by Tsujimoto et al. (30). Cells (1 ϫ 10 6 ) were exposed to 125 I-TNF (Amersham) at 4°C for 2.5 h and then washed with ice-cold medium and shifted to 37°C by adding prewarmed medium and further incubated at 37°C. At the times indicated, culture fluids were harvested, and trichloroacetic acid was added to a final concentration of 10% to quantitate degradation of internalized TNF by the cells. The soluble counts were determined after removal of the precipitate by centrifugation at 1500 ϫ g for 20 min. The cells were washed once with ice-cold PBS and incubated for 5 min at 4°C with 2 ml of 50 mM glycine-HCl buffer (pH 3.0) containing 150 mM NaCl. After removal of the glycine buffer, cells were washed twice and solubilized in 0.1% SDS. 125 I radioactivity found in the glycine buffer and that found in solubilized cells, respectively, represented surface-bound TNF and internalized intracellular TNF.
Nuclear Extracts and Electrophoretic Mobility Shift Assays-Human breast carcinoma MCF7 cells (15 ϫ 10 6 ) were incubated for 90 min in the presence or absence of TNF. Adherent cells were then trypsinized, and nuclear extracts were prepared according to the procedure of Dignam et al. (31). Gel mobility shift assays were performed with a synthetic double-stranded 31-mer oligonucleotide containing the B sequences of the human immunodeficiency virus long terminal repeat, 5Ј-end-labeled with [␥-32 P]ATP using the T4 kinase (32).
RNA Extraction and Northern Blot Analysis-Total RNA was extracted from the tumor cell lines according to the method of Chomczynski and Sacchi (33). RNA (15 g/lane) were electrophoresed in a 1.2% agarose gel and transferred to nitrocellulose membrane Hybond-C (Amersham). The membrane was then hybridized overnight at 42°C with the probe labeled with [␣-32 P]dCTP using a Megaprime DNA labeling system (Amersham). The hybridized membrane was washed and exposed to Hyperfilm-MP (Amersham). The blot was stripped by boiling in 0.1% SDS and probed again with a ␤-actin probe as a control for equal RNA loading.
Metabolic Labeling, Extraction, and Analysis of Cellular Phospholipids-For phospholipid labeling, cells were incubated with RPMI medium containing 5% FCS and 1 Ci/ml [9,10-3 H]palmitic acid (35.9 Ci/mmol, NEN du Pont, France). After 48 h of incubation, the radioactive medium was removed and cells were incubated for another 2-4 h in culture medium. Cells (3 ϫ 10 6 ) were resuspended in 1 ml of culture medium supplemented with 10 mM HEPES and treated with TNF at various times. Lipids were extracted by the method of Bligh and Dyer (34) and were separated by thin layer chromatography (TLC) using chloroform/methanol/water (100:42:6, by volume) followed by a second step using petroleum ether/diethylether/acetic acid (80:20:1, by volume) or hexane/diethylether/formic acid (55:45:1, by volume) as developing solvent systems. Radioactive lipid spots, detected with a Berthold radiochromatoscan and upon exposure to iodine vapor, were scraped into scintillation fluid and counted. The positions of ceramide on TLC plates were determined by comparison with concomitantly run 3 H-lipid extracts from MCF7 cells treated with exogenous bacterial SMase (100 milliunits/1.5 ϫ 10 6 cells). Statistical analysis was performed using Student's t test.
Analysis of Cellular SM Content-Cells (5 ϫ 10 6 ) were washed twice in phosphate-buffered saline, and lipids were extracted by the method of Bligh and Dye (34) and separated on TLC using chloroform/methanol/ water (70:35:5, by volume). The various spots detected after exposure to iodine vapor were determined for phosphorus content according to Böttcher et al. (35).
Assay for SMase Activity-The SMase assay was performed, as described previously by Wiegmann et al. (24). Cells were incubated with or without TNF for various times, and the stimulation was stopped by placing 2-3 ϫ 10 6 cell aliquots in a methanol-dry ice bath. To measure acidic SMase activity, cell pellets were resuspended in 0.1% Triton X-100 and incubated for 15 min at 4°C before homogenization. About 100 g of cellular lysate protein were incubated for 1 h at 37°C in a buffer containing 250 mM sodium acetate (pH 5.0), 1 mM EDTA, and [choline-methyl- 14

Lack of Signaling through TNF Receptors in the TNF-resistant Variant of MCF7-
In an attempt to examine the mechanism of TNF-resistance acquisition by tumor cells, we established a TNF-resistant variant R-A1, derived from MCF7 cells (25). As shown in Fig. 1A, R-A1 cells were resistant to TNF compared to the parental MCF7 cells. Flow cytometry analysis (Fig. 1B) indicates that while p55 TNF-R was highly expressed in parental MCF7 cells (80%), a lower level of p55 expression (30%) was observed in R-A1. Both cell lines displayed marginal expression of the p75 receptor (10%). Data of binding experiments (Fig. 1C) using 125 I-radiolabeled TNF show that receptor-bound TNF was rapidly internalized by TNF-sensitive MCF7 cells. In contrast, very little TNF binding and no TNF internalization were detected in R-A1 cells. Electrophoretic mobility shift assays for NF-B activation were performed to further evaluate the response of these cells to TNF. Treatment with TNF induced NF-B translocation in MCF7 but had no effect in R-A1 cells (Fig. 1D). The data of binding and internalization experiments are consistent with the failure of TNF to induce NF-B activation in R-A1. The hypothesis that resistance to TNF exhibited by R-A1 cells may be due to altered signaling through TNF receptors was examined next.
p55 TNF-R Expression in R-A1 Cells by Gene Transfection Restored NF-B Activation but Not TNF-induced Cell Lysis-Based on the above observations and as the p55 TNF receptor was reported to be responsible for TNF cytotoxicity signaling in most cellular models, we attempted to correct TNF signaling by transfecting R-A1 cells with a wild-type p55 expression vector, pMPSVEH-hup55 TNF-R. Following screening by fluorescence-activated cell sorting analysis using anti-TNF-R-p55 monoclonal antibody (htr-9), the stable transfected cells expressing a high level of cell surface p55 receptor (more than 70%) were selected for further study. Gel shift experiments were first performed to examine early response to TNF, i.e. NF-B activation, in the transfected cells. As shown in Fig. 2A, exposure of three representative R-A1 p55-transfected clones (clones 1001, 2101, and 3024) to TNF (50 ng/ml) resulted in the activation of NF-B, indicating that the TNF signaling pathway leading to NF-B activation was functional. Interestingly, despite NF-B activation, these p55-transfected clones remained resistant to TNF, even when a high concentration of TNF (200 ng/ml) was used (Fig. 2B). These data clearly indicate that wild-type p55 receptor expression and the resulting NF-B activation in R-A1 cells are not sufficient to trigger TNF cytotoxic activity.
The TNF Resistance of p55-transfected R-A1 Cells Is Not Associated with MnSOD and A20 Gene Expression-Strong evidence has been provided about the role of MnSOD and A20 in cell protection against the cytotoxic action of TNF (12,38). We therefore investigated the mRNA expression of these genes in the p55-transfected R-A1 cells to determine whether they were involved in the resistance to TNF cytotoxicity observed in these cells. Northern blot analysis showed that TNF treatment induced a similar increase in MnSOD and A20 mRNA expression in sensitive MCF7 and in p55-transfected R-A1 cells (Fig.  3). As expected, no increase in the mRNA expression level of either gene was detectable in control R-A1 cells due to the absence of TNF signaling. These data suggest that the resistance of these transfected cells was not directly related to Mn-SOD and A20 gene expression and further emphasize TNF signaling efficiency in these p55-transfected cells.
Failure of TNF to Induce SMase Activation and Ceramide Generation in TNF-resistant p55-transfected Cells-Ceramide, the product resulting from the breakdown of sphingomyelin, serves as the second messenger in the apoptotic signaling pathway (17). Recently, ceramide was reported to be an essential mediator in TNF-induced cell killing (39,40). We therefore investigated the possible interference of the sphingomyelin/ ceramide pathway with the mechanism of TNF resistance in MCF7-derivative cells. As shown in Fig. 4A, a concomitant increase in intracellular ceramide (ϳ2.4-fold increase) was detected in MCF7 at 10 -20 min of TNF incubation, preceded by a rapid SM hydrolysis (Ͼ30% decrease in SM content) that reached the maximum level within 5-10 min after TNF treatment. The measurement of neutral and acidic SMases indicates that the SM breakdown and ceramide generation in MCF7 cells was correlated with a significant induction of these two SMase activities (20% and 40% of control for neutral and acidic SMases, respectively) after 5-15 min of TNF incubation (Fig.  4B). In contrast, TNF did not induce neither ceramide generation nor SM hydrolysis in R-A1 and p55-transfected cells (clones 1001 and 3024) (Fig. 4A). This is consistent with the failure of TNF to stimulate both SMase activities in these cells (Fig. 4B). However, the defect of the SMase activation was not related to a decrease in basal SMase activities in the resistant cells, since the basal SMase activities were not significantly different in the three cell lines tested (e.g. N-SMase activities were 231, 385, and 258 pmol/h/mg proteins for MCF7, R-A1, and clone 1001 cells, respectively).
Cellular SM Content in TNF-sensitive and -resistant MCF7 Cells-In order to compare the basal level of SMase substrate in TNF-sensitive and resistant cells, the analysis of cellular SM content was performed. As shown in Table I, the percentage of SM (as compared to total phospholipids) was significantly higher in parental MCF7 cells (28 -30% of increase) than in R-A1 and clone 1001 cells. Moreover, the basal SM content was 2-fold higher in MCF7 than in the two resistant counterparts (20.9 nmol for MCF7 versus 10.7 nmol and 12.4 nmol for R-A1 and clone 1001 cells, respectively). This result suggests that the different responses induced by TNF in these cells may be dependent on the cellular SM content.
Exogenous SMase and Ceramide Trigger Cell Death in TNFresistant Cells-To determine whether the activation of SMase and the production of ceramide could overcome the resistance of transfected R-A1 cells, we examined the susceptibility of these cells to exogenous bacterial-derived SMase and synthetic cell-permeable ceramide. The addition of SMase (Fig. 5A) or C 6 -ceramide (Fig. 5B) was able to induce the killing of p55transfected cells as well as that of control R-A1 and parental MCF7 cells in a dose-dependent manner. The cytolytic effect of SMase and C 6 -ceramide on these cells was specific, since the addition of phospholipase D (data not shown) or C 6 -dihydroceramide (Fig. 5B) at equivalent concentrations failed to induce cell killing. Furthermore, DNA fragmentation analysis indicates that C 6 -ceramide killed both TNF-sensitive and TNFresistant MCF7 cells through an apoptotic pathway (Fig. 6). After 24 or 48 h of treatment of TNF-sensitive MCF7 cells, more apoptotic cells were observed upon C 6 -ceramide treatment than that upon TNF treatment, indicating that ceramide triggers apoptosis of these cells in a more direct manner than TNF.
TNF Induces the Release of Arachidonic Acid in Sensitive MCF7 but Not in p55-transfected R-A1 Cells-TNF has been described to be capable of activating PLA 2 and inducing the release of arachidonic acid (AA) from membrane phospholipids in several sensitive target cells (41,42). In addition, ceramide was recently reported to be capable of activating the PLA 2 /AA pathway (24). Initial experiments using dexamethasone, an inhibitor of PLA 2 , indicated that the addition of this component at a subtoxic concentration (100 ng/ml) efficiently inhibited (4-fold) the killing of parental MCF7 cells by TNF (Fig. 7A). Therefore, the involvement of PLA 2 in TNF signaling in MCF7 and transfected R-A1 cells was assessed by measuring the release of AA. As shown in Fig. 7B, TNF induced a significant release of 3 H-labeled AA metabolites (165% of control) in sensitive MCF7 cells after 18 h of incubation. No stimulation of AA release was observed during short term (0 -6 h) incubations (data not shown). In contrast, the increase in the release of 3 H-labeled AA in R-A1-and p55-transfected clones was not detected at any of times tested (Fig. 7B), suggesting that the  resistance of these cells to TNF may also be related to altered AA release.

DISCUSSION
In contrast to the rapid progress that has been made in defining gene products capable of regulating TNF-induced cell death, the knowledge of the molecular components involved in cell resistance to TNF remains limited. We attempted to delineate the functional role of some second messengers in the acquisition of tumor resistance to TNF by comparing a TNFsensitive human breast cancer cell line MCF7 with its R-A1 variant selected for resistance to TNF. This TNF-resistant variant derived from MCF7 was found to be susceptible to anti-Fas-induced cell lysis as well as to the chemotherapeutic drug adriamycin (data not shown), compelling evidence that the intracellular cell death pathway is functional in these cells. As compared to the parental MCF7 cells, the resistance of R-A1 cells to TNF correlated with a low level of p55 TNF receptor expression and an absence of TNF signaling through TNF-Rs. Although functional wild-type p55 receptor expression was reestablished in R-A1 cells by gene transfer as well as subsequent NF-B activation in response to TNF, these cells remained totally resistant to the cytotoxic action of TNF. It should be noted that transfection of rodent cells using the same vector efficiently triggered the cytotoxic effect of TNF (26). Our obser-vations suggest that NF-B is not sufficient to induce cell killing, and confirm that the activation of this nuclear factor and apoptosis are coincidental but that these two activities are separable. This is also in agreement with the report of Dbaibo et al. (43), indicating that the growth inhibitory effect of TNF is dissociated from the activation of NF-B in Jurkat lymphoblastic leukemia cells.
Cell resistance to the cytotoxic action of TNF is thought to be an active process requiring the synthesis of TNF-inducible proteins, since this resistance can be overcome by protein synthesis inhibitors in some experimental systems (44). Overexpression of several TNF-inducible early response genes, such as MnSOD and A20, has been reported to protect cells against TNF cytotoxicity (12,15). MnSOD acts as a superoxide radical scavenger, and its presence correlates with cellular protection against TNF cytotoxicity. A20 zinc finger protein is the product of a cytokine-induced primary response gene, and its overexpression inhibits TNF-induced activation of PLA 2 and TNFmediated apoptosis (38). Although the involvement of MnSOD and A20 in cellular protection has been established, our data indicate that both genes can be induced by TNF in p55-transfected resistant cells at similar level as compared to that in TNF-sensitive MCF7 cells, suggesting that the resistance of these cells is not related to overexpression of these two genes. Ceramide has emerged as a potent second messenger in TNF signaling, and a substantial amount of evidence has been accumulated in favor of ceramide functioning as a selective mediator of the cytotoxic/cytostatic effect of TNF (16,17). Ceramide generated by the activation of sphingomyelinase has been reported to mediate TNF-induced apoptosis in the human monocyte-like U937, human leukemic HL-60, and murine fibrosarcoma cell lines (39,40). This study shows that TNF can activate both neutral and acidic SMases in human breast tumor MCF7 cells. A defect in the activation of these enzymes is apparently sufficient to confer resistance to TNF in R-A1 cells. Indeed, when this defect was overcome by adding exogenous SMase or ceramide, the susceptibility of R-A1 cells to apoptosis was restored. This suggests that the stimulation of ceramideactivated enzymes may constitute an important step in the regulation of programmed cell death. It is worthy to note that in our study N-SMase was more rapidly activated than A-SMase, confirming that the activation requirements of A-SMase differ from that of N-SMase. This is also in agreement with the report by Wiegmann et al. (24), which suggests that Nand A-SMase activations may be triggered by distinct pathways. When specific inhibitors of two SMases become available, the determination of the nature of SMase involved in TNFinduced apoptotic cell death would be of major interest. It is tempting to speculate that the failure of TNF to induce DNA fragmentation and apoptosis in resistant cells may be related to structural membrane organization of SM, which may constitute a limiting step for the generation of this second messenger. Our data also demonstrate that the basal level of total SM content was higher in sensitive parental MCF7 cells, as com-pared to the resistant counterparts (R-A1 and clone 1001 cells). On the other hand, both sensitive and resistant cells showed a similar basal level of SMase activity. Thus, the lower level of basal SM content and the absence of SMases activation in resistant cells may represent a double blockage for ceramide generation. These results confirm the hypothesis from our previous report indicating that the failure of TNF to induce either SM hydrolysis or apoptosis in resistant myeloid leukemia KG1a cells was correlated with the SM pool used for TNF signaling, which is predominantly located in the inner leaflet of the plasma membrane (45). One could speculate that the lower TNF-hydrolyzable SM pool in p55-transfected R-A1 cells and the absence of SMase activation would explain the failure of TNF signaling to induce SM hydrolysis and ceramide generation in these cells. Although the sphingomyelin/ceramide pathway was reported to be capable of signaling NF-B translocation in HL-60 cells (46,24), we showed here that TNF induced NF-B translocation in p55-transfected resistant cells in the absence of ceramide generation. This is in agreement with other reports demonstrating that exogenous addition of a short chain ceramide to Jurkat cells or the inhibition of ceramide pathway had no effect on NF-B activation (47,48), and further confirming that NF-B activation by TNF can be independent of endogenous cellular ceramide generation.
Alternatively, non-induction of apoptosis in the p55-transfected cells could be due to an abnormal expression of other genes such as the proto-oncogene bcl2, involved in the regulation of apoptosis (49,50). However, the involvement of bcl2 in the resistance of R-A1 cells can be excluded, since the parental TNF-sensitive MCF7 cells displayed a higher bcl2 expression level than the resistant R-A1 cells (data not shown), suggesting that there is no correlation between bcl2 expression and the magnitude of TNF-induced apoptosis in these cells. In addition, we and others have reported that bcl2 acts downstream of ceramide preventing ceramide-induced cell death but not ceramide accumulation in at least two models of chemotherapyinduced cell death (51). 2 Our demonstration that TNF failed to generate ceramide in p55-transfected R-A1 clones overrides the implication of bcl2 in this phenomenon.
A complex pattern of integrated signals may be generated in response to an elevation of cellular ceramide content. Indeed, recent studies support the view that ceramide produced by N-SMase triggers the mitogen-activated protein kinase cascade via ceramide-activated protein kinase, which presumably results in the activation of PLA 2 (24,52,53). This is consistent with several lines of evidence suggesting the involvement of PLA 2 in the cytotoxic pathway of TNF (37,41,46,54). In addition, the AA generated by the activation of PLA 2 was reported to activate sphingomyelin hydrolysis in HL-60 cells (55). Data obtained in our studies indicate that the release of arachidonic acid is altered in p55-transfected TNF-resistant cells. This might be a consequence of the defect in ceramide generation. Although ceramide generation occurred rapidly (10 -20 min) following TNF treatment in the sensitive cells, the activation of PLA 2 and the AA release induced by TNF could not be detected in the first hours (0 -6 h) after TNF treatment. The AA release probably does not precede TNF-stimulated SM hydrolysis in our system, but is involved as a later biochemical response to the action of TNF. This is consistent with the report of Wiegmann et al. (24), suggesting that PLA 2 activation occurs as a later event in the TNF signaling pathway. Whether ceramide and AA function independently or in coordination to transduce TNF cytotoxic effect requires further investigation. Taken together, our data suggest that a selective defect in TNF signaling associated with an alteration in sphingomyelinase activation can, at least in part, confer resistance to TNF-induced cell death.
An insight into signal transduction by p55 TNF-R1 has resulted from the identification of the TNF-R1-associated protein TRADD, which interacts with the death domain of p55 and signals both cell death and NF-B activation (56). Following TNF treatment, the association of TRADD and TNF-R1 occurs rapidly in U937 cells (57). In addition, TRADD interacts with TRAF2 (58) and FADD/MORT1 (59, 60) leading, respectively, to NF-B activation and apoptosis induction in the overexpression systems. The importance of the signaling complex assembly is also demonstrated by the fact that dominant-negative derivative of FADD/MORT1 abrogated CD95(Fas)-induced apoptosis and ceramide generation in a B lymphoma cell line (61). One can suggest that in the case of TNF, ceramide generation may also be a downstream event, e.g. post-TRADD and/or post-FADD/MORT1 activation. We have obtained data indicating a comparable TRADD protein expression level in parental MCF7 as well as in resistant R-A1 and clone 1001 cells, and that transient TRADD-transfection induced apoptosis and NF-B activation in all these cells. However, no effect of TRADD overexpression on ceramide accumulation could be detected (data not shown). It seems unlikely that TRADD alone triggers cell death signaling through ceramide pathway. Whether TRADD/FADD complex formation occurs and interacts with the SM/ceramide pathway under physiological conditions remains to be determined. It would be of major interest to decipher the possible cross-talk between diverse TNF-R associated signaling molecules, such as FAN protein (62), and the sphingolipid messengers implicated in the TNF cytotoxic signaling pathway. Understanding of the molecular and biochemical mechanisms of tumor cell resistance to the cytocidal effect of TNF may ultimately provide new approaches to enhance the therapeutic efficacy of TNF against human malignancies.