Involvement of Protein Kinase C-β and Ceramide in Tumor Necrosis Factor-α-induced but Not Fas-induced Apoptosis of Human Myeloid Leukemia Cells*

The role of protein kinase C-β (PKC-β) in apoptosis induced by tumor necrosis factor (TNF)-α and anti-Fas monoclonal antibody (mAb) in the human myeloid HL-60 leukemia cell line was studied by using its variant HL-525, which is deficient in PKC-β. In contrast to the parental HL-60 cells, HL-525 is resistant to TNF-α-induced apoptosis but sensitive to anti-Fas mAb-induced apoptosis. Both cell types expressed similar levels of the TNF-receptor I, whereas the Fas receptor was detected only in HL-525 cells. Transfecting the HL-525 cells with an expression vector containing PKC-β reestablished their susceptibility to TNF-α-induced apoptosis. The apoptotic effect of TNF-α in HL-60 and the transfectants was abrogated by fumonisin, an inhibitor of ceramide generation, and by the peptide Ac-YVAD-BoMK, an inhibitor of caspase-1 and -4. Supplementing HL-525 cells with exogenous ceramides bypassed the PKC-β deficiency and induced apoptosis, which was also restrained by the caspase-1 and -4 inhibitor. The apoptotic effect of anti-Fas mAb in HL-525 cells was abrogated by the antioxidantsN-acetylcysteine and glutathione and by the peptide z-DEVD-FMK, an inhibitor of caspase-3 and -7. We suggest that TNF-α-induced apoptosis involves PKC-β and then ceramide and, in turn, caspase-1 and/or -4, whereas anti-Fas mAb-induced apoptosis utilizes reactive oxygen intermediates and, in turn, caspase-3 and/or -7.

TNF-␣ and Fas ligand (or specific agonistic monoclonal antibodies) induce apoptosis by binding to their respective death domain-containing receptors, TNF-RI and Fas (15). This domain is a protein-protein interaction motif that orchestrates the assembly of a signaling complex leading to the recruitment of the proapoptotic protease caspases. These proteases, related to the Caenorhabditis elegans death gene Ced-3, are cysteine aspartases that can be divided into two classes on the basis of the lengths of their N-terminal prodomains. Caspase-1, -2, -4, -5, -8, and -10 have long prodomains, whereas caspase-3, -6, -7, and -9 have short prodomains (16). Recently, caspases were shown to be key executioners of apoptosis mediated by various inducers, including TNF-␣ and anti-Fas mAb (16 -19).
The human HL-60 myeloid leukemia cell line is often used as a model system to study apoptosis in myeloid progenitor cells (1,20,21). From HL-60 cells, we have developed in our laboratory a variant cell line, HL-525 (22), that is deficient in PKC-␤ gene expression (23,24). These cell lines, therefore, serve as useful cell models for studying critical cellular events that require PKC, in particular PKC-␤ isozyme. Previously, it was reported that PKC is antagonistic to apoptosis mediated by certain inducers (1,25,26). Additional work, however, has indicated that PKC can act not only as a negative regulator of cell death but also a potentiator of the apoptotic effect of other inducers (20,21,27). The reason for this discrepancy may be due to the presence of discrete PKC isozyme patterns in different cell systems, with specific isoforms or combinations of isoforms having different and sometimes opposing effects. This study was initiated to (a) examine the role of one of these isozymes, PKC-␤, in TNF-␣ and Fas-mediated apoptosis in human myeloid progenitor cells and (b) determine the temporal relationship of this PKC with other apoptotic mediators, including ceramides, reactive oxygen intermediates, and caspases, in this process. murine goat immunoglobulin (Cy3 TM ) was from Jackson ImmunoResearch (West Grove, PA). TNF-␣ and the in situ apoptosis detection kit were from Roche Molecular Biochemicals. The ProtoBlot II AP system with stabilized substrate was purchased from Promega (Madison, WI).
Cells and Cell Culture-The human myeloid HL-60 leukemia cell line was originally obtained from R. C. Gallo of the NCI, National Institutes of Health. The HL-525 cells were established in our laboratory and have been described previously (22). The cells were incubated in tissue culture plates with RPMI 1640 medium supplemented with 10% heat-inactivated fetal calf serum (Intergen Co., Purchase, NY), penicillin (100 g/ml), streptomycin (100 g/ml), and 2 mM L-glutamine (Life Technologies, Inc.) at 37°C in an humidified atmosphere containing 8% CO 2 .
Transfection of Cells-Stable transfectants were obtained by electroporation using a Bio-Rad gene pulser apparatus with a capacitance extender in 0.4-cm gap electroporation cuvettes (Eppendorf Scientific, Madison, WI) as described previously (23). The pMV7-RP58 plasmid (kindly provided by I. B. Weinstein, Columbia University) contained both the full-length rat PKC-␤1 cDNA and the bacterial neomycin phosphotransferase gene (neo) that confers resistance to the antibiotic G418 (Geneticin, Life Technologies, Inc.). The pMV7 plasmid contained neo only. For each transfection, 5 ϫ 10 6 cells in 0.2 ml of medium were mixed with 10 g of supercoiled plasmid DNA and 0.2 ml of phosphatebuffered sucrose (272 mM sucrose, 7 mM Na 2 HPO 4 , pH 7.4) in a total volume of 0.5 ml. The cells were electroporated at 250 V and allowed to recover in 10 ml of serum-supplemented RPMI medium for 24 h prior to selection in a medium containing 0.5 mg/ml Geneticin. The Geneticinresistant transfectants were obtained by limited dilution in 24-well plates and tested for PKC-␤ expression and phorbol 12-myristate 13acetate inducibility of macrophage markers. The selected clones were maintained in Geneticin-containing medium. Because of their instability, we used pMV7-RP58 transfectants within 2 months (23), before they lost their functionally restored PKC-␤ phenotype.
Immunofluorescence-The immunostaining procedures were carried out at 4°C by using either 96-microwell plates or tissue culture chamber slides (Nunc, Inc., Naperville, IL). The cells (5 ϫ 10 5 /ml) were washed twice with PBSA (PBS solution containing 1% bovine serum albumin and 0.1% NaN 3 ) and incubated for 45 min with IgG 1 or 10 -50 g/ml specific mAb. The cells were then washed twice with PBSA and incubated for an additional 45 min with a goat anti-mouse secondary antibody conjugated to CY3 TM . After a wash with PBSA, the slides were mounted with phosphate-buffered Gelvatol TM (Becton Dickinson, Sunnyvale, CA). Fluorescence was examined by using a Micro-Tome Mac Digital Confocal system (VayTek, Fairfield, IA) attached to a Leitz Orthoplan fluorescence microscope.
Western Blotting-Pellets containing 10 9 cells were lysed in buffer (1% Triton X-100, 0.1% SDS, 1% sodium deoxycholate, 50 mM NaCl, 20 mM Tris-HCl, pH 7.4, 5 mM EDTA, 5 mM EGTA, 1 mM phenylmethylsulfonyl fluoride, 20 g/ml leupeptin, 20 g/ml aprotinin, and 1 mM sodium orthovanadate) by Dounce homogenization and stirred for 30 min on ice. Nuclei and cellular debris were removed by centrifugation at 16,000 ϫ g for 15 min. Samples containing equal protein amounts were subjected to SDS-polyacrylamide gel electrophoresis in 10% gels. Proteins were electroblotted onto polyvinylidene difluoride membranes or were stained with Coomassie Brilliant Blue R-250 to verify equal loading. Immunoblots were incubated with 1% bovine serum albumin in TBST (20 mM Tris-HCl (pH 7.5), 150 mM NaCl, and 0.05% Tween-20) for 30 min. The blocking solution was decanted and replaced with solutions containing mAbs to either Fas, TNF-RI, or TNF-RII in TBST at 50 g/ml and incubated with agitation for 2 h. After three TBST washes, the immunoblots were incubated with alkaline phosphatase-conjugated anti-IgG for 30 min, washed three times in TBS buffer (20 mM Tris-HCl (pH 7.5) and 150 mM NaCl), and then incubated with alkaline phosphatase substrate. When the color of the reaction developed to the desired intensity, the reaction was quenched by washing the immunoblots in deionized water.
Apoptosis Detection by the Terminal Deoxynucleotidyltransferase-mediated dUTP Nick End Labeling (TUNEL) Method-The experimental procedure for the TUNEL was conducted as described previously (29). Briefly, cell pellets were washed twice in PBS/1% bovine serum albumin at 4°C, adjusted to 10 6 cells/ml, and then transferred into a V-bottomed 96-well microtiter plate. A freshly prepared paraformaldehyde solution (4% in PBS, pH 7.4) was added to the cell suspension for 30 min at room temperature. After being washed twice, the cells were permeabilized by incubation with 0.1% Triton X-100 in 0.1% sodium citrate for 2 min at 4°C. As a positive control, 1 mg/ml DNaseI was added to permeabilized and fixed cells to induce DNA strand breaks. The TUNEL reaction was carried out with using the manufacturer's supplied buffer and terminal transferase and nucleotides for 60 min at 37°C in a humidified atmosphere in the dark. As a negative control, buffer without terminal transferase was added to the cells. After a wash, the pellet was resuspended in 10 l of phosphate-buffered Gelvatol TM and transferred onto slides. The incorporation of the fluorescein-labeled nucleotides into DNA strand breaks was detected by using the VayTek confocal fluorescent microscope.
Caspase Assay-Caspase activities in cell lysates were assayed with fluorogenic pNA substrates. Briefly, 10 l of cell lysate samples containing equal protein concentrations, were incubated for 1 h at 37°C in 0.1 ml of protease buffer (100 mM Hepes buffer (pH 7.6), 120 mM NaCl, 5 mM KCl, 1.2 mM MgSO 4 , 8 mM glucose, and 1% bovine serum albumin) in the presence of 0.1 mg/ml substrate. Formation of the pNA product in the supernatant was assayed at 405 nm.

Induction of Apoptosis in HL-60 and HL-525 Cells by TNF-␣
and Anti-Fas mAb-Initially, we examined the apoptotic effect of TNF-␣ and anti-Fas mAb in HL-60 and its variant HL-525 cells, because these agents have been reported to induce apoptosis (1,2). To assess apoptotic changes, we used an in situ method for detection of fluorescein-labeled DNA strand breaks, which is based on the TUNEL method (29,30). Unlike detection by the typical "DNA ladder" on agarose gels, the TUNEL method detects apoptosis at a single-cell level (29,30). By using this method we were able to show that treatment of HL-60 cells with 2 ϫ 10 3 units/ml TNF-␣ resulted in a time-dependent increase in the percentage of cells exhibiting apoptosis, whereas only a limited apoptotic effect was observed after incubation with the anti-Fas mAb (Fig. 1). The inverse situation was observed in HL-525 cells: TNF-␣ was inactive, whereas the anti-Fas mAb (but not the IgG control) caused a time-dependent increase in the percentage of cells exhibiting apoptosis (Fig. 1). The induction of apoptosis caused by TNF-␣ and anti-Fas mAb in HL-60 and HL-525 cells, respectively, was also dose-dependent (data not shown). These results indicate that TNF-␣ and anti-Fas mAb have divergent apoptotic effects in HL-60 and HL-525 cells.
TNF-␣ and Fas mAb induce apoptosis by binding to their respective receptors, TNF-RI (and possibly TNF-RII) and Fas (15,(31)(32)(33)(34)(35). Because the absence of such receptors may account for the inability of their respective ligands to generate an apoptotic effect, we compared the expression of these receptors in HL-60 and HL-525 cells by Western blotting. We detected similar levels of the TNF-RI and the weakly expressed TNF-RII in both cell types, whereas the Fas receptor was detected in HL-525 cells but not in HL-60 cells (Fig. 2). In addition, to demonstrate that these proteins were not only expressed but also manifested on the cell surface, we conducted immunostaining of viable HL-60 and HL-525 cells using the same antibodies used in the Western blotting experiments. The results (data not shown) demonstrated a similar pattern and level of cell surface manifestation of these receptors as seen in the Western blot. These results suggest that the inability of HL-60 cells to undergo anti-Fas mAb-induced apoptosis is caused by the absence of the Fas receptor in these cells, whereas the inability of TNF-␣ to induce apoptosis in HL-525 cells is due either to a defective TNF-RI/II or to an impaired TNF-␣ signaling pathway. It has been suggested that cooperation of both TNF-RI and TNF-RII may mediate the TNF-␣induced apoptosis process (31)(32)(33)(34)(35). One would speculate that receptor ligation with neutralizing mAbs to the individual TNF-␣ receptors could distinguish which receptor mediated the apoptotic effect of TNF-␣ in HL-60 cells. By using specific anti-TNF-RI and TNF-RII mAbs (at 10 g/ml), we found that induction of apoptosis in HL-60 cells by TNF-␣ was attenuated by more than 70% using the anti-TNF-RI mAb, whereas the anti-TNF-RII mAb was ineffective, suggesting that TNF-RII is not critical for this process.

Role of Protein Kinase C-␤ in TNF-␣-induced Apoptosis-We
have previously demonstrated that HL-525 cells are deficient in PKC-␤ gene expression and that restoration of PKC-␤ gene expression results in recovery of sensitivity to phorbol 12myristate 13-acetate-induced macrophage differentiation (22)(23)(24). To determine whether deficient expression of PKC-␤ or another PKC isozyme plays a similar role in the resistance to TNF-␣-induced apoptosis, we screened the HL-60 and HL-525 cell lines for the expression of nine PKC isoenzymes using immunofluorescence. We found that, with the exception of PKC-␤, which is expressed in HL-60 but not in HL-525 cells (23,24), the expression of the other PKC isoenzymes showed the same pattern between HL-60 and HL-525 cells. Both cell lines highly express PKC-␣, -, and -and rack-1, weakly express PKC-⑀, -, and -␦, and express little or no PKC-(data not shown). Therefore, we examined the possibility that PKC-␤ isozyme is involved in TNF-␣-induced apoptosis. We isolated stable HL-525 clones transfected with an expression plasmid containing the full-length PKC-␤ cDNA and neo, the gene that confers Geneticin resistance. As a control, we isolated HL-525 clones transfected with a plasmid containing neo only. Stable transfectants were then screened for their expression of PKC-␤ mRNA by Northern blotting. Our results indicate that PKC-␤ gene expression was restored in the HL-525 cells transfected with the PKC-␤ expression plasmid, whereas the control cells remained deficient (Fig. 3A).
Because the PKC-␤ transfectants are unstable and lose their functional PKC-␤ within 2 months (23), these clones were used in the experiments prior to a major drift in this phenotype. The transfectants with functional PKC-␤ regained susceptibility to TNF-␣-induced apoptosis (Fig. 3B). No restoration of TNF-␣induced apoptosis was observed in HL-525 cells transfected with the control vector (Fig. 3B). On the other hand, restoration of PKC-␤ gene expression in HL-525 cells reduced the susceptibility of these cells to anti-Fas mAb-induced apoptosis (Fig.  3B) and diminished Fas immunostaining from 100% to about 40% of the cell population. Taken together, these results implicate PKC-␤ in the signal transduction pathway leading to TNF-␣-mediated apoptosis in the HL-60 cell system; the inabil- ity of TNF-␣ to induce apoptosis in HL-525 cells seems to be due to an insufficient level of this PKC isoenzyme. The inability of anti-Fas mAb to induce apoptosis in HL-60 cells may be attributed to a lack of the Fas receptor, which seems to be negatively regulated by PKC-␤.
To demonstrate that PKC activity, rather than just the presence of the introduced PKC-␤ protein, was necessary for TNF-␣ activity, we employed inhibitors of PKC and PKA/PKG. Pretreatment of HL-60 cells with 25 M H7, a specific inhibitor of PKCs, repressed the ability of 2000 units/ml TNF-␣ to induce apoptosis by 87%, whereas 100 M HA1004, an inhibitor of PKA and PKG enzyme activities, was without significant effect. Therefore, the mere presence of a PKC protein is not sufficient; PKC must exert activity as well to transduce the apoptosis-inducing properties of TNF-␣ in HL-60 cells.
Involvement of Ceramides and Reactive Oxygen Intermediates in TNF-␣-and Fas-induced Apoptosis-Because ceramides and ROIs were reported to be involved in various pathways leading to apoptosis (3, 10, 11, 36 -39), it was of interest to investigate whether they are also involved in TNF-␣-and Fasmediated apoptosis in HL-60 and HL-525 cells, respectively. To delineate the specific mediators involved in these pathways, we incubated these cells with 0.5 M fumonisin, an inhibitor of ceramide generation (36,37), or with either 1 mM N-acetylcysteine or 10 mM glutathione, scavengers of ROIs (38,39), 20 min before and during TNF-␣ and anti-Fas mAb treatment. At these doses, the inhibitors themselves did not affect the viability of untreated HL-60 or IgG-treated HL-525 cells (Fig. 4A).
Our results indicated that fumonisin effectively inhibited TNF-␣-mediated apoptosis in HL-60 cells but not anti-Fas mAbmediated apoptosis in HL-525 cells (Fig. 4A). On the other hand, the antioxidants N-acetylcysteine and gluthatione effectively inhibited apoptosis induction by the anti-Fas mAb in HL-525 cells and to a limited degree by TNF-␣ in HL-60 cells (Fig. 4A).
These results implicate ceramides as mediators of TNF-␣induced apoptosis in HL-60 cells and ROIs in anti-Fas mAbinduced apoptosis. It was, therefore, of interest to determine whether supplying ceramides exogenously to HL-525 cells would bypass the PKC-␤ deficiency and would induce apoptosis. In a similar fashion, it was of interest to determine whether exogenous H 2 O 2 , which generates intercellular ROIs, would induce apoptosis in HL-60 cells. For this reason, we cultured HL-60 and HL-525 cells with 10 M of either cell permeable c2-ceramide or D-erythrosphingosine, a ceramide precursor (37), or with 10 M H 2 O 2 . In both HL-60 and HL-525 cells, treatment with these agents effectively induced the cells to undergo apoptosis (Fig. 4B). Moreover, the fraction of the cells showing apoptotic changes after treatment with these inducers varied only slightly between the two cell types (Fig. 4B). Taken together, these results indicate that ceramides appear to be mediators of TNF-␣-induced apoptosis in the HL-60 cell system and that activation of PKC-␤ in this process precedes the generation of ceramides. The results also implicate ROIs as mediators of anti-Fas mAb-induced apoptosis in HL-525 cells and as contributing factors in TNF-␣-induced apoptosis.
Effect of Caspase Inhibitors on TNF-␣-and Anti-Fas mAbinduced Apoptosis-Caspases appear to be essential for the execution of apoptosis because they cleave critical cellular substrates (16). Different studies have reported that caspase-1, which cleaves YVAD-type substrates, and/or caspase-3, which cleaves DEVD-type substrates, are involved in Fas-and TNF-␣-induced apoptosis (17-19, 40, 41). It was therefore of interest to determine whether caspase-1 and -3 are also involved in apoptosis induction in HL-60 and HL-525 cells by these inducers. A way to address this question is through the use of the available irreversible caspase inhibitors, AcYVAD-BoMK and z-DEVD-FMK, which are derivatives of the corresponding substrates (16,17,41). However, a recent study of inhibitor specificity found that Ac-YVAD-BoMK inhibits both caspase-1 and -4, whereas z-DEVD-FMK inhibits both caspase-3 and -7 (16).
Prior to the use of these inhibitors, we determined whether caspase-1, -3, -4, and -7 were present in HL-60 and HL-525 cells using fluorogenic substrates (Ac-YVAD-pNA for caspase-1 and -4 and Ac-DEVD-pNA for caspase-3 and -7). The results indicated that both cell types exhibit a similar specific activity of 30 Ϯ 10 nmol of pNA/g of protein/h for caspase-1 and/or -4 and 20 Ϯ 5 nmol of pNA/g of protein/h for caspase -3 and/or -7.
To investigate the involvement of caspase-1 and/or -4 in TNF-␣-induced apoptosis and caspase-3 and/or -7 in anti-Fas mAb-induced apoptosis in HL-60 and HL-525 cells, we incubated these cells with 10 -100 nM of the caspase inhibitors Ac-YVAD-BoMK and z-DEVD-FMK, respectively, 20 min before and during the treatment with the inducers. The caspase-1 and -4 inhibitor Ac-YVAD-BoMK abated the TNF-␣-mediated apoptosis in HL-60 cells and had little effect on anti-Fas mAbinduced apoptosis in HL-525 cells (Fig. 5A). In contrast, the caspase-3 and -7 inhibitor z-DEVD-FMK abated the apoptotic effect of anti-Fas mAb in HL-525 cells but had a limited effect on TNF-␣-mediated apoptosis in HL-60 cells (Fig. 5B). Moreover, the effect of these inhibitors was dose-dependent (Fig. 5,  A and B). The inhibitors by themselves at these dosages did not affect the cell viability of untreated HL-60 or HL-525 cells or IgG-treated HL-525 cells (Fig. 5, A and B). In another experiment, we incubated PKC-␤-transfected HL-525 cells with 100 nM of the caspase inhibitors 20 min before and during treatment with TNF-␣. At this dosage, the inhibitors did not affect the viability of untreated transfected cells. Similar to results found in HL-60 cells (Fig. 5A), the caspase-1 and -4 inhibitor Ac-YVAD-BoMK, but not the caspase-3 and -7 inhibitor, effectively blocked TNF-␣-induced apoptosis in the PKC-␤-transfected HL-525 cell (Fig. 5C). Thus, restoration of PKC-␤ gene expression to HL-525 cells in and of itself restores both susceptibility to TNF-␣ and utilization of the same mediators as HL-60 cells.
We also tested the effect of these inhibitors at 100 nM on ceramide-and H 2 O 2 -induced apoptosis. The results indicated that Ac-YVAD-BoMK but not z-DEVD-FMK reduced ceramideinduced apoptosis by about 80% in both HL-60 and HL-525 cells, whereas z-DEVD-FMK but not Ac-YVAD-BoMK reduced by more than 50% H 2 O 2 -induced apoptosis in both cell types. These results implicate caspase-1 and/or -4 in apoptosis induction by TNF-␣ and its mediator, ceramide, in HL-60 cells and caspase-3 and/or -7 in apoptosis induction by anti-Fas mAb and its mediator, ROIs. DISCUSSION PKC is known to modulate apoptosis; depending upon the cell system and the apoptosis inducer, one can conclude that the effect is either permissive (20,27) or antagonistic (1,25,26). The explanation for these conflicting results most likely lies with the facts that (a) PKC constitutes a family of at least 12 different isoenzymes, each with its own characteristic and cellular function (42), and different cell types often have distinct combinations of certain of these isoenzymes, and (b) the commonly used PKC activators or inhibitors do not exhibit isoenzyme specificity (43). Thus, the permissive or antagonistic action of individual PKC isoenzymes in apoptosis cannot be determined from experiments using such agents.
Here, by using cell types that are either proficient or deficient in a specific PKC isoenzyme, we were able to provide evidence that in the HL-60 cell system the PKC-␤ isoenzyme is a mediator of TNF-␣-induced apoptosis. In contrast, there appears to be a negative correlation between PKC-␤ expression and manifestation of the Fas receptor and apoptosis induction by the anti-Fas mAb.
These conclusions were derived from our studies with the HL-60 cell line and its PKC-␤-deficient variant, HL-525 (23,24). In these two cell types, the two common inducers of apoptosis, TNF-␣ and anti-Fas mAb (1, 2), exhibited a divergent apoptotic effect: TNF-␣ effectively induced apoptosis in HL-60 but not in HL-525 cells, whereas anti-Fas mAb exhibited the inverse effect. This finding initially raised the possibility that the inability of these inducers to evoke apoptosis in the corresponding cell types was due to lack of their respective receptors, TNF-RI, TNF-RII, or Fas (15,(31)(32)(33)(34)(35). Using appropriate mAbs, we found that this was not the case for TNF-␣, because both the HL-60 and the HL-525 cells exhibited similar levels of the TNF-RI and the more weakly expressed TNF-RII proteins. However, such a receptor difference between cell lines may account for the resistance of HL-60 cells to anti-Fas mAbinduced apoptosis, as HL-60 cells express undetectable levels of the Fas receptor. With regards to which of the two TNF receptors is involved in apoptosis induction by TNF-␣, it appears that the death domain-containing TNF-RI mediates this induction in HL-60 cells, as mAb against this receptor blocked apoptosis, whereas mAbs against the TNF-RII were ineffective in neutralizing the apoptotic effect of TNF-␣. The role of TNF-RII in the TNF-␣-induced apoptosis pathway has not been successfully demonstrated as yet, although there are some reports suggesting that TNF-RII mediates signaling events, which could play a similar role as in the TNRI pathway, including phosphatidylinositol-generated activation of MAP kinases, c-Jun N-terminal kinase, and NFB (31,35,43,44).
Because absence of the TNF-RI or TNF-RII could not account for the resistance of HL-525 cells to TNF-␣-induced apoptosis, we tested for the possibility that the defect is in a signal transduction step involving PKC-␤, as we have previously found with relationship to macrophage differentiation (23, 24). Our results indicated that this was the indeed the case, as restoration of PKC-␤ in the HL-525 cells, by transfection with an appropriate expression vector (23,24), reestablished the susceptibility of HL-525 cells to TNF-␣-induced apoptosis. Restoration also correlated to a down-regulation of the Fas receptor and to a decreased susceptibility to anti-Fas mAb-induced apoptosis, implying that PKC-␤ may act as a negative regulator of Fas mAb-induced apoptosis.
Taken together, the results indicate that a difference in a PKC isoenzyme can influence whether a cell type is permissive or antagonistic to apoptosis induction. This may explain to some degree the apparent discrepancies in previous studies involving PKC (1,20,(25)(26)(27). For instance, it was reported that PKC-down-regulation of TNF-␣-induced apoptosis correlates with enhanced expression of NF-B-dependent protective genes (45), PKC-influences the sensitivity to TNF-␣ (46), PKC-⑀ is required for the protective effect of phorbol 12-myristate 13-acetate in TNF-␣-induced apoptosis (47), and the cytosolic translocation of PKC-␦ and -⑀ plays an important role in ceramide-mediated apoptosis (48). In our cell system, differences in the levels of these other PKC isoforms most likely do not account for the contrasting response of the HL-60 and HL-525 cells to TNF-␣, as we have determined that these isoforms display a similar pattern of expression in these two cell types. Therefore, it appears most likely that PKC-␤ or some other kinase that is directly influenced by PKC-␤ expression is responsible for mediating TNF-␣-induced apoptosis in HL-60 cells.
It has been shown previously that ceramide and ROIs are mediators of apoptosis (3,10,11,(37)(38)(39)(40). It was therefore of interest to determine whether this observation applies to apoptosis induced by TNF-␣ in HL-60 cells and by anti-Fas mAb in HL-525 cells. To test for this possibility, we included in our studies an inhibitor of ceramide synthesis, fumonesin (37,38), and the antioxidants N-acetylcysteine and glutathione (39,40), which are scavengers of ROIs. We found that inhibition of ceramide generation abolished the apoptotic effect of TNF-␣ in HL-60 cells but had no apparent effect on anti-Fas mAb-induced apoptosis in HL-525 cells. In contrast, the antioxidants inhibited the apoptotic effect of anti-Fas mAb in HL-525 cells but had a limited effect on TNF-␣-induced apoptosis in HL-60 cells. These results implicate ceramides as major mediators of TNF-␣-induced apoptosis in HL-60 cells and ROIs in anti-Fas-MA-induced apoptosis in HL-525 cells (Fig. 6).
Our conclusions are compatible with previous observations made in myeloid cells, including HL-60 cells, showing that ceramides and ROIs are involved in the apoptosis process mediated by TNF-␣ (1,49,50) and that ROIs are important mediators of the Fas pathway (9,10). However, our results argue against the involvement of ceramides in Fas-mediated apoptosis in the HL-60 cell system (51,52). These conflicting results could be explained by the fact that the various processes that occur during apoptosis, evoked by an inducer, vary with the cell type (18,(53)(54)(55). It is not clear which of these processes occur in all or most cell types and which occur only in some. It is not clear how much overlap there is between pathways and to what degree disparate inducers share common mediators to achieve the same outcome, apoptosis.
We showed that PKC-␤ and ceramide are two signal transduction components in the TNF-␣ pathway. Next, we were interested in determining the temporal relationship of these mediators in this process. To delineate the role of PKC-␤ and ceramide in the TNF-pathway, we used PKC-␤-deficient HL-525 cells. We assumed that if we supplemented these cells with exogenous ceramides, the ceramides would either (a) induce the cells to undergo apoptosis, suggesting that PKC-␤ is upstream of ceramide, or (b) not have an effect, suggesting that PKC-␤ is downstream of ceramide. Our study showed that the ceramides were indeed able to induce apoptosis in HL-525 cells, indicating that, in the sequential events leading to TNF-␣induced apoptosis, PKC-␤ is upstream of the ceramides (Fig. 6). In this context, ceramide-induced apoptosis has been shown to be accompanied by up-regulation of c-jun gene transcription and DNA binding activity of the transcription factor AP-1 (56), inhibition of c-myc and BCL-2 expression (57), and activation of a cowpox virus protein (CrmA)-insensitive protease (58), all events that, using our evidence, would occur after PKC-␤ activity.
Caspases, which are related to the product of the C. elegans death gene Ced-3 (16), are considered to be one of the primary executioners of apoptosis (17). Most attention has focused on the interleukin 1-␤-converting enzyme-like protease/caspase-1, partly because of the progress made by Hengartner and Horvitz (59) and Kuida et al. (60). Other reports suggest that this protease is not the only one involved in apoptosis (17, 41, 61). Caspase-3 is considered as another key executioner of apoptosis, being responsible either partially or entirely for the proteolytic FIG. 6. Scheme of TNF-␣ and Fasinduced apoptosis signaling in HL-60 and HL-525 cells. TNF-␣ after interaction with TNF-RI induces the activation of PKC-␤, then the generation of ceramide (and to a lesser degree the production of ROI), and in turn the activation of caspase-1 and/or -4, to mediate apoptosis. On the other hand, anti-Fas mAb-induced apoptosis does not require PKC-␤. Anti-Fas mAb apoptosis utilizes an ROI-dependent pathway in which ROI production is followed by the activation of caspase-3 and/or -7. The inhibitory effect of PKC-␤ or Fas manifestation and the suppression effect of fumonisin (an inhibitor of ceramide production), antioxidants, such as N-acetylcysteine and glutathione (ROI scavengers), and Ac-YVAD-BoMK or z-DEVD-FMK (caspase inhibitors) on apoptosis induction are represented by vertical bars.
cleavage of many key proteins, such as the nuclear enzyme poly-(ADP-ribose) polymerase (62). A number of reports have implicated caspase-1 and/or caspase-3 in Fas-and TNF-␣-induced apoptosis in different cell types (18,19,41,42). In our experiments, the presence of an irreversible caspase-1 and -4 inhibitor resulted in a blocked TNF-␣-induced apoptosis in HL-60 cells but had a limited effect on anti-Fas mAb-induced apoptosis in HL-525 cells. On the other hand, the presence of an irreversible caspase-3 and -7 inhibitor abrogated the apoptotic effect of anti-Fas mAb in HL-525 cells but had little effect on TNF-␣-induced apoptosis in HL-60 cells. These results implicate caspase-1 and/or -4 in the TNF-␣-induced apoptotic pathway and caspase-3 and/or -7 in the anti-Fas mAb-induced apoptotic pathway (Fig. 6). In some reports, caspases were suggested to be involved in the apoptotic process either upstream (50,51) or downstream (63) of the ceramides. Based on our results, caspase-1 and/or -4 seems most likely to be involved downstream of the ceramides because the caspase-1 and -4 inhibitor abrogated the ceramide-induced apoptosis. A similar situation can be invoked for ROIs because the caspase-3 and -7 inhibitor reduced the apoptotic effect of H 2 O 2 , a generator of ROIs.
Apoptosis involving TNF-RI is believed to be mediated by different signaling pathways, depending upon which adaptor protein is recruited to the cytoplasmic domain of the receptor. For instance, recruitment of the signal transducer FADD involves activation of specific caspases, whereas two other signal transducers, RIP and TRAF-2, involve the mitogen-activated protein kinase/extracellular signal-regulated kinase kinase-1 and G-CkR (TNF-responsive serine/threonine related kinase) cascade (64 -68). Based on our results and these reports, we can suggest the following scheme: binding of TNF-␣ to TNF-RI induces receptor oligomerization (34), followed by the recruitment of adaptor proteins (such as FADD) (34) to the death domain and the activation of the phosphatidylcholine-specific phosopholipase C (27,69,(71)(72)(73). This enzyme hydrolyses inositol phospholipids to generate 1Ј-2Ј diacylglycerol (27,69,(71)(72)(73). The binding of diacylglycerol and calcium to the regulatory domain of PKC-␤ leads to allosteric activation of this kinase (27,73,74); followed by intramolecular autophosphorylation (14). Activated PKC-␤ is then translocated to different subcellular sites to phosphorylate and activate other proteins (70). At this point, we can speculate that PKC-␤ would activate the endosomal acidic sphingomyelinase, which mediates the hydrolysis of membrane sphingomyelin to ceramide (69). The generated ceramide evokes the recruitment and triggering of specific caspases, perhaps through the involvement of an appropriate adaptor protein. It is, however, still not well known how such an interaction may initiate the caspase pathway.
Conclusions-The present study identifies and delineates signaling factors involved in TNF-␣and anti-Fas mAb-induced apoptosis in the HL-60 cell system (Fig. 6). We suggest that PKC-␤, ceramide, and caspase-1 and/or -4 are key mediators of the TNF-␣-induced apoptotic pathway. In this pathway, TNF-␣ interacts with TNF-RI, causing the activation of PKC-␤, which is followed by the generation of ceramides and, in turn, the activation of caspase-1 and/or -4. We suggest that ROIs may also contribute, to a limited degree, to TNF-␣-induced apoptosis (Fig. 6). On the other hand, we conclude that anti-Fas mAb-induced apoptosis does not require PKC-␤; actually, PKC-␤ may perhaps negatively regulate this event, most likely by down-modulating the Fas receptor. The anti-Fas mAb, after interaction with its receptor, utilizes an ROI-and caspase-3 and/or -7-dependent pathway (Fig. 6).