Serine palmitoyltransferase regulates de novo ceramide generation during etoposide-induced apoptosis.

The de novo pathway of sphingolipid synthesis has been identified recently as a novel means of generating ceramide during apoptosis. Furthermore, it has been suggested that the activation of dihydroceramide synthase is responsible for increased ceramide production through this pathway. In this study, accumulation of ceramide mass in Molt-4 human leukemia cells by the chemotherapy agent etoposide was found to occur primarily due to activation of the de novo pathway. However, when the cells were labeled with a substrate for dihydroceramide synthase in the presence of etoposide, there was no corresponding increase in labeled ceramide. Further investigation using a labeled substrate for serine palmitoyltransferase, the rate-limiting enzyme in the pathway, resulted in an accumulation of label in ceramide upon etoposide treatment. This result suggests that the activation of serine palmitoyltransferase is the event responsible for increased ceramide generation during de novo synthesis initiated by etoposide. Importantly, the ceramide generated from de novo synthesis appears to have a distinct function from that induced by sphingomyelinase action in that it is not involved in caspase-induced poly (ADP-ribose)polymerase proteolysis but does play a role in disrupting membrane integrity in this model system. These results implicate serine palmitoyltransferase as the enzyme controlling de novo ceramide synthesis during apoptosis and begin to define a unique function of ceramide generated from this pathway.

It is increasingly apparent that sphingolipids, and in particular ceramide, are important mediators in regulating the response to stress of a cell. The agents that induce ceramide generation include physiological factors, such as tumor necrosis factor and the Fas ligand, as well as therapeutic agents, such as chemotherapy drugs and radiation. Many of these agents induce ceramide generation via the hydrolysis of sphingomyelin by the activation of one or more sphingomyelinases. Additional studies, however, have begun to implicate ceramide generated from the de novo pathway of sphingolipid synthesis as having a signaling function (1)(2)(3)(4)(5)(6)(7)(8).
Studies of de novo sphingolipid biosynthesis have been advanced by the realization that a class of fungal metabolites known as fumonisins share structural similarities with the sphingoid backbone. During investigation of the effects of fumonisin on sphingolipid metabolism in hepatocytes, it was observed that the synthesis of complex sphingolipids was significantly inhibited. It was also determined that the primary site of action of fumonisin was dihydroceramide synthase (9), an enzyme in the de novo pathway that catalyzes the N-acylation of sphinganine to produce dihydroceramide.
Because ceramide has been implicated as a regulatory molecule in apoptosis, more recent studies have used fumonisins to investigate the potential role of ceramide from the de novo pathway in this process. These studies have demonstrated that fumonisin is able to attenuate apoptosis induced by daunorubicin, camptothecin, tumor necrosis factor-␣, and phorbol ester (1,3,5,6). Because fumonisin inhibits dihydroceramide synthase, it generally has been assumed that this is the regulatory step in the de novo pathway during apoptosis.
In this study, we have used the chemotherapy agent etoposide to activate de novo ceramide synthesis in Molt-4 human leukemia cells. In preliminary experiments, we were unable to find any evidence for activation of dihydroceramide synthase, and we became interested in determining the regulatory point in de novo ceramide synthesis under apoptotic conditions with etoposide. Using intact cell radiolabeling techniques and cellfree enzyme assays, we determined that serine palmitoyltransferase, the initial and rate-limiting enzyme in the pathway, is activated during apoptosis and governs the production of ceramide.
We were also interested in elucidating a regulatory function for ceramide generated de novo in apoptosis, and our present studies demonstrate that it has its predominant effects on membrane-related events in apoptosis. Importantly, unlike short-chain ceramide or ceramide generated from sphingomyelinase action, this ceramide is dissociated from caspase activation. This study provides the first evidence that serine palmitoyltransferase is a regulated enzyme during apoptosis and provides further evidence that ceramide generated de novo functions as a regulatory molecule in mediating membranerelated apoptotic events.
[ 3 H]Sphinganine was synthesized as described previously (10). [9, H]palmitoyl CoA (60 Ci/mmol) and L-[ 3 H]serine (20 Ci/mmol) were purchased from American Radiolabeled Chemicals (St. Louis, MO). A rabbit polyclonal antibody raised to an epitope in the automodification domain of human poly(ADP-ribose) polymerase (PARP) 1 was a gift of * The costs of publication of this article were defrayed in part by the payment of page charges. This article must therefore be hereby marked "advertisement" in accordance with 18 U.S.C. Section 1734 solely to indicate this fact.
ʈ To whom correspondence should be addressed: Dept. of Biochemistry and Molecular and Biology, Medical University of South Carolina, 173 Ashley Ave., Charleston, SC 29425. Fax: 843-953-0843.
Cell Culture-Molt-4 cells were obtained from ATCC (Manassas, VA) and maintained at 37°C and 5% CO 2 in RPMI medium with 10% fetal bovine serum. Cells were subcultured prior to reaching a density of 2 ϫ 10 6 /ml.
Ceramide Mass Measurements-Molt-4 cells were seeded at 5 ϫ 10 5 /ml (3 mls/well in a 6-well plate). After treatments, ceramide mass was determined using the diglyceride kinase assay as described previously (11). Briefly, 5 ϫ 10 5 cells were lysed in a mixture of chloroform and methanol (1:2), and lipids were extracted according to Bligh and Dyer (12). Lipids were solubilized in dioleoylphosphatidylglycerol/␤octylglucoside mixed micelles for 30 min at 37°C. Three g of membranes from E. coli strain N4830/pJW10 overexpressing diglyceride kinase were added, and the reaction was initiated by the addition of 3 Ci of [␥-32 P]ATP (3000 Ci/mmol) in the presence of 1 mM carrier ATP. After 30 min at room temperature, the reaction was quenched with chloroform/methanol (1:2) and the lipids extracted by the Bligh-Dyer method. After evaporation of the chloroform phase under N 2 , the lipid pellet was suspended in 40 l of chloroform/methanol (4:1) and spotted on silica gel 60 TLC plates. The lipids were chromatographed in a solvent system of chloroform/acetone/methanol/acetic acid/water (10:4: 3:2:1). After autoradiography, the location of ceramide phosphate was determined based upon the R F (0.48) of phosphorylated ceramide standards. Radioactivity was quantified by liquid scintillation spectrometry. Ceramide mass was then determined using the slope of the standard curve and normalized to lipid phosphate.
Lipid Phosphate Determination-Lipid phosphate was measured by acid hydrolysis of a Bligh-Dyer extract from 10 6 Molt-4 cells. The chloroform phase of the extract was evaporated under N 2 , and the lipids were incubated at 150°C for 6 h in 0.6 ml of 10 N H 2 SO 4 /70% HClO 4 / H 2 O (9:1:40). After acid hydrolysis, 0.6 ml of H 2 O, 0.5 ml of 0.9% ammonium molybdate, and 0.2 ml of 9.0% ascorbic acid were added and incubated for 30 min at 45°C. Inorganic phosphate was detected by absorbance at 820 nm and quantified based upon a standard curve of K 2 HPO 4 .
Radiolabeling of Cells-Molt-4 cells were suspended at 5 ϫ 10 5 /ml in RPMI medium with 2% fetal calf serum. Five ml of this suspension were added to each flask, and cells were labeled with either 0.1 Ci of [ 3 H]sphinganine (0.045 Ci/nmol) or 5 Ci of [ 3 H]palmitate (43 Ci/ mmol). Concurrent with the addition of radionuclide, etoposide and inhibitors of de novo synthesis were added. After 6 h, cells were harvested by centrifugation at 4°C. Lipids were then extracted according to Bligh and Dyer (12). Lipid extracts from cells labeled with [ 3 H]sphinganine were immediately run on TLC. Lipids extracted from [ 3 H]palmitate-labeled cells were first submitted to mild alkaline hydrolysis prior to separation by TLC. The solvent system employed was chloroform/methanol/2 N NH 4 OH, 4:1:0.1. After chromatography, a ceramide standard (R F ϭ 0.77) was identified by the use of iodine vapors. The TLC plates were then sprayed lightly with En 3 hance, and radioactivity was visualized by autoradiography after 48 h at -80°C. The radioactive spot migrating with the same R F as the ceramide standard was scraped from the plate and quantified by liquid scintillation spectrometry.
RT-PCR of Human Serine Palmitoyltransferase Subunits-Total cellular RNA was isolated from 2.5 ϫ 10 6 human Molt-4 cells at each time point that had been treated with 10 M etoposide or left untreated at 37°C. RNA was subjected to RT-PCR on a Roche Molecular Biochemicals LightCycler as follows: a 20-l reaction containing 200 ng of RNA, 0.25 M each primer, 5 mM MgCl 2 , 1ϫ LightCycler RT-PCR reaction mixture SYBR Green, and 1ϫ LightCycler RT-PCR enzyme mixture was amplified in a LightCycler capillary. The sample was reverse transcribed at 55°C for 10 min and heated to 95°C for 30 s. The resulting cDNA was amplified for 45 cycles by denaturation at 95°C for 1 s, primer annealing at 55°C for 10 s, and elongation at 72°C for 20 s. One fluorescence reading was taken in each cycle following the elongation step. The primers used were as follows: human b-actin, 5Ј-TCC-TGTGGCATCCACGAAACT-3Ј and 5Ј-GAAGCATTTGCGGTGGACGA-T-3Ј; human serine palmitoyltransferase subunit 1 (hLCB1), 5Ј-AGGA-GTCACTGAACACTATG-3Ј and 5Ј-AGCTCTCTCCAGTTCTTCCT-3Ј; and human serine palmitoyltransferase subunit 2 (hLCB2), 5Ј-GTGG-ATGTTATGATGGGAACG-3Ј and 5Ј-CATACGTCGTCTCGTCAAAG-3Ј. The GenBank TM accession numbers for these genes are as follows: b-actin, X00351; hLCB1, Y08685; and hLCB2, Y08686.
Enzyme Assays-Serine palmitoyltransferase was assayed as described previously (13). Briefly, enzyme activity in 100 g of microsomal membranes was determined in 100 mM Hepes (pH 8.3), 5 mM dithiothreitol, 2.5 mM EDTA (pH 7.4), and 50 M pyridoxal 5Ј-phosphate. The reaction was initiated by the addition of 200 M palmitoyl CoA and 2 Ci of L-[ 3 H]serine, with a final serine concentration of 1 mM. Reactions were incubated for 10 min at 37°C prior to termination with 0.2 ml of 0.5 N NaOH. Organic soluble counts were extracted as described previously and quantified by liquid scintillation counting (13).
Dihydroceramide synthase was assayed according to Bose et al. (1). Briefly, the dihydroceramide synthase activity in 50 g of microsomal membranes was determined in 20 mM Hepes (pH 7.4), 2 mM MgCl 2 , 20 M fatty acid-free bovine serum albumin, and 20 M sphinganine. The reaction was initiated by the addition of 70 M palmitoyl CoA containing 0.5 Ci of [ 3 H]palmitoyl CoA and allowed to incubate for 30 min at 37°C. Lipids were extracted by the method of Bligh and Dyer (12) and resolved by thin layer chromatography in a solvent system of chloroform/methanol/2 N NH 4 OH (4:1.0:0.1). Radioactivity comigrating with a ceramide standard was scraped from the plate and quantified by liquid scintillation counting.
Western Blotting-PARP proteolysis was determined as described previously (14). Approximately 5 ϫ 10 5 cells were lysed by boiling in Laemmli buffer, and the lysate was electrophoresed on a 6% polyacrylamide gel. After transferring proteins to nitrocellulose, the blot was incubated for 1 h with rabbit anti-PARP (1:2000), and detection was performed by enhanced chemilumenescence using peroxidase-conjugated goat anti-rabbit (1:5000).
Cell Viability Assays-Molt-4 cells were suspended at 5 ϫ 10 5 /ml in RPMI medium with 2% fetal calf serum. Two ml of the cell suspension were aliquoted into the wells of a 12-well plate. Cells were treated simultaneously with etoposide and fumonisin. After 9 or 24 h, 0.5 ml of cells was removed from the well, harvested by centrifugation, and suspended in 0.2 ml of phosphate-buffered saline. Following the addition of 0.2 ml of 0.4% trypan blue solution, cells were examined with a light microscope for trypan blue staining.

Induction of de Novo Synthesis of Ceramide by Etoposide-
The chemotherapy agent etoposide initiates an apoptotic response by inhibiting topoisomerase II, resulting in singlestranded DNA breaks (15). Through unknown mechanisms, this insult by etoposide results in the release of cytochrome c from the mitochondria (16), activation of caspases (17), DNA fragmentation (18), and subsequent cell death (reviewed in Ref. 19). Because ceramide has been shown to be an inducer of these events (20 -22), we were interested in determining whether etoposide elevated ceramide levels and, if so, in determining the origin of ceramide. As demonstrated in Fig. 1, treatment with etoposide resulted in a nearly 3-fold elevation of ceramide after 6 h. In the presence of fumonisin, an inhibitor of de novo ceramide synthesis, etoposide-induced ceramide accumulation was decreased by nearly 75% after 6 h.
Radiolabeling of Cells with a Serine Palmitoyltransferase Substrate but Not a Dihydroceramide Synthase Substrate Results in Increased Incorporation of Radiolabel into Ceramide upon Etoposide Treatment-To determine whether induction of de novo ceramide generation was occurring by activation of dihydroceramide synthase as previously reported using daunorubicin (1), cells were radiolabeled with [ 3 H]sphinganine, the substrate for this enzyme (Fig. 2), and simultaneously treated with 10 M etoposide. Using this method and measuring [ 3 H]ceramide over a 9-h time course with etoposide, no increased incorporation of label into ceramide was observed (Fig.  3A). This result suggests that neither dihydroceramide synthase nor the subsequent enzyme in de novo synthesis, dihydroceramide desaturase, is responsible for increased synthesis of ceramide.
In a similar experiment using a 6 h time period of etoposide treatment in the absence or presence of fumonisin, the incorporation of [ 3 H]sphinganine into [ 3 H]ceramide was significantly inhibited (Fig. 3B), consistent with the reported inhibition of dihydroceramide synthase by fumonisin (23).
The lack of an increase in de novo ceramide synthesis using a dihydroceramide synthase substrate indicated a prior step in the pathway as the site of activation. Cells were therefore radiolabeled with [ 3 H]palmitate, a precursor to the serine palmitoyltransferase substrate palmitoyl CoA, and simultaneously treated with etoposide. An increase of over 3-fold in [ 3 H]ceramide was observed by 9 h (Fig. 4A), suggesting that serine palmitoyltransferase, the rate-limiting enzyme in de novo synthesis of sphingolipids, is activated during de novo synthesis of ceramide under apoptotic stress. Moreover, both fumonisin and cycloserine, an inhibitor of serine palmitoyltransferase (24), inhibited the incorporation of [ 3 H]palmitate into [ 3 H]ceramide during a 6 h time period of etoposide treatment (Fig. 4B).
Regulation of Serine Palmitoyltransferase Activity-Previously, it had been demonstrated that during irradiation of keratinocytes, RNA levels of LCB2, a serine palmitoyltransferase subunit, were up-regulated (25). Therefore, in order to understand the regulation of serine palmitoyltransferase activity during etoposide-induced apoptosis, we initially determined mRNA levels of the LCB1 and LCB2 subunits of serine palmitoyltransferase by RT-PCR. No increases in the message level of either subunit were observed. In contrast, the serine palmitoyltransferase RNA was degraded by 6 h of etoposide treatment (Table I). Because up-regulation of RNA was not responsible for increased enzyme activity, in vitro enzyme assays for serine palmitoyltransferase were conducted using microsomes from either control-or etoposide-treated cells. Over a 6-h course of treatment, serine palmitoyltransferase activity was elevated early (by 0.5 h) and sustained throughout the treatment (Fig. 5). These results rule out etoposide-induced activation of serine palmitoyltransferase by up-regulation of RNA and suggest activation by covalent modification or allosteric regulation of the enzyme.
Dissociation of de Novo Synthesized Ceramide from Caspase Activation-Because we and others have previously demonstrated that short-chain ceramide induces PARP proteolysis by caspase activation (21,26) and because etoposide had previously been demonstrated to cause PARP proteolysis (18), we were interested in determining whether the de novo generation of ceramide by etoposide was also instrumental in this caspase-mediated event. In the presence of fumonisin, etoposide-induced de novo ceramide generation returned to basal levels yet PARP proteolysis was unaffected (Figs. 4B and 6). This result demonstrates that de novo ceramide generation is not necessary for the activation of a PARP-cleaving caspase(s) in this cell system. In addition, we have also examined the effect of fumonisin on lamin B cleavage by etoposide, reportedly occurring by caspase-6 (27), and found no inhibition (data not shown). Additional studies using the chromogenic caspase substrates YVAD-pNA, DEVD-pNA, and IETD-pNA were also conducted. These peptides function as substrates for group I, group II, and group III caspases, respectively (28). Whereas etoposide induced both DEVD-pNA and IETD-pNA peptidase activity, fumonisin was without an inhibitory effect (data not shown). These results are consistent with prior studies demonstrating etoposide activation of caspase-3 (of which DEVD-pNA is a substrate), and caspase-6 (of which IETD-pNA is a substrate) in human leukemia cells (29). Furthermore, they provide evidence for the dissociation of de novo generated ceramide from caspase activation and serve to distinguish its function from

Serine Palmitoyltransferase Regulation in Apoptosis
both ceramide generated from sphingomyelinases and from exogenous short-chain ceramides, both of which have been implicated in activation of caspases (21,26,30).
Effect of Fumonisin on Cell Death-Finally, we were interested in determining whether the inhibition of de novo generated ceramide by fumonisin protected cells from death. Because noticeable cell death did not occur during the first 6 h of treatment with etoposide, we assayed cell death at 9 or 24 h after treatment and observed nearly 15 and 75% cell death, respectively, as determined by trypan blue staining (Fig. 7). When the cells were treated during this period with either fumonisin or zVAD-fmk, a pancaspase inhibitor, cell death was approximately 7% after 9 h and 30% after 24 h. Interestingly, when fumonisin and zVAD-fmk were added together, near complete protection from cell death was observed. These results suggest that ceramide from de novo synthesis and caspases contribute to independent pathways of death during etoposide-induced apoptosis. DISCUSSION The results from this study provide the first evidence that the initial and rate-limiting enzyme in the de novo pathway of  sphingolipid biosynthesis, serine palmitoyltransferase, is activated during apoptosis. They also provide further support for a regulatory role in apoptosis of ceramide generated from the de novo pathway. Currently, very little is known about the function of serine palmitoyltransferase other than its role in sphingolipid synthesis for housekeeping functions, but studies in Saccharomyces cerevisiae have begun to define a role for the enzyme in stress response signaling. A yeast strain lacking serine palmitoyltransferase activity has been identified that is unable to grow unless supplemented with sphingoid bases (31). Suppressors of this mutation have also been isolated that are able to grow at ambient temperature but are unable to survive hyperosmolar or heat stress (32). The suppressor strains could be rescued, however, either by transfection with the serine palmitoyltransferase or by supplying exogenous sphingoid bases (33), thus implicating serine palmitoyltransferase in both heat and osmotic stress responses.
In mammalian cells, it has been demonstrated that sphingoid bases are capable of down-regulating serine palmitoyltransferase (34) and that the activity of the enzyme progressively increases during the differentiation process of neuronal cells in culture (35). Moreover, a recent report indicated that 48 h after UV irradiation of keratinocytes, mRNA levels for the LCB2 subunit of serine palmitoyltransferase were up-regulated 1.7-fold and that this corresponded to a 1.5-fold increase in enzyme activity (25).
Our results suggest that caution should be used in interpret-ing data obtained with the dihydroceramide synthase inhibitor, fumonisin. An inhibition of ceramide accumulation with fumonisin, although implicating de novo synthesis, does not imply that dihydroceramide synthase is the regulatory enzyme. Such a result would also be consistent with a regulatory role of any of the preceding enzymes in the pathway. In two recent studies of phorbol ester-and daunorubicin-induced apoptosis mediated by de novo ceramide synthesis, in vitro assays of dihydroceramide synthase indicated that the V max was increased 1.6 -1.7fold after treatment (1,5). However, in our intact cell [ 3 H]sphinganine labeling experiment using etoposide or using an in vitro enzyme assay (data not shown), we found no evidence that dihydroceramide synthase is activated during de novo synthesis in response to etoposide. The enzymes of the de novo pathway (Fig. 2) leading to the production of ceramide reside on the endoplasmic reticulum (36). In studies using hepatocytes, it was determined that the specific activity of serine palmitoyltransferase is considerably less than that of other enzymes in the pathway, including ketosphinganine reductase and dihydroceramide synthase. This fact, in conjunction with the observation of low sphingoid base levels in cells, has led to the conclusion that serine palmi- toyltransferase is the rate-limiting enzyme in the pathway of sphingolipid biosynthesis (13). Because the concentrations of reactants for a rate-limiting enzyme are generally much higher than the products, the activation of such an enzyme would be expected to shift the reaction toward equilibrium and result in a deregulation of the pathway. Under the experimental conditions in this study, activation of serine palmitoyltransferase would therefore have the consequence of leading to accumulation of an end product of the pathway.
Under normal conditions of de novo sphingolipid synthesis, ceramide is not typically considered an end product because it serves as a substrate for sphingomyelin synthase and glucosylceramide synthase, enzymes of complex sphingolipid synthesis. Whereas the enzymes of the de novo pathway (Fig. 2) leading to the production of ceramide reside on the endoplasmic reticulum (36), the sphingomyelin and glucosylceramide synthases are located on the Golgi apparatus and/or plasma membrane (37,38). Therefore, our data demonstrating the accumulation of ceramide from de novo synthesis indicate that either the transport of ceramide to these locations or the respective synthases may serve as additional regulatory points in this pathway during conditions of stress.
Additionally, if etoposide induced the accumulation of de novo ceramide by inhibiting one of the enzymes that use ceramide as a substrate, similar accumulations of ceramide should have been observed with both palmitate and sphinganine as substrates in the intact cell labeling experiments. The observation that elevated ceramide was seen only with palmitate as a substrate further supports the conclusion that ceramide accumulation is primarily due to activation of serine palmitoyltransferase.
We and others had previously shown that short-chain ceramide can induce PARP proteolysis and caspase activation (21,26). It has also been reported that ceramide produced by the action of sphingomyelinases is important for caspase activation (30). Our results demonstrating a lack of involvement of de novo synthesized ceramide in the activation of caspases suggest that this pool of ceramide may have a unique signaling or regulatory function. Because the de novo pathway synthesizes ceramide on the endoplasmic reticulum and because a signaling pool of ceramide has also been identified at the plasma membrane (39), it is plausible that these respective pools of ceramide employ different effector mechanisms.
These studies with etoposide do not exclude the possibility that other inducers of de novo synthesis and/or other cell types regulate additional enzymes in the de novo pathway. Also, the results do not exclude the possibility of additional functions for de novo generated ceramide in the apoptotic response. However, it is clear that care should be exercised in defining the main sites of biochemical regulation in the de novo pathway using a combination of enzymatic and labeling studies. Care should also be exercised in determining what specific aspects of apoptosis are regulated (or not regulated) by the de novo pathway.
The results from the trypan blue experiment suggest that de novo ceramide generated during apoptosis exerts a key regulatory function in effecting membrane damage. Furthermore, the additive and protective effect of fumonisin and a caspase inhibitor on cell death provide evidence that ceramide from de novo synthesis and caspases are activating independent pathways in apoptosis and suggest the model proposed in Fig. 8.
In summary, the results from this study begin to define serine palmitoyltransferase as an important regulatory step in the de novo pathway of ceramide synthesis during apoptosis. Moreover, they dissociate this pathway from a role in caspase activation and implicate it in mediating membrane-related events in apoptosis.