Ceramide inhibits phospholipase D in a cell-free system.

Recent evidence in whole cells has implicated ceramide in the regulation of phospholipase D (PLD). In intact HL-60 cells, phorbol myristate acetate (PMA) activated PLD as measured by [3H]palmitate-labeled phosphatidylcholine conversion to phosphatidylethanol in the presence of 2% ethanol. C6-Ceramide completely inhibited PLD activation after 4 h of treatment and was maximally active at 10 μM. The activity was structurally specific in that the structural analogs 4,5-dihydro-C6-ceramide and dioctanoylglycerol were inactive. Although ceramide inhibited PMA-induced activation of PLD, it did not inhibit translocation of protein kinase C (PKC) to the membrane in response to PMA. In a cell-free system, we confirmed that PLD is activated by guanosine 5′-O-(3-thiotriphosphate (GTPγS); however, ceramide had no effect on this activity under a variety of conditions. Activation of PLD by GTPγS was synergistically enhanced by the addition of PKC activators. This upstream effect was inhibited rapidly and specifically by ceramide (30 μM). Recombinant ARF plus PKCα substituted for crude cytosol in the activation of PLD, and this activity was inhibited by C6-ceramide. Taken together, these data show that ceramide interferes with PKC-mediated activation of PLD.

PLD 1 has been extensively investigated for the last several years. It has been implicated in the regulation of inflammatory and immune responses, cellular trafficking, and cell growth (1). This broad range of effects has complicated the elucidation of the mechanism of activation of PLD but has also led to the belief that there is a family of enzymes that probably possess distinct activation pathways (1). The best supported mechanisms of activation involve PKC or the small soluble GTPbinding proteins Rho and ARF (1)(2)(3)(4). We have previously demonstrated a role for PLD in fibroblast mitogenesis and identified that PLD is inhibited in senescent cells. The inhibition appeared to be mediated by the sphingolipid ceramide.
The role of sphingolipids in signaling mechanisms is now being revealed by careful study of their regulation and biological effects. The sphingomyelin cycle, with regulated conversion of sphingomyelin to ceramide (5,6), is now recognized as a key pathway in cell growth, differentiation, and apoptosis (7). We recently discovered a role for sphingolipids in cellular senescence (8,9). Sphingomyelinase activity and ceramide levels are elevated in senescent fibroblasts. Ceramide was able to induce a senescent phenotype, as measured by its ability to induce several parameters of senescence and to specifically inhibit PLD (8). Ceramide has also been demonstrated to inhibit PLD in neutrophils (10) and rat fibroblasts (11,12).
Here we attempt to develop our understanding of the mechanism of ceramide inhibition of PLD. We chose the HL-60 system, involving ARF-activated PLD (3,4,13), as a model for the study of PLD regulation. We demonstrate that ceramide inhibits PLD activation by PMA in HL-60 cells. Our results also confirm that PLD is activated by GTP-binding protein and show that this activation is augmented by PKC. Analysis of the effect of ceramide on PLD in a cell-free system revealed that it is this augmentation by PKC that is the target of ceramide inhibition.

EXPERIMENTAL PROCEDURES
Materials-Myristoylated recombinant ARF (5.7%) was a generous gift of Dr. Richard Kahn (Emory University). Recombinant PKC␣ was a generous gift of Dr. Doriano Fabbro (Ciba Geigy, Basel, Switzerland). Silica Gel 60 thin-layer chromatography plates were from Whatman, and solvents were supplied by Fisher Scientific. Dipalmitoylphosphatidyl(methyl-[ 3 H])choline and [ 3 H]palmitic acid was from DuPont NEN. Solvents were from Mallinckrodt Chemical Works (analytical grade). Fetal bovine serum and RPMI was from Life Technologies, Inc. Phosphatidylethanol and dioleoylglycerol were from Avanti Polar Lipids, Inc. ATP was from Pharmacia Biotech Inc. D-Erythro-N-hexanoylsphingosine and D-erythro-N-[ 14 C]hexanoyl-sphingosine were synthesized as described (14). Other reagents were purchased from Sigma.
Cell Culture-HL-60 cells (American Type Culture Collection) were grown in RPMI (Life Technologies, Inc.) supplemented with 5% NaHCO 3 and 10% fetal bovine serum. Cells were maintained at 37°C in 5% CO 2 and subcultured two times per week. Routine tests verified that the cells were free of Mycoplasma. C 6 -Ceramide Uptake and Metabolism-HL-60 cells were seeded at 2.0 ϫ 10 5 cells/ml. D-Erythro-N-[ 14 C]hexanoyl-sphingosine suspended in ethanol was added to a final concentration of 10 M (1 l/ml cells). Cells were harvested at various times, and radioactivity in the media and the cells was quantitated. Cells were then extracted (15), and the lipids were separated by thin layer chromatography (chloroform:methanol, 80: 20). Areas that showed radioactivity by autoradiography were scraped and quantitated by scintillation spectrometry.
PKC Translocation-HL-60 cells (2.5 ϫ 10 5 /ml in 50 ml) were treated with C 6 ceramide or ethanol (0.1%) for 4 h and then with 100 nM PMA for 20 min at 37°C. Cells were harvested as described (16) and analyzed by Western blot analysis. Protein was estimated by Bradford analysis (17).
Western Blots-Samples (100 g protein from HL60 cells) were run on 10% SDS-polyacrylamide gel electrophoresis and electrophoretically transferred to nitrocellulose at 4°C overnight (8). Blots were then washed with 5% nonfat dry milk in PBS for 1 h at 20°C and incubated with PKC isoenzyme-specific antisera (18) at a dilution of 1:500 at 1:1000 with or without competing peptide (20 -40 g) for 2 h at 20°C. The blots were washed three times with 5% nonfat dry milk in PBS for 15 min at 20°C and then once with PBS. The blots were incubated with secondary antiserum (goat anti-rabbit linked to horseradish peroxidase) for 2 h at 20°C and then washed three times with PBS. Blots were developed using ECL under conditions described by the manufacturer (Amersham Corp.).
Whole-cell PLD Assay-Cells (2.5 ϫ 10 5 in 1 ml) were seeded in 24-well plates. Cells were rested and radiolabeled with [ 3 H]palmitic acid (3 Ci, added in 3 l ethanol) in serum-free media for 24 h, which resulted in approximately 95% being incorporated into the phosphatidylcholine fraction. Cells were also pretreated with 10 M C 6 -ceramide for the indicated time. Cells were stimulated in 2% ethanol with 100 nM PMA for the indicated times. Incubations were stopped by the addition of 1 ml methanol on ice. Cells were transferred to a glass tube and extracted (15). Lipids were analyzed by thin layer chromatography using solvent A for phosphatidylcholine analysis, chloroform:methanol: acetic acid:H 2 O (50:25:8:4), or solvent B for phosphatidylethanol analysis, the upper phase of a mixture of ethyl acetate:iso-octane:acetic acid:H 2 O (80:50:20:100) (19).
Cell-free PLD Assay-Cells were treated as described (3). Briefly, HL-60 cells (5-8 ϫ 10 8 ) were pelleted and then washed in PBS. The cell pellet was suspended in approximately 2 ml of homogenization buffer and then nitrogen cavitated. The cavitate was centrifuged 40,000 rpm for 40 min in a TL-100.3 rotor (Beckman Instruments). The pellet was resuspended in homogenization buffer (2 ml) using a 19-gauge needle and was recentrifuged, as was the supernatant. This pellet was resuspended in 500 l. Proteins were stable for at least 2 weeks when diluted with an equal volume of 100% glycerol, frozen in a dry ice/ethanol bath, and stored at Ϫ90°C. Assays were performed essentially as described previously (3) except that 3.4 g of membrane protein and 7.1 g of cytosol protein were used, and the radiospecific activity of the substrate dipalmitoylphosphatidyl(methyl-[ 3 H])choline was increased such that at the same molar concentration, 120,000 dpm radiolabel were added per tube.

RESULTS
Previous studies from our laboratory and others demonstrated that ceramide inhibits PLD activation in fibroblasts (8,11,12) and neutrophils (10). To define the target for ceramide, we used a cell-free system from HL-60 cells to investigate the action of ceramide on PLD. In intact HL-60 cells, PLD was inhibited by ceramide. Fig. 1 demonstrates that PMA stimulated PLD activity 2-3-fold in HL-60 cells. C 6 -Ceramide inhibited PMA-induced PLD in a dose-and time-dependent manner such that PLD activity decreased as early as 1 h after ceramide treatment and was at base-line unstimulated levels by 4 h of ceramide treatment. C 6 -Ceramide was effective at concentrations as low as 2 M, and maximal activity was reached by 10 M C 6 -ceramide.
To measure how quickly ceramide was reaching its putative internal target and whether a metabolite might be the active species, we studied the uptake and metabolism of [ 14 C]C 6ceramide. C 6 -Ceramide was taken up quite slowly with 17, 28, 37, and 42% being associated with the cells after 0.25, 0.5, 1.0, and 2.0 h, respectively. After 24 h, the amount of cell-associated radiolabel had dropped to 24%. [ 14 C]C 6 -Ceramide was also metabolized slowly. The starting material was 96% pure, and 91% remained as [ 14 C]C 6 -ceramide after 2 h. After 24 h, 78% of the radioactivity still chromatographed as [ 14 C]C 6 -ceramide. Since PLD inhibition by ceramide occurred within 4 h of ceramide treatment, these results suggest that ceramide and not one of its metabolites is the likely active species.
Ceramide inhibition of PMA-activatible PLD was specific in that C 2 -ceramide was equally as active as C 6 -ceramide, but the related 4,5-dihydroceramides had little or no activity (Fig. 2). Removal of the fatty acid (forming sphingosine) substantially reduced the activity. Introduction of a long chain fatty acid (stearate) eliminated its effect, although this is likely due to poor solubility. We chose to use C 6 -ceramide for the majority of our studies since it is more similar to the parent compound but has sufficient solubility.
Since PKC is the intracellular phorbol ester receptor, which is known to activate PLD, we next examined the effect of C 6 -ceramide on PKC translocation. Cells were treated with 10 M C 6 -ceramide for 24 h and then challenged with PMA for 20 min. Ceramide did not inhibit translocation of PKC␣ or ␤II (the only detectable PKC isoforms in HL-60) under conditions that it inhibited PLD activation (Fig. 3). Ceramide was also unable to inhibit purified PKC activity (16). Therefore, ceramide appeared to inhibit PLD activity downstream of PKC.
To delineate the mechanism of ceramide inhibition of PLD, we used the cell-free system described by Brown et al. (3). HL-60 cells were nitrogen-cavitated, and PLD activity was reconstituted by mixing membrane and cytosol preparations with GTP␥S and Ca 2ϩ . C 6 -Ceramide had no effect under these conditions (Fig. 4), indicating that the inhibition of PLD by ceramide was not a direct effect on the enzyme. Varying the ceramide concentration, temperature of preincubation, and preincubating cytosol and membrane compartments separately with C 6 -ceramide also failed to inhibit PLD activity (data not shown).
We considered that ceramide may exert its effect upstream of FIG. 1. PLD activation by PMA and inhibition by ceramide. Cells (2.5 ϫ 10 5 in 1 ml) were seeded in 24-well plates. Cells were rested and radiolabeled with [ 3 H]palmitic acid (3 Ci, added in 3 l ethanol) in serum-free media for 24 h. Cells were also pretreated with 10 M C 6 -ceramide for the indicated time (A) or treated for 24 h with the indicated concentration of C 6 -ceramide (B). Cells were stimulated in 2% ethanol with 100 nM PMA for the 20 min. The cells were placed on ice, and then 1 ml methanol was added to stop the reaction. Wells were washed with an additional 1 ml of methanol, and lipids were extracted (15). Lipids were separated by thin-layer chromatography (solvent B), and phosphatidylethanol (PEt) was quantitated by liquid scintillation spectrometry. These experiments were performed in duplicate (mean Ϯ S.E. (bars)) and are each representative of three experiments. the GTP␥S-activable protein. Since PKC-induced activation of PLD in intact cells was inhibited by ceramide ( Fig. 1), we tested the effect of PKC activators on the cell-free PLD activity. The addition of PKC activators to membrane and cytosol fractions in the absence of GTP␥S had no effect on PLD activity (Fig. 5). When these were added in combination with GTP␥S, however, we observed a synergistic increase in activity by 5-fold over background (Fig. 5). This effect of PKC activation was ATP-dependent, indicating that it was mediated by phosphorylation. These results suggested that PKC activation of PLD may be an important control site in this two-step activation of PLD in HL60 cells.
To test the hypothesis that ceramide-mediated inhibition of PLD was through inhibition of the PKC-activable component, we optimized and characterized the PKC-dependent, cell-free system. This was initially performed by studying the protein dependence of the PKC-activable PLD. Varying the membrane protein concentration while keeping the cytosol protein fixed at 7.1 g showed that optimal PKC activation of PLD was obtained in the presence of 3.4 g of membrane protein (Fig. 6A). Varying the cytosol protein concentration while keeping the membrane protein concentration constant at 3.4 g showed that optimal PKC-activable PLD was seen with 7.1 g of cytosol protein (Fig. 6B). These optimal protein concentrations were then used to survey other requirements and potential activators/inhibitors of PKC-activable PLD (Fig. 7). The Ca 2ϩ requirement was found to have a narrow optimum at 4 mM total Ca 2ϩ (data not shown). The activity was completely inhibited by high levels of Zn 2ϩ , Mn 2ϩ , and Ca 2ϩ . PLD activity is thought to be regulated by tyrosine kinases (20,21). We wanted to determine if this regulation involved the PKC-activable component of PLD. The tyrosine kinase inhibitor genistein did not inhibit PKC-activable PLD, but curiously, the phosphotyrosine phosphatase inhibitor orthovanadate was totally inhibitory (Fig. 7). Also, the ceramide-activated protein phosphatase/protein phosphatase 2A inhibitor okadaic acid was without effect (Fig. 7).
Once the PKC activation of PLD was established and its primary components and requirements characterized, we proceeded to test the possibility that C 6 -ceramide would act at this level. C 6 -Ceramide was added to the protein components first, mixed, and then the activators Ca 2ϩ , GTP␥S, PMA, and MgATP were added, and the incubations were performed as above. Our results showed that ceramide inhibited the PKCdependent activation of GTP-binding protein-dependent PLD activity in a dose-dependent fashion such that as little as 1 M of C 6 -ceramide was effective at inhibiting PKC-activable PLD with maximal effects seen with 30 M of C 6 -ceramides (Fig. 8). This inhibition was not through sequestration of PMA because PMA concentrations from 50 nM to 3 M were used, and ceramide inhibition was unchanged (data not shown). These data demonstrate that the mechanism of inhibition of PLD by ceramide is via inhibiting the PKC-activable component. Coupled with the lack of inhibition of PKC translocation by ceramide and its inability to inhibit PKC directly (Ref. 22 and data not shown), these data localize the ceramide target between PKC and the GTP-binding protein in the activation of PLD in HL60 cells.
In HL60 cells, the GTP␥S-responsive element involved in PLD activation was shown to be cytosolic and was identified as ARF (3,4); however, in neutrophils the GTP-binding protein component was shown to be in the membrane and was identified as Rho (2). In our HL60 cells, the PLD activity appears to be in the membrane (Ref. 3 and data not shown). PLD activation by GTP␥S, however, requires a cytosolic component in that the addition of GTP␥S to membrane in the absence of cytosol only slightly stimulated PLD activity from 1195 Ϯ 476 to 1399 Ϯ 269 dpm phosphorylcholine, and Western blot analysis showed that ARF was greater than 95% cytosolic (data not shown). Therefore, we studied the role of ARF in PLD activation and its inhibition by ceramide.
Our first goal was to reconstitute membrane PLD activation using defined cytosolic components, i.e. ARF and PKC. Substitution of recombinant 5.7% myristoylated ARF1 for cytosol reconstituted GTP␥S activation (Fig. 9A). The addition of baculovirus-expressed PKC␣ also stimulated PLD activity. This was in contrast to what was seen using cytosol as the source of PKC, where cytosol and PKC activators alone (i.e. in the absence of GTP␥S) did not activate PLD. The addition of rARF1 and PKC␣ stimulated high levels of PLD activity (Fig. 9A), thereby completely reconstituting the activation pattern seen with crude cytosol.
Next we defined the requirements for the activators (GTP␥S, PMA, and ATP). The substitution of rARF for cytosol substantially activated PLD in the presence of GTP␥S (Fig. 9B). Significantly greater activation was achieved by the addition of PMA and ATP (Fig. 9B), which may be due to a small amount of PKC that is membrane-associated in this preparation. The substitution of rPKC␣ for cytosol greatly stimulated PLD in washed membranes (Fig. 9C). This required only the PKC activators PMA and ATP, whereas no further activation by GTP␥S was observed. Complete reconstitution of the two-step activation pathway with rARF1 and rPKC␣ required both GTP-binding protein and PKC activation (Fig. 9D) because it showed progressively greater activity with the addition of GTP␥S, PMA plus ATP, and GTP␥S plus PMA plus ATP.
We next evaluated the inhibitory effect of C 6 -ceramide on this well-defined system (Fig. 10). C 6 -Ceramide was unable to inhibit ARF-activable PLD; however, C 6 -ceramide completely abolished the synergistic activation by PKC. Higher concentrations of C 6 -ceramide were required to maintain the lipid:protein ratio the same as that used in the crude system. These results with defined components recapitulate the observations in the crude system. They also define the least system required to preserve the effects of PKC and the action of ceramide. Importantly, they suggest, in inhibiting PLD, the target for ceramide resides in the membrane. DISCUSSION These studies demonstrate that ceramide inhibits PMA-activated PLD in intact HL-60 cells. More importantly, this study demonstrates ceramide inhibition of membrane-bound PLD using a cell-free assay that has allowed us to gain insight into the mechanism of ceramide action. We have shown that ceramide inhibits PKC activation of GTP-binding, protein-dependent PLD without inhibiting the GTP-binding protein or the PLD directly.
After characterizing the components of PLD activation in the HL60 cell reconstituted system, we identified the mechanism by which ceramide was inhibiting PLD. Our data using both the whole cytosol and the rARF/rPKC␣ system demonstrate that the augmentation of ARF-dependent PLD activity by PKC is the target of ceramide inhibition. These experiments also show that ceramide does not act directly on the PLD enzyme or the GTP-binding protein and that ceramide does not have nonspecific effects on the membrane or substrate vesicles. On the other hand, ceramide specifically inhibited PMA-activation of PLD in cells and in the reconstituted cell-free system. Re- constitution of PLD activation with washed membranes plus two recombinant cytosolic proteins, ARF and PKC␣, demonstrates that other cytosolic proteins are not required for activation. A role for membrane-associated Rho is also possible (2), although we were not able to rule that out in these studies.
Our finding that ceramide does not directly inhibit phosphatidylinositol bisphosphate-dependent PLD or its activation by GTP␥S is supported by a recent study by Nakamura et al. (23). They concluded that ceramide inhibited PLD through inhibition of a Ca 2ϩ channel, thereby inhibiting PKC activation of PLD.
It is unlikely that ceramide acts via its metabolism to sphingosine since our intact cell data demonstrate that sphingosine does not inhibit PLD as potently as C 6 -ceramide, and in the cell-free assay, no preincubation is necessary for maximal effect. Also, sphingosine has been reported to activate PLD in intact fibroblasts (12,24).
The discovery that ceramide inhibits PLD activation in this more defined system indicates that ceramide might inhibit the interaction of PKC with ARF/PLD without affecting PKC translocation per se. In addition to our finding that PKC translocation is not inhibited by ceramide, other studies show that ceramide does not inhibit PKC phosphorylation activity in vitro (16,22). Jones and Murray (11) also found inhibition of bradykinin stimulation of PLD activity by C 2 -ceramide in the intact mouse epidermal cell line HEL-37. In that study, the authors demonstrate only slight inhibition of translocation of PKC␣ in response to C 2 -ceramide. Blobe et al. (25) have shown that PKC␣ interacts specifically with actin filaments. It is possible that this interaction is affected by ceramide. In any case, our reconstituted system and the results of others (26 -28) implicate a novel role for PKC in signal transduction that is independent of classical PKC action i.e. translocation and phosphorylation, and that it is this action that is inhibited by ceramide. Although the regulatory pathways for phospholipase(s) D remain far from clear, a considerable amount of information has emerged recently. It is apparent that within a certain cell type, there are more than one PLD, as illustrated by Reinhold et al. (29), Massenburg et al. (30), Tyagi et al. (31), and Song and Foster (32) (see also Ref. 1). One PLD is activated by fatty acids, whereas another requires PKC. The small molecular mass GTP-binding proteins ARF and Rho each activate a PLD (1), possibly working together (33,34) and on the same enzyme.
Some studies have shown a synergistic activation of PLD by protein kinases and the small GTP-binding proteins (35,36). We also found this synergy, where activation of the GTP-binding protein alone stimulates PLD significantly and the addition of PKC substantially augments GTP-binding protein activation of PLD. This activation by PKC required ATP and PMA in our studies; however, ATP␥S and AMP-PNP were equally effective as ATP in the rARF plus rPKC assay system (data not shown; see also Ref. 37). Olson et al. (38) reported that PKC activation of PLD is ATP-dependent. On the other hand, Conricode et al. FIG. 10. rARF1/rPKC␣-activated membrane PLD is inhibited by ceramide. Assays were performed as described in Fig. 9. C 6 -Ceramide was added at the indicated concentration to the protein components just prior to the enzyme, activators, and the substrate. Data are the mean Ϯ range (bars, S.E.) of duplicate measurements and are representative of three experiments. GTPS, GTP␥S. (26) reported direct activation of PLD by PKC in the absence of ATP, and Singer et al. (27) reported separation of the PLD activation domain from the catalytic domain of PKC following trypsin treatment, suggesting that PKC-mediated phosphorylation is not required. Clearly, there are still some gaps in our knowledge of this pathway. The recently cloned PLD1 gene (39) should help clarify some of these issues.
The PLD that is inhibited by ceramide has the following characteristics: 1) it requires cooperative activation by ARF and PKC; 2) it is located in the membrane; 3) it is dependent on phosphatidylinositol bisphosphate; and 4) it is inhibited by free fatty acid (data not shown) (30). Recently, several laboratories have shown that ceramide plays a role in the regulation of PLD in intact neutrophils (10), fibroblasts (8,11,12), basophilic leukemia cells (23), and in HL-60 cells as shown here. Taken together, these studies suggest that ceramide may be a general regulator of PLD. This is particularly interesting in that ceramide seems to play a role in regulating PLD activation in both the mitogenic (fibroblasts and HL60 cells) and the inflammatory (neutrophils and basophilic leukemia cells) responses.
The results from this study provide mechanistic insight into how ceramide may interfere with cell growth and proliferation. We and others have shown that ceramide has an important role in the regulation of cell growth, differentiation, and apoptosis. We have demonstrated previously that there is cross-talk between the phospholipid and sphingolipid signaling pathways in nutrient deprivation (40) and in apoptosis (41). Recently, we (9) have demonstrated a role for ceramide as an endogenous mediator of cellular senescence in that ceramide levels and sphingomyelinase activity are elevated in senescent cells (but not in quiescent cells). Exogenous ceramide appears to induce a senescent phenotype in young human diploid fibroblasts. Furthermore, we found that diacylglycerol generation and PLD activity is inhibited in senescent cells, most probably due to ceramide accumulation (8). Here, using HL60 cells, we confirm that ceramide inhibits PLD activation and uncover a potential mechanism by which this occurs.
In conclusion, our studies shed new light on the current understanding of PLD regulation and the mechanism of ceramide action. We hope to identify the mechanism of ceramide inhibition through careful study of PKC activation. In this regard, ceramide will enable in-depth study of the mechanism of activation of PLD. We also provide a model for understanding the target of ceramide action in the regulation of cell growth, apoptosis, and cellular senescence.