Ceramide inhibition of phospholipase D and its relationship to RhoA and ARF1 translocation in GTPγS-stimulated polymorphonuclear leukocytes

Polymorphonuclear leukocytes (PMNs) serve an important role in host defense by ingestion (phagocytosis) and destruction of microbes. Destruction is accomplished by release of reactive oxygen intermediates and granule enzymes to the phagosome. We and others have demonstrated that PMN phagocytosis, degranulation, and oxidant production are regulated by phospholipase D (PLD).^1-7  PLD produces phosphatidic acid (PA) which may act directly as a second messenger or be converted to diacylglycerol (DAG) by phosphatidate phosphohydrolase. DAG activates protein kinase C (PKC)⁸  and may contribute to granule-membrane fusion.⁹  Two mammalian PLDs, PLD1 and PLD2, have been described¹⁰ ; however, recent PMN studies have only begun to distinguish between the 2 isoforms. PLD1 activation is dependent on the small guanosine triphosphateases (GTPases) adenosine diphosphate (ADP)–ribosylation factor (ARF) and RhoA, one or more PKC isoforms, and phosphatidylinositol-bisphosphate (PIP₂), whereas PLD2 activity appears not to involve the GTPases but is primarily stimulated by PIP₂.¹⁰  Activation of PLD in PMN and HL-60 cells is dependent on ARF and RhoA,^11,12  suggesting that PLD1 is the primary isozyme being measured in previous PMN studies. Both ARF1 and ARF6 have been implicated in PLD activation in phagocytic cells.^13,14 

Sphingolipid signaling is involved in cell growth, differentiation, and apoptosis.¹⁵  During PMN phagocytosis, plasma membrane-associated neutral sphingomyelinase is activated, resulting in accumulation of ceramide within the cell.^3,16  Significant ceramide accumulation occurs concurrent with a decline in the rate of phagocytosis, suggesting that ceramide is a negative regulator of PMN function. The exogenous addition of the cell-permeable analog C₂-ceramide has been shown to inhibit phagocytosis, degranulation, and oxidant production in PMNs through its inhibition of PLD activity.^1,3  The inhibition of PLD by ceramide is well established^17-19 ; however, the mechanism of inhibition has not been defined. Others have found C₂-ceramide to inhibit ARF and RhoA translocation to the plasma membrane in HL-60 cells.¹⁹  Other lipids found to inhibit PLD include the polyisoprenol phosphate, presqualene diphosphate, which directly inhibits PLD1b and superoxide generation,²⁰  and certain lysophospholipids which inhibit rat brain PLD.²¹ 

In this study we hypothesized that direct activation of PLD by way of guanosine 5′-[gamma-thio]triphosphate (GTPγS) mobilization of ARF and RhoA would allow evaluation of ceramide effects that could not be observed with receptor activation. We used 2 methods to bypass receptor activation of PLD, streptolysin-O permeabilization and a cell-free system. By using each method we could activate PLD and inhibit activation with C₂-ceramide. In permeabilized PMNs we stimulated degranulation with GTPγS through the same pathway.

D-erythro-C₂-ceramide, D-erythro-C₆-ceramide, and D-erythro-dihydro-C₂-ceramide were purchased from Matreya (Pleasant Gap, PA). Antibodies against lactoferrin were a gift from Dr Niels Borregaard, State University Hospital, Copenhagen, Denmark. Antibodies against ARF1 and ARF6 were kindly provided by Dr Sylvain Bourgoin, Université Laval, Québec, Canada. The antibody against RhoA was purchased from Santa Cruz Biotechnology, Santa Cruz, CA. Isozyme-specific PKC antibodies were obtained from BD Transduction Laboratories, San Diego, CA. Recombinant RhoA was purchased from Calbiochem (San Diego, CA).

PMNs were isolated from peripheral venous blood from healthy volunteers as previously described.³  Volunteers gave informed consent according to the Helsinki protocol. Data reported herein are pooled from experiments using several different donors. For experiments in which cells were disrupted, PMNs were pretreated with 5 mM diisopropyl fluorophosphate on ice for 5 minutes, then washed 3 times with phosphate-buffered saline (PBS).

PMNs were first incubated with the indicated concentrations of lipids for 30 minutes at 22°C. In later experiments it was determined that preincubation was not necessary for inhibition when PMNs were permeabilized, so ceramide was added along with the other reagents. PMNs (20 × 10⁶ cells/mL final) in Dulbecco phosphate buffer (50 mM HEPES (N-2-hydroxyethylpiperazine-N′-2-ethanesulfonic acid), pH 7.4, 100 mM KCl, 20 mM NaCl, 2 mM MgCl₂, 1 mM EGTA (ethyleneglycoltetraacetic acid), and 1 g/L dextrose) with 1 mg/mL bovine serum albumin (BSA) were added to tubes containing (final concentrations) 75 U/mL streptolysin O (Sigma, St Louis, MO), GTPγS as indicated in Figure 1, 1 mM adenosine triphosphate (ATP), and 3 μM Ca⁺⁺. Samples were incubated for 10 minutes at 37°C, then placed on ice 3 minutes. PMNs were removed by centrifugation, and supernatants were assayed for lactoferrin by enzymelinked immunosorbent assay (ELISA) as previously described.²²  All data were analyzed using 2-sided Student t tests.

PMNs were first labeled with 1-O-[³H]-octadecyl-sn-glycero-3 phosphocholine ([³H]-lyso-PAF) (Amersham, Arlington Heights, IL) as previously described.³  Incubations were carried out as described for lactoferrin experiments except for the addition of 1% ethanol for 5 minutes at 37°C prior to the stimulation/permeabilization step. Cell pellets were extracted according to the method of Van Veldhoven and Bell.²³  Lipids were separated in a thin-layer chromatography (TLC) system as previously described,²  and the plate was exposed to film. Radioactive phosphatidylethanol bands were scraped for scintillation counting.

PMNs were first [³H]-labeled as described in “PLD activity.” PMNs (2 × 10⁶/mL) were incubated with 10 μM C₂-ceramide, dihydro-C₂-ceramide, or no lipid (control cells) for 30 minutes at 22°C. Cells were then suspended at 1 × 10⁸/mL in disruption buffer (100 mM KCl, 3 mM NaCl, 10 mM piperazine diethanesulfonic acid [pH 7.0], 3.5 mM MgCl₂, 1 mM PMSF (phenylmethylsulfonyl fluoride), 100 μg/mL soybean trypsin inhibitor, 10 μM pepstatin, and 10 μg/mL each of aprotinin and leupeptin) and probe sonicated twice for 15 seconds. Cytosol and plasma membrane fractions were separated as described previously.¹  Briefly, samples were ultracentrifuged over discontinuous 15% and 35% sucrose gradients. Plasma membrane fractions were taken from the 15% 35% interface, then pelleted. Membranes were homogenized and assayed for protein with a bicinchoninic acid (BCA) assay. Equal amounts of membrane protein were used for each sample, and cytosol from only control cells was used. Before activation, samples were incubated with 1% ethanol. In some experiments, recombinant RhoA was then added to the cytosol. Cytosol was activated with 10 μM GTPγS (5 minutes, 37°C) then combined with membrane protein isolated from controls or cells pretreated with lipid (10 minutes, 37°C). Samples were extracted and TLC was performed as described in “PLD activity.”

Aliquots were taken from samples in PLD experiments in the reconstituted system. These aliquots were ultracentrifuged, and the pellets were solubilized in sample buffer to run 12.5% sodium dodecyl sulfate–polyacrylamide gel electrophoresis (SDS-PAGE). Proteins were transferred, and Western blots were performed with antibodies against ARF1, ARF6, PKCα and β, and RhoA as described previously.²⁴  Scanning densitometry was performed to quantify ARF1 and RhoA.

PMNs were incubated with C₂-ceramide, dihydro-C₂-ceramide, or buffer for 30 minutes at 22°C. Intact cells were then incubated with 100 nM N-formyl-1-methionyl-1-leucyl-1-phenylalamine (FMLP) for 10 minutes at 37°C. Alternatively, PMNs were streptolysin O–permeabilized, then stimulated as described in “PLD activity in cell-free system” with 10 μM GTPγS for 10 minutes at 37°C. Cells were probe sonicated, and cytosol was separated as described in “PLD activity in cell-free system,” then samples were prepared for 12.5% SDS-PAGE and Western blotting with antibody against RhoA.

PMNs were incubated with C₂-ceramide, then disrupted and fractionated as described in “Protein translocation in intact and permeabilized PMNs.” Lipids were extracted from the fractions by the method of Van Veldhoven and Bell,²³  then C₂-ceramide was assayed by the method of Preiss et al.²⁵ 

Previous studies by us and others^1,3,19  have demonstrated C₂-ceramide inhibition of PLD activity. However, many of these studies involved cell-free systems, wherein some functions cannot be assayed, or receptor-mediated PLD activation, in which parallel pathways may also be activated and confound the signaling model. We permeabilized PMNs with streptolysin O to introduce GTPγS and, thus, directly activate PLD and monitored PMN activation by evaluating lactoferrin release from specific granules. GTPγS stimulated the release of lactoferrin in a dose-dependent manner and reached maximum effect at a concentration of 10 μM. C₂-ceramide (50 μM) significantly inhibited lactoferrin release, whereas the inactive analog dihydro-C₂-ceramide did not (Figure 1). Some lactoferrin release occurred in unstimulated controls as a result of cell warming and handling, and ceramide inhibition was observed even at that low level of release. During streptolysin-O permeabilization, PMNs were at a cell density of 20 × 10⁶/mL, in contrast to our previous studies, in which PMNs were suspended at 2 × 10⁶/mL. In the latter studies, 10 μM C₂-ceramide was adequate to inhibit degranulation.¹  In the permeabilized PMNs, 50 μM C₂-ceramide was required. To evaluate possible nonspecific interactions between C₂-ceramide and streptolysin O, we measured C₂-ceramide in membrane and cytosol fractions of intact and streptolysin O–permeabilized PMNs after incubations. Streptolysin O did not prevent ceramide association with the membrane, and ceramide concentration did not significantly affect the percentage of the total associated with the membrane (data not shown). To inhibit function, intact PMNs required preincubation with C₂-ceramide,³  but permeabilized PMNs did not, suggesting that ceramide entered the cell through the pores created by the toxin.

Lactoferrin release was also measured in the presence of C₆-ceramide. C₆-ceramide inhibited lactoferrin release more potently than C₂-ceramide at the lower doses tested. At 20 μM, C₆-ceramide inhibited release significantly, whereas C₂-ceramide did not (Figure 2).

To determine whether ceramide had a direct effect on lactoferrin catalysis or regulation, we incubated PMNs with C₂- or C₆-ceramide and measured the total lactoferrin in detergent-lysed whole-cell preparations. Ceramide incubation had no effect on total lactoferrin present (data not shown).

PLD activity was also tested in streptolysin O–permeabilized PMNs, under the same conditions as lactoferrin release. GTPγS stimulated PLD in a dose-dependent fashion, peaking at 10 μM with a 10-fold increase over baseline (Figure 3). GTPγSat100 μM did not stimulate PLD significantly more than did 10 μM (data not shown). C₂-ceramide (50 μM) significantly inhibited PLD activity, whereas dihydro-C₂-ceramide treatment enhanced PLD activity. These results demonstrate that ceramide inhibits PLD activity and PLD-dependent functions in cells directly stimulated with GTPγS, as it does in receptor-stimulated PMNs.¹ 

To further evaluate the mechanisms underlying PLD activation, we used a reconstituted cell-free system in which [³H]-lyso-PAF–labeled PMNs were treated with C₂-ceramide, then disrupted, and cytosol was separated from plasma membrane. The cytosol was activated with 10 μM GTPγS, then the membrane fraction from control or ceramide-treated cells was added to the cytosol to reconstitute PLD activity. PLD activity increased about 7-fold with stimulation (Figure 4). Activity was inhibited significantly by C₂-ceramide but not by dihydro-C₂-ceramide, demonstrating results similar to those observed in permeabilized PMNs. As ceramide was not added after cell disruption, inhibition suggests that ceramide remains bound to cell membranes during the disruption and fractionation processes.

Others have found that ARF and RhoA translocation to the plasma membrane are partially inhibited by 50 μM C₂-ceramide in HL-60 cells.¹⁹  To address the effect of ceramide in PMN function, first we used the reconstituted cell-free system of GTPγS activation. After incubating cytosol with plasma membrane, samples from the cell-free system were centrifuged again to separate the fractions. These fractions were Western blotted and probed for ARF1, ARF6, and RhoA. ARF1 association with the plasma membrane was not significantly affected by prior C₂-ceramide treatment (Figure 5), suggesting that ceramide does not prevent GTPγS from interacting with proteins in this system. Similarly, the amount of ARF6 associated with the plasma membrane was not affected by C₂-ceramide (data not shown). However, RhoA translocation was significantly inhibited by prior treatment with 10 μM C₂-ceramide, whereas it was not inhibited by dihydro-C₂-ceramide (Figure 6), suggesting that the mechanism of PLD inhibition by ceramide would be a failure of RhoA translocation to PMN plasma membranes containing PLD. To further test this hypothesis, we evaluated RhoA translocation to the membrane in intact PMNs and in streptolysin O–permeabilized PMNs. In contrast to the cell-free system, translocation was not inhibited by C₂-ceramide (Figure 7), suggesting that the inhibition of PLD by ceramide is not necessarily mediated through RhoA.

If ceramide were to block the interaction between RhoA and PLD, addition of exogenous RhoA might overcome ceramide inhibition. In the cell-free system, the addition of up to 32 μg/mL (approximately 1.4 μM) exogenous RhoA did not restore PLD activity (Figure 8A). However, the exogenous RhoA, initially added to the cytosolic fraction, was capable of binding to the membrane fraction in the presence of C₂-ceramide (Figure 8B), providing more evidence that ceramide does not only inhibit PLD through an effect on RhoA.

PKCα, and likely PKCβ, are involved in positive regulation of PLD1.²⁶  The PKCs translocate to the plasma membrane on cell stimulation where they can interact with PLD. Thus, we evaluated samples from the cell-free system for PKC translocation. Prior treatment with C₂-ceramide had no effect on the amount of PKCα or β associated with the plasma membrane (data not shown), as we observed previously in intact PMNs.¹ 

Phosphorylation processes appear to be involved with PLD activation; for example, tyrosine kinase inhibitors reduce PLD activation and membrane recruitment of ARF, RhoA, and PKCα.²⁷  However, ARF1 and RhoA translocation were not affected by using apyrase to hydrolyze the ATP in a cell-free system from HL-60 leukemic cells.¹⁹  We incubated the fractions in our cell-free system with 5 U/mL apyrase for 10 minutes at 37°C before GTPγS activation. Neither PLD activity nor RhoA translocation was affected by apyrase treatment (data not shown), indicating that activation and translocation were not energy-dependent.

In this study we demonstrated functional activity in streptolysin O–permeabilized PMNs by release of lactoferrin from specific granules. We determined that activation of ARF and RhoA by GTPγS in permeabilized PMNs led to both activation of PLD and degranulation. Permeabilization allowed GTPγS to enter PMNs, bypassing receptor activation. Lactoferrin release and PLD activation responded to similar doses of GTPγS and were both inhibited by treatment with C₂-ceramide, consistent internally and with previous studies.¹  Enhancement of PLD in permeabilized PMNs by the generally inactive dihydro-C₂-ceramide may have occurred because of potential metabolism of dihydroceramide to sphingosine. C₆-ceramide was more potent than C₂-ceramide in inhibiting lactoferrin release, suggesting endogenous ceramide with its increased carbon chain length would be even more potent in inhibiting function in vivo. In the cell-free system, PLD was similarly activated by GTPγS and inhibited by C₂-ceramide. In the cell-free system we observed inhibition of RhoA translocation to the plasma membrane by ceramide pretreatment. However, ceramide did not block RhoA translocation in intact or permeabilized PMNs.

In other cell systems, translocation of ARF and RhoA to the cell membrane is inhibited by 20 to 50 μM C₂-ceramide, 2- to 5-fold more than was needed to block RhoA, but not ARF, in the cell-free system in this study.^19,28  Inhibition of RhoA translocation by ceramide in the cell-free system was specific, as RhoA was inhibited by a concentration of 10 μM C₂-ceramide but not the inactive analog dihydro-C₂-ceramide. This concentration of C₂-ceramide was previously shown not to result in significant damage to PMNs as assessed by lactate dehydrogenase release, while specifically inhibiting phagocytosis, degranulation, and nicotinamide adenine dinucleotide phosphate (NADPH) oxidase, but not chemotaxis.^1,3  In contrast to the cell-free system, in the intact PMN, cytochalasin B treatment by itself led to RhoA association with the membrane which was not affected by ceramide treatment.¹  Similarly, in intact PMNs incubated with FMLP alone, or in permeabilized PMNs activated with GTPγS, RhoA association with the membrane was not affected by ceramide. Addition of exogenous RhoA was unable to overcome ceramide inhibition of PLD in the cell-free system, despite binding to the membrane fraction where PLD is located.

The discrepancy in Rho translocation between cell-free and permeabilized systems may indicate a more complex interaction between ceramide, PLD, and G proteins. For example, in the cell-free system, ceramide-treated membranes may be remodeled such that ceramide affects Rho binding to PLD but not to the membrane itself. Ceramide treatment of intact and permeabilized PMNs may inhibit PLD through another mechanism, such as disruption of the interaction of PLD with PIP₂ as initially suggested by Singh et al,²⁹  Using a cell-free system of dimethyl sulfoxide (DMSO)–differentiated (PMN-like) HL-60 cells, Singh et al²⁹  found that PLD activation is inhibited by ceramide at low concentrations of PIP₂. This inhibition is antagonized by excess PIP₂, consistent with the idea that ceramide inhibits a PLD-PIP₂ interaction. We attempted these experiments in our PMN cell-free system, applying PIP₂ and ceramide in liposomes of phosphatidylcholine and phosphatidylethanolamine to recombined cytosol and membrane fractions.²⁹  However, we were unable to replicate the HL-60 results, finding no PIP₂ enhancement of PLD activity in the presence or absence of ceramide (data not shown). The PLD in PMN is primarily localized in secretory vesicles,³⁰  which fractionate with the plasma membrane in the method we used. The PLD in these PMN membrane fraction vesicles may be unavailable to the exogenous PIP₂, in contrast to the PLD of HL-60 cells which is in the plasma membrane. Singh et al²⁹  also observed that the activities of truncated mutants of PLD1 are inhibited by ceramide. These data together suggest that ceramide interacts with the catalytic core of PLD and that ceramide inhibits PLD more effectively when PIP₂ concentrations are rate limiting.²⁹  Our data do not preclude the possibility of direct ceramide interaction with PLD. Additionally, overexpression of RhoA or its downstream effector Rho kinase increases activity of phosphoinositide 4-phosphate (PI4-P) 5-kinase and resulting levels of PIP₂,³¹  suggesting RhoA could affect PLD both directly as a cofactor and indirectly by regulating PIP₂, another cofactor. We measured PI4-P 5-kinase activity by the phosphorylation of PI4-P using the method of Oude Weernink et al³¹  but found no inhibition by ceramide (data not shown).

Ceramide inhibition of RhoA translocation in the cell-free system could be related to phosphorylation/dephosphorylation events. Ceramide begins to accumulate after cell activation, at a time early enough to indicate a negative regulatory role.^3,32  Ceramide activates several protein kinases and phosphatases^15,33,34 ; thus, ceramide possibly could affect RhoA function indirectly through stimulation of phosphorylation or dephosphorylation. Phosphorylation of RhoA by protein kinase A enhances its dissociation from the membrane,^35,36  suggesting RhoA association with the membrane may be regulated by phosphorylation. To evaluate whether a phosphorylation process was required to mobilize RhoA, we treated the PMNs with apyrase to deplete the cells of ATP. We found that apyrase treatment to hydrolyze endogenous ATP did not affect RhoA translocation. Other studies suggested that PKC regulates translocation of RhoA in an ATP-independent manner,¹⁰  consistent with our observations. However, PKC did not appear to be involved in translocation of RhoA in the cell-free system, based on ceramide inhibition of RhoA translocation but not translocation of PKCα or β.

In summary, our finding of ceramide inhibition of RhoA translocation in the cell-free system does not preclude the direct interaction of ceramide and PLD. Ceramide may interfere with binding of cofactors to PLD or production of PIP₂ through Rho kinase effects on a phosphatidylinositol kinase.