Recent evidence suggests that phospholipase D (PLD) can be regulated through its association/dissociation to lipid rafts. We show here that modifying lipid rafts either by cholesterol depletion using methyl-β-cyclodextrin and filipin or by conversion of sphingomyelin to ceramide with exogenous bacterial sphingomyelinase (bSMase) markedly activated the PLD of human PBMC. bSMase was the most potent PLD activator, giving maximal 6- to 7-fold increase in PLD activity. Triton X-100-treated lysates prepared from control PBMC and from bSMase-treated cells were fractionated by centrifugation on sucrose density gradient. We observed that bSMase treatment of the cells induced a larger ceramide increase in raft than in nonraft membranes and displaced both the Src kinase Lck and PLD1 out of the raft fractions. In addition, the three raft-modifying agents markedly inhibited the lymphoproliferative response to mitogenic lectin. To examine further the potential role of PLD activation in the control of lymphocyte responses, we transiently overexpressed either of the PLD1 and PLD2 isoforms in Jurkat cells and analyzed the phorbol ester plus ionomycin-induced expression of IL-2 mRNA, which is one of the early responses of lymphocyte to activation. We observed a 43% decrease of IL-2 mRNA level in Jurkat cells overexpressing PLD1 as compared with mock- or PLD2-transfected cells, which indicates that elevated PLD1, but not PLD2, activity impairs lymphocyte activation. Altogether, the present results support the hypothesis that PLD1 is activated by exclusion from lipid rafts and that this activation conveys antiproliferative signals in lymphoid cells.
Phospholipase D (PLD),4 which hydrolyzes phosphatidylcholine (PC) to phosphatidic acid (PA), is a widely distributed enzyme that has been implicated in various mammalian physiological responses such as membrane trafficking, reorganization of actin cytoskeleton and cell proliferation (1, 2). In intact cells, PLD activity is usually low and is transiently increased by agonists of many cell surface receptors, including hormones, neurotransmitters, growth factors, and cytokines. Many of these receptors also stimulate phospholipase C activity leading to increases in cellular Ca2+ and diacylglycerol (DAG). Two mammalian PLD genes, PLD1 and PLD2, have been identified, each of them giving rise to splice variants. Both PLD1 and PLD2 enzymes are strictly dependent on phosphatidylinositol 4,5-bisphosphate (PIP2) for their activity. PLD1 has a low basal activity and is activated by small GTP-binding proteins belonging to the ADP ribosylation factor and ρ families and by protein kinase C. PLD2 is usually found constitutively active and is only modestly activated by ADP ribosylation factor proteins and protein kinase C (1, 2). With a few exceptions such as HL60 cells, which only express PLD1, human Jurkat T cells, which only express PLD2, or the murine T cell line EL4, which is virtually devoid of PLD, most mammalian tissues and cell lines express both PLD isoforms although at different levels (3). Both PLD1 and PLD2 are membrane associated. From microscopy localization studies using overexpressed-tagged PLDs, a consensus seems to emerge indicating that PLD2 is preferentially found at the plasma membrane while PLD1 is preferentially located in the endosomal/lysosomal compartment (2). However, there is no general agreement about the exact location of endogenous enzymes in cells. In the past few years, several groups have reported the presence of PLD enzymes in caveolae, which are specialized domains of the plasma membrane related to sphingolipid- and cholesterol-rich microdomains or rafts (2). These lipid domains characterized by their insolubility in cold nonionic detergents are also present in cells devoid of caveolae such as lymphoid cells (4). We shown have recently that a substantial part of PLD1 activity and protein was located to the detergent-insoluble membranes of human PBMCs (5). Furthermore, we have also shown that the partial disorganization of PBMC rafts by enrichment of the cells with the polyunsaturated fatty acid docosahexaenoic acid (DHA) induces a shift of PLD1 out of the rafts and a concomitant activation of the enzyme. These findings led us to the hypothesis that the association of PLD1 with lipid rafts hampers its activity and, on the opposite, that PLD1 exclusion from rafts stimulates the enzyme.
The functional consequences of PLD activation on mitogenic signaling are not very well understood. In most cell types, PLD activation seems to favor cell proliferation or differentiation and to protect against apoptosis (1). However, the role of PLD in T cell activation remains controversial. Although some reports have described an activation of PLD following ligation of the TCR/CD3 complex (6, 7), other studies failed to demonstrate efficient CD3 coupling to PLD, either in normal peripheral lymphocytes (8) or in T cell lines (9). It has been shown that TCR ligation was accompanied by a fast and large increase in PA level occurring in the first minutes of T cell activation (10, 11). However, this early peak could be attributed to PA derived from DAG through the sequential activation of phospholipase C-γ and DAG kinase because it was totally suppressed by DAG kinase inhibitors (10, 11). Only when phospholipids of lymphocyte membrane were modified by enrichment with DHA (5, 12) or the monohydroxylated derivative of arachidonic acid, 12-HETE (13), were we able to observe PLD activation following mitogenic stimulation. In both cases, this was accompanied by an inhibition of the proliferative response (13, 14). In B lymphocytes, activation of a PC-specific PLD has also been associated to an inhibition of the proliferative response (15).
To further support the hypothesis that PLD1 exclusion from lipid rafts stimulates enzyme activity and to determine whether this might be a general mechanism of PLD activation, we studied the effect of raft disruption on the PLD activity of human PBMCs. For this purpose, we used two cholesterol-depleting agents, methyl-β-cyclodextrin (MBCD), a carbohydrate molecule containing a cholesterol-binding pocket that depletes membrane cholesterol, and filipin, an antifungal polyene that sequesters cholesterol within membranes (16). Because sphingomyelin is required to maintain the integrity of rafts, we also used bacterial sphingomyelinase (bSMase) to alter the properties of rafts in PBMCs (17). Among the raft-modifying agents tested, we found that bSMase was the most potent PLD activator. Thus, we further examined the influence of bSMase cell treatment on the lipid and protein content of rafts isolated from PBMCs by floatation on sucrose density gradient. In particular, we investigated whether bSMase treatment could alter PLD1 isoform location.
Because we found that the three raft-modifying and PLD-activating agents decreased the lymphocyte response to mitogenic lectins, we set out to further investigate the physiological consequences of PLD activation. For this purpose, we transiently transfected Jurkat T cells with GFP-PLD1 and -PLD2 constructs and examined the effect of overexpression on lymphocyte function. We first examined the consequences of PLD1 or PLD2 overexpression on spontaneous Jurkat cell proliferation. Then, because IL-2 is one of the first genes expressed during the early events of T cell activation (18), we investigated the effect of PLD1 or PLD2 transfection on IL-2 mRNA expression. Altogether, the present results support the hypothesis that activation of PLD1, but not PLD2, conveys antiproliferative signals in lymphoid cells.
Materials and Methods
Reagents and chemicals
RPMI 1640 medium with 25 mM HEPES and bicarbonate, NaCl, dextran, Histopaque-1077, trypan blue, PMSF, leupeptin, aprotinin, tris[hydroxymethyl] amino-methane, sucrose, human delipidated albumin (human serum albumin), peroxidase-conjugated cholera toxin B subunit, nonhydroxy fatty acid ceramide from bovine brain, C2- and C6-ceramides, sphingomyelinase from Staphylococcus aureus, filipin III, MBCD, Tri Reagent and agarose were all purchased from Sigma-Aldrich. Triton X-100 was obtained from Fluka (Sigma-Aldrich). [5,6,8,9,11,12,14,15-3H]Arachidonic acid (37 kBq/ml, specific activity 7400 GBq/mmol), MP Hyperfilm, ECL, HRP-conjugated anti-mouse or anti-rabbit IgG Abs were purchased from Amersham Biosciences.[32P-γ]-ATP was obtained from PerkinElmer Life Sciences. DAG kinase from Escherichia coli and anti-Lck Ab were obtained from TEBU (Le Perray-en-Yvelines, France). Immobilon P membrane was purchased from Millipore. One-step PCR kit was purchased from Qiagen. FCS and the oligonucleotides used as primers in the RT-PCR analysis of PLD and IL-2 were purchased from Invitrogen Life Technologies. β-Actin primer pair was from Promega.
Preparation of human PBMCs
Peripheral blood was obtained from healthy subjects who had not taken any medication for 2 wk before blood donation (Etablissement Français du Sang). Venous blood was drawn into citrate-phosphate-dextran anticoagulant. PBMCs were separated by dextran sedimentation and density gradient centrifugation through Histopaque 1077 and then washed three times with RPMI 1640 medium by low-speed centrifugation to more thoroughly eliminate the contaminating platelets. PBMCs were then adjusted to a concentration of 2 × 107cells/ml in RPMI 1640 medium (with HEPES and bicarbonate). All steps were conducted at room temperature. Under such conditions, cell viability established by the trypan blue exclusion test was always >95%. Flow cytometry analyses of cell preparations after staining with specific mAbs showed that ∼65–70% of the isolated cells were CD3+ T cells (T3 Coulter clone), 4–6% were CD19+ B cells (B4 Coulter clone), 16–24% were CD11b+ monocytes (MO1 Coulter clone), and 4–6% were CD41a+ platelets (GP IIb IIIa; Immunotech).
PLD was determined on the basis of its transphosphatidylation activity. Freshly isolated cells were incubated for 1 h at 37°C in the presence of [3H]arachidonic acid in 0.1% ethanol and then washed three times in RPMI 1640 medium. Labeled cells were resuspended at a concentration of 2 × 107cells/ml in RPMI 1640 medium containing 5 μM human delipidated albumin and incubated with 1–5 μg/ml filipin, 10–40 mM MBCD, or 1–1000 mU/ml bacterial sphingomyelinase for 30 min at 37°C in the presence or in the absence of 1% butanol. Filipin was dissolved in DMSO, and DMSO solutions were then appropriately diluted with culture medium. The final DMSO concentration was 0.05%. Incubations were terminated by addition of ethanol and acidification of the medium to pH 3–4 with 2 N HCl. Unlabeled phosphatidylbutanol was added as a carrier. Lipids were extracted with chloroform/ethanol (6:3, by volume) according to Boukhchache and Lagarde (19) in the presence of 50 μM butylhydroxylated toluene. Phosphatidylbutanol was separated on bidimensional TLC (silicagel G60 plates; Merck) using chloroform/methanol/28% ammonia (65:35:5.5, by volume) for the first migration, and ethylacetate/isooctane/acetic acid (9:5:2, by volume) for migration in the second dimension, as described previously (13). Spots stained by Coomassie brilliant blue R were scraped off, mixed with Picofluor (Packard), and the radioactivity was determined by liquid scintillation counting. Phosphatidylbutanol was identified by comparing Rf with that of an authentic phosphatidylbutanol sample spotted on the same plate as a standard. The radioactivity associated with phosphatidylbutanol was expressed as percentage of the radioactivity incorporated in total phospholipids.
PBMCs were lysed in a solution containing 0.1% SDS, 1 mM EDTA, and 0.1 M Tris buffer (pH 7.4). The gelatinous mixture was homogenized by five passages through a 19-gauge needle fixed to a 2-ml syringe. Cholesterol was determined enzymatically on aliquots of suitably diluted lysates using a commercial assay kit (Sigma-Aldrich) according to the manufacturer’s recommendations. Proteins were determined on aliquots of the same lysates by the method of Schaffner and Weissmann (20).
Isolation of lipid rafts from human PBMCs
Lipid rafts were isolated as described previously (5). Briefly, control or SMase-treated PBMCs (5 × 108 cells) were homogenized in 1 ml of ice-cold lysis buffer (25 mM Tris (pH 7.5), 150 mM NaCl, and 5 mM EDTA) supplemented with a mixture of protease inhibitors (protease inhibitor mixture; Sigma-Aldrich) and 1 mM orthovanadate. After centrifugation at 800 × g at 4°C for 10 min, the postnuclear supernatant was incubated with Triton X-100 at a final concentration of 1% for 1 h at 4°C. The lysate was then adjusted to 1.3 M sucrose by the addition of an equal volume of 2.6 M sucrose, placed at the bottom of an ultracentrifuge tube, and a step sucrose gradient (0.2–0.9 M with 0.1 M steps, 1 ml each) was placed on top. It was centrifuged at 200,000 × g for 16 h in an SW41 rotor (Kontron) at 4°C. One-milliliter fractions were recovered from the bottom to the top of the gradient. The sucrose concentration of each fraction was determined with a Brix refractometer (Merck). Fractions were stored at −80°C until lipid analyses. Proteins were determined by the method of Shaffner and Weissmann (20) using BSA as a standard. Gradient protein recovery relative to total proteins in the postnuclear supernatant varied from 75 to 87.5% and was similar in gradients prepared with control PBMCs (81.03 ± 4.87%, n = 3) and with bSMase-treated PBMCs (80.30 ± 2.40%, n = 3).
Ceramide and cholesterol assay
The amount of ceramide present in the gradient fractions was quantified by the DAG kinase assay (21) as 32P incorporated upon phosphorylation of ceramide to ceramide 1-phosphate by diacylglycerol kinase from E. coli. Ceramide 1-phosphate was resolved by TLC using chloroform/methanol/acetic acid (65:15:5, by volume) as solvent. The level of ceramide was determined by comparison to a concomitantly run standard curve comprised of known amounts of standard ceramide. Results are expressed as ng ceramide/ml fraction. Gradient ceramide recovery relative to total ceramide in the postnuclear supernatant was 51.72 ± 3.09% (n = 3) for gradients prepared with control PBMCs and 58.33 ± 5.7% (n = 3) for gradients prepared with bSMase-treated PBMCs (NS).
Cholesterol was determined enzymatically on aliquots of the gradient fractions, using a commercial assay kit (Sigma-Aldrich) according to the manufacturer’s recommendations. Results are expressed as μg cholesterol/ml fraction. Gradient cholesterol recovery relative to total cholesterol present in the initial crude homogenate (before elimination of nuclear pellets) was 54.45 ± 0.90% (n = 3) for gradients prepared with control PBMCs and 57.38 ± 3.62% (n = 3) for gradients prepared with bSMase-treated PBMC (NS).
Twenty microliters of each gradient fraction was dotted onto Immobilon using a Hybri-Dot Manifold apparatus (Bethesda Research Laboratory, Cambridge, U.K.). Membranes were rinsed with distilled water and blocked with 5% serum albumin in TBS Tween 20 for 1 h. After appropriate washes, membranes were incubated with HRP-conjugated cholera toxin and developed with ECL reagent as described previously (5). The luminograms were quantitated using cooled digital charge-coupled device camera system (ImageMaster VDS-CL; Amersham Biosciences) and ImageQuant software. Results are expressed relative to the sum of the intensity of the various spots taken as 100.
Lck and PLD Western blotting
For Lck determination, proteins from 20 μl of gradient fractions were separated on SDS-PAGE and electrotransferred on Immobilon P membrane. Lck immunoreactive proteins were visualized using ECL detection kit and x-ray film autoradiography. For PLD1 Western blotting, proteins from 700 μl of each fraction were precipitated with ice-cold acetone, and pellets were dissolved in 80 μl of Laemmli buffer containing 4 M urea. Proteins were separated on 8% acrylamide gel in 4 M urea. PLD1 was detected with PLD1 antiserum (dilution 1/2000) prepared by Dr. S. Bourgoin (Laval University, Quebec, Canada). The luminograms were quantitated as described for GM1 assay. Results are expressed relative to the sum of the intensity of the various bands taken as 100.
Cell viability and proliferation
PBMCs were cultured in microtiter culture plates at 2 × 105 cells/well in a final volume of 200 μl of RPMI 1640 supplemented with 2 mM glutamine, 100 μg/ml streptomycin, 100 U/ml penicillin, and 10% decomplemented FCS (Invitrogen Life Technologies) in the absence or in the presence of MBCD (10, 15, 20, and 40 mM final concentration), filipin (1, 2, and 5 μg/ml) or bSMase (1, 5, 10, 100, and 1000 mU/ml final concentration). Stock solutions of filipin were prepared in DMSO and then appropriately diluted to a final DMSO concentration of 1%. Corresponding controls were made with 1% DMSO. At the initiation of the culture 5 μg of Con A/ml were added to the cell suspensions or cells were incubated in the same conditions without mitogen. Cultures were incubated at 37°C in an air-CO2 (95:5) atmosphere. After a 96-h incubation, cell proliferation was measured by the MTT colorimetric assay (Roche Diagnostics) as described previously by Mosmann (22). Briefly, 10 μl of the MTT labeling reagent (0.5 mg/ml final concentration) was added to each well. Cells were incubated further for 4 h and then solubilized according to the Cell Proliferation Kit I recommendations. The number of viable cells was directly correlated to the difference of absorbance measured at 550 and 690 nm. Cell viability was normalized relative to controls (incubated without drug and Con A). Cell proliferation was measured as the difference of absorbance between assays with and without Con A and normalized relative to controls (initially incubated without drug) taken as 100.
Jurkat cell culture and transient transfection
For stock culture, Jurkat T cells were grown in RPMI 1640 medium supplemented with 10% FCS, 2 mM glutamine, 50 μg/ml penicillin-streptomycin, and 20 mM HEPES at 37°C in a 95% air-5% CO2 humidified atmosphere. The medium was changed every 48 h. For transfection experiments, Jurkat cells were suspended in complete culture medium at the concentration of 2 × 106 cells/ml. The hPLD1b- and hPLD2-carrying pCDNA3 plasmids were provided by Dr. M. Record (Institut National de la Santé et de la Recherche Médicale Unité 563, Toulouse, France). The GFP-PLD1b and GFP-PLD2 constructs were prepared by inserting the hPLD1b and hPLD2 coding sequences at the SalI site, and EcoRI and SalI sites, respectively, of the pEGFP-C1 vector polylinker (BD Clontech). Four-micrograms of pEGFP-PLD1b DNA or 4 μg of pEGFP-PLD2 DNA was mixed with diluted Fugene6 (Roche Applied Science) (94 μl serum-free medium + 6 μl of Fugene) and left in contact for 30 min. The mix was then added dropwise to 2 × 106 Jurkat cells suspended in 1 ml of complete culture medium. The cells were plated (5 × 105cells/well) and cultured for 48 h in complete culture medium before RT-PCR and proliferation analyses. Mock-transfected Jurkat cells were transfected with 4 μg of pEGFP DNA (empty vector) in the same experimental conditions.
RNA isolation and semiquantitative RT-PCR analyses
Total RNA was isolated from control, mock-transfected, PLD1-transfected, or PLD2-transfected Jurkat cells using Tri Reagent according to the manufacturer’s instructions. Primers for the amplification of PLD1a, PLD1b, and PLD2 were designed on the basis of published human sequences as described in Ref.5 (expected sizes of amplified fragments: 446, 332 and 329 bp, respectively). For IL-2, the sense and antisense primers were 5′-CACTAAGTCTTGCACTTGTCAC-3′ and 5′-CCTTCTTGGGCATGTAAAACT-3′, respectively (expected size of amplified fragment: 186 bp). β-Actin primers were used as internal control to normalize the data. Reaction products were electrophoresed on 1% agarose gel impregnated with ethidium bromide and visualized by UV transillumination.
Jurkat cell spontaneous proliferation
Control Jurkat cells and Jurkat cells transfected with the empty vector, the PLD1b cDNA or the PLD2 cDNA were suspended at a concentration of 5 × 105 cells/ml in RPMI 1640 medium supplemented with 0.1% FCS, 2 mM glutamine, 50 μg/ml penicillin-streptomycin, and 20 mM HEPES, seeded in 96-well plates (5 × 104 cells/well), and cultured at 37°C in a 95% air-5% CO2 humidified atmosphere for 24 and 48 h. At the end of the incubation period, 10 μl of MTT was added, cells were incubated further for 4 h and then solubilized as described above.
Values are presented as means ± SE of n independent experiments. All data were compared by ANOVA (Statview II for Macintosh) followed by protected t test. A value of p ≤ 0.05 was considered statistically significant.
Effect of lipid raft disruption on PLD activity of human PBMCs
We have shown previously that in human PBMCs a substantial part of PLD1 was associated to detergent-insoluble lipid rafts and that translocation to detergent-soluble nonraft membranes was accompanied by an increase in activity (5). To ascertain whether PLD1 association to lipid rafts can regulate its activity, we studied the effect of lipid raft disruption on PLD activity in intact cells. Incubation of PBMCs with increasing amounts of MBCD for 30 min markedly stimulated PLD activity (Fig. 1,A) while decreasing the cholesterol content of cell membranes (Fig. 1,B). Interestingly, MBCD concentrations, which were effective to induce a significant 2- to 3-fold increase of PLD activity, also significantly reduced cholesterol level by 40–50%. Similarly, incubation of PBMCs with 2 or 5 μg/ml of the cholesterol sequestering agent filipin (Fig. 1,C) significantly activated PLD (1.9- and 2.2-fold, respectively). Thus, these results indicate that pharmacological disruption of lipid rafts by agents reducing or sequestering cholesterol was able to stimulate lymphocyte PLD activity. Besides cholesterol, sphingomyelin is a major component of raft domains (16). We reasoned that treating cells with bSMase would alter raft structure and modify the pattern of associated proteins. Incubation of PBMCs with increasing amounts of bSMase strongly stimulated PLD activity (Fig. 1 D). A highly significant 4-fold increase was observed with SMase concentrations as low as 5 mU/ml, and a maximal 6.7-fold increase was obtained with 10 mU/ml. Thus, various treatments able to modify lipid composition of rafts induced PLD activation, possibly through disorganization of these structures and exclusion of PLD1.
Effect of bSMase on PBMC microdomains
To determine whether PLD activation induced by lipid modification of rafts was due to PLD1 translocation to nonraft membranes, we characterized in detail the effects of bSMase on lymphocyte microdomains. After fractionation of Triton X-100-treated PBMC lysates on a discontinuous sucrose density gradient, most part of the ganglioside GM1 was recovered in the low-density fractions isolated between 13 and 22% sucrose (Fig. 2,A). These detergent-insoluble membrane domains were also enriched in cholesterol (Fig. 2,C), showing that these fractions had the typical lipid composition of rafts. The treatment of PBMCs with 10 mU/ml bSMase, the lowest concentration, which gave maximal PLD increase before Triton X-100 extraction and membrane fractionation, did not modify the distribution of GM1 throughout the gradient (Fig. 2 B). The only observed modification was a significant decrease of the cholesterol content in the raft fraction nos. 6–8 (14.1 ± 0.7 μg/ml, 31.5% of total, in control PBMCs vs 9.5 ± 1.6 μg/ml, 22% of total, in bSMase-treated PBMCs, n = 3, p = 0.03). These results show that, at the concentration used, bSMase only induced a partial disruption of lipid rafts.
In gradients prepared from control PBMCs, most part of ceramide (∼58%) was recovered in the low-density fractions corresponding to lipid rafts (Fig. 3,A). This observation is in line with that reported for some caveolae containing cells (23) After bSMase treatment of the cells, the overall ceramide content of the gradient fractions was increased markedly (Fig. 3 B). It is noteworthy that ceramide level increased more in the raft fractions (∼7-fold) than in nonraft membranes (∼3.5-fold).
Src family kinases such as Lck are known to be associated with the cytoplasmic layer of rafts (16). In control resting PBMCs, a substantial part of the Lck protein (26.3 ± 1.8%) was associated with the raft fractions, as shown by Western blotting experiments (Fig. 4,A). Exposure of the cells to 10 mU/ml bSMase (Fig. 4 B) for 30 min before cell fractionation induced a significant shift of Lck toward the dense fractions of the gradient, with 7.9 ± 2.0% only remaining associated with rafts (p = 0.002, n = 3). These results show that, as it has been reported for cholesterol depleting agents (24, 25), the partial disorganization of lipid rafts induced by low bSMase concentrations was able to release the protein tyrosine kinase Lck out of the rafts.
As shown in Fig. 5,A, a substantial part of PLD1 protein was associated with the detergent-insoluble membranes of control PBMCs. It is noteworthy that the pooled raft fractions (6, 7, 8) contained 18 times less total proteins (0.34 mg/ml) than the dense fractions (6.13 mg/ml), which indicates a high enrichment of PLD1 in the light fractions. When PBMCs were treated with 10 mU/ml bSMase, the PLD1 distribution pattern was drastically modified (Fig. 5 B), only 5.9 ± 3.4% of the protein remaining in the detergent insoluble fractions vs 28.9 ± 6.5% in control PBMCs (p = 0.02, n = 4). Taken together, these results show that a mild perturbation of lipid rafts due to localized hydrolysis of sphingomyelin and ceramide accumulation is sufficient to release Lck and PLD1 out of these structures.
Effect of raft disruption on cell viability and proliferation
To investigate whether raft disrupting treatments, in addition to inducing activation of PLD, interfere with the lymphoproliferative response, we measured cell proliferation in response to the mitogenic lectin Con A using the tetrazolium salt-formazan colorimetric method and in parallel checked the effects of the treatments on cell viability (Fig. 6). As shown in Fig. 6,A, MBCD treatment of the cells inhibited the lymphocyte response to mitogen by >80%, whereas cell viability was only decreased by 22% at 15 mM and up to 45% at 40 mM. These results indicate that the antiproliferative effect of MBCD cannot be totally accounted for by cytotoxicity. Similar observations could be made for filipin (Fig. 6,B), with an inhibition of the proliferative response of >80% and only 25% decrease of cell viability at a 2 μg/ml dose. Concerning bSMase treatment (Fig. 6 C), it proved able to inhibit cell proliferation within a concentration range of 1–100 mU/ml, which was totally devoid of cytotoxicity, and cell viability being significantly decreased only at the highest used concentration of 1 U/ml. Interestingly, the inhibition of the mitogen-induced proliferation by exogenous bSMase was clearly dose dependent, with an IC50 ∼10 mU/ml, a concentration devoid of cytotoxic effects, and that gave maximal PLD stimulation in intact PBMCs. Altogether these results show that in treatment conditions that induced PLD activation, lymphocyte proliferation was markedly hampered. In contrast, addition of exogenous C2- or C6-ceramides (1–10 μM) did not modify the lymphocyte response (data not shown).
Effect of transient PLD1 overexpression on Jurkat T cell activation
RT-PCR experiments showed that the leukemic Jurkat T cells mainly expressed PLD2 and very little, if any, PLD1 (Fig. 7,A). To test the hypothesis that PLD1 plays a role in the control of lymphoid cell response, we transfected Jurkat cells with a plasmid vector encoding a GFP-PLD1b fusion protein and used PLD2-transfected cells as control. Both GFP-PLD1b and GFP-PLD2 were expressed efficiently in Jurkat cells, as evidenced by RT-PCR (Fig. 7, B and C), and by fluorescence microscopy examination of the live cells (Fig. 7, D and E). Furthermore, GFP-PLD1-transfected cells exhibited a 4-fold higher PLD activity than cells transfected with the empty vector (Fig. 7,F), showing that the expressed GFP-PLD1 fusion protein was catalytically active. The activity of GFP-PLD2 transfected cells was ∼2-fold increased. To investigate the effect of PLD overexpression on Jurkat cell spontaneous proliferation, control cells, PLD1- and PLD2-transfected cells, and cells transfected with the empty vector were cultured for 24 and 48 h without effectors. Interestingly, the number of control, empty vector- and PLD2-transfected cells was markedly increased by 90, 76, and 89%, respectively, between hour 24 and hour 48 of culture, whereas only a small increase (24%) was noticed in PLD1-transfected cells (Fig. 7,G). These experiments strongly suggest that PLD1 conveys antiproliferative signals in these cells. An early response of Jurkat cells to activation by PMA and ionomycin is the expression of IL-2 mRNA, which starts after 1 h of treatment and reaches a maximum after 2 h of stimulation (26). Thus, we examined by RT-PCR IL-2 mRNA expression after 2 h of PMA and ionomycin treatment of empty vector-, PLD1-, and PLD2-transfected Jurkat cells. Results of these experiments indicated that PLD1 overexpression markedly lowered (−43%) the level of IL-2 mRNA transcripts (Fig. 7, H and I), whereas PLD2 overexpression did not lower IL-2 mRNA and even increased it (+40%). PLD1 thus seems to specifically decrease T cell activation and proliferation.
In the present study, we show for the first time a direct activation of basal PLD activity following treatment of intact unstimulated cells with agents able to alter lipid raft integrity, such as cholesterol binding agents or exogenous bacterial sphingomyelinase. A stimulation of agonist-induced PLD activity by MBCD or filipin has already been described in neuroblastoma M22 cells stimulated by 12-O-tetradecanoylphorbol-13-acetate (27) and in human neutrophils stimulated by fMLP (28). Although PLD activation by MBCD and filipin could be due to raft disorganization, we cannot exclude other raft-independent effects because these two agents are known to act on both raft and nonraft cholesterol (29).
Another way to alter the cohesion of lipid raft is to partly digest membrane sphingomyelin, which is mainly (up to 70% of total) found in lipid rafts (30). The strong PLD activation that we observed after treatment of human PBMCs with bSMase confirms further that the pharmacological disruption of lipid rafts can efficiently stimulate PLD in these cells. The stimulating effect was already significant with bSMase concentrations as low as 5 mU/ml, a maximal 6- to 7-fold increase being observed starting from 10 mU/ml. Conflicting results concerning the effect of bSMase on PLD activity have been reported, depending on the cell model considered. Thus, 100 mU/ml bSMase has been shown to modestly increase (50%) the basal PLD activity of fibroblasts (31) and rabbit cortical collecting duct cells (32), whereas it had no effect on the bradykinin-stimulated PLD activity in the same cells. On the opposite, bSMase used in similar experimental conditions completely abolished EGF- or TPA-induced PLD activation in A-431 cells while inducing a large increase in ceramide level (33). In the above-mentioned studies, the variations of PLD activity induced by exogenous bSMase have been attributed to a rise in ceramide level. However, we did not observe any effect of exogenous short-chain ceramides or of ceramide metabolism inhibitors (results not shown), on PLD activity, which supports a different mechanism of bSMase action. The ceramide measurements that we performed after fractionation of Triton X-100-treated PBMC lysates on sucrose gradient showed that ceramide production induced by 10 mU/ml bSMase mainly occurred in the detergent-insoluble fractions rather than in nonraft membranes. This SMase concentration did not induce a complete disaggregation of rafts because the distribution pattern of GM1 along the gradient fractions was not altered as compared with control PBMCs. SMase treatment rather induced changes in the properties of the detergent insoluble domains due to the conversion of sphingomyelin-rich to ceramide-rich rafts (34). Indeed, we observed a significant (32%) decrease in the cholesterol content of the raft fractions. Interestingly, Ito et al. (35) have reported a similar decrease in cholesterol of the detergent-resistant membrane fractions from rat astrocytes treated with bSMase. The present results are also in good agreement with those of London and London (36), showing that ceramides specifically displace cholesterol from lipid rafts while other lipid constituents remain raft-associated. In addition, we observed that bSMase treatment of PBMCs induced marked modifications at the level of raft-associated proteins, both Lck tyrosine kinase and PLD1 proteins being largely displaced from rafts. These results are in line with those of our previous study showing that destabilization of rafts by DHA enrichment of PBMCs also displaced these proteins toward the dense detergent soluble membranes (5). It is noteworthy that PLD1 exclusion from rafts correlates with a marked increase of PLD activity in intact cells whatever the agent used to disturb raft structures. These findings reinforce our hypothesis that the specialized microenvironment of rafts exerts a negative control on PLD1 activity. A similar negative control has been described for Lck tyrosine kinase which exhibits a much lower phosphotransferase activity in raft membranes than in the bulk of detergent-soluble membranes (25). It can be concluded that PLD activation by bSMase treatment of PBMCs is probably not due to a direct effect of ceramide but rather to changes in the physicochemical properties of the rafts resulting from the conversion of sphingomyelin to ceramide, inducing PLD1 exclusion and relocation to a different environment.
Several hypotheses can be put forward to explain that PLD1 activity was stimulated following its exclusion from lipid rafts. First, PLD1 can conceivably be maintained inactive in rafts due to interactions with raft resident proteins, as it has been demonstrated for PLD1-caveolin-1 interactions in caveolae (37). Raft exclusion then would release the inhibitory constraints. Second, PLD1 can be maintained inactive due to low availability of specific activators in rafts. It has been proposed that the myristoylated alanine-rich C kinase substrate (MARCKS), which resides in lipid rafts, diminishes the availability of PIP2 due to its propensity to bind PIP2 with a high affinity (38). It is noteworthy that MARCKS inhibits the PLC-catalyzed PIP2 hydrolysis at physiological concentrations (39). Thus, PLD maintained in rafts would be inactive due to low PIP2 availability. After exit from raft structures and relocation to nonraft membrane, which also contains PIP2, PLD would have free access to membrane PIP2 and thus become activated. Interestingly, the treatment of M22 neuroblastoma cells overexpressing MARCKS by cyclodextrin and filipin increased the PMA-stimulated PLD activity (27). Rafts are also known to be connected to the actin cytoskeleton. Actin and several actin binding proteins are potent inhibitors of PLD either directly by protein to protein contact or indirectly by sequestering PIP2 (40).
We and others have shown previously (14, 41) that DHA enrichment of human lymphocytes results in a marked decrease of their proliferative response to mitogens. In the present study, we show that raft disrupting agents also inhibit lymphocyte proliferation. Although MBCD and filipin are known to be cytotoxic, we observed strong antiproliferative effects at concentrations inducing limited cytotoxicity (≤25%) and marked PLD activation. Conflicting results concerning the effects of cholesterol sequestering agents on lymphocyte functions have been reported. Although there is agreement that both filipin and MBCD inhibit TCR-induced calcium mobilization (25, 42), the two cited groups have found opposite results relative to tyrosine phosphorylation, which is a key step in lymphocyte response. It is noteworthy that besides its raft disrupting effects, MBCD also induces non specific depletion of intracellular Ca2+ stores and plasma membrane depolarization, which might contribute to inhibit T cells responses independently of raft alteration (43, 44). Interestingly, bSMase inhibited cell proliferation without any cytotoxicity. At the lowest dose inducing maximal PLD activation, the proliferative response was inhibited by 51%, whereas viability was not affected. This inhibitory effect was probably independent of ceramide production per se because the addition of exogenous permeant ceramides did not impair the lymphocyte response, which is in line with literature data showing positive effects of ceramide on the proliferative response of murine splenic lymphocytes (45).
The fact that PLD activity was stimulated following raft disruption and that raft disruption inhibited lymphocyte activation does not necessarily mean that PLD activation is responsible for the inhibition of lymphocyte responses. Indeed, it is now well documented that the integrity of lipid rafts is required for correct lymphocyte activation and that raft lipid disruption inhibits lymphocyte responses through changes in the distribution of several critical proteins, among which the protein tyrosine kinase Lck (46). However, the present results showing that PLD1 overexpression significantly decreased IL-2 mRNA expression in activated Jurkat cells strongly support the hypothesis that PLD1 activation is one of the factors contributing to lymphocyte inhibition following lipid raft disruption. In contrast to PLD1, overexpression of the PLD2 isoform did not inhibit, but rather enhanced, IL-2 mRNA expression, showing that PLD1 effect was specific. There are several examples of differences in function of PLD1 and PLD2 in a given cell type (47, 48), these differences being possibly connected with their distinct subcellular location (2).
Decreased lymphocyte responses associated with PLD activation has also been reported for the antiproliferative cytokine IFN-β, which has been shown to decrease IL-2 production by Con A-activated PBMCs while inducing a sustained increase in phosphatidic acid level (49, 50). How PLD1 activation inhibits lymphocyte proliferation remains speculative. It may be hypothesized that the molecular species of second messengers PA and DAG generated by PLD through PC hydrolysis have reduced signaling properties, as compared with the more unsaturated ones issued from PIP2 hydrolysis (51). PA may also have specific intracellular targets involved in the inhibition of the proliferative response. Among the different enzymatic targets of phosphatidic acid, the protein tyrosine phosphatase Src homology region 2 domain-containing phosphatase 1 has been shown to bind PA with a high affinity, this interaction resulting in the stimulation of its phosphatase activity (52). Thus, it can be assumed that in lymphocytes with elevated PLD activity, Src homology region 2 domain-containing phosphatase 1, which is well recognized as a negative regulator of T cell function (53), could be activated. Another attractive hypothesis ensues from the recent observations that actin depolymerization must occur at the onset of immune response to allow the clustering of lipid rafts and the formation of the immune synapse (54). Indeed, blocking actin depolymerization results in an inhibition of lymphocyte activation. Because PLD activation is well known to induce actin polymerization, it may be supposed that increased PLD activity will counteract the initial depolymerization and then compromise cell response. Another possibility is that high PA level confers to lymphocytes a “preactivated state” leading to anergy or apoptosis upon subsequent mitogenic activation (55).
Although the precise mechanisms involved remain to be defined, our whole results support the hypothesis that in lymphoid cells PLD1 activation impairs the transduction of mitogenic signals.
We thank Dr. Sylvain Bourgoin (Laval University, Quebec, Canada) for providing hPLD1 and hPLD2 antisera, and Dr. Michel Record (Institut National de la Santé et de la Recherche Médicale Unité 563, Toulouse, France) for sharing plasmids carrying PLD cDNAs.
The authors have no financial conflict of interest.
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.
This work was supported by INSERM.
Abbreviations used in this paper: PLD, phospholipase D; PA, phosphatidic acid; PC, phosphatidylcholine; DAG, diacylglycerol; PIP2, phosphatidylinositol 4,5-bisphosphate; DHA, docosahexaenoic acid; MBCD, methyl-β-cyclodextrin; bSMase, bacterial sphingomyelinase; MARCKS, myristoylated alanine-rich C kinase substrate.