Sphingosine 1-phosphate induces platelet activation through an extracellular action and shares a platelet surface receptor with lysophosphatidic acid.

Sphingosine 1-phosphate (Sph-1-P) has been implicated as an intracellular second messenger in many studies. We investigated the metabolism of Sph-1-P and the mechanism by which Sph-1-P induces activation in enucleated and highly differentiated platelets. Platelets lack Sph-1-P lyase activity, possess persistently active sphingosine (Sph) kinase, and abundantly store Sph-1-P. Although exogenous Sph-1-P activated platelets, intracellular Sph-1-P, formed from exogenously added Sph by cytosolic Sph kinase, failed to do so. To support the notion that exogenous Sph-1-P stimulates platelets from outside, contact of platelet surfaces with immobilized Sph-1-P covalently linked to glass particles resulted in platelet activation. Furthermore, we detected the specific binding sites for radiolabeled Sph-1-P on the platelet surface, suggesting extracellular effects of Sph-1-P on plasma membrane receptors. This specific Sph-1-P binding was inhibited not by other sphingolipids but by lysophosphatidic acid (LPA), and platelet aggregation response to LPA was specifically desensitized by prior addition of Sph-1-P. Finally, internally stored Sph-1-P is released extracellularly upon stimulation, and the release correlated well with protein kinase C activation in intact platelets. These results suggest that Sph-1-P acts not intracellularly but intercellularly, following discharge from activated platelets, and shares a platelet surface receptor with LPA.

Preparation of Platelets-Citrated venous blood was obtained from healthy adult volunteers. Washed platelets were prepared and handled as described previously (16), unless stated otherwise. Bovine serum albumin (fatty acid-free) (1%) was added when indicated. For Sph-1-P lyase activity assay, outdated platelet concentrates obtained from the Oregon Red Cross (Portland, OR) were used.
Sph-1-P Lyase Activity Assay-Sph-1-P lyase activity was measured as described previously (6) except that [3-3 H]Sph-1-P instead of [4, H]dihydrosphingosine-1-phosphate was used as a substrate. The reaction products were applied to silica gel 60 HPTLC plates (Merck, Darmstadt, Germany), and the plates were developed in chloroform: methanol:acetic acid (50:50:1). In this polar solvent, all radioactive lyase metabolites (palmitaldehyde, palmitic acid, and palmitol) ran closely together near the front, and the whole region was scraped off and counted by liquid scintillation counting.

Metabolism of [ 3 H]Sph in
Platelets-Platelet suspensions (0.5 ml) were incubated with 1 M [ 3 H]Sph (0.2 Ci). At the indicated time points, the reaction was terminated by the addition of 1.875 ml of ice-cold chloroform:methanol:concentrated HCl (100:200:1), and lipids were extracted from the cell suspensions and analyzed for [ 3 H]Sph metabolism as described previously (16). Portions of lipids obtained from the lower chloroform phase were applied to silica gel 60 HPTLC plates, and the plates were developed in butanol:acetic acid:water (3:1: 1), followed by autoradiography. Silica gel areas containing radiolabeled sphingolipids were scraped off and counted by liquid scintillation counting.
Quantitative Measurement of Sph and Sph-1-P-Sph (31) and Sph-1-P (32)  Platelet Shape Change-Platelet shape change and aggregation were determined turbidometrically (33), as described (16). Calibration was performed with zero light transmission defined for platelet suspension and 100% transmission for the buffer.
Measurement of Intracellular Ca 2ϩ Concentration-Intracellular Ca 2ϩ concentration was measured using the Ca 2ϩ -sensitive fluorophore Fura-2 as described (16). Values of peak increases after addition of an agent were quantified.
Preparation and Application of Sph-1-P or Sph Immobilized on a Solid Support-Sph or Sph-1-P was immobilized on the surface of controlled-pore glass particles with a long chain alkylamine (CPG, Inc., Fairfield, NJ). Sph-1-P or Sph derivative possessing -carboxyl group was synthesized and linked to a long chain alkylamine of the glass particles through an amide linkage, as described previously (34). The long chain alkylamine acts as a spacer arm and avoids low steric availability of Sph-1-P or Sph immobilized on glass, enabling them to mimic parent (free) Sph-1-P or Sph. For control glass particles, the amino group was blocked by N-acetylation. Sph-1-P and Sph contents of the conjugates were 22 and 39 mol/g of dry glass particles, respectively. Extensive washing of the Sph-1-P-and Sph-attached glass particles with chloroform:methanol (1:2) did not result in dissolution of free lipids, which was confirmed by thin layer chromatography. The glass particles were added into platelet suspensions under constant stirring at 1000 rpm, with the Sph-1-P or Sph concentration at 40 M, followed by scanning electron microscopy (16) and measurement of intracellular Ca 2ϩ concentration.
[ 3 H]Sph-1-P Binding Assays-Binding assays were performed by incubating intact platelets (1 ϫ 10 8 cells), suspended in phosphatebuffered saline, with 2.1 nM [ 3 H]Sph-1-P. Reactions were initiated by the addition of the ligand and were incubated at 4°C for 1 h unless stated otherwise. The platelets were then washed with 1 mg/ml of bovine serum albumin, 150 mM NaCl, and 50 mM Tris-HCl, pH 7.5, three times. Radioactivity of the platelets was counted by liquid scintillation counting. Under those conditions, [ 3 H]Sph-1-P added remained intact, which was confirmed by separation with thin layer chromatography.
To examine the effect of a protease on [ 3 H]Sph-1-P binding, platelets were treated with 0.02% of the protease (type XXV) (Sigma) at room temperature for 30 min, and assays were carried out as described above.
Total binding was defined as the amount of radioactivity bound to platelets in the absence of competing ligand. Nonspecific binding was defined as the amount of [ 3 H]Sph-1-P binding that occurred in the presence of 50 M Sph-1-P. Specific binding was defined as the difference between total and nonspecific binding.

Sph-1-P Release from Platelets Labeled with [ 3 H]Sph-[ 3 H]
Sph-labeled platelet suspensions, to which 1% bovine serum albumin had been added, were centrifuged for 15 s at 12,000 ϫ g. Lipids were then extracted from the resultant medium supernatant and cell pellet and analyzed as described above. Albumin was included in the medium to prevent released Sph-1-P, a lipophilic molecule, from being nonspecifically attached to the plasma membrane surface and consequently underestimated. The percentage of Sph-1-P release into medium was calculated as 100 ϫ ([ 3 H]Sph-1-P in medium)/(total [ 3 H]Sph-1-P in medium plus cell pellet).
Protein Kinase C Activation in Intact Platelets-Protein kinase C activation in intact platelets was evaluated by pleckstrin (47-kDa protein) (35) phosphorylation as described (36).

Absence of Sph-1-P Lyase Activity in Platelets-When
[ 3 H]Sph was added into platelet homogenates under established assay conditions for Sph-1-P lyase (6), no lyase products (palmitaldehyde, palmitic acid, and palmitol) were formed. Under the same conditions, high lyase activity was detected in bovine (specific activity, 11.2 Ϯ 0.2 pmol/min⅐mg of protein, mean Ϯ S.D., n ϭ 3) and rat livers, as reported previously (6). Low but significant Sph-1-P lyase activity was detected in CMK and K562 cells (specific activities, 0.25 Ϯ 0.04 and 0.54 Ϯ 0.03 pmol/min⅐mg of protein, mean Ϯ S.D., n ϭ 3, respectively). CMK is a human megakaryoblastic cell line (37); K562 is a human erythroleukemia cell line capable of megakaryocytic differentiation (38). Our results confirm the absence of Sph-1-P lyase in platelets, as suggested by earlier studies (39,40). It is likely that blood stem cells lose Sph-1-P lyase activity during late-stage differentiation into platelets.
Independence of Sph Kinase Activity from Cell Activation in Platelets-When 1 M [ 3 H]Sph was added exogenously to platelet suspensions, the label was efficiently removed from the medium; uptake of [ 3 H]Sph was 96% at 5 min after the label addition. [ 3 H]Sph incorporated into platelets was rapidly converted to [ 3 H]Sph-1-P (Fig. 1A). Conversion was observed as early as at 10 s. Within 5 min, the level of Sph-1-P approached a plateau, with 45% conversion of [ 3 H]Sph to [ 3 H]Sph-1-P. At 3 h (the latest time point observed), 42% of the [ 3 H]Sph originally added remained as [ 3 H]Sph-1-P (Fig. 1A), indicating the stability of Sph-1-P in platelets. Sph at concentrations below 10 M did not activate platelets at all (Fig. 2B). Sph kinase, which converts [ 3 H]Sph into [ 3 H]Sph-1-P, is therefore considered to be present in platelets as an active enzyme under resting conditions. [ 3 H]Sph was also converted by N-acylation into Cer at later time points (Fig. 1A); 3, 7, and 10% of added [ 3 H]Sph was converted into Cer at 1, 2, and 3 h, respectively. The ratio of [ 3 H]Sph-1-P to [ 3 H]Sph increased rapidly to approximately 5:1 at 20 min and stayed relatively constant at this value until 3 h. This suggests that the cellular content of Sph-1-P is much higher than that of Sph in resting platelets.
The rapid conversion of [ 3 H]Sph into [ 3 H]Sph-1-P was not affected by the potent platelet activator thrombin (41) (Fig. 1B), which, under similar conditions, strongly induced platelet activation, including intracellular Ca 2ϩ mobilization and protein kinase C activation (data not shown). Other platelet activators (collagen, TPA, and epinephrine) and the platelet activation inhibitor prostaglandin E 1 (41) also had no effect on the [ 3 H]Sph conversion (data not shown). N,N-Dimethylsphingosine, which strongly inhibits Sph kinase, suppressed conversion of [ 3 H]Sph to [ 3 H]Sph-1-P (Fig. 1B), as we reported previously (42).
In summary, Sph kinase is highly active, even in resting platelets, and its activity is independent of intracellular events involved in platelet activation.

Determination of Sph and Sph-1-P Levels in Platelets-
We determined the mass levels of Sph-1-P and Sph extracted from platelets. The content of Sph-1-P and Sph was calculated as 1.41 Ϯ 0.04 nmol and 374 Ϯ 61 pmol (mean Ϯ S.D., n ϭ 3 and 5), respectively, per 10 9 platelets. The finding that the cellular content of Sph-1-P is much higher than that of Sph is consistent with the high Sph-1-P:Sph ratio obtained in [ 3 H]Sph labeling studies (Fig. 1A).
Platelet Activation Induced by Sph-1-P Immobilized on a Solid Support-The finding that only exogenous Sph-1-P activates platelets raises the possibility that Sph-1-P acts on these highly differentiated cells from outside. To test this possibility, we synthesized Sph-1-P immobilized on a solid support; a Sph-1-P derivative possessing an -carboxyl group was covalently linked to alkylamine-glass particles through amide linkage. Upon the addition of the immobilized Sph-1-P, platelets underwent shape change and aggregate formation, as determined by scanning electron microscopy (Fig. 3, C and D), and intracellular Ca 2ϩ mobilization (Fig. 4). Sph-bound glass particles (Fig.  3, A and B, and Fig. 4) or control glass particles (data not shown) had no effect on platelets. The fact that immobilized Sph-1-P mimics free Sph-1-P in terms of activating platelets strongly suggests that the site of Sph-1-P action resides not inside platelets but on the surface. This hypothesis would explain why exogenous addition of Sph-1-P but not Sph results in platelet activation in spite of rapid Sph conversion into Sph-1-P (which occurs intracellularly). The hypothesis is consistent with the observation that extracellular, but not intracellular (by microinjection), addition of Sph-1-P elicits intracellular Ca 2ϩ mobilization in Xenopus laevis oocytes (44).
[ 3 H]Sph-1-P Binding Studies-The above finding that Sph-1-P only acts from the extracellular face of platelet surface membranes can be best explained by the hypothesis that exogenous Sph-1-P acts on platelets via interaction with a plasma membrane receptor. Consequently, we performed [ 3 H]Sph-1-P binding studies using intact platelets. Fig. 5A shows the time course of specific binding of [ 3 H]Sph-1-P to platelets. The specific binding was time dependent, reached equilibrium by 1 h, and remained constant for at least 1 h. Saturation binding experiments were performed with [ 3 H]Sph-1-P (Fig. 5B). The specific binding was saturated around 2 nM of the ligand when platelets were incubated at 4°C for 1 h with or without cold 50 M Sph-1-P. The data were transformed by Scatchard analysis (Fig. 5C). Platelets were found to possess two binding sites for Sph-1-P; the K d values were estimated to be 110 nM and 2.6 M, and the numbers of the binding sites approximately 200/cell and 2400/cell, respectively. We examined the effect of a protease on the specific binding of [ 3 H]Sph-1-P. Treatment of platelets with a protease (type XXV) disrupted the specific binding of [ 3 H]Sph-1-P almost completely (data not shown), indicating that the binding sites are proteins that are located on the surface of platelets. The specific [ 3 H]Sph-1-P binding was not detected on K562 cells, which are capable of megakaryocytic differentiation but do not respond to Sph-1-P in terms of Ca 2ϩ mobilization (data not shown). These results indicate the pres-ence of a cell-surface receptor for Sph-1-P on platelets.
Competition binding experiments were performed using various sphingolipids and LPA (Fig. 6). Sph and C 2 -Cer were unable to compete for [ 3 H]Sph-1-P binding sites on platelets. Sphingosylphosphocholine only slightly inhibited the binding of [ 3 H]Sph-1-P. In contrast, LPA, which is a glycerophospholipid with a very similar structure to Sph-1-P (26), reduced the binding of [ 3 H]Sph-1-P to platelets as much as unlabeled Sph-1-P did.
Desensitization of Platelet Aggregation Response to LPA by Sph-1-P-LPA is a well established platelet agonist capable of inducing strong and irreversible aggregation response (45). Sph-1-P is also a platelet-aggregating agent, although its effect is weaker (16). Because the possibility that Sph-1-P and LPA may share a surface receptor on platelets was raised, desensitization studies in platelet aggregation were performed (Fig. 7). LPA (5 M) induced marked platelet aggregation, which was comparable with that induced by 5 g/ml of collagen (41). Prior addition of subthreshold concentrations of Sph-1-P inhibited the platelet aggregation response to LPA. In contrast, Sph-1-P did not affect the response to collagen. The specific desensitization of LPA-induced platelet aggregation by Sph-1-P might be related to the fact that Sph-1-P and LPA share a platelet surface receptor, although we cannot completely rule out the desensitization mechanism resulting from the modification of intracellular signaling pathways involved.
Sph-1-P Release from Activated Platelets-In view of the fact that Sph-1-P activates platelets from outside but is abundantly stored inside, we examined the possibility that Sph-1-P is released from platelets and acts intercellularly as a local mediator. When platelets were stimulated with 1 M TPA, which can act as a substitute for diacylglycerol and directly activates protein kinase C (41,46), 8,35,53,57, and 60% of stored Sph-1-P was released extracellularly 2, 5, 20, 60, and 120 min after challenge, respectively. This protein kinase C activator was also found to release Sph-1-P in a dose-dependent manner (Fig. 8A). The release correlated well with its effect on protein kinase C activation in intact platelets (Fig. 8B). Under the same conditions, thrombin, which produces diacylglycerol and FIG. 4. Platelet intracellular Ca 2؉ mobilization induced by Sph-1-P immobilized on a solid support. Platelets were challenged with Sph-or Sph-1-P-bound glass particles (arrow), and intracellular Ca 2ϩ concentration was monitored. Sph-1-P-bound particles elicited intracellular Ca 2ϩ mobilization (solid line), whereas Sph-bound particles did not (dotted line).

FIG. 3. Platelet activation induced by Sph-1-P immobilized on a solid support.
A and B, platelets were treated for 1 min with Sph-bound glass particles and examined by scanning electron microscopy. Platelets neither changed their shape nor formed clusters upon challenge with the Sph-bound glass. Note the presence of dispersed platelets and free glass particles without platelet attachment (arrows). C and D, platelets were treated for 1 min with Sph-1-P-bound glass particles and examined by scanning electron microscopy. Platelets underwent shape change with pseudopod formation and formed aggregates on Sph-1-P-bound particles. Note the scarcity of free platelets. Bars: A, B, and C, 10 m; D, 1 m.
activates protein kinase C as a result of phosphatidylinositol-4,5-bisphosphate hydrolysis (41,46), and 1-oleoyl-2-acetyl-glycerol, a membrane-permeable diacylglycerol (46), also caused marked Sph-1-P release (Fig. 9). The Sph-1-P release induced by these protein kinase C activators was inhibited by staurosporine (Fig. 9), an inhibitor of protein kinases, including protein kinase C (47). Furthermore, epinephrine, a weak platelet stimulator incapable of activating protein kinase C by itself (41), did not induce Sph-1-P release (data not shown). These findings indicate that Sph-1-P can be released from activated platelets and suggest that protein kinase C activation may be the mechanism involved.

DISCUSSION
In most cells, Sph-1-P is degraded to ethanolamine phosphate and fatty aldehyde by Sph-1-P lyase (1,5,6). Platelets lack this lyase activity. In contrast, these enucleated cells possess a highly active Sph kinase. Sph kinase is present in platelets as an active enzyme under resting conditions, and its activity is not related to cell activation by physiological agonists. It is not surprising that platelets, which possess high Sph kinase activity and lack Sph-1-P lyase activity, accumulate Sph-1-P abundantly. These findings on Sph-1-P-related metabolism in enucleated platelets contrast with previous findings on nucleated cells (7,8,14,17), in which researchers observed fast and extensive metabolism of Sph-1-P with a relative scarcity of cellular Sph-1-P and hypothesized that Sph kinase is the ratelimiting factor in Sph catabolism. The role of Sph-1-P as a mitogenic signaling molecule has been extensively studied in Swiss 3T3 fibroblasts. In these cells, Sph-1-P has been shown to possess properties that qualify it as an intracellular second messenger (7,8,48). Endogenous Sph-1-P is maintained at a low level, possibly because of degradation by a lyase. Sph kinase is relatively inactive in the resting state (only a small fraction of exogenous Sph is converted to Sph-1-P intracellularly); Sph kinase activity is stimulated and Sph-1-P level is H]Sph were stimulated with various concentrations of TPA for 5 min, and the percentage of Sph-1-P release into the medium was determined as described under "Experimental Procedures." B, 32 P i -loaded platelets were stimulated with 0 (a), 1 (b), 10 (c), 100 (d), or 1000 (e) nM TPA, and protein kinase C activation in intact platelets was evaluated by phosphorylation of pleckstrin, a protein kinase C substrate in platelets (25). The location of pleckstrin is indicated by an arrow.
transiently increased by specific growth factors (platelet-derived growth factor and serum). Recently, similar findings were reported for Fc⑀RI-mediated signal in the rat mast cell line (14). It seems obvious that Sph kinase regulation and hence the functional role of Sph-1-P in nonproliferative, terminally differentiated cells such as platelets differ from those in nucleated cells. The possibility of Sph-1-P playing a pivotal messenger role intracellularly is remote in platelets.
Sph-1-P is a platelet activator (16). The findings that extracellular but not intracellular Sph-1-P is capable of activating platelets, and that immobilized Sph-1-P mimics free Sph-1-P in terms of activating platelets, indicate that the site of Sph-1-P action resides not inside platelets but on the surface. The more convincing evidence for the site of Sph-1-P action being extracellular is our identification here of specific binding sites for [ 3 H]Sph-1-P on platelets, the first demonstration of a specific [ 3 H]Sph-1-P binding site being expressed on the surface of plasma membrane. It has been shown recently that signaling pathways of Sph-1-P are regulated by heterotrimeric GTPbinding proteins (9,13,23,24), whose activation is receptordependent (49). Furthermore, during the course of our present study, it was reported that only exogenously (not intracellularly) added Sph-1-P induces biological responses in guinea pig atrial myocytes (13) and N1E-115 neuronal cells (15). In addition to intracellular actions after passing the plasma membrane, activation of plasma membrane receptor(s) may be a critical mechanism by which Sph-1-P exerts biological responses in various cells.
Another important finding of ours is that the putative Sph-1-P receptor may be shared by LPA, a lysoglycerophospholipid that is similar to Sph-1-P in structure (26), capable of inducing platelet aggregation (45), and released from activated platelets (50). In support of this hypothesis, we found that platelet aggregation response to LPA was specifically desensitized by Sph-1-P. These findings are not consistent with several recent studies reporting lack of cross-desensitization between Sph-1-P and LPA in various nucleated cells (13,15,51). In this content, it is noted that platelets are shown to possess two different levels of binding sites, a high affinity site (K d , 110 nM) and a low affinity site (K d , 2.6 M), and one can assume that the low affinity site might be functional for platelets, judging from the facts that M order of Sph-1-P concentrations are needed to induce platelet shape change and aggregation. Although the molecular mechanism of our observation in platelets remains to be solved, one possibility may be that platelets possess a unique receptor for lysophospholipids, including Sph-1-P and LPA, leading to platelet aggregation. It is already established that platelets store a variety of biologically active molecules that are secreted upon stimulation (52). The secreted molecules interact with other platelets, plasma proteins, and the vessel wall. Sph-1-P was released from platelets, as expected given that Sph-1-P activates platelets from outside, but is abundantly stored inside. The Sph-1-P release may be mediated by protein kinase C, which is also highly expressed in platelets (53). We propose that Sph-1-P be added to the list of bioactive molecules stored in platelets and released from them upon stimulation, although, at present, we do not have the direct evidence for platelet activation caused by released Sph-1-P; the fact that not only Sph-1-P but also more potent lipid mediators such as thromboxane A2 are released from activated platelets makes our trial difficult. Previous reports have suggested that sphingolipid metabolites, including Sph-1-P, constitute a new class of intracellular second messengers in cell growth regulation and signal transduction (1,(3)(4)(5). However, as shown here, a sphingolipid can act intercellularly as a local mediator through its discharge from cells to regulate cellular functions in an autocrine or paracrine fashion.