Sphingosine 1-Phosphate Stimulates Cell Migration through a Gi-coupled Cell Surface Receptor

Sphingosine 1-phosphate (SPP) has been shown to inhibit chemotaxis of a variety of cells, in some cases through intracellular actions, while in others through receptor-mediated effects. Surprisingly, we found that low concentrations of SPP (10–100 nm) increased chemotaxis of HEK293 cells overexpressing the G protein-coupled SPP receptor EDG-1. In agreement with previous findings in human breast cancer cells (Wang, F., Nohara, K., Olivera, O., Thompson, E. W., and Spiegel, S. (1999) Exp. Cell Res. 247, 17–28), SPP, at micromolar concentrations, inhibited chemotaxis of both vector- and EDG-1-overexpressing HEK293 cells. Nanomolar concentrations of SPP also induced a marked increase in chemotaxis of human umbilical vein endothelial cells (HUVEC) and bovine aortic endothelial cells (BAEC), which express the SPP receptors EDG-1 and EDG-3, while higher concentrations of SPP were less effective. Treatment with pertussis toxin, which ADP-ribosylates and inactivates Gi-coupled receptors, blocked SPP-induced chemotaxis. Checkerboard analysis indicated that SPP stimulates both chemotaxis and chemokinesis. Taken together, these data suggest that SPP stimulates cell migration by binding to EDG-1. Similar to SPP, sphinganine 1-phosphate (dihydro-SPP), which also binds to this family of SPP receptors, enhanced chemotaxis; whereas, another structurally related lysophospholipid, lysophosphatidic acid, did not compete with SPP for binding nor did it have significant effects on chemotaxis of endothelial cells. Furthermore, SPP increased proliferation of HUVEC and BAEC in a pertussis toxin-sensitive manner. SPP and dihydro-SPP also stimulated tube formation of BAEC grown on collagen gels (in vitro angiogenesis), and potentiated tube formation induced by basic fibroblast growth factor. Pertussis toxin treatment blocked SPP-, but not bFGF-stimulated in vitro angiogenesis. Our results suggest that SPP may play a role in angiogenesis through binding to endothelial cell Gi-coupled SPP receptors.

The sphingolipid metabolite sphingosine 1-phosphate (SPP) 1 is a bioactive lipid that regulates diverse biological effects and signaling pathways (reviewed in Ref. 1). SPP increases cell proliferation (2,3) and opposes ceramide-mediated apoptosis (4 -7) through an intracellular action (8,9), yet some of its biological effects when added exogenously are due to binding to cell surface receptors. Pertussis toxin-sensitive G proteins are involved in some of the signaling pathways activated by SPP (10 -15), suggesting that it activates a receptor coupled to a G i /G o protein. In agreement, low concentrations of SPP activate G i protein-gated inward rectifying K ϩ channels only when applied at the extracellular face of atrial myocytes (16).
Several reports have demonstrated that SPP inhibits cell motility. SPP inhibited chemotactic motility of mouse melanoma B16, mouse fibroblast BALB/3T3 clone A31, and several tumor cell lines at nanomolar concentrations (17)(18)(19). Moreover, SPP immobilized on glass beads markedly inhibited melanoma cell motility. However, pertussis toxin treatment did not block the effect of SPP, suggesting that in these cells SPP acts through a cell surface receptor, independently of pertussis toxin-sensitive G-proteins (20). In contrast, SPP inhibits chemotaxis of human breast cancer cells only at high (micromolar) concentrations, acting independently of EDG-1 (21).
We have recently identified SPP as a ligand for the G-protein-coupled receptor, endothelial differentiation gene-1 (EDG-1) (22). EDG-1 binds SPP with remarkable specificity and high affinity (K D ϭ 8 nM) (9,22). Binding of SPP to EDG-1 resulted in inhibition of adenylate cyclase and activation of mitogen-activated protein kinase (both G i -mediated), but did not mobilize calcium from internal stores (9,23). In contrast, Okamoto et al. (12) found that in HEL cells overexpressing EDG-1, binding of SPP induced calcium mobilization (12).
Two other related G protein-coupled receptors, EDG-3 and EDG-5, have recently been shown to bind SPP with similar high affinity (14,24), to confer responsiveness to SPP of a serum response element-driven reporter gene when expressed in Jurkat cells, and to allow SPP-stimulated 45 Ca 2ϩ efflux in Xenopus oocytes (25). In agreement, overexpression of EDG-3 in Chinese hamster ovary cells led to phospholipase C activation and calcium mobilization induced by SPP, which was significantly inhibited by pertussis toxin (26). However, low con-centrations of SPP mobilize calcium from internal sources in BAEC in a pertussis toxin-sensitive manner without activation of phospholipase C (27), suggesting the involvement of novel, unidentified signaling pathways in SPP-induced release of intracellular calcium.
Although the biological functions of the EDG family of GPCR are not completely understood, the EDG-1 transcript was originally cloned as an immediate-early gene induced during differentiation of HUVEC, cells of the vessel wall accessible to platelet-derived ligands, into capillary-like tubules (28). Moreover, SPP signaling in HEK293 cells overexpressing EDG-1 leads, by a Rho-dependent mechanism, to formation of a network of cell-cell aggregates resembling the network formation of differentiated endothelial cells and P-cadherin expression (22). Because SPP is stored and released from activated platelets and serum concentrations of SPP are estimated to be approximately 0.5 M (29), about 60 times greater than the K D for binding to EDG-1, we suggested that SPP might play an important role in angiogenesis acting through EDG-1 (22).
Angiogenesis, the process of new vessel formation from preexisting ones, or neovascularization, is a critical event for a variety of physiological processes, such as wound healing, embryonic development, corpus luteum formation, and menstruation. However, angiogenesis can be activated in response to tissue damage and is important in certain pathological conditions such as tumor growth and metastasis, rheumatoid arthritis, diabetic retinopathy, psoriasis, and cardiovascular diseases (30). Conversely, in states of inadequate tissue perfusion, such as myocardial or limb ischemia, enhanced angiogenesis is essential and beneficial (31). Endothelial cell migration and formation of new capillary tubes are required events in the angiogenic response. In this study, we investigated the potential role of SPP in angiogenesis by examining regulation of endothelial cell motility, proliferation, and tube formation through SPP receptors.
SPP Binding Assay-HUVEC or BAEC were washed with binding buffer (20 mM Tris-HCl, pH 7.4, 100 mM NaCl, 15 mM sodium fluoride, 2 mM deoxypyridoxine, 0.2 mM phenylmethylsulfonyl fluoride, 1 g/ml aprotinin and leupeptin) and removed from dishes by scraping. Cells were then pelleted and resuspended in binding buffer containing 4 mg/ml BSA. 10 5 cells were incubated with 0.2 nM [ 32 P]SPP, synthesized enzymatically using recombinant sphingosine kinase (33) as described previously (9), in 0.2 ml of binding buffer plus 4 mg/ml BSA for 30 min at 4°C in the absence or presence of 1000-fold excess unlabeled SPP or other lipid competitors, added as 4 mg/ml fatty acid-free BSA complexes. Cells were then pelleted at 8,000 rpm in a microcentrifuge, washed twice with binding buffer containing 0.4 mg/ml fatty acid-free BSA, resuspended in binding buffer without BSA and bound [ 32 P]SPP quantitated by scintillation counting. The phosphatase and protease inhibitors were included in the binding assays as a precaution against the possibility that cells which may have been damaged during scraping might leak phosphatases or proteases which could cleave SPP or EDG receptors, respectively. In addition, it has been proposed that cell surface lipid phosphatases which can cleave exogenous SPP exist (34). Nevertheless, identical specific binding of [ 32 P]SPP was obtained in the absence of the protease and phosphatase inhibitors. It should be pointed out that SPP is not metabolized during the binding assay. When [ 32 P]SPP was incubated in the absence or presence of endothelial cells under the same conditions as the binding assay, no decrease in the amount of [ 32 P]SPP was detected by TLC nor did any additional bands appear.
Migration-Chemotactic migration of cells in response to a gradient of SPP was measured in a modified Boyden chamber as described previously (21). In brief, polycarbonate filters (5 m for BAEC and 8 m for HUVEC) were coated with gelatin (0.1%) overnight. Cells were harvested by trypsinization, washed with serum-free IMEM containing 0.1% fatty acid-free BSA, and were added to the upper wells (24multiwell Boyden microchambers) at 1 ϫ 10 5 cells per well; the lower wells contained SPP diluted in serum-free IMEM containing 0.1% BSA. After 2 h at 37°C in 5% CO 2 , non-migratory cells on the upper membrane surface were removed with a cotton swab and the cells which traversed and spread on the lower surface of the filter were fixed and stained with Diff-Quik. The number of migratory cells per membrane was enumerated using a microscope with a ϫ20 objective. Each data point is the average number of cells in four random fields, each counted twice. Each determination represents the average Ϯ S.D. of three individual wells. Checkerboard assays were carried out as described above except that various dilutions of SPP in fatty acid-free BSA were placed in the top and/or bottom wells of the Boyden chamber. In most of the experiments, unless indicated otherwise, cells were serum starved for 2 h prior to the assays.
[ 3 H]Thymidine Incorporation Assays-Cells were seeded at an initial density of 5 ϫ 10 4 cells per well in 24-well plates and allowed to attach overnight. Confluent BAEC were growth arrested in culture media without bFGF for 48 h and then treated for 24 h with SPP or bFGF. Since the sensitivity of BAEC to stimulation by bFGF and SPP decreased with passage number, all experiments were carried out with cells at less than passage 12. Confluent HUVEC were serum-starved for 24 h and then treated with different concentration of SPP in medium 199 containing 1% FCS and 1000 units/dl heparin for 24 h. [ 3 H]Thymidine (1 Ci/ml) was added 8 h before termination of the assay, and [ 3 H]thymidine incorporation into DNA was measured as described (38). Values are the means of triplicate determinations and standard deviations were routinely less than 10% of the mean.
In Vitro Angiogenesis-Three-dimensional collagen gel plates (24 well) were prepared by addition of 0.5 ml of a chilled solution of 0.7 mg/ml rat tail type I collagen in Dulbecco's modified Eagle's medium adjusted to neutral pH with NaHCO 3 . After formation of the collagen gel (about 1-2 mm thickness), BAEC were seeded at 50,000 cells/well. At 80% confluency, culture medium was changed to media without bFGF and incubation was continued for 48 h, by which time the cells formed a monolayer on the gel. The cells were then treated with different concentrations of bFGF or SPP as indicated. Cultures were maintained at 37°C for 48 h and then fixed with cold methanol. The gels were then soaked in phosphate-buffered saline/glycerol (1:1) and transferred onto glass slides. The extent of tube-like structures that formed in the gel was measured as total length per field using computerassisted imaging with a Hamamatsu C2400 video camera and a Zeiss Axioscope microscope to quantitate the extent of tube formation. Three culture wells were used for each sample, and three microscopic fields were examined for each well. Thus, each experimental point represents results from examination of nine microscopic fields.

Effect of SPP on Chemotaxis of EDG-1-overexpressing Cells-
SPP was previously shown to inhibit chemotaxis of mouse melanoma cells by binding to a putative cell surface receptor (20). Since we have recently identified EDG-1 as a receptor for SPP (22), it was of interest to examine the involvement of EDG-1 in SPP-regulated cell motility. Human embryonic kidney 293 fibroblasts stably expressing EDG-1 (HEK293-EDG-1) were selected for these studies since they have high levels of specific SPP binding, whereas SPP binding is nearly undetectable to parental and vector transfected cells (9,22). Surprisingly, low concentrations of SPP (10 -100 nM) did not inhibit, but rather enhanced chemotaxis of HEK293-EDG-1 cells by 1.5-1.9-fold, while chemotaxis of vector-transfected cells was not altered by these concentrations of SPP (Fig. 1A). The concentrations of SPP which stimulate chemotaxis of HEK293-EDG-1 cells are in the same range as the measured affinity of EDG-1 for SPP (K D ϭ 8 nM) (22) and correlate closely with binding and inhibition of forskolin-stimulated cAMP accumulation in these cells (9). These results suggest that low concentrations of SPP may increase chemotaxis by binding to EDG-1. Higher concentration of SPP (1-10 M), as previously reported in human breast cancer cells (21), inhibited, rather than stimulated, chemotaxis of both vector and EDG-1-transfected cells.
In contrast to undetectable levels of EDG-1 mRNA in parental and vector transfected HEK293 cells (9, 39), HEK293-EDG-1 cells express very high levels of EDG-1 mRNA as detected by Northern analysis (9) and RT-PCR ( Fig. 2A). Additionally, a low level of EDG-3 mRNA and EDG-5 mRNA were detected by RT-PCR in these cells ( Fig. 2A). Similar to our previous results (9), both unlabeled SPP and dihydro-SPP effectively competed with [ 32 P]SPP for binding to EDG-1 (data not shown), whereas, LPA and SPC were completely ineffective (Fig. 1B). Moreover, LPA, even at concentrations as high as 10 M, and for prolonged incubations, had no significant effect on SPP binding to HEK293-EDG-1 cells.
Expression of EDG Receptors in Endothelial Cells-Previously, it has been demonstrated that human endothelial cells express high levels of EDG-1 mRNA (28). Thus, it was of interest to examine whether the other SPP receptors, EDG-3 and EDG-5 (24,25,40), are also expressed in HUVEC as well as BAEC which, based on indirect evidence, have been proposed to have putative G i -coupled SPP receptors linked to calcium mobilization (27). Consistent with previous studies, RT-PCR analysis clearly demonstrated an EDG-1 PCR amplification product of about 1300 base pairs, in agreement with the predicted size (1284 base pairs), in HUVEC and BAEC (Fig.  2, B and C). HUVEC and BAEC also apparently expressed somewhat lower levels of EDG-3 and barely detectable EDG-5 mRNA. It should be noted that the bovine SPP receptor cDNAs have not yet been cloned and sequenced; however, our PCR primers were designed to cover highly conserved regions. Restriction analysis of the HUVEC RT-PCR products yielded fragments of the expected sizes confirming their identity (data not shown). The entire open reading frame of each of the three SPP receptors is encoded within a single exon (40,41), and RT-PCR primers which span an intron junction cannot be used to evaluate genomic DNA contamination. Therefore, controls without reverse transcriptase were performed in all cases (Fig. 2).
Effects of SPP on Chemotaxis of Endothelial Cells-Since SPP increased chemotaxis of HEK293 cells by apparently acting through EDG-1, the effect of SPP on chemotaxis of HUVEC and BAEC, which express EDG-1 and -3, was examined (Fig. 3,  A and B). SPP stimulated chemotaxis of both HUVEC and BAEC, reaching a maximum effect at 1 M (7-and 10-fold, respectively). Similar to the results with HEK293-EDG-1 cells, concentrations of SPP greater than 1 M were less effective, although significant enhancement of chemotaxis was still evident at higher concentrations. Interestingly, bFGF (20 ng/ml), a potent angiogenic factor (42), increased chemotaxis of BAEC to the same extent as did 1 M SPP.
It has recently been shown that directional migration toward appropriate agonist ligands can be triggered via receptors coupled to G i but not by agonists for receptors coupled to two other G proteins, G s and G q (43). Because biochemical evidence and the yeast two-hybrid system indicate that EDG-1 is capable of interaction with G␣ i1 and G␣ i3 (44), we investigated the possibility that G i proteins may be involved in the chemotactic response induced by SPP. To this end, HUVEC were treated with pertussis toxin, which ADP-ribosylates and inactivates G i and G o proteins, prior to addition of SPP. Pertussis toxin pretreatment completely abolished SPP-induced chemotaxis (Fig. 3C).
SPP Stimulates Directional Migration-We next determined whether the effect of SPP was mediated by enhanced directed migration in response to the gradient of chemoattractant (chemotaxis) or by increased random motility due to the presence of the chemoattractant itself (chemokinesis). Checkerboard as-  (Table I, bold), and also in the direction of the increasing chemokinetic gradient, i.e. when the concentration of SPP was the same in both the top and bottom chambers (Table I, italic), indicating that SPP stimulates both chemokinetic and chemotactic responses.
Binding of SPP and Dihydro-SPP Correlates with Stimulation of Cell Migration-It was important to determine whether [ 32 P]SPP was specifically bound to the endogenous SPP receptors on HUVEC. As shown in Fig. 4A, in HUVEC there was significant specific binding of both SPP and sphinganine 1-phosphate (dihydro-SPP), which lacks the double bond at the 4-position. Moreover, dihydro-SPP also markedly enhanced chemotaxis (Fig. 4B). The structure of SPP is similar to that of LPA, another serum-borne lysolipid that binds and signals through the related G i -coupled receptors, EDG-2 and EDG-4 (45)(46)(47). However, excess LPA did not compete with [ 32 P]SPP for binding to HUVEC, nor did it have a significant stimulatory effect on chemotaxis. SPC had no significant effect on specific [ 32 P]SPP binding and had a small but not statistically significant effect on chemotaxis (Fig. 4, A and B). Similar results were obtained with BAEC, where only SPP and dihydro-SPP, albeit less potently, competed with labeled SPP for binding, whereas LPA and SPC had no significant effect. These results are consistent with our previous observation that dihydro-SPP blocked binding to HEK293 cells overexpressing EDG-1 in a dose-dependent manner similar to unlabeled SPP (9,24), while neither LPA nor SPC had a significant effect on SPP binding (Fig. 1B). It should be pointed out that in these studies, all lysosphingolipids were added to cells as 0.4% BSA complexes, using conditions which were found to be optimal for binding of SPP to its receptor (9, 24), whereas these might not be optimal conditions for binding of LPA to EDG-2 and EDG-4. For example, LPA prepared as a 1% BSA complex promoted survival of Schwann cells, while SPP as a 0.01% complex was ineffective (48). We also examined the binding affinity of SPP for its putative receptor on endothelial cells by displacing bound [ 32 P]SPP with increasing concentrations of unlabeled SPP. 50% of the bound [ 32 P]SPP was competed at 10 nM unlabeled SPP in BAEC. Thus, binding of SPP to endothelial cells is also of high affinity and in agreement with the K d of EDG-1 (8.6 nM). These results indicate that there is an excellent correlation between the K d and the concentration-dependent effect of SPP on cell migration.
SPP Stimulates Proliferation of HUVEC and BAEC-Many angiogenic factors, in addition to enhancing chemotaxis, stimulate in vitro proliferation of endothelial cells (49 -51). Since SPP increased chemotaxis of endothelial cells, and has previously been shown to be a potent mitogen for diverse cell types (1,2,52), it was of interest to examine the effects of SPP on proliferation of endothelial cells. SPP treatment of HUVEC induced a dose-dependent increase of DNA synthesis as measured by [ 3 H]thymidine incorporation with a maximum effect at 0.1-1 M (Fig. 5A). Similar to SPP, 1 M dihydro-SPP also stimulated DNA synthesis in HUVEC by 1.83 Ϯ 0.1-fold, whereas LPA at a concentration as high as 10 M had no

FIG. 3. Effect of SPP on chemotaxis of HUVEC and BAEC.
A, HUVEC were allowed to migrate toward the indicated concentrations of SPP and chemotaxis was measured as described under "Experimental Procedures." B, BAEC were allowed to migrate toward different concentrations of SPP or 20 ng/ml bFGF as indicated, and chemotaxis was measured. Data are mean Ϯ S.D. of triplicate determinations. Similar results were obtained in at least three independent experiments. C, HUVEC were pretreated with vehicle or 200 ng/ml pertussis toxin for 3 h, and then allowed to migrate toward vehicle or 100 nM SPP as indicated. Asterisks indicate statistical significance determined by Student's t test (p Յ 0.01).

TABLE I Checkerboard analyses of HUVEC migration induced by SPP
The checkerboard assay was arranged with increasing concentrations of SPP in the lower chamber (bold indicates migration due to chemoattaction) or increasing concentrations of SPP in both upper and lower chambers, and cell migration assays were performed as described under "Materials and Methods." Data are mean Ϯ S.D. of triplicate determinations. SPP was also mitogenic for BAEC; however, a maximal effect in these cells required higher concentrations (1-10 M). bFGF has been reported to be a potent endothelial cell mitogen (56). Surprisingly, although bFGF stimulated proliferation of BAEC, at optimal concentrations it was only as effective as 1 M SPP and less effective than 10 M SPP (Fig. 5B). It should be noted that sensitivity of BAEC to SPP decreased with increasing passage number, similar to a previous report on the effect of passage number on bFGF responses (57). In addition to SPP, dihydro-SPP stimulated DNA synthesis, whereas we found that LPA at a concentration up to 10 M was not mitogenic, in fair agreement with previous studies where LPA only stimulated DNA synthesis in BAEC at concentrations around 30 M (57).
To investigate the possibility that a G i -coupled receptor may be involved in the proliferative response induced by SPP, endothelial cells were treated with pertussis toxin prior to addition of SPP. Both HUVEC and BAEC are more sensitive to pertussis toxin than Swiss 3T3 fibroblasts. In contrast with our previous studies with quiescent Swiss 3T3 fibroblasts (10,11), where half of the stimulated DNA synthesis was still evident even at the highest effective concentration of pertussis toxin, pertussis toxin pretreatment of HUVEC and BAEC completely inhibited the SPP-induced mitogenic response, while it had no significant effect on DNA synthesis induced by bFGF (Fig. 5C).
SPP Induces Capillary-like Tube Formation in Vitro-Later stages of angiogenesis require morphological alterations of endothelial cells, which result in lumen formation (31). Critical steps in angiogenesis, such as migration and differentiation, have been studied using an in vitro model of angiogenesis in which cultured endothelial cells are induced to invade a threedimensional collagen gel where they form a network of capillary-like structures or tubes when stimulated by angiogenic factors (58,59). This phenomenon is thought to mimic the formation of new blood vessels in vivo. In agreement with previous studies (59), confluent monolayers of BAEC treated with bFGF (50 ng/ml) formed networks of capillary-like tubular structures within the gel (Fig. 6A). In contrast, little invasion or network cord formation and only a few short capillary sprouts originating from untreated BAEC embedded in collagen gels were detected. SPP evoked a dose-dependent increase in the formation of capillary-like tubes of BAEC invading the collagen gel. Apparently thinner tubes were formed in response to lower doses of SPP than those formed in response to bFGF. Quantitative evaluation of tube formation revealed that SPP, similar to bFGF, markedly increased the length of the endothelial tubular structures (Fig. 6B). There was an additive effect when SPP was applied together with bFGF, suggesting that SPP can potentiate the effect of bFGF on in vitro angiogenesis. Similar to SPP, dihydro-SPP, also markedly enhanced capillary-like tube formation of BAEC, whereas LPA was completely inactive and SPC had a small but significant effect (Fig.  6C). Pertussis toxin not only inhibited endothelial cell migration and proliferation induced by SPP, it also markedly decreased the SPP-induced tube formation (Fig. 6C). In sharp contrast, pertussis toxin had no effect on tube formation induced by bFGF. DISCUSSION Previously, many studies have shown that SPP inhibits chemotaxis of diverse cell types (17,19,20,60,61). Although in human breast cancer cells, SPP inhibited chemotaxis independently of EDG-1 (21), a wealth of evidence suggests that in many other cell lines, SPP inhibits chemotaxis through unidentified cell surface receptors (17, 20, 62). Unexpectedly, we found in this study that binding of SPP to its G i protein-coupled receptor EDG-1 markedly increased cell motility. In accord with its affinity for EDG-1, SPP at nanomolar concentrations increased chemotaxis of EDG-1-transfected but not vectortransfected HEK293 cells, as well as HUVEC and BAEC which constitutively express EDG-1.
Although a recent study demonstrated that LPA can bind to EDG-1 leading to receptor phosphorylation, ERK activation, as well as Rho-dependent morphogenesis and P-cadherin expression (39), in our study, LPA, even at a concentration as high as 10 M, did not compete for binding of radiolabeled SPP to HEK293-EDG-1 cells. In agreement, LPA did not displace bound [ 32 P]SPP from endothelial cells nor did it stimulate chemotaxis of HUVEC. In contrast, dihydro-SPP, which binds to EDG-1 (9), EDG-3, and EDG-5 (24), was as potent as SPP in induction of chemotaxis. However, LPA, similarly to SPP, can mimic serum in inducing invasion of carcinoma and hepatoma cells into monolayers of mesothelial cells (63). Although the underlying mechanism of this effect is not clear, it may involve increased cell adhesion rather than enhanced cell motility. Moreover, in most other cell types, LPA stimulates chemokinesis and chemotaxis (62, 64 -66), which might be due to binding to its specific receptors, EDG-2 and EDG-4 (45)(46)(47)(48).
We have previously shown that the SPP-induced cAMP decrease (9) and ERK2 activation in EDG-1-transfected cells (22) were completely blocked by pretreatment with pertussis toxin, which uncouples G i from GPCR. Similarly, preincubation with pertussis toxin abolished the effect of SPP on migration of endothelial cells. Collectively, these findings suggest that binding of SPP to the serpentine receptor EDG-1 on the endothelial cell surface activates a pertussis toxin-sensitive G i protein crucial for chemotaxis. In agreement, it has recently been demonstrated that activation of G␣ i -coupled receptors and the subsequent release of G ␤␥ dimers is required to initiate signal transduction leading to directed cell migration (43,67). It should be noted that the receptors for all known leukocyte chemoattractants, including the chemokines, are members of a seven-transmembrane domain superfamily and coupled to a variety of G ␣ subunits (68,69). Moreover, pertussis toxin inhibits chemotaxis by preventing chemoattractant receptors from activating trimeric G proteins of the G i subfamily (67).
Endothelial cells are accessible to platelet-derived ligands in the serum, and might be the target of serum-borne SPP which regulates their proliferation, migration, and differentiation into capillary-like tubules, important aspects of angiogenesis (31). SPP is well established as a potent mitogen for diverse cell types (reviewed in Ref. 1). Indeed, we found that SPP also stimulated DNA synthesis in endothelial cells and this was completely blocked by pertussis toxin, suggesting a role for G i in this process. In agreement, SPP was recently reported to stimulate HUVEC proliferation (70), albeit at somewhat higher concentrations which resembled the dose-response curve that we found for BAEC. It is possible that differences in sensitivity to SPP might arise from differences in passage numbers as has been shown for responses to bFGF (57). Interestingly, SPP stimulated DNA synthesis in Swiss 3T3 fibroblasts only at high concentrations and in contrast to endothelial cells, was only partially inhibited by pertussis toxin (10). Moreover, in these cells, DNA synthesis was significantly and specifically increased by microinjection of SPP (9). Pertussis toxin reduced the level of DNA synthesis caused by exogenous SPP to approximately the same level as that induced by microinjected SPP (9). Thus it is likely that in fibroblasts, both intracellular as well as receptor-mediated responses to SPP are involved in its mitogenic effect. It is possible that there is a complex interplay between cell surface receptor signaling and intracellular targets for SPP, which can contribute to its mitogenic response in certain cell types. Thus, SPP, may act in a similar manner to other bioactive lipids, such as leukotriene B4 (LTB4), which act through cell surface receptors and also have intracellular targets. LTB4 is a potent chemoattractant that is primarily in-FIG. 6. Induction of in vitro angiogenesis in collagen gels by SPP is attenuated by pertussis toxin. BAEC were treated with the indicated concentration of SPP in the absence or presence of bFGF (50 ng/ml) and capillary tube formation on collagen gels was examined as described under "Experimental Procedures." A, representative phasecontrast micrographs of BAEC after 4 days of incubation in normal medium (upper panels) or in the presence of 50 ng/ml bFGF (lower panels) and the indicated concentrations of SPP. B, quantitative analysis of capillary tube formation. Data are expressed as length of tubes per square millimeter (n ϭ 4 pairs of duplicates). Asterisks indicate statistical significance determined by Student's t test (p Յ 0.01). C, BAEC were treated in the absence or presence of pertussis toxin (PT, 20 ng/ml), without or with the indicated lysophospholipids (1 M) or bFGF (50 ng/ml) and capillary tube formation on collagen gels was determined as described under "Experimental Procedures." Data are expressed relative to the maximum response elicited by bFGF. volved in activation of inflammatory cells by binding to its GPCR. However, it can also bind and activate the intranuclear transcription factor PPAR␣, resulting in the activation of genes that terminate inflammatory processes (71).
In this study, we also observed that SPP in addition to stimulating both random, nondirectional migration (chemokinesis) and directional migration (chemotaxis) of endothelial cells, also markedly enhanced morphogenetic differentiation of endothelial cells in vitro. BAEC cultured on collagen showed increased tube formation in the presence of SPP, demonstrating that SPP stimulates endothelial cell differentiation as well as migration. Moreover, pertussis toxin not only inhibited endothelial cell migration and proliferation induced by SPP, it also completely inhibited SPP-induced, but not bFGF-induced, formation of capillary-like tubes. Taken together, these results demonstrate that SPP has angiogenic activity in vitro acting through a G i -coupled cell surface receptor.
In contrast to its effect on HEK293-EDG-1 cells, SPP did not stimulate chemotaxis (Fig. 1), proliferation (9), or morphogenetic differentiation in vector-transfected HEK293 cells (22), further supporting the notion that the effects that we observed on chemotaxis, proliferation, and tube formation of endothelial cells are mediated by EDG-1. In contrast, parental and vectortransfected HEK293 cells do show activation of ERK mitogenactivated protein kinases in response to SPP which might be attributable to endogenous EDG-3 or EDG-5. Taken together with our results, these findings suggest that EDG-1 may mediate these stimulatory effects of SPP on endothelial cells. However, as the other known SPP receptor EDG-3, which is also expressed in HUVEC, has been shown in some cases to couple to pertussis toxin-sensitive G proteins as well as to Rho and phospholipase C signaling pathways (14,26), this receptor might also contribute to the effects of SPP on tube formation. Recently, the polycyclic anionic compound suramin has been shown to selectively antagonize SPP-activated calcium transients in EDG-3, but not in EDG-1 or EDG-5 expressing oocytes, with an IC 50 of ϳ22 M, suggesting that it is an antagonist selective for the EDG-3 GPCR isotype (72). However, addition of 100 M suramin did not abrogate the ability of SPP to induce tube formation in BAEC. 2 In contrast, suramin inhibited Rho-dependent neurite retraction induced by SPP in N1E-115 neuronal cells (73) and SPP-induced invasion of Tlymphoma cells (74). However, in agreement with its lack of effect on SPP-induced tube formation, suramin had no significant effect on proliferation, stress fiber formation, and focal adhesion kinase phosphorylation induced by SPP in Swiss 3T3 fibroblasts (75).
Similar to SPP, dihydro-SPP also markedly enhanced cell migration and capillary-like tube formation of BAEC, whereas LPA was completely inactive. Another structurally related analog of SPP, SPC, although it had no significant effect on binding of labeled SPP to endothelial cells, it had a small but significant stimulatory effect on tube formation, and at high concentrations, it also stimulated proliferation. These effects might be related to the potent action of SPC as a wound healing agent (55). Interestingly, high micromolar concentrations of SPC activated calcium transients in EDG-1, -3, and -5 expressing oocytes (72) and SRE-driven gene transcription in Jurkat T cells (25). While these observations suggest that SPC might be a very low affinity ligand for these EDG receptors, it is also possible that these effects were not mediated by SPC itself as it was recently found that commercial preparations of SPC are contaminated with highly potent alkenyl glycero-3-phosphates (76). Some of the downstream effects of SPP signaling through EDG-1, such as decreased cAMP (9,23) and activation of ERK2 (22), are pertussis toxin-sensitive, while others, such as morphogenetic differentiation, are pertussis toxin-insensitive but inhibited by the C3 exoenzyme (22) which blocks signaling through the small GTPase Rho (77). Thus it appears that EDG-1 can couple to G i proteins as previously reported (44), as well as to G 12/13 proteins which are thought to regulate Rho (78,79), and thus might also be important for SPP-induced chemotaxis and angiogenesis. Recently, it has been demonstrated that disruption of the gene encoding G␣ 13 impaired the ability of endothelial cells to organize into a vascular system and greatly impaired migratory responses (80), suggesting that in addition to G i , other proteins including G␣ 13 , might be required for regulation of cell movement. In agreement, T-lymphoma cell invasion is dependent on SPP receptor-mediated RhoA and phospholipase C signaling pathways which lead to pseudopodia formation (74).
In summary, in this study we have demonstrated that SPP has appropriate properties to be considered as a bona fide angiogenic factor, i.e. it stimulates chemokinetic and chemotactic motility, proliferation of vascular endothelial cells, and stimulates angiogenesis in vitro, similarly to the known angiogenic factor bFGF. Because bFGF and SPP have an additive effect on formation of capillary-like tubes by endothelial cells invading collagen gels, SPP may be a specific type of angiogenic factor. It is possible that SPP plays a role in normal blood vessel formation or in injury, when local production of SPP could be increased by activated platelets, and extravasation of intravascular fluid could also present SPP into tissues at concentrations sufficient to promote angiogenesis and wound healing. Elucidation of the molecular mechanisms by which SPP stimulates cell migration and angiogenesis might provide clues for development of new therapeutic agents to either promote or block these processes by targeting the EDG family of GPCR.