Sphingosine 1-Phosphate Induces Membrane Ruffling and Increases Motility of Human Umbilical Vein Endothelial Cells via Vascular Endothelial Growth Factor Receptor and CrkII* 210

Sphingosine 1-phosphate (S1P), a ligand for endothelial differentiation gene family proteins, is one of the most potent signal mediators released from activated platelets. Here, we report that S1P induces membrane ruffling of human umbilical vein endothelial cells (HUVECs) via the vascular endothelial growth factor receptor (VEGFR), Src family tyrosine kinase(s), and the CrkII adaptor protein. S1P induced prominent phosphorylation of CrkII in HUVECs, indicating that CrkII was involved in the S1P-induced signaling pathway. S1P-induced CrkII phosphorylation was blocked by pertussis toxin and overexpression of the carboxyl terminus of β-adrenergic receptor kinase, indicating that the βγ subunit of Gi was required for the phosphorylation. Notably, the S1P-induced CrkII phosphorylation was also abolished by inhibitors of VEGFR or Src family tyrosine kinases. By using Picchu, a real time monitoring protein for CrkII phosphorylation, we found that S1P induced rapid CrkII phosphorylation at membrane ruffles. Finally, we observed that expression of a dominant negative mutant of CrkII inhibited the S1P-induced membrane ruffling and cell migration. These results delineated a novel S1P signaling pathway that involves sequential activation of Gi-coupled receptor(s), VEGFR, Src family tyrosine kinase(s), and the CrkII adaptor protein, and which is responsible for both the induction of membrane ruffling and the increase in cell motility.

CrkII is an adaptor protein consisting of a Src homology 2 (SH2) domain and two SH3 domains (14). Alternative splicing of the human crk gene generates two Crk proteins, designated as CrkI and CrkII. CrkI lacks the carboxyl-terminal SH3 of CrkII. The SH2 domain of CrkII binds several phosphotyrosine-containing proteins, p130 Cas , paxillin, and Cbl, whereas the SH3 domain of CrkII binds to C3G, DOCK180, and Abl (15). Recently, CrkII associated with DOCK180 has received attention for its role in cell migration (16,17).
The involvement of CrkII in cellular migration and the induction of membrane ruffling has been studied extensively both biochemically and genetically (15). The Crk-DOCK180-Rac pathway is conserved from nematode to man and plays a critical role in the regulation of membrane ruffling and cellular migration. We have shown that one of the two major Crk SH3-binding proteins, DOCK180, binds to and activates Rac1 and Rac2 to induce membrane ruffling (18,19). A defect in ced-5, a homolog of DOCK180, inhibits migration of the distal tip cells and phagocytosis of apoptotic bodies (20,21). It has been proven genetically that ced-5 is downstream from ced-2, a homolog of crkII, and upstream from ced-10, a homolog of rac (21). In Drosophila melanogaster, nonfunctional mutation in myoblast city, a homolog of DOCK180, results in the failure of dorsal closure (22), suggesting defective cell migration.
CrkII becomes phosphorylated on Tyr 221 upon stimulation by the following growth factors: vascular endothelial growth factor (VEGF) (23), epidermal growth factor (EGF) (24), nerve growth factor (25), platelet-derived growth factor (PDGF) (26), and insulin-like growth factor I (27). Among these growth factors, VEGF induces the most prominent endothelial cell migration. Notably, CrkII is also phosphorylated upon S1P stimulation in NIH-3T3 cells (28), although the mechanisms by which CrkII is regulated downstream from the EDG receptor and how CrkII is involved in cell migration have not yet been elucidated.
This study investigates the molecular mechanism of S1Pinduced endothelial cell migration. The results demonstrate that S1P induces CrkII phosphorylation by the ␤␥ subunits of heterotrimeric G i protein (G i ␤␥)-mediated transactivation of VEGFR followed by the activation of Src family tyrosine kinase(s) and that CrkII is responsible for the membrane ruffling and cell motility induced by S1P in human umbilical vein endothelial cells (HUVECs).
Plasmids and Virus-cDNA coding CrkI replaced either at Arg 38 by Val (hereafter, R38V) or at Trp 169 by Leu (hereafter, W169L) was subcloned into the bicistronic promoter vector, pCXN2-FLAG-IRES-EGFP (41,42). We produced a recombinant adenovirus for the in vivo CrkII phosphorylation-monitoring protein, Picchu (43), by means of an Adeno-X expression system (CLONTECH). Briefly, Picchu consists of a yellow emitting mutant of green fluorescent protein (YFP), CrkII, and a cyan emitting mutant of green fluorescent protein (CFP) from the amino terminus (43). Upon phosphorylation on Tyr in Picchu corresponding to Tyr 221 of CrkII, SH2 binds to this phosphotyrosine, which causes intramolecular folding of Picchu and results in an increase in fluorescent resonance energy transfer from CFP to YFP. Adenovirus expressing both FLAG-tagged CrkI-W169L and EGFP was produced in a manner similar to that for adenovirus expressing Picchu. Recombinant adenovirus for the carboxyl terminus of ␤-adrenergic receptor kinase and for GFP was produced by the COS-TPC method as described previously (44). Adenovirus for carboxyl-terminal Src kinase (CSK) was obtained from S. Tanaka (University of Tokyo, Japan) (45).
Cells-HUVECs and COS-1 cells were purchased from American Type Culture Collection (Rockville, MD). HUVECs were cultured with Dulbecco's modified Eagle's medium supplemented with 10% fetal bovine serum, 2 mM L-glutamine, 0.1 mg/ml heparin, and 0.03 mg/ml endothelial cell growth supplement from Sigma, and used for experiments before passage 5. COS-1 cells were cultured in Dulbecco's modified Eagle's medium supplemented with 10% fetal bovine serum. HU-VECs cultured on a collagen-coated 35-mm diameter glass base dish (Asahi Techono Glass Co., Tokyo) were transfected with 3 g of plasmid DNA using LipofectAMINE PLUS Reagent (Invitrogen Corp.) or infected with adenovirus at the appropriate multiplicity of infection for more than 24 h before the stimulation with reagents for imaging.
Immunoprecipitation and Immunoblotting-HUVECs in 10-cm plates or COS-1 cells in six-well dishes were starved for 8 h and stimulated with reagents with or without pretreatment as indicated in the figure legends. Cells were exposed to the reagents at 37°C for the time indicated in the figures, washed with buffered saline containing 10 mM Tris-HCl (pH 7.5) and 1 mM Na 3 VO 4 , and lysed in a lysis buffer (10 mM Tris-HCl (pH 7.5), 50 mM NaCl, 50 mM NaF, 1% Triton X-100, 5 mM EDTA, 2 mM Na 3 VO 4 , 0.1% bovine serum albumin, 20 g/ml aprotinin, 1 mM phenylmethlysulfonyl fluoride), and cleared by centrifugation at 15,000 ϫ g for 15 min. Aliquots of total cell lysate were subjected to immunoblotting with antibodies as indicated in the figures. The remaining lysate was subjected to immunoprecipitation using antibodies as indicated in the figures and protein A and G-Sepharose (Calbiochem), followed by SDS-PAGE and immunoblotting. Proteins reacting with primary antibodies were visualized by the ECL system (Amersham Biosciences, Buckinghamshire, UK) detecting peroxidase-conjugated secondary antibodies and analyzed with the LAS-1000 system (Fuji Film, Tokyo).
Quantitation of CrkII Phosphorylation-The intensities of the phosphorylated slower migrating form and nonphosphorylated faster migrating form of CrkII (24) were measured with an LAS-1000 image analyzer. Then, the percentage of the phosphorylated form was calculated for each sample. The ligand-induced increase in the phosphorylated CrkII was determined as the ratio to the control (prestimulation). Data from at least three independent experiments were averaged, and statistical significance was evaluated by Student's t test.
Fluorescent Resonance Energy Transfer Imaging-HUVECs infected with adenovirus for Picchu on a collagen-coated 35-mm diameter glass base dish were starved for 8 h and stimulated with 100 nM S1P. Cells were imaged on an Olympus IX-70 inverted microscope with a 75-W xenon arc lamp equipped with a MultiSpec Micro-Imager (Optical Insights, Santa Fe, NM) and a cooled CCD camera, CoolSNAP-HQ, controlled by MetaFluor (Roper Scientific, Trenton, NJ). CFP and YFP images were obtained simultaneously by a filter set consisting of an XF1071 excitation filter and an XF2034 dichroic mirror (Omega Optical, Inc., Brattleboro, VT). The emission ratio of YFP to CFP and the intensity of CFP were used for imaging of the phosphorylation of Picchu in the intensity modulated display mode controlled by MetaFluor.
Time-lapse Imaging, Quantitative Analysis of Membrane Ruffling, and Cell Motility Analysis-HUVECs transfected with either pCXN2-FLAG-CrkI-W169L-IRES-EGFP or pCXN2-FLAG-CrkI-R38V-IRES-EGFP were starved for 8 h and stimulated with S1P. A phase contrast image and a fluorescence image were recorded first, and then a sequential phase-contrast image was obtained every 30 s. A series of timelapse images was converted to a video using MetaMorph 4.6 software (Roper Scientific). S1P-induced membrane extension reflecting the membrane ruffling was quantitated by measuring the cell size before and after S1P stimulation. The cell size was analyzed by a region measurement tool included in the MetaMorph 4.6 software. Cell motility was analyzed as described previously (46). Briefly, HUVECs labeled with BCECF-AM (Molecular Probes, Eugene, OR) were spread on a collagen-coated glass-base dish and cultured in Dulbecco's modified Eagle's medium supplemented with 1% fetal bovine serum for 2 h before exposure to S1P in the presence or absence of AG1433. HUVECs infected with either adenovirus expressing GFP or adenovirus expressing both CrkI-W169L and EGFP were spread and stimulated with S1P. Cells labeled with BCECF-AM or expressing EGFP were tracked by a series of time-lapse images using fluorescent microscopy. The distance between the point at which the cell attached and the end point to which the cell moved was measured by tracing cells. The velocity was obtained as the distance divided by the period during tracing of cells and was analyzed by a cell tracking system included in the MetaMorph 4.6 software.

CrkII Is Phosphorylated upon S1P and VEGF Stimulation in
HUVECs-To explore the involvement of CrkII in S1P-induced migration of HUVECs, we examined whether CrkII became phosphorylated when HUVECs were stimulated with S1P. We have shown previously that the phosphorylated form of CrkII migrates more slowly than the wild-type CrkII on SDS-polyacrylamide gel (24). In the present study, we used this observation to detect the percentage of phospho-CrkII by SDS-PAGE, immunoblotting, and densitometry. In parallel, as a second measure of S1P stimulation, we also examined the S1P-induced activation of ERK by anti-phospho-ERK antibody. 100 nM S1P induced a statistically significant increase in level of phosphorylation of CrkII (p Ͻ 0.01) (Fig. 1A), and this con-centration was used in subsequent experiments. Upon S1P stimulation, phosphorylation of CrkII and ERK reached a plateau within 1 min and returned slowly to the basal level in 1 h (Fig. 1B). Both CrkII and ERK were phosphorylated in a doseand time-dependent manner upon S1P stimulation in HU-VECs. Because G protein-coupled receptors often transactivate receptor-type tyrosine kinases, we searched for growth factors that could induce CrkII phosphorylation in HUVECs. Among VEGF, EGF, basic fibroblast growth factor, hepatocyte growth factor, and PDGF, only VEGF induced significant CrkII phosphorylation (Fig. 1C). All growth factors except PDGF induced ERK phosphorylation, suggesting that VEGFR might be the tyrosine kinase receptor responsible for S1P-induced CrkII phosphorylation in HUVECs.
CrkII Phosphorylation Is G i -dependent in HUVECs-To test which heterotrimeric G protein was involved in S1P-induced CrkII phosphorylation in HUVECs, we treated HUVECs with PTX before stimulation. Both CrkII and ERK phosphorylation by S1P were abolished by PTX pretreatment (Fig. 2A, left panel), indicating that S1P induced phosphorylation of CrkII and ERK via G i . As expected, VEGF-dependent phosphorylation of CrkII and ERK was not affected by PTX ( Fig. 2A, right  panel). In addition, adenoviral-mediated overexpression of the carboxyl terminus of ␤-adrenergic receptor kinase, which sequesters ␤␥ subunits (47), inhibited S1P-induced phosphorylation of CrkII and ERK. Adenoviral GFP, as a negative control, did not inhibit S1P-induced CrkII and ERK phosphorylation (Fig. 2B, left panel). Neither GFP nor the carboxyl terminus of ␤-adrenergic receptor kinase affected VEGF-induced phosphorylation of CrkII and ERK (Fig. 2B, right panel). These data indicated that both CrkII and ERK were phosphorylated downstream from G i ␤␥ in HUVECs.
Src Family Tyrosine Kinases Are Involved in S1P-induced CrkII Phosphorylation in HUVECs-It has been shown that Src family tyrosine kinase(s) is required to transactivate EGFR in COS cells (48). Therefore, we examined the involvement of Src family tyrosine kinase(s) in S1P-induced CrkII phosphorylation in HUVECs by using a Src family tyrosine kinase inhibitor, PP2, and CSK, a tyrosine kinase that down-regulates Src activity. Pretreatment of HUVECs with PP2 blocked S1P-induced CrkII phosphorylation but not ERK phosphorylation (Fig. 3A, left panel). That Src family tyrosine kinase(s) was required for S1P-induced CrkII phosphorylation was confirmed by the expression of CSK (Fig. 3B, left panel). Surprisingly, VEGF-induced CrkII phosphorylation was also inhibited by PP2 and CSK (Fig. 3, A, right panel and B, right panel). Thus, Src family tyrosine kinase(s) appeared to function downstream Phosphorylated CrkII (p-CRKII) and nonphosphorylated CrkII (CRKII) were detected as slower and faster migrating forms of CrkII on the membrane probed with anti-Crk antibody. Relative CRK phosphorylation indicates the ratio of the poststimulation p-CrkII fraction of total Crk (Crk ϩ p-CrkII) to the prestimulation (control) p-CrkII fraction. The mean relative Crk phosphorylation is shown Ϯ the S.D. Each immunoblot result is a representative of at least three independent experiments. A significant difference from the control by t test is indicated as an asterisk (p Ͻ 0.05) or double asterisk (p Ͻ 0.01). B, cells were exposed to 100 nM S1P for the duration indicated at the top of the figure and analyzed as in A. C, HUVECs were exposed to a series of growth factors as indicated at top for 5 min and analyzed by immunoblotting as in A. p-Erk, phospho-ERK; bFGF, basic fibroblast growth factor; HGF, hepatocyte growth factor; PDGF-AB, platelet-derived growth factor A-chain/B-chain heterodimer. from VEGFR in S1P-induced CrkII phosphorylation in HU-VECs. In contrast to HUVECs, S1P did not induce CrkII phosphorylation (Fig. 3C, left panel), and Src family tyrosine kinases were not involved in EGF-induced CrkII phosphorylation (Fig. 3C, right panel) in COS-1 cells, although Src family tyrosine kinases were required for S1P-induced ERK phosphorylation (Fig. 3C, left panel).
EGFR Is Not Required for S1P-induced CrkII Phosphorylation in HUVECs-To exclude the involvement of EGFR in S1P-induced CrkII phosphorylation in HUVECs, HUVECs were pretreated with AG1478, an EGFR kinase inhibitor, before S1P stimulation. As expected, in the presence of AG1478 neither CrkII nor ERK phosphorylation was inhibited upon S1P stimulation in HUVECs (Fig. 4A, left panel). The inhibition of EGFR by AG1478 in HUVECs was confirmed by examining the EGF-induced ERK phosphorylation (Fig. 4A, right  panel). In COS-1 cells, S1P-dependent (Fig. 4B, left panel) and EGF-dependent (Fig. 4B, right panel) phosphorylation of ERK were abrogated by pretreatment with AG1478, consistent with previous reports that EGFR is transactivated by G i to activate ERK (48,49).
VEGFR but Not EGFR Is Transactivated in S1P-stimulated HUVECs-We examined whether the EGFR and the VEGFR were phosphorylated upon S1P stimulation in HUVECs. Only VEGFR-2 was phosphorylated upon S1P stimulation in HU-VECs (Fig. 5, top and middle panel). Although VEGFR-1 and -2 are expressed in HUVECs (29), we could not test whether VEGFR-1 was phosphorylated because there was no specific and sensitive antibody to VEGFR-1 available for this analysis. In contrast to VEGFR-2, EGFR was phosphorylated in S1Pstimulated COS-1 cells (Fig. 5, bottom panel) but not in S1Pstimulated HUVECs (Fig. 5, top panel), suggesting that the S1P-induced EGFR activation was cell type-specific.
VEGFR Is Required for S1P-induced CrkII Phosphorylation-We further confirmed that VEGFR was responsible for the S1P-induced CrkII phosphorylation by the use of three VEGFR inhibitors. AG1433 and SU5614 inhibit the tyrosine kinase activity of VEGFR-2 and PDGFR, whereas VEGF-tyrosine kinase inhibitor is a specific inhibitor for VEGFR-1 and -2. As expected, all of these inhibitors blocked S1P-induced phosphorylation of CrkII but not that of ERK (Fig. 6A, upper panel). VEGF-induced ERK phosphorylation in HUVECs was completely inhibited by the VEGFR inhibitors (Fig. 6A, lower panel), demonstrating the effectiveness of these compounds. As an additional control, we also tested AG1296, an inhibitor specific for PDGFR, which would not be expected to inhibit S1P signaling to ERK and CrkII (Fig. 1C). As expected, AG1296 did not abrogate phosphorylation of either ERK or CrkII upon S1P stimulation (Fig. 6B, left panel). The inability of AG1296 to inhibit VEGFR signaling was also confirmed (Fig. 6B, right  panel). Therefore, although we cannot exclude the involvement of VEGFR-1 without a VEGFR-1-specific inhibitor, S1P-induced CrkII phosphorylation required VEGFR-2. In contrast to CrkII phosphorylation, S1P-induced ERK phosphorylation was mediated mostly by VEGFR-independent pathway(s).
CrkII Phosphorylation at the Membrane Ruffling-To examine where and when CrkII was involved in S1P-induced mem-  Fig. 1. B, HUVECs infected with adenovirus expressing GFP (Adeno-GFP) or adenovirus expressing CSK (Adeno-CSK) for 24 h were starved for 8 h and exposed to S1P (left panel) or VEGF (right panel). C, COS-1 cells pretreated with PP2 were exposed to S1P (left panel) or EGF (right panel) and analyzed for CrkII and ERK phosphorylation. Relative Crk phosphorylation was analyzed as described in the legend of Fig. 1.

FIG. 4. EGFR-independent CrkII and ERK phosphorylation by S1P in HUVECs.
A, HUVECs were pretreated with 10 nM AG1478 for 20 min prior to S1P (left panel) or EGF (right panel) stimulation. CrkII and ERK phosphorylation were analyzed as described in the legend of Fig. 1

. B, COS-1 cells were analyzed as in A.
brane ruffling, we used the phosphorylation indicator of the Crk chimeric unit, Picchu, for monitoring CrkII phosphorylation on Tyr 221 in living HUVECs. Phosphorylation of CrkII, shown by red hues, was most prominent at membrane ruffles in HUVECs after S1P stimulation ( Fig. 7A and supplemental  material). Similar phosphorylation of CrkII at membrane ruffles was also observed in VEGF-stimulated HUVECs (Fig. 7B and supplemental material). The intensity of Picchu, which reflects the concentration of CrkII, was increased at membrane ruffles. These observations indicated that CrkII was recruited to and phosphorylated at membrane ruffles upon S1P stimulation.
VEGFR Is Required for S1P-induced Membrane Ruffling and Cell Motility-To examine the requirement of VEGFR, we tested the effect of AG1433, an inhibitor for VEGFR kinase, on membrane ruffling and cell motility upon S1P stimulation (Fig.  8). HUVECs responded to S1P and showed prominent membrane ruffling (Fig. 8A, upper panel, and supplemental material), whereas HUVECs pretreated with AG1433 showed less membrane ruffling upon S1P stimulation (Fig. 8A, lower panel, and supplemental material). Because S1P-induced membrane ruffling was accompanied by spreading of the peripheral plasma membrane, we quantitated the effect of AG1433 by measuring the area of cells. S1P increased cell size by about 30%. This S1P-induced increase in cell size was completely inhibited by AG1433 (Fig. 8A, right panel). Furthermore, we examined the effect of AG1433 on S1P-induced cell motility of HUVECs stimulated with S1P (Fig. 8B). S1P accelerated the migratory velocity of HUVECs (Fig. 8B, left and center panel); however, this acceleration was inhibited by pretreatment with AG1433 (Fig. 8B, right and center panel). These results suggested that VEGFR was required for S1P-induced membrane ruffling and cell motility.
CrkII Is Required for S1P-induced Membrane Ruffling and Cell Motility-To examine whether CrkII was required for S1Pinduced membrane ruffling and cell motility, we used dominant negative mutants of CrkI, CrkI-W169L and CrkI-R38V. CrkI is a splicing variant of CrkII and lacks the carboxyl-terminal SH3 domain. CrkI-W169L consists of an intact SH2 and a nonfunctioning SH3 domain. CrkI-R38V consists of a nonfunctioning SH2 and an intact SH3 domain. We have shown previously that CrkI-W169L and CrkI-R38V work as dominant negative mutants for CrkI/II (42,50). Phosphorylation of CrkII upon S1P stimulation was abolished by adenovirus-mediated overexpression of CrkI-W169L (Fig. 9A, left panel). In addition, VEGF-induced CrkII phosphorylation was completely inhibited by overexpression of CrkI-W169L (Fig. 9A, right panel), suggesting that Crk functioned downstream from VEGFR. Furthermore, HUVECs transfected with CrkI-W169L did not show any membrane ruffling after S1P stimulation (Fig. 9B and  supplemental material). Overexpression of CrkI-R38V also inhibited S1P-induced membrane ruffling (data not shown). Finally, we examined the effect of overexpression of CrkI-W169L on cell motility of HUVECs. S1P accelerated the migratory velocity of HUVECs infected with the control adenovirus, Adeno-GFP (Fig. 9C, left and center panel); however, the S1Pinduced acceleration of migration was completely abolished by the adenovirus carrying the dominant negative CrkI-W169L gene (Fig. 9C, center and right panel). These observations indicated that CrkII was required for S1P-induced membrane ruffling and cell migration. DISCUSSION Angiogenesis, which is observed in wound repair, tumorigenesis, and tissue ischemia, is an integral feature of vascular sprouting, branching, and remodeling and is coordinated by endothelial cells, vascular smooth muscle cells, and mesenchymal cells (51). VEGF, angiopoietin, and ephrin have been shown to be key molecules in the promotion of angiogenesis via activation of the VEGFR, Tie, and Eph expressed on vascular endothelial cells, respectively (51). Recently, S1P has been identified as another potent angiogenic factor because it promotes prominent endothelial cell migration (11,52,53). However, despite these extensive studies, the molecular mechanism of S1P-induced endothelial cell migration is not yet clearly understood. In this study, we delineate a novel signaling pathway required for S1P-triggered cell migration via sequential activation of G i ␤␥, VEGFR, Src family tyrosine kinase(s), and CrkII.
Our data support the idea that CrkII is involved in the S1P-stimulated signaling pathway in HUVECs. CrkII was phosphorylated upon S1P stimulation in HUVECs, indicating that CrkII functions downstream from the EDG receptor in HUVECs, just as it functions downstream from nerve growth factor receptors in PC12 cells and EGFR in NRK cells in response to nerve growth factor and EGF, respectively (24,25). To explore where and how CrkII functions, we monitored CrkII phosphorylation in S1P-stimulated HUVECs by Picchu, which reflects CrkII phosphorylation on Tyr 221 . Our data showed that CrkII was phosphorylated at the site of membrane ruffling. CrkII was also localized in newly assembling focal complexes at the leading edge, where CrkII was likely to bind to p130 Cas upon S1P stimulation. 2 In addition, it has been reported that p130 Cas and CrkII are colocalized at membrane ruffles upon S1P stimulation (9) and that the CrkII and p130 Cas complex functions as a critical molecular switch in directing the membrane ruffling and cell migration (16). Our results together with these reports suggest that CrkII is involved in S1P-induced membrane ruffling in HUVECs. Furthermore, the inhibition of S1P-induced membrane ruffling and cell motility by the dominant negative mutant of CrkII supports our proposal. Because Rac activation via the CrkII⅐DOCK180 complex has already been demonstrated both biochemically and genetically (18,20,21), our finding connects the S1P signal transduction cascade to the actin reorganization machinery via Rac.
The present study is the first to demonstrate transactivation of VEGFR and thereby to link G i -coupled EDG receptors to CrkII phosphorylation. S1P-induced CrkII phosphorylation is dependent upon PTX and ␤␥ subunits. This observation suggests that among the eight EDG family proteins, those coupled 2 A. Endo and N. Mochizuki, unpublished results. with G i are responsible for the CrkII phosphorylation in HU-VECs. Such candidates in HUVECs are EDG-1 and -3 (11,54). Many G i -coupled receptors are known to transactivate EGFR and PDGFR in various cell types and in response to a wide variety of ligands (49,55). These two tyrosine kinase receptors are, to the best of our knowledge, the only two receptors identified in G i -mediated transactivation. The transactivation of EGFR was first reported in Rat-1 fibroblasts stimulated with endothelin-1, lysophosphatidic acid, or thrombin (56). PDGFR is also transactivated by angiotensin II in vascular smooth muscle cells (57) and by dopamine in CHO-K1 cells (58).
In contrast to other cell types, there was no evidence that the EGFR and the PDGFR are involved in S1P signaling in HU-VECs. EGFR was not tyrosine-phosphorylated upon S1P stim-ulation. The EGFR inhibitor AG1478 did not abrogate either CrkII phosphorylation or ERK activation upon S1P stimulation. In addition, PDGFR may not be expressed in HUVECs, as suggested by the findings that ERK was not activated by PDGF stimulation and that S1P-stimulated CrkII phosphorylation was not abrogated by a PDGFR inhibitor. Recently, heparinbinding EGF, a membrane-bound EGF-like ligand, has been shown to transactivate EGFR in COS cells stimulated with lysophosphatidic acid (55) and in vascular smooth muscle cells stimulated with angiotensin II (59). If this is a general mechanism for the transactivation of receptor tyrosine kinases, VEGF, which is the only inducer for CrkII phosphorylation tested in this study, may be released from the cell membrane of S1P-stimulated HUVECs.
FIG. 6. VEGFR-mediated CrkII phosphorylation upon S1P stimulation in HUVECs. A, HUVECs pretreated with a tyrosine kinase inhibitor, 20 M AG1433, 10 M SU5614, or 5 M VEGF-TKI, were exposed to S1P (upper panel) or VEGF (lower panel) and analyzed as described in the legend of Fig. 1. VEGF-TKI is an inhibitor of both VEGFR-1 and -2. B, HUVECs pretreated with 10 M AG1296, a PDGFspecific inhibitor, were exposed to S1P (left panel) or VEGF (right panel) and analyzed as in A.
We have shown that activation of VEGFR and subsequent activation of Src family tyrosine kinase(s) are required for S1P-induced CrkII phosphorylation in HUVECs. We have not identified which of the VEGFR family proteins plays the principal role in the CrkII phosphorylation because antibodies highly specific to each of the VEGFR family proteins are not yet available. VEGFR-2 was phosphorylated by S1P, and CrkII phosphorylation was abolished by VEGFR-2 inhibitors. In addition, CrkII phosphorylation upon either VEGF or S1P stimulation was inhibited by CSK and PP2, indicating that VEGFR indirectly induces CrkII phosphorylation. Furthermore, S1Ppromoted membrane ruffling and cell motility were diminished by an inhibitor for VEGFR-2. This is consistent with the previous finding that VEGFR-2 is responsible for migration of HUVECs (60). Thus, among VEGFR family proteins, VEGFR-2 appears to be a candidate for S1P-induced CrkII phosphorylation and increase in motility of HUVECs.
In conclusion, we have delineated a novel S1P signaling pathway involving a sequential activation of G i -coupled receptor(s), VEGFR, Src family tyrosine kinase(s), and the CrkII adaptor protein. Furthermore, we have demonstrated that this novel signaling pathway is responsible for the membrane ruffling and for the increase in cell motility. FIG. 7. Spatio-temporal imaging of phosphorylation of CrkII upon S1P stimulation in HUVECs. A, HUVECs infected with adenovirus expressing Picchu were stimulated with 100 nM S1P. The emission ratio of YFP to CFP and the intensity of CFP were used for imaging of phosphorylation of Picchu in the intensity modulated display mode. Red and blue hues indicate high and low emission ratio, respectively. The intensity of each hue reflects the intensity of the fluorescence from CFP of Picchu. B, HUVECs expressing Picchu were stimulated with VEGF, and the membrane ruffles were magnified. Representative intensity modulated display images are shown.
FIG. 8. VEGFR is required for S1P-promoted membrane ruffling and cell motility. A, HUVECs were stimulated with 100 nM S1P for 20 min in the absence or presence of 20 M AG1433, an inhibitor for VEGFR kinase. Cells were imaged before and after S1P stimulation (left panels). The percent increase in cell size was analyzed by measuring the cell area before and after S1P stimulation in the absence or presence of AG1433 using a measurement tool included in the Meta-Morph 4.6 software (right bar graph). B, HUVECs labeled with fluorescein were unstimulated (left panel), stimulated with S1P (center panel), or stimulated with S1P in the presence of AG1433 (right panel). 100 cells were monitored for 6 h after stimulation, and their velocities were calculated using a cell tracking program. Cells moving faster than 1.5 m/min are shown in the black column. Cell labeling and velocity calculation were performed as described under "Experimental Procedures." FIG. 9. Crk is required for S1P-promoted membrane ruffling and cell motility of HUVECs. A, HUVECs infected with either adenovirus expressing GFP (Adeno-GFP) or both CrkI-W169L and EGFP (Adeno-CRKI DN) were stimulated with 100 nM S1P (left panel) or 30 ng/ml VEGF (right panel) and analyzed as described in the legend of Fig. 1. B, HUVECs transfected with pCXN2-FLAG-CrkI-W169L-IRES-EGFP (indicated by the arrows) and untransfected HUVEC were stimulated with 100 nM S1P. The phase contrast image and epifluorescent image for EGFP are shown overlaid at time 0 (min). Cells were exposed to S1P for the period indicated at the bottom of the figure. A representative overlaid image before SIP stimulation and phase contrast images after S1P stimulation are shown. Arrowheads indicate the membrane ruffling. C, HUVECs infected with adenovirus expressing GFP were unstimulated (left panel) or stimulated with S1P (center panel). Those infected with adenovirus expressing both CrkI-W169L and EGFP were stimulated with S1P (right panel). 150 cells were monitored for 6 h after stimulation, and their velocities were calculated using a cell tracking program included in the MetaMorph 4.6 software. Cells moving faster than 1.5 m/min are shown in the black column.