Protein Kinase C-ϵ Regulates Sphingosine 1-Phosphate-mediated Migration of Human Lung Endothelial Cells through Activation of Phospholipase D2, Protein Kinase C-ζ, and Rac1*

The signaling pathways by which sphingosine 1-phosphate (S1P) potently stimulates endothelial cell migration and angiogenesis are not yet fully defined. We, therefore, investigated the role of protein kinase C (PKC) isoforms, phospholipase D (PLD), and Rac in S1P-induced migration of human pulmonary artery endothelial cells (HPAECs). S1P-induced migration was sensitive to S1P1 small interfering RNA (siRNA) and pertussis toxin, demonstrating coupling of S1P1 to Gi. Overexpression of dominant negative (dn) PKC-ϵ or -ζ, but not PKC-α or -δ, blocked S1P-induced migration. Although S1P activated both PLD1 and PLD2, S1P-induced migration was attenuated by knocking down PLD2 or expressing dnPLD2 but not PLD1. Blocking PKC-ϵ, but not PKC-ζ, activity attenuated S1P-mediated PLD stimulation, demonstrating that PKC-ϵ, but not PKC-ζ, was upstream of PLD. Transfection of HPAECs with dnRac1 or Rac1 siRNA attenuated S1P-induced migration. Furthermore, transfection with PLD2 siRNA, infection of HPAECs with dnPKC-ζ, or treatment with myristoylated PKC-ζ peptide inhibitor abrogated S1P-induced Rac1 activation. These results establish that S1P signals through S1P1 and Gi to activate PKC-ϵ and, subsequently, a PLD2-PKC-ζ-Rac1 cascade. Activation of this pathway is necessary to stimulate the migration of lung endothelial cells, a key component of the angiogenic process.

In the vasculature, S1P is a key regulator of vascular maturation and angiogenesis under physiological and pathological conditions. Angiogenesis, or new blood vessel formation, is critical for normal embryonic vascular development and in tumor metastasis. Although targeted deletion of S1P 2 or S1P 3 in mice has no adverse effect on embryogenesis, deletion of S1P 1 caused failure of vascular development leading to a massive hemorrhage and embryonic lethality between E12.5 and E14.5 (24). Endothelial cell (EC) migration is an essential component of angiogenesis that is regulated by growth factors, bioactive molecules, and intracellular signaling (25). Among the various agonists, S1P has emerged as a potent angiogenic, and vascular maturation factor and considerable evidence exists for S1Pinduced endothelial cell proliferation (4), migration (26 -28), chemotaxis (29), and endothelial cell remodeling (30). Based on a number of studies using inhibitors, siRNA, dn mutants, or genetically engineered mice, it is becoming evident that several signaling pathways including Rho/Rac, phosphatidylinositol 3-kinase, Akt, MAPKs, PKC, and changes in intracellular Ca 2ϩ are involved in S1P-induced EC migration (3,7,8,12,31). We recently demonstrated that PLD activation by S1P regulates ERK1/2 activation (31) and interleukin-8 secretion in human bronchial epithelial cells (22,32). Furthermore, involvement of lipid phosphate phosphatase-1 in regulating lysophosphatidic acid (LPA)-induced phosphatidate (PA) generation and fibroblast migration suggests a role for PLD2 in fibroblast migration, wound healing, and tumor metastasis (33).
In ECs, very little is known regarding the role of S1P-induced PLD activation and generation of PA in cell migration, wound healing, and angiogenesis. Therefore, in the present study we investigated the role of S1P on human pulmonary artery endothelial cell (HPAEC) and established how the activation of the PKC isoform(s) is involved in upstream and downstream signaling of PLD1 and/or PLD2 in relation to the stimulation of cell migration. Our results show that physiologically relevant concentrations of S1P markedly stimulated HPAEC migration, which was sensitive to pertussis toxin (PTx) and a S1P 1 antagonist. Furthermore, evidence is provided for the role of PKC-⑀, but not PKC-, in S1P-induced PLD activation and the PLD2mediated stimulation of PKC-, Rac1, and cell migration.
Cell Culture-HPAECs (passage number 3) were purchased from Cambrex Inc. (Walkersville, MD) and cultured in complete endothelial growth medium (EGM)-2 medium (46). The cells (passage number 5-8) in 35-or 100-mm dishes or glass coverslips were used for all the experiments.
Endothelial Cell Migration-HPAEC were cultured in 12-or 6-well plates to ϳ95% confluence and then starved in the serum-free EGM-2 medium for 1-3 h or in EBM-2 medium containing 0.1% FBS for 18 -24 h. The cell monolayer was wounded by scratching across the monolayer with a 10-l standard sterile pipette tip. The scratched monolayer was rinsed twice with serum-free medium to remove cell debris and incubated with varying concentrations of S1P. The area (ϳ1 cm 2 total) in a scratched area was recorded at 0 and 16 -24 h using a Hamamatsu digital camera connected to the Nikon Eclipse TE2000-S microscope with ϫ10 objective and MetaVue software (Universal Imaging Corp.). Images were analyzed by the Image J software. The effect of S1P and other agents on cell migration/wound healing was quantified by calculating the percentage of the free area not occupied by cells compared with an area of the initial wound that was defined as closure of wounded area.
Electrical Cell Substrate Impedance Sensing (ECIS) Assay-HPAEC were cultured in 8-well ECIS electrode arrays (8W1E, Applied Biophysics, NY) to ϳ95% confluence and starved in the serum-free EBM-2 medium for 1-3 h. An elevated field (3 V at 40,000 Hz for 10 s) was applied to wound the cells on the electrode. Either complete medium or medium containing S1P (100 -1000 nM) was added, and wound healing was monitored for 10 -20 h by measuring the transendothelial electrical resist-ance using the ECIS equipment (11,47). In all experiments S1P was complexed with 0.1% BSA.
Infection of HPAECs with Adenoviral Vectors-cDNA for wild type and catalytically inactive mutants of PLD1, PLD2, and dominant negative Rac1 were subcloned into the pShuttle-CMV vector (32). The recombinant plasmid was linearized and transfected into HEK293 cells to generate replication-defective adenovirus. Generation of purified virus (10 10 plaque-forming units/ml) was carried out by the University of Iowa Gene Transfer Vector Core. Purified adenovirus (1-10 m.o.i. or plaqueforming units/cell) in complete EGM-2 medium was added to HPAECs grown to ϳ80% confluence in 6-well plates or 60-or 100-mm dishes. After 24 h, the virus-containing medium was replaced with complete EGM-2 medium. Vector control or infected cells were subjected to scratch and wound-healing ECIS assays, and immunoprecipitates or cell lysates were analyzed by Western blotting.
Measurement of PLD Activation by S1P-HPAECs in 35-mm dishes were labeled with [ 32 P]orthophosphate (5 Ci/ml) in phosphate-free DMEM for 18 -24 h at 37°C in 5% CO 2 and 95% air. Cells were then challenged with EBM-2 medium alone or EBM-2 containing S1P plus 0.1% BSA in the presence of 0.1% 1-butanol (22,32). In some experiments incubations were also carried out in the presence of 0.1% 3-butanol that served as additional controls. Incubations were terminated by the addition of 1 ml of methanol:HCl (100:1 v/v), cells were scraped into glass tubes, and lipids were extracted by the addition of 1 ml of methanol:HCl (100:1 v/v), 2 ml of chloroform, and 0.8 ml of 1 N HCl. [ 32 P]Phosphatidylbutanol (PBt), formed as a result of PLD activation and transphosphatidylation of [ 32 P]PA to 1-butanol, but not butan-3-ol, was separated from the total lipid extract by thin layer chromatography on 1% potassium oxalate plates with the upper phase of ethyl acetate:2,2,4-trimethyl pentane:glacial acetic acid:water (65:10:15:50 v/v) as the developing solvent system (22,32). Unlabeled PBt was added as carrier during separation of labeled lipids that were visualized by exposure to iodine vapor. Radioactivity associated with PBt was quantified by liquid scintillation counting, and all values were normalized to 10 6 dpm in total lipid extract. [ 32 P]PBt formed in control and S1P-challenged samples was expressed as dpm/dish or percent control.
Measurement of PKC-⑀ and PKC-Activation-HPAECs were cultured in 100-mm dishes to ϳ95% confluence, starved in EBM-2 medium containing 0.1% FBS for 3 h, stimulated with S1P for 5-10 min, washed with cold phosphate-buffered saline containing 1 mM vanadate, and lysed with 500 l of lysis buffer containing 20 mM Tris-HCl (pH 7.5), 150 mM NaCl, 1 mM Na 2 EDTA, 1 mM EGTA, 1% Triton X-100, 2.5 mM sodium pyrophosphate, 1 mM ␤-glycerophosphate, 1 mM Na 3 VO 4 , 1 mM dithiothreitol, 1 g/ml leupeptin, 1 g/ml aprotinin, and protease inhibitors from EDTA-free Complete tablets (Roche Applied Science). Cells were subsequently sonicated twice for 15 s and then centrifuged at 10,000 ϫ g for 15 min. Supernatants were collected and incubated overnight with polyclonal anti-PKC-⑀ or anti-PKC-antibody at 4°C. The immunoprecipitates were washed 3 times with lysis buffer and 2 times with kinase buffer (20 mM HEPES (pH 7.4), 25 mM ␤-glycerophosphate, 10 mM MgCl 2, 1 mM EGTA, 1 mM sodium orthovanadate, and 1 mM dithiothreitol) and resuspended in 100 l of kinase buffer. The activity of PKC was measured in 100 l of kinase buffer containing 25 g of myelin basic protein as an exogenous substrate to which 10 M ATP, 2 g of dioleoylglycerol, 12 g of phosphatidylserine, and 20 -40 l of immunoprecipitate were added. Incubations were carried out for 10 min at 30°C and terminated by the addition of 20 l of Laemmli sample buffer. Samples were then boiled for 5 min and analyzed for phosphorylation of myelin basic protein by Western blotting with antiphosphoserine antibody. HPAECs grown to ϳ95% confluence in 35-mm dishes were starved for 3 h in 0.1% FBS in EBM-2 without growth factors, and monolayers were scratched and challenged with medium containing 0.1% BSA or 1.0 M of S1P, dihydrosphingosine 1-phosphate (DHS1P), ceramide1-P (C8:0), ceramide1-P (C18: 1), and LPA for 16 h. In A and B, phase contrast images captured at 0 and 16 h after the scratch are shown. C shows the migration of cells into a "wound" that was scratched and exposed to various agonists previously. The values are the mean Ϯ S.E. for three independent experiments, and each experiment was carried out in triplicate.
Rac1 Activation Assay-HPAECs were cultured in 100-mm dishes to ϳ50% confluence for siRNA transfection or to ϳ95% confluence for adenoviral infection or inhibitor treatment. Cells were starved in EBM-2 medium containing 0.1% FBS for 3 h before stimulation with S1P for 2-15 min, cell lysates were subjected to immunoprecipitation with PAK-1 PBD, and Rac1 activation was evaluated using the Rac1 Activation assay kit as per the manufacturer's instruction (Upstate).
Immunofluorescence Microscopy-HPAECs grown on coverslips (18 mm) or chamber slides were starved for 3 h in EBM-2 containing 0.1% FBS before treatment with S1P (100 -1000 nM) for 5-60 min. Cells were fixed in 3.7% paraformaldehyde in phosphate-buffered saline for 10 min, washed 3 times with phosphate-buffered saline, permeabilized with methanol for 4 min at Ϫ20°C, blocked with 2% BSA in TBST, incubated for 1 h with appropriate primary antibody (1:200 dilution), washed with TBST, and stained for 1 h with secondary antibody (1:200 dilution) in TBST containing 2% BSA. Cells were examined using a Nikon Eclipse TE2000-S immunofluorescence microscope and a Hamamatsu digital camera with ϫ60 oil immersion objective and Meta Vue software.
RNA Isolation and Real Time RT-PCR-Total RNA was isolated from HPAECs grown on 35-mm dishes using TRIzol reagent according to the manufacturer's instruction. iQ SYBR Green Supermix was used to do the real time measurements using iCycler by Bio-Rad. 18 S (sense, 5Ј-GTAACCCGTT-GAACCCCATT-3Ј, and antisense, 5Ј-CCATCCAATCGG-TAGTAGCG-3Ј) was used as a housekeeping gene to normalize expression. The reaction mixture consisted of 0.3 g of total RNA (target gene) or 0.03 g of total RNA (18 S rRNA), 12.5 l of iQ SYBR Green, 2 l of cDNA, 1.5 M target primers, or 1 M 18 S rRNA primers in a total volume of 25 l. For all samples reverse transcription was carried out at 25°C for 5 min followed by cycling to 42°C for 30 min and 85°C for 5 min with iScript cDNA synthesis kit. Amplicon expression in each sample was normalized to its 18 S rRNA content. The relative abundance of target mRNA in each sample was calculated as 2 raised to the negative of its threshold cycle value times 10 6 after being normalized to the abundance of its corresponding 18 S rRNA (housekeeping gene) (2 Ϫ(primer threshold cycle) 2 Ϫ(18 S threshold cycle) ϫ 10 6 ). All primers were designed by inspection of the genes of interest using Primer 3 software. Negative controls consisting of reaction mixtures containing all components except target RNA were included with each of the RT-PCR runs. To verify that amplified products were derived from mRNA and did not represent genomic DNA contamination, representative PCR mixtures for each gene were run in the absence of the RT enzyme after first being cycled to 95°C for 15 min. In the absence of reverse transcription, no PCR products were observed.
siRNA Transfection-HPAECs grown to ϳ50% confluence in 6-well plates or chamber slides were transfected with Gene Silencer (Gene Therapy System, Inc. San Diego, CA)-transfecting agent containing scrambled siRNA (50 -100 nM) or siRNA for target proteins (50 -100 nM) in serum-free EBM-2 medium according to the manufacturer's recommendation. To optimize conditions for efficient transfection, HPAECs were transfected with Fl-Luciferase GL2 Duplex siRNA (target sequence. 5Ј-CGTACGCGGAATACTTCGA-3Ј, Dharmacon) as a positive control. After 3 h of transfection, 1 ml of fresh complete EGM-2 medium containing 10% FBS was added, and cells were cultured for additional 72 h and analyzed for mRNA levels by real time PCR or protein expression by Western blotting. Scrambled siRNA control or siRNA-transfected cells were subjected to scratch or wound healing experiments as described earlier.
Statistical Analysis-Analysis of variance and Student-Newman-Keul's test were used to compare the means of two or more different treatment groups. The level of significance was set to p Ͻ 0.05 unless otherwise stated. Results are expressed as mean Ϯ S.E.

S1P, but Not LPA or Ceramide-1-phosphate, Stimulates
Migration of HPAECs-To determine the role of exogenous S1P, dihydro-S1P, LPA, or ceramide 1-phosphate in EC motility, we employed two complimentary assays of wound closure that measure the non-directional, chemokinetic movement.
Among the four lipid phosphates tested, S1P was the most potent stimulator of HPAEC migration in both the scratch and ECIS wound healing assays (Fig. 1, A-C). Furthermore, the wound closure was dose-dependent, with ϳ50% of the wounded area remaining free of cells after exposure to S1P (10 nM) for 16 h as compared with ϳ25% with 100 nM S1P (supple- . S1P activates PKC-␦, -⑀, and -isoforms in HPAEC. HPAECs (ϳ95% confluence) were treated with EBM-2 alone or EBM-2 containing S1P (1 M) for 10 min, and cell lysates were subjected to immunoprecipitation (IP) with PKC-␣, -␦, -⑀, and -isoform-specific antibodies as described under "Experimental Procedures." In A, PKC activity was performed in the immunoprecipitates using myelin basic protein (MBP) as a substrate. Phosphorylation of myelin basic protein was determined by Western blotting with anti-phosphoserine antibody. Shown is a representative blot from three independent experiments. B shows quantification of the data from A, and values were normalized to total PKC isoform in the immunoprecipitate. The values are the mean Ϯ S.E. for three independent experiments. HPAEC (ϳ85% confluence) were infected with adenoviral constructs of vector control or dnPKC-␣, -␦, -⑀, andisoforms for 24 h before "scratch" migration assay with and without S1P. In A HPAECs were infected with vector control or dnPKC-␣, -␦, -⑀, and -isoforms (5 m.o.i.) for 24 h. The cells were scratched for the migration assay and challenged with EBM-2 medium alone or medium containing S1P (1.0 M) in the presence of 0.1% BSA. The migration of cells was recorded by imaging at 0 and 16 h after scratch, and the average area by which the gap between the ECs closed over 16 h was determined. The values are the mean Ϯ S.E. for three independent experiments in triplicate and expressed as % control. *, significantly different from cells challenged with EBM-2 without S1P (p Ͻ 0.05); **, significantly different from cells stimulated with S1P (p Ͻ 0.05); ***, significantly different from cells treated with S1P (p Ͻ 0.01). In B, HPAECs were pretreated with PKC-⑀ peptide inhibitor (10 M) or myristoylated PKC-peptide inhibitor (10 M) for 1 h before scratch assay with S1P, and cell migration was determined 16 h post-S1P (1.0 M) challenge. The values are the mean Ϯ S.E. for three independent experiments in triplicate and expressed as % of control. *, significantly different from cells challenged with EBM-2 without S1P (p Ͻ 0.01); **, significantly different from cells stimulated with S1P (p Ͻ 0.05); ***, significantly different from cells exposed to S1P (p Ͻ 0.001).
mental Fig. 1, A-D). Our results on S1P-induced cell migration are in agreement with a previous report (26). In contrast, LPA and ceramide 1-phosphate did not stimulate cell migration significantly in HPAECs, and this finding differs from an earlier report on LPA-induced migration of bovine aortic and human umbilical vein ECs (27). S1P 1 Is Required for S1P-induced HPAEC Migration-S1P exerts its action through G-protein-coupled S1P 1-5 receptors (7-9). To determine which S1P receptor subtype(s) is involved in S1P-induced cell migration, we first examined the expression of S1P receptors by real time RT-PCR using specific primers derived from human sequences (supplemental Table 1). Analyses of total RNA by real time RT-PCR showed that all the five S1P receptors were expressed in the order of S1P 1 Ͼ S1P 3 Ͼ S1P 2 Ͼ S1P 4 Ͼ S1P 5 (supplemental Fig. 1A). Complementing the mRNA expression profile, Western blot analyses showed S1P 1 to be abundantly expressed in HPAECs (supplemental Fig.  2B). Because S1P receptors are coupled to heterotrimeric G proteins (7-9), we examined the effect of treating HPAECs with 100 ng/ml of PTx for 3 h. This significantly blocked basal as well as S1P-mediated cell migration and wound healing (supplemental Fig. 2, C and D). These results suggest that S1P 1 is the predominant receptor expressed in HPAECs and the involvement of G i in transducing signals from S1P ligation to its receptors in cell migration.
To obtain evidence that S1P 1 is involved in S1P-induced cell migration and wound healing, we tested the ability of SB649146, which reduces constitutively active S1P 1 and blocks S1P binding (48), on S1P-mediated cell motility. Pretreatment of HPAECs with SB649146 (10 M) for 1 h reduced S1P-induced migration ( Fig. 2A). Furthermore, SB649146 decreased the basal migration of cells in the absence of added S1P (Fig. 2A). A similar effect of SB649146 on S1Pinduced EC migration was observed in ECIS wound healing assay (Fig. 2B).
To further characterize the role of S1P 1 in S1P-mediated cell migration, HPAECs were transfected with S1P 1 siRNA (100 nM) for 72 h, cells were wounded and challenged with S1P (1 M) for 16 h, and migration was evaluated. S1P 1 siRNA effectively down-regulated mRNA expression of S1P 1 , but not S1P 4 , and protein expression of S1P 1 (ϳ60%) (Fig. 2C). Furthermore, the S1P-induced migration of HPAECs was almost completely attenuated by knockdown of S1P 1 (Fig. 2D). These combined results establish the role for S1P 1 in regulating S1Pinduced cell migration in HPAECs.
PLD2, but Not PLD1, Is Involved in S1P-induced HPAEC Migration-To further characterize the role of PLD1 and PLD2 in S1P-induced cell migration and wound healing, HPAECs were infected with the adenoviral constructs (10 m.o.i.) encoding catalytically inactive hPLD1-K898R and mPLD2-K758R mutants for 24 h before S1P challenge. Overexpression of mPLD2-K758R, but not hPLD1-K798R, attenuated S1P-induced cell migration and wound healing, as evidenced by scratch and ECIS assay (Fig. 5, A-D). The role of PLD2 in cell migration was further investigated by knocking down PLD2 expression with PLD2 siRNA. HPAECs were transfected with PLD2 siRNA (50 and 100 nM) for 48 h, and the efficacy of the siRNA was determined by real time RT-PCR for mRNA levels and Western blotting for protein. Transfection of HPAECs with siRNA for PLD2 specifically blocked the mRNA expression of PLD2 without affecting PLD1 expression (Fig. 6A). Furthermore, treatment with PLD2 siRNA (50 nM) for 48 h attenuated S1P-induced cell migration in the scratch assay as compared with cells transfected with scrambled siRNA (Fig.  6B). In parallel experiments, transfection of cells with PLD1 siRNA had no effect on S1P-induced migration (results not shown). These results demonstrate that PLD2, but not PLD1, regulates S1P-induced migration of HPAECs.
PKC-⑀, but Not PKC-, Regulates S1P-induced PLD Activation-Having established a role for PKC-⑀ and -and PLD2 in S1P-induced migration, we further characterized the signaling cascades of PKC-⑀ and -in S1P-induced PLD activation using peptide inhibitors. HPAECs were incubated with PKC-⑀ peptide inhibitor (10 M, 2 h) or myristoylated PKC- peptide (10 M, 2 h). In parallel experiments, cells were labeled with [ 32 P]orthophosphate overnight before stimulation with S1P for activation of PLD. Pretreatment of cells with either PKC-⑀ peptide inhibitor or myristoylated PKCpeptide attenuated translocation of the respective isoform to cell periphery as determined by immunofluorescence microscopy (supplemental Fig. 4, A and B). Additionally, as shown in supplemental Fig. 4, C and D, the S1P stimulation of [ 32 P]PBt formation by treatment with S1P for 5 min was blocked by dnPKC-⑀ and PKC-⑀ peptide inhibitor but not by dnPKCand myristoylated PKCpeptide. These results demonstrate that S1P activated PKC-⑀ upstream of PLD and that PKCis downstream of PLD in HPAECs.
PLD2 siRNA Attenuates S1P-induced PKC-, but Not PKC-⑀, Activation in HPAECs-To further characterize the role of PLD2 in S1P-induced PKCactivation, HPAECs grown on glass coverslips or 35-mm dishes were transfected with PLD2 siRNA (50 nM) for 72 h, and cells were challenged with S1P for 5 min. Activation of PKC-⑀ and PKCwas then determined by immunocytochemistry or Western blotting. S1P treatment stimulated translocation of PKC-⑀ and PKCisoforms to the plasma membrane (Fig. 7, A and C), and activation of PKC-⑀ and -isoforms compared with control cells (Fig. 7, B and D). Similarly, knockdown of PLD2 with PLD2 siRNA blocked S1P-induced translocation and phosphorylation of PKC-, but not PKC-⑀ (Fig. 7,  A-D). These results further establish that PLD regulates S1P-induced PKC-, but not PKC-⑀, activation in HPAECs.
PLD2 Regulates S1P-induced HPAEC Migration via Rac1-Earlier studies demonstrated that Rac1 regulates S1P-induced EC migration (50); however, the role of PLD in Rac1 activation is largely unknown. Therefore, we characterized the link between PLD signaling and Rac1 activation in S1P-medi- FIGURE 7. Effect of PLD2 siRNA on S1P-induced activation of PKC-⑀ and PKC-in HPAECs. HPAECs grown to ϳ60% confluence in 35-mm dishes or glass coverslips were transfected with Gene Silencer reagent plus scrambled (sc) siRNA or Gene Silencer reagent plus PLD2 siRNA (50 nM) for 72 h as described under "Experimental Procedures." In A and C, cells transfected with scrambled or PLD2 siRNA, as indicated above, were challenged with EBM-2 or EBM-2 containing S1P (1.0 M) plus 0.1% BSA for 5 min, washed, fixed, permeabilized, and probed with anti-PKC-⑀ or PKC-antibodies. Shown are representative immunofluorescence micrographs from three independent experiments. The redistribution of PKC-⑀ or PKC-to cell periphery was quantified using an image analyzer as described under "Experimental Procedures." Values are the mean Ϯ S.E. from three independent experiments and are expressed as relative pixels. In B, cells transfected with scrambled or PLD2 siRNA were challenged with S1P (1.0 M) for 5 min, total cell lysates were subjected to immunoprecipitation (IP)with anti-PKC-⑀ antibody, and immunoprecipitates were assayed for PKC-⑀ activity with myelin basic protein as a substrate by Western blotting (WB) with anti-phosphoserine antibody. The results of S1P-induced phosphorylation of MBP by PKC-⑀ immunoprecipitates were calculated by densitometry. The values are the mean Ϯ S.E. for three independent experiments and are normalized to total PKC-⑀ in the immunoprecipitates. In D, cells transfected with scrambled or PLD2 siRNA were challenged with S1P (1.0 M) for 5 min, total cell lysates (250 g of total protein) were immunoprecipitated with anti-PKC-antibody, and immunoprecipitates were separated by SDS-PAGE and Western blotted with anti-phospho-PKC-and anti-PKC-antibodies. The results of S1P-induced phosphorylation of PKC-were calculated by densitometry. The values are the mean Ϯ S.E. for three independent experiments and are normalized to total PKC-in the immunoprecipitates. N.S., not significant.
ated HPAEC migration. S1P (1 M) activated translocation of Rac1 to cell plasma membrane in a time-dependent manner (Fig. 8A). Activation of Rac1 by S1P was also verified by immunoprecipitation of Rac1-GTP with PAK-1 PBD. S1P increased the activation of Rac1 with a peak at 2 min after S1P challenge (Fig. 8B). To demonstrate a role for Rac1 in cell migration, HPAECs were infected with adenoviral dnRac1 (1-10 m.o.i.) for 48 h or transfected with Rac1 siRNA (50 nM) for 72 h. Although infection of HPAECs with the dnRac1 at varying multiplicities of infection for 24 h resulted in overexpression of the protein (Fig. 8C), knockdown of Rac1 with Rac1 siRNA downregulated Rac1 expression by ϳ85% (Fig. 8D). Furthermore, as shown in Fig. 8, C and D, dnRac1 or Rac1 siRNA significantly reduced S1P-induced closure of wound.
Next, we investigated the role of PLD2 in S1P-induced Rac1 activation. HPAECs either on glass coverslips or 60-mm dishes were transfected with PLD2 siRNA (50 nM) for 72 h or infected with mPLD2-K758R mutant (10 m.o.i.) for 24 h. These treatments attenuated S1P-induced translocation of Rac1 to cell plasma membrane and decreased S1P-induced association of Rac1 with p21-activated protein (Fig. 9, A-D). These results demonstrate that PLD2 regulates S1P-induced activation of Rac1 in HPAECs.
PKC-Regulates S1P-induced Rac1 Activation-Having established a role for PLD2 in PKC and Rac1 activation by S1P, we determined the role of PKCin Rac1 stimulation. HPAECs were pretreated with myristoylated PKCpeptide (10 M) for 2 h before S1P challenge. As shown in Fig. 10, A and B, myristoylated PKCpeptide attenuated S1P-induced phosphorylation of PKCand S1P-mediated translocation of Rac1 to cell plasma membrane, respectively. These results demonstrate that PKCregulates S1P-induced Rac1 activation in HPAECs.

DISCUSSION
Migration of ECs is an important physiological process required for wound healing, repair of blood vessels, and angiogenesis during normal and also pathological conditions such as atherogenesis and balloon angioplasty. Several studies demonstrated a role for platelet-derived growth factor, vascular endothelial growth factor, and fibroblast growth factor as well as ECM-derived proteins such as fibronectin, vitronectin, collagen, and laminin in regulating EC migration and angiogenesis (25,51). Recently, S1P was shown to be a potent EC migratory factor (26,27,52) that promotes wound healing; however, the signaling mechanisms of S1P-induced EC migration in the pulmonary endothelium are poorly defined. The present work employed both a wound healing and transendothelial electrical resistance assay that detects mainly chemokinesis to characterize signaling pathways regulating S1P-mediated migration of human lung ECs. Our novel findings are that (i) S1P activates both PKC-⑀ and -isoforms in HPAECs, (ii) stimulation of PKC-⑀ by S1P, but not PKC-, activates PLD2, and (iii) activation of PLD2 by PKC-⑀ through S1P in turn stimulates migration via PKCand Rac1. Furthermore, we observed that S1P 1 is the predominant S1P receptor expressed in HPAECs, which is coupled to G i in transducing signals from S1P to intracellular targets such as PLD. S1P is a potent and pleotropic bioactive sphingolipid that mediates cellular responses such as proliferation, differentiation, tumor cell invasion, cell migration, and angiogenesis via G-protein-coupled S1P 1-5 (7)(8)(9). Earlier studies showed that S1P regulates cell motility, and whether it is stimulatory or inhibitory depends upon the cell type and the expression of HPAECs grown to ϳ90% confluence on glass coverslips, or 35-mm dishes were starved overnight in EBM-2 containing 0.1% FBS without any growth factors. In A, cells on coverslips were stimulated with EBM-2 or EBM-2 containing S1P (1.0 M) plus 0.1% BSA for 2, 5, and 15 min, washed, fixed, permeabilized, probed with anti-Rac1 antibody, and examined by immunofluorescence microscopy using a ϫ60 oil objective. Translocation of Rac1 to cell periphery was semiquantified using an image analyzer as described under "Experimental Procedures." The values are the mean Ϯ S.E. for three independent experiments. In B, activated Rac1 was immunoprecipitated (IP) from total cell lysates (500 g of total protein) from control and S1P (1.0 M)-treated cells using PAK-1 PBD-agarose beads as described under "Experimental Procedures." Rac-1-GTP bound to PAK-1 PBD were separated by SDS-PAGE, transferred to nitrocellulose, and probed with anti-Rac1 antibody. Shown is a representative blot from three independent experiments, and Rac1 activation was quantified from the Western blots (WB) by image analysis and normalized to total ␤-actin in the cell lysates. The values are the mean Ϯ S.E. from three independent experiments. In C, HPAECs grown to 80% confluence in 35-mm dishes were infected with empty vector or adenoviral vector containing cDNA for dnRac1 (1, 5,  different S1P-receptors. Although S1P stimulates migration of ECs (26,27), it inhibited chemotactic cell motility of several tumor cell lines (53,54) and vascular smooth muscle cells (55,56). S1P induces EC migration and angiogenesis in human umbilical vein (2) and bovine aortic ECs (26) that require the expression of S1P 1 and S1P 3 (2,10,12). The S1P-mediated migration of human umbilical vein ECs was sensitive to C3 exotoxin (10), and PTx (27,28) and required activation of integrin ␣ v ␤ 3 via Rho, but not Rac (10). Furthermore, in one study PKC inhibitor showed no effect on S1Pinduced migration of human umbilical vein ECs, but inhibition of phospholipase C abrogated the S1P response (57). Although vascular endothelial growth factorstimulated migration and proliferation of ECs was dependent on the nitric oxide-mediated decrease in PKC-␦ activity (58), sustained PKC-␦ activation by lysophosphatidylcholine also inhibited syndecan-4-dependent assembly/ disassembly of focal adhesions necessary for migration of bovine aortic ECs (59). In the present study we demonstrated a role for PKC-⑀ and -, but not -␣ and -␦, isoforms in S1P-induced migration of HPAECs using overexpression of dnPKC isoforms and peptide inhibitors. Our results show that dnPKC-⑀ and PKC-⑀ peptide inhibitor, but not dnPKCor myristoylated PKCpeptide, attenuated S1P-induced PLD activation in HPAECs. We observed some variation in the control levels of cell migration from ϳ25% (Figs. 1 and 2A and 6B) to ϳ50% (supplemental Figs. 2 and 4) when we used different batches of primary HPAECs. This variation can be attributed to differences in responses from individual donors. Nevertheless, our results consistently demonstrate distinct roles for PKC-⑀ and -in S1P-induced cell migration, and we show for the first time that S1P-mediated stimulation of PLD2 is regulated by PKC-⑀, whereas PKCactivation is downstream of PLD2. The values are the mean Ϯ S.E. for three independent experiments. Veh, vehicle. B shows Western blot (WB) analysis and quantification of Rac1 activation by S1P, as determined by PAK-1 PBD assay in scrambled and PLD2 siRNA-transfected HPAECs. The values are the mean Ϯ S.E. for three independent experiments normalized to total ␤-actin. IP, immunoprecipitation. In C, HPAECs (ϳ80% confluence on glass coverslips) were infected with empty vector or adenoviral vectors containing cDNA for mPLD2 mutant (mu, K758R) (10 m.o.i.) in complete EGM-2 medium for 24 h. Cells were challenged with EBM-2 plus 0.1% BSA or EBM-2 plus 0.1% BSA containing S1P (1.0 M) for 5 min, fixed as indicated above, and probed with anti-Rac1 antibody. Shown is a representative immunofluorescence micrograph from three independent experiments. Translocation of Rac1 to cell periphery was quantified using an image analyzer as described under "Experimental Procedures." The values are the mean Ϯ S.E. for three independent experiments. In D, cell lysates (500 g of total protein) from empty vector and mPLD2 mutant-infected cells, as described for C, were immunoprecipitated with PAK-1 PBD-agarose, and immunoprecipitates were subjected to SDS-PAGE and Western blotting with anti-Rac1 antibody. Activation of Rac1 was quantified by image analysis and normalized to ␤-actin in total cell lysates. The values are the mean Ϯ S.E. for three independent experiments. N.S., not significant.
We demonstrated earlier that S1P stimulated PLD1 and PLD2 activity in Beas-2B bronchial epithelial cells (22). In HPAECs, the S1P-induced PLD activation was PTx-sensitive and was blocked by SB649146 and S1P 1 siRNA, demonstrating the involvement of S1P 1 coupled to G i . Furthermore, the participation of PLD2, but not PLD1, in S1P-induced cell migration was established by expression of mPLD2-K758R and siRNA for PLD2 that blocked S1P-induced cell migration. By contrast, infection of cells with hPLD1-K898R or PLD1 siRNA had no effect on S1P-dependent cell migration. These results provide the first direct evidence for the participation of PLD2, but not PLD1, in S1P-induced EC migration. This work supports earlier studies with rat fibroblasts that showed a role for PLD2, but not PLD1, in LPA-induced migration (32). A potential explanation for this differential participation by PLD could be differences in the subcellular localization of PLD1 and PLD2 in mammalian cells. PLD1 is localized in the cytosol, Golgi apparatus, nucleus, and plasma membrane, whereas PLD2 is primarily found in the plasma membrane (41,60,61).
A role for PLD and PA generated via PLD has been implicated in the secretion of proinflammatory cytokines (22), activation of phagocytic NADPH oxidase (34), and regulation of neutrophil/fibroblast/cancer cell migration (33,62). It is not clear how PA generated by either PLD1 or PLD2 activation regulates cell migration. Our studies show for the first time that PLD regulates Rac1 via PKCand cell migration in HPAECs; however, the mechanism(s) of PLD/PA-dependent activation of PKCis yet to be defined. PKCis insensitive to diacylglycerol and Ca 2ϩ (36,63), but it can be activated by acidic lipids such as PA, phosphatidylinositol, and phosphatidylinositol 3,4,5-trisphosphate (64). Although PA can physically bind to and activate PKCin vitro and in vivo (38,39,65), it is unclear if PKChas any domain structure for PA binding. It has been reported that Rac-associated type I phosphatidylinositol-4-phosphate 5-kinase (PIP5K), but not type II PIP5K, is stimulated by PA (66). It is possible that stimulation of type I phosphatidylinositol-4-phosphate 5-kinase by PA in cells results in enhanced phosphatidylinositol-4,5-bisphosphate generation, which is phosphorylated by phosphatidylinositol 3-kinase to form phosphatidylinositol 3,4,5-trisphosphate. It is of interest that regulators of PLD1, namely RalA and Arf6 (67,68), are implicated in cell migration. The formation of stress fibers is linked to migration, and PLD activity increases stress fiber formation (69). Thus, factors that regulate PLD activity and responses to PLD signals co-ordinate and support cell motility in a variety of normal and cancer cells. Interestingly, our results show Rac1 is part of signaling cascade involved in S1P-mediated cell migration, and PLD2 regulates Rac1 activation via PKCin HPAECs. In addition to Rac1, infection of HPAECs with dn Cdc42 or Rho (5 m.o.i.) partially reduced S1P-induced cell migration, confirming a role for these Rho family of GTPases in migration. Rac1 transduces signals from agonists to induce changes in cell motility, cytoskeletal organization, cell morphology, and adhesion. Although several earlier studies showed that growth factors such as vascular endothelial growth factor and nerve growth factor caused Rac1 stimulated lamellipodia formation and cell migration (70,71), the signaling pathway(s) of S1Pmediated Rac1 activation has been not studied. S1P-stimulated Rac1 activation is linked to cytoskeletal organization and barrier regulation of human lung ECs (72). In this study we identified PKCas a downstream target of PLD2 in mediating S1P signaling to Rac1 in HPAECs. Our current results on PLD2-dependent activation of Rac1 are in contrast to an earlier report on Rac1-dependent regulation of PLD1b by antigens in RBL-2H3 cells (73). Furthermore, we observed that dnRac1 or Rac1 siRNA had no effect on S1P-induced PLD activation, translocation of PKC-⑀ and PKC-to cell periphery, and transient increase of [Ca 2ϩ ] i , further confirming that Rac1 is downstream to PLD, PKC-⑀/PKCactivation, and changes in intracellular Ca 2ϩ (results not shown). Thus, Rac1 is not a prerequisite for initiation of S1P signaling via S1P 1 /G i in HPAECs. FIGURE 10. Inhibition of S1P-induced Rac1 activation by myristoylated PKC-peptide in HPAECs. In A, HPAECs grown on 35-mm dishes (ϳ90% confluence) were pretreated with myristoylated PKC-peptide (10 M) for 2 h before S1P (1.0 M) challenge for 5 min. Cell lysates from control and S1P-treated cells (500 g of protein) were subjected to immunoprecipitation with anti-PKC-antibody, and immunoprecipitates were separated by SDS-PAGE, probed with anti-phospho (p) PKC-and anti-PKC-antibodies, and quantified by image analysis. Shown is a representative Western blot (WB) from three independent experiments. The values for phosphorylation of PKC-are the mean Ϯ S.E. from three independent experiments and are normalized to total PKC-in the immunoprecipitates. In B, HPAECs grown on glass coverslips (ϳ90% confluence) were pretreated with PKC--myristoylated peptide (10 M) for 2 h and then challenged with S1P (1.0 M) for 5 min. Cells were washed, fixed, permeabilized, and probed with anti-Rac1 antibody. A representative immunofluorescence micrograph from three independent experiments is shown. Quantification of Rac1 translocation was performed using image analysis software, and values are the mean Ϯ S.E. from three independent experiments. Veh, vehicle.
Furthermore, S1P-induced transient [Ca 2ϩ ] i was sensitive to PTx (ϳ85% inhibition) and SB649146 (ϳ80% inhibition) treatment, suggesting participation of S1P 1 linked to G i in the response (results not shown). Although, we have not studied the essential role of [Ca 2ϩ ] i in endothelial cell motility, studies carried out in HUVECs indicate that [Ca 2ϩ ] i signal is linked to S1P-induced cell migration via tyrosine phosphorylation of focal adhesion kinase (2,57). S1P-induced cell migration is not completely attenuated by blocking PLD2 with catalytically inactive mutant/siRNA (Figs. 5B and 6B), which may reflect an incomplete inhibition of PLD2. However, we predict that additional pathways independent of PLD2 might be involved in mediating cell motility in HPAECs. For example, in HPAECs S1P also promotes Tiam1/ Rac1 activation via phosphatidylinositol 3-kinase (PI3K) (74), suggesting a potential role for PI3K signaling in cell motility. Earlier, we demonstrated a role for phosphatidylinositol 3-kinase (PI3K) in S1P-induced EC migration (75); however, it is unclear if PI3K signals via PKCand/or Rac1 in cell migration. Interestingly, blocking phosphatidylinositol 3-kinase with LY294002 (1-25 M) did not attenuate S1P-induced PLD activation, indicating the absence of phosphatidylinositol 3-kinasedependent activation of PLD signaling by S1P in HPAECs. 4 Furthermore, specific guanine nucleotide exchange factors couple growth factor signaling to the Rho family of GTPases (71,76). The effect of vascular endothelial growth factor on Rac1-dependent motility of human umbilical vein ECs is regulated by Vav2 guanine nucleotide exchange factor (GEF) (69), and our preliminary results show that S1P-induced Rac1 activation is partly dependent on Tiam1, another GEF for Rac1. 4 However, the role of Tiam1 in PLD2-dependent regulation of Rac1 via PKC-is unclear, and future studies will address the role of guanine nucleotide exchange factors in S1P-induced cell motility. S1P is a natural component of plasma and is present in nM to M levels (7,46). Although stimulation of human lung EC migration by S1P is important for normal blood vessel function, S1P also stimulates neovascularization and tumor cell metastasis. Precise understanding of S1P-mediated signaling is important in developing novel therapeutic agents against targets regulating EC motility. Our identification of PKC-⑀ as an activator of PLD2 regulating S1P-induced migration and coupling PLD2 to Rac1 via PKC-indicates that PKC-⑀ and PLD2 could be potential targets for inhibiting excessive angiogenesis.