Introduction

Top Abstract Introduction Materials and Methods Results Discussion References

Angiogenesis is a complex biologic process involving a sequence of events that begins with endothelial cell detachment and invasion through the basement membrane, directional cell migration toward a chemoattractant gradient, proliferation, tube formation, and the establishment of a new intact vascular barrier. Several angiogenic factors have been identified, including vascular endothelial growth factor (VEGF), which ligates its specific receptor tyrosine kinases and promotes several aspects of the angiogenic process, including endothelial cell permeability, proliferation, and migration (1). Recently, several studies have clearly defined platelet phospholipids as potential angiogenic factors, most notably sphingosine-1-phosphate (S1P), which is released by platelets during blood clotting and is a potent and selective endothelial cell chemoattractant (2). S1P is found in nanomolar to micromolar concentrations in plasma and serum, accounts almost exclusively for the strong endothelial cell chemotactic activity of serum, and was remarkably more effective than optimal concentrations of VEGF and basic fibroblast growth factor (5). S1P is a potent stimulus for endothelial cell proliferation (6) and was strikingly effective in promoting angiogenesis in the developing chick embryo chorioallantoic membrane and the avascular mouse cornea (5). S1P binds to a family of G protein-coupled receptors encoded by the endothelial differentiation gene (Edg) that have been cloned and studied (7). Although intracellular accumulation may occur in a receptor-independent manner (8, 9), predominant biologic responses are due to ligation of G protein-coupled Edg receptors that regulate the stimulus/coupling events (10, 11). Thus, lysophospholipid growth factors are attractive angiogenesis factor candidates because they can be introduced in high levels from activated platelets at sites of endothelial cell injury.

Despite a marked increase in information regarding S1P effects on endothelial cell function, there is a paucity of information regarding the signal transduction events involved in S1P-mediated endothelial cell migration. In this study, we examined and contrasted the signaling mechanisms of S1P with VEGF on endothelial cell migration, an essential component of angiogenesis. Our studies indicate that at concentrations found in the serum, S1P shares with VEGF the ability to stimulate substantial macrovascular and microvascular endothelial cell chemotactic migration in a p38 mitogen-activated protein (MAP) kinase activity-dependent manner. However, compared with VEGF, S1P-mediated migration proceeds through Edg receptors that are coupled to a pertussis toxin (PTX)-sensitive G₁₂ protein involving Rho guanosine triphosphatase (GTPase). Furthermore, unlike VEGF, neither myosin light chain kinase (MLCK), p60^src, nor phosphatidylinositol-3' kinase activities are directly involved in the S1P response. These studies suggest that platelet-derived phospholipid growth factors use unique G protein-coupled cytoskeletal and motility signaling mechanisms to accomplish direct chemotaxis that may be involved in a critical linkage between coagulation, inflammation, and angiogenic processes.

	Materials and Methods

Top Abstract Introduction Materials and Methods Results Discussion References

Reagents

Chemicals, reagents and S1P were obtained from Sigma Chemical Co. (St. Louis, MO), unless noted otherwise. Transwell chemotaxis chambers (6.5 mm diameter, 8 µm pore size) were purchased from Costar, Inc. (Cambridge, MA). S1P was dissolved in methanol at 1 mM. Chemotaxis filters were soaked in 20 µg/ml rat-tail collagen (Boehringer Mannheim, Indianapolis, IN) in 0.1% acetic acid for 1 h at room temperature and dried overnight. Lyophilized recombinant human VEGF and recombinant human hepatocyte growth factor (HGF) (R&D Systems, Minneapolis, MN) were resuspended (100 µg/ml) in an aqueous solution of human serum albumin (0.4 mg/ml). PTX, UO126, SB203580, ML-7, LY294002, and PP2 were purchased from Calbiochem (La Jolla, CA). C3 exotoxin was from List Biological Laboratories (Campbell, CA). Y27632 was from Upstate Biotechnology (Lake Placid, NY). Myosin light chain (MLC) antibody was produced in rabbit against baculovirus-expressed and purified smooth muscle MLC by Biodesign International (Kennebunk, ME), rabbit anti-phospho-p44/42 MAP kinase (MAPK), rabbit anti-phospho-p38 MAPK, and p38 MAPK antibodies were purchased from New England Biolabs (Beverly, MA). Monoclonal anti-pan extracellular regulated kinase (ERK) antibody was obtained from Transduction Labs (Lexington, KY). Rabbit anti-Edg-1 antibody was purchased from Exalpha Biologicals (Boston, MA). Edg-1 and Edg-3 antisense oligonucleotides as well as the control oligonucleotides were synthesized by the DNA Analysis Facility at Johns Hopkins University (Baltimore, MD). FuGENE 6 Transfection Reagent was purchased from Boehringer Mannheim.

Bovine Pulmonary Artery Endothelial Cell Cultures

Bovine pulmonary artery endothelial cells were obtained frozen at 16 passages from American Type Culture Collection (CCL 209; Rockville, MD), cultured in complete medium (12), and used at passage 19-24. Bovine lung microvascular endothelial cells were purchased from Vec Tech Technologies (Troy, NY). All endothelial cell cultures were maintained in Dulbecco's modified Eagle's medium (Life Technologies, Rockville, MD), supplemented with 20% (vol/vol) colostrum-free bovine serum (Irvine Scientific, Santa Ana, CA), 15 µg/ml endothelial cell growth supplement (Upstate Biotechnology), 1% antibiotic and antimycotic solution (penicillin, 10,000 U/ml; streptomycin, 10 µg/ml; and amphotericin B, 25 µg/ml; K.C. Biologicals, Lenexa, KS), and 0.1 mM nonessential amino acids (Life Technologies). The endothelial cell cultures were maintained at 37°C in a humidified atmosphere of 5% CO₂/95% air and grew to contact-inhibited monolayers with typical cobblestone morphology. Cells from each primary flask were detached with 0.05% trypsin and resuspended in fresh culture medium and passaged into 100-mm or 35-mm dishes for transfection, 35-mm dishes for MAPK activity and MLC phosphorylation determinations, six-well plates for antisense treatments, or Transwells for migration measurements.

Endothelial Cell Chemotaxis

Assessment of endothelial cell migration was performed as recently described (5) with minor modifications. Endothelial cells (bovine macrovascular and microvascular) were dislodged after brief trypsinization and dispersed into homogeneous single cell suspensions that were washed extensively with M199/0.1% acid-free bovine serum albumin (migration medium) and resuspended in the same medium at 10⁶ cells/ml. To assess migration from established monolayers, cells (10⁵) were dispersed onto collagen-coated chemotaxis filters that partition Transwell inserts into upper and lower chambers. Migration medium (300 µl) was placed in the lower chambers and the cells allowed to adhere for 1 h at 37°C. In some experiments, media on both sides of filters were replaced with specific antagonists for 30 min after cell attachment. The medium in the lower chambers was then removed and cells were challenged by adding 300 µl fresh migration medium containing chemoattractants to the lower chamber. Unless otherwise indicated, migration was allowed to proceed for 2 h at 37°C. Cells remaining attached to the upper surface of the filters were carefully removed with cotton swabs, whereas cells migrating to the lower surface of the filters were fixed and stained with Diff-Quik staining set (Dade Behring Inc., Newark, DE) according to the manufacturer's recommendations. Filters were allowed to dry under a vented hood, carefully removed from Transwells, mounted on glass slides with Cytoseal mounting medium (VWR, Bridgeport, NJ), and finally examined by light microscopy (×20; Nikon Eclipse TE 300; Nikon, Melville, NY). Images of at least six consecutive fields horizontally across each filter were captured by a Sony Digital Photo camera (model DKC 5000; Sony, Tokyo, Japan). The number of migrating cells per field was enumerated and the average was calculated. Data were expressed as number of migrating cells per field, fold increase of stimulated migration over basal or percentage inhibition of chemoattractant-stimulated migration by antagonists, which represent the mean (± standard error [SE]) value from three separate experiments performed in duplicates within each experiment unless specified otherwise.

MLC Phosphorylation in Intact Endothelium

Endothelial cell monolayers grown in 35-mm dishes were analyzed for MLC phosphorylation by urea polyacrylamide gel electrophoresis (PAGE) as previously described by Garcia and coworkers (12), followed by Western immunoblotting with specific anti-MLC antibodies. The blot was scanned on a Bio-Rad densitometer (Bio-Rad Laboratories, Hercules, CA), and the percentage of MLC phosphorylation was determined by dividing the total of the phosphorylated and unphosphorylated areas. This method takes advantage of the fact that the mono- and diphosphorylated forms of MLC migrate more rapidly than does unphosphorylated MLC and are independent of sample loading. Stoichiometry (in mol/mol) was calculated with the formula [P₁ + 2(P₂)]/U + P₁ + P₂, where U is the percent unphosphorylated endothelial cell MLCK, P₁ is the percent monophosphorylated MLC, and P₂ is the percent diphosphorylated MLC. The diphosphorylated MLC is multiplied by a factor of 2 to reflect the presence of two phosphate groups per light chain. The percentage of light chain phosphorylation was calculated by adding the densitometric values of phosphorylation state for each isoform, i.e., unphosphorylated, monophosphorylated, and diphosphorylated.

MAPK Activation

MAPK activation was determined as previously described (13) by the immunoblotting of endothelial cell lysates with specific phospho-ERK and phospho-p38 antibodies, which indicate the enhanced catalytic activity of the enzymes. Briefly, after S1P or VEGF challenge, cells were lysed in 2× Laemmli's sodium dodecyl sulfate (SDS) sample buffer and boiled for 5 min. The lysates were separated by 12% SDS-PAGE and electrophoretically transferred to nitrocellulose membranes for Western immunoblotting as we have previously described (13). ERK1/2 phosphorylation was detected by 1 µg/ml of rabbit anti-phospho-p44/42 MAPK, p38 MAPK phosphorylation was determined by rabbit anti-phospho-p38 MAPK (1 µg/ml), total ERK protein was detected by monoclonal anti-pan ERK antibody (50 ng/ml), and total p38 MAPK protein was detected by rabbit anti-p38 MAPK antibody (1 µg/ml). After incubation with peroxidase-conjugated secondary antibodies (goat antirabbit immunoglobulin [Ig]G, 1:10,000 dilution [Sigma]; or goat antimouse IgG, 1:10,000 dilution [Bio-Rad Labs], immunoreactive proteins were visualized using an enhanced chemiluminescent detection system according to the manufacturer's directions (Amersham, Little Chalfront, Buckinghamshire, UK). The relative intensities of the protein bands were quantified by scanning densitometry.

Cotransfection of Specific Plasmids with Enhanced Green Fluorescent Protein and Subsequent Cell Sorting

To enrich transfected cells for determination of endothelial cell migration, we co-transfected the construct of interest (-adrenergic receptor kinase 1 [ARK1], G₁₂, and Rho GTPase dominant/negative [DN Rho] construct) with plasmids with enhanced green fluorescent protein (pEGFP) and selected green fluorescent protein (GFP)-positive cells as recently described (14). The ARK minigene plasmid (pRK-ARK1-495-689) contains the carboxyl terminus of ARK (the G binding domain) and was kindly provided by Dr. Walter J. Koch (Duke University, Durham, NC). Briefly, 50 to 60% confluent bovine endothelial cells plated in 100-mm dishes were incubated with DNA/FuGENE 6 mixture prepared as follows: 50 µl FuGENE 6 was diluted in 1 ml OPTI-MEM, incubated at room temperature for 15 min and added to 5 µg DNA containing 1 µg pEGFP and 4 µg ARK1, G₁₂, DN Rho, or their respective empty vector controls. The mixture was incubated at room temperature for 30 min and added to the dishes with 10 ml fresh complete medium. After 40 h incubation, cells were trypsinized, centrifuged, and resuspended in complete medium. Cells expressing GFP were isolated by fluorescence-activated cell sorting (FACS) using a FACStar^plus cell sorter (Becton Dickinson Immunocytometry Systems, San Jose, CA).

Phosphothioate Oligonucleotide Treatment

Endothelia grown in six-well plates to approximately 80% confluence were rinsed with OPTI-MEM, and 1 ml OPTI-MEM containing mixtures of either 100 µg of the antisense oligonucleotides specific for Edg-1 and Edg-3 as well as controls (sense or scrambled sequence) complexed with 10 µg lipofectamine (Life Technologies) was added to cells after 20 min incubation at room temperature. The cells were incubated with the mixtures of oligonucleotides and lipofectamine for 4 h and 1 ml culture medium was added to each well and incubation was continued for 20 h. Medium was then replaced by 1 ml M199 only containing 100 µg oligonucleotide. After 24 h incubation, cells were used in the migration assay. To confirm that Edg-1 expression was effectively reduced by Edg-1 antisense treatment, proximately 2.5 × 10⁵ cells, treated or untreated, were lysed with 80 µl SDS buffer. A total of 20 µl of the lysates was separated by 12% SDS-PAGE and immunoblotted with anti-Edg-1 antibody. The published sequences of the oligonucleotides used are the following: Edg-1 antisense, 5'-GAC GCT GGT GGG CCC CAT-3'; scrambled Edg-1, 5'-TGA TCC TTG GCG GGG CCG-3'; Edg-3 antisense, 5'-CGG GAG GGC AGT TGC CAT-3'; scrambled Edg-3, 5'-ATC CGT CAA GCG GGG GTG-3' (15).

	Results

Top Abstract Introduction Materials and Methods Results Discussion References

S1P Stimulates Bovine Lung Macrovascular and Microvascular Endothelial Cell Migration

We first contrasted the capacity of S1P and VEGF to induce the migration of bovine pulmonary artery endothelial cells (BPAECs) and bovine lung microvascular endothelial cells (BLMVECs) seeded on collagen-coated Transwell filters. Both agents increased endothelial cell migration in a dose-dependent manner as shown in Figure 1. S1P significantly induced BPAEC migration at concentrations as low as 10 nM with a maximum stimulation at 1 µM (8.9 ± 1.7-fold increase), whereas maximal VEGF-mediated migration was reduced when compared to S1P (2.8 ± 0.4-fold, 10 ng/ml) (Figure 1A). Further increases in the concentrations of both S1P and VEGF produced less migration, a common finding with chemotactic agents (16). Comparisons of microvascular and macrovascular endothelial cell migration in response to S1P or VEGF revealed similar concentration response curves; however, BLMVEC migration was less robust, achieving only 25 to 50% of the BPAEC level of migration at each concentration used (Figure 1B). Figure 1C depicts representative light microscopic images of BPAEC that have migrated to the abluminal surfaces of the Transwell membranes in response to S1P (1 µM), VEGF (10 ng/ml), and HGF (20 ng/ml). Whereas all three angiogenic factors demonstrated the capacity to induce endothelial cell chemotaxis, our results clearly demonstrate the exceptional potency of S1P as an endothelial cell chemoattractant.

View larger version (38K): [in this window] [in a new window]   View larger version (112K): [in this window] [in a new window]   Figure 1.   Endothelial cell migration toward S1P and VEGF. Both S1P- and VEGF-induced endothelial cells migrated in a dose-dependent manner. S1P-induced migration was maximal at 1 µM and VEGF induced maximum migration at 10 ng/ml both in bovine pulmonary artery endothelial cells (BPAEC) (A) and in bovine lung microvascular endothelial cells (BLMVEC) (B). Data represent the mean value of the fold increase of chemoattractant-induced migration over basal migration (± SE, n = 3 independent experiments performed in duplicate). (C) A light microscopic image of BPAECs (stained using Diff-Quik staining set) that have migrated to the abluminal surface of the 8 µm-filter Transwells. In comparison studies, the chemotactic response to S1P consistently exceeded that produced by VEGF and HGF. The concentrations of S1P, VEGF, and HGF are 1 µM, 10 ng/ml, and 20 ng/ml, respectively.

Role of Heterotrimeric G Protein and Edg Receptors in S1P- and VEGF-Induced Endothelial Cell Chemotaxis

To study the contribution of heterotrimeric G protein S1P-induced endothelial cell migration, we pretreated BPAEC with PTX (1 µg/ml, 1 h) to inactivate G_i proteins via adenosine diphosphate (ADP) ribosylation of  subunits as we have previously described (17). As shown in Figure 2A, PTX completely abolished S1P-induced endothelial cell migration, whereas VEGF-induced migration was preserved, consistent with the notion that VEGF induces endothelial cell activation through receptor tyrosine kinase signaling that is not G_i-linked. To further define the elements of G_i that are responsible for transducing the S1P migration signal downstream, we coexpressed the pEGFP with plasmids encoding either the carboxyl terminus of ARK (the binding domain) or a plasmid encoding a peptide that blocks G₁₂ function (G₁₂ minigene) (18). Cells expressing the reporter plasmid pEGFP were isolated using FACS to enrich the cells that express genes of interest where GFP overexpression alone does not affect S1P-induced migration (14). In these experiments, expression of the ARK minigene or the G₁₂ inhibitory minigene significantly inhibited S1P-stimulated endothelial cell migration (Figure 2B), whereas empty vector expression (pRK5 or pCDNA3.1, respectively) had no effect. Consistent with the results obtained with PTX pretreatment, these transfection experiments failed to alter VEGF-stimulated endothelial cell migration. These data suggest that each heterotrimeric G_i protein subunit (G₁₂ and G) participates in the S1P signaling cascade leading to endothelial cell migration.

View larger version (22K): [in this window] [in a new window]   View larger version (28K): [in this window] [in a new window]   View larger version (34K): [in this window] [in a new window]   Figure 2.   Edg-1 and Edg-3 receptor coupling via Gi2 and subunits is required for S1P-induced endothelial cell migration. (A) The role of Gi protein in S1P- and VEGF- induced migration is depicted. In these studies, BPAECs were incubated with PTX (1 µg/ml) for 1 h while attaching to Transwell filters. The medium in the lower chambers was then replaced with 1 µM S1P, 10 ng/ml VEGF, or vehicle and incubated for 2 h. Data represent numbers of migrating cells per field (20×) (mean ± SE from three independent experiments performed in duplicate). (B) The requirement for Gi2 and G subunit involvement in S1P-induced endothelial cell migration is shown. Subconfluent (50 to 60%) endothelial cells in 100-mm dishes were cotransfected with pEGFP and either Gi2, the ARK minigene, or their respective empty vectors. After 40 h of transfection, the GFP-positive cells were isolated by flow cytometry and subjected to the endothelial cell migration assay in the presence of S1P (1 µM), VEGF (10 ng/ ml), or vehicle. Data are presented as the percent inhibition of the chemoattractant-stimulated migration in minigene-transfected cells as opposed to vector-transfected cells. (C) The role of the Edg family receptors in S1P-mediated endothelial cell chemotaxis is shown. Bovine pulmonary endothelia grown in six-well plates were incubated with Edg-1 and Edg-3 antisense (18 mer oligonucleotides) or their respective control oligonucleotides (100 µg/ml, 48 h). Endothelial cell migration in response to S1P (1 µM) was significantly reduced by Edg-1 antisense and Edg-3 antisense but not by scrambled Edg-1 and Edg-3 controls, whereas migration to VEGF (10 ng/ml) was not affected. Data are presented as the percent inhibition of chemoattractant-induced migration produced by Edg antisense as opposed to scrambled oligonucleotide controls. (Inset) Edg-1 expression was significantly reduced by Edg-1 antisense treatment. Cell lysates from an equal number of cells (containing ~ 20 µg of total protein) were run on 12% SDS-PAGE and immunoblotted with anti-Edg-1 antibody (1 µg/ml). (1) Vehicle control. (2) Edg-1 antisense. (3) Scrambled Edg-1 oligonucleotides.

Because many, but not all, cellular events regulated by S1P are mediated through G protein-coupled Edg receptors, we next tested the hypothesis that S1P enhanced endothelial cell migration proceeds via Edg receptor ligation. Cells were incubated with antisense oligonucleotides (100 µg/ml) for 48 hrs to reduce the expression of either Edg-1 or Edg-3 and the results were compared with corresponding scrambled Edg-1 or Edg-3 oligonucleotides. In Edg-1 antisense-treated cells, S1P-stimulated migration was reduced by 50% compared with scrambled Edg-1 oligo-treated cells. Consistently, as evidenced by Western immunoblotting, the expression of Edg-1 protein was dramatically reduced after treatment with Edg-1 antisense but not with the scrambled Edg-1 oligo. In addition, migration of S1P-stimulated endothelial cells previously treated with Edg-3 antisense oligonucleotide was reduced by 25%. Predictably, neither Edg-1 nor Edg-3 antisense pretreatment altered VEGF-stimulated migration, indicating that specific S1P ligation of G protein-coupled Edg-1 and Edg-3 receptors evokes endothelial cell motility and angiogenic responses.

Effect of S1P and VEGF on Endothelial Cell MLC Phosphorylation and MAPK Activity

Endothelial cell migration is a complicated multistep process that includes cell cytoskeletal rearrangement and contraction, likely regulated via MLCK-dependent and -independent contractile signaling pathways (19, 20). For example, thrombin-induced endothelial cell contraction is highly dependent on the phosphorylation of MLCs by MLCK (12), whereas other agonists (phorbol 12-myristate 13-acetate, PTX) stimulate endothelial cell cytoskeletal rearrangement via signaling pathways that involve activation of mitogen-activated proline-directed kinases (MAPK) without an increase in MLCK activity (13, 21, 22). Therefore, to study the signaling pathways involved in S1P- and VEGF-induced endothelial cell migration, we initially investigated the effect of S1P and VEGF on MLC phosphorylation, ERK, and p38 MAPK activation. Endothelial cells challenged with either S1P (1 µM) or VEGF (10 ng/ml) for 2, 10, 30, and 60 min did not significantly increase MLC phosphorylation (data not shown), whereas thrombin (100 nM, 2 min) dramatically increased MLC phosphorylation, particularly with the increased stoichiometry of diphosphorylated MLCs. However, under identical conditions, S1P and VEGF stimulated significant MAPK activation as detected by Western blotting using phosphorylated ERK or p38 MAPK-specific antibodies (Figure 3B). S1P- but not VEGF-induced ERK activation was abolished by pretreatment with PTX (1 µg/ml, 1 h) (Figure 3B), suggesting ERK activation by S1P was G_i-dependent and therefore may participate in S1P-induced signaling leading to endothelial cell chemotaxis.

View larger version (37K): [in this window] [in a new window]   View larger version (43K): [in this window] [in a new window]   Figure 3.   S1P and VEGF activate ERK and p38 MAPKs but do not increase MLCK activity. (A) S1P and VEGF fail to alter MLCK activity or to increase MLC phosphorylation. BPAEC monolayers were incubated with S1P (1 µM, 10 min), VEGF (10 ng/ml, 10 min), or thrombin (100 nM, 2 min). MLCK activity was determined by the status of MLC phosphorylation. The unphosphorylated, monophosphorylated, and diphosphorylated MLC isoforms were separated on urea gel. Stoichiometry was calculated as described in MATERIALS AND METHODS. Data are mean ± SE (n = 3). As a positive control, thrombin greatly enhanced the diphosphorylation of MLC, implicating increased MLCK activity. (Inset) A representative urea gel electrophoresis. Additional times were evaluated (2, 30, and 60 min) and produced similar results, i.e., that MLC phosphorylation is not affected by either S1P or VEGF. (B) Both S1P and VEGF enhance ERK and p38 MAPK; however, only SIP-stimulated ERK activation was abolished by PTX. Endothelial cells challenged with S1P (1 µM, 10 min) or VEGF (10 ng/ml, 10 min), with or without PTX (1 µM, 1 h), were lysed and the lysates analyzed by Western immunoblotting. Activation of ERK was detected with anti-phospho-p44/42 MAPK, activation of p38 was detected with anti-phospho-p38 MAPK. Total MAPK protein was measured with anti-pan-ERK or anti-pan-p38 to ensure equal sample loading.

Regulation of S1P- and VEGF-Induced Signaling and Endothelial Cell Chemotaxis

We next examined the roles of ERK and p38 MAPK activation in S1P- and VEGF-induced endothelial cell migration. Pretreatment with UO126 (10 µM, 30 min), a potent inhibitor of the upstream ERK kinase, MEK, did not affect either S1P- or VEGF-induced endothelial cell migration (Figure 4), although S1P- and VEGF-stimulated ERK activation was completely abolished (data not shown). In contrast, pretreatment with SB203580 (20 µM, 30 min) to inhibit p38 MAPK activity, partially attenuated S1P-induced migration (40% reduction) and significantly reduced VEGF-induced migration (80% inhibition). We also investigated the involvement of other signaling pathways in S1P- and VEGF-mediated migration such as p60^src tyrosine kinase and phosphatidylinositol-3' kinase (PI-3' kinase). Treatment with the p60^src inhibitor PP2 (10 µM, 30 min) or PI-3' kinase inhibitor LY294002 (50 µM, 30 min) greatly reduced basal migration and completely abrogated VEGF-stimulated migration (100% inhibition; Figure 4B). In contrast, whereas the absolute number of migrating endothelial cells was reduced, the fold increase in endothelial cell migration evoked by S1P was unaffected (Figure 4A). Together, these findings strongly indicate differential regulation of S1P- and VEGF-induced signaling pathways leading to endothelial cell migration.

View larger version (20K): [in this window] [in a new window]   Figure 4.   Effect of protein kinase pharmacologic inhibitors on S1P- and VEGF-induced endothelial cell migration. Bovine endothelial migration responses to S1P (1 µM; A) and VEGF (10 µg/ml; B) were assayed after 30 min incubation with SB203580 (20 µM), an inhibitor of p38 MAPK activity, the MEK inhibitor UO126 (10 µM), the PI-3' kinase inhibitor LY294002 (50 µM), or the recently described selective src family tyrosine kinase inhibitor PP2 (10 µM). Data are presented as the percent inhibition of chemoattractant-induced migration by kinase inhibitors as compared with vehicle pretreatment. Both S1P and VEGF require p38 MAPK activity; however, S1P and VEGF differ substantially in each of the other signaling pathways evaluated, i.e., PI-3' kinase and p60src.

Regulation of S1P- and VEGF-Induced Cytoskeletal Rearrangement in Endothelial Cell Chemotaxis

Cellular locomotion is a complex, poorly understood process that depends on precisely orchestrated cytoskeleton reorganization (20). Because neither S1P nor VEGF produced a significant increase in MLC phosphorylation, we asked whether constitutive MLCK activity and MLC phosphorylation, the stoichiometry of which is generally measured to be ~ 0.4 mol/mol MLC in resting cells (12, 23), are critical to S1P- and VEGF-mediated endothelial cell migration. Although the levels of MLC phosphorylation were not altered by either S1P or VEGF, inhibition of MLCK activity by ML-7 attenuated VEGF-stimulated migration (~ 60% inhibition; Figure 5). As with p60^src inhibition, ML-7 treatment decreased the cell numbers that migrated to vehicle as well as S1P without altering the fold increase in migration stimulated by S1P, suggesting the independence of S1P-induced migration on MLCK activity. The Rho family of small GTPases also plays a critical role in the regulation of actin cytoskeleton dynamics and levels of MLC phosphorylation. Recent studies have found that interruption of the Rho GTPase signaling pathway inhibits angiogenesis in human umbilical vascular endothelial cells and inhibits human prostate cancer cell migration (24, 25). We next studied the functions of Rho and its major effector kinases, Rho-associated coiled-coil forming protein serine/threonine kinases (ROCK), in S1P- and VEGF-induced endothelial cell chemotaxis. Incubation with C₃ exotoxin (40 µg/ml, 48 h), known to ADP-ribosylate and consequently inactivate Rho GTPases (26), significantly reduced S1P-stimulated endothelial cell migration (Figure 5A), whereas VEGF-stimulated endothelial cell migration was unaffected (Figure 5B). Next, endothelial cells, FACS-enriched to express a plasmid encoding a DN Rho GTPase construct, exhibited significant decreased motility (30 ± 8% reduction) in response to S1P but not to VEGF challenge or cells transfected with the empty vector (pEF-BOS-HA). Finally, the direct involvement of Rho GTPase in S1P-induced chemotaxis was further confirmed by incubation with Y27632 (10 µM, 30 min) which inhibits both ROCK-I and ROCK-II by binding to the catalytic site with extremely high affinity (27). Rho kinase inhibition in this manner produced substantial decreases in S1P-induced migration (49 ± 4% inhibition), whereas VEGF-mediated chemotaxis was unaffected (Figure 5). Taken together, these findings strongly indicate the critical participation of Rho GTPase signaling pathways in S1P-activated endothelial cells leading to directed migration.

View larger version (17K): [in this window] [in a new window]   Figure 5.   Effect of MLCK and Rho GTPase inhibition on S1P- and VEGF-stimulated endothelial cell migration. BPAEC migration to S1P (1 µM; A) and VEGF (10 ng/ml; B) were assayed after incubation with the MLCK inhibitor ML-7 (10 µM, 30 min), Rho-associated kinase inhibitor Y27632 (10 µM, 30 min), Rho GTPase inhibitor C3 exotoxin (40 µg/ml, 48 h), or cotransfection with DN Rho and pEGFP. Data are presented as percent inhibition of chemoattractant-stimulated migration. These studies strongly indicate that Rho GTPase activity is an absolute requirement for S1P- but not VEGF-mediated chemotaxis. Conversely, MLCK activity, while not increased or required for the S1P response, was an essential component of VEGF-induced migration.

	Discussion

Top Abstract Introduction Materials and Methods Results Discussion References

S1P is a biologically active lipid generated by hydrolysis of glycerophospholipids and sphingomyelin in the membranes of activated cells and formed by the phosphorylation of sphingosine by sphingosine kinase. Although the fate of S1P is transient due to dephosphorylation to sphingosine (28), S1P is a normal constituent in plasma with submicromolar levels in serum consistent with the notion that S1P is released during whole blood coagulation. S1P exerts diverse effects on cells ranging from proliferation to suppression of caspase-dependent apoptotic activities and is now recognized as an important courier of cellular information. Only very recently, however, has S1P been unequivocally demonstrated to be a potent endothelial cell chemoattractant when released from activated platelets (3, 4). Furthermore, the cloning of a high affinity S1P receptor from endothelial cells stimulated to assume an angiogenic phenotype in vitro (29) suggests a potential role for S1P in angiogenesis (2). We recently reported direct examination of S1P-induced angiogenic activities and found S1P to be the primary angiogenic factor present in serum (4), capable of inducing angiogenesis in avian chorion and the avascular mouse cornea while potentiating growth factor-mediated angiogenesis in vivo. Whether the mechanism by which S1P exerts this angiogenic activity is shared by other angiogenic agents has remained unclear. In this report, we have contrasted the regulation of endothelial cell chemotaxis evoked by S1P to that elicited by the well-accepted angiogenic factor VEGF. Unlike VEGF, S1P produced endothelial chemotaxis via G₁₂-linked Edg-1 and EDG-3 receptor ligation with downstream signaling to Rho GTPase and p38 MAPK activities, indicating that angiogenic responses occur via agonist-specific signaling pathways.

Even though intracellularly generated S1P exerts a number of biologic responses (8), consistent with a growing literature, our data indicate that S1P-mediated endothelial cell chemotaxis follows ligation of the Edg family of receptors. Edg receptors are developmentally regulated and differ in tissue distribution and include Edg-1, -3, -5, and possibly -8 (30). Our data indicate Edg-1 and Edg-3 as primary S1P receptors involved in bovine endothelial cell chemotaxis (Figure 2C), consistent with expression studies in human endothelial cells where Edg-1 and Edg-3 were abundantly expressed and Edg-5 expression was undetectable (15). Furthermore, most in vitro (4) and in vivo (15) angiogenic activity elicited by S1P was inhibited by Edg-1 and Edg-3 antisense oligonucleotides, indicating that the role of Edg-5 in the angiogenic and chemotactic responses is questionable.

Edg family receptors have been variably described as coupled to three heterotrimeric G proteins, G_i, G_12/13, and G_q (33). We have provided several lines of evidence to indicate that S1P signals involved in endothelial chemotaxis are transduced via Edg-G₁₂ coupling. S1P-activated endothelial cell migration was abolished by pre-incubation with PTX (Figure 2A), demonstrating the involvement of a G_i-coupled receptor. Furthermore, two complementary strategies to interrupt G protein signaling via transfection with plasmids encoding either a specific G₁₂ inhibitory peptide or a ARK competing peptide (which inhibits signaling) significantly reduced S1P- but not VEGF-induced endothelial cell migration (Figure 2B). The targets involved in S1P-mediated endothelial cell migration are unknown. Ras and subsequently p42/p44 ERK MAPK are well recognized as activated by signaling, and both VEGF and S1P significantly increased the activity of ERK MAPK (Figure 3B). Our data are consistent with several prior studies in COS cells and smooth muscle that S1P activates ERK MAPK in a PTX-sensitive G_i-dependent manner. However, profound attenuation of ERK activity consistently failed to alter either VEGF- or S1P-mediated migration (Figure 4). These results confirm recent studies in human aortic endothelium and fetal bovine heart endothelial cells that demonstrated significant migration and ERK activation in response to S1P but only a minimal effect of ERK inhibition on endothelial cell migration (6, 34).

The subunits have also been demonstrated to target the activation of p60^src and PI-3' kinase pathways, stimulus/coupling events that participate in cell migration responses to specific agonists. Binding of VEGF to its receptor tyrosine kinase promotes recruitment and activation of PI-3' kinase subunits (35), which leads to phosphorylation and activation of PI-3' kinase-dependent serine/threonine kinase Akt (36). Akt directly phosphorylates bovine endothelial nitric oxide synthase (eNOS) on serine 1179, resulting in eNOS activation and nitric oxide production (37), actin reorganization, and consequently, cell migration (38, 39). In our studies, however, the pharmacologic modulation of these pathways failed to reduce S1P-mediated endothelial cell chemotaxis, whereas the VEGF response was particularly susceptible to PI -3' kinase inhibition (Figure 4). The Edg receptors are also reported to couple through G proteins linked to tyrosine kinases such as p60^src (40) and S1P activates p60^src in a PTX-dependent fashion (41). However, despite significant inhibition of src kinase activity and reduction in basal endothelial cell chemotaxis by the p60^src inhibitor PP2, our results do not indicate a major role for p60^src activity in S1P-induced chemotaxis as the magnitude of S1P-induced endothelial cell migration was not altered (Figure 4).

Our results confirm the critical role of the cytoskeleton in mediating endothelial chemoattractant responses and highlight the differential Rho GTPase-dependent and -independent signaling pathways involved in the motility response. As we failed to demonstrate an increase in MLC phosphorylation with either VEGF or S1P (Figure 3A), on first blush, it would appear that our data indicate the lack of either MLCK or Rho kinase involvement in endothelial cell motility. However, inhibition of either MLCK or Rho kinase, two enzymes that work in concert to increase MLC phosphorylation, produced significant agonist-specific reduction of endothelial cell chemotaxis. For example, several strategies to reduce Rho activities decreased S1P chemotaxis (Figure 5), whereas VEGF was not affected, suggesting both overlapping and distinct pathways for accomplishing cell locomotion. These studies also indicate the possibility that the presumably highly spatially localized activities of these enzymes at focal adhesion may escape biochemical detection when using whole cell lysates. Miura and colleagues (42), using much higher concentrations of S1P, recently detected slight increases in monophosphorylated MLC in human umbilical endothelia. Unlike S1P signaling, we failed to find direct evidence supporting a role of Rho GTPase signaling pathways in VEGF-stimulated chemotaxis. Gingras and associates (43) recently reported that transient overexpression of a dominant/active Rho A mutant increased the tyrosine phosphorylation of VEGF receptor (VEGFR)-2. However, overexpression of a DN Rho A had no effect on VEGFR-2 phosphorylation and activation, which is consistent with our finding that overexpression of DN Rho A did not influence VEGF-induced endothelial cell migration.

Our data indicate that concentrations of S1P less than 1 µM activate G₁₂-coupled Edg receptors, leading to stimulated migration in a Rho kinase-dependent manner; however, the exact manner by which S1P-Edg receptor ligation activates Rho GTPase and the mechanism of Rho regulation of endothelial cell chemotaxis is unknown. It is of considerable interest to us that recent reports indicated that Edg-1, the preferred receptor for S1P, activates members of the G_i family but not G_s, G_q, or G_12/13 (11). S1P is also capable of binding to Edg-3 and Edg-5, which, unlike Edg-1, may be coupled to G_q or G_12/13 (10), thereby explaining the ability of S1P to activate Rho. Conversely, oligomers may also participate in Rho recruitment (44). Whether Edg-3 is coupled to the G_12/13, which then mediates activation of the Rho guanosine triphosphate pathway and thereby increases cytoskeletal rearrangement, is not clear.

Finally, our data indicate a significant role for p38 MAPK activity in both VEGF- and S1P-induced endothelial cell migration, findings consistent with the report of Rousseau and coworkers (49) that demonstrated the requirement of p38 MAPK in VEGF-induced endothelial chemotaxis. We have shown recently p38 MAPK to be a critical signaling system in producing endothelial cell actin cytoskeletal rearrangement, paracellular gap formation, and endothelial cell permeability (J. G. N. Garcia and colleagues, and A. D. Verin and associates, unpublished data). The p38 MAPK activation elicited by S1P appears to occur via a receptor-G_i protein pathway as PTX completely inhibited this response, but the cytoskeletal targets remain unidentified. p38 MAPK activates the cytoskeleton via MAP kinase-activated protein (MAPKAP)-dependent phosphorylation of actin binding proteins such as HSP27 (50, 51). VEGF-induced endothelial cell migration appears to require the concerted activation of p38-mediated HSP27 activation (49); however, whether Edg-1 or Edg-3 ligation by S1P mediates endothelial cell motility via similar p38 MAPK pathways involving cytoskeletal targets such as HSP27 requires further examination.

In summary, we have compared signaling events elicited by two potent angiogenic factors and endothelial cell chemotaxins, S1P and VEGF. S1P, when released from platelets at sites of blood coagulation, would appear to act in an autocrine or juxtacrine pathway through ligation of Edg-1 and Edg-3 receptors linked to downstream effectors that include p38 MAPK and Rho kinase, but not p60^src, PI-3' kinase, or MLCK. Given that S1P belongs to a growing list of key endothelial cell growth factors that regulate vascular function and homeostasis, S1P participation in platelet-endothelial cell interactions may be critical to understanding the complexity of coagulation, wound healing processes, and angiogenesis and pathologic responses to macrovascular damage.