Syndecan-4 Is Required for Thrombin-induced Migration and Proliferation in Human Vascular Smooth Muscle Cells*

Thrombin is a mitogen and chemoattractant for vascular smooth muscle cells (SMCs) and may contribute to vascular lesion formation. We have previously shown that human SMCs, when stimulated with thrombin, release basic fibroblast growth factor (bFGF), causing phosphorylation of FGF receptor-1 (FGFR-1). Treatment with bFGF-neutralizing antibodies (anti-bFGF) or heparin inhibits thrombin-induced DNA synthesis. We concluded that thrombin may stimulate entry into the cell cycle via bFGF release and FGFR-1 activation. In the present study, we demonstrate a requirement for not only FGFR-1 but also syndecan-4, a transmembrane heparan-sulfate proteoglycan. Inhibition of syndecan-4 expression using small interfering RNA (siRNA) resulted in reduced DNA synthesis by human SMCs after stimulation with thrombin (10 nmol/liter). Anti-bFGF antibody, which inhibits DNA synthesis in control cells, had no inhibitory effect when syndecan-4 expression was reduced by siRNA. Thrombin- or bFGF-induced SMC migration, determined in Boyden chamber assays, was reduced in cells treated with syndecan-4 or FGFR-1 siRNA or by anti-bFGF. Thrombin induced phosphorylation of extracellular signal-regulated kinase (ERK) 1/2 in a biphasic pattern. Although thrombin-mediated ERK phosphorylation at 5 min was not affected by syndecan-4 or FGFR-1 siRNA, ERK phosphorylation at later time points was reduced. We conclude that thrombin-released bFGF binds to syndecan-4 and FGFR-1, which is required for thrombin-induced mitogenesis and migration.

Vascular smooth muscle cell (SMC) 1 proliferation and migration are key events in atherosclerosis and restenosis after vas-cular injury (1,2). The G-protein-coupled receptor agonist thrombin may contribute to disease progression through stimulation of mitogenic signaling. In a previous study with cultured human SMCs, we demonstrated that thrombin releases basic fibroblast growth factor (bFGF), which is bound to the pericellular matrix and causes phosphorylation of the FGF receptor-1 (FGFR-1) (3). Preventing FGFR-1 phosphorylation by bFGF-neutralizing antibodies or by heparin inhibits thrombin-induced mitogenesis. Since heparin binds bFGF and inhibits bFGF-FGFR-1 interactions, we hypothesized that an endogenous heparan sulfate proteoglycan (HSPG) may be involved in mediating the effects of thrombin-released bFGF and thus be required for thrombin-induced mitogenesis.
Syndecan-4 is a transmembrane HSPG with a core protein of 35 kDa carrying mainly heparan sulfate side chains (4). Syndecans and the HSPG glypican have been shown to bind bFGF and increase bFGF-FGFR-1 interactions (5). Of the family of syndecans (syndecan-1 to -4), syndecan-4 is the only member with an intracellular domain that is capable of binding phosphatidylinositol 4,5-bisphosphate and protein kinase C ␣ (PKC␣), which allows it to support PKC-mediated signaling (6 -8). Recently, several unique functions have been described for syndecan-4. It clusters in focal contacts and affects integrin function, thereby modulating cell adhesion, migration, and morphology (9 -13). Since thrombin activates FGFR-1 by releasing bFGF into the pericellular matrix and syndecan-4 functions as a mediator for bFGF signaling and a cofactor for FGFR-1 (for a review, see Ref. 13), we tested the possibility that thrombin-induced mitogenesis and migration require syndecan-4 for binding thrombin-released bFGF.

EXPERIMENTAL PROCEDURES
Materials-Antibodies (Abs) against syndecan-4 and FGFR-1 were from Santa Cruz Biotechnology (Santa Cruz, CA). Ab against phosphorylated ERK1/2 was from Cell Signaling Technology (Beverly, MA). Ab to an HSPG neo-epitope, exposed after heparitinase III digestion (anti-⌬HS, clone 3G10), was from Seikagaku Corp. (Tokyo, Japan). Protein A-agarose was from Roche Diagnostics. Neutralizing Ab against human bFGF was a generous gift from Dr. Michael A. Reidy (University of Washington) (14). ERK1/2 antiserum was a generous gift from Dr. Karen Bornfeld (University of Washington). Human ␣-thrombin was from American Diagnostica (Greenwich, CT). Recombinant bFGF, heparin (porcine intestinal mucosa), and heparitinase I, II, and III were from Sigma. All cell culture solutions were from Invitrogen.
Cell Culture-Human aortic SMCs were prepared by the explant technique as described previously (15). Cells were cultured in Dulbecco's modified Eagle's medium supplemented with 15% fetal bovine serum, 200 units/ml penicillin, and 200 g/ml streptomycin. Cells at passage 5-12 were used for experiments.
Cells were transfected with siRNAs (final concentration 10 nmol/ liter) by calcium phosphate precipitation for 15 h, as described (16). Cells were washed three times with phosphate-buffered saline, once with medium containing 15% fetal bovine serum, and allowed to recover for at least 9 h. Cells were detached with trypsin, counted, and reseeded for experiments in 15% serum.
For Western blot analysis, samples were boiled in Laemmli sample buffer and subjected to SDS-PAGE followed by transfer to nitrocellulose. Primary antibodies used were monoclonal Ab against syndecan-4 (clone 5G9, Santa Cruz Biotechnology), anti-⌬HS (clone 3G10, Seikagaku Corp.), anti-ERK1/2 (Cell Signaling), anti-actin (Sigma), and anti-FGFR1 (Santa Cruz Biotechnology). Protein was visualized by enhanced chemiluminescence (kit from Amersham Biosciences) according to the manufacturer's protocol. Results were quantified by densitometry of films using ImageQuant software (Amersham Biosciences).
DNA Synthesis-SMCs (7,500 -8,750 cells/cm 2 ) were incubated in serum-free medium for 48 h with a change of medium 24 h before stimulation. Cells were stimulated by the addition of thrombin (10 nmol/liter) or bFGF (1 ng/ml). [ 3 H]Thymidine (1 Ci/ml) was added 16 -18 h after stimulation. After 26 -28 h, cells were washed three times with ice-cold phosphate-buffered saline followed by incubation in 10% trichloroacetic acid overnight at 4°C. Cells were washed in trichloroacetic acid, and DNA was solubilized in 0.1 N NaOH. Radioactivity was measured by liquid scintillation counting.
Boyden Chamber Migration Assay and Cell Spreading-Modified Boyden chamber assays were performed for 5 h at 37°C with 48-well microchemotaxis chambers (Neuro Probe) and polycarbonate filters (10-m pores; Nucleopore Corp.) coated with monomeric collagen (100 g/ml Vitrogen 100 in 0.1 mol/liter acetic acid, Collagen Corp., Palo Alto, CA), as described (15). Cells (20,000/well) were added to the upper chamber, and chemoattractants or serum-free medium were added to the lower chamber. Migration was measured as the number of cells/high power field (ϫ100) that had migrated across the membrane. Assays were performed in quadruplicate. For cell spreading experiments, cells were seeded on glass slides coated with monomeric collagen (100 g/ml Vitrogen 100 in 0.1 mol/liter acetic acid) and allowed to spread for up to 6 h in serum-free medium. Cells were fixed and stained with Diff-Quick staining solution (Baxter, Detroit, MI) according to the manufacturer's instructions.
Statistics-All experiments were performed at least three times in duplicate, triplicate, or quadruplicate. Statistical analysis was performed using one-way ANOVA followed by Bonferroni's multiple comparison test or by a paired two-tailed t test as indicated. Values of p Ͻ 0.05 were considered significant.

Inhibition of Syndecan-4 and FGFR-1 Expression by siRNA in Human Vascular
SMCs-Specific siRNA significantly reduced RNA levels for syndecan-4 and FGFR-1 in human SMCs, respectively, but had no effect on RNA of syndecan-2 or glypi-
can, two proteoglycans that can both bind bFGF (5). RNA oligonucleotides with a scrambled sequence derived from syndecan-4 or FGFR-1 siRNA had no effect on RNA levels (Fig. 1).
Syndecan-4 protein levels were determined in extracts from human aortic SMCs by Western blotting. Although the specific siRNA inhibited syndecan-4 protein expression by over 90%, a matched scrambled RNA oligonucleotide had no effect on syndecan-4 expression ( Fig. 2A). To determine whether other HSPGs in the cell layer were affected by syndecan-4 siRNA, an antibody (anti-⌬HS, clone 3G10) that detects an HSPG neoepitope after heparitinase digestion was used on Western blots of SMC lysates prepared as described under "Experimental Procedures" (Fig. 2B). Multiple bands, similar to those described by others using this antibody (20), were detected. A double band of 35-37 kDa, presumably syndecan-4 core protein, was strongly reduced in cells treated with syndecan-4 siRNA. These data indicate that whereas other HSPGs are present, only syndecan-4 is altered by siRNA to syndecan-4. An effective siRNA was also generated to inhibit FGFR-1, as determined in immunoprecipitates from SMC lysates (Fig. 2C).
Cell Spreading Is Impaired in SMCs with Reduced Syndecan-4 Expression-Because syndecan-4 is involved in the organization of the cytoskeleton (9, 21), we investigated whether syndecan-4 siRNA affects cell spreading. When compared with controls, syndecan-4 siRNA-treated cells spread less at 1 and 3 h in serum-free medium. However, by 6 h, all cells were equally spread on collagen (Fig. 3). Cell spreading in 15% serum appeared to be normal, and also, the number of cells that attached in 15% serum overnight was not decreased in syndecan-4 siRNA cells (Fig. 3B) when compared with control cells. These data suggest that the lack of syndecan-4 delays cell spreading on collagen. Delayed spreading was not observed in cells with decreased FGFR-1 expression, indicating distinct functions of syndecan-4 and FGFR-1.

Syndecan-4 and FGFR-1 Are Required for Thrombin-and bFGF-induced Migration-In
Boyden chamber assays, thrombin-or bFGF-induced migration of SMCs transfected with syndecan-4 siRNA was impaired when compared with the migration of scrambled RNA-treated control cells (Fig. 4A). In contrast, serum-induced migration was not affected by syndecan-4 siRNA. SMCs transfected with FGFR-1 siRNA also migrated less toward thrombin than control cells, and unlike control cells, bFGF-neutralizing antibodies had no inhibitory effect on thrombin-induced migration (Fig. 4B). The migration pattern of untreated SMCs was not different from control cells treated with scrambled RNA matched for either syndecan-4 or FGFR-1 siRNA (data not shown).
Syndecan-4 and FGFR-1 Are Required for Thrombin-induced bFGF-dependent DNA Synthesis-SMCs transfected with syndecan-4 or FGFR-1 siRNA or with matched scrambled RNA were stimulated with thrombin or bFGF. In cells treated with syndecan-4 siRNA, DNA synthesis in response to thrombin or bFGF was reduced (Fig. 5A). A bFGF-neutralizing antibody inhibited thrombin-induced DNA synthesis in control cells but had no inhibitory effect when syndecan-4 expression was reduced by siRNA. Also of interest, inhibition of thrombin-induced DNA synthesis by heparin was only significant in control cells. A nonspecific IgG had no effect in either experimental group (Fig. 5A).
The effects of reduced FGFR-1 expression were similar to those observed with reduced syndecan-4 expression. Thrombininduced DNA synthesis was inhibited in cells with reduced FGFR-1 expression, and bFGF-neutralizing antibodies exerted no further inhibitory effect (Fig. 5B). Heparin significantly inhibited DNA synthesis only in control cells. In contrast to control cells, stimulation with recombinant bFGF of cells treated with FGFR-1 siRNA did not result in increased DNA synthesis.
Late Phase of Thrombin-induced ERK1/2 Phosphorylation Requires Syndecan-4 and FGFR-1-To determine whether ERK 1/2 is involved in syndecan-4-and FGFR-1-controlled intracellular pathways that are activated by thrombin, nontreated SMCs and SMCs transfected with scrambled RNA, syndecan-4 siRNA, or FGFR-1 siRNA were stimulated with thrombin from 5 min to 4 h. The extent of ERK phosphorylation at 5 min in control cells and siRNA cells was the same, whereas at later time points, it was reduced in SMCs transfected with either syndecan-4 siRNA (Fig. 6A) or FGFR-1 siRNA (Fig. 6B). DISCUSSION In previous studies, we and others have shown that thrombin-induced mitogenesis and migration require the activation of a secondary ligand-receptor system (15,22). In rat SMCs, thrombin transactivates the EGF receptor by releasing hepa-rin-binding EGF-like growth factor (HB-EGF) from the cell surface (15). In contrast, we recently demonstrated that in human SMCs, thrombin does not transactivate the EGF receptor. Instead, thrombin causes rapid release of bFGF into the pericellular matrix with subsequent FGFR-1 phosphorylation, which can be blocked by bFGF-neutralizing antibodies and by heparin (3). In the present study, we have demonstrated that the cell surface heparan sulfate proteoglycan syndecan-4 is required for thrombin-induced mitogenesis and migration by using syndecan-4 siRNA to specifically decrease syndecan-4 core protein (Fig. 2B). We found that human SMCs with reduced syndecan-4 or FGFR-1 migrate and proliferate less in response to thrombin or bFGF (Figs. 4 and 5). The magnitude of this effect is similar to the level of inhibition of thrombinmediated mitogenesis obtained by treating normal cells with bFGF-neutralizing antibodies (3). In contrast, the bFGF-neutralizing antibody did not alter thrombin-or bFGF-mediated migration and proliferation of FGFR-1 knockdown SMCs. These data indicate that the thrombin-induced bFGF-dependent pathway is mediated by both syndecan-4 and FGFR-1.
Syndecan-4, as well as syndecan-1 and -2 and glypicans, can bind bFGF and increase bFGF-FGFR-1 interactions (5). In addition, syndecan-4 is known to play a unique role in bFGFdependent signal transduction (9,13,23). The cytoplasmic tail of syndecan-4 forms a complex with phosphatidylinositol 4,5bisphosphate and PKC␣, which promotes PKC␣ activation (6 -8). It has been proposed that a serine/threonine protein phosphatase becomes activated after bFGF binds to its tyrosine kinase receptor. This phosphatase dephosphorylates the cytoplasmic tail of syndecan-4, increasing its affinity for phosphati- dylinositol 4,5-bisphosphate and promoting dimerization (and possibly higher order multimers) of the syndecan-4 cytoplasmic tail. This in turn increases the binding and activation of PKC␣ (24). It is known that thrombin-induced signaling involves activation of PKC isoforms and translocation of PKC␣ into focal domains (25). Therefore, since syndecan-4 activates PKC␣ (6 -8), it is possible that thrombin may exert its effects on migration and proliferation, in part, via promoting bFGF and syndecan-4 dependent PKC␣ activation.
Thrombin also induces signal transduction through the MAPK pathway, and prolonged ERK activation is required for cell cycle progression induced by various mitogens, including thrombin (26). In rat SMCs, the EGF receptor co-stimulatory pathway was required to sustain ERK activity beyond 30 min (15). In human SMCs, we found that this sustained activation of ERK by thrombin required syndecan-4 and FGFR-1 (Fig. 6), in part explaining the role of syndecan-4 and FGFR-1 in thrombin-mediated SMC proliferation (Fig. 5). It would be of interest to further investigate the mechanisms involved in syndecan-4 dependent bFGF-mediated ERK activation.
Our observation of impaired spreading of syndecan-4 siRNA cells on collagen-coated slides (Fig. 3) is in agreement with previous reports that syndecan-4 is involved in cell spreading through the assembly of focal contacts (9,13). This may explain the reduced migration of these SMCs (Fig. 4). Although we did not directly investigate focal adhesion formation, the observation that syndecan-deficient cells were fully spread by 6 h suggests that human SMCs have sufficient compensatory mechanisms to form focal adhesions. This is consistent with observations using fibroblasts from syndecan-4-deficient animals. Although these cells are able to form normal focal adhesions (13,27), focal adhesion formation is impaired when the cells are cultured on the cell binding fragment of fibronectin in the presence of culture medium with the heparin-binding fragment of fibronectin (27). Of interest, these syndecan-4-deficient mice develop normally, but when challenged, they exhibit impaired wound healing and decreased angiogenesis in the granulation tissue (28,29). Thus, the specific role of syndecan-4 for focal adhesion formation, especially in human cells, is uncertain and requires further study.
In summary, we demonstrate a requirement for syndecan-4 and FGFR-1 for thrombin-induced bFGF-dependent migration, mitogenesis, and sustained ERK activation in human vascular SMCs.