The chondroitin sulfate moiety mediates thrombomodulin-enhanced adhesion and migration of vascular smooth muscle cells

Background Thrombomodulin (TM), a transmembrane glycoprotein highly expressed in endothelial cells (ECs), is a potent anticoagulant maintaining circulation homeostasis. Under inflammatory states, TM expression is drastically reduced in ECs while vascular smooth muscle cells (VSMCs) show a robust expression of TM. The functional role of TM in VSMCs remains elusive. Methods We examined the role of TM in VSMCs activities in human aortic VSMCs stimulated with platelet-derived growth factor-BB (PDGF-BB). Using rat embryonic aorta-derived A7r5 VSMCs which do not express TM, the role of the chondroitin sulfate (CS) moiety of TM in VSMCs was delineated with cells expressing wild-type TM and the CS-devoid TM mutant. Results Expression of TM enhanced cell migration and adhesion/spreading onto type I collagen, but had no effect on cell proliferation. Knocking down TM with short hairpin RNA reduced PDGF-stimulated adhesion and migration of human aortic VSMCs. In A7r5 cells, TM-mediated cell adhesion was eradicated by pretreatment with chondroitinase ABC which degrades CS moiety. Furthermore, the TM mutant (TMS490, 492A) devoid of CS moiety failed to increase cell adhesion, spreading or migration. Wild-type TM, but not TMS490, 492A, increased focal adhesion kinase (FAK) activation during cell adhesion, and TM-enhanced cell migration was abolished by a function-blocking anti-integrin β1 antibody. Conclusion Chondroitin sulfate modification is required for TM-mediated activation of β1-integrin and FAK, thereby enhancing adhesion and migration activity of VSMCs. Electronic supplementary material The online version of this article (10.1186/s12929-018-0415-7) contains supplementary material, which is available to authorized users.


Background
Migration and proliferation of vascular smooth muscle cells (VSMCs) are key events in the intimal thickening of atherosclerosis and neointimal formation induced by vascular injury [1]. During vascular remodeling, VSMCs in the media transform from a quiescent, contractile phenotype into an activated, synthetic phenotype and migrate to the intima. VSMCs in the intima may undergo replication, active production of extracellular matrix (ECM) or even secretion of pro-inflammatory factors, resulting in intimal thickening and atherosclerosis progression. Aberrant migration and proliferation of VSMCs can be triggered by excessive secretion of various cytokines and growth factors [2]. Thrombomodulin (TM), a transmembrane glycoprotein in endothelial cells (ECs), is a critical player in maintaining vascular thromboresistance. TM acts as a cofactor in protein C activation through regulating thrombin substrate affinity and plays a critical role in the anticoagulant pathway [3]. Under physiological conditions, TM is highly expressed on the endothelial surface and is considered an endothelial marker. In disease states associated with inflammation, such as atherosclerosis and vein graft thrombosis, TM expression in ECs is markedly downregulated, leading to thrombotic complications and an increase in local inflammation and matrix proteolysis [4,5]. Interestingly, in advanced atherosclerotic plaques, TM is detected in VSMCs of both the intima and the media [6,7]. Thus, intimal VSMCs were proposed to be a relevant source of TM and form a surface relatively resistant to thrombosis. We previously found that platelet-derived growth factor (PDGF)-BB, the most potent mitogen and chemoattractant of VSMCs in vitro, induces TM expression in VSMCs through Src kinase and PI3-kinase/Akt/mTOR-dependent signaling pathway. Furthermore, PDGF-BB and TM co-localize in the media and the neointima following carotid ligationinduced injury [8]. Similarly, TM expression is induced in distension-caused urinary bladder injuries and is involved in the PDGF-stimulated migration of urinary bladder smooth muscle cells [9]. Recent evidence further showed that the forced expression of TM in TM-deficient A2058 melanoma cells enhances cell adhesion, migration, and invasion through the activation of focal adhesion kinase (FAK) [10].
TM consists of five domains, including an N-terminal lectin-like domain, followed by six epidermal growth factor (EGF)-like repeats, a serine/threonine-rich domain, a transmembrane domain, and a cytoplasmic domain [11]. The activation of protein C requires thrombin binding to the EGF-like domain and is enhanced by a second site binding to the chondroitin sulfate (CS) moiety attached to the serine/threonine-rich domain [12]. In addition to an anticoagulant function, previous studies have revealed multiple biological roles for TM including anti-inflammation [13], cell-cell adhesion [14], cell-matrix adhesion and migration [10], cell proliferation, and angiogenesis [15], which are linked to different domains of TM. While TM peptides consisting of EGFlike repeats and serine/threonine-rich domains stimulate cell proliferation, migration, and angiogenesis [15], whether the CS moiety of TM participates in these processes remains unclear. Previous studies have suggested that proteoglycans containing CS play a significant role in regulating cell adhesion and migration in a variety of cell types [16][17][18][19]. The CS moiety on CD44, a transmembrane glycoprotein, modulates cell migration on type I collagen in melanoma cells, which is abolished by chondroitinase ABC (ChABC), an enzyme that cleaves CS from the core protein [16]. Similarly, CD44-associated CS proteoglycan mediates EC migration and adhesion on fibrinogen [17]. In addition, CS proteoglycans regulate cell growth, differentiation, and migration of neural precursor cells [20].
We hypothesized that the CS moiety of TM has functional roles in VSMCs adhesion and migration analogous to the CS of other membrane proteins. We tested our hypothesis by examining the effect of ChABC and the CS-devoid TM mutant expression on VSMCs adhesion and migration. Our results indicate that TM expression promotes cell spreading, adhesion, and migration involving the activation of β 1 -integrin and FAK and the CS moiety is required for these processes.

Lentivirus-based knockdown of TM expression
pLKO.1-puro-shTM, pCMV-ΔR8.91 and pMD.G were obtained from National RNAi core facility (Academia Sinica, Taiwan). The lentivirus stock was prepared as previously described [21]. HASMCs were infected with a viral stock in complete medium for 5 days and were processed for the subsequent experiments.

Transient transfection
A7r5 cells at 40-60% confluence were transfected with Fugene HD (Roche Applied Science) at a 2.5:1 ratio to plasmid DNA. The amount of DNA was adjusted if necessary. The medium was replaced 12 or 24 h later, and the cells were harvested 48 or 72 h later. Transfection efficiency was consistently 30-50% as determined by green fluorescent protein expression.

Site-directed mutagenesis
TM DNA within the pEGFP-N1 vector was mutated at serine 490 and 492 to alanine using the QuikChange™ site-directed mutagenesis kit from Stratagene (La Jolla, CA, USA) following the manufacturer's instructions. DNA sequencing confirmed the mutations.

Adhesion assay
The ability of VSMCs to adhere to ECM molecules was quantified as previously described [23]. Clear-bottom 96-well plates were coated with 40 μg/ml collagen type I or 5 μg/ml fibronectin in PBS for 16 h at 4°C. Nonspecific binding was blocked with 2% BSA in PBS. HASMCs were stimulated with 10 ng/ml human recombinant PDGF-BB (R&D Systems) for 6 h before adhesion assay. A7r5 cells were transfected with 1 μg of pEGFP, pEGFP-TM, or pEGFP-TM S490,492A for 12-24 h and cultured in regular medium for 36-48 h. Cells were trypsinized, washed, and treated both with and without ChABC (0.5 U/ml) for 1 h at 37°C. The cells (10 4 cells/100 μl) were added to the coated 96-well plates and incubated for 30 min at 37°C. Non-adherent cells were removed by washing with DMEM. Adherent cells were fixed with 4% paraformaldehyde, stained with 0.1% crystal violet for 25 min, and solubilized with 0.5% triton X-100. The optical density at 595 nm (OD 595) was determined using a microplate reader.

Wound healing assay
Fully confluent cell cultures were wounded with a 100-μl sterile pipette tip as previously described [24]. Wound healing assay was conducted in a CO 2 incubator. Each scratch was randomly photographed at four separate sites along the length with time-lapse video microscopy over the succeeding 24 h. Migration velocity of A7r5 cells expressing GFP, TM-GFP, or TM S490, 492A -GFP across the wound edges was measured manually. In some experiments, a hamster IgM (clone Ha2/5, BD Pharmingen) or the control IgM (G235-1) was used. The migration rate of HASMCs was determined by the ratio of the wound area at 24 h to the initial wound area.

Invasion assay
To determine the invasive ability of A7r5 cells, the upper sides of the transwell polycarbonate membrane filters, with 8-μm pore size (Corning Inc.), were coated with 40 μg/ml collagen type I or 5 μg/ml fibronectin. Cells (20,000/well) in serum-free medium were seeded in the upper chamber, and the bottom chamber contained medium with 10% FBS. Cells were incubated for 16 h at 37°C. Following incubation, the cells that had invaded and attached to the lower surface of the membrane were fixed with 4% paraformaldehyde and stained with 4′,6diamidino-2-phenylindole (DAPI, diluted 1000× in PBS) (Sigma-Aldrich). Cell number was counted in five random fields per membrane under an inverted microscope.

TM and pFAK immunofluorescence
HASMCs were serum-starved for 36 h and treated with or without PDGF-BB (10 ng/ml) for 6 h. A7r5 cells were transfected with pEGFP, pEGFP-TM or pEGFP-TM S490,492A (1 μg) for 12 h and cultured in fresh medium for 36-60 h. Cells were detached and seeded onto 40 μg/ml type I collagen-coated cover slip for 3 h. The cells were fixed for 10 min with 4% paraformaldehyde and permeabilized for 15 min with 0.1% Triton X-100 in phosphate buffered saline (PBS)/BSA. The cells were incubated at 4°C overnight with anti-TM (mouse clone D-3, Santa Cruz Biotechnology) or anti-pFAK Y397 antibody (BD611807, 1:100 in 3% BSA/PBS), followed by Alexa 546-conjugated goat anti-mouse IgG (Molecular Probes; 1:200). The cell nuclei were stained with DAPI, and cells were observed under a confocal microscope (Leica TCS SP5 II).

Statistical analysis
Data are presented as mean ± SEM of n independent experiments. Statistical analysis was performed with Student's t test for comparison between two groups. For comparisons among multiple groups, one-way ANOVA, followed by Dunnett multiple comparison was used. P values smaller than 0.05 were considered significant.

VSMCs expressed TM both with and without chondroitin sulfate (CS) moiety
We previously reported that HASMCs express TM under PDGF stimulation but not at quiescence [8]. On the contrary, A7r5 cells did not express TM mRNA in the presence or absence of PDGF treatment (Additional file 1: Supplementary data, Table S1 and Figure  S1). Therefore, we applied both HASMCs and A7r5 cells to examine the functional roles of TM in VSMCs. We first examined TM expression in PDGF-stimulated HASMCs and A7r5 cells transfected with TM cDNA. Compared to quiescent cells (Fig. 1a, Lane 1, serum starvation for 48 h), PDGF treatment profoundly increased TM expression in HASMCs. TM mainly existed as a1 00 kDa form, but a diffused, high-molecular-mass band of approximately 180-200 kDa was also present (Fig. 1a, Lane 2). TM possesses four potential sites for O-linked glycosylation, which supports the post-translational attachment of a CS moiety, a stretch of approximately 20 repeating disaccharide units with a trisaccharide terminus [25]. HASMCs treated with ChABC (0.5 U/ml) substantially reduced the high-molecular-mass form (Fig. 1a, Lane 3), indicating that TM expressed in HASMCs was modified by CS.
The CS moiety of TM is indispensable for TM to perform its anticoagulant activity [12]. However, additional roles for the CS moiety of TM have not been defined. To address this question, we generated the TM mutation of Ser 490 and Ser 492 to Ala, which was shown to block the CS modification of TM [26]. Transfection with pEGFP-TM S490, 492A had no effect on TM expression (Fig. 1b, Lanes 2 and 4). A7r5 cells expressing TM-GFP exhibited a pattern similar to that of PDGF-stimulated HASMCs with both bands showing higher molecular mass resulting from the GFP tag. The high-molecularmass bands were eliminated with ChABC treatment. In addition, A7r5 cells expressing TM S490, 492A exhibited only the low-molecular-mass band. These results suggested that TM underwent CS modification in both HASMCs and A7r5 cells through serine 490 and 492. The cellular localization of TM in PDGFstimulated HASMCs and TM-expressing A7r5 cells was examined. TM was mainly localized on the plasma membrane (Fig. 1c, b & d) and CS-deficient TM S490, 492A displayed a similar distribution to wild-type TM (Fig. 1c, e).
The CS moiety mediates TM-enhanced VSMCs adhesion and spreading TM functions as a Ca 2+ -dependent cell-to-cell adhesion molecule [14] and was recently found to bind to fibronectin through the lectin-like domain [10]. To assess the effect of TM on VSMCs adhesion to the ECM, we performed cell adhesion assay on plates coated with type I collagen which is a major ECM in the neointima [27]. HASMCs were infected with a lentiviral vector encoding luciferase shRNA (shLuc) or TM shRNA (shTM). The infection with shTM, but not shLuc, for 3 to 5 days markedly decreased TM expression in HASMCs, showing the specificity of the engineered lentivirus (Fig. 2a). The adhesion of TM-depleted HASMCs was markedly suppressed (41.0% ± 9.6% of control, P < 0.05) (Fig. 2b). Cell surface CD44-related CS proteoglycan was reported to mediate melanoma cell adhesion and migration on type I collagen [16]. Therefore, we hypothesized that TM affects the adhesive and/or migratory capabilities of VSMCs through the CS component and examined it with A7r5 adhesion onto type I collagen. Expression of TM-GFP significantly increased A7r5 adhesion (Fig. 2c, 163.5% ± 15.9% of GFP, P < 0.01), which was reversed by ChABC treatment (26.3% ± 5.7% of TM-GFP, P < 0.05). Moreover, expression of TM S490, 492A -GFP had no effect on cell adhesion onto type I collagen. To examine whether TM-enhanced cell adhesion of VSMCs is specific to type I collagen, we compared A7r5 cells adhesion to type I collagen-, fibronectin-, and non-coated surface. TM expression markedly enhanced cell adhesion to both ECM molecules with stronger effect detected when fibronectin was used. In contrast, expression of TM S490, 492A -GFP inhibited cell adhesion onto fibronectin. As a comparison, expression of another TM mutant with lectin-like domain deletion (TMΔL-GFP) enhanced cell adhesion to both molecules less effectively compared to TM-GFP expression (Additional file 1: Supplementary data, Figure S2). These results indicated that TM promotes VSMCs adhesion and the CS moiety mediates the effect. Next, we examined whether TM affects cell spreading. A7r5 cells expressing GFP, TM-GFP or TM S490, 492A -GFP were re-suspended and allowed to adhere onto immobilized type I collagen for 30 min. Cells expressing TM-GFP exhibited accelerated spreading and appeared flattened with extended lamellipodia. In contrast, most GFP-and TM S490, 492A -GFP-expressing cells remained round. The spreading area of TM-expressing cells was increased (2.3 ± 0.18 fold control), whereas cells expressing CS-deficient TM S490, 492A showed no difference in spreading area (Fig. 2d).

The CS moiety is required for TM-enhanced VSMCs migration
VSMCs migration across the internal elastic lamella into the intima is a key process in neointima formation [28]. To investigate the role of TM in VSMCs migration, wound healing assay was performed under PDGF treatment for 24 h in HASMCs infected with lentivirus encoding luciferase or TM shRNA. Knocking down TM expression in HASMCs attenuated wound closure compared to the control group (Fig. 3a, TM knockdown: 40.4% ± 6.1%; Luc knockdown: 70.5% ± 9.7%, P < 0.05). To further elucidate the role of the TM CS moiety in VSMCs migration, A7r5 cells expressing TM-GFP, TM S490, 492A -GFP or GFP were analyzed with wound-healing assay.
Cells expressing TM-GFP exhibited markedly higher migration velocity (8.6 ± 0.72 μm/h) compared with the GFP control (5.0 ± 0.3 μm/h), whereas cells expressing TM S490, 492A -GFP (6.3 ± 0.9 μm/h) showed no effect on cell mobility (Fig. 3b). We further examined the effect of expressing TM-GFP, TM S490, 492A -GFP, and TMΔL-GFP on A7r5 cells invasion through type I collagen-and fibronectin-coated transwells. Similar to that found with adhesion assay, TMenhanced cell invasion was more pronounced with fibronectin coating. Expression of TM S490, 492A -GFP inhibited cell invasion through fibronectin-coated wells, whereas TMΔL-GFP was less effective than TM-GFP in enhancing cell invasion (Fig. 3c). These results indicated that the CS moiety mediates TM-promoted mobility of VSMCs.

TM-enhanced VSMCs migration involves β 1 -integrin and focal adhesion kinase
CS on CD44 is implicated in mediating EC adhesion and migration [17]. Furthermore, another CS-containing glycoprotein, PG-M/versican, binds to β 1 -integrin, thereby activating FAK and cell adhesion [29]. We next explored whether β 1 -integrin is involved in TM-increased cell mobility. A7r5 cells expressing GFP-tagged TM or TM S490, 492A were applied to wound healing assay in the presence of PDGF and control IgM or a function-blocking antibody against rat integrin β 1 (Ha 2/5). Blocking integrin β 1 activation eliminated the migration promoted by TM-GFP, whereas the mobility of cells expressing TM S490, 492A -GFP was not affected (Fig. 4). These results suggest that TM-promoted cell migration is mediated by β 1 integrin activation and CS modification is required for this effect. We then examined whether TM affects FAK activation. Immunofluorescence of phosphorylated FAK was more intense in TM-GFP-expressing A7r5 cells than cells expressing GFP or TM S490, 492A -GFP and can be seen at focal adhesions of pseudopods (Fig. 5a). Furthermore, FAK phosphorylation was significantly increased in cells stably expressing TM-GFP, whereas cells expressing a b c Fig. 3 Thrombomodulin is involved in VSMCs migration. a HASMCs infected with a lentivirus encoding luciferase (shLuc) or TM (shTM) shRNA were cultured in regular medium. Cell migration was analyzed by a wound-healing assay. Each monolayer was scratched, incubated for 24 h, and followed by image acquisition. The open area (scratch) was quantified with ImageJ software from 5 different experiments and expressed as mean ± SEM. *P < 0.05 compared with shLuc. b A7r5 cells transfected with pEGFP, pEGFP-TM or pEGFP-TM S490, 492A were cultured for 36 h and serum-starved for 24 h. Wound healing assay was performed under PDGF-BB treatment for 24 h. Cell motility was quantified from 4 independent experiments and expressed as mean ± SEM. c A7r5 cells transfected with pEGFP, pEGFP-TM, pEGFP-TMΔL or pEGFP-TM S490, 492A underwent invasion assay in transwells without coating or coated with type I collagen or fibronectin. Cells seeded in the upper chamber were incubated for 16 h, those attached to the lower surface of the membrane were fixed, stained with DAPI, and GFPexpressing cells counted. *P < 0.05, **P < 0.01, ***P < 0.001, compared with the GFP-expressing group; # P < 0.05, ## P < 0.01, ### P < 0.001, compared with the TM-GFP-expressing group TM S490, 492A -GFP exhibited no difference from control. In contrast to FAK activation, the expression of FAK was not affected by TM expression in VSMCs (Fig. 5b and c).
TM expression has no effect on PDGF-stimulated proliferation of VSMCs TM was shown to decrease the proliferation of tumor cell lines derived from patients with malignant melanoma [30]. On the contrary, recombinant TM containing 6 EGF-like repeats increases DNA synthesis in ECs and VSMCs [6,15]. To determine whether TM affects VSMCs proliferation, A7r5 cells expressing GFP-tagged TM or CS-deficient TM S490, 492A were stimulated with PDGF and their entry into S phase was indicated by Ki-67 staining. The percentage of Ki-67-positive nuclei was calculated among GFPexpressing cells. In control cells, 15.8% ± 1.6% nuclei were Ki-67-positive and no difference was detected in the percentage of Ki-67-positive cells among GFP-, TM-GFP-or TM S490, 492A -GFP-expressing cells (Additional file 1: Supplementary data, Figure S3A and B). To substantiate these results, the percentage of Ki-67-positive cells was examined in A7r5 cells stably expressing vector only, TM-GFP, and TM S490, 492A -GFP. Likewise, no difference was detected in the percentage of Ki-67positive cells (Fig. 6a and b) or cell number among three groups (Fig. 6c). These results suggested that TM expression did not affect VSMCs proliferation.

Discussion
In this study we described a new function of TM in VSMCs. We showed that TM enhances cell adhesion, spreading, and migration of VSMCs. Moreover, we provide evidence showing that the CS moiety at the serine/threonine-rich domain of TM mediates these functions.
Smooth muscle cell migration is an integral part of vascular development, atherogenesis, and injury-induced vascular remodeling [1,31]. Migration is a multistep process which involves cell protrusion, adhesion, spreading, and contraction [32]. During cell migration, cell adhesion to the substratum and interaction with ECM proteins play a central role in the mobile process. Extensive evidence indicates that soluble TM has roles in cell migration, thereby promoting angiogenesis and wound healing. A soluble recombinant TM fragment consisting of the second and third domains (rTMD23) stimulates ECs migration in vitro and angiogenesis both in vitro and in vivo [15]. Similarly, recombinant TM fragments consisting of the three N-terminal domains (rTMD123) and rTMD23 promote cutaneous wound healing [33]. Both cell spreading and wound closure are slower in TM-deficient keratinocytes compared to their wild-type counterpart, suggesting a role for TM in cell migration [33]. Indeed, TM was shown to mediate PDGF-stimulated migration activity of cultured bladder smooth muscle cells [9]. Furthermore, forced expression of TM promotes cell adhesion and migration of TM-deficient A2058 melanoma cells over a fibronectin-coated surface [10]. Results from this study agree fully with those detected in A2058 cells and suggest that TM enhances VSMCs motility, at least in part, through promoting cell adhesion. It's noteworthy that TM was reported to inhibit the migration and invasion of different cancer cells [34][35][36][37], probably through its role in preventing epithelial-mesenchymal transition [35]. The mechanisms underlying these apparently opposite roles of TM in epithelial cancer cells and VSMCs warrant further investigation. Furthermore, the impact of TM-enhanced VSMCs migration activity in vascular remodeling remains to be determined as previous evidence showed that TM overexpression inhibits arterial overdilation-induced neointima formation [38].
The CS moiety bound to the Ser/Thr-rich domain is indispensable in the anti-coagulant function of TM, but whether the CS moiety of TM performs other functions remained elusive. Our results showing that removing the CS moiety, either through ChABC-catalyzed degradation or via expressing the CS-devoid mutant, eliminated the TM effect on VSMCs adhesion and migration establish a new role for the CS moiety in mediating TM-enhanced  [16][17][18][19][39][40][41]. Given our previous results that the lectin-like domain of TM binds to fibronectin [10], we were surprised to find that CS deficiency, not the lectin-like domain deletion, eliminates VSMCs adhesion and migration on fibronectin (Additional file 1: Figure S2 and Fig. 3C). Furthermore, TM-enhanced cell adhesion and migration were more pronounced with fibronectin coating than collagen coating. These results implicate the CS moiety of TM in binding fibronectin, a notion supported by an earlier study showing that CS of melanoma cells binds to a specific domain of fibronectin and can either act independently or coordinate with an integrin to enhance melanoma cell adhesion to fibronectin [40].
CS proteoglycans coordinate with integrins to function in cell adhesion and migration [39][40][41]. Reduction of endogenous CS proteoglycan expression inhibits the motility, migration, and adhesion of fibrosarcoma cells, whereas treatment with exogenous CS proteoglycans dose-dependently stimulates cell motility and migration via a JNK-dependent pathway [18]. A recent study showed that chondroitin/dermatan sulfate proteoglycans play a key role in the adhesion and migration of VSMCs. The DS-epi1 −/− VSMCs, which exhibit reduced iduronic acid both on the cell surface and in secreted CS proteoglycans, have fewer focal adhesion sites, decrease directional migration, and reduce the activation and expression of FAK compared to the wild-type VSMCs [19]. The activation of FAK downstream of β 1 integrin signaling is a major pathway underlying cell migration activity. Our results that β 1 integrin mediates TMenhanced VSMCs migration and the fact that TMenhanced FAK activation is dependent on the CS moiety again point to the activation of β 1 integrin-FAK in mediating TM-enhanced VSMCs migration. In this context, the serine/threonine-rich domain of TM was shown to interact with β 2 integrin expressed on peripheral blood mononuclear cells in vitro [42]. Further study is required to unravel how the CS moiety of TM interacts with c b a Fig. 6 Thrombomodulin expression did not affect PDGF-stimulated proliferation of A7r5 cells. A7r5 cells were transfected with pEGFP, pEGFP-TM, or pEGFP-TM S490,492A , serum-starved for 48 h, and treated with PDGF-BB for 24 h. Cells were fixed and stained with DAPI (blue) and Ki-67(red). a shows representative results; (b) summarizes the ratio of Ki-67-positive cells; (c) summarizes total cell numbers in cells expressing vector alone, TM-GFP, and TM S490, 492A -GFP from at least 3 independent experiments. Results were expressed as mean ± SEM. Scale bar: 100 μm integrin-FAK pathway as we were unable to detect the binding complex between TM and β 1 integrin (CP Pai and MJ Jiang, unpublished results).
The role of TM in cell proliferation appears to vary depending on the form of TM expressed and secreted by cells. A soluble TM fragment consisting of the EGF-like repeats (rTMD2) stimulates proliferation in various cell types, including ECs, VSMCs, and Swiss 3T3 cells [6,15,43], and was shown to act through a fibroblast growth factor receptor in ECs [44]. On the contrary, membrane-bound TM overexpression and a soluble recombinant TM fragment consisting of the entire extracellular domains inhibit both thrombinstimulated VSMC proliferation in culture [45,46] and vascular injury-induced neointima formation [38,47]. These contradictory results illustrate the complex roles of TM reflecting the interaction of various domains of TM with other proteins and possibly among different TM domains. Our results are consistent with those reported in cultured urinary bladder smooth muscle cells, both showing no effect of increased TM expression on PDGF-stimulated cell proliferation [9]. It is well established that TM soluble fragments are detected in serum and urine in various forms and concentrations depending on physiological conditions [48]. Therefore, the role of TM in VSMCs proliferation in vivo may vary under different pathophysiological conditions.

Conclusion
The CS moiety at the serine/threonine-rich domain mediates TM-enhanced VSMCs adhesion, spreading, and migration activity through the activation of β 1 integrin and a FAK pathway.