Modulation of Thrombomodulin-dependent Activation of Human Protein C through Differential Expression of Endothelial Smads*

Protein C is a plasma protease that in its active form plays a central role in the regulation of vascular function by modulating thrombosis, inflammation, and apoptosis. A central player in this pathway is the cytokine-regulated receptor thrombomodulin (TM), which functions as a co-factor for the thrombin-dependent generation of activated protein C. We have found that tumor necrosis factor-β (TGF-β)-dependent suppression of TM on endothelial cells is differentially regulated by endothelial Smad6s and Smad7. Overexpression of Smad6s resulted in activation of a TGF-β reporter alone and enhanced TGF-β response. Moreover, Smad6s overexpression suppressed TM and subsequently reduced activated protein C generation. Antisense inhibition of Smad6s expression enhanced the TM-dependent activation of protein C, whereas blocking the inhibitory Smad7 by antisense resulted in reduced TM-dependent activation of protein C. The effect of Smad6s appeared to be due, at least in part, to up-regulation of TGF-β itself. Immunohistochemisty studies in normal versusatherosclerotic vessels showed that TM levels were suppressed in the endothelium over plaque. Consistent with the in vitro data, we found differential expression of Smad6s and Smad7 in normalversus atherosclerotic vessels, with Smad6s expression low in normal vessels but elevated in atherosclerotic vessels. In contrast, the opposite expression pattern was observed for Smad7. Overall, our results suggest that the relative balance of these intracellular Smads modulate the balance of endothelial function with regard to protein C activation.

Protein C is a plasma protease that in its active form plays a central role in the regulation of vascular function by modulating thrombosis, inflammation, and apoptosis. A central player in this pathway is the cytokineregulated receptor thrombomodulin (TM), which functions as a co-factor for the thrombin-dependent generation of activated protein C. We have found that tumor necrosis factor-␤ (TGF-␤)-dependent suppression of TM on endothelial cells is differentially regulated by endothelial Smad6s and Smad7. Overexpression of Smad6s resulted in activation of a TGF-␤ reporter alone and enhanced TGF-␤ response. Moreover, Smad6s overexpression suppressed TM and subsequently reduced activated protein C generation. Antisense inhibition of Smad6s expression enhanced the TM-dependent activation of protein C, whereas blocking the inhibitory Smad7 by antisense resulted in reduced TM-dependent activation of protein C. The effect of Smad6s appeared to be due, at least in part, to up-regulation of TGF-␤ itself. Immunohistochemisty studies in normal versus atherosclerotic vessels showed that TM levels were suppressed in the endothelium over plaque. Consistent with the in vitro data, we found differential expression of Smad6s and Smad7 in normal versus atherosclerotic vessels, with Smad6s expression low in normal vessels but elevated in atherosclerotic vessels. In contrast, the opposite expression pattern was observed for Smad7. Overall, our results suggest that the relative balance of these intracellular Smads modulate the balance of endothelial function with regard to protein C activation.
Endothelial dysfunction plays a critical role in the development of both chronic (atherosclerosis) and acute (thrombotic) cardiovascular disease (1)(2)(3)(4). An important modulator of endothelial function is activated protein C. Protein C is a plasma serine protease that has a well described role in maintaining normal hemostatic balance (5)(6)(7)(8). Upon activation by thrombin, activated protein C (APC) 1 acts as a feedback inhibitor of thrombin generation by inactivating the activated forms of Factors V and VIII, thus inhibiting the prothrombinase and Factor Xase enzyme complexes, respectively. In addition, recent studies have demonstrated that activated protein C plays a key role in regulating endothelial genes associated with cell survival and anti-apoptotic activities, thereby directly modulating intracellular signaling and pathways that protect the interface between the vessel wall and the soluble environment (9,10). Activated protein C has been shown to be an effective antithrombotic in a wide variety of venous and arterial thrombosis models (reviewed in Refs. 7, 11, and 12), and activated recombinant human protein C has demonstrated efficacy in the treatment of severe sepsis (13).
The major physiologic factor controlling the activation of protein C by thrombin is the endothelial surface membrane protein thrombomodulin (TM) (14). As thrombin is generated, it complexes with TM to form an enzyme complex that no longer has procoagulant activity (converting fibrinogen to fibrin) but instead has anti-thrombotic activity by converting zymogen protein C to APC. Areas of the human aorta at high risk for atherosclerosis have reduced TM and increased prothombinase activity (15), and TM has been shown to be reduced in the endothelium over atherosclerotic plaque (16). In general, atherosclerotic lesions appear to be associated with acquired defects in the protein C system (17), and poor response to APC generation has been suggested as a prominent risk predictor of advanced arterial disease (18) and ischemic stroke risk (19). Thus, TM controls the balance between the pro-coagulant activities of thrombin (fibrin generation, platelet activation) and anticoagulant activity (APC generation) and determines a condition of either normal homeostasis or vessel pathology.
The level of TM on endothelial cells has been shown to be modulated by cytokines, including transforming growth factor-␤ (TGF-␤). Both TGF-␤1 and TGF-␤2 have been shown to down-regulate thrombomodulin mRNA expression in cultured human endothelial cells (20,21), and increased TGF-␤ correlates with decreased TM-containing vessels in sustained local endothelial dysfunction (22). In this study, we have explored the role of TGF-␤ signaling in controlling the TM-dependent activation of protein C on endothelial cells, focusing on the role of endothelial Smad proteins. Most of the Smad family of proteins positively control signal transduction of the various members of the TGF-␤ family (reviewed in Refs. [23][24][25][26][27][28]; however, two Smads (Smad6 and Smad7) inhibit TGF-␤ signal transduction. Smad6 has been shown to inhibit signaling by the TGF-␤ superfamily (29 -32) and plays a role in the development of the cardiovascular system (33). The Smad7 protein has been extensively studied and shown to prevent phosphorylation of receptor-activated Smads, thereby inhibiting TGF-␤-induced signaling responses (34 -37). Recently, we described the characterization of an endothelial splice variant of Smad6, designated Smad6s, that showed differential activity relative to the inhibitory Smads in a Xenopus model, being an antagonist of * The costs of publication of this article were defrayed in part by the payment of page charges. This article must therefore be hereby marked "advertisement" in accordance with 18 1 The abbreviations used are: APC, activated protein C; CAT, chloramphenicol acetyltransferase; SFM, serum free medium; TGF-␤, tumor necrosis factor-␤; TM, thrombomodulin; DMEM, Dulbecco's modified Eagle's medium; ELISA, enzyme-linked immunosorbent assay; PBS, phosphate-buffered saline. the bone morphogenic protein pathway but an agonist of the activin pathways (38). Using overexpression and antisense modulation, we show that Smad6s positively mediates TGF-␤ repression of TM and subsequent reduction in APC generation in human endothelial cells. In contrast, the inhibitory Smad7 significantly reduced TGF-␤ down-regulation of APC generation. Furthermore, we show that Smad6s and Smad7 are counter-regulated in normal versus atherosclerotic plaques in a manner consistent with the down-regulated TM levels and APC generation in the endothelium of atherosclerotic vessels.

EXPERIMENTAL PROCEDURES
Reagents and Cell Culture-Thrombin and hirudin were obtained from Sigma. Human protein C was produced as described previously (39). The chromogenic substrate, S2366, was obtained from Chromogenix. TGF-␤1 and TGF-␤3 were purchased from R & D Systems. Human umbilical vein endothelial cells were obtained from Clonetics (San Diego, CA) and grown as described previously (40). SVHA-1 cells, an SV40-transformed human aortic endothelial cell line, were used to obtain high transfection rates needed for the antisense experiments. Cells were maintained in DMEM/F-12 (3:1), a medium comprised of a 3:1 v/v mixture of Dulbecco's modified Eagle's medium and Ham's nutrient mixture F-12. DMEM/F-12 (3:1) and fetal bovine serum were purchased from Invitrogen. The basal medium was supplemented with 10 nM selenium, 50 M 2-aminoethanol, 20 mM HEPES, 50 g/ml gentamycin, and 5% fetal bovine serum. All other reagents were of the highest quality available.
DNA Transfection and Chloramphenicol Acetyltransferace (CAT) ELISA Assay-The vector pOCAT2336 for measuring TGF-␤ response and Smad expression vectors were described previously (38). Endothelial cells were seeded in six-well plates to 80% confluence. DNA was transfected at a concentration of 1 g for pOCAT2336 and 5 g for the Smad vectors with Invitrogen's Lipofectin reagent and used accordingly to manufacturer's instruction. Expressed CAT protein was measured using a CAT ELISA kit from Roche Molecular Biochemicals and performed according to manufacturer's protocol. The plates were read kinetically and data expressed in mOD/min.
Antisense Phosphorothioate Oligodeoxynucleotides-The oligonucleotides used in antisense experiments were synthesized with phosphorothioates and C-5 propyne pyrimidines following standard protocols (41). Antisense oligodeoxynucleotides were designed to hybridize to the region of the Smad6s or the Smad7 mRNA encompassing the initial ATG. The sequence of the antisense oligodeoxynucleotide for Smad6s was: 5Ј-GGTTTGCCCATTCTGGACAT-3Ј, and the sense strand control oligodeoxynucleotide for Smad6s was: 5Ј-ATGTCCAGAATGGGCAAACC-3Ј. The sequence of the antisense oligodeoxynucleotide for Smad7 was: 5Ј-GATCGTTTGGTCCTGAACAT-3Ј, and the sense strand control oligodeoxynucleotide for Smad7 was: 5Ј-ATGTTCAGGACCAAACGATC-3Ј. Oligonucleotides were introduced to cells following a modification of a procedure as outlined in Ref. 41. Briefly, cells were plated in 96-well COSTAR plates at a density of 2000 or 5000 cells/well and allowed to attach overnight. After washing monolayers with serum-free medium (SFM), a 10 or 20 M concentration of each oligonucleotides was introduced in 100 l of SFM. Control wells containing SFM with vehicle alone were included in addition to the sense strand oligonucleotides controls. After an overnight incubation in the presence of oligos, each condition was rinsed with SFM and re-charged with oligonucleotides overnight as above. On the fourth day each experimental condition was treated with or without TGF-␤ at a concentration of 1 ng/ml in the maintenance medium. The antisense was shown to inhibit the Smad protein levels by Western blot analysis. Cell surface TM levels were assayed as described previously (42). TM-dependent protein C activa- tion was performed as follows: conditioned medium was removed and replaced with 100 l of SFM containing 25 g/ml human protein C and 0.5 units/ml thrombin. After 1-h incubation at room temperature, 75 l was removed to a 96-well plate well containing 50 l of 10 units/ml hirudin in activation buffer (20 mM Tris, pH 7.4, 150 mM NaCl) and incubated 5-10 min with agitation. Activated human protein C was then assayed by adding 50 l of the chromogenic substrate S2366 and measuring the change in absorbance at 405 nm on a 5-min kinetic run using a Molecular Devices ThermoMax plate reader.
TFG-␤ Promoter Activation and Secretion-TGF-␤3 (43) and TGF-␤1 promoter (44) constructs driving CAT expression (1 g) were co-transfected with Smad6s vector (5 g), Smad7 (5 g), or with a control pCIneo vector (Promega, Madison, WI). Transfection using Lipofectin in serum-free medium on ECV304 cells (ATCC CRL1998), previously plating cells at 3 ϫ 10 5 cells per well in a six-well plate with DMEM/F-12 medium in 5% fetal bovine serum. After 24 h, the cells were washed twice with phosphate-buffered saline (PBS), and 2 ng/ml TGF-␤1 or TGF-␤3 (R & D Systems) was added in serum-free DMEM medium containing 100 g/ml Cohn's fractionated bovine serum albumin. The cells were incubated overnight at 37°C, and supernatants were collected, and levels of endogenous latent TGF-␤1 and TGF-␤3 secreted into the supernatant were evaluated with ELISA kits according to manufacturer (R & D systems). The cells were washed twice with PBS, lysed, and CAT activity expressed was measured kinetically. Lysates were normalized using a BCA assay measuring total protein concentration.
Immunohistochemistry-All tissue specimens were retrieved from the tissue bank of Lilly Research Laboratories. These tissues were obtained from the Cooperative Human Tissue Network using an institutional review board-approved protocol. All human samples were derived from surgical specimens obtained during the period extending from 1996 to 1999. Tissues were fixed overnight in zinc-buffered formalin and then transferred to 70% ethanol prior to processing through paraffin. Five-micrometer sections were microtomed and the slides baked overnight at 60°C. The slides were then deparaffinized in xylene and rehydrated through graded alcohols to water. Antigen retrieval was performed by immersing the slides in Accutuf tissue unmasking solution (Accurate Chemical) for 10 min at 90°C (in a water bath), cooling at room temperature for 10 min, washing in water, and then proceeding with immunostaining. All subsequent staining steps were performed on the Dako immunostainer; incubations were done at room temperature, and Tris-buffered saline plus 0.05% Tween 20, pH 7.4 (TBS; Dako Corp.) was used for all washes and diluents. Thorough washing was performed after each incubation. Slides were blocked with protein blocking solution (Dako) for 5 min; after washing, a 10 g/ml amount of the particular SMAD antibody (or irrelevant control antibody) was added to the slides and incubated for 30 min. A biotinylated secondary antibody plus streptavidin-horseradish peroxidase kit (Dako LSAB2) was then utilized along with a DAB chromagen and peroxide substrate to detect the bound antibody complexes. The slides were briefly counterstained with hematoxylin, removed from the autostainer, and dehydrated through graded alcohols to xylene. Coverslips were mounted with a permanent mounting medium. Scoring was based on a blind evaluation of the intensity and localization of staining using light microscopy as reviewed by two board-certified pathologists, with negative being the total absence of detectible staining.

TGF-␤ Suppresses APC Generation on Human Endothelial
Cells-As described previously and shown in Fig. 1A, the level of surface TM is suppressed on human endothelial cells treated with TGF-␤1. To determine whether this reduction in TM resulted in a concomitant reduction in the ability of cells to support APC generation, cells were treated with TGF-␤, and the rate of thrombin-catalyzed activation of exogenous zymogen protein C was determined. As shown in Fig. 1B, the reduction in TM by TGF-␤ resulted in a significant reduction in the ability to support APC generation. The amount of APC generation was directly proportional to the relative level of throm- bomodulin present on the cell surface (inset).
Smad6s Enhances TGF-␤ Suppression of TM and APC Generation-Because our previous studies in Xenopus had shown that Smad6s had differential activity in modulating TFG-␤ family pathways, we examined its role in TGF-␤ signaling in endothelial cells. Co-transfection experiments were performed with a Smad6s mammalian expression vector and the TGF-␤ reporter plasmid (pOCAT2336) containing the CAT gene under the control of TGF-␤-inducible plasminogen activator inhibitor-1 gene promoter. Endothelial cells transiently transfected with the reporter construct responded to TGF-␤ with an approximate 5-fold increase in reporter activity ( Fig. 2A). Interestingly, transfection of cells with the Smad6s expression plasmid resulted in a similar increase in TGF-␤-dependent promoter activity and enhanced the response in combination with TGF-␤. The ability of Smad6s to induce a TGF-␤-like response in the reporter assay was confirmed by examining its effect on TM-dependant APC activation. As shown in Fig. 2B, transfection of cells with the Smad6s expression vector resulted in a significant suppression in APC generation, similar to that observed with treatment of cells with 1 ng/ml TGF-␤. In control experiments with inhibitory Smad7 we showed complete inhibition of TGF-␤ response on both the reporters ( Fig.  2A) and TM levels (not shown), as would have been expected from previous studies.
Inhibition of Smad6s with Antisense Blocks TGF-␤ Inhibition of APC Generation-While the overexpression of Smad6s suggested a role in negatively regulating TM and APC generation, we confirmed these results by inhibiting endogenous Smad6s with antisense oligonucleotide. Endothelial cells were treated with sense or antisense oligonucleotides followed by stimulation with TGF-␤. Smad6s antisense, but not a sense control, blocked the inhibition of TM by TGF-␤ (Fig. 2C) and the inhibition of APC generation (Fig. 2D). We also determined the effect of inhibiting Smad7 with antisense, and as expected (since Smad7 inhibits TGF-␤ response), the TGF-␤ response was enhanced as shown by an increased inhibition of TM in the presence of TGF-␤ (Fig. 2C).
Smad6s Induces TGF-␤-The ability of Smad6s to mimic a TGF-␤ response suggested that possibly Smad6s can induce TGF-␤ expression. We examined the effect of Smad6s overexpression on TGF-␤ levels secreted into the culture medium and on a reporter construct driven by TGF-␤ promoters. As shown, Smad6s overexpression was capable of increasing the amount of TGF-␤1 secreted from the cell (Fig. 3A). We also assessed the effect on TGF-␤3 and found a similar ϳ6-fold increase in secreted TFG-␤3 activity. This appeared to be the result of increased expression as indicated by the ability of Smad6s to induce both the TGF-␤1 and TGF-␤3 promoter (Fig. 3B). These results suggested that the level of Smad6s in endothelial cells can alter the level of TGF-␤ and thus the relative level of TM and APC generation.
Differential Regulation of TM and Smads in Vascular Disease-The data above demonstrate a differential effect of Smad6s and Smad7 on the modulation of TM on human endothelial cells in culture. To determine whether this relationship was observed in endothelial cells in vivo, we examined normal and atherosclerotic human cardiovascular tissues with TM-, Smad6-, and Smad7-specific antibodies as described previously (38). As shown in Fig. 4A, we observed a significant decrease in the level of TM in atheroslerotic coronary vessels. Consistent with the data above, we observed that Smad6s levels were undetectable in normal vessels and overexpressed in the endothelium over the plaque, as well as in the plaque vasculature. In contrast, Smad7 was expressed in normal vascular endothelium but significantly decreased in the plaque endothelium. This relationship of differential expression is clearly demonstrated in Fig. 4B following analysis of 35 vessels for expression of Smad6s and Smad7. These data are consistent with the observations in cultured endothelial cells and suggest that TM levels, and therefore APC generation, are controlled by differential expression of the Smad proteins in vivo. DISCUSSION Under normal circumstances, the vascular endothelium displays a number of regulatory mechanisms to modulate coagulation, inflammation, and vascular function to maintain homeostatic balance in the local environment (45,46). As this balance is disturbed, pathological processes, including thrombosis and atherosclerosis, ensue. While we have long considered APC as an important antithrombotic modulator, providing feedback inhibition of coagulation, the emerging data suggest that APC is also an agent that effectively modulates the balance of anti-inflammatory and anti-apoptotic systems in response to injury (9,10). Thus, factors that suppress protein C activation will compromise endothelial function and promote pathogenic processes.
As reviewed above, recent studies have clearly linked low TM levels and defects in APC generation to vessel disease. Moreover, studies have demonstrated that TGF-␤ is overexpressed in fibroproliferative vascular lesions (47) and appears to be most active in lipid-rich aortic intimal lesions (48), consistent with the overexpression of Smad6s. TM suppression is recognized as a marker of sustained endothelial dysfunction (22), and its down-modulation clearly is the linked to increased TGF-␤ in the vessel and to the promotion of thrombogenesis at the sites of injury (20). Our results provide new mechanistic understanding for the control of APC generation. In these studies we find that the balance of two TGF-␤ signaling molecules, Smad7 and Smad6s, can control the activation of APC by controlling the relative level of endothelial thrombomodulin. This appears to be at the level of controlling both TGF-␤ levels and signaling. Moreover the relative balance of these two molecules appears to correlate with TM levels in normal versus diseased vessels.
The protein C pathway plays a unique and central role in modulating vascular function. In states of systemic inflammatory activation, loss of protein C results in a compromised ability to modulate coagulation, inflammatory, and cell survival functions, leading to vascular dysfunction (reviewed in Refs. 10 and 49). The results presented here suggest that TGF-␤ plays a prominent role in controlling the activation of the protein C pathway and that targeting Smad signals may provide new opportunities for therapeutic intervention in treating and preventing vascular dysfunction.