Thrombin Inhibits Tumor Cell Growth in Association with Up-regulation of p21 waf/cip1 and Caspases via a p53-independent, STAT-1-dependent Pathway*

Thrombin, a multifunctional protein, has been found to be involved in cellular mitogenesis, tumor growth, and metastasis, in addition to its well known effects on the initiation of platelet aggregation and secretion and the conversion of fibrinogen to fibrin to form blood clots. These properties of thrombin rely on its action as a serine protease, which cleaves the N-terminal region of a 7-transmembrane G protein receptor (protease-ac-tivated receptor, PAR-1), thus exposing a tethered end hexapeptide sequence capable of activating its receptor. Little is known about its effect on genes that regulate the cell cycle. This study was undertaken to investigate the possible mechanisms by which thrombin regulates tumor cell growth in several tumor cell lines: human CHRF megakaryocyte, DU145 prostate, MDAMB231 and MCF7 breast, U3A fibrosarcoma, and 2 murine fibroblast cell lines, MEFp53( 2 / 2 ) and CD STAT( 2 / 2 ). We have found that thrombin under the conditions of culture employed inhibits cell growth by both up-regulation of p21 waf/cip1 and induction of caspases via its PAR-1 receptor. The increased expression of p21 waf/cip1 by thrombin was p53 independent, STAT1 dependent, and protein synthesis independent. This was associated with tyrosine phosphorylation of JAK2 and STAT1, and nuclear translocation of STAT1. Induction of apoptosis is also PAR-1-specific, STAT1-dependent, and associated with up-regulation of caspases 1, 2,

In mammalian cells, the cell cycle is regulated by the ordered activation of a group of enzymes known as the cyclin-dependent kinases (CDC2 and CDK2 1 to CDK5). The activity of these enzymes is controlled in part by their association with regulatory subunits called cyclins (cyclin A to E) (1,2). In normal cells, CDKs exist predominantly in multiple quaternary complexes, each containing a CDK, cyclin, proliferating cell nuclear antigen, and the p21 protein. However, in many transformed cells, such as SV40 T antigen-transformed cells, p21 waf/cip1 is lost from these multiprotein complexes; overexpression of p21 waf/cip1 inhibits the proliferation of mammalian cells (3)(4)(5). It is clear that p21 waf/cip1 plays an important negative regulatory role in control of cell proliferation.
Thrombin is a multifunctional protein which converts fibrinogen to fibrin, activates blood platelet adhesion, aggregation, and secretion, induces chemotaxis and adhesion of other cells, and elicits mitogenic responses in primary endothelial cells, vascular smooth muscle cells, and fibroblasts as well as mitogenesis and metastasis of tumor cells (6 -14). All of these properties of thrombin rely on its action as a serine protease. The action of thrombin on cells is induced via a protease-activated receptor(s), PAR-1, PAR-3, and possibly PAR-4 (15)(16)(17)(18). Activation of the PAR thrombin receptors requires the cleavage of an N-terminal region of a 7-transmembrane G protein receptor. The newly formed N-terminal region of PAR-1 acts as a tethered ligand and activates thrombin's cellular responses. Synthetic peptides corresponding to the new N-terminal region of PAR-1 such as a 14-mer or 6-mer (thrombin receptor activation peptide (TRAP)) can substitute for thrombin receptor activation (19,20). Activation of the thrombin receptor stimulates a number of phospholipid-directed enzymes, generating both lipid-soluble and water-soluble second messengers (6). Thrombin stimulates phosphoinositidase C activity which generates inositol triphosphate and diacylglycerol. Both of these second messages are involved in intracellular Ca 2ϩ mobilization and protein kinase C activation, respectively. Thrombin stimulates jun, fos, Ras and MAP kinases which may be involved in mitogenesis (6,(21)(22)(23). On the other hand, it has been reported that the rapid induction of discrete intracellular signaling mechanisms by thrombin, including the Raf-1/MAP kinase pathway, appears to be insufficient alone to promote vascular smooth muscle cell mitogenesis (23).
Little is known regarding the effect of thrombin on genes regulating the cell cycle. This study was undertaken to investigate the possible mechanisms by which thrombin regulates tumor cell growth. We report that thrombin can inhibit tumor cell growth. This is associated with the up-regulation of the universal inhibitor p21 waf/cip1 via a p53-independent, STAT1dependent pathway as well as induction of apoptosis in association with the up-regulation of caspases, which is also mediated by a STAT1-dependent pathway.

MATERIALS AND METHODS
Cell Culture and Reagents-CHRF, a human megakaryocyte cell line, was a gift from Dr. M. Lieberman, University of Cincinnati Medical Center, Cincinnati, OH, and maintained in RPMI 1640 medium supplemented with 10% fetal calf serum and 1% penicillin and streptomycin. Human prostate cancer cell line DU145 and breast cancer cell lines MDAMB231 and MCF7 were obtained from the ATCC and grown in Dulbecco's modified Eagle's medium containing 10% fetal calf serum and antibiotics. MEFp53(Ϫ/Ϫ) cells (mouse embryonic fibroblasts) were derived from p53 knockout mice, and kindly provided by Dr. Joseph Schlessinger, New York University Medical Center. Mouse fibroblasts CDϩ and CDϪ were derived from parental and STAT knockout mice, respectively, and kindly supplied by Dr. David Levy, New York University Medical Center. Both mouse fibroblast cell lines were grown in Dulbecco's modified Eagle's medium containing 10% FCS and antibiotics. U3A, STAT 1(Ϫ/Ϫ), human fibrosarcoma and 2B, STAT(ϩ/ϩ) human fibrosarcoma were obtained from Dr. Richard Pine (Public Health Research Center, New York). For experimental studies, cells were grown to approximately 80% confluence and then switched to serumfree media containing 0.1% bovine serum albumin for starving overnight. In some experiments, DU145 cells were cultured in media containing 5% FCS. Various concentrations of thrombin were then added to the cells. Thrombin and the thrombin receptor-activation peptide (TRAP 14-mer), hirudin (thrombin inhibitor), and cycloheximide were purchased from Sigma. Anti-p21, anti-caspase 3, anti-JAK2, anti-STAT1, and anti-phosphotyrosine antibodies were obtained from Santa Cruz.
DNA Synthesis Assay-Cells were seeded at 1 ϫ 10 4 /well in 96-well microtiter plates. Various concentrations of thrombin, ranging from 0.25 to 1 units/ml were then added to the wells. After 24 h, cells were pulsed with 10 M bromodeoxyuridine (BrdUrd) for 4 h at 37°C. Br-dUrd incorporation into newly synthesized cellular DNA was analyzed by immunoassay according to the method described by the manufacturer (Roche Molecular Biochemicals).
Immunoblot Assays-After incubation, cells were washed with PBS, and lysed in 0.5 ml of 1% Triton X-100 in PBS containing 10 mM phenylmethylsulfonyl fluoride and aprotinin (10 g/ml). Twenty g of total protein were applied to 10% SDS-polyacrylamide gel electrophoresis, and then transferred to nitrocellulose. Filters were incubated with preblocking solution (10 mM Tris, pH 7.6, 0.15 M NaCl, 5% non-fat skim milk, 2% bovine serum albumin, and 0.1% Tween 20) for 2 h at room temperature, followed by 16 h at 4°C with rabbit anti-human p21 waf/cip1 antibody or rabbit anti-human caspase 3 antibody (Santa Cruz) in the same solution. After three washes in PBS, blots were incubated with goat anti-rabbit IgG conjugated with peroxidase and developed using chemiluminescence with luminol.
Immunoprecipitation-Total cell extracts were prepared by lysing cells in lysis buffer containing 1% Triton X-100, 300 mM NaCl, 50 mM HEPES (pH 7.6), 1 mM Na 3 VO 4 , 1 mM dithiothreitol, 1 mM phenylmethylsulfonyl fluoride, and 5 g/ml aprotinin and leupeptin. After centrifugation, the supernatants were incubated with either anti-STAT1 or JAK2 antibodies for 2 h at 40°C, followed by addition of protein G beads and further incubation for 2 h. The immune complexes were washed five times with lysis buffer and then subjected to immunoblotting analysis as indicated above. Tyrosine phosphorylation of STAT1 and JAK2 were detected by anti-phosphotyrosine antibody.
Immunochemical Staining-DU145 cells, cultured on chamber slides were starved in RPMI 1640 media containing 0.1% bovine serum albumin for 16 h and then stimulated with thrombin for 30 min. The cells were fixed with cold methanol and then washed with PBS. After treatment with H 2 O 2 , the cells were incubated with anti-STAT1 (1:300) antibody for 2 h and visualized with avidin-biotin complex (Vector Laboratories Inc., Burlingame, CA) and 3,3-diaminobenzidine. The cells were counterstained with methanol blue. Cells, treated without anti-STAT1 antibody, served as control.
Terminal dUTP Nick-end Labeling Assay-DU145 cells were cultured in media containing 1% FCS overnight, and then treated with 1 unit/ml thrombin for 24 h. The cells were smeared on slides and fixed with 10% formamide. The solution containing terminal deoxynucleotidyl transferase enzyme and digoxigenin nucleotides was applied to the slides (Oncor, Gaithersburg, MD). After washes, the antidigoxigenin antibody-peroxidase conjugate was added to the slides, followed by staining with substrate 3,3-diaminobenzidine. Methanol green solution was used for counterstaining.

RESULTS
Thrombin Inhibits Proliferation of CHRF Cells-CHRF cells express PAR-1 and respond to stimulation by thrombin. The effect of thrombin on CHRF cells was examined by the incorporation of BrdUrd, a measure of DNA replication and therefore an indicator of cellular activity in the S phase of the cell cycle. CHRF and DU145 cells were incubated with or without various concentrations of thrombin for 16 h and pulse-labeled with BrdUrd. As shown in Fig. 1A, the amount of BrdUrd incorporated with thrombin-treated cells (1 units/ml) was only 20 -25% of the incorporation seen with untreated cells. The maximum change in thrombin concentration-dependent impairment of cell growth was at 1 unit/ml thrombin. The same experiment was also performed with DU145 cells in media containing 5% FCS. Thrombin inhibited the incorporation of BrdUrd into DNA. This effect of 1.5 unit/ml thrombin could be reversed by hirudin (p ϭ 0.003), demonstrating thrombin specificity (Fig. 1B).
Thrombin Up-regulates the Expression of p21 waf/cip1 in CHRF Cells-Cell proliferation is controlled by cell cycle regulatory factors which include cyclins and cyclin-dependent kinases. The p21 waf/cip1 protein is a universal inhibitor of cyclin kinases and plays an important role in inhibiting cell proliferation. Therefore, the expression of p21 waf/cip1 mRNA was studied in CHRF cells following stimulation. The effect of continuous thrombin exposure (at various concentrations) on p21 waf/cip1 up-regulation in CHRF cells was prominent at 1, 3, 6, and 24 h (Fig. 2A). Although the effect was noted with as little as 0.25 units/ml thrombin, the maximum effect (3-4-fold) was generally noted at 0.5 unis/ml (5 nM). A similar up-regulation of p21 waf/cip1 was noted after a relatively short 10-min exposure to thrombin (followed by washing) at 1 and 3 h of further incubation with disappearance of the effect at 6 h (Fig.  2B). Similar 5-fold induction was noted with 2 breast cell lines, MDAMB431 and MCF7, following 3 h of incubation with 1 unit/ml thrombin (Fig. 2C) and a prostate cell line, DU145, at 3, 6, and 24 h of incubation (Fig. 3A). Immunoblot analysis of p21 waf/cip1 protein confirmed the Northern analysis. Fig. 2D demonstrates markedly enhanced p21 waf/cip1 protein following 3 h of thrombin stimulation of CHRF cells, with maximum effect at 0.5-1 units/ml.
Up-regulation of p21 waf/cip1 by Thrombin Is p53-independent-p21 waf/cip1 is regulated by the p53 pathway (4). Two approaches were used to study whether increased p21 waf/cip1 expression by thrombin is p53-dependent. Prostate cancer cell line, DU145, with a defect in the gene encoding for p53, and mouse embryonic fibroblasts (MEF) derived from a p53 "knockout" mouse were studied for p21 waf/cip1 up-regulation by thrombin. As shown in Fig. 3, A and B, the p21 waf/cip1 mRNA level was clearly increased in both cell lines treated with thrombin. These results suggested that thrombin-stimulated up-regulation of p21 waf/cip1 expression is p53-independent.
Up-regulation of p21 waf/cip1 by Thrombin Is STAT 1-dependent-To study whether induction of p21 waf/cip1 by thrombin required ongoing protein synthesis, we stimulated CHRF cells with thrombin (1 unit/ml) in the presence or absence of cycloheximide (50 g/ml), a potent protein synthesis inhibitor. Approximately 3-fold up-regulation of p21 was observed in cells treated with thrombin or cycloheximide plus thrombin indicating that ongoing protein synthesis was not required for upregulation of p21 waf/cip1 by thrombin (Fig. 4A). It has been reported that interferon ␥ induces the expression of p21 waf/cip1 via the STAT pathway (24). In examining the possible role of STAT1 in thrombin induced up-regulation of p21 waf/cip1 , we looked at the level of expression of p21 waf/cip1 in thrombinstimulated fibroblasts derived from a STAT1 knockout mouse. The increased expression of p21 waf/cip1 upon thrombin stimulation was observed in CDϩ cells not in STAT1-deficient fibroblasts (CDϪ) (Fig. 4B). The same result was also obtained with 2B and U3A fibrosarcoma cells, which contain functional or deficient STAT1, respectively (data not shown). This result suggested that thrombin up-regulation of p21 required the activation of STAT1. The activation of STAT1 in thrombininduced expression of p21 waf/cip1 was further examined using the mobility shift assay (Fig. 4C). Nuclear extracts prepared were examined for STAT1 binding activity. Administration of thrombin resulted in a marked increase in STAT1 activity as measured by mobility shift assay. Activation of the JAK-STAT pathway by thrombin was further investigated by tyrosine phosphorylation of the two proteins. As shown in Fig. 4, D and E, STAT1 and JAK2 were tyrosine phosphorylated after stimulation of CHRF cells with thrombin for 15 min. Thrombin-dependent nuclear translocation of STAT1 was analyzed by immunostaining using anti-STAT1 antibody. As shown in Fig. 4F, thrombin induced the translocation of STAT 1 from the cytoplasm to the nucleus. Positive signals were not observed in control cells (data not shown).
Thrombin Up-regulation of p21 waf/cip1 in DU145 Cells Is Specific and Mediated via PAR-1 Activation- Fig. 5 demonstrates inhibition of thrombin up-regulation of p21 waf/cip1 by hirudin (lanes 1-4) and PAR-1 receptor specificity (lanes 1 and  5). Similar results were noted with the up-regulation of caspase 3 by thrombin (Fig. 6D). Thrombin receptor analysis by RT-PCR revealed the presence of PAR-1 in all cell lines examined (DU145, CHRF, MDAMB231, and MCF7). The expression of PAR-3 and PAR-4 was only detected in CHRF cells, not the other 3 cell lines tested (data not shown).
Thrombin Up-regulates the Expression of Caspases-Cells were treated with or without various concentrations of thrombin and the total RNAs isolated. Northern blot analysis was performed using specific probes. As shown in Fig. 6A, thrombin induced caspases 1, 2, and 3 in DU145 cells. The maximum effects were observed at 1 unit/ml thrombin with 5-fold increased expression of caspase 1-and 4-fold increased expression of caspases 2 and 3. A 4-fold up-regulation of caspase 3 by thrombin was also observed in CHRF cells. The increased expression of caspase 3 was not observed in cells derived from STAT 1 knockout mice treated with 1 unit/ml thrombin, but was detected in mouse cells containing wild type STAT1 (Fig.  6B). Up-regulation of caspase 3 was also detected at the protein level in both DU145 and CHRF cells (Fig. 6C).
Induction of Apoptosis by Thrombin-DU145 cells were treated with or without 1 unit/ml thrombin for 24 h, and cells further analyzed by terminal dUTP nick-end labeling assay. The apoptotic cells were clearly detected in cells treated with thrombin, while no apoptotic cells were observed in control cells (data not shown).
Effect of Thrombin on Growth of CDϪ and STATϪ/Ϫ Cells-The requirement of STAT1 for thrombin-induced p21 waf/cip1 and caspase 3 induction suggested that STAT1(Ϫ/Ϫ) cells would be more resistant to thrombin-induced impaired cell growth. Cell growth of CDϩ versus CDϪ cells were therefore measured by BrdUrd incorporation as in Fig. 1. CDϪ cells were 1.4 -1.5-fold more resistant to thrombin-induced impaired cell growth than CDϩ cells at 1, 1.5, 2, and 2.5 units/ml thrombin (two experiments, data not shown). DISCUSSION A novel and important observation of the present study is the observation that thrombin induces growth inhibition of several tumor cell lines under the conditions studied. This finding indicates that thrombin, aside from its well known effects on blood coagulation, platelet aggregation, and cell proliferation, may also directly inhibit the growth of cells. Many growth factors and cytokines have dual effects on cell growth (25). For example, epidermal growth factor stimulates the growth of many cell lines, but inhibits growth in some breast cancer cell lines (26). Interleukin-6 stimulates the proliferation of hepatocytes but inhibits the growth of some myeloma cell lines (27). Thrombin has been shown to be able to induce the proliferation of endothelial cells, smooth muscle cells, and fibroblasts (6). However, in comparison to platelet-derived growth factor and fibroblast growth factor-stimulated cell growth, thrombin-stimulated vascular smooth muscle proliferation is delayed and requires the de novo expression of one or more autocrine mitogens (23). Thrombin may also have a dual biphasic effect on tumor cell growth, with low concentration inducing mitogenesis and higher concentration impairing cell growth, as we have observed with 3 other cell lines. 2 It is also conceivable that various types of cells respond to thrombin differently, possibly on the basis of their thrombin receptor repertoire, or other intracellular messenger factors. The level of prothrombin in serum ranges from 1 to 5 nM. Therefore, the concentration of thrombin produced locally could be high enough to inhibit the growth of some types of cells via up-regulation of p21 waf/cip1 and/or induce cell death via apoptosis.
An increase of the amount of p21 waf/cip1 relative to the amount of cyclin-bound CDK may convert active CDK complexes into inactive ones. Whether the genes that control the cell cycle are regulated by thrombin is not well defined. In this study, we have demonstrated that thrombin-induced expression of p21 waf/cip1 is associated with thrombin receptor activa-tion. Thrombin-induced expression of p21 waf/cip1 was inhibited by hirudin, a specific thrombin antagonist. Activation by TRAP 14-mer confirmed that activation of the PAR-1 thrombin receptor could induce increased p21 waf/cip1 expression. It should be noted that although TRAP can activate non-thrombin receptor PAR-2 as well, it has a substantially low affinity for this receptor which cannot be activated by thrombin (28). Although thrombin can activate PAR-3 and PAR-4, they were not present on the DU145, MDAMB231, and MCF7 cells studied. In addition, the TRAP 14-mer used in this study can only activate PAR-1, not activate PAR-3 and PAR-4. Thus, the increased expression of p21 waf/cip1 by thrombin is induced by PAR-1 activation.
Expression of p21 waf/cip1 is induced by the tumor suppressor protein p53, which facilitates G 1 arrest in response to DNA damage (3,4). It has been reported that expression of p21 is rarely p53-independent, i.e. p21 waf/cip1 expression is not preserved in cells of p53 knockout mice (3,4) and MyoD, a skeletal muscle-specific transcriptional regulator as well as antioxidant induces p53-independent activation of p21 waf/cip1 (29,30). Therefore, the possible involvement of p53 in thrombin-stimulated p21 waf/cip1 expression was studied using the p53 knockout cell line as well as the tumor cell line deficient in p53. The thrombin-induced up-regulation of p21 waf/cip1 expression in the 2 cell lines strongly suggests that the increased expression of p21 waf/cip1 is p53-independent. It has been reported that the expression of p21 waf/cip1 is directly induced by activated STAT1 protein, which specifically recognizes the conserved STAT-responsive elements in the promoter of p21 waf/cip1 (24). Our study indicates that JAK2 and STAT1 are activated by thrombin. Furthermore, p21 waf/cip1 mRNA was induced by thrombin in CDϩ cells but not in STAT-1 knockout CDϪ cells. Thus, STAT-1 is required in these cells for increased expression of p21 waf/cip1 in response to thrombin. STAT proteins are known to be activated by a wide variety of ligands including an increasing number of cytokines whose receptors belong to the large cytokine-receptor superfamily as well as a number of other growth factors, whose receptors exhibit tyrosine kinase dependent activity (31). Except for angiotensin II, ligands that bind to 7-transmembrane domain receptors have not been reported to activate STATs (32). It is of interest that phosphorylation of JAK2, which could induce STAT protein activation, has been reported in thrombin-stimulated human platelets (33). We hypothesize that Gprotein-coupled receptor activation by thrombin induces tyrosine phosphorylation of JAK2 which creates recruitment sites for STAT transcription factor activation, leading to its dimerization and nuclear translocation.
Thrombin inhibited the growth of several tumor cell lines. To determine whether up-regulation of p21 waf/cip1 alone is sufficient for the inhibition of the growth of several lines of tumor cells, the role of thrombin in the induction of apoptosis was also investigated. Our results, in agreement with our recent observations with murine B16F10 melanoma and human HCT8 colon carcinoma cells, 2 and that recently reported for neurons and astrocytes (34), indicate that thrombin also induces apoptosis in other cell lines examined, via the PAR-1 receptor and therefore may be a growth inhibitor for certain types of cells. It is of interest that CDϪ cells were 1.5-fold more resistant to thrombin-induced impaired cell growth than CDϩ cells. The absence of complete resistance with CDϪ cells suggests that other mechanisms may be involved in thrombin-induced impaired cell growth other than p21 induction or apoptosis; or that the counterbalancing effect of enhanced cell growth induced by thrombin, possibly by the release of platelet-derived growth factor and basic fibroblast growth factor may play a role (35,36).
The up-regulation of caspases by thrombin was demonstrated in DU145, CHRF, and mouse fibroblasts, but not in mouse cells without functional STAT1, indicating that induction of caspases by thrombin requires the STAT1 pathway. It is of interest in this regard that STAT1 is required for efficient constitutive expression of caspases 1, 2, and 3 in human fibrosarcoma cells and STAT1-null cells are resistant to apoptosis by tumor necrosis factor-␣ (37). The present study demonstrates that thrombin, like tumor necrosis factor-␣, can induce apoptosis via STAT1-mediated up-regulation of caspases.
Our study establishes, for the first time, a link between PAR-1 receptor activation with the STAT signal pathway, which leads to cell cycle control and apoptosis. This observation broadens our understanding of the mechanism of PAR-1 activation and its effect on cell growth, and could possibly lead to therapeutic approaches for the treatment of cancer. For example, an adenoviral vector containing a constitutively active PAR-1 (38) regulated by a promoter specific for a malignant tissue may be of clinical benefit. Studies designed to explore this interesting hypothesis are currently in progress.