Plasmin-induced Migration Requires Signaling through Protease-activated Receptor 1 and Integrin α9β1*

Plasmin is a major extracellular protease that elicits intracellular signals to mediate platelet aggregation, chemotaxis of peripheral blood monocytes, and release of arachidonate and leukotriene from several cell types in a G protein-dependent manner. Angiostatin, a fragment of plasmin(ogen), is a ligand and an antagonist for integrin α9β1. Here we report that plasmin specifically interacts with α9β1 and that plasmin induces of cells expressing migration recombinant α9β1 (α9-Chinese hamster ovary (CHO) cells). Migration was dependent on an interaction of the kringle domains of plasmin with α9β1 as well as the catalytic activity of plasmin. Angiostatin, representing the kringle domains of plasmin, alone did not induce the migration of α9-CHO cells, but simultaneous activation of the G protein-coupled protease-activated receptor (PAR)-1 with an agonist peptide induced the migration on angiostatin, whereas PAR-2 or PAR-4 agonist peptides were without effect. Furthermore, a small chemical inhibitor of PAR-1 (RWJ 58259) and a palmitoylated PAR-1-blocking peptide inhibited plasmin-induced migration of α9-CHO cells. These results suggest that plasmin induces migration by kringle-mediated binding to α9β1 and simultaneous proteolytic activation of PAR-1.

Integrins are a family of heterodimeric cell adhesion receptors that regulate many critical cellular processes including cell migration, adhesion, signal transduction, and gene expression. By controlling such cellular behaviors, they play a role in mediating important biological functions such as embryonic development, differentiation, and wound healing as well as pathological processes such as inflammation, angiogenesis, and tumor progression. Integrin ␣ 9 ␤ 1 is constitutively expressed in liver and smooth and skeletal muscles as well as in squamous and airway epithelium (1)(2)(3). ␣ 9 ␤ 1 is also expressed on neutrophils and is up-regulated after neutrophil activation (4,5). Its role in activated neutrophils may involve binding to vascular cell adhesion molecule on endothelial cells resulting in chemotaxis across endothelial monolayers by interaction with vascular cell adhesion molecule 1 during inflammation (5,6). ␣ 9 ␤ 1 recognizes many other ligands such as tenascin-C (7) and os-teopontin (8), both extracellular matrix ligands, and ADAMs (a disintegrin and metalloprotease) (9,10). Thus, this integrin may potentially play a crucial role in inflammatory responses and metastasis as well as during development and wound healing.
Some coagulation cascade proteases play a role in wound healing, tissue remodeling, angiogenesis, and metastasis, both by extracellular matrix degradation and by regulating cell migration and proliferation (11). One such protease, plasmin, is generated from its precursor plasminogen. Plasminogen is first converted to the two-chain serine protease plasmin by the cleavage of a single Arg 561 -Val 562 peptide bond by urokinasetype plasminogen activator (uPA), 1 and plasmin serves as both the substrate and enzyme for the generation of angiostatin (12). Angiostatin contains either the first three or first four kringle domains of plasminogen (13,14). After plasminogen is cleaved by uPA, the kringle domain of plasmin is still attached to the catalytic domain by interchain-disulfide linkage.
It has been demonstrated in various systems that plasmin can activate signaling leading to protein phosphorylation, Ca 2ϩ mobilization, and activation of phospholipase C and protein kinase C (15)(16)(17)(18). Plasmin has been shown to increase arachidonate release in bovine artery endothelial cells and to trigger leukotriene B4 release in peripheral monocytes in a pertussis toxin-sensitive manner (15,18). A few recent reports suggest that plasmin can induce cell migration, e.g. in human peripheral blood monocytes, and other processes important in angiogenesis (19). However, very little is known regarding the mechanisms and molecules participating in plasmin signaling. Recent studies suggest that plasmin may also mediate its effects through activation of a G protein-coupled receptor-mediated signaling pathway (20,21). This is supported by the fact that some effects of plasmin are sensitive to pertussis toxin, which inhibits G i/o family proteins, suggesting the involvement of such a G protein (15,18,22). However, the signal transduction mechanism downstream of plasmin activation remains largely unknown.
Protease-activated receptors (PARs) are seven transmembrane-spanning G protein-coupled receptors that are activated by N-terminal cleavage to expose a tethered ligand (23,24). G protein-coupled receptors are known to induce stimulation of heterotrimeric G proteins and can regulate various signaling cascades leading to cellular responses such as gene expression, mitogenesis, and cell motility. Thrombin, the major protease in the coagulation cascade, elicits its responses through activation of PARs. PAR-1 (thrombin receptor) and PAR-4 are expressed on human platelets, and their activation mediates platelet aggregation and plays a role in hemostasis and thrombosis. PARs also have been shown to play a role in cell invasion during breast carcinoma metastasis and placental implantation (25). The role of PARs in metastasis and invasion relies on its ability to regulate chemotaxis, mitogenesis, and adhesion by PAR-1 signaling in a ␣ v ␤ 5 -dependent manner (26). Recent reports indicate that plasmin may signal through PAR-1 (22). It is somewhat surprising that PAR-1, which is an ideal substrate for thrombin, also serves as the signaling receptor for plasmin that has very different substrate specificity. However, the concept is emerging that co-receptors play important roles in regulating the activation of PARs. PAR-1 has been shown to be activated efficiently by receptor-targeted proteases in the initiation of coagulation (by the tissue factor-factor VIIa-factor Xa complex) (27) and of the anticoagulant pathway (by activated protein C bound to the endothelial cell protein C receptor) (28). Whether integrins play similar roles as co-receptors in plasmin signaling through PARs has not been explored.
In this study, we demonstrate that plasmin, a serine protease and the parent molecule of angiostatin, specifically binds to integrin ␣ 9 ␤ 1 through its kringle domains to induce signaling. The pro-migratory activity of plasmin requires ␣ 9 ␤ 1 and the catalytic activity of plasmin. We show that PAR-1 is involved in plasmin-induced cell migration using PAR agonist peptides and small molecule PAR-1 inhibitors. Thus, both ␣ 9 ␤ 1 and PAR-1 are critical for plasmin-induced cell migration.
Cells-D. Sheppard kindly provided Chinese hamster ovary (CHO) cells expressing human ␣ 9 ␤ 1 (designated ␣9-CHO cells). As a control, CHO cells were transfected only with vector (pBJ-1) together with the neomycin gene and selected for G418 resistance (designated mock-CHO cells).
Adhesion Assays-Adhesion assays were performed as previously described (30,31). We coated wells in 96-well Immulon-2 microtiter plates (Dynatech Laboratories, Chantilly, VA) with 100 l of phosphate-buffered saline (10 mM phosphate buffer, 0.15 M NaCl, pH 7.4) containing substrates at a concentration of 50 -500 nM and incubated 1 h at 37°C. We blocked the remaining protein-binding sites by incubating with 0.2% BSA (Calbiochem) for 1 h at room temperature. Cells (10 5 cells/well) in 100 l of Hepes-Tyrode buffer (10 mM Hepes, 150 mM NaCl, 12 mM NaHCO 3 , 0.4 mM NaH 2 PO 4 , 2.5 mM KCl, 0.1% glucose, 0.02% BSA) supplemented with 2 mM MgCl 2 were added to the wells and incubated at 37°C for 1 h unless stated otherwise. After non-bound cells were removed by rinsing the wells with the same buffer, bound cells were quantified by measuring endogenous phosphatase activity (32). For inhibition assays, cells were preincubated with mAbs and peptides on ice for 10 min before cells were added to plasmin-coated wells.
Migration Assays-Cell migration was analyzed using tissue culturetreated 24-well Transwell plates (Costar, Cambridge, MA) with polycarbonate membranes of a pore size of 8 m. The lower side of the filter was coated with various concentrations of substrates and blocked with 0.5% BSA. Coated filters were placed into serum-free migration buffer (Dulbecco's modified Eagle's medium, 10 mM Hepes, 0.5% BSA, and 1ϫ penicillin-streptomycin), and cells (100 l) suspended in the same buffer (2 ϫ 10 5 cells/ml) were added to the upper chamber. The cells were incubated at 37°C in 5% CO 2 for 20 h. Cells on the upper surface of the membrane were removed by gently wiping with a cotton swab, and cells that migrated to the lower surface of the filters were fixed and stained with 0.5% crystal violet in 2% ethanol and counted using high magnification microscopy. The result is the mean cell number of four randomly selected, high magnification microscopic fields from duplicate wells in two or three separate experiments. To verify quantification of the mean number of migrated cells, we took a picture of the stained membrane and determined the number of cells on the entire lower surface of the membrane using ImageJ software (rsb.info.nih.gov/ij/). We obtained essentially identical results using either method. Antiintegrin mAbs (10 g/ml), angiostatin (500 nM), or other inhibitory peptides or mAbs were preincubated with cells for 15 min at 37°C prior to the assay. When the effect of aprotinin on plasmin-induced cell migration was tested, aprotinin at designated concentrations was incubated with immobilized plasmin. Free aprotinin was removed before starting the migration assays.
Staining Stress Fibers-Stress fiber staining was carried out as described previously (34). Visualization of cell spreading and morphol-FIG. 1. Plasmin binding to ␣ 9 -CHO cells is dose-dependent and inhibited by mAbs. A, adhesion of ␣ 9 -CHO cells to plasmin as a function of plasmin-coating concentrations. Adhesion of ␣ 9 -CHO cells to plasmin was also compared with that of mock (vector only)-transfected CHO cells. Adhesion assays were carried out as described under "Experimental Procedures." Plasmin at different concentrations was incubated in wells of 96-well microtiter plates, and the wells were then blocked with BSA. Cells were added in Tyrode-Hepes buffer supplemented with 1 mM MgCl 2 . After incubating for 1 h at 37°C, bound cells were quantified as described. Data are shown as the mean Ϯ S.D. from three separate experiments. B, the effect of anti-integrin and antikringle agents on adhesion of ␣ 9 -CHO cells to plasmin. 100 nM plasmin was used for coating wells. mAbs Y9A2 (anti-␣ 9 ), KH72 (anti-␣ 5 ), P1F6 (anti-␣ v ␤ 5 ), and mAb 51 (anti-plasminogen kringle) were used at 4 g/ml. ⑀-Aminocaproic acid, a Lys analogue, was used at 200 mM. RGD and RGE peptides were used at 100 M.
ogy on substrates and actin fiber staining was carried out in cells that were incubated in 0.5% serum for 18 h and resuspended in Hepes-Tyrode buffer with 2 mM MgCl 2 at 2 ϫ 10 4 cells/coverslip before plating on poly-L-lysine-, fibronectin-, plasmin-, or angiostatin-coated coverslips. After allowing cells to spread for 90 min at 37°C, they were fixed in 3% paraformaldehyde and visualized by phase-contrast microscopy. For actin fiber staining, the fixed cells were permeabilized with 0.5% Triton X-100. The actin fibers of these fixed cells were then stained with fluorescein isothiocyanate-conjugated phalloidin. Actin fiber morphology was visualized using confocal microscopy.
Immunoblotting of PAR-1 in CHO Cells-Membrane proteins of CHO cells were prepared essentially as described previously (33). Confluent CHO cells were harvested and resuspended in 0.25 ml of extraction buffer (10 mM Tris/HCl buffer, pH 7.4, containing 10 mM NaCl, 25 mM NaF, 1 mM EDTA, 2 mM EGTA, 1 mM dithiothreitol, 1 mM phenylmethylsulfonyl fluoride, and protease inhibitors leupeptin, aprotinin, pepstatin, and bestatin) and lysed by sonication. The lysate was centrifuged at 15,000 ϫ g for 20 min, and the supernatant was further centrifuged at 100,000 ϫ g for 1 h. The pellet was resuspended in 100 l of the extraction buffer containing 1.0% Nonidet P-40, rocked on ice for 1 h, and centrifuged at 15,000 ϫ g for 15 min. The supernatant (containing membrane proteins) was analyzed by SDS-PAGE using 10% acrylamide gels (15 g of protein/lane) and transferred to Immobilon-P membrane (Millipore) for immunoblotting. The membrane was blocked with 5% nonfat dry milk in phosphate-buffered saline and incubated with a rabbit polyclonal antibody against PAR-1 (H-111, Santa Cruz Biotechnology, Santa Cruz, CA) or with control rabbit serum at 4°C overnight. Bound IgG was detected using horseradish peroxidase-linked anti-rabbit IgG (Cell Signaling Technology, Beverly, MA) and ECL detection reagents (Amersham Biosciences).

RESULTS
Plasmin Binding to Integrin ␣ 9 ␤ 1 -Because ␣ 9 ␤ 1 on endothelial cells binds to angiostatin but not to plasminogen (35), it is possible that proteolytic activation of plasminogen to plasmin produces a similar conformational change to expose the integrin-binding sites located in the kringle domains. We found that plasmin supported the adhesion of ␣ 9 -CHO cells to a greater extent than it did of mock-transfected CHO cells (Fig.  1A) (30). Anti-␣ 9 mAb Y9A2 significantly blocked the adhesion of ␣ 9 -CHO cells to plasmin, but control anti-␣ 5 (KH72) or anti-␣ v ␤ 5 mAb (P1F6) did not block adhesion (Fig. 1B). RGD peptides did not completely block the adhesion of ␣ 9 -CHO cells to plasmin (consistent with other reported ␣ 9 ␤ 1 -ligand interactions). This finding also indicates that the integrin-binding sites within the kringle domain of plasminogen are exposed by proteolytic activation of plasminogen to plasmin.
The kringle domains of plasminogen have multiple Lys-binding sites (36). We have shown that a Lys analogue (⑀-aminocaproic acid) almost completely blocked plasmin binding to ␣ 9 ␤ 1 . Also, a mAb against plasminogen kringles (mAb 51) blocked interaction between plasmin and this integrin (Fig. 1B). These results suggest that plasmin specifically binds to ␣ 9 ␤ 1 through the kringle domains as in the case of angiostatin.
We previously found that angiostatin did not induce cell spreading or stress fiber formation in ␣ v ␤ 3 -expressing ␤ 3 -CHO cells (35). Here. we find that ␣ 9 -CHO cells also did not spread or form stress fibers on angiostatin (Fig. 2). In contrast, plasmin induced cell spreading and stress fiber formation in ␣ 9 -CHO cells (Fig. 2), suggesting that plasmin transduces signals that cause cytoskeletal reorganization upon binding to integrins. Control experiments with mock-CHO (vector-transfected) on these substrates demonstrate that these cells do not spread on poly-L-lysine, angiostatin, or plasmin (data not shown). Thus, plasmin specifically signals through integrin ␣ 9 ␤ 1 (Fig. 2) and integrin ␣ v ␤ 3 (35) to induce cell spreading, whereas plasmin does not signal through endogenous integrins on mock-transfected CHO cells (e.g.

Plasmin Induces Haptotaxis of ␣ 9 -CHO Cells in an Integrinand Catalytic Activity-dependent Manner, and Angiostatin (K1-3) Blocks the Plasmin-induced
Migration-It has been reported that plasmin is a potent chemoattractant for human peripheral monocytes (19). We studied whether plasmin induces migration in an integrin-dependent manner using a modified Boyden chamber assay. Notably, plasmin induced haptotaxis of ␣ 9 -CHO cells but not significantly of mock-transfected CHO cells (Fig. 3A). ␣ 9 -CHO cells migrated at a much higher rate than mock-transfected CHO cells. This is consistent with the previous observation that ␣ 9 ␤ 1 significantly enhances cell migration (37). An anti-␣ 9 mAb Y9A2 effectively blocked plasmin-induced haptotaxis of ␣ 9 -CHO cells, whereas the anti-␣ v ␤ 5 mAb P1F6 did not block migration (Fig. 3B), suggesting that ␣ 9 ␤ 1 integrin in particular is critical for plasmin-induced migration. Notably, we found that soluble angiostatin effectively blocked plasmin-induced haptotaxis of ␣ 9 -CHO cells in a dosedependent manner (Fig. 3C).
Because angiostatin does not induce migration of these cells (35), the serine protease activity of plasmin may be important for plasmin-induced migration. Consistently, aprotinin, a ser-

FIG. 2. Plasmin induces spreading of ␣ 9 -CHO cells, but not mock-CHO.
Top, ␣ 9 -CHO cells were plated on glass coverslips coated with angiostatin (fragment K1-3, 500 nM) or plasmin (200 nM) in the absence of serum and allowed to spread for 2 h at 37°C, and then they were fixed in 3.7% formaldehyde and visualized by phase-contrast microscopy to assess cell morphology. Mock-CHO cells on K1-3 or plasmin showed morphology similar to that of ␣ 9 -CHO cells on K1-3 (not shown). Bottom, ␣ 9 -CHO cells were plated on glass coverslips coated with poly-L-lysine (50 g/ml), angiostatin (K1-3, 500 nM), or plasmin (200 nM) in the absence of serum, allowed to spread, and then fixed and stained for actin stress fiber formation, and their cytoskeletal morphology was visualized by confocal microscopy as described under "Experimental Procedures." ine protease inhibitor, effectively blocked the haptotaxis of these cells on plasmin (Fig. 3D). Mock-transfected CHO cells do not migrate to a significant extent on plasmin and are not affected by aprotinin under the conditions used. These results suggest that the catalytic activity of plasmin is critical for plasmin-mediated migration and that blocking plasmin-induced migration of endothelial cells is a potential mechanism by which angiostatin can exert its effects.
The Role of PAR in Plasmin Signaling.-It has recently been reported that plasmin transduces signals through activating PAR-1 (22), but it is unclear whether integrins are required for this signaling process. We have tested whether PARs are involved in plasmin-induced migration of ␣ 9 -CHO cells. Tethered ligands (PAR agonist peptides TFLLRN for PAR-1, SLIGRL for PAR-2, and GYPGQV for PAR-4) activate PARs without proteolysis. We tested whether PAR agonist peptides can mimic the possible plasmin activation of PARs and stimulate cell migration. We used angiostatin that binds to ␣ 9 ␤ 1 but does not by itself induce migration of ␣ 9 -CHO cells as a haptotactic substrate. We found that PAR-1 agonist, but not PAR-2 or PAR-4 agonist, significantly induced the migration of ␣ 9 -CHO cells on angiostatin (Fig. 4). These results suggest that PAR-1 is a likely candidate that mediates plasmin-induced signaling.
RWJ 58259 is a synthetic PAR-1 antagonist that efficiently blocks PAR-1-mediated signaling (38). We found that RWJ 58259 inhibited plasmin-induced migration of ␣ 9 -CHO cells (Fig. 5). These results support the notion that PAR-1 is primarily responsible for plasmin-induced migration of ␣ 9 -CHO cells. We also found that a palmitoylated peptide, which specifically recognizes the third intracellular loop of PAR-1 and blocks G protein activation (pal1, also known as pepducin) (39), can significantly block PAR-1 agonist peptide-induced migration of ␣ 9 -CHO cells, whereas a similar palmitoylated inhibitory peptide for PAR-4 (pal4) did not induce migration (Fig. 6A). Mocktransfected CHO cells did not respond to stimulatory PAR-1 and PAR-4 agonist peptides or inhibitory pal1 and pal4 peptides. This finding suggests that PAR-1 signaling specifically cooperates with ␣ 9 ␤ 1 . Plasmin-induced migration of ␣ 9 -CHO cells is also significantly inhibited by pal1 but not by pal4 (Fig.  6B). These experiments further support a role for PAR-1 in plasmin-induced migration of ␣ 9 -CHO cells. We have confirmed the expression of PAR-1 in CHO and ␣ 9 -CHO cells by immu- FIG. 3. Plasmin induces migration of ␣ 9 -CHO cells and is inhibited by mAbs, angiostatin, or aprotinin. A, migration of ␣ 9 -CHO and mock-transfected CHO cells was determined as a function of plasmin concentration (nM). Migration assays were performed as described under "Experimental Procedures" using modified Boyden chambers. B, cell migration was measured on membranes coated with 0.5% BSA (control, "no coat") or plasmin (200 nM) in the presence of anti-integrin mAbs (10 g/ml) (Y9A2, anti-␣ 9 ; P1F6 anti-␣ v ␤ 5 ). C, cell migration on plasmin (200 nM) was measured in the presence of increasing concentrations of angiostatin (K1-3). D, the effect of a serine protease inhibitor aprotinin on cell migration on plasmin (200 nM) was determined. Immobilized plasmin was incubated with increased concentrations of aprotinin. Data are shown as the mean Ϯ S.D. from three separate experiments done in triplicate relative to the migration level of ␣ 9 -CHO cells on non-coated wells. noblotting (Fig. 6C), which is consistent with the previous report that CHO cells express functional hamster PAR-1 (27). These results demonstrate that integrins ␣ 9 ␤ 1 serves as a migration-stimulating co-signaling receptor for plasmin and displays specificity to PAR-1. DISCUSSION We have previously shown that, whereas angiostatin binds to integrin ␣ 9 ␤ 1 , plasminogen does not bind (35). These results establish that plasmin, similar to angiostatin, binds to integrin ␣ 9 ␤ 1 through the kringle domains. Our results suggest that the ␣ 9 ␤ 1 -binding sites in the kringle domains are exposed by proteolytic cleavage of plasminogen. It should be noted that the kringle domain of plasmin is still attached to the catalytic domain by interchain-disulfide linkage. This study demonstrates that plasmin binding to ␣ 9 ␤ 1 is required for cell migration and that the catalytic activity of plasmin is also essential for this process. ␣ 9 -CHO cells showed much more pronounced migration than mock-transfected CHO cells, which is consistent with the previous observation (37), and plasmin further enhanced the migration of ␣ 9 -CHO cells.
␣ 9 ␤ 1 -expressing CHO cells did not form stress fibers on poly-L-lysine or angiostatin (K1-3) but did spread and form stress fibers on plasmin. Control experiments with mock-CHO cells on these substrates demonstrate that these cells do not spread well on poly-L-lysine, angiostatin, or plasmin. These results suggest that plasmin signaling requires ␣ 9 ␤ 1 . However, mock-CHO cells do show a low level of binding and migration, which may be because of the binding of other integrins (e.g. ␣ 5 ␤ 1 , ␣ v ␤ 5 , and ␣ v ␤ 1 ) that are endogenous in CHO cells. Such endogenous receptors may also be the cause of the partial inhibition of adhesion of ␣ 9 -CHO cells seen on plasmin by the RGD peptide and the anti-␣ 9 mAb.
We also provide direct evidence that PAR-1 is involved in plasmin-induced signaling in ␣ 9 -CHO cells using PAR agonist peptides that substitute for the catalytic activity of plasmin and induce migration of ␣ 9 -CHO cells on angiostatin. Consistently, we also found that the inhibition of PAR-1 activation by treatment with a small molecule inhibitor, RWJ 58259, or a palmitoylated inhibitory peptide against PAR-1 (pal1) is effective in blocking plasmin-induced migration of ␣ 9 -CHO cells. Taken together, the present evidence suggests that PAR-1 is critical for plasmin-mediated migration of ␣ 9 -CHO cells.
Plasmin has an ϳ10-times lower affinity for PAR-1 in comparison to thrombin (40). Thus, the activation of PAR-1 requires a mechanism to concentrate plasmin on the cell surface. The binding of plasmin to ␣ 9 ␤ 1 would significantly increase the cell surface plasmin concentration. The binding of plasmin to the integrins through the kringle domain would also prevent the inactivation of plasmin, because circulating abundant ␣ 2 -antiplasmin rapidly inactivates free plasmin dependent on interactions with the kringle domain of plasmin (for a review, see Ref. 41). In addition, uPA and uPA receptor are widely expressed by proliferating tumor cells (for review, see Ref. 42). Because uPA/uPA receptor binding can convert plasminogen to plasmin, tumor cells will generate plasmin locally and a high pericellular plasmin level can be expected in tumor cells. It has recently been reported that plasmin induces PAR-1 activation, leading to mitogen-activated protein kinase activation at very low plasmin concentrations on fibroblast (22). This is not consistent with a previous report (43) that claims that plasmin cleaves PAR-1 only inefficiently. We suspect that plasmin may be concentrated at the cell surface of fibroblasts through binding to one or more integrins in this case as well.
The present study establishes that plasmin-induced cell migration requires both the binding of plasmin to integrin ␣ 9 ␤ 1 through the kringle domains and activation of PAR-1 by the catalytic activity of plasmin. We have reported that endothelial cells (bovine artery endothelial cells) express both ␣ 9 ␤ 1 and ␣ v ␤ 3 (35) and are involved in plasmin signaling. We also found using fluorescence-activated cell sorter analysis that calf pul- monary artery endothelial cells, which have been widely used for angiogenesis studies (44 -47), express ␣ 9 ␤ 1 and ␣ v ␤ 3 . 2 These findings suggest that both ␣ 9 ␤ 1 and ␣ v ␤ 3 may support plasmin-induced signals in endothelial cells and may be involved in the inhibitory effect of angiostatin.
We have previously shown that plasmin-induced haptotaxis of ␤ 3 -CHO cells required the catalytic activity of plasmin and binding to ␣ v ␤ 3 integrins and that angiostatin effectively blocked plasmin-induced haptotaxis of ␤ 3 -CHO cells. ␣ v ␤ 3 integrin is highly expressed in angiogenic endothelial cells in tumors, wounds, or inflammatory tissues (30). Although angiostatin has been shown to inhibit angiogenesis by blocking endothelial cell proliferation and migration (48,49), there was conflicting data on the underlying mechanism of action. Our study (35) provides insight into a possible mechanism by which angiostatin may inhibit angiogenesis when ␣ v ␤ 3 integrin is involved. The present findings indicate that angiostatin would simultaneously block plasmin-induced migration of cells that express ␣ v ␤ 3 -as well as ␣ 9 ␤ 1 -integrins.
Angiostatin may similarly exert effects on ␣ 9 ␤ 1 -dependent migration in neutrophils, hepatocytes, smooth muscle, or epithelial cells (that express ␣ 9 ␤ 1 ). It has been shown that the ␣ 9 -cytoplasmic domain (like the highly homologous ␣ 4 -cytoplasmic domain) preferentially enhances cell migration (37). It is interesting that both ␣ 9 ␤ 1 and ␣ 4 ␤ 1 integrins are expressed on leukocytes, which are cells that are required to migrate rapidly at sites of inflammation. However, ␣ 9 ␤ 1 and ␣ 4 ␤ 1 appear to regulate different signaling pathways that both lead to increased migration. The present study used cells overexpressing ␣ 9 ␤ 1 to study plasmin-induced migration and signaling in a manner specific to ␣ 9 ␤ 1 . The use of this artificial system would be justified and essential to study ␣ 9 ␤ 1 -specific plasmin signaling considering that endothelial cells, monocytes, neutrophils, and other natural cells express several integrins other than ␣ 9 ␤ 1 that interact with plasmin and angiostatin (e.g. ␣ v ␤ 3 and perhaps other integrins). Elucidating the signaling pathways downstream of ␣ 9 ␤ 1 and their potential inhibitory agents (e.g. angiostatin, PAR-1 inhibitory compounds, serine protease inhibitors, and so on) may be relevant to understanding processes such as inflammation and wound healing in these systems.