Domain-dependent action of urokinase on smooth muscle cell responses

Domain-dependent action of urokinase on smooth muscle cell responses William J. Tanski, MD, Allison J. Fegley, MD, Elisa Roztocil, BS, and Mark G. Davies, MD, PhD, Rochester, NY Background: Single-chain urokinase-type plasminogen activator (sc-uPA) is one of the key serine proteases involved in modulating cellular and extracellular matrix responses during tissue remodeling. Sc-uPA is composed of three domains: aminoterminal fragment (ATF), kringle domain, and carboxyterminal fragment (CTF). sc-uPA is readily cleaved into these three domain fragments in vitro, each of which is biologically active; however, their roles in the microenvironment of the vessel wall are poorly understood. Purpose: The purpose of this study was to determine the role of each domain of sc-uPA on vascular smooth muscle cell (SMC) proliferation and migration. Methods: SMCs were cultured in vitro. Assays of DNA synthesis, cell proliferation, and migration were performed in response to sc-uPA, ATF, kringle, and CTF in the presence and absence of the plasmin inhibitors ⑀-aminocaproic acid (EACA) and aprotinin, the G␣i inhibitor pertussis toxin, and the mitogen-activated protein kinase 1 (the upstream regulator of the extracellular-signal regulated kinase [ERK]) inhibitor PD98059. Results: sc-uPA produced dose-dependent increases in DNA synthesis and cell proliferation. These responses were dependent on the CTF domain and were sensitive to plasmin inhibitors, pertussis toxin, and PD98059. Sc-uPA also induced SMC migration, which could be elicited by both ATF and kringle. Migration to sc-uPA, ATF, and kringle was both pertussis toxin and PD98059 sensitive, but importantly was plasmin-independent. Conclusion: sc-uPA induces SMC proliferation and migration, which are domain-dependent and mediated in part by G␣i-linked, ERK-dependent processes, while only the mitogenic response is protease dependent. These findings suggest that migration is linked to a G-protein coupled nonprotease receptor, while proliferation is associated with a G-protein coupled protease receptor. (J Vasc Surg 2004;39:214-22.)

Recurrent stenosis, the anatomic result of intimal hyperplasia and remodeling, remains an obstacle to the longterm patency of conventional and endoluminal vascular interventions.1 Intimal hyperplasia involves an integrated program of cell proliferation, migration, and extracellular matrix modulation. Proliferation and migration of smooth muscle cells (SMCs) involves the complex regulation of proteases, integrins, and extracellular molecules within the microenvironment of the vessel wall. Urokinase plasminogen activator (uPA), a serine protease, is the primary plasminogen activator in tissue remodeling,2 and is secreted as a single chain subunit (sc-uPA) and localized to the membrane by its receptor, the uPA receptor (uPAR).3 Within this microenvironment uPA protease activity is controlled by plasminogen activator inhibitor-1 (PAI-1).4 sc-uPA is cleaved in vitro into three fragments: aminoterminal (ATF), kringle, and carboxyterminal (CTF). Matrix metalFrom the Department of Surgery, University of Rochester, Strong Memorial Hospital. Dr Davies supported by the American College of Surgeons Junior Faculty Award and by the Mentored Clinical Scientist Development Award, NIH-NHLBI/Lifeline Foundation (K08 HL 67746). Dr Tanski supported by NIH National Research Service Award (F32 HL69598-01). Competition of interest: none. Additional material for this article may be found online at www.mosby. com/jvs. Reprint requests: W. J. Tanski, MD, University of Rochester/Strong Memorial Hospital, Department of Surgery, Box SURG, 601 Elmwood Ave, Rochester, NY 14642 (e-mail: [email protected]). 0741-5214/2004/$30.00 ⫹ 0 Copyright © 2004 by The Society for Vascular Surgery. doi:10.1016/S0741-5214(03)01031-0

214

loproteinase (MMP)–3 cleaves uPA in vitro into ATF and CTF fragments with the same functional domains as those used in our experiments5; the likelihood that similar cleavage products occur in vivo is high. Each fragment has unique characteristics in the microenvironment of the vessel wall.6,7 We tested the hypothesis that sc-uPA induces SMC proliferation and migration through G␣i-linked receptors and a protease-mediated, extracellular signal-regulated kinase (ERK)– dependent process. METHODS Experiment design. Sprague-Dawley rat vascular SMCs were cultured in vitro. Standard assays of DNA synthesis (thymidine incorporation), cell proliferation (manual cell counting), and migration (linear wound assay) were used in response to sc-uPA, ATF, kringle, and CTF in the presence and absence of the plasmin inhibitors aprotinin and ⑀aminocaproic acid (EACA), the G␣i inhibitor pertussis toxin, the G␣q inhibitor GP-2A, and the mitogen activated protein kinase 1 (MEK1) inhibitor PD98059. Materials. Single-chain uPA, ATF, kringle, and CTF were purchased from American Diagnostica (Greenwich, Conn). According to the manufacturer, sc-uPA was purified from the media of an sc-uPA–secreting kidney cell line, and purified with affinity column chromatography, resulting in 100% zymogen with a specific activity of 90,000 IU/mg urokinase equivalent. CTF was produced by controlled proteolysis of uPA isolated from human urine, and contains more than 95% low molecular weight uPA (CTF) and less than 5% high molecular weight uPA (intact uPA).

JOURNAL OF VASCULAR SURGERY Volume 39, Number 1

The specific activity of the resultant product was 160,000 IU/mg, and PAI-1 binding was confirmed. ATF is a recombinant product representing amino acids 1 to 143 of human uPA, including the growth factor–like and kringle domains, and specifically binds with high affinity to the uPA receptor. EACA, aprotinin, and pertussis toxin were purchased from Sigma (St Louis, Mo). PD98059 was purchased from Calbiochem (La Jolla, Calif). GP-2A was purchased from Biomol (Plymouth Meeting, Pa). Peroxidase-conjugated anti-rabbit immunoglobulin G (IgG) antibody was purchased from Jackson ImmunoResearch Laboratories (West Grove, Pa). Peroxidase-conjugated antimouse IgG antibody was purchased from BioRad Laboratories (Hercules, Calif). Phospho-ERK antibody was purchased from Promega (Madison, Wis). Total ERK antibody was purchased from BD Transduction Laboratories (Lexington, Ky). Delbecco modified Eagle medium (DMEM) and deionized phosphate-buffered saline solution were purchased from Cellgro (Herndon, Va). [H3]-thymidine incorporation. The technique used for [H3]-thymidine incorporation has been described in detail.8 In brief, 24-well cluster dishes (Falcon Plastics, Madison, SD) were inoculated with 5 ⫻ 104 cells per well and allowed to reach about 70% confluence. The medium was changed to DMEM for 2 days to induce cell quiescence. [H3]-thymidine (1 ␮Ci/mL; ICN, Irvine, Calif) was added with serum (10% fetal bovine serum), sc-uPA (0.1100 nmol/L), or domain fragments (0.1-100 nmol/L), with and without pharmacologic inhibitors. Incorporation of [H3]-thymidine into acid-precipitable material was measured after 24 hours of incubation (air plus 5% carbon dioxide at 37°C). Aliquots were counted in 5 mL of Cytoscint (ICN Research Products, Costa Mesa, CA) with a scintillation counter (Beckman Instruments, Fullerton, Calif). Cell proliferation. The technique used to assay cell proliferation has been described.8 Cell proliferation in response to serum (10% fetal calf serum), sc-uPA (10 nmol/ L), or domain fragments (10 nmol/L) with and without pharmacologic inhibitors was determined (see Results for time points). Cells from each well were removed with trypsinization, and were counted with a hemocytometer (Hausser Scientific, Horsham, Pa). Wound assay. The technique used for wound assay has been described.9 In brief, confluent SMCs in 60 mm2 dishes were starved for 24 hours. Each dish was divided into a 2 ⫻ 3 grid, and a linear wound was made in each hemisphere of the dish. Photomicrographs were taken of the intersections of the linear wound and each grid line (eight fields per dish) at time 0 and at 24 hours. Cells were allowed to migrate at 37°C in DMEM with or without sc-uPA (10 nmol/L), ATF (10 nmol/L), kringle (10 nmol/L), or CTF (10 nmol/L). Subsequently, migration in response to these agonists was examined in the presence of aprotinin (100 units/mL), EACA (100 ␮mol/L), pertussis toxin (100 ng/mL), GP-2A (10 ␮mol/L), or PD98059 (25 ␮mol/L). Eight fields from each dish were averaged. Trials with each reagent or inhibitor were per-

Tanski et al 215

formed in at least six separate dishes, and the results were averaged. Western blot technique. The technique for in vitro cell stimulation and Western blotting for phosphorylated and total ERK1/2 has been described in detail.9 In brief, SMCs at 80% confluence were starved for 24 hours, stimulated with agonists alone or with inhibitors, and harvested (see Results for time points). Equal amounts of protein lysates (25 ␮g) were loaded onto 12.5% polyacrylamide gels, separated with electrophoresis, and transferred to nitrocellulose membranes (BioRad). After blocking in 5% nonfat milk, blots were immunostained with antibodies to phosphorylated or total ERK. Protein bands were visualized with anti-rabbit (phospho-ERK) or anti-mouse (total ERK) IgG horseradish peroxidase– conjugated antibodies, followed by enhanced chemiluminescence (Amersham, Arlington Heights, Ill), according to the manufacturer’s protocols. Band intensity was measured with Gel-Imaging software (Kodak, Rochester, NY). Data and statistical analysis. All data are presented as mean ⫾ SEM. Statistical differences between groups were tested with the Kruskal-Wallis nonparametric test, with post hoc Dunn multiple comparison correction where appropriate. P ⬍ .05 was regarded as significant; nonsignificant differences are expressed as P ⫽ NS. RESULTS sc-uPA induces smooth muscle cell migration. scuPA induced migration of SMCs in a wound assay (Fig 1, A). This response was reproduced with ATF and kringle, but not with CTF (Fig 1, A). Incubation with the plasmin inhibitors EACA and aprotinin did not inhibit migration stimulated by sc-uPA, ATF, or kringle (Fig 1, B). The migration response to ATF was significantly decreased with addition of pertussis toxin, suggesting a receptor-linked, G␣i-mediated response (Fig 1, C). The G␣q inhibitory peptide, GP-2A, had no effect on migration induced with ATF. Cell migration to ATF was sensitive to am MEK1 inhibitor (PD98059), suggesting that the response is ERKdependent. Similar migration results were obtained for sc-uPA and kringle (Fig 1, D, E online only). The Western blot technique demonstrated that scuPA, ATF, and kringle induced time-dependent increases in ERK phosphorylation, with maximum ERK activation at 15 to 20 minutes (Fig 3, A-C). There were no changes in total ERK at any time for any of the time courses studied (data not shown). Phosphorylation of ERK by sc-uPA was sensitive to plasmin inhibitors (Fig 2, D, online only; Fig 3, A); phosphorylation of ERK by ATF and kringle was not (Fig 2, E, F, online only; Fig 3, B, C). ERK phosphorylation induced by Sc-uPA, ATF, and kringle was sensitive to both G␣i and MEK1 inhibition (Fig 2, A; Fig 2, D-F, online only; Fig 3, A-C). sc-uPA induces SMC proliferation. Sc-uPA produced a concentration-dependent increase in DNA synthesis (Fig 4, A, B) and a time-dependent increase in cell

JOURNAL OF VASCULAR SURGERY January 2004

216 Tanski et al

proliferation (Fig 4, C). These responses were induced by CTF, but not by ATF or kringle (Fig 4, A, B). Incubation with EACA and aprotinin at concentrations that markedly diminish plasmin activity significantly inhibited sc-uPA– mediated DNA synthesis (Fig 5, A). DNA synthesis stimulated with CTF and sc-uPA was significantly decreased at preincubation with pertussis toxin, suggesting a receptorlinked, G␣i-mediated response (Fig 5, B; C, online only). Incubation with GP-2A had no effect. Both sc-uPA and CTF induced thymidine incorporation that was sensitive to an MEK1 inhibitor (PD98059), suggesting that these responses were ERK-dependent (Fig 5, B; C, online only). Both sc-uPA and CTF produced time-dependent increases in ERK phosphorylation that were sensitive to inhibitors of plasmin, G␣i, and MEK1 (Fig 2, A-C; D, online only; Fig 3, A, D). DISCUSSION

Fig 1. A, Migration in response to single-chain urokinase-type plasminogen activator (sc-uPA; 10 n mol/L), amino-terminal fragment of uPA (ATF; 10 nmol/L), kringle (10 n mol/L), and carboxy-terminal fragment of uPA (CTF; 10 nmol/L) in the wound assay. Confluent plates were scratched, and reagents were added at time 0; cells were then allowed to migrate over 24 hours. Larger decreases in wound area (more negative values) represent increased migration. Values represent mean ⫾ SEM percent of control for six experiments (**P ⬍ .01 vs control). B, Migration in response to sc-uPA (10 n mol/L), ATF (10 n mol/L), and kringle (10 n mol/L) in the wound assay, with and without the plasmin inhibitors ⑀-aminocaproic acid (EACA, 100 ␮mol/L) and aprotinin (100 units/mL). Values represent mean ⫾ SEM of percent of

Studies suggest that plasminogen activators and plasmin have a significant role in development of intimal hyperplasia. After injury to the rat carotid artery, there are increases in expression of uPA and tissue-type plasminogen activator (tPA) and in plasminogen activator activity on SMCs within the vessel wall.10 Tranexamic acid, a plasmin inhibitor, reduces SMC migration after carotid balloon injury.11 When uPA is placed in a pluronic gel around an injured carotid artery, there is a significant increase in development of intimal hyperplasia.12 Expression of PAI-1, an endogenous plasminogen activator inhibitor, is increased sevenfold 3 hours after injury, and returns to baseline by 2 days; it then increases again from days 4 to 7.13,14 Elevated PAI-1 in the rat carotid artery enhances thrombosis and endothelial cell regeneration and inhibits intimal thickening.15 Overexpression of uPA and deficiency of PAI-1 in transgenic mice promote neointimal formation in response to electrical injury.16-18 However, in mice without uPAR, neointima formation is unchanged.4 Plasmin is clearly important, and it appears that in arterial injury plasmin generation by uPA, but not tissue-type plasminogen activator, is involved in stimulation of cell migration and neointima formation.19 Plasminogen knockout mice do not develop intimal hyperplasia compared with control mice, and show no MMP-9 activation (although MMP-2 activation is unchanged).20 Lamfers et al,21 using a vector composed of ATF and bovine pancreas trypsin inhibitor, showed a reduction in neointima formation, presumably by inhibiting SMC migration. The present in vitro data demonstrate roles for sc-uPA and, more important, its individual domains in both proliferation and migration of SMCs.

4™™™™™™™™™™™™™™™™™™™™™™™™™™™™™™™™™ Fig 1. (Continued) control for six experiments. C, Migration response to ATF in the wound assay with and without pertussis toxin (PTx; 100 ng/mL), GP-2A (a G␣q inhibitory peptide; 10 ␮mol/L), or a mitogen-activated protein kinase–1 inhibitor (PD98059, PD; 25 ␮mol/L). Values represent mean ⫾ SEM of percent of control for six experiments (**P ⬍ .01, ATF alone).

JOURNAL OF VASCULAR SURGERY Volume 39, Number 1

Tanski et al 217

Fig 2. A, Densitometric measurements of time-dependent extracellular signal-regulated kinase (ERK)–1/2 phosphorylation in response to carboxy-terminal fragment (CTF; 10 nmol/L) in the presence and absence of plasmin inhibitors aprotinin (100 units/mL) and ⑀-aminocaproic acid (EACA; 100 ␮mol/L). ERK phosphorylation is expressed as increase over time 0 control. B, C, Maximal ERK phosphorylation for single-chain urokinase-type plasminogen activator (sc-uPA; 10 nmol/L [B, C]), amino-terminal fragment of uPA (ATF; 10 nmol/L [B]), kringle (10 nM, [B]), and carboxy-terminal fragment of uPA (CTF; 10 nmol/L [C]) in the presence and absence of G␣i inhibitor pertussis toxin (PTx; 100 ng/mL) and mitogen-activated protein kinase–1 inhibitor PD98059 (PD; 25 ␮mol/L). All values represent mean ⫾ SEM of percent of control for three experiments (*P ⬍ .05, **P ⬍ .01, vs agonist alone).

sc-uPA is a chemotactic factor in many cell lines, and apparently requires ATF, which has a uPAR binding domain. We have shown here that sc-uPA stimulates migra-

tion through both ATF and kringle fragments of the molecule, and does not require protease activity. We have also observed that ATF stimulates SMC migration when

218 Tanski et al

JOURNAL OF VASCULAR SURGERY January 2004

Fig 3. A-D, Representative Western blots of extracellular signal-regulated kinase (ERK)–1/2 phosphorylation in response to single-chain urokinase-type plasminogen activator (sc-uPA; 10 n mol/L [A]), amino-terminal fragment (ATF; 10 n mol/L [B]), kringle (10 n mol/L [C]), and carboxy-terminal fragment (CTF; 10 n mol/L [D]) in the presence and absence of the plasmin inhibitors aprotinin (100 units/mL) and ⑀-aminocaproic acid (EACA; 100 ␮mol/L), G␣i inhibitor pertussis toxin (PTx; 100 ng/mL), and mitogen-activated protein kinase–1 inhibitor PD98059 (25 ␮mol/L). P44, Phosphorylated form of ERK2; p42, phosphorylated form of ERK1. Cells were pretreated with inhibitors for 30 minutes. Blots are representative of three experiments for each time course. Bands labeled with agonists (eg, uPA) or inhibitors (eg, pertussis toxin) represent positive and negative controls, respectively.

placed in the bottom well of a modified Boyden chamber with a Matrigel-coated membrane, supporting its role as a chemoattractant in an invasion model (unpublished observations). Others have shown that cell migration depends on specific mutant sc-uPA molecules containing either ATF or

kringle.22,23 The respective importance of each of these domains in migration is not resolved. The response to ATF may be due to binding and cleavage of a glycosyl phosphatidyl inosital (GPI)⫺linked uPA receptor (uPAR). This receptor is composed of three domains (D1D2D3); on bind-

JOURNAL OF VASCULAR SURGERY Volume 39, Number 1

Tanski et al 219

Fig 3. (Continued.)

ing with ATF it is cleaved into a cell-bound D1 fragment containing the GPI tail and a soluble D2D3 fragment. D2D3 then binds to the low-affinity receptor of chemotactic peptide fMPL, FRAL-1, to induce a G␣i-mediated migration response. The ATF response is G␣i-linked, but not G␣q-linked, and is ERK-dependent; similar results were observed in a modified Boyden chamber with a Matrigel-coated membrane (unpublished observations). We cannot exclude the possibility that these effects are G␤␥-

mediated. Little is known about the mechanism whereby kringle induces migration. We have demonstrated that it is partially G␣i-linked, but not G␣q-linked, and is ERK dependent. Kringle may induce ERK activation through receptor-dependent and receptor-independent mechanisms, which may explain why pertussis toxin did not eliminate the migration response entirely. Commercially made ATF contains the kringle fragment, and thus may also act through both receptor-dependent and receptor-independent mech-

220 Tanski et al

JOURNAL OF VASCULAR SURGERY January 2004

Fig 4. A and B, In vitro [3H]- thymidine incorporation into DNA of cultured rat vascular smooth muscle cells in response to 0.1 to 100 n mol/L of single-chain urokinase-type plasminogen activator (sc-uPA), carboxy-terminal fragment of uPA (CTF), amino-terminal fragment of uPA (ATF), and kringle. B represents results for each agonist at 10 n mol/L. Values represent mean ⫾ ⫾ SEM of increase over control for four experiments (**P ⬍ .01 vs control). C, Cell proliferation in response to sc-uPA (10 n mol/L) and CTF (10 n mol/L) as measured by cell counting. ATF and kringle at 10 n mol/L did not stimulate cell proliferation (data not shown). Reagents were added at time 0. Serum (10%) and DMEM represent positive and negative controls, respectively. Values represent mean ⫾ SEM of increase over control for six experiments.

JOURNAL OF VASCULAR SURGERY Volume 39, Number 1

Tanski et al 221

Fig 5. A, In vitro [3H]-thymidine incorporation into DNA of cultured rat vascular smooth muscle cells over 24 hours in response to single-chain urokinase-type plasminogen activator (sc-uPA; 10 n mol/L) and carboxy-terminal fragment of uPA (CTF; 10 n mol/L) in the presence and absence of plasmin inhibitors aprotinin (100 units/mL) and ⑀-aminocaproic acid (100 ␮mol/L). Values represent mean ⫾ SEM of percent of control for five experiments (**P ⬍ .01 vs sc-uPA or CTF alone). B, In vitro [3H]-thymidine incorporation into DNA of cultured rat vascular smooth muscle cells over 24 hours in response to CTF (10 n mol/L) in the presence and absence of the G␣i inhibitor pertussis toxin (PTx; 100 ng/mL), mitogen-activated protein kinase–1 inhibitor PD98059 (25 ␮mol/L), and G␣q inhibitory peptide GP-2A (10 ␮mol/L). Values represent mean ⫾ SEM of percent of control for six experiments (**P ⬍ .01 vs CTF alone).

anisms. The ability of ATF and kringle to induce migration may explain why, in vivo, mice without uPAR do not exhibit reduced neointima formation.4 In vitro data suggest that plasmin and MMPs cleave uPA into its respective fragments, and CTF and kringle fragments may compensate for inability of ATF to cleave and activate absent uPAR. Transgenic murine models suggest that uPA has a significant role in stimulating proliferation and that it is plasmin-dependent. Using mutants of sc-uPA, others have shown that cell proliferation requires the CTF portion.22,23 CTF contains a protease domain, and leads to plasmin generation, whereas plasmin inhibition reduces DNA synthesis. Plasmin induces ERK-dependent SMC proliferation (unpublished observations). CTF may mediate SMC proliferation by three separate mechanisms. First, plasmin may activate MMPs, leading to epidermal growth factor receptor– dependent mitogen-activated protein kinase activation.20,24 Inhibition of plasminogen activation decreases plasmin and thus MMP activation. Activation of epidermal growth factor receptor by sc-uPA and CTF may explain our incomplete inhibition of DNA synthesis in the presence of pertussis toxin. Second, plasmin interacts with the protease-activated receptor PAR-1 (a thrombin receptor), possibly using annexin as a cofactor, in a manner distinct from the protease thrombin.25 The PAR-1 receptor is G protein– coupled, and may bind to either G␣i or G␣q; our data suggest that it is G␣i-dependent and may involve G␤␥. Third, preliminary data from other authors suggest that CTF may bind to a novel cell surface receptor independent of uPAR and PAR-1, which remains to be identified. Inasmuch as no plasmin was added during CTF stimulation in

our experiments, it is possible that this third mechanism is indeed significant and that plasmin inhibitors affect other necessary intermediates downstream of G protein– coupled receptors. Alternatively, some plasminogen may remain in cell culture despite removal of serum during “starving.” sc-uPA is a pluripotent molecule with the ability to induce both cell migration and proliferation through G protein– coupled receptors. Migration is protease-independent, whereas cell proliferation is protease-dependent and may be mediated through G protein– coupled, proteasedependent receptors.

REFERENCES 1. Davies MG, Hagen P-O. Pathobiology of intimal hyperplasia. Br J Surg 1994;81:1254-69. 2. Biazi F. The urokinase receptor: a cell surface regulated chemokine. APMIS 1999;107:96-101. 3. Biasi F, Vassalli J-D, Dang K. Urokinase type plasminogen activator: proenzyme, receptor and inhibitors. J Cell Biol 1987;104:801-4. 4. Shen GX. Vascular cell-derived fibrinolytic regulators and atherothrombotic vascular disorders. Int J Mol Med 1998;1:399-408. 5. Lijnen HR. Matrix metalloproteinases and cellular fibrinolytic activity. Biochemistry (Moscow) 2002;67:92-8. 6. Resnati M, Guttinger M, Valcamonica S, et al. Proteolytic cleavage of the urokinase receptor substitutes for agonist-induced chemotactic effects. EMBO J 1996;15:1572-82. 7. Fazioli F, Resnati M, Sidenius N, et al. A urokinase sensitive region of the human urokinase receptor is responsible for its chemotactic activity. EMBO J 1997;16:7279-86. 8. Davies MG, Mason DP, Tran PK, Deou J, Hawkins S, Clowes AW. G-protein expression and intimal hyperplasia after arterial injury: a role for Galpha(i) proteins. J Vasc Surg 2001;33:408-18.

222 Tanski et al

9. Tanski W, Roztocil E, Davies MG. Sphingose-1-phosphate induces G␣i-coupled, PI3K/ras dependent smooth muscle cell migration. J Surg Res 2002;108:98-106. 10. Clowes AW, Reidy MA, Clowes MM, Belin D. Smooth muscle cells express urokinase during mitogenesis and tissue-type plasminogen activator during migration in injured rat carotid. Circ Res 1990;67:61-7. 11. Bendeck MP, Zempo N, Clowes AW, Galardy RE, Reidy MA. Smooth muscle cell migration and matrix metalloproteinase expression after arterial injury in the rat. Circ Res 1994;75:539-45. 12. Plekhanova OS, Solomatina MA, Domogatskii SP, Naumov VG, Tkachuk VA, Parfenova EV, et al. Urokinase stimulates whereas tissue plasminogen activator attenuates blood vessel stenosis. Ross Fiziol Zh Im I M Sechenova 2001;87:584-93. 13. Hasentaub D, Forough R, Clowes AW. Plasminogen activator inhibitor type I and tissue inhibitor of metalloproteinases-2 increase after arterial injury in rats. Circ Res 1997;80:490-6. 14. Hamdan AD, Quist WC, Gagne JB, Feener EP. ACE inhibition suppresses PAI-1 expression in the neointima of balloon injured rat aorta. Circulation 1996;93:1073-8. 15. Hasentaub D, Lea H, Clowes AW. Local plasminogen activator inhibitor type 1 overexpression in rat carotid artery enhances thrombosis and endothelial regeneration while inhibiting intimal thickening. Arterioscler Thromb Vasc Biol 2000;20:853-9. 16. Carmeliet P, Moons L, Lijnen R, Janssens S, Lupu F, Collen D, et al. Inhibitory role of PAI-1 in arterial wound healing and neointima formation. Circulation 1997;96:3180-91. 17. Carmeliet P, Moons L, Herbert J-M, Crawley J, Lupu F, Lijnen R, et al. Urokinase but not tissue plasminogen activator mediates arterial neointima formation in mice. Circ Res 1997;81:829-39. 18. Peng l, Bhatia N, Parker AC, Zhu Y, Fay WP. Endogenous vitronectin and PAI-1 promote neointima formation in murine carotid arteries. Arterioscler Thromb Vasc Biol 2002;22:934.

JOURNAL OF VASCULAR SURGERY January 2004

19. Schafer K, Konstantinides S, Riedel C, Thinnes T, Muller K, Dellas C, et al. Different mechanisms of increased luminal stenosis after arterial injury in mice deficient for urokinase or tissue-type plasminogen activator. Circulation 2002;106:1847. 20. Lijnen HR, VanHoef B, Lupu F, Moon L, Carmeliet P, Collen D. Function of the plasminogen/plasmin and MMP systems after vascular injury in mice with targeted inactivation of fibrinolytic genes. Arterioscler Thromb Vasc Biol 1998;18:1035-45. 21. Lamfers ML, Lardenoye JH, deVries MR, Aalders MC, Engelse MA, Grimbergen JM, et al. In vivo suppression of restenosis in ballooninjured rat carotid artery by adenovirus-mediated gene transfer of the cell surface directed plasmin inhibitor ATF BPTI. Gene Ther 2001;8: 534-41. 22. Stepanova V, Mukhina S, Kohler E, Resink TJ, Erne P, Tkachuk VA. Urokinase plasminogen activator induces human smooth muscle cell migration and proliferation via distinct receptor dependent and proteolysis-dependent mechanisms. Mol Cell Biochem 1999;195:199-206. 23. Nguyen DH, Hussaini IM, Gonias SL. Binding of uPA to its receptor in MCF-7 cells activates ERK1 and 2, which is required for increased motility. J Biol Chem 1998;273:8502-607. 24. Prenzel N, Zwick E, Daub H, Leserer M, Abraham R, Wallasch C, et al. EGF receptor transactivation by G-protein-coupled receptors requires metalloproteinase cleavage of proHB-EGF. Nature 1999;402:884-8. 25. Pendurthi UR, Ngyuen M, Andrade-Gordon P, Petersen LC, Rao LV. Plasmin induces Cyr61 gene expression in fibroblasts via proteaseactivated receptor-1 and p44/42 mitogen-activated protein kinasedependent signaling pathway. Arterioscler Thromb Vasc Biol 2002;22: 1421-6. Submitted Apr 9, 2003; accepted Jun 21, 2003.