Title: Maspin Inhibits Cell Migration in the Absence of Protease Inhibitory Activity
Abstract: Maspin is a member of the serpin family of protease inhibitors and is a tumor suppressor gene acting at the level of tumor invasion and metastasis. This in vivo activity correlates with the ability of maspin to inhibit cell migrationin vitro. This behavior suggests that maspin inhibits matrix-degrading proteases, such as those of the plasminogen activation system, in a similar manner to the serpin PAI-1. However, there is controversy concerning the protease inhibitory activity of maspin. It is devoid of activity against a wide range of proteases, in common with other non-inhibitory serpins, but has recently been reported to inhibit plasminogen activators associated with cells and other biological surfaces (Sheng, S. J., Truong, B., Fredrickson, D., Wu, R. L., Pardee, A. B., and Sager, R. (1998) Proc. Natl. Acad. Sci. U. S. A. 95, 499–504; McGowen, R., Biliran, H., Jr., Sager, R., and Sheng, S. (2000) Cancer Res. 60, 4771–4778). We have compared the effects of maspin with those of PAI-1 in a range of situations in which plasminogen activation is potentiated, reflecting the biological context of this proteolytic system: urokinase-type plasminogen activator bound to its receptor on the surface of tumor cells, tissue-type plasminogen activator specifically bound to vascular smooth muscle cells, fibrin, and the prion protein. Maspin was found to have no inhibitory effect in any of these situations, in contrast to the efficient inhibition observed with PAI-1, but nevertheless maspin inhibited the migration of both tumor and vascular smooth muscle cells. We conclude that maspin is a non-inhibitory serpin and that protease inhibition does not account for its activity as a tumor suppressor. Maspin is a member of the serpin family of protease inhibitors and is a tumor suppressor gene acting at the level of tumor invasion and metastasis. This in vivo activity correlates with the ability of maspin to inhibit cell migrationin vitro. This behavior suggests that maspin inhibits matrix-degrading proteases, such as those of the plasminogen activation system, in a similar manner to the serpin PAI-1. However, there is controversy concerning the protease inhibitory activity of maspin. It is devoid of activity against a wide range of proteases, in common with other non-inhibitory serpins, but has recently been reported to inhibit plasminogen activators associated with cells and other biological surfaces (Sheng, S. J., Truong, B., Fredrickson, D., Wu, R. L., Pardee, A. B., and Sager, R. (1998) Proc. Natl. Acad. Sci. U. S. A. 95, 499–504; McGowen, R., Biliran, H., Jr., Sager, R., and Sheng, S. (2000) Cancer Res. 60, 4771–4778). We have compared the effects of maspin with those of PAI-1 in a range of situations in which plasminogen activation is potentiated, reflecting the biological context of this proteolytic system: urokinase-type plasminogen activator bound to its receptor on the surface of tumor cells, tissue-type plasminogen activator specifically bound to vascular smooth muscle cells, fibrin, and the prion protein. Maspin was found to have no inhibitory effect in any of these situations, in contrast to the efficient inhibition observed with PAI-1, but nevertheless maspin inhibited the migration of both tumor and vascular smooth muscle cells. We conclude that maspin is a non-inhibitory serpin and that protease inhibition does not account for its activity as a tumor suppressor. Proteolytic activity is a key event in cell migration and invasion, being required to dynamically modulate interactions between the cell and its surrounding extracellular matrix (1Werb Z. Cell. 1997; 91: 439-442Google Scholar). Multiple protease systems are implicated in this process, including the serine proteases of the plasminogen activation system (2Ellis V. Murphy G. FEBS Lett. 2001; 506: 1-5Google Scholar). In the pericellular environment the activity of this system is regulated by binding of the proteases or their zymogens to specific cell-surface receptors or binding sites. The plasminogen activator uPA 1The abbreviations used are: uPA, urokinase-type plasminogen activator; uPAR, cellular receptor for uPA; tPA, tissue-type plasminogen activator; RSL, reactive-site loop; PrP, prion protein; VSMC, vascular smooth muscle cells; AMC, amido-4-methylcoumarin; α1-PI, α1-proteinase inhibitor; S, stressed; R, relaxed binds to uPAR, a well characterized glycosylphosphatidylinositol-anchored membrane protein (3Ploug M. Ellis V. FEBS Lett. 1994; 349: 163-168Google Scholar); tPA binds to cell-surface proteins on cell types including endothelial (4Hajjar K.A. Jacovina A.T. Chacko J. J. Biol. Chem. 1994; 269: 21191-21197Google Scholar) and VSMC (5Ellis V. Whawell S.A. Blood. 1997; 90: 2312-2322Google Scholar, 6Werner F. Razzaq T.M. Ellis V. J. Biol. Chem. 1999; 274: 21555-21561Google Scholar); and plasminogen binds to multiple cell-surface molecules (7Hawley S.B. Green M.A. Miles L.A. Thromb. Haemostasis. 2000; 84: 882-890Google Scholar). These interactions lead to assembly of complexes on the cell surface that greatly increase plasmin generation (5Ellis V. Whawell S.A. Blood. 1997; 90: 2312-2322Google Scholar, 8Ellis V. Behrendt N. Danø K. J. Biol. Chem. 1991; 266: 12752-12758Google Scholar, 9Ellis V. Whawell S.A. Werner F. Deadman J.J. Biochemistry. 1999; 38: 651-659Google Scholar). The activity of tPA is also potentiated by binding to protein cofactors, including fibrin (10Hoylaerts M. Rijken D.C. Lijnen H.R. Collen D. J. Biol. Chem. 1982; 257: 2912-2919Google Scholar) and PrP (11Ellis V. Daniels M. Misra R. Brown D.R. Biochemistry. 2002; 41: 6891-6896Google Scholar). This powerful proteolytic system is inhibited by members of the serpin (serine protease inhibitor) family, in particular PAI-1 (SERPINE1), which can inhibit free, cofactor-bound and cell-associated plasminogen activators (6Werner F. Razzaq T.M. Ellis V. J. Biol. Chem. 1999; 274: 21555-21561Google Scholar, 12Ellis V. Wun T.-C. Behrendt N. Rønne E. Danø K. J. Biol. Chem. 1990; 265: 9904-9908Google Scholar). The protease inhibitory activity of PAI-1 has been shown to be of importancein vivo, inhibiting VSMC migration (13Carmeliet P. Moons L. Lijnen R. Janssens S. Lupu F. Collen D. Gerard R.D. Circulation. 1997; 96: 3180-3191Google Scholar) and regulating tumor angiogenesis (14Bajou K. Masson V. Gerard R.D. Schmitt P.M. Albert V. Praus M. Lund L.R. Frandsen T.L. Brunner N. Dano K. Fusenig N.E. Weidle U. Carmeliet G. Loskutoff D. Collen D. Carmeliet P. Foidart J.M. Noel A. J. Cell Biol. 2001; 152: 777-784Google Scholar), and its expression correlates with disease progression and prognosis in human cancers (15Duffy M.J. Clin. Chem. 2002; 48: 1194-1197Google Scholar). Some serpins have biological activities independent of protease inhibition. For example, PAI-1 binds to vitronectin, modulating cell adhesion and migration (16Stefansson S. Lawrence D.A. Nature. 1996; 383: 441-443Google Scholar). Other serpins lack intrinsic inhibitory activity. Examples of this are ovalbumin, thyroid-binding globulin (SERPINA6), angiotensinogen (SERPINA8), and pigment epithelium-derived factor (SERPINF1), which has neurotrophic and anti-angiogenic activity (17Silverman G.A. Bird P.I. Carrell R.W. Church F.C. Coughlin P.B. Gettins P.G. Irving J.A. Lomas D.A. Luke C.J. Moyer R.W. Pemberton P.A. Remold-O'Donnell E. Salvesen G.S. Travis J. Whisstock J.C. J. Biol. Chem. 2001; 276: 33293-33296Google Scholar, 18Dawson D.W. Volpert O.V. Gillis P. Crawford S.E. Xu H. Benedict W. Bouck N.P. Science. 1999; 285: 245-248Google Scholar). Maspin (SERPINB5) is thought to be another non-inhibitory serpin. Maspin was first identified as a class II tumor suppressor in human breast cancer (19Zou Z. Anisowicz A. Hendrix M.J. Thor A. Neveu M. Sheng S. Rafidi K. Seftor E. Sager R. Science. 1994; 263: 526-529Google Scholar), and transfection of maspin into carcinoma cells reduces their metastatic potential in vivo (19Zou Z. Anisowicz A. Hendrix M.J. Thor A. Neveu M. Sheng S. Rafidi K. Seftor E. Sager R. Science. 1994; 263: 526-529Google Scholar, 20Shi H.Y. Zhang W. Liang R. Abraham S. Kittrell F.S. Medina D. Zhang M. Cancer Res. 2001; 61: 6945-6951Google Scholar). It is a predominantly cytoplasmic protein, but is also secreted to the cell surface (21Pemberton P.A. Tipton A.R. Pavloff N. Smith J. Erickson J.R. Mouchabeck Z.M. Kiefer M.C. J. Histochem. Cytochem. 1997; 45: 1697-1706Google Scholar), where it has been shown to reduce the migration of various cell types in vitro (22Sheng S. Pemberton P.A. Sager R. J. Biol. Chem. 1994; 269: 30988-30993Google Scholar, 23Sheng S. Carey J. Seftor E.A. Dias L. Hendrix M.J. Sager R. Proc. Natl. Acad. Sci. U. S. A. 1996; 93: 11669-11674Google Scholar) and to inhibit angiogenesis in both in vitro and in vivo models (24Zhang M. Volpert O. Shi Y.H. Bouck N. Nat. Med. 2000; 6: 196-199Google Scholar). These activities of maspin are consistent with those of a protease inhibitor, yet extensive biochemical characterization has failed to demonstrate a protease target for maspin, and it lacks key features of inhibitory serpins (25Pemberton P.A. Wong D.T. Gibson H.L. Kiefer M.C. Fitzpatrick P.A. Sager R. Barr P.J. J. Biol. Chem. 1995; 270: 15832-15837Google Scholar). Therefore the mechanisms underlying its biological activities are considered to be largely unresolved (26Hendrix M.J. Nat. Med. 2000; 6: 374-376Google Scholar). However, recent studies have suggested that maspin does exhibit inhibitory activity toward the plasminogen activators uPA and tPA, but only when these proteases are bound to macromolecular cofactors, that is tPA bound to fibrin (27Sheng S.J. Truong B. Fredrickson D. Wu R.L. Pardee A.B. Sager R. Proc. Natl. Acad. Sci. U. S. A. 1998; 95: 499-504Google Scholar) and uPA on the cell surface (28McGowen R. Biliran Jr., H. Sager R. Sheng S. Cancer Res. 2000; 60: 4771-4778Google Scholar, 29Biliran Jr., H. Sheng S. Cancer Res. 2001; 61: 8676-8682Google Scholar). Using techniques that we have previously established to investigate the activity and inhibition of cell-surface plasminogen activators, we demonstrate here that maspin has no inhibitory activity against these protease in either cellular environments or other situations in which their activities are potentiated and that reflect the biological context of this proteolytic system. Nevertheless, maspin was able to inhibit cell migration, strongly suggesting that this activity of maspin is not dependent on protease inhibition. Recombinant maspin expressed inSaccharomyces cerevisiae (25Pemberton P.A. Wong D.T. Gibson H.L. Kiefer M.C. Fitzpatrick P.A. Sager R. Barr P.J. J. Biol. Chem. 1995; 270: 15832-15837Google Scholar) was obtained from Andy Robertson (Department of Biochemistry, University of Iowa). tPA (Actilyse) was obtained from Boehringer-Ingleheim (Ingleheim, Germany). Recombinant PAI-1 was obtained from Calbiochem and its concentration determined by titration against tPA (6Werner F. Razzaq T.M. Ellis V. J. Biol. Chem. 1999; 274: 21555-21561Google Scholar). Lys-plasminogen (i.e. with Lys77 as N terminus) was obtained from Enzyme Research Laboratories (Swansea, UK). The soluble fibrin fragment preparation Desafib-X was obtained from American Diagnostica (Greenwich, CT). Recombinant PrP was prepared as described previously (11Ellis V. Daniels M. Misra R. Brown D.R. Biochemistry. 2002; 41: 6891-6896Google Scholar). The fibrosarcoma cell line HT-1080 and DU 145 prostate carcinoma cells were from ATCC, and VSMC of aortic origin were isolated and cultures as described previously (5Ellis V. Whawell S.A. Blood. 1997; 90: 2312-2322Google Scholar). Plasminogen activation by uPAR-bound uPA on the surface of HT-1080 and DU 145 cells was determined as described previously (8Ellis V. Behrendt N. Danø K. J. Biol. Chem. 1991; 266: 12752-12758Google Scholar). In brief, cells grown to confluence in 24-well plates were washed in phosphate-buffered saline to remove unbound uPA and incubated at 37 °C with varying fixed concentrations of plasminogen (20–200 nm), the plasmin specific substrate H-d-Val-Leu-Lys-AMC (0.25 mm), and varying concentrations of maspin or PAI-1. Plasmin generated by endogenously bound uPA was measured continuously as change in fluorescence in a SpectraMAX Gemini microplate reader (Molecular Device, Sunnyvale, CA) at λ 360/440 nm. Plasmin concentration was determined as δF and plasmin generation represented as δF versus time. Second-order inhibition rate constants were calculated from inhibition curves according to (30Liu W. Tsou C.L. Biochim. Biophys. Acta. 1986; 870: 185-190Google Scholar), as described previously (6Werner F. Razzaq T.M. Ellis V. J. Biol. Chem. 1999; 274: 21555-21561Google Scholar, 12Ellis V. Wun T.-C. Behrendt N. Rønne E. Danø K. J. Biol. Chem. 1990; 265: 9904-9908Google Scholar). Plasminogen activation by tPA bound to VSMC was determined essentially as described previously (5Ellis V. Whawell S.A. Blood. 1997; 90: 2312-2322Google Scholar, 6Werner F. Razzaq T.M. Ellis V. J. Biol. Chem. 1999; 274: 21555-21561Google Scholar). In brief, cells grown to confluence in 24-well plates were incubated with tPA (10 nm) for 20 min at 37 °C, washed extensively to remove unbound tPA, and plasminogen activation determined as described above. In these experiments plasmin generation was represented as F versustime2. tPA-catalyzed plasminogen activation stimulated by fibrin was determined by incubation of tPA (1.5 nm), Lys-plasminogen (25 nm), and varying concentrations of fibrin fragments in 0.05 m Tris-HCl, 0.1 mNaCl, pH 7.4, containing H-d-Val-Leu-Lys-AMC (0.25 mm). In preliminary experiments the fibrin concentration giving maximal stimulation (∼250-fold) was determined and found to be 250 μg/ml. This concentration was used for all subsequent experiments. Varying concentrations of either maspin or PAI-1 were included in these experiments and inhibition of plasminogen activation determined as described above. Similar experiments were performed to determine the effect of PrP on tPA inhibition by maspin. Recombinant PrP in its divalent metal ion-free form (apo-PrP) was included, in place of fibrin, at an optimal concentration of 50 μg/ml leading to more than a 250-fold stimulation of plasmin generation (11Ellis V. Daniels M. Misra R. Brown D.R. Biochemistry. 2002; 41: 6891-6896Google Scholar). Cell migration was determined using time-lapse video microscopy. VSMC were seeded into four-well plates at a density of 7500 cells/ml/well in Medium 231 containing Smooth Muscle Cell Growth Supplement (Cascade Biologics, Portland, OR). After 24 h the medium was changed to L15 air-buffered medium (Sigma), 0.1% bovine serum albumin containing varying concentrations of maspin (0–200 nm). Cell movement was recorded by computerized time-lapse video microscopy (Nikon, Kingston upon Thames, UK) with images acquired every 5 min for 15 h. 10–20 cells were tracked per movie and cell movement quantified using Lucia 32G/Magic 4.11 software (Nikon) and expressed as micrometer/hour. The best characterized pericellular proteolytic system is the uPA/uPAR system, which specifically activates cell-associated plasminogen. We have previously shown that uPA bound to cellular uPAR is efficiently inhibited by PAI-1, with kinetics similar to those in solution (12Ellis V. Wun T.-C. Behrendt N. Rønne E. Danø K. J. Biol. Chem. 1990; 265: 9904-9908Google Scholar). Fig.1 A shows the inhibition of endogenous uPA on HT-1080 fibrosarcoma cells. Increasing concentrations of PAI-1 (up to 20 nm) lead to a complete inhibition of uPA activity in a time-dependent manner, consistent with the standard serpin inhibitory mechanism. The calculated second-order inhibition rate constant, 4.1 × 106m−1 s−1, compares with 7.9 × 106m−1 s−1determined for uPA in solution. In sharp contrast to this, maspin at concentrations up to 200 nm completely failed to inhibit uPA activity (Fig.1 B). Decreasing the concentration of plasminogen in the experiment to greater than 10-fold below K m, to minimize possible competitive effects on the reaction with maspin, did not lead to an observable inhibitory effect. In the absence of cells, uPA bound to recombinant soluble uPAR was also not inhibited by maspin (data not shown). From these data in Fig. 1 B it can be estimated (assuming a minimum detection level of 5% inhibition) that the maximum value of the rate constant for uPA inhibition by maspin is ∼400 m−1 s−1, 4 orders of magnitude less than for inhibition by PAI-1. Experiments were also performed using DU 145 prostate carcinoma cells (as used in the study of McGowen et al. (28McGowen R. Biliran Jr., H. Sager R. Sheng S. Cancer Res. 2000; 60: 4771-4778Google Scholar)), and a similar lack of inhibition by maspin was observed (data not shown). tPA is also known to associate with certain cell types, and we have shown that VSMC bind tPA and stimulate its activity more than 100-fold (5Ellis V. Whawell S.A. Blood. 1997; 90: 2312-2322Google Scholar, 6Werner F. Razzaq T.M. Ellis V. J. Biol. Chem. 1999; 274: 21555-21561Google Scholar). This involves a putative receptor-mediated mechanism analogous to the uPAR-dependent mechanism for the activation of cell-associated plasminogen. Fig. 1, C and D, show that tPA specifically bound to VSMC is efficiently inhibited by PAI-1, but again no inhibition was detectable at maspin concentrations of up to 200 nm. Plasmin generation at the highest maspin concentration was consistently increased, but was also observed with the non-inhibitory serpin ovalbumin (data not shown), suggesting an additional stimulation possibly by cleaved serpin. A similar effect has previously been observed (27Sheng S.J. Truong B. Fredrickson D. Wu R.L. Pardee A.B. Sager R. Proc. Natl. Acad. Sci. U. S. A. 1998; 95: 499-504Google Scholar), and high concentrations of both maspin and ovalbumin led to a small stimulation of tPA activity in solution (data not shown). tPA activity is stimulated, in the absence of cells, by binding to fibrin. This involves the binding of tPA and plasminogen to fibrin as a catalytic "template" and direct effects on catalytic activity. In the presence of this very specific stimulatory mechanism, PAI-1 is still an effective inhibitor of tPA (Fig. 1 E), but maspin once again failed to manifest inhibitory activity. We have recently shown that certain conformations of PrP can enhance tPA-catalyzed plasminogen activation by greater than 300-fold by a template mechanism involving independent and specific interactions of PrP with plasminogen and tPA (11Ellis V. Daniels M. Misra R. Brown D.R. Biochemistry. 2002; 41: 6891-6896Google Scholar). Maspin was also unable to inhibit tPA activity in this environment (data not shown). Previous studies have correlated the inhibitory effects of maspin on cell migration to the inhibition of plasminogen activator activity (28McGowen R. Biliran Jr., H. Sager R. Sheng S. Cancer Res. 2000; 60: 4771-4778Google Scholar, 29Biliran Jr., H. Sheng S. Cancer Res. 2001; 61: 8676-8682Google Scholar). We have determined the effect of maspin on the migration of VSMC using time-lapse video microscopy. Fig. 2 shows that maspin inhibited VSMC migration in a biphasic manner, consistent with previous observations on other cell types (22Sheng S. Pemberton P.A. Sager R. J. Biol. Chem. 1994; 269: 30988-30993Google Scholar, 28McGowen R. Biliran Jr., H. Sager R. Sheng S. Cancer Res. 2000; 60: 4771-4778Google Scholar). Migration of HT-1080 cells was also inhibited in a similar manner (data not shown). Interestingly, the time course of migration in the presence of maspin was linear (Fig. 2,inset), suggesting that maspin exerts an immediate inhibitory effect. These experiments both verify the biological activity of the maspin used in these experiments and more importantly demonstrate that the inhibitory effect of maspin on cell migration, thought to be intimately involved in its tumor suppressing activity, is not a consequence of protease inhibition. The activity of the plasminogen activation system is regulated by two opposing mechanisms: cell-surface-binding sites and protein cofactors, which facilitate productive catalytic interactions with plasminogen and thereby potentiate plasmin generation, and serpin inhibitors, which temporally and spatially restrict the activities of the proteases. These mechanisms have been shown to have a complex interplay in vivo, for example, in the regulation of angiogenesis (31Devy L. Blacher S. Grignet-Debrus C. Bajou K. Masson V. Gerard R.D. Gils A. Carmeliet G. Carmeliet P. Declerck P.J. Noel A. Foidart J.M. FASEB J. 2002; 16: 147-154Google Scholar). The ability of maspin to inhibit cell migration and its tumor suppressing activity in vivo are consistent with the inhibition of pericellular proteases, particularly the plasminogen activators. However the data presented here show that in a wide variety of situations in which the functional activity of the plasminogen activation system is highly up-regulated, and in which PAI-1 is an extremely effective inhibitor, maspin has no detectable inhibitory effect. Despite this lack of protease inhibitory activity, maspin was nevertheless able to inhibit the migration of both VSMC and tumor cells in a biphasic manner consistent with previous reports (22Sheng S. Pemberton P.A. Sager R. J. Biol. Chem. 1994; 269: 30988-30993Google Scholar). Our observations are consistent with the molecular characteristics of maspin in relation to current knowledge of serpin mechanisms. This involves complex formation between protease and serpin and cleavage at the P1 residue of the reactive-site loop (RSL) followed by insertion of this loop into the major β-sheet as a new central strand and translocation of the protease to the opposite pole of the serpin leading to structural alterations in the now covalently bound protease (32Huntington J.A. Read R.J. Carrell R.W. Nature. 2000; 19: 923-926Google Scholar). This mechanism is critically dependent on a number of features of the serpin RSL, one being its length (32Huntington J.A. Read R.J. Carrell R.W. Nature. 2000; 19: 923-926Google Scholar, 33Zhou A. Carrell R.W. Huntington J.A. J. Biol. Chem. 2001; 276: 27541-27547Google Scholar). In the far majority of inhibitory serpins the RSL has 17 residues (determined from the Glu residue of the proximal hinge region to the reactive-site P1 residue) and 16 residues in the remainder (Fig.3). Although a three-dimensional structure is not available for maspin, its sequence suggests that it has the shortest RSL of both the inhibitory and non-inhibitory serpins. Arg340 is the putative P1 residue in maspin, giving an RSL of just 13 residues. A potential alternative P1 residue for cleavage by serine proteases with trypsin-like specificity is Lys345, which would give an 18-residue RSL. The length of neither of these RSLs appears to be compatible with protease inhibition. For maspin to have an RSL of 16 or 17 residues, the P1 residue would be either Gln343 or His344, neither of which is a suitable P1 residue for serine protease inhibition. Gln is not found as a P1 residue in any serpin, and His is found only in the "fast" isoform of α1-PI from guinea pig, a species with multiple α1-PI isoforms and homologs (34Suzuki Y. Yoshida K. Honda E. Sinohara H. J. Biol. Chem. 1991; 266: 928-932Google Scholar), suggesting that this protein may not be inhibitory. Another critical feature of the RSL is its sequence, as incorporation of the RSL into the β-sheet requires residues to be compatible with adopting β conformation and not to involve burial of unfavorable side chains (33Zhou A. Carrell R.W. Huntington J.A. J. Biol. Chem. 2001; 276: 27541-27547Google Scholar). Maspin lacks the Ala-rich sequence found in the RSL of most inhibitory serpins, instead having bulky or charged residues including Ile334 and Glu335. Pro337at the P8 position is particularly unfavorable, being a Thr residue in the majority of inhibitory serpins and a critical determinant of RSL insertion (35Harrop S.J. Jankova L. Coles M. Jardine D. Whittaker J.S. Gould A.R. Meister A. King G.C. Mabbutt B.C. Curmi P.M. Struct. Fold. Des. 1999; 7: 43-54Google Scholar). P14 is also important in regulating serpin inhibitory function and is also a Thr residue in the inhibitory serpins but Gly in maspin. The introduction of a P14 Thr → Gly mutation in PAI-1 leads to a significant reduction in inhibitory activity (36Lawrence D.A. Olson S.T. Muhammad S. Day D.E. Kvassman J.O. Ginsburg D. Shore J.D. J. Biol. Chem. 2000; 275: 5839-5844Google Scholar). Maspin also lacks the hinge region P12 Ala residue found in all inhibitory, but never in non-inhibitory, serpins. A corollary of the serpin mechanism is that RSL cleavage by non-target proteases induces a transition from a "stressed" (S) to a "relaxed" (R) form by incorporation of the cleaved RSL into the major β-sheet, equivalent to the insertion occurring during the inhibitory mechanism. However, it has previously been shown that maspin does not undergo this hallmark S → R transition on cleavage at Arg340, the putative P1 residue (25Pemberton P.A. Wong D.T. Gibson H.L. Kiefer M.C. Fitzpatrick P.A. Sager R. Barr P.J. J. Biol. Chem. 1995; 270: 15832-15837Google Scholar), consistent with the preceding structural considerations. Other non-inhibitory serpins also fail to undergo this conformational transition (37Stein P.E. Tewkesbury D.A. Carrell R.W. Biochem. J. 1989; 262: 103-107Google Scholar, 38Mast A.E. Enghild J.J. Pizzo S.V. Salvesen G. Biochemistry. 1991; 30: 1723-1730Google Scholar, 39Becerra S.P. Sagasti A. Spinella P. Notario V. J. Biol. Chem. 1995; 270: 25992-25999Google Scholar). These observations strongly suggest that maspin cannot be an inhibitory serpin, in agreement with our failure to detect inhibition of plasminogen activator activity under a wide range of conditions. Our conclusions differ from those of Sheng and co-workers, who claimed that maspin has protease inhibitory activity against both tPA bound to a fibrin surface (27Sheng S.J. Truong B. Fredrickson D. Wu R.L. Pardee A.B. Sager R. Proc. Natl. Acad. Sci. U. S. A. 1998; 95: 499-504Google Scholar) and uPA associated with tumor cells (28McGowen R. Biliran Jr., H. Sager R. Sheng S. Cancer Res. 2000; 60: 4771-4778Google Scholar, 29Biliran Jr., H. Sheng S. Cancer Res. 2001; 61: 8676-8682Google Scholar). This is not easily reconciled, but two lines of argumentation can be proposed that support our conclusions. The first is that the previously reported effects were not characteristic of the standard covalent serpin inhibitory mechanism, being more suggestive of competitive inhibition. Non-inhibitory serpins act as protease substrates and, at sufficiently high concentrations, will act as competitive inhibitors in the same way as other competing substrates. However, as the kinetic mechanism underlying the stimulation of plasminogen activation in the various situations studied here is a large reduction in theK m for plasminogen, the reaction with substrate plasminogen is highly favored and reactions with potential competing substrates equally disfavored. We have used plasminogen concentrations both above and below K m, but no effects indicative of maspin behaving as a competing substrate were observed. Therefore, our data suggest that surface-bound plasminogen activators are no different to the soluble proteases in their reactivity with maspin, with neither being inhibited. The second consideration, in the case of uPA, is that our observations are consistent with the known independence of the C-terminal catalytic domain from the N-terminal uPAR-binding domain (40Oswald R.E. Bogusky M.J. Bamberger M. Smith R.A.G. Dobson C.M. Nature. 1989; 337: 579-582Google Scholar) and our previous observations on the mechanism of enhancement of plasminogen activation by uPAR. We have shown that the catalytic activity of uPA is not affected by uPAR and that the enhanced plasminogen activation is due to the formation of catalytically favored complexes with cell-associated plasminogen (8Ellis V. Behrendt N. Danø K. J. Biol. Chem. 1991; 266: 12752-12758Google Scholar, 9Ellis V. Whawell S.A. Werner F. Deadman J.J. Biochemistry. 1999; 38: 651-659Google Scholar,41Ellis V. J. Biol. Chem. 1996; 271: 14779-14784Google Scholar). A consequence of this is that both uPAR-bound uPA and uPA in solution are inhibited by the plasminogen activator inhibitors PAI-1 and PAI-2 with similar kinetics (12Ellis V. Wun T.-C. Behrendt N. Rønne E. Danø K. J. Biol. Chem. 1990; 265: 9904-9908Google Scholar). Therefore, the reaction of receptor-bound uPA with other serpins would be expected to be similarly unaffected, i.e. maspin would not be expected to inhibit either free or cell-associated uPA. Maspin has been reported to bind specifically to the cell surface (23Sheng S. Carey J. Seftor E.A. Dias L. Hendrix M.J. Sager R. Proc. Natl. Acad. Sci. U. S. A. 1996; 93: 11669-11674Google Scholar,28McGowen R. Biliran Jr., H. Sager R. Sheng S. Cancer Res. 2000; 60: 4771-4778Google Scholar), raising the possibility that this interaction conformationally converts maspin into an inhibitory form or facilitates a reaction between uPA and maspin by close juxtaposition of the proteins. Neither of these possibilities are compatible with the previous considerations regarding the requirements for an inhibitory RSL, although the latter could favor competitive substrate-like behavior. However, our observations provide no evidence for such an effect under conditions where the inhibitory effects of PAI-1 are readily detected. Although it cannot be completely excluded that specific conditions favor inhibitory-like activity in maspin, the observations here that maspin inhibits cell migration in the absence of detectable protease inhibitory activity demonstrates that this is not the mechanism responsible for the biological activity of maspin. Our data are consistent with reports that maspin both directly and indirectly affects cell adhesion, a critical event in the regulation of cell motility. Maspin has recently been shown to bind directly to collagen, an interaction that may contribute to cell adhesion (42Blacque O.E. Worrall D.M. J. Biol. Chem. 2002; 277: 10783-10788Google Scholar). Interestingly, maspin has also been shown to alter the expression profile of integrins in breast carcinoma cells, in particular inducing the expression of the α5β1 fibronectin receptor (43Seftor R.E. Seftor E.A. Sheng S. Pemberton P.A. Sager R. Hendrix M.J. Cancer Res. 1998; 58: 5681-5685Google Scholar). Although we have shown here that maspin does not inhibit the activity of uPAR-bound uPA, the reported increase in α5β1 expression by maspin may potentially lead to an indirect effect on this proteolytic system. uPAR is known to associate with α5β1 (44Aguirre-Ghiso J.A. Liu D. Mignatti A. Kovalski K. Ossowski L. Mol. Biol. Cell. 2001; 12: 863-879Google Scholar), and recent observations in this laboratory indicate that this interaction may lead to a reduction in uPA binding and a concomitant reduction in cell-surface plasminogen activation. 2R. Bass, F. Berditchevski, and V. Ellis, unpublished observations. Therefore, despite lacking protease inhibitory activity, it is possible that maspin can indirectly influence the activity of the cell-surface plasminogen activation system and that this mechanism may contribute to its function as a tumor suppressor. We thank Andy Robertson (University of Iowa) for the gift of recombinant maspin.